Santee City Council
June 11, 2025

TO: Mayor and Councilmembers
Gary Halbert, Interim City Manager
Shawn Hagerty, City Attorney
FROM: James Jeffries, City Clerk
DATE: June 11, 2025
SUBJ: Updated Council Meeting Materials – June 11, 2025 PUBLIC HEARING:
(6) Public Hearing for the Development of Fanita Ranch, to Consider Certifying Final Recirculated Revised Environmental Impact Report Including Second Recirculated Sections (ENV-2025-0003, AEIS 2022-4, AEIS 2017-11), and Approving the Fanita Ranch Development Plan and Development Review Permit (DR-2025-0001), Vesting Tentative Map (TM2025-0001), and Conditional Use Permits for Public Parks (CUP2025-0001 and CUP-2025-0002), and a Fire Station (CUP-2025-0003). (Planning and Building – Sawa)
The attached correspondence for above mentioned Item was received and is provided for your consideration.

From: RANDY AVERY
Sent: Monday, June 9, 2025 5:41:07 PM
To: Sandi Sawa <[email protected]>
Subject: Fanita Ranch Project
I am totally against the Fanita Ranch Project. We are in a fire zone, and will have limited access to roads leaving our homes if there is another fire in our area.

The traffic is already unbearable and congested in every direction, adding 3000 more homes in the area is totally ludacris!
Why do we have to continue to try to protect our neighborhood? Betty Avery
on
From: C COSTANTINO JR
Sent: Monday, June 9, 2025 12:53 PM
To: Sandi Sawa <[email protected]>
Cc: John Minto <[email protected]>; Ronn Hall <[email protected]>; Dustin Trotter <[email protected]>; [email protected]; Laura Koval <[email protected]>
Subject: Citizen Response & Objections to Fanita Ranch Final EIR Please see the attached letter denoting various concerns & objections to the Fanita Ranch Development project, submitted for the scheduled Public Hearing during the June 11th, 2025 City Council Meeting.

Please provide confirmation of receipt of this email / letter.

Thank You,
Carl Costantino
June 7, 2025 Page 1 of 3
City of Santee
AƩn: Sandi Sawa, Director of Planning, Mayor John Minto, & the City Council of Santee, Re: Fanita Ranch Development, upcoming June 11th, 2025 Public Hearing As a concerned, long term (45+ years) resident of the City of Santee, this leƩer serves to voice my strong opposiƟon to the proposed Fanita Ranch project, for the following reasons:
1) This project is fundamentally incompaƟble with the current and projected capacity of Santee’s infrastructure. Santee is already experiencing the consequences of rapid, dense development without concurrent Infrastructure.

1.1. This development plans to add nearly 3,000 addiƟonal housing units, along with commercial spaces, parks, trails, and community faciliƟes, on approximately over 2,600 acres of land north of the city.

1.2. City Street Traffic congesƟon has worsened considerably, especially along Mission Gorge Rd, Cuyamaca Rd, Mast Blvd, & Magnolia Ave, with Cuyamaca & Magnolia being 2 of the 3 planned roadways into Fanita Ranch.

1.3. State Route 52, which is over capacity (effecƟvely the day it opened), is in gridlock mornings & aŌernoons, on a daily basis.

1.4. Adding an influx of thousands of new residents and vehicles from this development can only exacerbate these issues, & will fundamentally alter the character of our community, shiŌing away from a livable, open-space town into another traffic-clogged, overextended suburban grid.

1.5. The Fanita Ranch proposal indicates three disƟnct “villages”, yet the necessary road extensions (Fanita Parkway, Cuyamaca Street, & Magnolia Avenue) at this Ɵme remain largely theoreƟcal.

There is no funding plans nor evidence that the city or county is prepared to expand and maintain the road network, uƟliƟes, and emergency response services at the scale required to support such an expansion.

1.6. Simply put, we do not have the infrastructure, nor the funding, to accommodate this magnitude of growth without significant, long-term negaƟve consequences for current residents. 2) ConstrucƟon in this project’s locaƟon would also risk irreversible damage to area watershed, sensiƟve ecological preserves, including Mission Trails, Sycamore Canyon, & other nearby natural resources & wildlife habitats.

June 7, 2025 Page 2 of 3
3) Most importantly, 100% of the proposed development area is designated by CalFire as a “Very High Fire Hazard Severity Zone”. (Please See Page 3)

3.1. I can vividly recall watching the flames of the 2003 Cedar Fire traverse this area’s hillsides from Hwy 67 all the way into Kearny Mesa, and now Home Fed & Santee want to put 3000 homes, etc. into this very same High Fire-risk pathway?!?

3.2. We all saw the horrific destrucƟon of the Palisades & Eaton Fires in Los Angeles. Over 16,000 Structures burned. The Fanita Ranch development would simply be Ɵnder for a repeat of this disastrous Fire.

3.3. We have seen the consequences of poor wildfire planning in other parts of California — Santee must not follow that path.

Finally, & above all else . . .
4) The Fanita Ranch development has been repeatedly voted down by the ciƟzens of Santee, as well as ruled against by the Court.

4.1. This revised proposal conƟnues to fail to address the very real limitaƟons of Santee’s Infrastructure, and the “Very High” Fire Hazards, & Environmental risks involved.

4.2. To approve this project would demonstrate an uƩer lack of respect & consideraƟon for the ciƟzens of Santee, for whom you are sworn to represent.

4.3. The City of Santee must prioriƟze the well-being, safety, & stability of its current ciƟzens over the greed & ambiƟons of large-scale developers. Santee should not bear the burden of Home Fed’s bad investment.

Development does NOT inherently equal progress. True progress lies in improving the quality of life for the residents who already call Santee home.

I respecƞully urge your office & the City Council of Santee to do what is best for the ciƟzens of Santee, and reject the Fanita Ranch Housing Development.

Thank you for your Ɵme and consideraƟon.

Sincerely,
Carl & Mary CostanƟno,
Santee Residents for 45+ years
June 7, 2025 Page 3 of 3
From:

To: Sandi Sawa
Subject: Comments IRT Fanita Ranch EIR
Date: Wednesday, June 11, 2025 1:48:09 PM
Greetings,
I urge the Santee City Council to vote against approval of the Fanita Ranch Housing development.

Approving residential development that is located solely within a Very High Fire Hazard Severity Zone with very limited emergency egress is irresponsible and reckless. It would put the new residents, and the residents of existing surrounding neighborhoods at increased risk in event that large scale evacuations become necessary to escape a fast approaching fire storm.

The fire evacuation plan, dated 2022, included in the latest circulated EIR appears to assume evacuation times based on best case scenarios. It does not adequately consider that the only available routes must pass through existing established neighborhood on small side streets that are not designed to handle large volumes of vehicle traffic, such as would be encountered during a mass evacuation during a fast approaching fire storm. Such a firestorm occurred during the Cedar fire in 2003, the axis of which burned through the proposed Fanita Ranch development site at a very fast rate of spread. The fire EIR does not adequately account for traffic complications during a large scale evacuation, such as a disabled vehicle incident at one of the choke points which would severely obstruct evacuating traffic flow. Blocked and impassable streets during the recent Pacific Palisades fire illustrates how such an incident is a very real risk.

The City Councils continuous attempts to circumvent court rulings and block a citizen vote on the development give the appearance that the City is working exclusively on behalf of the developer, not for the citizens who elected them, and reflects negatively on the City.

Very respectfully,
David Kramer
Santee resident

From: GLORIA GERAK
To: Sandi Sawa; [email protected]
Subject: Fanita Ranch Final EIR
Date: Tuesday, June 10, 2025 1:03:24 PM
S. Sawa
We fell that your response to the citizen comments in the final EIR is not adequate. We are concerned about fire mitigation as an example: the Los Angeles Fires. Should there be a fire - there is no way the Citizens can exit safely due to the traffic on Mast Blvd. Additionally, not enough attention is being given to Air Quality and Noise. As a person who lives along Mast Blvd, the sound wall were built to mitigate the increased noise and the DB’s were to be lower than 60 DB during waking hours. We had meters in our yard, placed by the City and the sound was never at 60 DB - always higher. In other words, the sound walls failed from their initial installment. How will you mitigate this additional traffic noise? ENVIRONMENTAL STATUS: A Final Recirculated Revised EIR (State Clearinghouse Number SCH#2005061118), including Second Recirculated Sections of the Final Revised EIR will be presented to the City Council for certification in compliance with the California Environmental Quality Act (CEQA). Areas of significant and unmitigable impact that require a Statement of Overriding Considerations include Air Quality, Noise, Recreation, Transportation, and Utilities and Service Systems. The purpose of this notice is to give property owners in the vicinity of the subject property and other interested parties an opportunity to be informed of the proposal prior to action by the City Council. The time within which judicial review of a City Council decision must be sought is governed by Section 1094.6 of the California Code of Civil Procedure. A right to appeal a City Council decision is governed by the Code of Civil Procedure Section 1094.5. At the subject hearing any interested party may appear and be heard. The meeting will be broadcast live on public access channels 117 on Cox Cable and 99 on AT&T/U-verse, live streamed online, and recorded.

Respectfully:

Gloria Valenti Gerak
Ronald Gerak
Alexander Gerak
Santee, CA 92071
Office
Mobile
Please consider the environment before printing this e-mail CONFIDENTIALITY NOTICE, PLEASE READ: This e-mail message, including any attachments, is for the sole viewing and use of the intended recipient(s) and may contain confidential and privileged information within. Any unauthorized review, use, disclosure or distribution is prohibited.

From: CPB
Sent: Monday, June 9, 2025 8:08:56 PM
To: Sandi Sawa <[email protected]>
Subject: Fanita Ranch
> Dear Ms Sawa,
> I am against building in the Area under consideration for several reasons . Air Quality, > Noise, Our famous Santee Lakes Recreation area will also be affected a place our > Citizens enjoy would be destroyed by thousands of cars each day roaring bye and not to mention the danger of fires . Your own fire expert advised that the city and Home Fed hired said it would take me 2 hours to get to Mission Gorge Rd in event of a > Fire. I have lived in Santee for 41 years and the disregard of this City Representatives > Is alarming and they ignore our concerns for our safety . This is a working class area we drive to work and the thought of thousands of more cars trying to drive to the 52 not to > Mention getting to West Hills High School and Sycamore School is not safe for us > And our children . The proposed routes Fanita and Cuyamaca and Mast and your map > Attachment shows tentative - - - - marks for Magnolia is this because the > Representive for that area doesn’t want Magnolia involved ? >
> Sincerely
> Janis Barnhart
>
>
>
>
> Sent from my iPad
From: Yvonne Young
To: Sandi Sawa
Subject: Fanita Ranch Project
Date: Wednesday, June 11, 2025 8:42:40 AM
Good Afternoon,
Pushing through the Fanita Ranch project when the voters have already voted against it is ridiculous.

You are not listening to the parties that elected you. You have already wasted a lot of money going against your voters. I think this is irresponsible and is going to end up costing all of Santee’s residents more money for unnecessary lawsuit’s.

Calling this an Essential Housing Project is just ridiculous. It’s just another way to get around the will of the people. If it was really essential housing every house built would lie in the median income of all the residents of Santee. But this is not about essential housing it’s all about money!!!!!!!!!!!!! No wonder why people no longer have any belief in Government. They vote against something and the politicians disagree with the voters so they find some way to get around the voters, doesn’t sound like democracy to me.

Thanks
Jeff Young
From: Marie Weber
Sent: Friday, June 6, 2025 4:46:54 PM
To: Sandi Sawa <[email protected]>
Subject: Fanita Ranch
Sandi Sawa,
I have a question about the Fanita Ranch project which I would like included with the challenge to approve the development. I believe the judge ordered Santee city council to remove their past approval of this project. When I look at the Active Projects map, however, it is still listed as approved. https://www.cityofsanteeca.gov/documents/planning-building/active-projects/activeprojects-map.pdf So, did council ever rescind their approval? It's clearly listed as "Approved - Not Built". The map is not old, as I see projects listed from the current year (2025). I also question how this project fits Califormia's definition of Essential Housing. It seems as though Santee City Council has created it's own definition to avoid a vote of the citizens due to the overwhelming majority against Fanita Ranch.

Sincerely,
Marie Weber
From: Mary Fujimoto
To: Sandi Sawa
Cc: Dustin Trotter; John Minto
Subject: Fanita Ranch
Date: Wednesday, June 11, 2025 8:07:21 AM
Dear City of Santee,
It is rather frustrating that Fanita Ranch continues to appear as a done deal in the City of Santee’s eyes.

This project has fire risks to those residents that should buy there and traffic in the community will not be mitigated.

I have not been a fan of Fanita Ranch because of those hazardous (in the case of fire - look at Paradise- look at LA (Palisades and Alta Dena) issues and traffic but could learn to accept this project if the project came to a vote to residents as was supposed to happen as a result of Prop N. However, the City has done all it could do to circumvent and obstruct a vote of residents. Frankly, myself and others should not have to continually correspond on this project because our elected officials do not want to put the project on a ballot for residents to vote on.

Please just put the Fanita Ranch project up to a vote and let the chips fall where they will. At least it would be a decision of the residents.

Thank you,
Mary Fujimoto
From: Maureen Wallace
To: Sandi Sawa
Subject: Fanita Ranch
Date: Tuesday, June 10, 2025 2:35:00 PM
Sandi Sawa,
My name is Maureen Wallace and I live at the end of the current Fanita Ranch on Cadwell Road. I am opposed to any further development of the Fanita Ranch area.

First, the citizens of Santee continue to vote down the expansion of Fanita Ranch. I do not understand how the mayor and our city council can continue to support Fanita Ranch expansion. Aren't they supposed to support the opinions of their citizens? Having lived here for 39 years, every time Fanita Ranch is voted down, somehow the council and Mayor manage to bring it back up. I thought once you voted something down, that ended the discussion on it.

Second, homeowners living along the hillsides in Fanita Ranch have seen their homeowners insurance continually rise. Building more homes in this area is only going to make it harder for new homeowners to afford the insurance. In the last 2 years I have stood in my yard to see smoke coming from 3 fires that were "accidentally " started from Miramar. Always during warm weather and one fire we could see the fire burning up the hillside.

I am unable to attend the meeting tomorrow night but I hope my views will be shared. I know I am one of many who share these same feelings.

Sincerely,
Maureen Walllace

From: Tina Deesen
Sent: Saturday, June 7, 2025 5:42 PM
To: Sandi Sawa <[email protected]>
Subject: Fanita Ranch Final 2nd Recirculated Sections FREIR Dear Director Sawa and City Council,
Please respect the Democratic process and allow the citizens of Santee to make a final vote on the project.

Even though the proposed new homes are built with fire protections in mind, the older homes like mine are not. A fire in that area puts my home, which is very close to the proposed Fanita Ranch community, at huge risk. It's only right you take that into consideration as I am a current voting Santee citizen.

We deserve our vote to be heard.

Sincerely,
Tina Deesen
Yahoo Mail: Search, Organize, Conquer
From: SS DD
To: James Jeffries
Subject: Submitted Public Comments
Date: Tuesday, June 10, 2025 5:24:36 PM
Hello again. As I may not be able to attend tomorrow's meeting in person, I am requesting that the following comments are submitted both into the record, as well as distributed to each Council Member under the name "Truth", thank you.

Each comment is following each item number. There are a total of 4 separate comments (items: #2, #6, #8, #9).

If you have any questions, feel free to email me back.

Item 6: (Fanita)
Why did “…the Final Recirculated REIR [fail] to adequately Disclose the proposed project’s Inconsistency with the Santee General Plan…”? What lead to issues to begin with: “…On August 25, 2021, the City of Santee adopted Urgency Ordinance No. 592, declaring the need for an Essential Housing Program…”
What makes a Project “a Certified Essential Housing Project” versus any Other housing? Is it just because it contains a Ton of Units? The item says: “The Density Bonus Law is to be ‘interpreted liberally’…” – That’s Questionable. Everything must be Specified in Law. Does Density Bonus apply to so-called Sprawl developments like this? What about to Major developments that have 3 sub-developments within like this one? For Air Quality, it says the Fanita Ranch project:

It Conflicts with applicable Air Quality Plans… – What plans? And there will be a Significant Increase in traffic at intersections located Outside Santee’s jurisdiction, or no funding mechanism is currently available, or no feasible mitigation is available… – Which is it? And there will be an Increase in Vehicle Miles Traveled Above the calculated threshold… – Will that come back to SANDAG and their Regional Plan to Reduce VMT?”
“Under State CEQA Guidelines…, a public agency may Approve a project even though the project would cause a Significant effect on the Environment if the agency makes a fully informed and publicly disclosed decision that there is no feasible way to lessen or avoid the significant effect…, and Specifically Identified expected benefits from the project Outweigh…Reducing or Avoiding significant environmental impacts of the project…
The proposed Resolution…[finds] that economic, legal, social, technological or other benefits, including Region-Wide benefits, of the project Outweigh the Unavoidable adverse environmental effects, rendering adverse environmental effects ‘Acceptable.’” – Who is Region-Wide? – SANDAG or the State? Care to Specifically Identify the expected Benefits that Outweigh the Unavoidable adverse effects, as Required? Is the benefit the $105 million in Development Impact Fees? Is it to meet the State’s RHNA Housing Quota numbers? What’s the point of CEQA if Environmental Protection can be Waived?
This project will have businesses, a school (or 59 units), parks, farm, orchards and vineyards – How many?
Also a fire station – Who would build it? Who would own it? Also a potential solar farm – Why can’t that be a community farm or dog park for all of these future people?
Question: How do the Developers Not know if they need a school or not? Taking almost 30% of the land of the City – It’s kind of obvious they need a school and a Whole City to meet their needs. Has anyone considered the adverse effect of adding an entire village will include Changing the Character of the neighborhood?
How does this project fit in with CalFire's new Fire Maps? Is Fire Safety another adverse effect the State is willing to Waive?

June 11, 2025
Public Comment Fanita Ranch Development
To the Santee City Councilmembers and Mayor:

I write to express my deep concern with the revival of the Fanita Ranch Development project and urge the council to vote no for three distinct reasons:

1. Precedent setting development that disregards statutory requirements for citizen approval of new zoning through Measure N. Through the passage of this Measure, citizens must be allowed the right to vote on approval of projects if they are outside of the constraints of the General Plan. Increased capacity of this development, while baked in saving housing costs and providing affordable living, do not change the fact that the zoning should have been a consultation process. I’m concerned that habitat conserved lands seem to be a priority, but the fragmentation of this area into smaller pieces, with now more extensive trail system, and the accommodating vehicular and foot traffic, lead to serious concerns about the development and the way it should have been approved.

2. Serious habitat fragmentation that damages the livelihood and little remaining habitat for dense sage scrub animals like the Bell’s Sparrow. Sycamore Canyon and the surrounding protected lands around this new development are some of the last deep coastal sage scrub habitats in the County. Numerous species, including those protected under law like the California Gnatcatcher, will be harmed by the damage to this ecosystem. Bell’s Sparrow, a regional endemic also relies on these deep sage scrub habitats and with more encroachment, species like these two which DO NOT adapt well to manmade infrastructure, will be left with increasingly smaller habitats that will be even more susceptible to fire damage. Fragmentation from this development only speeds up the rate at which these species will struggle to survive in a changing world.

3. Hazardous construction into dangerous high fire risk zones. With this development, the City of Santee sets a precedent that it is willing to build housing into areas with extreme fire risk, knowing that families likely will not be able to insure and will suffer the consequences of fire damage wholly. San Diego is expected at minimum to experience equal or greater fire damage in the coming decades, and given the severity of fires in Los Angeles County, it’s possible San Diego experiences something that large too. Building into this habitat, which is at extreme fire risk, regardless of the preparatory steps, is an unwise choice for the safety and security of San Diegans that will move to the area. With increased fire mitigation strategies, more land will be damaged that vulnerable species rely on.

Please consider voting no on this motion, due to the dangerous consequences of this development.

Sincerely,
Zach Billot
San Diego Bird Alliance
From: Meredith Stevenson <[email protected]> Sent: Tuesday, June 10, 2025 4:59 PM
To: Sandi Sawa <[email protected]>
Subject: Comment Letter on FREIR for Fanita Ranch
Your attachment(s) were cleaned by Check Point Sandblast Threat Extraction.

Click here to restore the original(s) or contact your system administrator.

You may be required to authenticate, in that case follow these instructions:

1. You will be directed to a page where you would be requested to specify your email address.

2. An email with verification code will be sent to you.

3. Copy the code and return to the attachment recovery page.

4. The email with original attachments will be released to your mailbox.

Please exercise discretion when requesting to release suspicious attachments.

Good afternoon,
Attached please find a letter from the Center for Biological Diversity regarding the FREIR for Fanita Ranch. We will also send the letter via FedEx. The references are available here and on a flash drive we will send tomorrow. Please confirm that you received the letter and were able to download the references.

Thank you,
Meredith Stevenson
Staff Attorney
Center for Biological Diversity
574-309-5620
[email protected]
June 10, 2025
Sent via email, with references via FTP and FedEx
Sandi Sawa, AICP Director of Planning & Building
Department of Development Services
City Hall, Building 4
10601 Magnolia Avenue
Santee, California 92071
Telephone: (619) 258-4100, extension 167
Email: [email protected]
Re: Second Recirculated Sections of the Final Revised Environmental Impact Report for Fanita Ranch, SCH# 2005061118
Dear Ms. Sawa:

This letter is submitted on behalf of the Center for Biological Diversity (the “Center”) regarding the Second Recirculated Final Revised Environmental Impact Report (“2025 FREIR”) for Fanita Ranch (the “Project”) (State Clearinghouse No. 2005061118).

This FREIR is the latest environmental document prepared pursuant to the California Environmental Quality Act (“CEQA”), intended to address the deficiencies the San Diego County Superior Court noted in the 2022 First Recirculated REIR for the Project (“2022 REIR”).

The Center is a non-profit, public interest environmental organization dedicated to the protection of native species and their habitats through science, policy, and environmental law. The Center has over 1.7 million members and online activists throughout California and the United States. The Center has worked for many years to protect imperiled plants and wildlife, open space, air and water quality, and overall quality of life for people in Santee and throughout San Diego County. 2 I. The 2025 FREIR Still Lacks an Adequate Analysis of the Project’s Impacts Relating to Wildfire and Emergency Evacuation.

CalFire’s updated 2025 maps and the FREIR confirm that a Very High FHSZ now covers the entire Project site (Cal Fire, 2025), yet the FREIR fails to adequately assess the Project’s wildfire ignition and public safety risks and adopt feasible mitigation measures to reduce or avoid the Project’s fire safety impacts. Among other things, the City failed to fully consider new information from the recent Eaton and Palisades Fires (LA Fires), which indicates an increased likelihood that the Project will increase ignition risk and pose safety risks to residents.

As stated in the Center’s prior comments, the Project site has historically been extremely susceptible to wildfire. As the 2022 REIR acknowledges, “wildfire has occurred and would likely occur in the Project vicinity again.” (2022 REIR, EIR Appendix P-1.) The Project fire map further indicates that virtually the entire Project site has been burned at least once, with the vast majority having burned numerous times. (Id., Figure 5.) Additionally, the High FHSZ landscapes surrounding the Project area further increase the site’s wildfire risk. (Cal Fire, 2025.)
Given the extremely high risk of wildfire in the area, and the past history of repeated burnings at the Project site, it remains critical that the City prepare a revised EIR that adequately discloses and analyzes the Project’s wildfire impacts in light of the recent LA Fires under similar conditions, as well as difficulties in evacuation during those fires.

A. The FREIR Fails to Acknowledge or Adequately Analyze the Increase in Fire Risk Resulting from the Project in Light of the Recent LA Fires.

The FREIR remains deficient because it fails to acknowledge or adequately analyze conditions on the Project site similar to the conditions that contributed to the 2025 LA Fires. CEQA requires a revised EIR or subsequent EIR when significant new information indicates (1) “[a] new significant environmental impact would result from the project”; or (2) “[a] substantial increase in the severity of an environmental impact would result unless mitigation measures are adopted that reduce the impact to a level of insignificance.” (Pub. Res. Code § 21166; CEQA Guidelines § 15088.5(a); CEQA Guidelines, § 15162(a)(2), (a)(3)(A), (B).) But here, the FREIR dismisses pertinent new information from the LA Fires because the Los Angeles wildfires occurred 150 miles from the City with “different fuel types, weather, and terrain.” (2025 FREIR at 1-9.) 3 This conclusion is unsupported. As explained in prior comments, the Project site has similar fire conditions to those that gave rise to the 2025 Eaton and Palisades Fires.

For example, chaparral, one of the most widespread vegetation communities on the Project site, fueled both the Eaton and Palisades fires and poses a similar threat on the Project site. (McKenzie, 2025; 2020 DREIR at 4.3-6.) Specifically, chamise is prominent in both the Santa Monica Mountains where the Palisades Fire began and on the Project site. (2020 DREIR at 4.3-6; Bland, 2025.) Additionally, similar weather poses a risk, as the Santa Ana winds that fueled the Los Angeles Fires threaten to fuel fires at the Project site. (2022 REIR at 4.18-15; Laskowski, 2025.) And terrain also bears similarities: the Project site encompasses “steep topography that can facilitate fire spread” (2022 REIR at

4.18-15), while the steep slopes, canyons and valleys along the Santa Monica and San Gabriel Mountains added to dangerous fire conditions and contributed to the LA Fires’ severity. (Top 20, 2025; Graff, 2025.)
B. The FREIR Improperly Relies on Project Features to Avoid Assessing Relevant New Information.

Instead of analyzing the significance of these similar conditions, the FREIR relies on project features to keep wildfire ignition risks below the significance threshold.

Specifically, the FREIR emphasizes that the Project is “a new, master-planned, and ignition-resistant community” that utilizes project features such as “fuel modification zones, fire-hardened structures, code-compliant access, fire flow and hydrants, automatic interior fire sprinklers, flying ember protections, and ongoing maintenance, among others.” (2025 FREIR at 1-9.) The FREIR explains that the Project’s features render the Project so different than the LA Fire communities that the FREIR need not consider new information from the LA Fires. (2025 FREIR at 1-9.) This is incorrect for several reasons.

First, longstanding case law provides that substituting mitigating design features for impact analysis violates CEQA. (People ex rel. Bonta v. County of Lake (2024) 105 Cal.App.5th 1222, 1234 [county could not avoid analyzing wildfire risk and effectiveness of measures to reduce wildfire risk reduction by deeming them “design features”]; Lotus v. Department of Transportation (2014) 223 Cal.App.4th 645, 658 [“The failure of the EIR to separately identify and analyze the significance of the impacts . . . before proposing mitigation measures . . . precludes both identification of potential environmental consequences arising from the project and also thoughtful analysis of the sufficiency of measures to mitigate”]; San Joaquin Raptor Rescue Center v. County of Merced (2007) 149 Cal.App.4th 645, 663 [“A mitigation measure cannot be used as a 4 device to avoid disclosing project impacts”].) This makes sense, considering an EIR must first inform the public and decisionmakers of a project’s significant impacts, and then propose mitigation measures and project features to eliminate them, supported by substantial evidence. In other words, relying on project features to mitigate potential impacts of fast-moving fires like the recent LA Fires does not eliminate the requirement to first analyze the significance of the potential impacts.

The FREIR touts the Orchard Hills development, in Irvine, CA as an example of how Fanita Ranch’s project features will perform in a wildfire. (2025 FREIR at 1-9.) However, unlike this Project, Orchard Hills includes nearly 200-foot fuel modification zones, fire walls to protect the development from embers, and perimeter roads in consideration of Santa Ana wind direction. (Flemming, 2025.) As a result, the FREIR fails to provide adequate support for its conclusion that this Project will perform the same as Orchard Hills.

Second, even if effective, the project features do not address other reasonably foreseeable impacts of a fire similar to the recent LA Fires on the Project site. Fires, especially the hotter and longer-burning variety that have overtaken California in recent decades, can prove disastrous for plant and animal life. If native habitat fire regimes are disrupted, the habitats they provide can become degraded. (Keeley 2005; Keeley 2006.) When fires occur too frequently, type conversion occurs, and native shrublands are replaced by non-native grasses and forbs that burn more frequently and more easily, ultimately eliminating native habitats and biodiversity while increasing fire threat over time (Keeley 2005; Keeley 2006; Syphard et al. 2009.) Wildfires can have a long-lasting negative effect on habitat, and can impair animals’ movement (Jennings 2018), mating ability, foraging, and reproductive success. (See Syphard et al. 2007 [“With more fires occurring in close proximity to human infrastructure, there may also be devastating ecological impacts if development continues to grow farther into wildland vegetation.”].) This could have serious consequences for special-status species in the project area such as mountain lions (Blakey et al., 2022; Jennings, 2018), whose populations are already struggling in much of the state due to lack of connectivity and genetic isolation. (Benson et al., 2019; Gustafson et al., 2021.) As a result, the City must take into account new information on the LA Fires’ impacts on biological resources and mitigate accordingly.

Third, the project features do not address the evacuation concerns the Center raised in prior comments. The FREIR claims that new information on evacuation difficulties from the LA Fires is irrelevant due to several project features that were not 5 present in all LA Fire evacuation zones: automatic sprinkler systems (Appendix U at 3), on-site sheltering, primary and secondary access points (Appendix U at 2), and less vulnerable structures. (Appendix U at 4.) However, while potentially helpful, these project features do not consider or address older residents’ unwillingness to evacuate the LA Fires, their difficulties due to physical constraints, logistical challenges for older residents, immediate health risks to older evacuees, and new information regarding health risks of wildfire smoke. (See Center DREIR Comments at 7-11.) Instead, to address these concerns, the FREIR offers only reassurances that “for Fanita Ranch, there could be evacuation, or temporary relocation into other parts of the community or into Santee with a short duration.” (Appendix U at 4 [emphasis added].) Appendix U also suggests that “transportation or other special requirements can be provided during an emergency evacuation”; however, this is not required. (Appendix P2 at 21 [stating that “[t]ransportation should be accessible to all populations, including people with disabilities and other access and functional needs” not that it must].) The FREIR also fails to address the Center’s concerns regarding new information on impacts from smoke inhalation and particularly its impact on older residents.

Appendix U states only: “Regarding smoke-related health issues, wildfires occurring in and around Fanita Ranch would be short duration as the fuels burned and the fire was driven around the project, further downwind.” (Appendix U at 4.) However, even short periods of smoke inhalation can impact residents’ health and is associated with increased risk of exacerbation of pre-existing respiratory and cardiovascular disease, as well as premature mortality. (Melton, 2023.) A revised EIR must take into account new information on smoke inhalation and mitigate accordingly.

The FREIR also fails to adequately address the Center’s concerns regarding evacuation alerts. Instead, Appendix U incorrectly states that Emergency Alerts of evacuation areas in Santee requires a “layered approach,” including wireless emergency alerts (text/email), Alert San Diego (Reverse 911), Television, Radio, social media, and in the field law enforcement and fire personnel. (Appendix U at 3.) As noted in prior comments, the Project’s Wildland Fire Evacuation Plan relies only on text message/email alerts and does not require other methods that likely saved many residents during the Los Angeles fires such as police cruisers and helicopters with blowhorns, door-to-door notifications, television notifications, and changeable message signs. (2022 REIR, Appendix P2 at 9.) Considering new information from the LA Fires, electronic notifications are not sufficient: not only are they prone to error, but they also may not be suited to the Project’s older population, which is less likely to receive disaster warnings 6 (Melton, 2023) and may lack access to critical internet communications to monitor fire conditions, preparedness, and evacuation notices. (Center DREIR Comments at 7-8; Melton, 2023; Lin, 2025; Lee et al., 2021; Courtin & Knapp, 2017.) In short, a project built in a location known to have very high or high wildfire risk cannot compensate for wildfire hazards simply through a fire-resistant design. Because the FREIR fails to acknowledge significant wildfire impacts based on new information, the FREIR also fails to mitigate them to reduce these impacts. The public—including future residents of the Project and existing residents nearby—have a right to know the full extent of the Project’s impacts on wildfire ignition and evacuation. “Omission of material necessary to informed decision-making and informed public participation is prejudicial.” (Sierra Club v. County of Fresno, (2018) 6 Cal.5th 502, 515.) II. Conclusion
Thank you for the opportunity to submit comments on the 2025 FREIR for Fanita Ranch. We urge the City to prepare a new EIR for the Project that fully complies with CEQA and recirculate the revised EIR. Because significant new information has become available, the City must reevaluate and incorporate this information, as well as new research and studies on the Project’s impacts that have become available in the last few years.

Given the possibility that the Center will be required to pursue legal remedies in order to ensure that the City complies with its legal obligations including those arising under CEQA, we would like to remind the City of its statutory duty to maintain and preserve all documents and communications that may constitute part of the “administrative record” of this proceeding. (§ 21167.6(e); Golden Door Properties, LLC v. Superior Court (2020) 53 Cal.App.5th 733, 762.) The administrative record encompasses any and all documents and communications that relate to any and all actions taken by the City with respect to the Project, and includes “pretty much everything that ever came near a proposed [project] or [] the agency’s compliance with CEQA . . . .” (County of Orange v. Superior Court (2003) 113 Cal.App.4th 1, 8.) The administrative record further includes all correspondence, emails, and text messages sent to or received by the City’s representatives or employees, that relate to the Project, including any correspondence, emails, and text messages sent between the City’s representatives or employees and the Applicant’s representatives or employees. Maintenance and preservation of the administrative record requires that, inter alia, the City (1) suspend all 7 data destruction policies; and (2) preserve all relevant hardware unless an exact replica of each file is made.

Please add the Center to your notice list for all future updates to the Project and do not hesitate to contact the Center with any questions at the email listed below.

Sincerely,
Meredith Stevenson
Staff Attorney
2100 Franklin Street, Suite 375
Oakland, CA 94612
[email protected]
8 REFERENCES
Benson, J. F., Mahoney, P. J., Vickers, T. W., Sikich, J. A., Beier, P., Riley, S. P. D., Ernest, H. B., & Boyce, W. M. (2019). Extinction vortex dynamics of top predators isolated by urbanization. Ecological Applications, 29(3), e01868– e01868.
https://digitalcommons.unl.edu/cgi/viewcontent.cgi?article=2039&context=natres papers
Blakey, R. V., Sikich, J. A., Blumstein, D. T., & Riley, S. P. D. (2022). Mountain lions avoid burned areas and increase risky behavior after wildfire in a fragmented urban landscape. Current Biology, 32(21), 4762-4768.e5.

Bland, A. (2025, Jan., 15). ‘Literally off the charts’: LA’s critically dry conditions stun scientists as fires rage. CalMatters. Available at:
https://calmatters.org/environment/wildfires/2025/01/dry-danger-zone-californiafires-climate-change/ Cal Fire. (2025). Top 20 Most Destructive California Wildfires. Available at:
https://34c031f8-c9fd-4018-8c5a-4159cdff6b0d-cdn-endpoint.azureedge.net/- /media/calfire-website/our-impact/firestatistics/top20_destruction.pdf?rev=057c3d89da86403290fcbcef630fc692&hash= E7D05B4DCE9C3C0CE857221695C54FEF
Cal Fire. (2025). Updated Fire Hazard Safety Zones. Available at:
https://osfm.fire.ca.gov/what-we-do/community-wildfire-preparedness-andmitigation/fire-hazard-severity-zones Courtin, E., & Knapp, M. (2017). Social isolation, loneliness and health in old age: A scoping review. Health & Social Care in the Community, 25(3), 799–812.

Flemming, J. (2025, Feb. 19). Could this Irvine neighborhood be the blueprint for a more fire-resistant L.A.? LA Times. Available at:
https://www.latimes.com/california/story/2025-02-19/could-this-irvineneighborhood-be-the-blueprint-for-a-more-fire-resistant-l-a Graff, A. et al. (2025, Jan. 11). Dry Vegetation Fuels L.A. Fires as Wind Speeds Drop.

New York Times. Available at:
https://www.nytimes.com/2025/01/11/us/mandeville-brentwood-fires-vegetationwinds.html 9 Gustafson, K. D., Gagne, R. B., Buchalski, M. R., Vickers, T. W., Riley, S. P. D., Sikich, J. A., Rudd, J. L., Dellinger, J. A., LaCava, M. E. F., & Ernest, H. B. (2021).

Multi‐population puma connectivity could restore genomic diversity to at‐risk coastal populations in California. Evolutionary Applications.

Jennings, M. (2018). Effects of Wildfire on Wildlife and Connectivity.

Keeley, J. E. (2005). Fire as a threat to biodiversity in fire-type shrublands. In Planning for biodiversity: Bringing research and management together. USDA Forest Service General Technical Report PSW-GTR-195.

Keeley, J. E. (2006). Fire management impacts on invasive plants in the western United States. Conservation Biology, 20(2), 375–384. https://doi.org/10.1111/j.1523- 1739.2006.00339.x
Lee, H. Y., Kanthawala, S., Choi, E. Y., & Kim, Y. S. (2021). Rural and non-rural digital divide persists in older adults: Internet access, usage, and attitudes toward technology. Gerontechnology, 20(2).

Lin, S. (2025, Feb.) California wildfires are burning deeper into urban areas like Altadena and finding new victims. LA Times. Available at:
https://www.latimes.com/california/story/2025-02-25/california-wildfiresvulnerable-victims McKenzie, B. (2025, Jan. 14). Weather, Wind and Flammable Flora Fuel California Fires.

Available at: https://news.virginia.edu/content/weather-wind-and-flammable-florafuel-californiafires#:~:text=%E2%80%9CThe%20native%20vegetation%20is%20chaparral,very %20volatile%20and%20burns%20hot.%E2%80%9D
Melton, C. C., De Fries, C. M., Smith, R. M., & Mason, L. R. (2023). Wildfires and older adults: A scoping review of impacts, risks, and interventions. International Journal of Environmental Research and Public Health, 20(13), 6252.

Syphard, A. D., Radeloff, V. C., Keeley, J. E., Hawbaker, T. J., Clayton, M. K., Stewart, S. I., Hammer, R. B., Syphard, A. D., Radeloff, V. C., Keeley, J. E., Hawbaker, T.

J., Stewart, S. I., & Hammer, R. B. (2007). Human influence on California fire regimes. Ecological Society of America, 17(5), 1388–1402.

Syphard, A. D., Radeloff, V. C., Hawbaker, T. J., & Stewart, S. I. (2009). Conservation threats due to human-caused increases in fire frequency in mediterranean-climate ecosystems. Conservation Biology, 23(3), 758–769.
10 Waddick, K. (2025, Jan. 8). 3 reasons California's wildfires got so dangerous so fast. USA Today. Available at:
https://www.usatoday.com/story/news/nation/2025/01/08/geography-los-angeleswildfires/77550014007/ Extinction vortex dynamics of top predators isolated by urbanization JOHN F. BENSON,
1,2,8 PETER J. MAHONEY,3 T. WINSTON VICKERS,4 JEFFA. SIKICH,5 PAUL BEIER,6 SETH P. D. RILEY,2,5 HOLLY B. ERNEST,
7 AND WALTER M. BOYCE4
1 School of Natural Resources, University of Nebraska, Lincoln, Nebraska 68583 USA 2 La Kretz Center for California Conservation Science, Institute of the Environment and Sustainability, University of California, Los Angeles, California 90095 USA 3
School of Environmental and Forest Sciences, University of Washington, Seattle, Washington 98195 USA 4 Karen C. Drayer Wildlife Health Center, School of Veterinary Medicine, University of California, Davis, California 95616 USA 5 National Park Service, Santa Monica Mountains National Recreation Area, Thousand Oaks, California 91360 USA 6 School of Forestry, Northern Arizona University, Flagstaff, Arizona 86011 USA 7 Department of Veterinary Sciences, University of Wyoming, Laramie, Wyoming 82071 USA Citation: Benson, J. F., P. J. Mahoney, T. W. Vickers, J. A. Sikich, P. Beier, S. P. D. Riley, H. B. Ernest, and W. M. Boyce. 2019. Extinction vortex dynamics of top predators isolated by urbanization. Ecological Applications 00(00):e01868. 10.1002/eap.1868 Abstract. Extinction risk is elevated in small, isolated populations due to demographic and genetic interactions. Therefore, it is critical to model these processes realistically in population viability analyses (PVA) to inform local management and contribute to a greater understanding of mechanisms within the extinction vortex. We conducted PVA’s for two small mountain lion populations isolated by urbanization in southern California to predict population growth, extinction probability, and loss of genetic diversity with empirical data. Specifically, we (1) provide the first PVA for isolated mountain lions in the Santa Ana Mountains (SAM) that considers both demographic and genetic risk factors and (2) test the hypothesis that variation in abundance and mortality between the SAM and Santa Monica Mountains (SMM) result in differences in population growth, loss of heterozygosity, and extinction probability. Our models predicted 16–21% probability of local extinction in the SAM due purely to demographic processes over 50 yr with current low levels or no immigration. Our models also predicted that genetic diversity will further erode in the SAM such that concern regarding inbreeding depression is warranted unless gene flow is increased, and that if inbreeding depression occurs, rapid local extinction will be highly likely. Dynamics of the two populations were broadly similar, but they also exhibited differences driven by larger population size and higher mortality in the SAM. Density-independent scenarios predicted a rapidly increasing population in the SMM, whereas growth potential did not differ from a stable trend in the SAM.

Demographic extinction probability and loss of heterozygosity were greater in the SMM for density-dependent scenarios without immigration. However, higher levels of immigration had stronger, positive influences on both demographic viability and retention of genetic diversity in the SMM driven by lower abundance and higher adult survival. Our results elucidate demographic and genetic threats to small populations within the extinction vortex, and how these vary relative to demographic structure. Importantly, simulating seemingly attainable increases in connectivity was sufficient to greatly reduce extinction probability. Our work highlights that conservation of large carnivores is achievable within urbanized landscapes, but requires land protection, connectivity, and strategies to promote coexistence with humans.

Key words: demographic stochasticity; extinction; heterozygosity; inbreeding; mortality; population viability analysis; Puma concolor; urbanization.

INTRODUCTION
Demographic and genetic processes, and interactions
between them, influence probability of extinction for
small, isolated populations (Saccheri et al. 1998,
O’Grady et al. 2006). Specifically, deterministic stressors, demographic and environmental stochasticity, and inbreeding depression can all contribute to increased
extinction probability (Mills and Smouse 1994, Beissinger et al. 2008). However, the relative influence of these processes in different wildlife populations remains difficult to predict and empirical demonstrations are
rare (Palomares et al. 2012, Wootton and Pfister 2013).

The predicted decline to extinction of small populations from these interacting processes is referred to as the
extinction vortex (Gilpin and Soul
e 1986). Modeling
dynamics of small, isolated populations provides critical information to local conservation efforts and also
Manuscript received 15 June 2018; revised 30 October 2018; accepted 3 January 2019. Corresponding Editor: Aaron
Wirsing. 8E-mail: [email protected]
Article e01868; page 1
Ecological Applications, 0(0), 2019, e01868
© 2019 by the Ecological Society of America
contributes to a more general understanding of the
forces influencing extinction.

All populations with small numbers of breeding individuals are likely to be destabilized by demographic stochasticity (Lande 1993, Morris and Doak 2002) and
are also the most likely to suffer from inbreeding depression (Mills and Smouse 1994). However, variation in local environmental conditions, and resulting differences in demographic structure, can influence population
growth, the rate at which genetic diversity is lost, and extinction probability (Stacey and Taper 1992, Reed
2005). Prior to the onset of inbreeding depression, some small populations continue to exhibit strong survival
and reproduction (Benson et al. 2016a), while others
suffer from poor demographic performance due to ongoing deterministic stressors (Caughley 1994). Comparing dynamics of isolated populations of the same species but of varying abundance, and with different vital rates and associated deterministic stressors, will increase our
understanding of demographic and genetic processes of
small populations.

Mountain lions (Puma concolor) exist at low density,
have female-biased sex ratios, and often exhibit highly skewed male reproductive success (Johnson et al. 2010,
Riley et al. 2014). These traits reduce effective population size (Mills and Smouse 1994) and have made mountain lions important study species for investigating small population dynamics (Johnson et al. 2010, Benson et al. 2016a). Indeed, one of the clearest demonstrations of
inbreeding depression driving a population to the brink of extinction was with endangered Florida panthers (a
subspecies of mountain lions, Puma concolor coryi;
Johnson et al. 2010). Panthers exhibited reduced fitness from inbreeding depression and declined to fewer than
30 individuals; however, extinction was avoided and the population rapidly increased following genetic restoration (Johnson et al. 2010).

Small, isolated populations of mountain lions have
also persisted within the highly urbanized landscape of southern California in the Santa Monica Mountains
(SMM) and Santa Ana Mountains (SAM) northwest
and southeast of Los Angeles. These two populations
exhibit the lowest genetic diversity documented for the species aside from Florida panthers (Ernest et al. 2014, Riley et al. 2014). Recently, a population viability analysis (PVA) indicated that mountain lions in the SMM population exhibited strong survival and reproduction
and predicted generally stable population growth for the next 50 yr (Benson et al. 2016a). However, this PVA also predicted potential for extinction due purely to demographic factors, as well as rapid loss of genetic diversity that raised concern about inbreeding depression (Benson et al. 2016a). An earlier PVA for mountain lions occupying the SAM indicated that the population was demographically unstable and that additional habitat loss would lead to a high risk of extinction (Beier 1993). This PVA explicitly considered the influence of corridors and habitat loss on extinction probability due to
demographic processes, but ignored potential effects of inbreeding depression. Furthermore, most of the demographic rates came from the literature rather than from empirical data collected within the SAM (Beier 1993).

Given the isolation and low genetic diversity documented for this population (Ernest et al. 2014, Gustafson et al. 2017), as well as additional fragmentation of the available habitat that has occurred (Burdett et al. 2010), an updated PVA constructed with empirical
genetic and demographic data is needed for mountain
lions in the SAM to evaluate the influence of interactions between genetics, demography, and landscape connectivity in this heavily human-dominated landscape.

The SMM and SAM are both occupied by small populations of mountain lions in similar habitats isolated by anthropogenic barriers and exhibiting low levels of
genetic diversity (Ernest et al. 2014, Riley et al. 2014).

However, there are notable differences in demographic
structure of the two populations that could have consequences for population dynamics and viability. First, the estimated number of breeding adults in the SAM was
approximately twice that estimated for the SMMs (Beier
1993, Ernest et al. 2014, Riley et al. 2014, Benson et al. 2016a). Differences in abundance were clearly related to the smaller patch of available habitat within the SMM
relative to the SAM (Beier 1993, Benson et al. 2016a).

Importantly, smaller population and habitat island size are strong predictors of reduced genetic diversity
(Frankham 1995). Second, survival rate of radiocollared adult mountain lions in the SAM, where the main cause
of death was collisions with vehicles, was lower than
other unhunted populations (Vickers et al. 2015). In
contrast, adult survival in the SMM was as high or
higher than most unhunted populations and the main
cause of death was intraspecific strife (Riley et al. 2014, Benson et al. 2016a). This could have important implications because mountain lion population growth is most strongly influenced by adult female survival (Lambert et al. 2006, Benson et al. 2016a). Thus, comparing the dynamics of these populations will inform conservation efforts and provide empirical insight into the influence of variation in demographic structure (i.e., abundance and survival rate) on the relative influence of demographic and genetic processes, and how they interact to influence extinction risk. Such research would represent an important case study for understanding
the dynamics of isolated populations and provide insight into management strategies for maintaining viable
populations of top predators within human-dominated
landscapes.

We used the individual-based population model of
Benson et al. (2016a) parametrized with empirical
demographic and genetic data collected during longterm studies of mountain lions in our focal populations to model dynamics and viability. We constructed starting populations with empirical, multi-locus genotypes that
reflected the age, sex, and genetic structure of the current populations and projected models forward to estimate
Article e01868; page 2 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
stochastic population growth, extinction probability,
and measures of genetic diversity over the next 50 yr. We used these model projections to address multiple questions regarding the viability of populations of top predators in isolated mountain ranges within highly urbanized landscapes. First, we investigated the dynamics and viability of mountain lions in the SAM to evaluate the influence of demographic and genetic processes on probability of extinction. Second, we hypothesized that variation in population abundance and mortality patterns in small, isolated mountain lion populations would result
in differences in population growth, the rate of loss of genetic diversity, and extinction probability. We predicted that reduced adult survival would result in lower population growth and greater extinction probability
due purely to demographic processes (P1a). We also
tested the alternative prediction that the greater number of breeding adults and carrying capacity in the SAM
would offset the lower survival and result in similar
growth and extinction probability between the two populations (P1b). Next, we predicted that genetic diversity would erode more quickly in the SMM population given
the smaller number of individuals and smaller amount
of available habitat (P2). Finally, we predicted that
reductions in vital rates due to inbreeding depression
would result in high probability of extinction for both populations (P3). We provide the first PVA for mountain lions in the SAM that explicitly models both demographic and genetic processes. More broadly, our results elucidate how variation in abundance, carrying capacity, vital rates, and sources of mortality influence mechanisms underlying the extinction vortex for isolated populations in fragmented landscapes. Thus, our work provides a case study that will help to inform conservation of isolated wildlife populations in human-dominated landscapes.

MATERIALS AND METHODS
Study area
We studied mountain lions in two isolated mountain
lion populations occupying mountain ranges southeast
(SAM: Orange, Riverside, and San Diego Counties) and
northwest (SMM: Los Angeles and Ventura Counties;
Fig. 1) of the city of Los Angeles. The SAM population
Santa Monica
Mountains
Santa Ana Mountains
Eastern Peninsular Range
Los Angeles
San Diego
Chino Hills
15 91
405
0 75 150 km
FIG. 1. Greater Los Angeles, southern California, USA showing the location of the Santa Monica (blue polygon) and Santa Ana (red polygon) Mountains within which we studied population dynamics of mountain lions. Also shown are other nearby mountain ranges, major (white lines) and more minor (gray lines) roads, and areas where natural habitat has been replaced by urbanization (dark gray) and agriculture (lighter gray).

Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 3 inhabited approximately 1,533 km2 in the SAM, a portion of the Peninsular Ranges including federal, state, county, and private lands. The SMM population inhabited approximately 600 km2 in the Santa Monica Mountains, part of the Santa Monica Mountains National Recreation Area, a unit of the National Park Service that included an assemblage of federal, state, and privately owned lands. The areas occupied by both populations
were bordered by a combination of anthropogenic (freeways, development, agriculture) and natural (Pacific Ocean) barriers that have drastically restricted movement of mountain lions between the populations and surrounding areas. Both were characterized by a Mediterranean climate, with cool, wet winters and hot, dry summers. Vegetation consisted mainly of mixed chaparral, coastal sage scrub, oak woodlands and savannahs, riparian woodlands, and nonnative annual grasslands.

Mountain lions were the only remaining large carnivore
and the only wild ungulates were mule deer (Odocoileus
hemionus). Both study areas have been described extensively elsewhere (Burdett et al. 2010, Riley et al. 2014, Vickers et al. 2015, Benson et al. 2016b).

Capture and monitoring
We captured mountain lions using Aldrich foot-snares
or cable restraints, baited cage-traps, or by treeing them with trained hounds. We deployed global positioning
system (GPS) or very high frequency (VHF) radio-collars on adult and subadult mountain lions. In the SMM, we also captured 3–5 week old kittens at natal dens by
hand and implanted VHF transmitters in their peritoneal cavities (Moriarty et al. 2012). We monitored survival and determined causes of mortality of radioinstrumented mountain lions as described previously (Beier and Barrett 1993, Vickers et al. 2015, Benson
et al. 2016a). We monitored reproduction of all collared females in the SMM using GPS telemetry to locate natal
dens and count kittens (Moriarty et al. 2012). In the
SAM, all capture and handling was conducted under
Protocol 10950/PHS, Animal Welfare Assurance number
A3433-01, with capture and sampling procedures
approved in Protocol number 17233 by the Animal Care
and Use Committee at the University of California,
Davis, and Memoranda of Understanding and Scientific
Collecting Permits from the California Department of
Fish and Wildlife (CDFW). In the SMM, animal capture and handling protocols were approved by the National Park Service Institutional Animal Care and
Use Committee under protocol PWR_SAMO_Riley_Mt.Lion_2014.A3.

Genotyping
We genotyped all captured mountain lions at 44
(SAM) or 54 (SMM) microsatellite loci using laboratory
methods and markers described previously (Ernest et al. 2014, Riley et al. 2014). Briefly, we extracted DNA from blood or tissue using DNeasy Blood & Tissue Kit
(QUIAGEN, Valencia, California, USA). The loci used
for genotyping mountain lions in both populations conformed to expectations for Hardy-Weinberg and linkage equilibria (Ernest et al. 2014, Riley et al. 2014). We used many of the same genotypes analyzed by Ernest et al.
(2014) and Riley et al. (2014) to parameterize our model but also included genotypes from mountain lions captured more recently. We also genotyped mountain lions from samples obtained from areas adjacent to our focal
populations to simulate immigration in our models.

Demographic parameters
We separated mountain lions into three age classes for
parameter estimation. Kittens were dependent offspring with their mother (0–14 months), subadults were independent animals prior to reproduction (females, 14– 25 months; males, 14–42 months), and adults were breeding animals (females, >25 months; males, >42 months; Benson et al. 2016a). We estimated sex and age-class
specific survival rates using the Kaplan-Meier estimator generalized for staggered entry (Pollock et al. 1989)
implemented in R version 3.1.3 (R Development Core
Team 2015) with the package “survival”. We estimated
survival for adults and subadults separately for the SMM and SAM using empirical data from each population. We
used survival data collected during 1987–1993 (Beier and Barrett 1993) and 2003–2016 (Vickers et al. 2015; T. W.

Vickers et al., unpublished data) for the SAM, and during 2002–2015 for the SMM (Riley et al. 2014, Benson et al. 2016a). Females in the model bred in the first month after reaching adulthood and again following loss or independence of kittens, consistent with documentation in our field study (Benson et al. 2016a). We estimated the probability of females having two, three, or four kittens in a litter based on the proportion of these litter sizes
documented in the SMM during 2004–2017 (all input
demographic parameters are shown in Appendix S1:

Table S1). Although the samples sizes used to estimate
demographic parameters were relatively small numerically, they should be representative given the small size of the populations.

Model overview
We used the individual-based population model for
mountain lions of Benson et al. (2016a) that incorporated demographic and environmental stochasticity, as well as a simple form of density dependence. We did not have data to understand the influence of catastrophes on vital rates of mountain lions in these populations, so our model assumes these unpredictable events do not occur
during our projections. We began models with starting
populations of individuals that reflected the sex, age, and genetic structure of the populations and projected
the models forward to estimate the demographic and
genetic structure of future populations. In the SAM, we Article e01868; page 4 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
combined information from published estimates of population density and available habitat, as well as information from our ongoing 15-yr field study to assemble the starting population. Beier and Barrett (1993) and Beier (1993) estimated 2,070 km2 of available habitat for the SAM population. We adjusted this estimate by subtracting 506 km2 to remove the Chino Hills that are now isolated from the SAM by highway 91 and no longer occupied by mountain lions (Fig. 1). We further reduced the available habitat by 2% to reflect habitat loss during 1993–2017 based on estimates of Burdett et al. (2010).

Thus, our estimate of available habitat for mountain
lions in the SAMs was 1,533 km2
. Beier and Barrett
(1993) estimated mountain lion density to be 0.7 females and 0.35 males/100 km2
. We applied estimates of mountain lion density (0.7 females and 0.35 males/100 km2 ;
Beier and Barrett 1993) to our habitat area estimate,
which yielded 11 adult females and 5 adult males. We
also included 9 kittens and 4 subadults. Although we did not formally estimate population density in our study,
the abundance:habitat area ratios we used from Beier
and Barrett (1993) agreed with observations made using
telemetry, genetic analysis, and camera trapping during our intensive 15-yr study. We assigned empirical genotypes at 44 loci to all starting individuals in the SAM.

The starting population for the SMM population was 15
mountain lions (including 5 adult females and 2 adult
males) with empirical genotypes at 54 loci as described by Benson et al. (2016a).

We ran simulations consisting of 5,000 population
projections of 50 yr unless noted otherwise. Although
researchers sometimes attempt to predict extinction
probability farther into the future (e.g., 100 yr), we followed the recommendation of Morris and Doak (2002:452) to avoid projecting population viability far into the future because of the increased uncertainty of predictions made over longer time periods. The population dynamics simulated by the model were a reflection of individual-based demographic processes specified by
empirical probability distributions estimated with data collected in both populations. We imposed mortality
(survival senescence) on all mountain lions of both sexes in the model that reached 15 yr of age (Benson et al.
2016a). We incorporated density dependence by imposing a maximum number of adult, breeding males (SMM, n = 2; SAM, n = 5) and females (SMM, n = 6; SAM,
n = 11) that could exist in the population at any given time. For the SMM, we felt confident that our estimates were the maximum numbers of breeding individuals that
could occupy the available habitat. The greater area and size of the SAM population contributed to uncertainty
in our carrying capacity estimates; thus, we also explored an alternate scenario with a greater carrying capacity of 7 adult males and 14 adult females. For all scenarios,
when all the adult slots of a given sex were occupied, we eliminated individuals of that sex that would have otherwise transitioned from sub-adults to adults. This process simulated density-dependent population regulation
through death or dispersal. Although the upper limits
for adult males and females were fixed, the number of
adults varied stochastically during model projections
due to variation in survival and reproduction. When
breeding occurred within the model, we assigned genotypes to resulting offspring based on principles of Mendelian genetics (i.e. 1 allele randomly inherited from each parent at each loci). Additional details of the model and submodels are provided by Benson et al. (2016a).

Submodels
Survival.—We incorporated environmental and demographic stochasticity into age-class-specific survival rates as in Benson et al. (2016a). Specifically, we generated environmentally stochastic monthly survival probabilities by transforming survival rates and their standard deviations estimated from each study population into
beta shape parameters using the betaval function in the R package popbio. At each monthly time step, we drew
a random survival value from this beta distribution,
which was used as the environmentally stochastic survival probability for all individuals of the same sex and age class during that time step. We then assessed demographically stochastic survival of each individual using a Bernoulli trial with the monthly survival probability as the threshold between survival and mortality.

Reproduction.—We designated reproductive males and
females in the starting population and, thereafter, randomly selected breeding animals from subadults eligible to transition to adults when openings became available.

Female age at first reproduction varied stochastically
between 25 and 33 months in our model. Males reaching
breeding status remained reproductive until death. If no adult males were present in our simulated populations,
males were allowed to begin breeding at 36 months as
the reason for delayed breeding in males is presumably
due to social constraints imposed by dominant adult
males. Breeding females were eligible to become pregnant until death whenever they did not have dependent offspring. Litter size varied stochastically by generating a random, uniform value between 0 and 1 for each
reproductive female and comparing the value to a cumulative probability distribution for litter sizes we documented. We determined the sex of each offspring using a Bernoulli trial with a probability of 0.5.

Immigration.—We assigned a fixed annual immigration
rate prior to starting a simulation. We transformed this into a monthly probability and assessed immigration
stochastically using Bernoulli trials during each monthly time step. We restricted immigration to subadult males.

Subadult males are more likely to disperse and to undertake longer and riskier dispersal events than females (Sweanor et al. 2000). Indeed, all immigration documented into the SMM and SAM populations has involved subadult males (Riley et al. 2014, Gustafson
Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 5 et al. 2017). We assigned genotypes to immigrants from
mountain lions genotyped in adjacent areas north and
east of the SMMs (n = 18) and east of the SAM
(n = 83). We modeled different immigration scenarios
ranging from no immigration to a rate of one immigrant
per year for our main analyses. We also modeled immigration rates for both populations based on immigration observed with radio-tracking and genetic analysis of
mountain lions within and adjacent to our focal populations. Specifically, we observed two immigrants in 15 yr in the SMM (Riley et al. 2014; S. Riley et al., unpublished data) and three immigrants in 15 yr in the SAM (Gustafson et al. 2017). We were conservative with
respect to modeling how much additional immigration
could occur in our main analyses so we limited these to 1 immigrant per year. However, we also conducted additional scenarios to explore the hypothetical influence of two immigrants per year. For additional details of all
submodels see Benson et al. (2016a).

Model outputs
Demography and extinction.—We estimated kt (Lambda
[population growth] at time t) as Nt/Nt
1, where
Nt is total population size at time t. We estimated ks
(stochastic lambda) across time periods of interest with the formula:
½
PNyears lnðktÞ
Nyears
We report median ks from the distribution of values
across all projections for simulations of interest. We estimated credible intervals for ks using the highest posterior density (HPD) derived using the R package coda (v. 0.17-1). We estimated probability of extinction as the proportion of projections that went extinct during a
given simulation and derived estimates of variability by conducting a nonparametric bootstrapping procedure
implemented in the R package boot (v. 1.3-17). We ran
1,000 bootstraps of 5,000 population projections to estimate uncertainty regarding extinction probability with 95% HPD intervals. We estimated the effective population size based on a census of the breeding animals in simulated populations using the formula: Ne = (4 9
NBF 9 NBM)/(NBF + NBM) (Crow and Kimura 1970),
where Ne is the effective population size, NBF is
the number of breeding females, and NBM is the number
of breeding males.

Genetic parameters.—We estimated measures of genetic
diversity from genotypes of mountain lions in populations simulated by our models 1–50 yr in the future using mean values across all projections. Specifically, we estimated expected (He) and observed (Ho) heterozygosity, individual inbreeding coefficient (Fis), the mean number of alleles per loci (NA), and the proportion of
polymorphic loci using the R package adegenet v. 2.0.0.

Our genetic predictions varied stochastically because
they were realistically linked to the stochastic demographic processes we modeled. Thus, by running 5,000 projections for each scenario, our models captured considerable environmental, demographic, and genetic stochasticity.

Elasticity analysis.—We investigated proportional sensitivity (elasticity) of ks to small (5%) increases in vital rates (Morris and Doak 2002). We conducted these analyses with the density-independent model to investigate which demographic parameters had the greatest influence on ks in the absence of density-dependent limitations. We calculated sensitivity values (S) for each demographic parameter:

S ¼ LogksðincreasedÞ
LogksðoriginalÞ
parameterðincreasedÞ
parameterðoriginalÞ
and elasticity (E) for each demographic parameter following Morris and Doak (2002):

E ¼ S  parameteroriginal
parameteradjusted !:

Inbreeding depression.—We simulated inbreeding depression by running population projections with input parameters reduced to reflect proportional changes in
age- and sex-specific survival rates documented between inbred and outbred Florida panthers following the
genetic restoration program (Hostetler et al. 2010,
Benson et al. 2011; see Appendix S2: Table S1).

RESULTS
Population viability in Santa Ana Mountains
Density-dependent simulations predicted stable median stochastic population growth over the next 50 yr in the SAM, regardless of the level of immigration
(Table 1). However, there was an 11–21% probability of
extinction across all immigration scenarios in the density-dependent simulations, inversely related to the level of immigration (Table 1, Fig. 2). The scenarios without immigration, or with the low level observed in our study, resulted in substantial loss of genetic diversity (e.g., 28–49% of expected heterozygosity) over 50 yr (Fig. 3;
Appendix S3, S4). Predicted loss of heterozygosity
decreased with higher levels of immigration, and
heterozygosity was largely maintained with one immigrant per year (Fig. 3; Appendix S3: Table S1, Appendix S4: Fig. S1). Other measures of genetic diversity including percent polymorphism, inbreeding coefficient, and the number of alleles per loci responded to varying degrees of isolation and immigration similarly
over time (Appendix S3: Table S1). When we explored
the influence of a larger carrying capacity in the SAM (7 adult males and 14 adult females), population growth
Article e01868; page 6 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
rate was similar (ks = 1.01 [0.92, 1.02], but extinction probability was reduced (10% with observed level of
immigration; Appendix S5: Table S1). All immigration
scenarios with higher carrying capacity yielded lower
probability of extinction, ranging from 12% with no
immigration to 5% with one immigrant per year
(Appendix S5: Table S1). Loss of genetic diversity slowed slightly and effective population size increased with greater carrying capacity, although diversity still
declined substantially with no immigration or the
observed level (Appendix S5: Table S2). Simulating
inbreeding depression in the SAM by reducing age-specific survival rates proportional to reductions documented in inbred Florida panthers, resulted in rapidly declining population growth (ks = 0.84, [0.61, 0.96]),
100% probability of extinction over fifty years, and median time to extinction of 11.7 yr (5.2, 23.5; Fig. 4).

Comparing dynamics of SAM and SMM
Median stochastic population growth rate predicted
by the density-dependent scenarios was similarly stable in the two populations (Table 1, Fig. 2). However, the
density-dependent scenarios for both populations also
predicted extinction probabilities of 16–28% over 50 yr with no or observed immigration (Table 1, Fig. 2).

Extinction probability due purely to demographic processes was reduced for both populations with higher levels of immigration, but more so for the SMM
(Table 1, Fig. 2). Density-independent scenarios predicted a rapidly increasing population in the SMM (ks = 1.17 [1.11, 1.22]), whereas the predicted trend in the SAM did not differ from stable (ks = 1.06 [0.89,

1.12]; Fig. 5). Sensitivity and elasticity analysis showed that adult female survival had the strongest influence
on density-independent population growth in both populations (Appendix S6: Table S1). Female subadult survival, female kitten survival, and litter size had moderate influence on population growth for both populations,
whereas male survival parameters had little influence
(Appendix S6: Table S1). When we explored the influence of two immigrants per year, extinction probability was further reduced and genetic diversity increased
beyond the starting values in 50 yr (Appendix S7: Tables S1, S2).

Genetic diversity declined rapidly in both populations
with no or observed immigration (Fig. 3; Appendix S3,
S4). No immigration resulted in a greater loss of genetic diversity for SMM (57% loss expected heterozygosity)
compared with SAM (49% loss; Fig. 3; Appendix S3,
S4). However, the SMM population responded more
strongly to increased levels of immigration as with one immigrant every 1–2 yr, the SMM retained more of its
genetic diversity over 50 yr relative to the SAM (Fig. 3; Appendix S3, S4). Similar to the SAM, simulating
inbreeding depression in the SMM resulted in predictions of declining population growth (ks = 0.89, [0.75,

0.96]), high probability of extinction (>99%) over 50 yr, and rapid median time to extinction (15.1 yr; Fig. 4).

DISCUSSION
Our modeling predicted a 16–21% probability of local
extinction for mountain lions in the SAM over the next
50 yr with the low level of immigration observed in our study or no immigration. Thus, demographic and environmental stochasticity leave the SAM population vulnerable to extinction even before considering inbreeding depression, consistent with earlier predictions for this population (Beier 1993). Furthermore, our results
TABLE 1. Demographic results predicted by individual-based population model for mountain lions in the Santa Ana and Santa Monica Mountain, Southern California, USA.

Parameter
No immigration Observed immigration 1 immigrant/2 yr 1 immigrant/1 yr Estimate 95% HPD† Estimate 95% HPD† Estimate 95% HPD† Estimate 95% HPD† Santa Anas
ks‡ 1.00 0.89, 1.01 1.00 0.95, 1.01 1.01 0.93, 1.02 1.01 0.95, 1.02 Extinction probability 0.22 0.20, 0.23 0.16 0.15, 0.17 0.11 0.10, 0.11 0.08 0.07, 0.09 Time to extinction (yr) 31 12, 50 31 11, 50 33 13, 50 33 13, 50 Adults (n)§ 8 0, 11 9 0, 15 9 0, 15 10 0, 15
NE§ 6 0, 11 6 0, 12 6 0, 12 7 3, 14
Santa Monicas
ks‡ 1.00 0.89, 1.02 1.01 0.93, 1.02 1.01 0.98, 1.02 1.01 1.00, 1.02 Extinction probability 0.29 0.28, 0.30 0.16 0.15, 0.17 0.04 0.04, 0.05 0.02 0.01, 0.02 Time to extinction (yr) 31 13, 50 31 13, 49 33 12, 50 32 11, 48 Adults (n)§ 5 0, 8 5 0, 8 6 0, 8 7 4, 8
NE§ 4 0, 6 4 0, 6 4 0, 6 5 3, 6
Notes: Estimates are median or mean estimates at year 50 based on 5,000 population projections.

HPD, highest posterior density; ks, stochastic population growth. † 95% highest posterior density credible intervals.
‡ Median value.
§ Effective population size (mean value).

Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 7 suggest that, unless gene flow is increased, genetic diversity will rapidly erode in the SAM, and that, if inbreeding depression occurs rapidly, local extinction will be highly likely. We acknowledge that it is impossible to
predict exactly when inbreeding depression will occur in a wild population, but our predictions with respect to
genetic diversity are alarming and far surpass proportional losses of heterozygosity suggested by previous researchers to warrant concern regarding inbreeding
depression (e.g., 5–10% loss in 100 yr; Soul
e et al. 1986, Allendorf and Ryman 2002). Importantly, our predictions suggest that the loss of genetic diversity in SAM mountain lions in the next 50 yr will approach proportional losses experienced in another population of the same species (Florida panthers) that nearly went extinct due to poor demographic performance associated with
inbreeding depression (Johnson et al. 2010;
Appendix S4). Simulating a higher carrying capacity of 7 adult males and 14 adult females resulted in reduced
extinction probability and slowed the loss of heterozygosity, highlighting benefits of even small increases in additional habitat and number of breeding adults
(Frankham 1995). However, even with higher carrying
capacity and abundance, the model predicted a 10%
probability of extinction and 24% loss of expected
heterozygosity over 50 yr.

Dynamics of the SAM and SMM populations were
broadly similar, but our simulations revealed differences in their dynamics caused by variation in deterministic
stressors, survival rates, and population abundance.

The density-independent scenarios provided partial
support for our prediction that lower survival in the
SAM would negatively influence growth rate (P1a).

Clearly, density independence is unrealistic given the
space limitations experienced by both populations;
however, these scenarios were instructive to compare
growth potential and dynamics. Density-independent
models predicted a rapidly increasing population for
the SMM, whereas density-independent ks in the SAM
did not differ significantly from a stable trend. Extinction probability was approximately three times greater (5.6%) in the SAM compared to the SMM (1.8%) in
the absence of density dependence. Furthermore, in our
density-dependent scenarios, higher levels of immigration (1–2 per year) in the SMM raised the credible interval of ks above 1, predicting a slightly increasing trend, whereas credible intervals overlapped 1 for all
predictions of ks in the SAM, even with similarly high
levels of immigration (Table 1; Appendix S7: Tables S1, S2). Clearly, realized population growth is limited by
available habitat in both populations, but growth
potential also appears to be limited by high humancaused mortality in the SAM. The leading cause of death for radiocollared mountain lions in the SAM
was vehicle strikes, which did not differ in frequency by age or sex class, and resulted in high mortality of
adults (Vickers et al. 2015). Although poor adult male
survival had relatively little influence on density-dependent population growth, it influenced extinction probability by causing male extinction in some simulations for this small population with a female-biased adult sex ratio. These dynamics appear to be realistic as there
was evidence of occasional male extinction in the SAM
during previous research (Beier 1993). Conversely, adult survival of both sexes was high in the SMM where
population growth appears to be mainly limited by the
lack of additional habitat. Subadults survive poorly in the SMM due to the difficulty of successfully
●
●
●
●
●
●
●
●
●
●

0.0 0.1

0.2 0.3
No
immigration
Observed
immigration
One immigrant
in four years
One immigrant
in two years
One immigrant
per year
Mean extinction probability
Population
●
●
Santa Ana Mountains
Santa Monica Mountains
FIG. 2. Estimated extinction probability (without considering potential inbreeding effects) in 50 yr for mountain lion populations in the Santa Ana and Santa Monica Mountains from an individual-based population model based on 5,000 projections and varying levels of immigration.

Article e01868; page 8 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
dispersing, as many young animals are killed by breeding males or hit by vehicles before or during dispersal (Riley et al. 2014). The difficulty of dispersal,
combined with high survivalof breeding adults in a
space-limited population provides few opportunities for mountain lions born in the SMM to breed.

One immigrant per year
One immigrant in 2 years
Observed immigration
No immigration
0 10 20 30 40 50

0.0 0.1

0.2 0.3

0.4 0.5

0.0 0.1

0.2 0.3

0.4 0.5

0.0 0.1

0.2 0.3

0.4 0.5

0.0 0.1

0.2 0.3

0.4 0.5
Time (yr)
Heterozygosity (expected)
Population
Santa Ana Mountains
Santa Monica Mountains
FIG. 3. Estimated expected heterozygosity over 50 yr for mountain lion populations in the Santa Ana and Santa Monica Mountains from and individual-based population model based on 5,000 projections and varying levels of immigration.

Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 9 The SMM population had a slightly higher probability of extinction with no immigration than the SAM in our density-dependent scenarios, but increasing immigration resulted in a more pronounced reduction in extinction probability for the SMM. In fact, with one
immigrant per year, extinction probability in the SMM
did not differ from that predicted by the density-independent model suggesting that increased connectivity could largely mitigate the effects of isolation and limited habitat in the SMM, at least with respect to demographic extinction risk. The lesser positive impact of immigration on demographic extinction probability in
the SAM was likely associated with the lower survival of adult males, which meant that tenure of immigrants successfully establishing as breeding adults was often shortlived. Although these comparisons were useful for evaluating the influence of variation in demographic structure on the dynamics of small populations, we recommend
cautious interpretation of these differences for practical purposes. Indeed, predictions regarding extinction probability from PVA are probably best viewed as relative assessments (Morris and Doak 2002).

The greatest long-term threat to both populations
appears to be the rapid loss of genetic diversity associated with their isolation from mountain lions in surrounding areas. With no immigration, the predicted rate FIG. 4. Density-dependent demographic projections from individual-based population model showing predicted population sizes for mountain lions in the Santa Ana and Santa Monica Mountains over 50 yr based on 5,000 projections when we simulated inbreeding depression with the observed level of immigration. 0 500
1,000
1,500
2,000
2,500
0 5 10 15 20 25
Time (yr)
Population size
Population
Santa Ana Mountains
Santa Monica Mountains
FIG. 5. Density-independent demographic projections from individual-based population model showing predicted mountain lion population size in the Santa Ana and Santa Monica Mountains over 25 yr based on 5,000 projections.

Article e01868; page 10 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
of loss of expected heterozygosity over 50 yr was greater for the smaller SMM population relative to the SAM
population. This provided support for our prediction
(P2) and is consistent with theoretical and empirical
work indicating that population abundance and habitat
island size are strong, positive predictors of genetic
diversity (Crow and Kimura 1970, Frankham 1995).

However, with immigration rates observed during the
last 15 yr, predicted loss of heterozygosity was similar in the two populations. Importantly, simulating increased
immigration and gene flow had a stronger positive influence on heterozygosity in the smaller SMM population.

Thus, although heterozygosity is lost more rapidly in
smaller populations, immigration events can also more
quickly reverse these losses and restore diversity. In the SMM, only one or two males generally breed at any one
time, such that when a radiocollared male immigrant
entered the population in 2009 and began breeding it
resulted in a rapid increase in population-level genetic diversity (Riley et al. 2014). A single breeding immigrant also positively influenced genetic diversity in the SAM (Gustafson et al. 2017), and relatively few immigrants
have similarly influenced small populations of other
large mammals (Vila et al. 2003, Hogg et al. 2006,
Adams et al. 2011). However, the key to maintaining
diversity in small populations is to ensure that immigration occurs consistently (Mills and Allendorf 1996), to prevent reversal of short-term diversity gains as immigrants begin breeding with their offspring (Riley et al. 2014, Benson et al. 2016a). In addition to the larger
population size, lower adult survival likely contributed to a reduced positive influence of immigration on genetic diversity in the SAM by limiting the reproductive success of immigrants. This finding further highlights the link between demographic and genetic factors in terms
of influencing extinction in small populations. Despite interesting differences, we stress that our models predict rapid loss of diversity in both populations, indicating that viability will likely be compromised by interactions between genetics and demography unless gene flow is
increased.

Mountain lions are not endangered in southern California and genetically diverse populations of mountain lions exist in areas such as the Sierra Nevada Mountains and other mountain ranges in southern California (Ernest et al. 2014, Riley et al. 2014). However, there is value to conserving viable populations of a native top predator within the SAM and SMM to maintain stable predator–
prey dynamics and naturally functioning ecosystems
within these isolated mountain ranges. This contention
echoes growing recognition among ecologists and managers that conservation efforts should prioritize ecological function and maintaining ecosystem processes across extensive geographic areas, rather than simply preserving minimum viable populations somewhere across the range of a species (Soul
e et al. 2003, Ritchie et al. 2012).

Predators and other highly interactive species may be
especially important to conserve in as many places as
feasible to maintain important species interactions and ecosystem functions (Soul
e et al. 2003, Lindenmayer
et al. 2008, Cadotte et al. 2011). Indeed, research from around the world has begun to highlight the potential
for conserving large predators within human-dominated
landscapes (Athreya et al. 2013, Chapron et al. 2014,
Riley et al. 2014). Our work suggests that conserving
mountain lions in isolated mountain ranges in greater
Los Angeles is feasible with relatively modest increases in landscape connectivity. If achieved over the longterm, this would be an important step toward maintaining intact, functioning ecosystems in these mountain ranges that lie within one of the most human-impacted
landscapes in the world.

Our results suggest mitigation strategies for mountain
lions in SAM and SMM should target two main threats:
isolation and mortality. Increasing connectivity between both populations and the areas across the freeways
should (1) decrease extinction probability due purely to demographic processes, and (2) maintain genetic diversity and prevent the onset of inbreeding depression.

Translocation of outbred animals can be effective to
quickly increase genetic diversity in threatened mountain lion populations (Johnson et al. 2010), but strategically located highway crossing structures (Gloyne and Clevenger 2001) allowing for dispersal and gene flow could be a more comprehensive long-term strategy. Our results suggest that maintaining genetic diversity in these populations would require at least one migrant every 1–2 yr.

Given the expense of erecting highway crossing structures, translocation would certainly be a less expensive strategy, especially in the short-term. Indeed, the estimated cost for a bridge to connect the SMM population with habitat north of the 101 Freeway (Fig. 1) is approximately US$60 million. However, our results indicate that animals would need to be translocated frequently
and indefinitely if connectivity is not improved, whereas a highway crossing structure would provide long-term
connectivity once erected. Furthermore, populations of
other species are also isolated by the freeways and other barriers surrounding these habitat islands (Delaney
et al. 2010, Riley et al. 2006). Thus, construction of
highway crossing structures, although unquestionably an expensive initial investment, would likely provide regular, consistent immigration of mountain lions and many other species that should increase the likelihood of maintaining healthy populations and intact ecosystems within these isolated mountain ranges. However, we certainly
do not discount the value of translocation as a management tool. Translocation may be an especially valuable option if proposed development further degrades or prevents improvement of currently available passageways, and if the significant financial challenges delay construction of new crossing structures. As a specific example, extensive residential and resort development projects
have been proposed for construction in the primary corridor area that has facilitated some movement of mountain lions between the SAM and habitat east of the Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 11 Interstate Highway 15 (Gustafson et al. 2017). Our
results show that further reduction in immigration and
gene flow, which are likely to occur with new development in corridor areas, would increase demographic extinction probability and hasten the loss of genetic
diversity.

Our sensitivity analyses and inbreeding simulations
show that increased mortality could have rapid, negative consequences for population growth and extinction
probability in both populations, supporting our prediction (P3). Despite the smaller population size, predicted demographic extinction probability in the SMM was
generally similar to that in SAM under the current levels of immigration largely because of the strong growth
potential afforded by higher adult female survival. If
female mortality increases in future years from the multitude of mortality agents documented in the SMMs (e.g., aggression from males, vehicle strikes, rodenticide poisoning) this could destabilize the population and
increase extinction probability. Thus, reducing mortality in both populations is important and should decrease
probability of extinction due to environmental and
demographic stochasticity. In addition to highway crossing structures, exclusionary fencing strategically implemented along roadways where mountain lions are killed can be effective at reducing mortality (Foster and Humphrey 1995), such as that recently constructed along SR 241 in the SAM (Vickers et al. 2015). Strategies to promote best practices for housing domestic animals could reduce mortality from depredation permits issued to kill mountain lions threatening livestock (Vickers et al.
2015). To reduce mortality in the SAM and SMM from
depredation mortality, the California Department of
Fish and Wildlife (CDFW) has recently changed policies
regarding depredation permit issuance in these two populations such that non-lethal deterrence methods must be attempted before lethal removal can occur (CDFW
2017).

Differences in demographic structure between the two
populations revealed important aspects that have contributed to their persistence and highlighted management priorities for both populations. The greater demographic vigor of the SMM population is critical to
its persistence, as a population with six to eight adults would clearly be at much higher risk of local extinction if survival and reproduction declined. Thus, in addition to management efforts to reduce mortality from documented causes such as vehicle strikes and rodenticide (Riley et al. 2014), it would be prudent to evaluate and monitor population dynamics of their main prey (mule
deer) in the SMM to ensure the prey base remains adequate to support strong survival and reproduction.

Greater population abundance in the SAM reduced
demographic extinction probability and slowed the erosion of genetic diversity in simulations without immigration. Thus, the larger population size is beneficial to the persistence of mountain lions in the SAM, especially
during periods when no immigration occurs. If
additional habitat loss or fragmentation reduced the
number of breeding adults that could occupy the SAM,
this would have negative consequences for both demographic extinction risk and loss of genetic diversity. For instance, a population as small as the SMM population,
but with the poor survival of the SAM would have a
higher probability of extinction than we documented for either population. As noted above (see Model overview), both habitat loss and isolation appear to have reduced
the population size over the last 25 yr. Our model results suggest it is critical to ensure that future habitat loss in the SAM is prevented and that fragmentation does not
isolate portions of the current population.

Although our model realistically models demographic
and genetic processes in these small populations with
empirical data, we acknowledge that our model and data
have limitations. For instance, although we were able to account for demographic and genetic processes, density
dependence, and varying levels of immigration, we did
not have sufficient data to understand the influence of catastrophes on the vital rates and viability of these populations. Two unpredictable forces that could potentially cause catastrophes include wildfires and disease outbreaks. Wildfires have become larger and more frequent in southern California shrubland ecosystems, and
increasingly destructive wildfires appear to be linked to expansion of the urban–wildland interface (Keeley et al. 1999). Wildfires have directly caused mortality of mountain lions within our study populations (Vickers et al. 2015) and can also temporarily displace mountain lions
(Jennings et al. 2016; S. P. D. Riley and J. A. Sikich unpublished data). Previous research on Iberian lynx (Lynx pardinus) has noted the potential that outbreaks of disease such as feline leukemia and reduced immune response associated with low genetic diversity could negatively affect population viability (Mill
an et al. 2009, Palomares et al. 2012). We note that catastrophic mortality associated with wildfires, disease, or other unpredictable events could substantially increase extinction probability above the predictions generated by our
model.

Our results demonstrate that small populations isolated by freeways and urbanization are subjected to elevated extinction risk due to interactions between demography and genetics. We agree with previous
authors that demographic and genetic risk factors for
small populations should not be considered in isolation (Mills and Smouse 1994, Soul
e and Mills 1998) and that both must be addressed in any comprehensive wildlife
conservation strategy within urbanized landscapes (Benson et al. 2016a). Indeed, other small, isolated populations of felids are threatened by a combination of limited habitat and mortality, such as the highly endangered Iberian lynx (Ferreras et al. 2001). Inbreeding depression and extinction vortex dynamics are also concerns for Iberian lynx, and their conservation will require restoring habitat and improving demographic parameters (Palomares et al. 2012). As urbanization increases globally, it Article e01868; page 12 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
will be necessary to (1) protect habitat patches large
enough to facilitate persistence of populations of large carnivores, (2) mitigate anthropogenic deterministic
stressors, and (3) restore and maintain connectivity
within and between habitat patches if we are to maintain populations and ecosystem processes (e.g., predator–prey interactions) within urban landscapes (Crooks 2002).

Our results also show that relatively small changes in
abundance and key demographic parameters can influence loss of genetic diversity as well as extinction probability due to non-genetic processes. The difficulty of conserving top predators in the modern world are well
documented (Woodroffe 2000, Ripple et al. 2014) and
our work further details the demographic and genetic
challenges facing large carnivores in human-dominated
landscapes. Yet our results also provide reason for optimism, as seemingly realistic increases in gene flow appear sufficient to substantially reduce probability of extinction of top predators due to combined demographic and
genetic threats within the second largest metropolitan
area in the United States. Long-term conservation of
mountain lions in greater Los Angeles would provide
compelling evidence that large carnivores and abundant
human populations are compatible, even within the most
intensely developed landscapes.

ACKNOWLEDGMENTS
We thank the many people who assisted with fieldwork. We thank R. Wayne, K. Gustafson, L. E. K. Serieys, J. Pollinger, D.

Fraser, and T. Drazenovich for assistance with DNA laboratory work. We thank all of the land owners and managers who
allowed field activities on their properties. Research in the SAM was supported by California State Department of Parks and Recreation, California Department of Fish and Game, The Nature Conservancy, The McBeth Foundation, The Anza Borrego Foundation, The Nature Reserve of Orange County, The National Science Foundation, The Foothill/Eastern Transportation Corridor Agency, San Diego County Association of Governments Environmental Mitigation Program, The San
Diego Foundation, Felidae Conservation Fund, The Mountain Lion Foundation, The Santa Rosa Plateau Foundation, the Institute for Wildlife Studies, and private donors. Research in the SMM was supported by the La Kretz Center for California Conservation Science at UCLA, California State Parks, the Santa Monica Mountains Conservancy/Mountains Recreation and Conservation Authority, and the Calabasas Landfill.

LITERATURE CITED
Adams, J. R., L. M. Vucetich, P. W. Hedrick, R. O. Peterson, and J. A. Vucetich. 2011. Genomic sweep and potential
genetic rescue during limiting environmental conditions in an isolated wolf population. Proceedings of the Royal Society B 278:3336–3344.

Allendorf, F. W., and N. Ryman. 2002. The role of genetics in population viability analysis. Pages 50–85 in S. R. Beissinger and D. R. McCullough, editors. Population viability analysis.

University of Chicago Press, Chicago, Illinois, USA.

Athreya, V., M. Odden, J. D. Linnell, J. Krishnaswamy, and U.

Karanth. 2013. Big cats in our backyards: persistence of large carnivores in a human dominated landscape in India. PLoS ONE 8:e57872.

Beier, P. 1993. Determining minimum habitat areas and habitat corridors for cougars. Conservation Biology 7:94–108.

Beier, P., and R. H. Barrett. 1993. The cougar in the Santa Ana Mountain Range, California. Final report. Orange County Cooperative Mountain Lion Study, Department of Forestry and Resource Management. University of California, Berkeley, California, USA.

Beissinger, S. R., J. M. Wunderle, J. M. Meyers, B. E. Sæther, and S. Engen. 2008. Anatomy of a bottleneck: diagnosing factors limiting population growth in the Puerto Rican parrot.

Ecological Monographs 78:185–203.

Benson, J. F., J. A. Hostetler, D. P. Onorato, W. E. Johnson, M.

E. Roelke, S. J. O’Brien, D. Jansen, and M. K. Oli. 2011.

Intentional genetic introgression influences survival of adults and subadults in a small, inbred felid population. Journal of Animal Ecology 80:958–967.

Benson, J. F., P. J. Mahoney, J. A. Sikich, L. E. Serieys, J. P. Pollinger, H. B. Ernest, and S. P. Riley. 2016a. Interactions between demography, genetics, and landscape connectivity increase extinction probability for a small population of large carnivores in a major metropolitan area. Proceedings of the Royal Society B 283:1–10.

Benson, J. F., J. A. Sikich, and S. P. D. Riley. 2016b. Population and individual-level resource selection patterns of mountain lions preying on mule deer along an urban-wildland gradient.

PLoS ONE 11:1–16.

Burdett, C. L., K. R. Crooks, D. M. Theobald, K. R. Wilson, E. E. Boydston, L. M. Boydston, R. N. Fisher, T. W. Vickers, S. A. Morrison, and W. M. Boyce. 2010. Interfacing models of wildlife habitat and human development to predict the future distribution of puma habitat. Ecosphere 1:1–21.

Cadotte, M. W., K. Carscadden, and N. Mirotchnick. 2011.

Beyond species: functional diversity and the maintenance of ecological processes and services. Journal of Applied Ecology 48:1079–1087.

California Department of Fish and Game Departmental Bulletin. 2017-07. https://nrm.dfg.ca.gov/FileHandler.ashx?Docu mentID=153021
Caughley, G. 1994. Directions in conservation biology. Journal of Animal Ecology 63:215–244.

Chapron, G., et al. 2014. Recovery of large carnivores in Europe’s modern human-dominated landscapes. Science 346:1517–1519.

Crooks, K. R. 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation. Conservation Biology 16:488–502.

Crow, J. F., and M. Kimura. 1970. An introduction to population genetics theory. Population (French Edition) 26:977.

Delaney, K. S., S. P. D. Riley, and R. N. Fisher. 2010. A rapid, strong, and convergent genetic response to urban habitat fragmentation in four divergent and widespread vertebrates.

PLoS ONE 5:1–11.

Ernest, H. B., T. W. Vickers, S. A. Morrison, M. R. Buchalski, and W. M. Boyce. 2014. Fractured genetic connectivity
threatens a southern California puma (Puma concolor) population. PLoS ONE 9:e107985.

Ferreras, P., P. Gaona, F. Palomares, and M. Delibes. 2001.

Restore habitat or reduce mortality? Implications from a population viability analysis of the Iberian lynx. Animal Conservation 4:265–274.

Foster, M. L., and S. R. Humphrey. 1995. Use of highway underpasses by Florida panthers and other wildlife. Wildlife Society Bulletin 23:95–100.

Frankham, R. 1995. Conservation genetics. Annual Review of Genetics 29:305–327.

Gilpin, M. E., and M. E. Soul
e. 1986. Minimum viable populations: processes of extinction. Pages 19–34 in M. E. Soul
e, Xxxxx 2019 EXTINCTION VORTEX DYNAMICS Article e01868; page 13 editor. Conservation biology: the science of scarcity and diversity. Sinauer Associates, Sunderland, Massachusetts, USA.

Gloyne, C. C., and A. P. Clevenger. 2001. Cougar Puma concolor use of wildlife crossing structures on the Trans-Canada highway in Banff National Park, Alberta. Wildlife Biology 7:117–124.

Gustafson, K. D., T. W. Vickers, W. M. Boyce, and H. B. Ernest. 2017. A single migrant enhances the genetic diversity of an inbred puma population. Royal Society Open Science 4:170115.

Hogg, J. T., S. H. Forbes, B. M. Steele, and G. Luikart. 2006.

Genetic rescue of an insular population of large mammals.

Proceedings of the Royal Society B 273:1491–1499.

Hostetler, J. A., D. P. Onorato, J. D. Nichols, W. E. Johnson, M.

E. Roelke, S. J. O’Brien, D. Jansen, and M. K. Oli. 2010.

Genetic introgression and the survival of Florida panther kittens. Biological Conservation 143:2789–2796.

Jennings, M. K., R. L. Lewison, T. W. Vickers, and W. M.

Boyce. 2016. Puma response to the effects of fire and urbanization. Journal of Wildlife Management 80:221–234.

Johnson, W. E., et al. 2010. Genetic restoration of the Florida panther. Science 329:1641–1645.

Keeley, J. E., C. J. Fotheringham, and M. Morais. 1999. Reexamining fire suppression impacts on brushland fire regimes.

Science 284:1829–1832.

Lambert, C. M., R. B. Wielgus, H. S. Robinson, D. D. Katnik, H. S. Cruickshank, R. Clarke, and J. Almack. 2006. Cougar population dynamics and viability in the Pacific Northwest.

Journal of Wildlife Management 70:246–254.

Lande, R. 1993. Risks of population extinction from demographic and environmental stochasticity and random catastrophes. American Naturalist 142:911–927.

Lindenmayer, D., et al. 2008. A checklist for ecological management of landscapes for conservation. Ecology Letters 11:78–91.

Mill
an, J., M. G. Candela, F. Palomares, M. J. Cubero, A.

Rodr
ıguez, M. Barral, J. de la Fuente, S. Almer
ıa, and L.

Leon-Vizca

ıno. 2009. Disease threats to the endangered Iberian lynx (Lynx pardinus). Veterinary Journal 182:114–124.

Mills, L. S., and F. W. Allendorf. 1996. The one-migrant-pergeneration rule in conservation and management. Conservation Biology 10:1509–1518.

Mills, L. S., and P. E. Smouse. 1994. Demographic consequences of inbreeding in remnant populations. American Naturalist 144:412–431.

Moriarty, J. G., L. Whited, J. A. Sikich, and S. P. Riley. 2012. Use of intraperitoneal radiotransmitters to study mountain lion (Puma concolor) kittens. Wildlife Society Bulletin 36:161–166.

Morris, W. F., and D. F. Doak. 2002. Quantitative conservation biology: theory and practice of population viability analysis.

Sinauer Associates, Sunderland, Massachusetts, USA.

O’Grady, J. J., B. W. Brook, D. H. Reed, J. D. Ballou, D. W.

Tonkyn, and R. Frankham. 2006. Realistic levels of inbreeding depression strongly affect extinction risk in wild populations. Biological Conservation 133:42–51.

Palomares, F., J. A. Godoy, J. V. Lόpez-Bao, A. Rodr
ıguez, S.

Roques, M. Casas-Marce, E. Revilla, and M. Delibes. 2012.

Possible extinction vortex for a population of Iberian lynx on the verge of extirpation. Conservation Biology 26:689–697.

Pollock, K. H., S. R. Winterstein, C. M. Bunck, and P. D. Curtis. 1989. Survival analysis in telemetry studies: the staggered entry design. Journal of Wildlife Management 53:7–15.

R Development Core Team. 2015. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

Reed, D. H. 2005. Relationship between population size and fitness. Conservation Biology 19:563–568.

Riley, S. P. D., J. P. Pollinger, R. M. Sauvajot, E. C. York, C.

Bromley, T. K. Fuller, and R. K. Wayne. 2006. A southern California freeway is a physical and social barrier to gene flow in carnivores. Molecular Ecology 15:17331741.

Riley, S. P., L. E. Serieys, J. P. Pollinger, J. A. Sikich, L. Dalbeck, R. K. Wayne, and H. B. Ernest. 2014. Individual behaviors dominate the dynamics of an urban mountain lion population isolated by roads. Current Biology 24:1989–1994.

Ripple, W. J., et al. 2014. Status and ecological effects of the world’s largest carnivores. Science 343:1241484.

Ritchie, E. G., B. Elmhagen, A. S. Glen, M. Letnic, G. Ludwig, and R. A. McDonald. 2012. Ecosystem restoration with
teeth: What role for predators? Trends in Ecology & Evolution 27:265–271.

Saccheri, I., M. Kuussaari, M. Kankare, P. Vikman, W. Fortelius, and I. Hanski. 1998. Inbreeding and extinction in a butterfly metapopulation. Nature 392:491–494.

Soul
e, M. E., J. A. Estes, J. Berger, and C. M. Del Rio. 2003.

Ecological effectiveness: conservation goals for interactive species. Conservation Biology 17:1238–1250.

Soul
e, M. E., M. E. Gilpin, W. Conway, and T. Foose. 1986.

The millennium ark: How long a voyage, how many staterooms, how many passengers? Zoo Biology 5:101–113.

Soul
e, M. E., and L. S. Mills. 1998. No need to isolate genetics.

Science 282:1658–1659.

Stacey, P. B., and M. Taper. 1992. Environmental variation and the persistence of small populations. Ecological Applications 2:18–29.

Sweanor, L. L., K. A. Logan, and M. G. Hornocker. 2000. Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology 14:798–808.

Vickers, T. W., J. N. Sanchez, C. K. Johnson, S. A. Morrison, R.

Botta, T. Smith, B. S. Cohen, P. R. Huber, H. B. Ernest, and W. M.

Boyce. 2015. Survival and mortality of pumas (Puma concolor) in a fragmented, urbanizing landscape. PLoS ONE 10:e0131490.

Vila, C., A. K. Sundqvist, Ø. Flagstad, J. Seddon, S. Bjorner- € feldt, I. Kojola, A. Casulli, H. Sand, P. Wabakken, and H.

Ellegren. 2003. Rescue of a severely bottlenecked wolf (Canis lupus) population by a single immigrant. Proceedings of the Royal Society B 270:91–97.

Woodroffe, R. 2000. Predators and people: using human densities to interpret declines of large carnivores. Animal Conservation Forum 3:165–173.

Wootton, J. T., and C. A. Pfister. 2013. Experimental separation of genetic and demographic factors on extinction risk in wild populations. Ecology 94:2117–2123.

SUPPORTING INFORMATION
Additional supporting information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/eap.1868/full

DATA AVAILABILITY
Data are available on Zenodo: https://doi.org/10.5281/zenodo.2548917 Article e01868; page 14 JOHN F. BENSON ET AL. Ecological Applications Vol. 0, No. 0
Report
Mountain lions avoid burned areas and increase risky
behavior after wildfire in a fragmented urban
landscape
Highlights
d After the wildfire, mountain lions avoided burned areas and increased risky behavior
d Post-fire risky behavior included increased road crossings and daytime activity
d Altered movement and space use post-fire increased risk of intraspecific conflict
d Risky behavior may increase mortality risk for animals in urban fire-prone regions
Authors
Rachel V. Blakey, Jeff A. Sikich,
Daniel T. Blumstein, Seth P.D. Riley
Correspondence
[email protected]
In brief
When wildfires and urbanization coincide,
animals trade off risks of burned
landscapes with those of anthropogenic
environments. Blakey et al. find that after
an exceptionally large wildfire in an
urbanized region, mountain lions avoided
burned areas and increased activities that
elevated their risk of negative interactions
with humans and conspecifics.

Blakey et al., 2022, Current Biology 32, 4762–4768
November 7, 2022 ª 2022 The Authors. Published by Elsevier Inc. https://doi.org/10.1016/j.cub.2022.08.082 ll
Report
Mountain lions avoid burned areas and
increase risky behavior after wildfire
in a fragmented urban landscape
Rachel V. Blakey,1,2,3,4,6,7,
* Jeff A. Sikich,3 Daniel T. Blumstein,1,2,5 and Seth P.D. Riley1,2,3,5 1La Kretz Center for California Conservation Science, Institute of the Environment and Sustainability, University of California, Los Angeles, CA, USA
2Department of Ecology and Evolutionary Biology, University of California, Los Angeles, CA, USA 3National Park Service, Santa Monica Mountains National Recreation Area, Thousand Oaks, CA, USA 4Present address: Biological Sciences Department, California State Polytechnic University, Pomona, 3801 W. Temple Ave, Pomona, CA 91768, USA
5These authors contributed equally
6Twitter: @blakey_rachel 7Lead contact
*Correspondence: [email protected]
https://doi.org/10.1016/j.cub.2022.08.082

SUMMARY
Urban environments are high risk areas for large carnivores, where anthropogenic disturbances can reduce fitness and increase mortality risk.1 When catastrophic events like large wildfires occur, trade-offs between acquiring resources and avoiding risks of the urban environment are intensified. This landscape context could lead to an increase in risk-taking behavior by carnivores if burned areas do not allow them to meet their energetic needs, potentially leading to human-wildlife conflict.2,3 We studied mountain lion behavior using GPS location and accelerometer data from 17 individuals tracked before and after a large wildfire (the 2018 Woolsey Fire) within a highly urbanized area (Los Angeles, California, USA). After the wildfire, mountain lions avoided burned areas and increased behaviors associated with anthropogenic risk, including more frequent road and freeway crossings (mean crossings increased from 3 to 5 per month) and greater activity during the daytime (means from increased 10% to 16% of daytime active), a time when they are most likely to encounter humans. Mountain lions also increased their amount of space used, distance traveled (mean distances increased from 250 to 390 km per month), and intrasexual overlap, potentially putting them at risk of intraspecific conflict. Joint pressures from urbanization and severe wildfire, alongside resulting risk-taking, could thus increase mortality and extinction risk for populations already suffering from low genetic diversity, necessitating increased connectivity in fire-prone areas.

RESULTS
Direct effects of wildfire on mountain lions
Direct and immediate effects of wildfire on mountain lions can include injury and mortality. Of the 11 individual mountain lions being tracked at the time of the Woolsey Fire that had the potential to be affected by it, two died or were presumed to have died during or soon after the fire.

Do mountain lions avoid burned areas after a large
wildfire?
At the population level, mountain lions avoided burned areas after the wildfire (Figures 1 and 2) and no individual animal showed significant selection for them. Males avoided burned areas more than females, as indicated by their generally larger and more negative effect sizes (Figure 1). Proportions of locations within burned areas compared before and after the fire showed the same trend as selection analyses (Table S1), specifically, much lower proportions of locations in burned areas post-fire.

Excluding the two males that had less than 10% of their pre-fire locations within the burn perimeter (P56 and P61, Table S1), all 3 males showed strong and significant avoidance (effect size
0.63 to
1.45). The response of females to the fire was more variable (Figure 1). The post-fire burned area use that did occur was concentrated (61%) in the patchily burned region in the southeast corner of the outer burn perimeter, and within the Simi Hills (north of the US-101 freeway) where the majority of the landscape (66%) burned (Figure 2).

Do mountain lions increase behaviors that put them at
anthropogenic risk after a large wildfire?
While there was support for mountain lions increasing use of urban areas after the wildfire, the magnitude of this increase was negligible (Figure 3A). The probability of urban use was low before the fire (4.3%), and while this increased after the fire, it remained low (5.4%); this 1% change was much lower than 4762 Current Biology 32, 4762–4768, November 7, 2022 ª 2022 The Authors. Published by Elsevier Inc.

This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ll

OPEN ACCESS
the range of variability in proportion of urban use by mountain lions across the population (0%–15%) (Table S2). Regardless of fire, mountain lions used urban areas rarely (mean for study animals was 5% of the time, including time periods before and after the fire) and use of urban areas was variable among individuals ranging from one female who used urban areas less than 1% of the time to two females who used urban areas > 10% of the time. All sex and age classes were variable in urban area use.

Consistent with our predictions, mountain lions tended to increase road crossings after the wildfire, with the fitted relationship indicating an increase from 3 crossings per month before the fire to 5 crossings per month 15 months after the fire (Figure 3B).

Mountain lions also increased their daytime activity after the fire from 10% of the day to 16% of the day, although the continuous response model indicated a potential slight increase prior to the wildfire event (Figure 3C). Our analysis pooled all major road crossings (major roads shown in Figure 4), though mortality risk (both perceived and actual) is likely to vary with the size and traffic volume of roads. California has the busiest roads in the USA and the busiest interstate in any USA city runs through our study area (I-405).4 The first successful crossing of the I-405 freeway over the 16 years of the broader study was recorded in the months after the fire; comparing crossing frequencies of the busy US-101 freeway, we observed roughly one crossing every 2 years before the fire, compared to one crossing every 4 months after the fire.

Do mountain lions increase behaviors that could
increase the risk of conflict with conspecifics after a large wildfire?
Mountain lions increased both their distance traveled and the amount of space used after the fire (Figures 3D and 3E). Distance travelled increased from 250 ± 48 (predicted 95 % confidence interval [CI]) km per month to  390 ± 48 km per month, a more than 50% increase from pre-fire distances. Although adult males either decreased or retained similar amount of space used after the fire, subadult males and all females, the groups most at risk in intraspecific encounters, increased their amount of space used by 15%–24%. Results of the age-sex class analyses should be interpreted cautiously due to the low number of individuals per class and wide confidence intervals (Figure 3E and Table S3).

Where analyzed, trends towards increases in spatial overlap in mountain lion landscape use after the fire did not perform better than the null model (Table S4), potentially due to the relatively low sample size and the confounding factor of two males perishing in the fire and an additional three males perishing of anticoagulant rodenticide poisoning and vehicular collision during the 15 months post-fire. However, we saw a trend towards an increase in spatial overlap after the fire between the dominant male and other males in the study area after the fire (Figure 4).

Additionally, mean observed overlap was greater for all agesex classes after the fire across all iterations of the model validation expressed as a proportion of male and female home ranges, though this difference was negligible for male-female overlap (Figure S1). Specifically, important components of intrasexual overlap in this territorial species more than doubled: overlap of the dominant male on other males increased from 10% to 23% post-fire (Figure 4C) and overlap between females increased from 7% to 18% post-fire (Figure S1).

DISCUSSION
In an urban landscape after the wildfire, we found support for the prediction that mountain lions avoided burned areas post-fire, and increased behavior that could expose them to risk. Changes in behavior by mountain lions post-fire are likely due to a complex trade-off balancing the necessity to acquire food and breed, while avoiding conspecific conflict and encounters with humans in a transformed and fragmented landscape. These kinds of trade-offs between anthropogenic disturbances and other major disturbance events are an increasing reality for carnivores persisting in human-dominated landscapes worldwide.5–7 Carnivores have varying responses to fire, and this is likely to be strongly influenced by how fire changes the structure of vegetation, and with it, the ability to capture prey.8,9 In the case of cursorial carnivores, such as wolves and coyotes, fire may increase their abilities to capture prey.10,11 Whereas ambush predators such as mountain lions, lynx, and African lions may require more heterogeneity, including retained vegetation cover, in postfire landscapes in order to successfully stalk prey.12–15 The mountain lions in our study mostly avoided burned areas in the 15 months after the fire. This contrasted with studies that indicate opportunistic use of burned landscapes by carnivores,7,16,17 but was consistent with Eby et al., 13 who found that despite abundant prey in burned areas, African lions (Panthera leo) avoided the burned landscape, likely due to reduced Figure 1. Results of resource selection analysis for burned areas after the fire by the 9 individual mountain lions who were tracked both before and after the fire
The two individuals assumed to have perished during or soon after the fire were excluded. Each point shows the effect size comparing selection for burned areas before and after the fire using step selection functions, for each individual mountain lion. The overall effect size was calculated using a metaanalytic approach and all error bars show 95% confidence intervals. Negative effect sizes indicate selection against burned areas while positive effect sizes indicate selection for burned areas following fire.

See also Table S1.
ll

OPEN ACCESS
Current Biology 32, 4762–4768, November 7, 2022 4763
Report
cover decreasing ambush hunting success. In the Santa Monica Mountains, the most intensive use of burned areas in our study occurred in areas surrounding a patchily burned area in the southeastern part of the outer burn perimeter of the Woolsey Fire (Figure 3), an area that was more heterogeneously burned and that included some sizable unburned patches. Use of these areas could be due to hunting advantages and prey availability in landscapes where burned areas are patchy, and near the edges of burns.14,18 We did not account for differences in burn severity across the landscape, which can be an important predictor of wildlife post-fire habitat use, because fires within Southern Californian shrubby vegetation tend to burn with uniformly high-intensity, stand-replacing fire.19 Our findings are overall consistent with the reduction in predator-prey interactions for ambush predators after the fire proposed by Doherty et al. 9 ,
and the need to find suitable habitat to capture prey is likely one of the drivers of the risk-taking behaviors we observed.2 There is extensive evidence globally that large carnivores avoid areas of high human footprint (areas of relatively greater human population and infrastructural development) in space and time.20,21 Our study indicated that even after a considerable disturbance that transformed the structure of over half the landscape used by the resident population, urban areas remained a strong deterrent. However, mountain lions did increase their exposure to anthropogenic risk by increasing road and freeway crossings and by increasing activity during the day when human activity is greatest. Human killings of mountain lions (in response to depredation of livestock) may be more likely in areas of intermediate housing density than in more urban areas,22 and vehicle strikes are also a very high cause of mountain lion mortality23 in this population. Therefore, mountain lions in our study area may be experiencing an assessment risk-response mismatch,
whereby the animals’ assessment of risk does not accurately reflect mortality risk.24
Reduction of suitable habitat after fire has the potential to result in greater risk of intraspecific conflict in carnivore populations within urban environments, where dispersal is constrained by multiple barriers. Though carnivore home ranges tend to be smaller and population densities higher in urban areas,25 during the study period, the population we studied presented a relatively extreme example, given that the Santa Monica Mountains, south of the 101 freeway, were being used by at least eight males (most being subadults), though its size is the equivalent of 1 to 2 home ranges for adult males.26,27 In this context, multiple behavioral changes by the mountain lions in our study, including a 50% increase in distance traveled, use of 15%–24% larger areas by females and subadults, and a trend towards greater intrasexual overlap, have the potential to increase the risk of intraspecific conflict, especially between males. In our study area, intraspecific conflict, specifically being killed by an adult male, is the biggest cause of mortality for subadult mountain lions, and adult males have also been recorded to kill adult females and kittens, including their own offspring and past mates.23,26 Intraspecific conflict (fatal or otherwise) is likely to be exacerbated in urban areas where barriers prevent subadults from dispersing into new territories.23,26,28 Therefore, after a severe wildfire, when space available for hunting and moving within cover is reduced, animals must trade-off energetic demand with perceived risk of encountering adult males, weighing behaviors that put them at greater risk of conflict against greater flexibility in space use and, potentially, diet.29 The increases in amount of space used and distance traveled that we observed could be influenced by multiple factors. A severe wildfire like the Woolsey Fire could allow mountain lions to move more efficiently by removing dense cover in the landscape and Figure 2. Study area within the Los Angeles
and Ventura County areas of California,
USA, showing locations of 17 individual
mountain lions in periods before and after
the 2018 Woolsey Fire
The study area includes the Santa Monica Mountains (south of the 101 freeway) and Simi Hills (north of the 101 freeway).
(A–F) Locations of 17 individual mountain lions
studied within the periods from 15 months prior the
fire (A) and 15 months after the fire (B)–(F) (in
3-month intervals) are shown in different colors
for each individual. Time periods shown include
15 months pre-fire to time of fire (A); time of fire
to 3 months post-fire (B); 3–6 months post-fire (C);
6–9 months post-fire (D); 9–12 months post-fire
(E); and 12–15 months post-fire (F). Of the 17 individuals, 12 were tracked both pre- and post-fire (though of these, 1 individual was suspected to
have perished in the fire and 1 individual died soon
after) and 5 individuals were tracked only after the
fire (Figure S3). Land use is shown by dark green
(natural areas), light green (altered open areas) and
gray (urban areas). The area burned by the Woolsey Fire (2018) is shown in white outline with white hatching. Freeways are shown in yellow.

See also Table S1.
ll

OPEN ACCESS
4764 Current Biology 32, 4762–4768, November 7, 2022
Report
due to the reduction in human recreational use in the short-term after fire.30,31Alternatively, increased space use could indicate an increase in avoidance of either humans or adult males, in the more sparse landscape where concealment is more challenging, given that mountain lions generally avoid open areas.27 Alternatively, or perhaps concurrently, hunting could be more difficult for mountain lions due to the lack of cover on the landscape to ambush deer, as observed for African lions in savanna habitats.13 All of these scenarios are likely to influence energy expenditure, indicating that a major disturbance, such as the wildfire in this study, could lead to energy deficits in carnivore populations.32
Our study was an opportunistic study of a population of mountain lions who were tracked before, during, and after a wildfire.

The limited number of individuals who were not impacted by the wildfire precluded a natural experiment (such as a BACI design), therefore we must consider the possibility of other factors that could have influenced the behavior of mountain lions in our system over the 30 months of the study. Variability in human activity is unlikely to have contributed to changes in mountain lion behavior because our study ended (March 2, 2020) prior to local and statewide restrictions on public movement due to COVID-19 in the state and county (beginning March 19, 2020). Over the study period, rainfall varied, with greater rainfall after the fire than before, and two and a half mule deer (Odocoileus hemionus) calving seasons (important periods for mountain lion hunting) occurred, with one and a half prior to the fire and one after the fire (Figure S2). We cannot rule out the possibility that fluctuations in, and interactions between, weather and mule deer abundance influenced mountain lion behavior during our study. However, it is unlikely that these variables resulted in the findings we report here. The greater rainfall after the fire would be expected to increase deer forage and subsequently decrease, rather than increase, mountain lion space use and therefore reduce road crossings.33,34 Further, given that mule deer tend to be crepuscular, the increase in daytime activity is unlikely to be explained by variability in environmental conditions changing deer abundance.35,36
Conservation implications
Our findings have important implications for the conservation of large carnivore populations living near urban areas, showing that wildfire can not only result in direct mortality, but could also influence carnivore behavior in ways that increase anthropogenic risks, like vehicular collisions and encounters with humans, as well as increase the risk of intraspecific conflict. These risks can interact.

For example, one subadult male in this study was hit and killed by a vehicle on a freeway immediately after an altercation with an uncollared adult male. Behavioral changes observed in this study (e.g., variable usage of burned areas, increased activity Figure 3. Predicted changes in risky behaviors by mountain lions after the 2018 Woolsey Fire, based on mixed effects models comparing probability of mountain lion use of urban areas
(A–D) Comparing probability of mountain lion use of urban areas (A), frequency of road crossings per month (B), proportion of day spent active (C), monthly distance traveled (D), and mean area of amount of space used over 3-month periods separated by sex and age class before and after the 2018 Woolsey Fire (E).

The periods before and after fire were defined by the 15 months prior to and following the Woolsey Fire.

Models used to predict relationships included a mixed effects logistic regression model (A), segmented linear mixed effects models (B) and (C), segmented mixed effects meta-regression (D), and a linear mixed effects model (E).

Error bars and bands show 95% confidence intervals around fitted relationships.

See also Table S2.
ll

OPEN ACCESS
Current Biology 32, 4762–4768, November 7, 2022 4765
Report
during the day, and increased distance traveled) could be indicative of increased hunting challenges or hunting flexibility. If the fire-transformed landscape reduces the ability of mountain lions to ambush deer, they might rely on other prey items, including smaller carnivores, which in turn put them at greater risk of poisoning from toxicants such as anticoagulant rodenticides.37 Greater risk-taking behaviors by carnivores living near urban areas could lead to increased mortality in populations already suffering from low genetic diversity, leading to increased extinction risk.38–40 As the world continues to urbanize and as we see increasing frequency of high severity fires in many of the world’s fire-prone landscapes,41 we are likely to see similar challenges for carnivore conservation in a broader range of global regions and taxa. Increasing the connectivity among urban habitat patches through a system of wildlife overpasses or underpasses,42 already known to be important for increasing genetic exchange, could be particularly critical in fire-prone areas when the quantity of already limited suitable habitat can be greatly reduced post-fire.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:
d KEY RESOURCES TABLE
d RESOURCE AVAILABILITY
B Lead contact
B Materials availability
B Data and code availability
d EXPERIMENTAL MODEL AND SUBJECT DETAILS
d METHOD DETAILS
B Study area
d QUANTIFICATION AND STATISTICAL ANALYSIS
B Study design
B Do mountain lions avoid burned areas after a large
wildfire?
B Do mountain lions increase behaviors that put them at anthropogenic risk after a large wildfire?
B Do mountain lions increase behaviors that could increase risk of conflict with conspecifics after a large wildfire?
B Resampling for model validation

SUPPLEMENTAL INFORMATION
Supplemental information can be found online at https://doi.org/10.1016/j. cub.2022.08.082.

A video abstract is available at https://doi.org/10.1016/j.cub.2022.08. 082#mmc3.

ACKNOWLEDGMENTS
We thank the many NPS biologists, technicians, and interns who assisted with fieldwork. We thank the La Kretz Center for California Conservation at UCLA, Figure 4. Observed overlap between the dominant adult male and subadult males before and after the 2018 Woolsey Fire
The dominant male (P30) is shown by a black line and subadult males are shown in colored points, different colors signify different individuals. Time periods include two 6-month periods before (8th May 2018–8th November 2018) and after (21st March 2019–10th September 2019) the 2018 Woolsey Fire.
(A–C) (A) indicates the period before the fire until the Woolsey fire, when P30 was dominant (8th May 2018–8th November 2018), and (B) shows a similar period of time ending with P30’s death (21st March 2019–10th September 2019). Before the fire, P30 regularly used the area within the fire perimeter and was rarely in the eastern half of the Santa Monica Mountains (A), whereas postfire, he occasionally moved through the burned area and largely relocated to the eastern end, overlapping extensively with multiple subadult males. (C) shows the mean (± SE) proportion of P30’s space use that overlaps with six other individual mountain lions (3 males and 3 females), tracked concurrently with him, before and after the fire. We defined P30 as the dominant male since he showed behaviors including territorial marking through scraping, breeding, and regular use of core natural areas.

Land use is shown by dark green (natural areas), light green (altered open areas), and gray (urban areas) and the extent of the Woolsey Fire (2018) is shown in white outline with white hatching. Primary and secondary roads are shown with gray lines with freeways in yellow (both were used in the road crossing analysis).

See also Figure S1.
ll

OPEN ACCESS
4766 Current Biology 32, 4762–4768, November 7, 2022
Report
and National Park Foundation for funding R.V.B.’s work on the project. We would also like to thank Prof. Youyi Fong for advice about mixed effects segmented regression, Prof. Wolfgang Viechtbauer for assistance with meta-analytic approaches, and Dr. Christen Fleming for assistance with continuous time movement models and extensions.

AUTHOR CONTRIBUTIONS
R.V.B., S.P.D.R, D.T.B, and J.A.S. conceived the ideas, J.A.S. and S.P.D.R collected the data, R.V.B. conducted the analysis, wrote the paper and made graphs. J.A.S, S.P.D.R., and D.T.B. edited and provided feedback on the manuscript throughout development.

DECLARATION OF INTERESTS
The authors declare no competing interests.

INCLUSION AND DIVERSITY
We worked to ensure sex balance in the selection of non-human subjects. One or more of the authors of this paper self-identifies as a member of the LGBTQ+ community. While citing references scientifically relevant for this work, we also actively worked to promote gender balance in our reference list.

Received: April 14, 2022
Revised: June 13, 2022
Accepted: August 31, 2022
Published: October 20, 2022

REFERENCES

1. Gehrt, S.D., Riley, S.P.D., and Cypher, B.L., eds. (2010). Urban Carnivores:

Ecology, Conflict and Conservation (The John Hopkins University Press).

2. Blecha, K.A., Boone, R.B., and Alldredge, M.W. (2018). Hunger mediates apex predator’s risk avoidance response in wildland–urban interface.

J. Anim. Ecol. 87, 609–622.

3. Nimmo, D.G., Avitabile, S., Banks, S.C., Bliege Bird, R., Callister, K., Clarke, M.F., Dickman, C.R., Doherty, T.S., Driscoll, D.A., Greenville, A.C., et al. (2019). Animal movements in fire-prone landscapes. Biol.

Rev. Camb. Philos. Soc. 94, 981–998.

4. US Department of Transportation (2013). New FHWA Report Reveals States with the Busiest Freeways. https://www.fhwa.dot.gov/pressroom/ fhwa1335.cfm.

5. Ripple, W.J., Estes, J.A., Beschta, R.L., Wilmers, C.C., Ritchie, E.G., Hebblewhite, M., Berger, J., Elmhagen, B., Letnic, M., Nelson, M.P., et al. (2014). Status and ecological effects of the world’s largest carnivores. Science 343, 1241484.

6. Rodrı´guez-Estival, J., and Mateo, R. (2019). Exposure to anthropogenic chemicals in wild carnivores: a silent conservation threat demanding long-term surveillance. Curr. Opin. Environ. Sci. Health 11, 21–25.

7. Jennings, M.K., Lewison, R.L., Vickers, T.W., and Boyce, W.M. (2016).

Puma response to the effects of fire and urbanization. J. Wildl. Manage. 80, 221–234.

8. Geary, W.L., Doherty, T.S., Nimmo, D.G., Tulloch, A.I.T., and Ritchie, E.G. (2020). Predator responses to fire: a global systematic review and metaanalysis. J. Anim. Ecol. 89, 955–971.

9. Doherty, T.S., Geary, W.L., Jolly, C.J., Macdonald, K.J., Miritis, V., Watchorn, D.J., Cherry, M.J., Conner, L.M., Gonza´ lez, T.M., Legge, S.M., et al. (2022). Fire as a driver and mediator of predator–prey interactions. Biol. Rev.

10. Cherry, M.J., Chandler, R.B., Garrison, E.P., Crawford, D.A., Kelly, B.D., Shindle, D.B., Godsea, K.G., Miller, K.V., and Conner, L.M. (2018).

Wildfire affects space use and movement of white-tailed deer in a tropical pyric landscape. For. Ecol. Manage. 409, 161–169.

11. Lino, S., Sillero, N., Torres, J., Santos, X., and A´ lvares, F. (2019). The role of fire on wolf distribution and breeding-site selection: Insights from a generalist carnivore occurring in a fire-prone landscape. Landsc. Urban Plan. 183, 111–121.

12. Vanbianchi, C.M., Murphy, M.A., and Hodges, K.E. (2017). Canada lynx use of burned areas: Conservation implications of changing fire regimes.

Ecol. Evol. 7, 2382–2394.

13. Eby, S., Mosser, A., Swanson, A., Packer, C., and Ritchie, M. (2013). The impact of burning on lion Panthera leo habitat choice in an African savanna. Curr. Zool. 59, 335–339.

14. Furnas, B.J., Goldstein, B.R., and Figura, P.J. (2022). Intermediate fire severity diversity promotes richness of forest carnivores in California.

Divers. Distrib. 28, 493–505.

15. Gigliotti, L.C., Curveira-Santos, G., Slotow, R., Sholto-Douglas, C., Swanepoel, L.H., and Jachowski, D.S. (2022). Community-level responses of African carnivores to prescribed burning. J. Appl. Ecol. 59, 251–262.

16. Monroy-Vilchis, O., Go´ mez, Y., Janczur, M., and Urios, V. (2009). Food Niche of Puma concolor in Central Mexico. Wildlife Biol. 15, 97–105.

17. Dees, C.S., Clark, J.D., and Manen, F.T.V. (2001). Florida Panther Habitat Use in Response to Prescribed Fire. J. Wildl. Manage. 65, 141–147.

18. Parkins, K., York, A., and Di Stefano, J. (2018). Edge effects in fire-prone landscapes: Ecological importance and implications for fauna. Ecol.

Evol. 8, 5937–5948.

19. Keeley, J.E. (2006). South Coast bioregion. In Fire in California’s Ecosystems, J.W. van Wagtendonk, N.G. Sugihara, S.L. Stephens, A.E.

Thode, K.E. Shaffer, and J.A. Fites-Kau, eds., pp. 350–390.

20. Tucker, M.A., Bo¨ hning-Gaese, K., Fagan, W.F., Fryxell, J.M., Van Moorter, B., Alberts, S.C., Ali, A.H., Allen, A.M., Attias, N., Avgar, T., et al. (2018).

Moving in the Anthropocene: Global reductions in terrestrial mammalian movements. Science 359, 466–469.

21. Gaynor, K.M., Hojnowski, C.E., Carter, N.H., and Brashares, J.S. (2018).

The influence of human disturbance on wildlife nocturnality. Science 360, 1232–1235.

22. Nisi, A.C., Benson, J.F., and Wilmers, C.C. (2022). Puma responses to unreliable human cues suggest an ecological trap in a fragmented landscape. Oikos, e09051.

23. Benson, J.F., Sikich, J.A., and Riley, S.P. (2020). Survival and competing mortality risks of mountain lions in a major metropolitan area. Biol.

Conserv. 241, 108294.

24. Smith, J.A., Gaynor, K.M., and Suraci, J.P. (2021). Mismatch Between Risk and Response May Amplify Lethal and Non-lethal Effects of Humans on Wild Animal Populations. Front. Ecol. Evol. 9, 1–13.

25. Sa
´ lek, M., Drahnı´kova´ , L., and Tkadlec, E. (2015). Changes in home range sizes and population densities of carnivore species along the natural to urban habitat gradient. Mamm Rev. 45, 1–14.

26. Riley, S.P.D., Serieys, L.E.K., Pollinger, J.P., Sikich, J.A., Dalbeck, L., Wayne, R.K., and Ernest, H.B. (2014). Individual behaviors dominate the dynamics of an urban mountain lion population isolated by roads. Curr.

Biol. 24, 1989–1994.

27. Riley, S.P.D., Sikich, J.A., and Benson, J.F. (2021). Big Cats in the Big City:

Spatial Ecology of Mountain Lions in Greater Los Angeles. J. Wildl.

Manage. 85, 1527–1542.

28. Crooks, K.R. (2010). Community ecology of urban carnivores. In Urban carnivores: Ecology, conflict, and conservation, S.D. Gehrt, S.P.D. Riley, and B.L. Cypher, eds. (The John Hopkins University Press), pp. 185–196.

29. Moss, W.E., Alldredge, M.W., and Pauli, J.N. (2016). Quantifying risk and resource use for a large carnivore in an expanding urban-wildland interface. J. Appl. Ecol. 53, 371–378.

30. Benson, J.F., Abernathy, H.N., Sikich, J.A., and Riley, S.P. (2021).

Mountain lions reduce movement, increase efficiency during the Covid-19 shutdown. Ecological Solutions and Evidence 2, e12093.

31. Wilmers, C.C., Nisi, A.C., and Ranc, N. (2021). COVID-19 suppression of human mobility releases mountain lions from a landscape of fear. Curr.

Biol. 31, 3952–3955.e3.
ll

OPEN ACCESS
Current Biology 32, 4762–4768, November 7, 2022 4767
Report

32. Houston, A.I., Prosser, E., and Sans, E. (2012). The cost of disturbance: A waste of time and energy? Oikos 121, 597–604.

33. Stoner, D.C., Sexton, J.O., Choate, D.M., Nagol, J., Bernales, H.H., Sims, S.A., Ironside, K.E., Longshore, K.M., and Edwards, T.C. (2018).

Climatically driven changes in primary production propagate through trophic levels. Glob. Chang. Biol. 24, 4453–4463.

34. Marshal, J.P., Krausman, P.R., and Bleich, V.C. (2005). Rainfall, temperature, and forage dynamics affect nutritional quality of desert mule deer forage. Rangel. Ecol. Manag. 58, 360–365.

35. Perez-Solano, L.A., and Gallina Tessaro, S. (2019). Activity patterns and  their relationship to the habitat use of mule deer (Odocoileus hemionus) in the Chihuahuan desert, Mexico. Therya 10, 323–328.

36. Relyea, R.A., and Demarais, S. (1994). Activity of desert mule deer during the breeding season. J. Mammal. 75, 940–949.

37. Riley, S.P.D., Bromley, C., Poppenga, R.H., Uzal, F.A., Whited, L., and Sauvajot, R.M. (2007). Anticoagulant Exposure and Notoedric Mange in Bobcats and Mountain Lions in Urban Southern California. J. Wildl.

Manage. 71, 1874–1884.

38. Gustafson, K.D., Gagne, R.B., Vickers, T.W., Riley, S.P.D., Wilmers, C.C., Bleich, V.C., Pierce, B.M., Kenyon, M., Drazenovich, T.L., Sikich, J.A., et al. (2019). Genetic source–sink dynamics among naturally structured and anthropogenically fragmented puma populations. Conserv. Genet. 20, 215–227.

39. Ruell, E.W., Riley, S.P.D., Douglas, M.R., Antolin, M.F., Pollinger, J.R., Tracey, J.A., Lyren, L.M., Boydston, E.E., Fisher, R.N., and Crooks, K.R. (2012). Urban habitat fragmentation and genetic population structure of bobcats in Coastal Southern California. Am. Midl. Nat. 168, 265–280.

40. Kimmig, S.E., Beninde, J., Brandt, M., Schleimer, A., Kramer-Schadt, S., Hofer, H., Bo¨ rner, K., Schulze, C., Wittstatt, U., Heddergott, M., et al. (2020). Beyond the landscape: Resistance modelling infers physical and behavioural gene flow barriers to a mobile carnivore across a metropolitan area. Mol. Ecol. 29, 466–484.

41. Bowman, D.M.J.S., Kolden, C.A., Abatzoglou, J.T., Johnston, F.H., van der Werf, G.R., and Flannigan, M. (2020). Vegetation fires in the Anthropocene. Nat. Rev. Earth Environ. 1, 500–515.

42. Gregory, A., Spence, E., Beier, P., and Garding, E. (2021). Toward Best Management Practices for Ecological Corridors. Land 10, 140.

43. Benson, J.F., Sikich, J.A., and Riley, S.P.D. (2016). Individual and population level resource selection patterns of mountain lions preying on mule deer along an urban-wildland gradient. PLoS One 11, e0158006.

44. Benson, J.F., Mahoney, P.J., Sikich, J.A., Serieys, L.E.K., Pollinger, J.P., Ernest, H.B., and Riley, S.P.D. (2016). Interactions between demography, genetics, and landscape connectivity increase extinction probability for a small population of large carnivores in a major metropolitan area. Proc.

Biol. Sci. 283, 20160957.

45. Gillespie, T.W., Ostermann-Kelm, S., Dong, C., Willis, K.S., Okin, G.S., and MacDonald, G.M. (2018). Monitoring changes of NDVI in protected areas of southern California. Ecol. Indic. 88, 485–494.

46. Viechtbauer, W. (2010). Conducting Meta-Analyses in R with the metafor Package. J. Stat. Softw. 36, 1–48.

47. Calenge, C. (2006). The package ‘‘adehabitat’’ for the R software: A tool for the analysis of space and habitat use by animals. Ecol. Modell. 197, 516–519.

48. R Development Core Team (2019). R: A language and environment for statistical computing. http://www.r-project.org/.

49. Signer, J., Fieberg, J., and Avgar, T. (2019). Animal movement tools (amt):

R package for managing tracking data and conducting habitat selection analyses. Ecol. Evol. 9, 880–890.

50. Senn, S., Gavini, F., Magrez, D., and Scheen, A. (2013). Issues in performing a network meta-analysis. Stat. Methods Med. Res. 22, 169–189.

51. SCAG (2005). Land-use dat for Ventura and Los Angeles counties (Southern California Association of Governments).

52. Bates, D., Machler, M., Bolker, B., and Walker, S. (2015). Fitting linear € mixed-effects models using lme4. J. Stat. Softw. 67, 1–48.

53. U.S. Census Bureau (2019). TIGER/Line Shapefiles (machine-readable data files). https://www.census.gov/geographies/mapping-files/ time-series/geo/tiger-line-file.html.

54. QGIS Development Team (2020). QGIS Geographic Information System v3.4.

55. Leos-Barajas, V., Photopoulou, T., Langrock, R., Patterson, T.A., Watanabe, Y.Y., Murgatroyd, M., and Papastamatiou, Y.P. (2017).

Analysis of animal accelerometer data using hidden Markov models.

Methods Ecol. Evol. 8, 161–173.

56. Hetem, R.S., Mitchell, D., De Witt, B.A., Fick, L.G., Maloney, S.K., Meyer, L.C.R., and Fuller, A. (2019). Body temperature, activity patterns and hunting in free-living cheetah: biologging reveals new insights. Integr. Zool. 14, 30–47.

57. McClintock, B.T., and Michelot, T. (2018). momentuHMM: R package for generalized hidden Markov models of animal movement. Methods Ecol.

Evol. 9, 1518–1530.

58. Calabrese, J.M., Fleming, C.H., and Gurarie, E. (2016). Ctmm: an R Package for Analyzing Animal Relocation Data As a Continuous-Time Stochastic Process. Methods Ecol. Evol. 7, 1124–1132.

59. Noonan, M.J., Fleming, C.H., Akre, T.S., Drescher-Lehman, J., Gurarie, E., Harrison, A.L., Kays, R., and Calabrese, J.M. (2019). Scale-insensitive estimation of speed and distance traveled from animal tracking data. Mov.

Ecol. 7, 35.

60. Getz, W.M., Fortmann-roe, S., Cross, P.C., Lyons, A.J., Ryan, S.J., and Wilmers, C.C. (2007). LoCoH : Nonparameteric Kernel Methods for Constructing Home Ranges and Utilization Distributions. PLoS One 2, e207.

61. Noonan, M.J., Tucker, M.A., Fleming, C.H., Akre, T.S., Alberts, S.C., Ali, A.H., Altmann, J., Antunes, P.C., Belant, J.L., Beyer, D., et al. (2019). A comprehensive analysis of autocorrelation and bias in home range estimation. Ecol. Monogr. 89, e01344.

62. Rstudio Team (2020). Rstudio: Integrated Development Environment for R.

63. Wickham, H. (2016). ggplot2: Elegant graphics for data analysis, 2nd ed. (Springer-Verlag).
ll

OPEN ACCESS
4768 Current Biology 32, 4762–4768, November 7, 2022
Report
STAR+METHODS

KEY RESOURCES TABLE

RESOURCE AVAILABILITY
Lead contact
Further information and requests for resources or reagents should be directed to and will be fulfilled by the lead contact, Rachel Blakey ([email protected]).

Materials availability
The study did not generate new unique reagents.

Data and code availability
The data and code generated during this study are available at Dryad: https://doi.org/10.5068/D1M97D.

EXPERIMENTAL MODEL AND SUBJECT DETAILS
We captured and tracked mountain lions using global positioning system (GPS) collars (Puma concolor) as part of a long-term study conducted by the National Park Service (2002–present).26,27,43 Mountain lions were captured using foot cable-restraints, baited cage-traps, or by treeing them with trained hounds; and immobilized with ketamine hydrochloride combined with medetomidine hydrochloride, administered intramuscularly. All animals were monitored throughout the time they were immobilized, during which time we estimated age, based on body size and tooth wear measurements. Age classes were: kittens (dependent offspring with their mother, 0-14 months), subadults (independent animals prior to reproduction: females 14-25 months, males 14-42 months), and adults (breeding animals: females >25 months, males >42 months).44 We fitted adult and subadult animals with Vectronic Aerospace GPS collars (Berlin, Germany; Vertex Plus and Vertex Lite models) equipped with VHF beacons. Animal capture and handling procedures were permitted through a scientific collecting permit with the California Department of Fish and Wildlife (SCP # 05636) and the National Park Service Institutional Animal Care and Use Committee (Protocol PWR_SAMO_Riley_Mt.Lion_2014.A3). For this study, we used locational and accelerometer data for 17 individual mountain lions, collected over a 2.5-year period between 2017 and 2020, encompassing a large wildfire event, the 2018 Woolsey Fire. Individuals tracked for the study included 9 females (5 adult, 2 subadults, and 2 subadults that became adults during the study period) and 8 males (2 adult, 1 subadult, 1 kitten, and 4 subadults that became adults during the study period). Age was calculated for each three-month period, and the male kitten was treated as a subadult for the purposes of the study, given that he was estimated to be close to subadult age ( 1 year old) and his mother was not observed during his capture.

We programmed collars to collect 8 locations per 24-hour period (7 at night, 1 during the day). The seven fixes at night were at 2 h intervals beginning at 5:00pm Pacific Standard Time (PST), while the day location was collected at 1:00pm PST. On average, 90% of programmed fixes for periods used in this study were successful, with individual mountain lion fix rates ranging from 69% to 98%.

Collars also collected activity data on two axes (X: anterior-posterior/surge, Y: lateral/sway), averaged across every 5 minute period. A third axis (Z: dorso-ventral/heave) was only available for two of the seventeen individuals, so these data were not used REAGENT or RESOURCE SOURCE IDENTIFIER
Deposited data
Data and code Dryad data repository Data and code are available at Dryad: https://doi.org/10.5068/D1M97D Software and algorithms
R v 3.6.1 R Development Core Team https://www.r-project.org/ Rstudio v. 1.3.1093 Rstudio Team https://www.rstudio.com/ adehabitatHR v. 0.4.16 The Comprehensive R Archive
Network (CRAN)
https://cran.r-project.org/web/packages/adehabitatHR
adehabitatLT v 0.3.25 CRAN https://cran.r-project.org/web/packages/adehabitatLT amt v. 0.1.4 CRAN https://cran.r-project.org/web/packages/amt ctmm v. 0.5.11 CRAN https://cran.r-project.org/web/packages/ctmm ggplot2 v. 3.3.0 CRAN https://cran.r-project.org/web/packages/ggplot2 lme4 v 1.1-23 CRAN https://cran.r-project.org/web/packages/lme4 metafor v. 2.4-0 CRAN https://cran.r-project.org/web/packages/metafor momentuHMM v. 1.5.1 CRAN https://cran.r-project.org/web/packages/momentuHMM QGIS v. 3.4 QGIS Development Team https://qgis.org
ll

OPEN ACCESS
Current Biology 32, 4762–4768.e1–e5, November 7, 2022 e1 Report
in the analysis. Accelerometer measurements were 99% successful on average, with all individuals recording > 96% of expected measurements.

METHOD DETAILS
Study area
We studied an urban population of mountain lions within Los Angeles and Ventura counties, California, in the Santa Monica Mountains and Simi Hills (3405’N, 11846’W) (Figure 2). All patches of natural habitat were bordered by major freeways, urbanization, agricultural development, or the Pacific Ocean. The study population in the Santa Monica Mountains, in particular, has been genetically isolated from nearby populations by roads and urbanization,26,38 leading to high extinction risk.44 Land-use was variable across the study area, and included federal, state, and local parklands, as well as urban areas consisting of high-density residential, commercial, and industrial areas, low-density rural or suburban residential areas, and agricultural areas. Natural vegetation in the study area consisted of mixed chaparral, coastal sage scrub, oak woodlands and savannas, riparian woodlands, and non-native annual grasslands.

The only wild, large ungulates were mule deer, which are the predominant prey for mountain lions in the region,43 and two-and-a-half mule deer (Odocoileus hemionus) calving seasons occurred during the study period (Figure S2). The climate of the study area was Mediterranean, with cool, wet winters and hot, dry summers. Rainfall varied over the study period, with greater rainfall after the fire than before.,. The area is prone to drought and wildfire,45 with two major wildfires occurring within less than a decade prior to this study, the Springs Fire in 2013, 9,814 ha, and the Woolsey Fire in 2018, 39,234 ha. The Woolsey Fire was the largest fire on record to have affected the Santa Monica Mountains and burned > 40% of the natural area in the Santa Monica Mountains and > 66% of the natural area in the Simi Hills (Figure 3).

QUANTIFICATION AND STATISTICAL ANALYSIS
Study design
We included locational data for 17 individual lions during 15 months leading up to and 15 months following the Woolsey Fire (2018).

Mountain lion tracking periods varied (Figure S3), and more individuals were tracked after the fire (F: 9; M: 6) compared to before the fire (F: 5; M: 7). We therefore used resampling methods that balanced numbers of individuals among age classes to validate our findings (Table S3).

Do mountain lions avoid burned areas after a large wildfire? To evaluate whether mountain lions decreased use of areas after they were burned in the Woolsey Fire, we compared selection coefficients for individual mountain lions derived from step selection functions before and after the fire using a meta-analytic approach.46 Individual mountain lions were excluded from this analysis if an adaptive Local Convex Hull (LoCoH), calculated from every location recorded during the study period (the period spanning 15 months before and after the focal fire), overlapped with the burned area from the focal fire by less than 10%, or if they were not tracked during both periods (both before and after fire). We used the adehabitatHR v0.4.16 package47 within the R v3.6.1 environment48 to fit LoCoH home ranges and used the maximum number of nearest neighbors as all those points which were within the maximum distance between any 2 points recorded for animals in this analysis.

We first fitted a separate step selection function to each individual mountain lion during the periods before and after the fire separately using the amt v 0.1.4 package.49 These functions compared observed ‘‘steps’’ (movements connecting successive locations) with random possible steps generated from distributions of turning angles and step lengths from the broader population. We used only night locations for the step selection analysis, defined as locations collected between one hour after sunset and one hour before sunrise. The observed and random (i.e., "available") steps were compared to estimate selection coefficients using a conditional logistic regression to match observed to related randomly selected steps as strata. We used a sample rate of 2 h with a tolerance of 1 h and generated 1000 random steps for each observed step. The high tolerance level was not necessary and unlikely to have influenced the analysis, given > 99.96 of steps were within ± 5 minutes of the 2 h interval. Steps were separated into ‘‘bursts’’ for each night, to ensure sample intervals were regular (2 h intervals between each step). We then calculated effect sizes (yi) representing the change in selection of areas within the fire perimeter before and after they were burned by subtracting the ‘‘before fire’’ coefficient (coefbefore) from the ‘‘after fire’’ coefficient (coefafter) for each individual. This meant that positive coefficients indicated selection for burned areas was higher after the fire, and negative values indicated that selection for burned areas was lower after the fire. We calculated the sampling standard error (sei) using the following approach recommended by Senn, Gavini, Magrez, & Scheen, 50 where sebefore and seafter are the standard errors of the selection coefficients before and after the fire for each individual and ri is the correlation between the coefficients before and after the fire. yi = coefafter
coefbefore
sei =
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi se2
after + se2before
ð2 3 ri 3 seafter 3 sebeforeÞ
q
ll

OPEN ACCESS
e2 Current Biology 32, 4762–4768.e1–e5, November 7, 2022 Report
Our sample size was small (5 males and 4 females tracked both before and after the fire), so we were chiefly interested in population-level selection for or against burned areas. We therefore estimated a population-level effect size using random effects metaanalysis46 using the metafor v. 2.4-0 package.46 Along with the step-selection analyses and for comparison with them, we calculated mountain lion use of areas within the burn perimeter before and after the fire as the number of point locations whose 10 m radius intersected with the burned area (to allow for some variability in GPS location and fire layer accuracy).

Do mountain lions increase behaviors that put them at anthropogenic risk after a large wildfire? We calculated three metrics associated with behaviors that may place mountain lions at additional risk from humans and anthropogenic threats: use of urban areas; number of road crossings; and proportion of daytime period active. We defined urban areas as commercial, and industrial areas and residential areas with R 2.5 houses/hectare identified within the Southern California Association of Governments land use map.51 This map was the most accurate available land-use data for the region, because later versions classified land uses at the parcel scale, rather than based on observed boundaries between different land uses. The dataset we used was reflective of the landscape throughout the study period from 2017–2020 for the broad development and alteredopen classifications that we used in these analyses. The geographic information system (GIS) program for the park monitors land use in and around SMMNRA as part of the National Park Service Inventory and Monitoring Program. We defined mountain lion use of urban areas before and after the fire as a binary variable where point locations whose 10 m radius intersected urban areas were recorded as used (1), and those locations whose buffer did not intersect with urban areas were unused (0). We compared use of urban areas before and after the fire using a mixed effects logistic regression with period (before and after fire) as a fixed effect and individual mountain lion as a random intercept using lme4 v 1.1-2352 (see Tables S3 and S5 for details of all analyses). We compared 3 models to investigate how the probability of mountain lion use of urban areas changed after the fire including: null (no effect of fire); step response (an abrupt change in urban use after the fire compared to before the fire); continuous response (a change in the relationship between urban use and time after the fire) (Table S5). We compared models using Akaike’s Information Criterion adjusted for small sample size (AICc) and identified the most parsimonious model as the model with the lowest AICc, that was separated from a less complex nested model by DAIC > 2. Modelled coefficients and fitted relationships are presented with 95% confidence intervals, and confidence intervals around the fixed effects were calculated for fitted relationships using parametric bootstrapping.

To quantify road crossing behavior, we first exported each month of locations for each mountain lion into a movement trajectory using the adehabitatLT v0.3.25 package.47 We classified a major road as all freeways and secondary roads using road data from the U.S. Census Bureau, (53), adding roads that had similar amount and speed of traffic based on observations by National Park Service biologists. Specific roads included are shown in Figure 4. We added a 50 m buffer (50 m either side) to each road, to allow for road width and spatial uncertainty in road and mountain lion datasets. Road crossings were identified manually as ‘‘minimum road crossings’’, using lines between two consecutive points that traversed any buffered road, using QGIS v. 3.4.54 When the line drawn between two consecutive point locations traversed a single road more than once, and the starting point was on one side of the road, whereas the ending point was on the other side, this was counted as one crossing. When the line drawn between two consecutive point locations traversed a road any number of times, but both starting and ending points were on the same side of the road, this was counted conservatively as zero crossings. As point locations were separated by a minimum of 2 h, we cannot discount the possibility of the animal taking an alternative (rather than the shortest) route to traverse between the two points. However, in all cases where we have recorded a crossing, the alternative route would have resulted in at least one road crossing, so our measure of ‘‘minimum road crossings’’ remains consistent with these possibilities.

We analyzed the relationship between road crossings and fire in a similar way to the urban use analyses (Tables S3 and S5). We used linear mixed effects models with the number of road crossings per individual per month as the response variable and individual mountain lion as a random intercept, and we used model selection to assess support for either an abrupt (step) response or a gradual (continuous) response to fire (Table S5). To account for unequal fix rates among months and individuals, we included fix rate (the number of locations recorded for an individual mountain lion during the month when road crossings were counted) as a fixed effect in all road crossings models.

To estimate the proportion of the daytime period spent active, we analyzed accelerometer data for lions where it was available (Figure S3 & Table S3). Given that we did not have field observations to inform our estimations of behavioral state, we used unsupervised Hidden Markov Models (HMMs) to estimate two states approximating ‘‘resting’’ and ‘‘active’’ behavior.55 The HMM method explicitly models temporal dependence which is inherent in accelerometer data and assumes that the observed acceleration data time series is driven by an unobserved (hidden) behavioral state process.55 We split the data into separate individuals.55 We fitted a 2-state HMM using two data streams (activity of the X and Y axes), for which we assumed Gaussian distributions. We estimated starting values for our two states by examining distributions of the two data streams. We also fitted HMMs considering time of day as a covariate (cosine(2*pi*(hour of day/24)) using starting values extracted from the simpler models. These models did not improve fit compared to the simpler models based on AICc, so we retained the simpler models. Prior to analysis we standardized activity measurements by dividing all values for separate individuals and collars by the maximum recorded value during the period the collar was worn by the animal, given collar tightness can affect acceleration values measured by the sensor.56 We fitted HMMs using the momentuHMM v1.5.1 package.57
Next, we separated daytime activity data, including all data collected from one hour after sunrise to one hour before sunset to avoid crepuscular periods.27 We removed 24 h periods from the dataset if they had < 95% of expected recordings. We then ll

OPEN ACCESS
Current Biology 32, 4762–4768.e1–e5, November 7, 2022 e3 Report
calculated the proportion of daytime active as the proportion of time that was classified as ‘‘active’’ using the HMM method. We used logit-transformed proportion of daytime active as the response variable in linear mixed effects models (LMM) with individual as a random intercept to account for variability in activity levels among individuals (Tables S3 and S5). Consistent with the urban use and road crossings analyses, we used model selection to assess support for either an abrupt (step) response or a gradual (continuous) response to fire (Table S5).

Do mountain lions increase behaviors that could increase risk of conflict with conspecifics after a large wildfire? We calculated three metrics to quantify behaviors that could place mountain lions at additional risk due to increased chance of conspecific interactions: distance travelled, amount of space used, and spatial overlap with other mountain lions.

We quantified distance travelled using a continuous time movement modelling (CTMM) approach.58 The continuous time approach aims to separate the sampling processes from the animal’s underlying movement processes by fitting a model accounting for the positional and velocity autocorrelation properties inherent in movement data, and then simulating multiple possible trajectories based on this model.59 We used model selection to fit a movement model to each monthly period for each individual mountain lion that best described the positional and velocity autocorrelation of the animal’s movement for that period. For 38 out of 257 individual-months analyzed, the movement showed no statistically significant evidence for velocity autocorrelation, so we were unable to estimate distance for these months. We estimated monthly distance travelled and variance of these estimates for the remaining 219 months. Given that the CTMM approach allows for estimation of uncertainty, we used a mixed effects meta-regression approach, fitted via restricted maximum likelihood, using estimated distance as the effect sizes and variance of distance as the sampling variances, with individual mountain lion as a random effect (Tables S3 and S5). Our estimated values of distance travelled were normally distributed around a mean of 330 ± 120 km (SD) per month. Moderators (covariates) were defined in the same way as fixed effects for the models of urban use, road crossings, and daytime activity (Table S5). We compared 3 models to investigate whether mountain lions changed their distance travelled after the fire including: null (no effect of fire on distance travelled); step response to fire (abrupt change in distance travelled after the fire); and continuous response (a change in the relationship between distance travelled and time after the fire) (Table S5). We fitted continuous time movement models and estimated distance travelled using the ctmm v 0.5.11 package.58 We quantified the amount of space used and estimated home range overlap using adaptive local convex hulls (LoCoH),60 implemented within the adehabitatHR v0.4.18. While we recognize that this method can underestimate the amount of space used and is sensitive to sampling rates,61 it performs well when animal movement is constrained by barriers like roads and urban areas,60 and our sampling rate was generally consistent among individuals. Since we were more interested in comparative space use (before and after fire), rather than absolute measurements of area, we believe this approach is robust.

We quantified the amount of space used by calculating the adaptive LoCoH for every individual mountain lion and every 3-month period which contained a R 75% fix rate (Table S3). We analyzed the relationship between amount of space used and fire using linear mixed effects models with individual mountain lion as a random intercept (Table S5). We used model selection to assess support for an abrupt (step) response to fire and did not investigate a gradual response to fire as space use was calculated for 3-month periods (Table S5). We also fitted models including the interaction between period (before and after fire) and age-sex class, given the known disparities between amount of space used across age-sex classes,27 though we interpret these results cautiously due to the low number of individuals in each group (Table S3).

We took two approaches to investigating changes in home range overlap before and after the fire. For the first approach, we focused on an adult male who held the largest territory within the Santa Monica mountains prior to the Woolsey Fire, P30, which we refer to as the ‘‘dominant male’’. We examined all animals that had the potential to overlap with P30 (individuals that used the Santa Monica Mountains area as part or all of their home range) and that were tracked at the same time as P30 for at least 3 months both before and after the fire. This resulted in a dataset of 6 mountain lions (3 males and 3 females), who were tracked for periods ranging from 5 to 11 months (both before and after fire) concurrently with P30. For space use calculations, we limited tracking periods to the same period of time before and after the fire for each individual. For each individual we calculated amount of space used over the period they were tracked concurrently with P30 using adaptive LoCoHs. We then calculated areal overlap of the LoCoH with the corresponding LoCoH for P30 during the same period. Given the small sample size (6 individuals with one measure of overlap per period for a total of 12 measures of overlap), we interpreted the results graphically rather than conducting a formal analysis. Our second approach to quantifying overlap involved calculating the overlap between every pair of mountain lions that were tracked during concurrent 3-month periods (Table S3). We restricted this to animals that used the same region (e.g., animals that exclusively used the Simi Hills portion of the study area were only compared to other animals that used this part of the study area). We analyzed the overlap data in the same way as the first overlap analysis, but separated into two datasets, one expressing overlap as proportions of female home ranges overlapped and the other expressing overlap as proportions of male home ranges overlapped. We fitted linear mixed effects models to each of those two datasets using pair category (male-male, male-female, female-female) as a fixed effect and overlap pair (pair of individual mountain lions for which overlap was calculated) as a random intercept (Table S5). Similar to the space use analysis, we used model selection to assess support for an abrupt (step) response to fire (Table S5).

All analyses were conducted within R v3.6.148 using Rstudio v. 1.3.1093, 62 all plots were made using ggplot2 v. 3.3.063 and all map figures were made using QGIS v. 3.4.54
ll

OPEN ACCESS
e4 Current Biology 32, 4762–4768.e1–e5, November 7, 2022 Report
Resampling for model validation
In order to account for the variability in sampling across individuals and age-sex classes, we resampled observations in each dataset 100 times to provide equal numbers of locations across sex and age classes and re-ran the model selection analysis. The specific approaches for each analysis are listed in Table S3. We recorded the percentage of iterations for which the most parsimonious models from the full dataset were selected as well as the proportion of models that resulted in fitted relationships in the same direction (e.g. greater or lower magnitude after compared to before fire) as the full-data model for all analyses. Where the majority of the relationships were in the same direction as the full dataset and the majority of iterations showed the same direction in relationships, we classified the relationships as robust. An additional validation step was performed for the urban use analysis. Given the female who used urban areas the most frequently (P75 — 15% of use was urban) was only sampled after the wildfire, we performed an additional check and removed her from the dataset and re-fit the models. We found that the strength and direction of the relationships were similar and that the same model type was found to be the most parsimonious, so we retained the full dataset.

Most of our analyses showed that the most parsimonious model and the direction of relationships were consistent across 100% of iterations, and we report only the exceptions below. In the analysis of urban use, models predicting abrupt changes were selected as the most parsimonious 76% of the time, with continuous responses to fire 24% of the time. In the road crossings analysis, 78% of model iterations showed an increase in road crossings after fire with 25% of models showing an abrupt change and 62% showing a continuous response. For the space-use analysis, direction of the relationships (increase in space use after fire) was consistent across 83 % of iterations. When space use was separated into sex and age classes, model selection was consistent across all iterations, but the consistency of relationship directions (increase or decrease after fire) varied among sex and age classes (adult male: 63 %, subadult male: 100 %, adult female: 81 %, subadult female: 92 %). ll

OPEN ACCESS
Current Biology 32, 4762–4768.e1–e5, November 7, 2022 e5 Report

Review
Social isolation, loneliness and health in old age: a scoping review Emilie Courtin and Martin Knapp
London School of Economics and Political Science, Personal Social Services Research Unit, London, UK Accepted for publication 21 October 2015
Correspondence
Emilie Courtin
London School of Economics and
Political Science
Personal Social Services Research Unit
Houghton Street, London WC2A 2AE,
UK
E-mail: [email protected]
What is known about this topic
• Social isolation and loneliness are
risk factors for poor mental and
physical health.
• They are particularly problematic
in old age due to reduced social
networks, decreasing economic
resources and changes in family
structures.

What this paper adds
• Loneliness is more frequently
researched than isolation;
depression and cardiovascular
health are the most researched
health outcomes.
• Still little is known about
interventions that would affect
Abstract
The health and well-being consequences of social isolation and loneliness in old age are increasingly being recognised. The purpose of this scoping review was to take stock of the available evidence and to highlight gaps and areas for future research. We searched nine databases for empirical papers investigating the impact of social isolation and/or loneliness on a range of health outcomes in old age.

Our search, conducted between July and September 2013 yielded 11,736 articles, of which 128 items from 15 countries were included in the scoping review. Papers were reviewed, with a focus on the definitions and measurements of the two concepts, associations and causal mechanisms, differences across population groups and interventions. The evidence is largely US-focused, and loneliness is more researched than social isolation. A recent trend is the investigation of the comparative effects of social isolation and loneliness. Depression and cardiovascular health are the most often researched outcomes, followed by well-being.

Almost all (but two) studies found a detrimental effect of isolation or loneliness on health. However, causal links and mechanisms are difficult to demonstrate, and further investigation is warranted. We found a paucity of research focusing on at-risk sub-groups and in the area of interventions. Future research should aim to better link the evidence on the risk factors for loneliness and social isolation and the evidence on their impact on health.

Keywords: health, loneliness, mental health, older people, scoping review, social isolation
loneliness and health, about causal
mechanisms or service use of
isolated or lonely older people.
• Future research should link the
evidence on risk factors for
loneliness and social isolation and
the evidence on their impact on
different health domains, with
longitudinal designs needed to
understand associations.

Introduction
An increasing number of older people are living alone and are at risk of being socially isolated (Victor et al. 2002, Savikko et al. 2005, Sundstrom€ et al. 2009). Social isolation has been identified as a risk factor for poor health, reduced well-being, mortality (e.g. Patterson and Veenstra 2010, Steptoe et al. 2013), depression (Heikkinen and Kauppinen 2004) and cognitive decline (Wilson et al. 2007). It has been argued that health risks associated with isolation and loneliness are equivalent to the well-established detrimental effects of smoking and obesity (Holt-Lunstad et al. 2010). Social isolation and loneliness are particularly problematic in old age due to decreasing economic and social resources, functional limitations, the death of relatives and spouses, and changes in family structures and mobility.

New research in the area is accumulating. We therefore sought to review the evidence on social isolation and loneliness, and their impacts on health, well-being and service use.
© 2015 John Wiley & Sons Ltd 799
Health and Social Care in the Community (2017) 25(3), 799–812 doi: 10.1111/hsc.12311 Our review addresses two research questions: (i)
What evidence exists on the relationships between isolation, loneliness and health? (ii) What are the limitations and gaps in the evidence base? We address these two questions by scoping the literature. We focus on
recent findings about the associations between social
isolation and poorer health and well-being outcomes,
differential effects across population groups, as well
as the methodological challenges associated with the
design and evaluation of interventions aimed at
reducing or addressing the consequences of isolation.

Methods
We followed the five-stage methodological framework for scoping studies suggested by Arksey and O’Malley (2005): (i) identify the research question; (ii) identify relevant studies; (iii) select studies; (iv) chart the data; and (v) collate, summarise and report the
results. The stages of the review are detailed below,
with steps 4 and 5 conflated together.

Identifying the research question
As stated above, this review is guided by two research
questions: (i) What evidence exists on the relationships between isolation, loneliness and health? (ii) What are the limitations and gaps in the evidence base? Definitions of both social isolation and loneliness have been debated (Nummela et al. 2011, Giuli et al. 2012). For
example, social isolation is often defined as the lack of integration of individuals in their social environment.

However, recent research has distinguished specific
components of isolation in old age (e.g. quality of relationships), with implications for measurement of social isolation (Cornwell and Waite 2009). We did not predefine social isolation or loneliness as we wanted to compare the definitions and associated measurements in the literature as part of our review. Given our purpose, a broad range of health outcomes was also included.

Identifying relevant studies
The scoping review identified, retrieved and evaluated information from empirical peer-reviewed articles that examined the impacts of isolation and/or loneliness on physical and mental health. We focused
on studies published between 2000 and 2013. This
timeframe was selected as it closely parallels that of
recently published reviews of interventions aimed at
reducing loneliness and isolation in old age (e.g.

Hagan et al. 2014). Our study differs from previous
reviews in that we focus on the impact of isolation
and loneliness on health in old age in particular.

We searched nine databases between July and
September 2013 (PubMed, SCOPUS, PsycINFO, Medline, International Bibliography of Social Sciences, Public Affairs Information Service, EconLit, and from
the Thomson Reuters Web of Knowledge platform,
Web of Science and Current contents connect).

We used a combination of search terms related to
our population group of interest, social isolation and
loneliness, and a broad range of physical and mental
health outcomes (Table 1).

Selecting studies
Only empirical papers with an English language
abstract were included. Because the aim of the scoping
was to describe the breadth of relevant research, we did not exclude studies based on the same sample of respondents. We considered all types of research design.

We applied the following exclusion criteria at two
stages of study selection (screening by title and
abstract, and full text):
• Studies focusing on countries other than Western
Europe and USA;
• Studies focusing on issues other than social isolation and loneliness; • Studies not assessing physical or mental health
outcome(s);
• Editorials, letters, book reviews;
• Studies covering population groups other than older
people (defined here as people aged 50 and over).

Figure 1 summarises the selection process. The
search yielded 11,736 articles (11,392 through the nine databases and 344 through a selection of websites).

When duplicates were removed there were 5342 references, of which 94% were excluded based on screening of the title and abstract. Full texts of the remaining 288 papers were accessed. Of these papers, 156 articles
(54%) were excluded. One hundred and twenty-eight
articles were included in the scoping review.

Table 1 Keywords and search terms employed in the database searches
Population or
target group Issue Health outcome
Aged Loneliness Health
Ageing Isolation Physical health
Ageing Social isolation Mental health
Senior Solitude Mental health problems
Elderly Mental disorder
Elder people Well-being
Old age Well-being
Older people Depression
Old people
800 © 2015 John Wiley & Sons Ltd
E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Charting the data, summarising and reporting the
findings
We used the ‘narrative review’ approach to collect
similar information on all studies (Pawson 2002). We
recorded information on first author, year of publication, study objectives, type of data and research design, study setting, sample size, issue studied, measurement of isolation or loneliness, health outcome, population group and main findings.

Results
Study context
Research on the impact of social isolation and loneliness on health constitutes a large and growing body of literature, with 54% of the included articles published between 2010 and 2013 (Figure 2).

Table 2 presents the main characteristics of the 128
studies. The papers spanned 15 countries. Half focused
on the USA, followed by the UK and the Netherlands.

Loneliness and social isolation attract multidisciplinary attention. First authors’ disciplines included medicine, psychology, epidemiology, public health and nursing.

Study design
Half the studies collected primary data. The vast
majority used a quantitative approach, with only 5%
being qualitative and 3% using mixed methods. Over
half of the studies used samples representative of the
population.

Definitions and measurements
Just over half of the 128 studies included a formal
definition of isolation or loneliness, mostly the latter.

Figure 1 Flow chart of the search strategy and results. © 2015 John Wiley & Sons Ltd 801
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License First, the concept of loneliness was usually defined
as an undesirable subjective experience, related to
‘unfulfilled intimate and social needs’ (Peplau and
Perlman 1982). The notion was often considered as
unidimensional as only 23% of studies on loneliness
contrasted different dimensions, such as sense of
belonging or the nature of discrepancies between
experienced and expected relationships (e.g. Martina
and Stevens 2006); or between social and emotional
loneliness (e.g. Dong et al. 2012, Drageset et al. 2013a, b). We found a variety of measures (Table 3). Seventeen studies used a single-item question on loneliness.

More complex measures have also been introduced.

Most studies (47%) used the UCLA Loneliness scale
(Russell 1996) or its shorter revised version. Although most studies using this scale focused on the USA, it
was also used by the studies based on the English
Longitudinal Survey of Ageing (e.g. Shankar et al.
2011). Seven studies used the de Jong Gierveld scale
(1987) – specifically designed to measure loneliness in old age.

Second, in the area of social isolation, most papers
defined the notion as a unidimensional concept, i.e.
an objective measure of the number of contacts with
family and friends. Only a few studies considered
isolation as a multidimensional concept, adding the
notion of quality of relationships (e.g. Cornwell and
Waite 2009, Ha and Ingersoll-Dayton 2011). Measurement of social isolation was heterogeneous. Most studies used an ad hoc index composed of measures
of marital status, household composition and number
of friends and relatives (e.g. Coyle and Dugan 2012,
Stafford et al. 2013), and counts of meetings with relatives (e.g. Tilvis et al. 2012). A few studies included a previously developed scale such as the Berkman–
Syme Social Network Index (Michael et al. 2001, Eng
et al. 2002, Rodriguez et al. 2011) or the Duke Social
Support Index (Hastings et al. 2008, Parsons et al.
2013).

Focus of the studies
Half the studies included in the scoping review
aimed at describing the association between isolation
or loneliness and a health outcome. A quarter investiFigure 2 Number of studies included in the scoping review, by year of publication.

Table 2 Overview of the studies characteristics
Variable Number of studies
Percentage
of studies
Study context
Country of the first author*
USA 65 51
UK 10 8
Netherlands 9 7
Discipline of the first author†
Medicine 24 18
Psychology 19 15
Public health 15 12
Epidemiology 15 12
Nursing 13 10
Study setting
Community-based 114 90
Facility-based 14 10
Sample size
Minimum 6
Median 430
Maximum 44,573
Type of data
Primary data 65 51
Secondary data 63 49
Study design
Case–control 1 1
Controlled before and after 2 2
Cross-sectional 68 52
Longitudinal 42 33
Mixed methods 4 3
Other 1 1
Qualitative 6 5
Randomised control trial 4 3
*Only the first three countries with the highest number of publications are reported here. †
Only the first five disciplines with the highest number of publications are reported. 802 © 2015 John Wiley & Sons Ltd
E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License gated at-risk population groups and 15% looked at
the mediating effects of isolation or loneliness
between stress and health (Lefranc
ois et al. 2000, Paul et al. 2006, Aanes et al. 2010), between depression and health (Bisschop et al. 2004, Wang et al. 2006), or
between alcohol consumption and all-cause mortality
(Greenfield et al. 2002). Describing or evaluating an
intervention was the aim of 7% of studies. The level
of health service use of isolated older people was
investigated in 2% of papers. A similar proportion of
papers described the health of isolated older people.

In terms of the issue studied, 53% of the 128 papers
focused on loneliness and 21% on social isolation.

Associations with health outcomes
A wide range of health outcomes was examined
(Table 4). Overall, we found a balance between mental
health and physical health. Across the 128 studies, only two did not find a negative association between social
isolation or loneliness and health (Wattanakit et al.
2005, Wilby 2011). It should be noted that a number of
studies looked at the impact of isolation and loneliness on biomarkers such as cortisol levels or C-reactive protein (a marker of systemic infection) (10 studies in total; 80% published between 2010 and 2013).

The most commonly studied outcome was depression, followed by cardiovascular health. We report here findings for these two outcomes only (Table 5).

We chose this focus because depression and cardiovascular are major contributors to the burden of disease in old age (Marengoni et al. 2008).

In our sample, 75% of studies on depression
looked at loneliness, whereas 72% of studies on cardiovascular health investigated social isolation.

Important variations were also found in terms of
study design. Only 25% of papers investigating
depression used longitudinal data, compared to
almost half of the studies focused on cardiovascular
health. Further details of these studies are provided
in Appendix S1.

Out of the 32 papers on depression, 25 looked at
its association with loneliness. A difficulty is that the two concepts are overlapping, and loneliness may be
a symptom of depression. However, recent literature
has found depression and loneliness to be separate
entities (Stek et al. 2005). The evidence reviewed
clearly shows that loneliness is an independent risk
factor for depression in old age (Alpass and Neville
2003, Adams et al. 2004, Paul et al. 2006, Theeke et al. 2012). Longitudinal research has confirmed these
findings: loneliness is an independent risk factor for
depression, controlling for a number of covariates
such as demographic characteristics, marital status,
social isolation and psychosocial risk factors
(Cacioppo et al. 2010). Gender differences were also
consistently reported. For instance, the detrimental
effect of living alone on depression was more often
due to loneliness for men than for women (Park et al.
2013).

Table 3 Most commonly used measures of loneliness
Measure Authors Description Examples of studies using this measure UCLA Loneliness
scale or its revised
version
Russell
(1996)
Twenty-item scale, with
each item rated from 1
(never) to 4 (often). A
number of studies use a
shorter revised version
(R-UCLA)
Beeson (2003); Cacioppo et al. (2002) Cacioppo et al.
(2006, 2010); Grov et al. (2010); Hackett et al. (2012); Hawkley et al. (2006, 2010a,b); Kahlbaugh et al. (2011); Krause-Parello (2008, 2012); Luo et al. (2012); Nezlek et al. (2002); Ong et al. (2012); Perissinotto et al. (2012); Poulin et al. (2012); Routasalo et al. (2009); Shankar et al.
(2011, 2013); Sorkin et al. (2002); Steptoe et al. (2004, 2013); Theeke et al. (2012); Theeke and Mallow (2013);
VanderWeele et al. (2011, 2012); Zebhauser et al. (2013); Fessman and Lester (2000); Adams et al. (2004); Norman
et al. (2013)
De Jong Gierveld
scale
de Jong
Gierveld
(1987)
Eleven-item scale,
combining emotional and
social loneliness
Alma et al. (2011); Jongenelis et al. (2004); La Grow et al. (2012); Martina and Stevens (2006); Newall et al. (2013); Wilson et al. (2007); Han and Richardson (2010)
Single item Item from the CES-D
scale of depressive
symptoms
Item on the frequency of
feelings of loneliness
Ayalon and Shiovitz-Ezra (2011); Beeson et al. (2000);
O’Luanaigh et al. (2011, 2012); Tiikkainen and Heikkinen (2005); Kvaal et al. (2013); Theeke (2010b).

Conroy et al. (2010); Holmen and Furukawa (2002); Losada et al. (2012); Nummela et al. (2011); Park et al. (2013); Patterson and Veenstra (2010); Paul et al. (2006); Stephens et al. (2010, 2011); Tilvis et al. (2011); Holwerda et al. (2012) Adapted from O’Luanaigh and Lawlor (2008).
© 2015 John Wiley & Sons Ltd 803
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Only three studies investigated the impact of
social isolation on depression. Sjoberg et al. (2013)
found interesting differences between two Swedish
cohorts of older people: frequency and perception of
social contacts were related to depressive symptoms
in the first cohort but not in the second. A mixedmethod study of chronic depression in older British Pakistani women found that the persistence of
depression was partly explained by social isolation
(Gask et al. 2011). Interestingly, one of the two studies which did not find an association between isolation and health focused on depression: Wilby (2011) found that depressed older people were not socially
isolated but were on the contrary more likely to
report contacts than non-depressed respondents. To
explain these findings, the authors emphasised the
need to better understand the quality and meaning of
different types of social relations in old age.

Cardiovascular health was the focus of 13% of the
studies included in the review. Social isolation has
been consistently found to be associated with coronary artery disease (Brummett et al. 2001), chronic heart failure (Friedmann et al. 2006), congestive heart failure (Murberg 2004) and hospitalisation due to
heart failure (Cene et al. 2012). In addition, two studies researched the impact of loneliness on cardiovascular health. Both reported that loneliness was associated with cardiovascular risk factors (Sorkin
et al. 2002, Kamiya et al. 2010). Differences across
population groups have not been well-researched to
date, but Wang et al. (2005, 2006) did find an association between social isolation and the progression of coronary artery disease, specifically among women.

Mechanisms and causal links
Potential mechanisms between isolation, loneliness
and health were the focus of 15% of the studies.

These depend on the outcome being studied, but a
Table 4 Health outcomes studied*
Outcome
Number
of studies
Percentage
of studies
Depression 32 25
Cardiovascular health 15 13
Quality of life and well-being 15 13
General health and physical function 11 9
Biological measures 10 8
Health and mental health 9 7
Mortality 6 4
Cognitive function 5 4
Mental health 4 3
Dementia 3 3
Disability 2 2
Stress-related reactions 2 2
Substance abuse 2 2
Anxiety 1 1
Passive death wishes 1 1
Physiological processes 1 1
Diabetes 1 1
Unspecified 1 1
Total 121 100
*Seven studies on service use and at-risk groups are excluded from this table.

Table 5 Overview of the characteristics of the studies focused on depression and cardiovascular health
Number
of studies
Percentage
of studies
Depression
Study setting
Community-based 27 84
Facility-based 5 16
Study design
Cross-sectional 18 57
Longitudinal 8 25
Mixed methods 2 6
Other 2 6
Qualitative 1 3
Randomised control trial 1 3
Focus
Loneliness 26 75
Social isolation 3 16
Other 2 9
Outcome measure
CES-D scale or its
revised version
20 62
Geriatric depression scale 8 26
Other scales* 2 6
DSM-IV medical diagnosis 2 6
Cardiovascular health
Study setting
Community-based 14 94
Facility-based 1 6
Study design
Cross-sectional 7 47
Longitudinal 7 47
Mixed methods 1 6
Focus
Loneliness 2 14
Social isolation 11 72
Other 2 14
Outcome measure
Cardiovascular and
urinary measures†
9 60
Medical diagnosis 4 27
Medical records‡ 2 13
*Other depression scales found in the reviewed literature include for instance the Hamilton Rating Scale for Depression, the Psychiatric Symptom Index or the Zung Depression Scale. †
Including the measurement of hypertension and cardiovascular activity, heart failure survival score, cardiac index, arteriography and angiography.
‡
Including reason for hospitalisation and cause of death. 804 © 2015 John Wiley & Sons Ltd
E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License number of studies have suggested that health behaviours (Cacioppo et al. 2002, Eng et al. 2002), poorer sleep quality (Cacioppo et al. 2002) and vital exhaustion (Eng et al. 2002) were potential mediators between loneliness and a range of physical health
outcomes. Other studies have investigated the role of
social mechanisms such as perceived togetherness
(Tiikkainen and Heikkinen 2005).

A third of the studies investigated causal links
between isolation, loneliness and health. The use of
longitudinal data enabled the issue of reverse causality between loneliness, isolation and health to be (partially) addressed. The evidence is mixed. Green
et al. (2008) found a cross-sectional association
between social networks and cognition and functional
status in old age, but not a longitudinal association.

Out of 42 studies which used panel data, 23 provided
a basis for inference about causal links by using, for
instance, a cross-lagged panel analysis (e.g. Cacioppo
et al. 2010, Hawkley et al. 2010b, Luo et al. 2012) or
by adjusting for health status at baseline (e.g. Michael et al. 2001, Eng et al. 2002, Bisschop et al. 2003, Patterson and Veenstra 2010, Perissinotto, Stijacic Cenzer and Covinsky 2010, Ayalon and Shiovitz-Ezra 2011,
Nummela et al. 2011, Holwerda et al. 2012, Udell
et al. 2012, Shankar et al. 2013).

The issue of reverse causality was particularly salient in the case of depression. Out of the 32 studies which looked at depression, eight used longitudinal
data and implemented appropriate analytical strategies to infer causal links. Again, the findings were mixed. Luo et al. (2012) found that loneliness both
affected and was affected by depression and functional
limitations over time. Cacioppo et al. (2006) also
demonstrated that loneliness and depressive symptoms in old age have strong reciprocal impact. A similar effect has been found between loneliness and subjective well-being in old age (VanderWeele et al.
2012). In contrast, cross-lagged panel analysis showed
that loneliness predicted subsequent changes in
depression but not vice versa (Cacioppo et al. 2010).

Life course approaches also provided interesting
evidence: the detrimental impacts of early trauma on
pulse pressure were partially dependent on the level
of perceived social isolation in old age, as older
adults with low isolation levels did not display a significant association between early trauma and the health outcome (Norman et al. 2013).

Differential impact of isolation and loneliness
Eleven studies focused on the differential impact of
isolation and loneliness, with 75% published since
2010. Among these studies, 54% used longitudinal
data. Although no clear pattern could be discerned
due to the small number of studies, mortality and
biological processes appeared to be the most commonly studied health outcomes in this area.

To date, results have been mixed. Tilvis et al.
(2011, 2012) found that groups of older people who
are isolated or lonely only partially overlap and that
only loneliness (and not social isolation) was an independent mortality risk factor in old age. Similar findings were reported in the Netherlands: feelings of loneliness rather than social isolation were found to
be a major risk factor for increased mortality in older men (Holwerda et al. 2012). Steptoe et al. (2013)
reached the opposite conclusion using UK data, as
the effects of loneliness in their study on mortality
were not independent of the demographic characteristics and health status of the respondents, contrary to the effects of social isolation. Other studies have
found that both isolation and loneliness were independent risk factors for a range of health outcomes (Shankar et al. 2011, 2013, Coyle and Dugan 2012).

Interventions
Associations between social isolation, loneliness and
health have received relatively little attention in the intervention literature, and the results of studies were quite modest. There have been studies looking at
other outcomes, including a reduction in loneliness
itself. We found only nine studies (7%) that evaluated
interventions.

The first type of intervention covered was
befriending initiatives. One of these programmes –
focused on older women – reported success in attracting lonely older people but not in improving the well-being of participants (Martina and Stevens 2006).

Different results were found for a club targeting men
in a care home, as participants reported a significant
reduction in their depression and anxiety levels
(Gleibs et al. 2011). A randomised control trial looking at the effects of psychosocial group rehabilitation on social functioning, loneliness and well-being of
older people also had mixed results. Routasalo et al.
(2009) reported that a large proportion of participants had found new friends via the programme and that
their well-being levels increased significantly. However, their loneliness scores were not affected by taking part in the programme, suggesting that there are other mechanisms at play. Similar results were found
for group activities. A randomised control trial of a
model of restorative home care on physical health
and social support showed significant improvements
in physical function but no changes in perceived
levels of social support (Parsons et al. 2013). Kahl-
© 2015 John Wiley & Sons Ltd 805
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License baugh et al. (2011) measured the effects of playing
console games on physical activities, loneliness and
mood: older people who engaged in games reported
lower levels of loneliness, but no difference was
found with the control group in terms of life satisfaction or physical activity.

Another type of intervention was professionallyled support for isolated carers. Telephone-based support for female carers of people with dementia was found to be associated with lower isolation and
depression after 6 months for older carers (Winter
and Gitlin 2006).

Finally, a number of promising trials were underway at the time we conducted this review. The Senior Connection programme fosters peer companionship
for older adults, with the aim of reducing suicide risk in later life: preliminary results from a randomised
control trial suggest that socially disconnected older
adults were at considerably higher risk of suicide
(Van Orden et al. 2013).

Service use
We found only three studies that focused on the level
of health service use of isolated or lonely older people, only one using longitudinal data. Social isolation was found by Mistry et al. (2001) to predict re-hospitalisation among isolated older American veterans.

Burr and Lee (2013) examined the association
between social relationships and dental care service
use among older adults: older people who exhibit
loneliness and are under financial strain were less
likely to visit a dentist. Finally, gender differences
were reported. Lower levels of isolation and a
supportive environment were predictive of receiving
preventive home visits for older Danish women but
not for their male counterparts (Avlund et al. 2008).

At-risk groups
One third of the studies included in the scoping
review explicitly explored differences across population groups. As reported previously, considerable gender differences are found in terms of the association between isolation, loneliness and health. Twothirds of the papers which looked at differences across population groups focused on gender. The
findings were mixed. For example, Zebhauser et al.
(2013) found that, although levels of loneliness were
equally distributed among men and women in their
study, loneliness had a detrimental impact only on
the mental health of men. In contrast, another study
found that women were more sensitive to the impact
of loneliness on biological responses (Hackett et al.
2012). A potential explanation is that men and
women experience different types of loneliness, with
different impacts on their physical and mental health
(Nummela et al. 2011).

Differential effects across age groups were the
focus of 16% of the studies. For instance, Ayalon and
Shiovitz-Ezra (2011) found that loneliness is a major
risk factor for passive death wishes for people aged
50 and over, but that the effect was not noticeable for respondents aged over 75.

In recent years, differences by socioeconomic or
ethnic backgrounds have also started to be investigated. For example, a US study found that Hispanic respondents who were socially isolated had a greater
risk of increased left ventricular mass compared to
isolated older people from other ethnic backgrounds
(Rodriguez et al. 2011). Poulin et al. (2012) also
showed that the association between perceived support and depression was stronger for elderly American people than for elderly Chinese people.

Finally, a number of studies focused on groups at
higher risk of isolation and loneliness and of associated negative health outcomes, including older people who are cancer survivors (Jaremka et al. 2013), unpaid carers (Jaremka et al. 2013), and substance
users (Smith and Rosen 2009) who are HIV-positive
(Grov et al. 2010) or who have a history of institutionalisation (Smith and Hirdes 2009).

Discussion
We set out to describe the available literature on the
relationships between social isolation, loneliness and
health outcomes. The research evidence on these
associations has significantly expanded since 2000.

We found that the majority of the available evidence
comes from the USA and has focused on loneliness.

It should be noted that a growing number of studies
are concerned with the differential impacts of loneliness and isolation on health. The most researched outcomes are depression and cardiovascular health.

Almost all studies included in our review found that
social isolation and loneliness have detrimental effects on physical and mental health in old age. Although
limited in number, longitudinal designs have allowed
researchers to investigate potential mechanisms and
causal pathways.

Our review highlights a number of gaps in the evidence base. First, a lack of consistency in the definition and measurement of isolation and loneliness considerably limits the comparisons that can be made between studies, and hence the broader conclusions that can be
drawn. Isolation and loneliness are multidisciplinary
concepts and to date there is no agreement across disci806 © 2015 John Wiley & Sons Ltd E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License plines as to the best way to define or measure them.

Loneliness and social isolation are also linked to other concepts such as the availability of social support or
lack thereof (e.g. social capital or social network). As noted by Valtorta and Hanratty (2010), this has contributed to the richness of the research findings but potentially limits their usefulness to policy makers and practitioners. Differences in measurements are also
problematic. Measures of loneliness, for instance, range from single-item questions to 20-item scales designed
to measure different dimensions. Although previous
studies have shown that a single-item measure correlates strongly with more sophisticated scales (Victor et al. 2005), recent research has emphasised the importance of contrasting different dimensions of loneliness (Coyle and Dugan 2012). The UCLA and de Jong Gierveld scales are the most commonly used measures of loneliness. A recent assessment of both scales has
shown the relative superiority of the de Jong Gierveld
scale for the study of middle-aged and older adults
(Penning et al. 2013).

There are similar debates in relation to social isolation (Cornwell and Waite 2009). Fiori et al. (2006) stress the importance of combining structural (e.g. number of
contacts) and functional (e.g. type and quality of support received) aspects of social isolation. There is at least one interesting way forward. As noted by O’Luanaigh and Lawlor (2008), integrating research on the drivers of loneliness and isolation with research on
their impacts on health would allow researchers to
understand better which dimensions are crucial to
include in their studies. Indeed, very broad and general measures that fail to distinguish between isolation, feelings of loneliness and their different dimensions
may not be able fully to detect the impacts on physical and mental health of older adults (Coyle and Dugan
2012), and therefore could ultimately hold back development of effective interventions.

Second, only a third of the studies included in the
review used a longitudinal design. The fact that most
studies are cross-sectional means that still relatively little is known about mechanisms and causal links.

More longitudinal studies are needed to disentangle
the independent or interacting effects of loneliness
and isolation on health. A better understanding of
these mechanisms is crucial for designing appropriate
interventions. Indeed, results from longitudinal studies have shown that the experiences of loneliness and isolation are not uniform across the life course. On
the contrary, older people may become lonelier or
more isolated, be chronically isolated or become so
because of trigger events such as retirement or
bereavement (Ha and Ingersoll-Dayton 2011, Bekhet
and Zauszniewski 2012). Studies adopting a life
course approach are also needed as they can provide
very useful insights on potential triggers of loneliness (Savikko et al. 2005), which have to be considered for
the design of interventions.

Health and social care service use by isolated older
people is also under-researched. Available studies
focusing on other population groups provide helpful
insights for future research. A Dutch study found that
general (medical) practitioners acknowledged the
importance of patients’ feelings of loneliness in their daily practice but that they had difficulty responding
to these feelings and faced a lack of therapeutic
options. The authors suggest that a distinction
between chronic and transitory loneliness is helpful
for general practitioners (van der Zwet et al. 2009).

Another study suggests that more isolated people in
the USA have lower access to adequate health information (Askeslon et al. 2011), while a Canadian study argued that front-line health professionals such as
nurses have a role to play (Wilson et al. 2011).

Our review also identified a paucity of research
on population sub-groups, despite evidence of ethnic
and socioeconomic differences in the impact of loneliness and isolation on health. It should also be noted that the available evidence focuses almost exclusively
on individual-level analyses. We suggest that to
understand the scope and magnitude of the impact of
loneliness and isolation on health, future research
should further take into account ecological factors
such as the characteristics of communities and neighbourhoods where older individuals live. ‘Ageing-inplace’ policies have become a key component of European strategies for older people, on the grounds
of both concerns about public expenditure as the
population ages and the desire of older people to
remain autonomous (Means 2007, Wiles et al. 2012).

In that context, thinking about isolation and loneliness together with access to transport, health, community and local social services is crucial.

Finally, we identified little published work on
interventions which have been evaluated for their
impacts on health outcomes, clearly indicating a gap
for evidence-based practice. Since 1984, nine reviews
of the loneliness intervention literature have been
published (Rook 1984, McWhirter 1990, Cattan &
White 1998, Findlay 2003, Cattan et al. 2005, Perese
and Wolf 2005, Hawkley and Cacioppo 2010, Theeke
2010a, Hagan et al. 2014). Most of these reviews consider possible ways to reduce loneliness and isolation, and only one focuses explicitly on the impact on
health outcomes (Cattan et al. 2005). It is not easy to draw out consistent messages on the most successful
methods from these reviews. According to Cattan
et al. (2005), only interventions with an educational
© 2015 John Wiley & Sons Ltd 807
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License element succeeded in reducing loneliness among
older people, whereas one-to-one interventions were
less successful. In contrast, Findlay (2003) found a
positive effect only for one-to-one interventions. More recently, Hagan et al. (2014) have identified 17 studies relating to loneliness interventions. Out of these, four studies (three on new technologies such as web-based
communications, one on group interventions) identified significant reductions in loneliness.

It was equally difficult to draw a consistent message from the evaluations considered in the present review. This is likely to be due to the difficulty of
adequately measuring concepts as complex as social
isolation and loneliness in old age. As noted above,
the lack of exploration of the societal and policy contexts within which these interventions are operating also renders it difficult to draw firm conclusions
about these interventions.

As the findings of further longitudinal research
are becoming available, potential causal mechanisms
have to be considered in the design of these interventions. A clear conceptual model of loneliness or isolation is also needed as different target populations will potentially respond differently to interventions
aimed at reducing social or emotional loneliness. As
noted by Hawkley and Cacioppo (2010), a crucial
question is whether modifying the feeling of loneliness can have an impact on health. To date, the available evidence is insufficient to provide an answer, and this should be a priority for future research.

Our scoping review has both strengths and weaknesses. It covered a broad range of health outcomes and included both loneliness and isolation as risk factors for poor health in old age. However, it has two main limitations, with implications for the scope of
the evidence covered. First, we only included studies
with English language abstracts and we limited our
search to western countries. A number of relevant
articles focused on South Asian countries, for example, have been excluded from this review as the limited research available points to important cultural differences in the meaning and experience of loneliness and isolation (Rokach et al. 2001). Also, we did not include ‘grey literature’ or doctoral theses. Second, we conducted a scoping review and not a systematic literature review. In that sense, we did not assess or exclude papers based on their quality.

Conclusion
Social isolation and loneliness are common among
older people and can be both negatively associated
with mental and physical health, although still relatively little is known about causal links. Our scoping review has shown that a sizeable body of literature,
coming from a variety of disciplines and making use
of a range of methods, is focused on assessing the
associations between isolation, loneliness and health.

An important limitation to the development of the
evidence base for both researchers and policy makers
is the diversity of definitions and measurements. One
way forward would be to pool the evidence from the
literature on the drivers of isolation and loneliness in old age and the research on their impact on health,
so that important domains and dimensions are measured. There is also a challenge in relation to interventions, as data are still very limited. To date, the evidence is not sufficient to determine whether modifying social isolation levels or feelings of loneliness will have an impact on subsequent health. Better
understanding of the causal pathways through which
loneliness and isolation affect health is needed to
inform the development of appropriate interventions.

Acknowledgements
This article presents independent research funded by
the NIHR School for Social Care Research. The views
expressed in this publication are those of the authors
and not necessarily those of the NIHR School for
Social Care Research or the Department of Health,
NIHR or NHS.

References
An asterisk (*) indicates that the reference has been
included in the scoping review.
*Aanes M.M., Mittelmark M.B. & Hetland J. (2010) Interpersonal stress and poor health: the mediating role of loneliness. European Psychologist 15 (1), 3–11. *Adams K.B., Sanders S. & Auth E.A. (2004) Loneliness and depression in independent living retirement communities:
risk and resilience factors. Aging & Mental Health 8 (6), 475–485.
*Alma M.A., Van der Mei S.F., Feitsma W.N., Groothoff
J.W., Van Tilburg T.G. & Suurmeijer T.P. (2011) Loneliness and self-management abilities in the visually impaired
elderly. Journal of Aging and Health 23 (5), 843–861.
*Alpass F.M. & Neville S. (2003) Loneliness, health and depression in older males. Aging & Mental Health 7 (3), 212–216.

Arksey H. & O’Malley L. (2005) Scoping studies: towards a methodological framework. International Journal of Social Research Methodology 8 (1), 19–32.

Askeslon N., Campo S. & Carter K. (2011) Completely isolated? Health information seeking among social isolates.

Health Education and Behavior 38 (2), 116–122.
*Avlund K., Vass M., Lund R., Yamada Y. & Hendriksen C. (2008) Influence of psychological characteristics and social relations on receiving preventive home visits in older men and women. European Journal of Ageing 5 (3), 191–201.
808 © 2015 John Wiley & Sons Ltd
E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License *Ayalon L. & Shiovitz-Ezra S. (2011) The relationship
between loneliness and passive death wishes in the
second half of life. International Psychogeriatrics 23 (10), 1677–1685.
*Beeson R.A. (2003) Loneliness and depression in spousal caregivers of those with Alzheimer’s disease versus noncaregiving spouses. Archives of Psychiatric Nursing 17 (3), 135–143.
*Beeson R., Horton-Deutsch S., Farran C. & Neundorfer M. (2000) Loneliness and depression in caregivers of persons with Alzheimer’s disease or related disorders. Issues in Mental Health Nursing 21 (8), 779–806.
*Bekhet A.K. & Zauszniewski J.A. (2012) Mental health
of elders in retirement communities: is loneliness a
key factor? Archives of Psychiatric Nursing 26 (3),
214–224.
*Bisschop M.I., Kriegsman D.M.W., van Tilburg T.G., Penninx B.W.J.H., van Eijk J. & Deeg D.J.H. (2003) The influence of differing social ties on decline in physical functioning among older people with and without
chronic diseases: the Longitudinal Aging Study Amsterdam. Aging Clinical and Experimental Research 15 (2), 164–173.
*Bisschop M.I., Kriegsman D.M.W., Beekman A.T.F. & Deeg D.J.H. (2004) Chronic diseases and depression: the modifying role of psychosocial resources. Social Science and Medicine 59 (4), 721–733.
*Brummett B.H., Barefoot J.C., Siegler I.C. et al. (2001) Characteristics of socially isolated patients with coronary artery disease who are at elevated risk for mortality. Psychosomatic Medicine 63 (2), 267–272. *Burr J.A. & Lee H.J. (2013) Social relationships and dental care service utilization among older adults. Journal of Aging and Health 25 (2), 191–220.

Cacioppo J.T., Hawkley L.C., Crawford L.E., Ernst J.M., Burleson M.H., Kowalewski R.B. et al. (2002) Loneliness and health: potential mechanisms. Psychosomatic Medicine 64 (3), 407–417.
*Cacioppo J.T., Hughes M.E., Waite L.J., Hawkley L.C. & Thisted R.A. (2006) Loneliness as a specific risk factor for depressive symptoms: cross-sectional and longitudinal
analyses. Psychology and Aging 21 (1), 140–151.
*Cacioppo J.T., Hawkley L.C. & Thisted R.A. (2010) Perceived social isolation makes me sad: 5-year cross-lagged analyses of loneliness and depressive symptomatology in the chicago health, aging, and social relations study. Psychology and Aging 25 (2), 453–463.

Cattan M. & White M. (1998) Developing evidence base
health promotion for older people: a systematic review
and survey of health promotion interventions targeting
social isolation and loneliness among older people. Internet Journal of Health Promotion 13,1–9.

Cattan M., White M., Bond J. & Learmouth A. (2005)
Preventing social isolation and loneliness among older
people: a systematic review of health promotion interventions. Ageing and Society 25, 41–67. *Cene C.W., Loehr L., Lin F.C., Hammond W.P., Foraker
R.E., Rose K. et al. (2012) Social isolation, vital exhaustion, and incident heart failure: findings from the Atherosclerosis Risk in Communities Study. European Journal of Heart Failure 14 (7), 748–753.
*Conroy R.M., Golden J., Jeffares I., O’Neill D. & McGee H. (2010) Boredom-proneness, loneliness, social engagement and depression and their association with cognitive function in older people: a population study. Psychology, Health & Medicine 15 (4), 463–473.
*Cornwell E.Y. & Waite L.J. (2009) Social disconnectedness, perceived isolation, and health among older adults. Journal of Health and Social Behavior 50 (1), 31–48. *Coyle C.E. & Dugan E. (2012) Social isolation, loneliness and health among older adults. Journal of Aging and Health 24 (8), 1346–1363.
*Dong X.Q., Chang E.S., Wong E. & Simon M. (2012) Perception and negative effect of loneliness in a Chicago Chinese population of older adults. Archives of Gerontology and Geriatrics 54 (1), 151–159.
*Drageset J., Eide G.E., Kirkevold M. & Ranhoff A.H.
(2013a) Emotional loneliness is associated with mortality among mentally intact nursing home residents with and
without cancer: a five-year follow-up study. Journal of Clinical Nursing 22 (1–2), 106–114.
*Drageset J., Eide G.E. & Ranhoff A.H. (2013b) Anxiety and depression among nursing home residents without cognitive impairment. Scandinavian Journal of Caring Sciences 27 (4), 872–881.
*Eng P.M., Rimm E.B., Fitzmaurice G. & Kawachi I. (2002) Social ties and change in social ties in relation to subsequent total and cause-specific mortality and coronary heart disease incidence in men. American Journal of Epidemiology 155 (8), 700–709. *Fessman N. & Lester D. (2000) Loneliness and depression among elderly nursing home patients. International Journal of Aging & Human Development 51 (2), 137–141.

Findlay R.A. (2003) Interventions to reduce social isolation amongst older people: where is the evidence? Ageing and Society 23 (5), 647–658.
*Fiori K.L., Antonucci T.C. & Cortina K.S. (2006) Social network typologies and mental health among older adults.

Journals of Gerontology – Series B Psychological Sciences and Social Sciences 61 (1), 25–32.
*Friedmann E., Thomas S.A., Liu F., Morton P.G., Chapa D., Gottlieb S.S. et al. (2006) Relationship of depression, anxiety, and social isolation to chronic heart failure outpatient mortality. American Heart Journal 152 (5), 940–948.
*Gask L., Aseem S., Waquas A. & Waheed W. (2011) Isolation, feeling ‘stuck’ and loss of control: understanding persistence of depression in British Pakistani women.

Journal of Affective Disorders 128 (1–2), 49–55.
*Giuli C., Spazzafumo L., Sirolla C., Abbatecola A.M., Lattanzio F. & Postacchini D. (2012) Social isolation risk factors in older hospitalized individuals. Archives of Gerontology and Geriatrics 55 (3), 580–585.
*Gleibs I.H., Haslam C., Jones J.M., Haslam S.A., McNeill J. & Connolly H. (2011) No country for old men? The role
of a ‘Gentlemen’s Club’ in promoting social engagement
and psychological well-being in residential care. Aging & Mental Health 15 (4), 456–466.
*Green A.F., Rebok G. & Lyketsos C.G. (2008) Influence of social network characteristics on cognition and functional status with aging. International Journal of Geriatric Psychiatry 23 (9), 972–978. *Greenfield T.K., Rehm J. & Rogers J.D. (2002) Effects of depression and social integration on the relationship
between alcohol consumption and all-cause mortality.

Addiction 97 (1), 29–38.
*Grov C., Golub S.A., Parsons J.T., Brennan M. & Karpiak S.E. (2010) Loneliness and HIV-related stigma explain
depression among older HIV-positive adults. AIDS Care-
© 2015 John Wiley & Sons Ltd 809
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Psychological and Socio-Medical Aspects of AIDS/HIV 22 (5), 630–639.
*Ha J.H. & Ingersoll-Dayton B. (2011) Moderators in the relationship between social contact and psychological distress among widowed adults. Aging & Mental Health 15 (3), 354–363.
*Hackett R.A., Hamer M., Endrighi R., Brydon L. & Steptoe A. (2012) Loneliness and stress-related inflammatory and neuroendocrine responses in older men and women. Psychoneuroendocrinology 37 (11), 1801–1809.

Hagan R., Manktelow R., Taylor B.J. & Mallett J. (2014) Reducing loneliness amongst older people: a systematic
search and narrative review. Aging & Mental Health 18 (6), 683–693.
*Han J. & Richardson V.E. (2010) The relationship between depression and loneliness among homebound older persons: does spirituality moderate this relationship? Journal of Religion and Spirituality in Social Work 29 (3), 218–236. *Hastings S.N., George L.K., Fillenbaum G.G., Park R.S., Burchett B.M. & Schmader K.E. (2008) Does lack of social support lead to more ED visits for older adults? American Journal of Emergency Medicine 26 (4), 454–461.

Hawkley L.C. & Cacioppo J.T. (2010) Loneliness matters: a theoretical and empirical review of consequences and
mechanisms. Annals of Behavioral Medicine 40, 218–227.
*Hawkley L.C., Masi C.M., Berry J.D. & Cacioppo J.T.
(2006) Loneliness is a unique predictor of age-related differences in systolic blood pressure. Psychology and Aging 21 (1), 152–164.
*Hawkley L.C., Preacher K.J. & Cacioppo J.T. (2010a) Loneliness impairs daytime functioning but not sleep duration.

Health Psychology 29 (2), 124–129.
*Hawkley L.C., Thisted R.A., Masi C.M. & Cacioppo J.T.
(2010b) Loneliness predicts increased blood pressure: 5- year cross-lagged analyses in middle-aged and older
adults. Psychology and Aging 25 (1), 132–141.
*Heikkinen R.-L. & Kauppinen M. (2004) Depressive symptoms in late life: a 10-year follow-up. Archives of Gerontology and Geriatrics 38 (3), 239–250. *Holmen K. & Furukawa H. (2002) Loneliness, health and
social network among elderly people-a follow-up study.

Archives of Gerontology and Geriatics 35 (3), 261–274.

Holt-Lunstad J., Smith T.B. & Layton J.B. (2010) Social relationships and mortality risk: a meta-analytic review. PLoS Medicine 7, 7.
*Holwerda T.J., Beekman A.T.F., Deeg D.J.H., Stek M.L., van Tilburg T.G., Visser P.J. et al. (2012) Increased risk of mortality associated with social isolation in older men:
only when feeling lonely? Results from the Amsterdam
Study of the Elderly (AMSTEL). Psychological Medicine 42 (4), 843–853.
*Jaremka L.M., Andridge R.R., Fagundes C.P., Alfano C.M., Povoski S.P., Lipari A.M. et al. (2013) Pain, depression, and fatigue: loneliness as a longitudinal risk factor. Health Psychology 19, 19.
de Jong Gierveld J. (1987) Developing and testing a model of loneliness. Journal of Personality and Social Psychology 53, 119–128.
*Jongenelis K., Pot A.M., Eisses A.M.H., Beekman A.T.F., Kluiter H. & Ribbe M.W. (2004) Prevalence and risk
indicators of depression in elderly nursing home patients:
the AGED study. Journal of Affective Disorders 83 (2–3), 135–142.
*Kahlbaugh P.E., Sperandio A.J., Carlson A.L. & Hauselt J. (2011) Effects of playing wii on well-being in the elderly:
physical activity, loneliness, and mood. Activities, Adaptation and Aging 35 (4), 331–344. *Kamiya Y., Whelan B., Timonen V. & Kenny R.A. (2010)
The differential impact of subjective and objective aspects of social engagement on cardiovascular risk factors. BMC Geriatrics 10, 81.
*Krause-Parello C.A. (2008) The mediating effect of pet attachment support between loneliness and general health in older females living in the community. Journal of Community Health Nursing 25 (1), 1–14. *Krause-Parello C.A. (2012) Pet ownership and older
women: the relationships among loneliness, pet attachment support, human social support, and depressed mood. Geriatric Nursing 33 (3), 194–203.
*Kvaal K., Halding A.G. & Kvigne K. (2013) Social provision and loneliness among older people suffering from chronic physical illness. A mixed-methods approach.

Scandinavian Journal of Caring Sciences 28 (1), 104–111. *La Grow S., Neville S., Alpass F. & Rodgers V. (2012)
Loneliness and self-reported health among older persons in New Zealand. Australasian Journal on Ageing 31 (2),
121–123.
*Lefranc
ois R., Leclerc G., Hamel S. & Gaulin P. (2000) Stressful life events and psychological distress of the very old: does social support have a moderating effect?
Archives of Gerontology and Geriatrics 31 (3), 243–255. *Losada A., Marquez-Gonzalez M., Garcia-Ortiz L., GomezMarcos M.A., Fernandez-Fernandez V. & Rodriguez-Sanchez E. (2012) Loneliness and mental health in a representative sample of community-dwelling spanish older adults. Journal of Psychology 146 (3), 277–292.
*Luo Y., Hawkley L.C., Waite L.J. & Cacioppo J.T. (2012) Loneliness, health, and mortality in old age: a national longitudinal study. Social Science & Medicine 74 (6),
907–914.

Marengoni A., Winblad B., Karp A. & Fratiglioni L. (2008) Prevalence of chronic diseases and multimorbidity among the elderly population in Sweden. American Journal of Public Health 98 (7), 1198–1200. *Martina C.M.S. & Stevens N.L. (2006) Breaking the cycle of loneliness? Psychological effects of a friendship enrichment program for older women. Aging & Mental Health 10 (5), 467–475.

McWhirter B.T. (1990) Loneliness: a review of the current literature, with implications for counseling and research.

Journal of Counseling and Development 68, 417–422.

Means R. (2007) Safe as houses? Ageing in place and vulnerable older people in the UK. Social Policy & Administration 41 (1), 65–85. *Michael Y.L., Berkman L.F., Colditz G.A. & Kawachi I.
(2001) Living arrangements, social integration, and
change in functional health status. American Journal of Epidemiology 153 (2), 123–131.

Mistry R., Rosansky J., McGuire J., McDermott C. & Jarvik L. (2001) Social isolation predicts re-hospitalization in a group of older American veterans enrolled in the
UPBEAT Program. Unified Psychogeriatric Biopsychosocial Evaluation and Treatment. International Journal of Geriatric Psychiatry 16 (10), 950–959. *Murberg T.A. (2004) Long-term effect of social relationships on mortality in patients with congestive heart fail810 © 2015 John Wiley & Sons Ltd E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License ure. International Journal of Psychiatry in Medicine 34 (3), 207–217.
*Newall N.E.G., Chipperfield J.G., Bailis D.S. & Stewart T.L. (2013) Consequences of loneliness on physical activity
and mortality in older adults and the power of positive emotions. Health Psychology 32 (8), 921–924.
*Nezlek J.B., Richardson D.S., Green L.R. & Schatten-Jones E.C. (2002) Psychological well-being and day-to-day social interaction among older adults. Personal Relationships 9 (1), 57–71.
*Norman G.J., Hawkley L., Ball A., Berntson G.G. &
Cacioppo J.T. (2013) Perceived social isolation moderates the relationship between early childhood trauma and
pulse pressure in older adults. International Journal of Psychophysiology 88 (3), 334–338. *Nummela O., Sepp€anen M. & Uutela A. (2011) The effect of loneliness and change in loneliness on self-rated health (SRH): a longitudinal study among aging people. Archives of Gerontology and Geriatrics 53 (2), 163–167.

O’Luanaigh C. & Lawlor B.A. (2008) Loneliness and the
health of older people. International Journal of Geriatric Psychiatry 23, 1213–1221. *O’Luanaigh C., O’Connell H., Chin A.V., Hamilton F.,
Coen R., Walsh C. et al. (2011) Loneliness and vascular biomarkers: the Dublin Healthy Ageing Study. International Journal of Geriatric Psychiatry 27 (1), 83–88. *O’Luanaigh C., O’Connell H., Chin A.V., Hamilton F.,
Coen R., Walsh C. et al. (2012) Loneliness and cognition in older people: the Dublin Healthy Ageing study. Aging & Mental Health 16 (3), 347–352.
*Ong A.D., Rothstein J.D. & Uchino B.N. (2012) Loneliness accentuates age differences in cardiovascular responses to social evaluative threat. Psychology and Aging 27 (1),
190–198.
*Park N.S., Jang Y.R., Lee B.S., Haley W.E. & Chiriboga D.A. (2013) The mediating role of loneliness in the relation between social engagement and depressive symptoms among older Korean Americans: do men and women differ? Journals of Gerontology Series B-Psychological Sciences and Social Sciences 68 (2), 193–201.
*Parsons J.G.M., Sheridan N., Rouse P., Robinson E. & Connolly M. (2013) A randomized controlled trial to determine the effect of a model of restorative home care on physical function and social support among older people.

Archives of Physical Medicine and Rehabilitation 94 (6), 1015–1022.
*Patterson A.C. & Veenstra G. (2010) Loneliness and risk of mortality: a longitudinal investigation in Alameda County, California. Social Science & Medicine 71 (1), 181–186.
*Paul C., Ayis S. & Ebrahim S. (2006) Psychological distress, loneliness and disability in old age. Psychology, Health and Medicine 11 (2), 221–232.

Pawson R. (2002) Evidence-based policy: the promise of ‘realist synthesis’. Evaluation 8 (3), 340–358.

Penning M., Liu G. & Chou P. (2013) Measuring loneliness among middle-aged and older adults: the UCLA and de
Jong Gierveld Loneliness Scales. Social Indicators Research 18 (3), 1–20.

Peplau L. & Perlman D. (1982) Perspectives on loneliness.

In: L. Peplau & D. Perlman (Eds) Loneliness: a sourcebook of current theory, research and therapy, pp. 1–18. WileyInterscience, New York.

Perese E. & Wolf M. (2005) Combating loneliness among
people with severe mental illness: social network interventions’ characteristics, effectiveness and applicability.

Issues in Mental Health Nursing 26, 591–609.
*Perissinotto C.M., Stijacic Cenzer I. & Covinsky K.E. (2012) Loneliness in older persons: a predictor of functional
decline and death. Archives of Internal Medicine 172 (14), 1078–1083.
*Poulin J., Deng R., Ingersoll T.S., Witt H. & Swain M. (2012) Perceived family and friend support and the psychological well-being of American and Chinese elderly persons. Journal of Cross-Cultural Gerontology 27 (4), 305–317. *Rodriguez C.J., Elkind M.S., Clemow L., Jin Z., Di Tullio M., Sacco R.L. et al. (2011) Association between social isolation and left ventricular mass. American Journal of Medicine 124 (2), 164–170.

Rokach A. et al. (2001) The effects of culture on the meaning of loneliness. Social Indicators Research 53 (1), 17–31.

Rook K.S. (1984) Promoting social bonding: strategies for helping the lonely and socially isolated. American Psychologist 39, 1389–1407. *Routasalo P.E., Tilvis R.S., Kautiainen H. & Pitkala K.H. (2009) Effects of psychosocial group rehabilitation on
social functioning, loneliness and well-being of lonely, older people: randomized controlled trial. Journal of
Advanced Nursing 65 (2), 297–305.

Russell D.W. (1996) UCLA Loneliness Scale (Version 3): reliability, validity, and factor structure. Journal of Personality Assessment 66 (1), 20–40.
*Savikko N., Routasalo P., Tilvis R.S., Strandberg T.E. & Pitk€al€a K.H. (2005) Predictors and subjective causes of loneliness in an aged population. Archives of Gerontology and Geriatrics 41 (3), 223–233.
*Shankar A., McMunn A., Banks J. & Steptoe A. (2011)
Loneliness, social isolation, and behavioral and biological health indicators in older adults. Health Psychology 30 (4), 377–385.
*Shankar A., Hamer M., McMunn A. & Steptoe A. (2013)
Social isolation and loneliness: relationships with cognitive function during 4 years of follow-up in the English Longitudinal Study of Ageing. Psychosomatic Medicine 75 (2), 161–170.
*Sjoberg L., Ostling S., Falk H., Sundh V., Waern M. &
Skoog I. (2013) Secular changes in the relation between social factors and depression: a study of two birth cohorts of Swedish septuagenarians followed for 5 years. Journal of Affective Disorders 150 (2), 245–252.
*Smith T.F. & Hirdes J.R. (2009) Predicting social isolation among geriatric psychiatry patients. International Psychogeriatrics 21 (1), 50–59. *Smith M.L. & Rosen D. (2009) Mistrust and self-isolation:
barriers to social support for older adult methadone clients. Journal of Gerontological Social Work 52 (7), 653–667. *Sorkin D., Rook K.S. & Lu J.L. (2002) Loneliness, lack of emotional support, lack of companionship, and the likelihood of having a heart condition in an elderly sample.

Annals of Behavioral Medicine 24 (4), 290–298.
*Stafford M., Gardner M., Kumari M., Kuh D. & BenShlomo Y. (2013) Social isolation and diurnal cortisol patterns in an ageing cohort. Psychoneuroendocrinology 3 (13), 252–257.
*Stek M.L., Vinkers D.J., Gussekloo J., Beekman A.T.F., van der Mast R.C. & Westendorp R.G.J. (2005) Is depression
in old age fatal only when people feel lonely? American Journal of Psychiatry 162 (1), 178–180.
© 2015 John Wiley & Sons Ltd 811
Social isolation, loneliness and health in old age
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License *Stephens C., Alpass F. & Towers A. (2010) Economic hardship among older people in New Zealand: the effects of low living standards on social support, loneliness, and mental health. New Zealand Journal of Psychology 39 (2), 49–55. *Stephens C., Alpass F., Towers A. & Stevenson B. (2011) The effects of types of social networks, perceived social support, and loneliness on the health of older people:
accounting for the social context. Journal of Aging and Health 23 (6), 887–911.
*Steptoe A., Owen N., Kunz-Ebrecht S.R. & Brydon L.
(2004) Loneliness and neuroendocrine, cardiovascular,
and inflammatory stress responses in middle-aged men
and women. Psychoneuroendocrinology 29 (5), 593–611.
*Steptoe A., Shankar A., Demakakos P. & Wardle J. (2013) Social isolation, loneliness, and all-cause mortality in older men and women. Proceedings of the National Academy of Sciences of the United States of America 110 (15),
5797–5801.

Sundstrom G., Fransson E., Malmberg B. & Davey A. (2009) € Loneliness among older Europeans. European Journal of
Ageing 6 (4), 267–275.

Theeke L.A. (2010a) A current state of the science on loneliness and the health of older adults. Gerontologist 50, 3. *Theeke L.A. (2010b) Sociodemographic and health-related risks for loneliness and outcome differences by loneliness status in a sample of U.S. older adults. Research in Gerontological Nursing 3 (2), 113–125. *Theeke L.A. & Mallow J. (2013) Loneliness and quality of life in chronically ill rural older adults. American Journal of Nursing 14, 14.
*Theeke L.A., Goins R.T., Moore J. & Campbell H. (2012) Loneliness, depression, social support, and quality of life in older chronically ill appalachians. Journal of Psychology 146 (1–2), 155–171.
*Tiikkainen P. & Heikkinen R.L. (2005) Associations
between loneliness, depressive symptoms and perceived
togetherness in older people. Aging & Mental Health 9 (6), 526–534.
*Tilvis R.S., Laitala V., Routasalo P.E. & Pitk€al€a K.H. (2011) Suffering from loneliness indicates significant mortality risk of older people. Journal of Aging Research 2090–2204. *Tilvis R.S., Routasalo P., Karppinen H., Strandberg T.E., Kautiainen H. & Pitkala K.H. (2012) Social isolation, social activity and loneliness as survival indicators in old age; a nationwide survey with a 7-year follow-up. European Geriatric Medicine 3 (1), 18–22. *Udell J.A., Steg P.G., Scirica B.M. et al. (2012) Living alone and cardiovascular risk in outpatients at risk of or with atherothrombosis. Archives of Internal Medicine 172 (14), 1086–1095.

Valtorta N. & Hanratty B. (2010) Loneliness, isolation and health of older adults: do we need a new research
agenda? Journal of the Royal Society of Medicine 105,
518–522.
*Van Orden K.A., Stone D.M., Rowe J., McIntosh W.L., Podgorski C. & Conwell Y. (2013) The senior connection:
design and rationale of a randomized trial of peer companionship to reduce suicide risk in later life. Contemporary Clinical Trials 35 (1), 117–126. *VanderWeele T.J., Hawkley L.C., Thisted R.A. & Cacioppo J.T. (2011) A marginal structural model analysis for loneliness: implications for intervention trials and clinical practice. Journal of Consulting and Clinical Psychology 79 (2), 225–235.
*VanderWeele T.J., Hawkley L.C. & Cacioppo J.T. (2012) On the reciprocal association between loneliness and subjective well-being. American Journal of Epidemiology 176 (9), 777–784.

Victor C., Scambler S., Shah S., Cook D., Harris T., Rink E. et al. (2002) Has loneliness amongst older people
increased? An investigation into variations between
cohorts. Ageing & Society 22 (5), 585–597.

Victor C., Grenade L. & Boldy D. (2005) Measuring loneliness in later life: a comparison of differing measures.

Reviews in Clinical Gerontology 15 (1), 63–70.
*Wang H.X., Mittleman M.A. & Orth-Gomer K. (2005) Influence of social support on progression of coronary artery disease in women. Social Science & Medicine 60 (3), 599–607. *Wang H.X., Mittleman M.A., Leineweber C. & Orth-Gomer
K. (2006) Depressive symptoms, social isolation, and progression of coronary artery atherosclerosis: the Stockholm female coronary angiography study. Psychotherapy and
Psychosomatics 75 (2), 96–102.
*Wattanakit K., Williams J.E., Schreiner P.J., Hirsch A.T. & Folsom A.R. (2005) Association of anger proneness,
depression and low social support with peripheral arterial disease: the Atherosclerosis Risk in Communities Study.

Vascular Medicine 10 (3), 199–206.
*Wilby F. (2011) Depression and social networks in community dwelling elders: a descriptive study. Journal of Gerontological Social Work 54 (3), 246–259.

Wiles J.L., Leibing A., Guberman N., Reeve J. & Allen R. (2012) The meaning of “ageing in place” to older people.

The Gerontologist 52 (3), 357–366.
*Wilson R.S., Krueger K.R., Arnold S.E., Schneider J.A., Kelly J.F., Barnes L.L. et al. (2007) Loneliness and risk of Alzheimer’s disease. Archives of General Psychiatry 64 (2), 234–240.

Wilson D.M., Harris A., Hollis V. & Mohankumar D. (2011) Upstream thinking and health promotion planning for
older adults at risk of social isolation. International Journal of Older People Nursing 6 (4), 282–288.
*Winter L. & Gitlin L.N. (2006) Evaluation of a telephonebased support group intervention for female caregivers of community-dwelling individuals with dementia. American
Journal of Alzheimer’s Disease and Other Dementias 21 (6), 391–397.
*Zebhauser A., Hofmann-Xu L., Baumert J., H€afner S.,
Lacruz M.E., Emeny R.T. et al. (2013) How much does it
hurt to be lonely? Mental and physical differences
between older men and women in the KORA-Age Study.

International Journal of Geriatric Psychiatry 29 (3), 245–252. van der Zwet J., Koelewijn-van Loon M.S. & van den Akker M. (2009) Lonely patients in general practice: a call for revealing GPs’ emotions? A qualitative study Family Practice 26 (6), 501–509.

Supporting Information
Additional supporting information may be found in
the online version of this article at the publisher’s
web-site.
812 © 2015 John Wiley & Sons Ltd
E. Courtin & M. Knapp
13652524, 2017, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/hsc.12311, Wiley Online Library on [10/06/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 

 




 


 



 



!"
#

$ 

%& '
( ))

 #* #
 *)
+
 ,
' % 

+

+
-.)*
&/

)+
/#
&01 . 23 45
6 *
& 
+

 *&7
#

 )+#*&8 

/#
 9 :)
(* #))
 +

%(


# 
+

*



;#<+%+
:
 
< %+

 =" 

) +&
 



 


 
 

 



 


!"#$%
&#"%'$()*

 
  

 

 

 

 

 






 
   



  



 






 
 



!






  "

"


 
 
 

#
 




 
#

 

$ 
  % 
&


 







  

 

 


 

 
' 
 




 
  





(
)
  

 
   

*


  



 


  




+, - ./01
23 




 
 
 
 
4*
 
 #
 


 




 
5
%6



  
&
7


 

6

  %




    #


   

 

 



  
 







 

 











 



 

 ! 
 


"#

!











 

 


 " $


  








 

    







 


 
 %
&

'
( 




 )*+*,-./0-,+1,2 3421*5/ 6


 



 
 



 7 



8



 


 


 

 


 


 






  

 
 








 







  


 !"#$%&'"%()*'"



 +













 
  








  
 







  


   

  








 !




  




"
 
!

#

!

$%&'()*+,-(./0123&*)-34*5(1



 


6 





78 
 

 
98 
! 




  




 



 

"

  

 




 





 












 
 
 

  
 




 







 




 


 
 
 
  


 



 




! "
#$ 

  
 

  



 "

%
"
&


 








'
() 






'





 

 





"


 




 
* 





"# 
 

 $

 

 



  
 "# 

  
   

"
+
 ,

 %

 

 "

, 
$



 ,
 


"-

 
 

 

 




, 




 . 

"








  


 
  

 

 
 
 













 





  






! 
 

 
"



 

 

" 


 #
 
$
 


 










  

 





   



 
%&'()*+'&,-)%.*/+01-)+2.3(2% 
# 

4

4





#

5


 




$ 






 



 6  





 





  


 #

 
"


$5 



 

 






7

  
 



89


   




4 
  
 




 

:7

 8;4 


 


:


 







 



 









 




 

  


 



 











 

 


!

"#$%




%&

 
 



'







 (





 


 




 














)
 *+,







 





  










- 

.


 



 



.




"#$/

 

0123 4556789






 




 
":,




 ,









 



 


;
 
 








 







 

 


 

 





 








 

 






 

 





 

 




 
 













 


 
 

 
   
 

 

 




  

  





 

 

!

 " 




  #
 !


 
$%


 

 
&

 
 
'

(!












 




!"#$% !
& #'!"#() '


*
 "&(

# +," #!

-
 


. 


 
/ 
 /
0

 /



  

12

/3
4

5
/ 0
0











 


 ! " !#
$%
 &'()*'+),-
./
01 23/14.
5/1
6781/ 792
/ 8071:/ 7
/392

1/37;29 <

=>5
1 87

17:?19/
7
@AB
9C/30:1
.
5/1
2;DAE

/
17FG;
H0:/  37025
18/564I:; /1B

/71
/

0/23
9FC381B9/276



 






 








 






!



 


"#$%&
'

(





 )


*
+ ,-./0123456./ 



+

+
(


 7



+









 +89!


 
&


 : 
 ) &:


;&<

 *

'



)
(
 
(
=
 



  


;(7



$>8




 

  

 







!"#$
%&
'(!)*+,-.
" /0123450
67897:;<=>?<@ABC<@D EFB<G@HCIJK;LMNOPQRSTUVWRPUMXYLZ[S\Y]QZ^_`abcdefghijklemnffoepkqfrh enfsltufnmrvkepwfgjxvle uhklvf tjimwfgtkryhkzr{fxk|emklht}~jlep|krhlefhrx}ihrft~€rtmvp}‚fwrpk xhnmr{ vkƒfvkmritfpkvehƒfklepgmtf{elftknhjxvihrlpmefqxp„fk‚vt}wf{lpmhrpvklfƒtpmrƒjkefvl pxmkvfgmtf‚le qm{fkl‚pkl‚ƒj mr{p fnknplehgy kzr{fxkjrvftuprvplht}fwpijlmhrtvfk~defqxp„fnpkqjtrm{pxhr{lefhƒk gleftmv{fkhg†prvfwmx‡pr}h‚kpmvˆfrmie‰pkofl‚pvmwkhriemfgnmle fyhkzr{fxk‡hjrl}mtfŠfƒptlufr‚h tmvp}rm{el~d fgmtlh ehj{epklfƒpt gjxhgvt}wf{lpmhrpvletfp rvle rm{eqht vg‹rimh lefrhtle~ŒŽ‘’“”•–—˜™š›œ—”š“ž”–Ÿ ¡ŸšŒ•¡  



 



















 



 





 !

"#$ 

  



 


!





!

!

%

&








 


! 







 '

&




(
)







!



 *
+
 
 






 +

!
%
+



 
  




 


!
 ,
 

 





 

 
 
-
%



 - 



* +

'




%
  




 




%

!


 








   

 


 
.

#/!


.

0



12
!
 !



 3

% 
4

!








5$!
4




!


 -

 
2
!


0
-




 




 

 


&
&!


 


4 

6&
%
7 %8

0




 !
 








  

 













&



!)
!

 





 






6

90
&

9

!


 
*
+
0
 1!


  


!
%


&


 %

:




 





!

!







 




 


  

  
  



 






  

 !"
# 
 
 $ %&'()*&'+,-./,01-21345 ,671489:;4<=12->1.?,6@1:-A B)&C'D*EFGH,->.1I,67.<201654.I<4=12-J.9,K426.1:-L349,<M5216I-. 312-46.9,51-A7;1 .->16.N4826.9,-5K420415;.6K482 I1<. -A This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi:

10.1111/EVA.13341
This article is protected by copyright. All rights reserved Multi-population puma connectivity could restore genomic diversity to at-risk coastal populations in California
Running title: Population genomics of California pumas
Kyle D. Gustafson*, Roderick B. Gagne, Michael R. Buchalski, T. Winston Vickers, Seth P.D. Riley, Jeff A. Sikich, Jaime L. Rudd, Justin A. Dellinger, Melanie E.F. LaCava, Holly B. Ernest *Corresponding author
KDG: [email protected], 870-972-3174, Arkansas State University, Department of Biological Sciences, Jonesboro, AR 72401, ORCID 0000-0003-1869-4023 RBG: [email protected], 610-925-6713, University of Pennsylvania School of Veterinary Medicine, Department of Pathobiology, Wildlife Futures Program, Kennett Square, PA, USA MRB: [email protected], 916-508-1385, California Department of Fish and Wildlife, Rancho Cordova, CA 95670
TWV: [email protected], 949-929-8643, University of California - Davis, Karen C. Drayer Wildlife Health Center, School of Veterinary Medicine, Davis, CA 95616 SPDR: [email protected], 805-370-2358, National Park Service, Santa Monica Mountains National Recreation Area, 401 W. Hillcrest Dr., Thousand Oaks, CA, 91360.

JAS: [email protected], 805-370-2395, National Park Service, Santa Monica Mountains National Recreation Area, 401 W. Hillcrest Dr., Thousand Oaks, CA, 91360.

JLR: [email protected], 916-358-2790, California Department of Fish and Wildlife, Rancho Cordova, CA 95670
JAD: [email protected], 916-261-3610, California Department of Fish and Wildlife, Rancho Cordova, CA 95670
MEFL: [email protected], Wildlife Genomics and Disease Ecology Laboratory, Department of Veterinary Sciences, University of Wyoming, Laramie, WY 82071 HBE: [email protected], Wildlife Genomics and Disease Ecology Laboratory, Department of Veterinary Sciences, University of Wyoming, Laramie, WY 82071

ACKNOWLEDGMENTS
Accepted Article
This article is protected by copyright. All rights reserved We thank the multiple agencies and people who provided samples and expertise including California Department of Fish and Wildlife (D. Clifford, R. Botta, J. Colby, S. Torres), California State Parks, The Nature Conservancy, University of California Davis Wildlife Health Center (T.

Drazenovich), University of California Los Angeles, The National Park Service, and the US Geological Survey. We thank University of Wyoming personnel, S. Love Stowell, A. Gustafson, L.

Johnson, B. Godwin, and C. A. Buerkle. Computational resources were provided by Advanced Research Computing Center (2018) Teton Computing Environment, Intel x86_64 cluster. University of Wyoming, Laramie, WY https://doi.org/10.15786/M2FY47. This research was also supported by the Arkansas High Performance Computing Center which is funded through multiple National Science Foundation grants and the Arkansas Economic Development Commission. Funding for this work was provided by the California Department of Fish and Wildlife (H.B.E.), Excellence Chair funds (H.B.E), National Science Foundation Ecology of Infectious Disease program grant (DEB 1413925). We thank three anonymous reviewers for their constructive comments.

AUTHOR CONTRIBUTIONS
KDG, MRB, JAD, and HBE designed the study. KDG, RBG, JLR, and MEFL performed laboratory research. MRB, TWB, SPDR, JAS, and JAD performed field research. KDG, RBG, MRB, MEFL, and HBE analyzed the data. KDG, RBG, MRB, TWV, SPDR, JAS, JLR, JAD, MEFL, and HBE wrote the paper.

Accepted Article
This article is protected by copyright. All rights reserved DR. KYLE D GUSTAFSON (Orcid ID : 0000-0003-1869-4023)
DR. RODERICK B GAGNE (Orcid ID : 0000-0002-4901-5081)
DR. MELANIE E.F. LACAVA (Orcid ID : 0000-0001-7921-9184) Article type : Original Article

ABSTRACT
Urbanization is decreasing wildlife habitat and connectivity worldwide, including for apex predators, such as the puma (Puma concolor). Puma populations along California’s central and southern coastal habitats have experienced rapid fragmentation from development, leading to calls for demographic and genetic management. To address urgent conservation genomic concerns, we used double-digest restriction-site associated DNA (ddRAD) sequencing to analyze 16,285 genome-wide single-nucleotide polymorphisms (SNPs) from 401 pumas sampled broadly across the state. Our analyses indicated support for 4–10 geographically nested, broad- to finescale genetic clusters. At the broadest scale, the 4 genetic clusters had high genetic diversity and exhibited low linkage disequilibrium, indicating pumas have retained genomic diversity statewide. However, multiple lines of evidence indicated substructure, including 10 finer-scale genetic clusters, some of which exhibited fixed alleles and linkage disequilibrium. Fragmented populations along the Southern Coast and Central Coast had particularly low genetic diversity and strong linkage disequilibrium, indicating genetic drift and close inbreeding. Our results demonstrate that genetically at-risk populations are typically nested within a broader-scale group of interconnected populations that collectively retains high genetic diversity and heterogeneous fixations. Thus, extant variation at the broader scale has potential to restore diversity to local populations if management actions can enhance vital gene flow and recombine locally sequestered genetic diversity. These state- and genome-wide results are critically important for Accepted Article
science-based conservation and management practices. Our nested population genomic analysis This article is protected by copyright. All rights reserved highlights the information that can be gained from population genomic studies aiming to provide guidance for the conservation of fragmented populations.

Key words: conservation genetics, mountain lion, nested population structure, population genetics, Puma concolor, SNP

INTRODUCTION
Human development is reducing habitat on a global scale, undermining efforts to conserve ecosystem structure and function (Newbold et al., 2016). Reports of fragmented wildlife populations and the increasing need for human housing and associated agriculture and energy have emphasized the necessity for development to avoid impacting the long-term sustainability of wildlife populations (Jordan et al., 2007; Kiesecker et al., 2011; Saha & Paterson, 2008). One of the most developed states in the United States is California, which contains the largest census size with over 39 million people (US Census, 2019). Although the development of California has led to historical extirpations of other apex predators, such as the grizzly bear (Ursus arctos; Herrero, 1970) and gray wolf (Canis lupus; Schmidt, 1991), the puma (Puma concolor; also known as mountain lion and cougar) has maintained a widespread distribution throughout the state (Dellinger et al., 2020a).

The puma is a large-bodied felid that originated in South America, migrated and expanded throughout North America, and experienced a human-induced range restriction to the western United States, with an extant remnant population in Florida (Culver, Johnson, PeconSlattery, & O’Brien, 2000). Currently, approximately half of all apparent puma habitat in California is conserved and the remainder could be subject to further development (Dellinger et al., 2020a). Much of the inland areas of California have continous stretches of protected habitat (Dellinger et al., 2020a) supporting puma populations with high genetic diversity and large effective populations sizes (Gustafson et al., 2019). However, movement corridors among coastal mountain ranges are increasingly being degraded by human development (Burdett et al., 2010; Suraci, Nickel, & Wilmers, 2020; Zeller et al., 2017). Despite the natural long-range dispersal abilities of pumas (Gonzalez-Borrajo, López‐ Bao, & Palomares, 2017), interstate highways limit dispersal via avoidance and direct mortality in some urban areas (Riley et al., Accepted Article
2014; Vickers et al., 2015). Although human-caused mortality from vehicle collisions and lethal This article is protected by copyright. All rights reserved removal after wildlife–livestock conflicts are concerns (Guerisoli, Luengos Vidal, Caruso, Giordano, & Lucherini, 2020; Torres, Mansfield, Foley, Lupo, & Brinkhaus, 1996), a larger concern for long-term population viability is the genetic isolation of pumas within small or shrinking patches of habitat, which has led to high levels of intraspecific competition and mortality (Benson, Sikich, & Riley, 2020) and low genetic diversity in some areas (Ernest et al., 2014; Gustafson et al., 2019; Riley et al., 2014).

Previous studies have reported that two isolated puma populations in southern California, including the Santa Ana Mountains and the Santa Monica Mountains (Fig. 1), had the lowest genetic diversity estimates measured throughout the range of P. concolor (Ernest et al., 2014; Riley et al., 2014), apart from the endangered Florida panther (P. c. coryi). In both the Santa Ana and Santa Monica Mountains, phenotypic evidence of inbreeding depression has been observed, similar to Florida panthers (Ernest et al., 2014; Huffmeyer, Sikich, Vickers, Riley, & Wayne, In Press; Roelke, Martenson, & O’Brien, 1993). For both populations, freeway traffic is isolating pumas (Ernest et al., 2014; Riley et al., 2014; Vickers et al., 2015) and contemporary gene flow has been severely limited. Detailed pedigree analyses following the immigration of one male into each region showed evidence of natural genetic rescue (Ernest et al., 2014; Gustafson, Vickers, Boyce, & Ernest, 2017; Riley et al., 2014). Although migrant effects were positive, projection models predict the extirpation of these populations in 50 years without enhanced demographic dispersal and gene flow (Benson et al., 2016; 2019).

Recently published genome-resequencing data that included 4 pumas from California, 2 from the Santa Monica Mountains and 2 from the Central Coast North region in the Santa Cruz Mountains, indicated these individuals had ~20–40% of their genomes represented as long runs of homozygosity, resulting from recent inbreeding (Saremi et al., 2019). However, these runs of homozygosity were not shared among individuals, and different populations exhibited different homozygous haplotypes, suggesting genetic restoration (Hedrick, 2005; Tallmon, Luikart, & Waples, 2004) is possible because genetic variation still exists.

The complex distribution of pumas throughout California along a continuum of high genetic diversity populations occupying abundant habitat, to strongly isolated populations displaying evidence of inbreeding depression, requires a thorough characterization of statewide genomic diversity to achieve proper conservation. In this study, our objective was to characterize Accepted Article
patterns of genomic diversity at varying geographic scales. Such an approach has the potential to This article is protected by copyright. All rights reserved aid conservation strategies because it can identify at-risk, low-diversity local populations that would benefit from restored gene flow within a broader geographic region. We identified 16,285 single-nucleotide polymorphisms (SNPs) from 401 individuals using a double-digest, restrictionsite associated DNA sequencing method (ddRAD; Peterson, Weber, Kay, Fisher, & Hoekstra, 2012). Specifically, our aims were to determine population genomic structure, genetic diversity, evidence for selection, and linkage disequilibrium.

METHODS
Sample collection and DNA extraction
We obtained 354 tissue samples collected by the California Department of Fish and Wildlife between 2011–2017 from pumas either hit-by-car (~6%), found dead (~2%), poached (<1%), or through depredation permits (>90%) which had never been used in any previous genetic survey. Samples were well-distributed throughout the state, except for smaller populations in smaller mountain ranges. To bolster our sample size in the Los Angeles region of southern California, we added the only remaining DNA extracts (N = 144) from pumas collected between 2002–2015 (Riley et al. 2014; Vickers et al., 2015). After genomic and bioinformatic filtering (described below), we retained 401 out of 498 samples in the final dataset, which spanned the majority of puma habitat in California, excluding desert regions (Fig. 1). For samples that lacked a precise GPS location, we used the nearest address or town where they were collected as their GPS point. Samples were stored at -80˚C until DNA was extracted using Omega Bio-tek Mag-Bind Blood & Tissue DNA HDQ Kits (Omega Bio-tek, #M6399-01) with a manufacturer-designed protocol for the Kingfisher Duo Prime (ThermoFisher Scientific, #5400110) automated DNA purification system. We measured the concentration of DNA from each sample using a Qubit 3.0 fluorometer (Invitrogen, #Q33216) with Qubit dsDNA highsensitivity kits (Invitrogen, #Q32854).

Double digest restriction site associated DNA library preparation and sequencing We reduced the genome size of our samples and identified single-nucleotide polymorphisms (SNPs) using modifications to the double digest restriction-site associated DNA Accepted Article
sequencing (ddRAD) protocols developed by Peterson et al. (2012). We used a library This article is protected by copyright. All rights reserved construction scheme which pooled 48 samples per library based on barcode availability, cost effective multiplexing, and sufficient coverage per individual. For each pooled library, we first normalized DNA concentrations to the sample with the lowest concentration within a library, with the goal to be above 200 ng DNA starting material in 25 µL elution buffer (>8 ng/µL). The library with the lowest normalized starting concentration for each sample had 17.8 ng/µL DNA, whereas the library with the highest starting material had 51.6 ng/µL DNA. We used digestion enzymes and protocols established with previous puma work (Trumbo et al., 2019). After DNA was normalized, we double-digested the DNA from each individual using NlaIII (New England BioLabs, #R0125S) and EcoRI (New England BioLabs, #R3101S) restriction enzymes (37 °C for 3 hours, then held at 4 °C) at manufacturer-recommended enzyme concentrations and used AMPure XP beads (Beckman Coulter, #A63881) at a 1.5X ratio to retain only DNA from the digestion. We omitted the Peterson et al. (2012) DynaBeads cleanup step and again used the Qubit to measure DNA concentrations and to guide another round of normalization. After normalization, the library with the lowest per-sample concentration had 2.1 ng/µL (in 29 µL) and the library with the highest per-sample concentration had 8.1 ng/µL.

We then ligated 48 uniquely barcoded P1 adaptors (e.g., P1.1 through P1.48) and two common P2 adapter pairs (i.e., P2.1 and P2.2) to each sample’s double-digested fragments using the protocols of Peterson et al. (2012) to identify individual puma samples. Following ligation with individual barcodes, we pooled all 48 samples into a single tube and used AMPure XP beads to clean the library. We used TE buffer (rather than molecular-grade water) as the final step in this cleanup, which is recommended by the manufacturer for running size selection in the Pippin Prep (Sage Science, Beverly, Massachusetts). We selected fragments ranging from 375– 475 bp (including 75 bp of adapters) using 2% dye-free gels run on a Pippin Prep. To minimize random polymerase chain reaction (PCR) duplicate errors, we split the library and ran 5 highfidelity Phusion (New England BioLabs, #M0530) PCRs for 12 cycles on a SimpliAmp thermal cycler (ThermoFisher Scientific, #A24811). We then recombined the 5 PCR products and used an AMPure XP bead clean up on the amplified library. Sample concentrations after size selection averaged 2.0 ng/µL DNA (range 0.82–3.7) and, after the PCR, averaged 8.2 ng/µL (range 3.6–

15.0). We shipped the unfrozen DNA with freezer packs to the University of Oregon’s Genomics and Cell Characterization Core Facility (https://gc3f.uoregon.edu/) for 150 bp single-end Accepted Article
sequencing on an Illumina HiSeq 4000 (Illumina, San Diego, CA).

This article is protected by copyright. All rights reserved Bioinformatic SNP filtering
We ran standard quality control analyses using program FastQC v0.11.5 (Andrews, 2010). We used the process_radtags program in the Stacks v2.55 (Catchen, Hohenlohe, Bassham, Amores, & Cresko, 2013) package to de-multiplex the reads based on unique barcodes, to assign each sequence to an individual puma sample, to remove sequences with a Phred quality score below 20 (99% accuracy), and to remove Illumina adapter sequences from the data. We then aligned reads for each individual to PumCon1.0—the Puma concolor draft reference genome —using program bwa (Li & Durbin, 2009). We identified and filtered SNPs with Samtools (Li et al., 2009). We discarded loci with a mapping quality score below 20, minimum base quality less than 20, with more than two alleles at a site, and with a maximum depth greater than 100. We skipped indels and used only a random SNP per read to reduce linkage disequilibrium.

Using vcftools, we tested the effects of multiple filtering parameters on our dataset, specifically looking at which parameters produced unreliable and inconsistent heterozygosity estimates, inbreeding coefficients, and relatedness values. We retained loci with a minor allele frequency ≥ 0.05 as lower frequency SNPs could be sequencing error. The relationship between minimum depth of reads per individual and heterozygosity was asymptotic and plateaued at about 3–4 reads. To be conservative, we selected a minimum depth of 4 reads per individual to reliably acquire genotypes based on both alleles. We also retained SNPs that had genotypes for at least 50% of the individuals. We iteratively removed samples with more than 50% missing data to maximize the number of SNPs retained in the dataset. Being more conservative with the percent of missing data decreased the number of SNPs in the final dataset but did not affect heterozygosity estimates, inbreeding coefficients, and relatedness values. We scanned for duplicate samples using relatedness values in vcftools, but, as expected, found none because all DNA samples were removed from dead pumas. We also removed two potentially contaminated samples based on negative F statistics in vcftools.

In each library of 48 samples, we strategically included puma samples from across a large geographic area so libraries would have no correlation with spatial location. For example, there was no significant difference between mean sample latitudes (F7;309 = 1.108, p = 0.358) or longitudes (F7;309 = 1.533, p = 0.155) among libraries. However, because the southern California Accepted Article
libraries constructed from pre-existing extracts were from a small geographic region, there ended This article is protected by copyright. All rights reserved up being some latitudinal (F10;395 = 33.76, p < 0.001) and longitudinal (F10;395 = 33.89, p < 0.001) mean differences between those libraries and the libraries constructed from the new samples.

However, as indicated below, there were no detectable biases of including the southern California libraries in any analyses.

To test for library-effect biases (i.e., differences among sequencing lanes), we used BayeScan to identify outlier SNPs while treating sequencing lanes as “populations” and using a false discovery rate of 0.05 (Foll & Gaggiotti, 2008). There were no outlier loci among any of the libraries, including the southern California libraries. We also assessed bias with various genetic structure analyses. Genotypes resulting from the pre-existing DNA extracts consistently clustered with those genotypes resulting from the new samples collected from southern California. With no apparent library-effect biases, we retained 16,285 bi-allelic variants (mean ± SD = 12,245 ± 2749) with a mean depth at each locus of 11.7 ± 5.1 and a mean depth per locus per individual of 11.7 ± 7.1.

Population structure and outlier loci
We used multiple approaches to identify genetic clusters of individuals, including a linear principal components analysis (PCA) and a spatially-explicit population structure analysis in program R (R Core Team, 2020). We ran the PCA using adegenet 2.1.1 (Jombart, 2008) and the structure analysis in tess3r 1.1.0 (Caye, Deist, Martins, Michel, & François 2016). We used adegenet::colorplot to present linear structure identified by the first 3 principal component axes.

In tess3r, we ran 20 replicates for each K (1–20) at 100,000 iterations each. We kept the most highly supported model (i.e., “best” based on cross-entropy scores) within each of the 20 replicates. To test for evidence of loci under selection, we identified outlier loci among populations (Narum & Hess, 2011) using BayeScan and tess3r with the Benjamini–Hochberg statistical correction and the recommended α-value of 0.0001.

Genetic diversity, effective population size, genetic differentiation, and linkage decay For each genetic cluster identified in tess3r, we calculated observed heterozygosity (HO), gene diversity (HS), and allelic richness (Ar) using hierfstat::basic.stats (Goudet, 2005; Nei, 1987). To test for Wahlund effects within broad-scale clusters, we used t-tests to test for Accepted Article
differences between HO and HS. We calculated private alleles (Ap) using poppr::private_alleles This article is protected by copyright. All rights reserved (Kamvar, Tabima, & Grünwald, 2014). We used NeEstimator 2.1 (Do et al., 2014) to estimate effective population size (Ne), using the linkage disequilibrium model, random mating, allele frequencies >0.05, and with a correction factor of 19 haploid chromosomes (Hsu, Rearden, & Luquette, 1963) as recommended by (Waples, Larson, & Waples, 2016). We used hierfstat::pairwise.neifst and hierfstat::pairwise.WCfst to estimate pairwise genetic differentiation based on FST according to Nei (1987) or Weir and Cockerham (1984).

We used Plink 2.0 (Purcell et al., 2007) to estimate linkage disequilibrium among loci (-- ld-window-r2 0 --ld-window 999999 --ld-window-kb 8000). To determine the level of nonrandom segregation of alleles across the genome, we assessed linkage decay in each genetic cluster by plotting the correlation of loci (R
2 ) based on genomic distance between SNPs. We
correlated loci using binned intervals of 100,000bp from 0 to the maximum scaffold size of PumCon1.0. Meiosis should break up linkage, resulting in low R 2 values. However, populations
experiencing strong selection, low mutation, inbreeding, low migration, or strong genetic drift will have higher R
2 values. In short, SNPs that are close together on chromosomes are expected to be correlated (i.e., inherited as chromosomal/haplotype segments), but SNPs far away are expected to assort randomly during recombination. However, if sequences are too similar, which they may be in small and inbred populations, we are not be able to detect events of crossing over despite their occurrence, resulting in higher estimates of linkage disequilibrium, which is still an important indicator of genetic diversity and Ne.

RESULTS
Population structure and outlier loci
We recovered 16,285 SNPs that were randomly distributed among 125 draft-genome scaffolds. The first three axes of the PCA accounted for 14.6% of the variance and indicated there were 4 broad-scale genetic clusters distributed across California (Fig. 2). When each puma was plotted on a map of California (Fig. 2A), the 4 clusters were geographically concordant with the Sierra Nevada (SN), North Coast (NC), Central Coast (CC), and Southern Coast (SC). The first eigenvector separated the negative-valued CC and SC groups from the positive-valued SN and NC (Figs. 2B & 2C). The second eigenvector separated negative-valued CC from positivevalued SC (Fig. 2B). Finally, the third eigenvector separated negative-valued NC from all other Accepted Article
groups (Fig. 2C).

This article is protected by copyright. All rights reserved A spatially-explicit population structure analysis indicated that there was broad- to finescale nested genetic structure with support for 4–10 genetic clusters (Fig. 3). Root mean square error (inset plot in K = 2 panel of Fig. 3) and cross-entropy scores (inset plot in K = 3 panel of Fig. 3) provide statistical evidence for nested genetic structure; values begin to curve at K = 4 and there is a major increase in variance at K = 5, but there is a steady increase in statistical support at higher K values. However, single pumas formed individual clusters at K > 10 at which point K lost biological meaning. When K was set to 4, the genetic clusters corresponded to the broad-scale genetic groups identified by the PCA (Figs. 2 & 3). Briefly, at K = 5, pumas in the Central Coast North (CC-N) emerged; at K = 6, the Eastern Sierra Nevada (ESN) cluster separated from the Western Sierra Nevada (WSN); at K = 7, the Santa Ana (SA) cluster separated from the Eastern Peninsular Range (EP); at K = 8, the San Gabriel–San Bernardino (SGSB) cluster emerged; at K = 9, the Klamath–Cascades (KC) cluster emerged; and at K = 10, the Central Coast South (CC-S) cluster separated from Central Coast Central (CC-C; Fig. 3). We observed no significant evidence for outlier loci using the Benjamini–Hochberg statistical correction in tess3r nor BayeScan for either K = 4 or K = 10.

Genetic diversity, effective population size, genetic differentiation, and linkage decay For K = 4, calculations of observed heterozygosity (HO), gene diversity (HS), polymorphic loci (Poly), allelic richness (Ar), and the private alleles (Ap) indicate that the Sierra Nevada cluster had higher genetic diversity than the Southern Coast, Central Coast, and North Coast (Table 1). Although significant, the North Coast was the only broad-scale genetic cluster that did not exhibit a strong Wahlund effect (i.e., significantly lower HO compared to HS; SN: t=-

50.6, p<0.001; SC: t=-48.2, p<0.001; CC: t=-58.5, p<0.001; NC: t=-10.6, p<0.001) or finer-scale substructure. Effective population sizes were not reported for broad-scale clusters because substructure introduced major biases (i.e., near-zero values) into Ne estimates.

Broad-scale genetic clusters were moderately differentiated based on FST estimates which ranged from ~0.1–0.2 (Table 2). The Sierra Nevada cluster was least differentiated from the others and the lowest FST estimates were between the Sierra Nevada and the North Coast clusters. In contrast, the Southern Coast cluster was the most differentiated from the others and Accepted Article
the highest FST estimates were between the Southern Coast and the North Coast, followed by the This article is protected by copyright. All rights reserved Southern Coast and the Central Coast. At the broad scale, the linkage decay plot indicated that linkage disequilibrium (LD) was lowest in the Sierra Nevada and slightly increased in the Central Coast, Southern Coast, and North Coast clusters (Fig. 4A). When ignoring population assignments, California pumas (N = 401) had a LD R
2 of ~0.3 which decreased rapidly to less
than 0.1 at a distance of 0.3 Mbp, then approached 0 at farther distances. Nearly the same result was observed in the Sierra Nevada. The Central Coast also had a major reduction in LD with distance but did not fall under 0.1 until ~3 Mbp in distance. In contrast, the Southern Coast and North Coast started with an LD R
2 of ~0.4 which remained above 0.1 even at distance of 8 million bp (Fig. 4A).

The nested genetic clusters within the Sierra Nevada — including KC, WSN, and ESN — had the highest genetic diversity estimates, as well as the highest estimates of Ne. Only the WSN had an Ne above 50, a threshold commonly considered to be sustainable over the long-term (Table 1; Franklin, 1980). Pairwise FST estimates among nested genetic clusters within the Sierra Nevada suggested weak substructure, with little genetic differentiation (i.e., pairwise FST < 0.05), indicating substantial gene flow throughout this region (Table 2). Within the Sierra Nevada, the ESN showed slightly higher LD than KC or WSN, and all three retained a high proportion of polymorphic loci (i.e., 87–91%).

The nested genetic clusters within the Southern Coast — including EP, SGSB, and the SA — exhibited lower genetic diversity estimates when compared to the Sierra Nevada, as well as large differences when compared to each other (Table 1). Estimates were generally lowest in SA, whereas EP and SGSB had similar overall estimates. However, both SA and SGSB had extremely low estimates of Ne. Unlike the Sierra Nevada, nested genetic clusters within the Southern Coast had moderate to strong genetic differentiation from one another (pairwise FST values ~0.1–0.2; Table 2). Except for the moderate differentiation with EP (i.e., pairwise FST of ~0.1), SA was the most differentiated among the 10 finer-scale genetic clusters (pairwise FST values range: ~0.2–0.3). The SGSB cluster had relatively lower pairwise FST estimates with the Sierra Nevada and EP clusters, moderate FST estimates with CC-C and CC-S, and was more strongly differentiated from the CC-N and NC. The EP cluster showed similar patterns of differentiation but was least differentiated from the geographically adjacent SA and SGSB Accepted Article
clusters. Although EP exhibited LD estimates similar to the Southern Coast as a whole, SGSB This article is protected by copyright. All rights reserved and SA started with a high LD R
2 of ~0.5 which decreased to just above 0.3 at a distance of 0.3 Mbp, then remained high (above 0.25) at farther distances (Fig. 4).

The nested genetic clusters within the Central Coast exhibited the most variation in estimates of genetic diversity (Table 1). The CC-C cluster had the highest diversity within the region, including the largest estimate of Ne. The CC-S cluster had intermediate levels of diversity but exhibited the lowest Ne in the region. The CC-N cluster had as low, or lower, genetic diversity estimates than most of the 10 fine-scale genetic clusters examined overall, but had one of the higher Ne estimates outside of the Sierra Nevada. Differentiation within the Central Coast was moderate overall (pairwise FST ~0.06–0.15) and appeared to correlate with distance (i.e., CC-N more differentiated from CC-S than CC-C; Table 2). Within the Central Coast, CC-C had the lowest LD R
2 values (Fig. 4). The CC-N cluster had higher LD values, especially at lower distances between SNPs, and CC-S had among the highest LD R2values, comparable to those of SGSB and SA in the Southern Coast.

Finally, the NC had genetic diversity estimates that were lower than the Sierra Nevada and comparable to the Southern Coast and Central Coast, with an Ne estimate of 14.1 (Table 1).

Overall, the NC showed strong differentiation from the other fine-scale genetic clusters with the exception of KC and WSN for which differentiation was moderate (Table 2). The linkage decay plot indicates the NC had similar LD R
2 values to that of ESN and EP (Fig. 4).

DISCUSSION
Our analyses of genetic diversity and linkage disequilibrium based on 16,285 SNPs from 401 pumas throughout California demonstrated that the complex geography and land use patterns in California result in equally complex patterns of gene flow and population structure. The highdensity SNP data provided resolution to detect both four broad-scale genetic clusters with high genetic diversity, as well as substructure at a finer scale that we designate as 10 genetic populations with highly variable genetic diversity. Our data further support the notion that puma populations in California form a “horseshoe” network around the Central Valley with San Francisco Bay acting as a barrier to gene flow along the coast (Gustafson et al., 2019). For the Sierra Nevada cluster, the nested finer-scale populations had consistently high genetic variation.

However, within the coastal groups, genetic variation within certain fine-scale genetic Accepted Article
populations was concerningly low, while others appeared to have retained sufficient variation as This article is protected by copyright. All rights reserved to be capable of serving as sources of genetic rescue under various management scenarios to restore connectivity. In fact, our linkage decay analysis indicated that populations with low genetic diversity and high linkage disequilibrium may not necessarily share the same fixed loci, consistent with what was suggested by Saremi et al. (2019). Specifically, when individuals from nested populations were combined within the 4 broader-scale groups, linkage decay values were much lower, indicating variation still exists among populations. Therefore, maintaining and enhancing connectivity within and among broad-scale groups could increase genetic diversity to entire regions and could decrease the apparent effects of genetic drift and inbreeding to some atrisk coastal populations (Ernest et al., 2003; 2014; Gustafson et al., 2017; Riley et al., 2014).

The support for four broad-scale genetic groups from SNPs is different than previous studies using microsatellites (Ernest et al., 2003; Gustafson et al., 2019) indicating the importance of using genomic methods in the study of broader-scale wildlife conservation genetics. Our data further support the claim that the Sierra Nevada region is a major refugium of puma genetic diversity in California (Gustafson et al., 2019). Therefore, it is important to protect the Sierra Nevada group from habitat degradation and foster conservation actions that can enhance gene flow with the North Coast, Central Coast, and Southern Coast clusters as well as with the Great Basin to the east (Gustafson et al., 2019). The broad-scale Southern Coast group is least connected to the other genetic clusters in the state but had higher genetic diversity and more private alleles than the Central Coast or North Coast. This indicates that the Southern Coast group retains unique genomic variation that must be conserved in order to maximize genetic diversity among pumas in California. Further, our finding of greater genetic diversity at lower latitudes is consistent with a previous study of gene flow among puma populations across southwestern North America, which found both higher microsatellite allelic diversity and a greater number of private alleles among pumas in southern Arizona and New Mexico (McRae, Beier, Dewald, Huynh, & Keim, 2005). Those authors suggested the pattern was consistent with recolonization of North America following a late-Pleistocene extinction (Culver et al., 2000); range expansion from the south was accompanied by decreasing diversity in more northern populations because of serial founder events. Our finding of high genetic diversity in the Southern Coast group suggests the genetic legacy of recolonization is generally consistent across Accepted Article
the contemporary range of pumas in North America.

This article is protected by copyright. All rights reserved Although the four major genetic clusters are highly consistent among our structure analysis and PCA, there was also statistical support for substructure (i.e., tess3r results and moderate to high pairwise FST values within and among the broad-scale groups) indicating 10 genetic populations at a finer scale. Generally, the 10 genetic populations identified with SNPs correspond strongly to those identified in previous studies using microsatellite markers and different samples (Ernest et al., 2003, Gustafson et al., 2019). However, the northern-most Klamath–Cascade population was not observed previously with microsatellites (Gustafson et al., 2019). This is likely because there were very few pumas available for analysis in the Klamath or Cascade Mountains during the 2019 microsatellite study. It is also possible that 42 microsatellites may not have been sufficient to detect the low genetic differentiation (FST =

0.022) observed between the Klamath–Cascade and Western Sierra Nevada populations. The 10 populations varied considerably in genetic diversity estimates (HO range 0.22–0.32; HS range

0.24–0.33; Poly range: 0.63–0.91; Ar range: 1.24–1.33), effective population sizes (Ne range 2.3–

54.4), and genetic differentiation (FST range: 0.22–0.32), as discussed below.

A major difference between this and previous studies is the observation that pumas in the Central Coast North population have genetic diversity estimates as low as those in the Santa Ana and Central Coast South populations, which are highly isolated by urbanization and transportation infrastructure and exhibit evidence of inbreeding depression (Benson et al., 2020; Ernest et al., 2014; Gustafson et al. 2017; Riley et al., 2014; Vickers et al., 2015). Our results are consistent with those of Saremi et al. (2019), which indicated that inbreeding metrics between pumas from the Santa Monica Mountains (in Central Coast South) and pumas from the Santa Cruz Mountains (in Central Coast North) were similar. Interestingly, Ne for the Central Coast North was much higher than both the Santa Ana and Central Coast South populations. These observations are consistent with a large breeding population experiencing genetic drift due to dispersal barriers to the north (i.e., San Francisco Bay) and gene flow occurring only with the Central Coast Central population to the south. This pattern could also be driven by carrying capacity processes associated with habitat limitations (Dellinger, Gustafson, Gammons, Ernest, & Torres, 2020b). If dispersal is limited by continued development southeast of the Central Coast North population, rapid genetic drift and inbreeding may ensue (Mills & Allendorf, 1996; Wang, 2004) and local extinctions may occur as predicted in the Central Coast South and Santa Accepted Article
Ana populations (Benson et al., 2016; 2019). Thus, puma population viability will be facilitated This article is protected by copyright. All rights reserved when land management agencies and land developers in the region work proactively to preserve or enhance wildlife corridors.

Notably, the San Gabriel–San Bernardino population had the lowest Ne, but had intermediate levels of genetic diversity. Occasional migrants could alter Ne estimates and temporarily inflate estimates of heterozygosity (Gustafson et al., 2017). We suggest this could also be the result of metapopulation dynamics—i.e., a small local population with frequent turnover located at the intersection of dispersal corridors for the Sierra Nevada, Central Coast, and Southern Coast groups. Although the genetics of this population are complex and somewhat uncertain, this region is of critical importance for maintaining statewide puma gene flow.

Enhancing connectivity through the Transverse Ranges (including the Tehachapi Mountains, Sierra Pelona, San Gabriel Mountains, and San Bernardino Mountains; Fig. 1B) is a critical conservation priority in order to maintain gene flow between the Southern Coast populations and the Sierra Nevada or Central Coast groups.

The three populations with the lowest Ne, including the San Gabriel–San Bernardino, Santa Ana, and Central Coast South populations, have the smallest available amount of habitat (Dellinger et al., 2020b), and had the highest linkage disequilibrium throughout their genomes.

As we observed, there was great variation among populations in the decay curves, with the Central Coast North population having the next highest linkage disequilibrium after these three populations. Given the genetic diversity, Ne, and linkage data, the San Gabriel–San Bernardino and Central Coast North populations may be approaching levels of genetic drift and inbreeding similar to the well-monitored and genetically depauperate Santa Ana and Central Coast South populations (Ernest et al., 2014; Gustafson et al., 2017; Riley et al. 2014).

Populations with intermediate genetic diversity include the North Coast, Central Coast Central, and Eastern Peninsular Range. Measures of genetic diversity were lower than expected for the North Coast population given there are no obvious anthropogenic barriers to gene flow with the Klamath–Cascade, Western Sierra Nevada, or pumas from Oregon (Gustafson et al., 2019). However, the majority of our samples from this genetic cluster came from just north of the San Francisco Bay, an area of substantial human density and restricted gene flow on three sides. Thus our results may not be truly representative of this region as a whole and may represent the most isolated pumas on a “peninsula” of habitat. Future studies would benefit from Accepted Article
increased sampling throughout this genetic cluster, north to (and including) Oregon. Nonetheless, This article is protected by copyright. All rights reserved pumas and other animals would benefit if decisions for future development between the North Coast and Sierra Nevada consider the future connectivity of private timber land holdings along the coast with the inland National Forests.

The Central Coast Central population has ample habitat for maintaining a breeding population (Dellinger et al., 2020b). Given the apparent absence of gene flow across the Central Valley, this population may be the only consistent source of migrants for the Central Coast North and Central Coast South, which have concerningly low levels of genetic diversity and evidence of inbreeding. Thus, we consider the Central Coast Central population to be essential for the long-term viability of both adjacent populations and urge that habitat in this region is not fragmented further.

Despite having less than half of the overall habitat of the Central Coast Central population (Dellinger et al., 2020b), the Eastern Peninsular Range population has roughly similar genetic diversity estimates, but a much lower Ne. Dispersal in and out of the Eastern Peninsular Range is extremely limited and the degree to which pumas disperse across the border between USA and Mexico remains unknown (Gustafson et al., 2019). Given that the Eastern Peninsular Range is the only population known to exchange individuals with the Santa Ana population, management actions which enhance gene flow between these areas remain critical to the recovery of pumas in the Santa Ana Mountains.

Our linkage decay analysis suggests that in the Central Coast South, San Gabriel–San Bernardino, Santa Ana, and perhaps the Central Coast North populations, pumas may have long runs of homozygosity that are identical-by-descent. This is consistent with the genome resequencing results of Saremi et al. (2019) in the Santa Cruz (i.e., Central Coast North) and Santa Monica Mountains (i.e., Central Coast South), which suggested close and recent inbreeding led to runs of homozygosity. Although Saremi et al. (2019) sequenced individuals only from California populations known to have low genetic diversity, our linkage decay results from populations throughout the state indicate that the genome-level problems of inbreeding are not universal throughout California. Instead, the Klamath–Cascades, Western Sierra Nevada, Eastern Sierra Nevada, Central Coast Central, and the Eastern Peninsular Range populations all have low linkage disequilibrium throughout the genome. Additionally, when the inbred populations are analyzed with their broad-scale group, linkage decay curves demonstrated the Accepted Article
potential for gene flow with adjacent populations to reduce linkage to negligible levels. We This article is protected by copyright. All rights reserved observed up to 30–37% of the SNPs as fixed in the Central Coast–South, Santa Ana, and Central Coast North populations. Our linkage decay curves and the resequencing results of Saremi et al. (2019) demonstrate that fixed regions of the genome often differ among populations. Thus, genetic restoration is possible even among genetically depauperate populations. When considering that genetic diversity is much higher in several California puma populations than in those heavily studied along urban coasts, there is high potential for the long-term persistence of pumas throughout the majority of the state.

Genetic restoration or rescue has been successfully demonstrated for isolated, large-felid populations, such as the African lion (Panthera leo; Miller et al., 2020) and Florida panther (P. concolor; Ralls et al., 2018). There has also been calls for genetic rescue of other large-felids such as isolated populations of tigers (Panthera tigris; Armstrong et al., 2021) and leopards (Panthera pardus; Perez, Geffen, & Mokady, 2006). Thus, it is becoming increasingly evident that large-bodied cats and other apex predators will need habitat and connectivity for long-term evolutionary survival. Natural events of genetic restoration among fragmented populations of pumas in California (Ernest et al., 2014; Gustafson et al. 2017; Riley et al., 2014) combined with our linkage decay analysis indicates pumas and other apex predators may need to be managed in a metapopulation framework that incorporates genomic data (Farquharson et al., 2021).

We tested for outlier loci using multiple methods (Narum & Hess, 2011) but found no evidence of local adaptation when K = 4 or K = 10. Detection of outlier loci with RADseq is limited by the reduced representation of the genome, yet it has often been shown to be an effective approach (Catchen et al., 2017). Pumas are long-distance dispersers (Hawley et al., 2016; Sweanor, Logan, & Hornocker, 2000) and inhabit all major mountain ranges in California (Dellinger et al., 2020b), suggesting local adaptation may be unlikely. Our results provide preliminary evidence that outbreeding depression resulting from potential active genetic management may be of minimal concern (Frankham et al., 2011). Recent modelling (Kyriazis, Wayne, & Lohmueller, 2021) does suggest, however, that attempts to maximize genetic diversity in a population can introduce hidden deleterious recessive mutations, enhancing extinction risk.

The modelling of Kyriazis et al. (2021) has faced criticisms (Garcia-Dorado & Caballero, 2021), however, and Ralls, Sunnucks, Lacy, and Frankham (2020) argue that the benefits of increasing genetic diversity outweigh the risks. Thus, managers could consider actions (e.g., wildlife Accepted Article
overpasses/underpasses, translocation of individuals between populations, etc.) to improve This article is protected by copyright. All rights reserved viability of some coastal populations, as was empirically demonstrated to have shifted the trajectory of Florida panther population from extinction (Ralls et al., 2018). However, we suggest whole genome resequencing methods better suited for detecting selection (Fuentes-Pardo & Ruzzante, 2017) be implemented before such efforts, especially over long distances. Managers would also need to consider other risks as well, such as the movement of pathogens or the ethical implications of moving large carnivores (Bevins et al., 2012). Wildlife managers will have to weigh these concerns against their obligation to minimize the risks of extirpation such as those predicted for the Santa Ana and Central Coast South populations (Benson et al., 2019), and shown here to be a concern in the Central Coast North population as well. Should connectivity be re-established, then these factors, as well as possible local adaptation, should be weighed carefully. It is our opinion that current efforts to construct or improve wildlife crossing structures that can facilitate natural movement among coastal populations should be considered the primary management strategy for conserving viable puma populations in that region.

CONCLUSION
Our population genomic analyses provide decision makers a contemporary and thorough evaluation of the genetic diversity, effective population sizes, and connectivity of puma populations throughout California. These state- and genome-wide results are critically important for conservation and management practices in California, especially considering the increasing demand for development and the current political climate surrounding the petition to list pumas in Southern and Central California as Threatened under the California Endangered Species Act (Yap & Rose, 2019). In brief, puma populations are widespread throughout the mountains of California. Populations range from major genetic sources to populations with issues of low genetic diversity and inbreeding. Multiple lines of evidence suggest inbred populations do not share the same runs of homozygosity and therefore genetic diversity could be restored through enhanced gene flow. Current challenges to puma populations are highly regional and should be addressed by focusing on how natural geography and human development impacts puma habitat and movements locally. Attention is understandably given to those populations that are highly imperiled, but it is important to note that California has several thriving populations throughout the state which represent an important resource for any genetic management strategy. Protecting Accepted Article
tracts of contiguous habitat to preserve large populations will provide greater protection for the This article is protected by copyright. All rights reserved species as a whole. Specifically, further fragmentation of habitat in the Sierra Nevada group could be catastrophic to population viability of pumas in the state because it serves as a genetic refugium. Protecting, enhancing, and creating movement corridors to allow statewide “steppingstone” connectivity at broad and fine scales will allow for the migrants needed to counteract the local extirpations faced by some coastal populations.

REFERENCES
Advanced Research Computing Center. (2018). Teton Computing Environment. University of Wyoming. doi: 10.15786/M2FY47
Andrews, S. (2010). FastQC: A quality control tool for high throughput sequence data Retrieved from http://www.bioinformatics.babraham.ac.uk/projects/fastqc/.

Armstrong, E. E., Khan, A., Taylor, R. W., Gouy, A., Greenbaum, G., Thiéry, A., ... & Ramakrishnan, U. (2021). Recent evolutionary history of tigers highlights contrasting roles of genetic drift and selection. Molecular Biology and Evolution, 38(6), 2366-2379.

Benson, J. F., Mahoney, P. J., Sikich, J. A., Serieys, L. E., Pollinger, J. P., Ernest, H. B., & Riley, S. P. (2016). Interactions between demography, genetics, and landscape connectivity increase extinction probability for a small population of large carnivores in a major metropolitan area. Proceedings of the Royal Society B: Biological Sciences, 283(1837), 20160957. doi: 10.1098/rspb.2016.0957
Benson, J. F., Mahoney, P. J., Vickers, T. W., Sikich, J. A., Beier, P., Riley, S. P., ... & Boyce, W. M. (2019). Extinction vortex dynamics of top predators isolated by urbanization. Ecological Applications, 29(3), e01868. doi: 10.1002/eap.1868 Benson, J. F., Sikich, J. A., & Riley, S. P. (2020). Survival and competing mortality risks of mountain lions in a major metropolitan area. Biological Conservation, 241, 108294. doi:

10.1016/j.biocon.2019.108294
Bevins, S. N., Carver, S., Boydston, E. E., Lyren, L. M., Alldredge, M., Logan, K. A., ... & Accepted Article
VandeWoude, S. (2012). Three pathogens in sympatric populations of pumas, bobcats, This article is protected by copyright. All rights reserved and domestic cats: implications for infectious disease transmission. PLoS One, 7(2), e31403. doi: 10.1371/journal.pone.0031403
Burdett, C. L., Crooks, K. R., Theobald, D. M., Wilson, K. R., Boydston, E. E., Lyren, L. M., ... & Boyce, W. M. (2010). Interfacing models of wildlife habitat and human development to predict the future distribution of puma habitat. Ecosphere, 1(1), 1-21. doi:

10.1890/ES10-00005.1
Catchen, J., Hohenlohe, P. A., Bassham, S., Amores, A., & Cresko, W. A. (2013). Stacks: an analysis tool set for population genomics. Molecular ecology, 22(11), 3124-3140. doi:

10.1111/mec.12354
Catchen, J. M., Hohenlohe, P. A., Bernatchez, L., Funk, W. C., Andrews, K. R., & Allendorf, F.

W. (2017). Unbroken: RADseq remains a powerful tool for understanding the genetics of adaptation in natural populations. Molecular ecology resources, 17(3), 362-365. doi:

10.1111/1755-0998.12669
Caye, K., Deist, T. M., Martins, H., Michel, O., & François, O. (2016). TESS3: fast inference of spatial population structure and genome scans for selection. Molecular Ecology Resources, 16(2), 540-548. doi: 10.1111/1755-0998.12471 Culver, M., Johnson, W. E., Pecon-Slattery, J., & O'Brien, S. J. (2000). Genomic ancestry of the American puma (Puma concolor). Journal of Heredity, 91(3), 186-197. doi:

10.1093/jhered/91.3.186
Dellinger, J. A., Cristescu, B., Ewanyk, J., Gammons, D. J., Garcelon, D., Johnston, P., ... & Torres, S. G. (2020a). Using mountain lion habitat selection in management. The Journal of Wildlife Management, 84(2), 359-371. doi: 10.1002/jwmg.21798 Dellinger, J. A., Gustafson, K. D., Gammons, D. J., Ernest, H. B., & Torres, S. G. (2020b).

Minimum habitat thresholds required for conserving mountain lion genetic diversity. Ecology and Evolution, 10(19), 10687-10696. doi: 10.1002/ece3.6723 Do, C., Waples, R. S., Peel, D., Macbeth, G. M., Tillett, B. J., & Ovenden, J. R. (2014).

NeEstimator v2: re‐ implementation of software for the estimation of contemporary effective population size (Ne) from genetic data. Molecular ecology resources, 14(1), 209-214. doi: 10.1111/1755-0998.12157
Accepted Article
This article is protected by copyright. All rights reserved Ernest, H. B., Boyce, W. M., Bleich, V. C., May, B., Stiver, S. J., & Torres, S. G. (2003).

Genetic structure of mountain lion (Puma concolor) populations in California. Conservation Genetics, 4(3), 353-366. doi: 10.1023/A:1024069014911 Ernest, H. B., Vickers, T. W., Morrison, S. A., Buchalski, M. R., & Boyce, W. M. (2014).

Fractured genetic connectivity threatens a southern California puma (Puma concolor) population. PloS one, 9(10), e107985. doi: 10.1371/journal.pone.0107985 Farquharson, K. A., McLennan, E. A., Wayne, A., Smith, M., Peel, E., Belov, K., & Hogg, C. J. (2021). Metapopulation management of a critically endangered marsupial in the age of genomics. Global Ecology and Conservation, 31, e01869.

Foll, M., & Gaggiotti, O. (2008). A genome-scan method to identify selected loci appropriate for both dominant and codominant markers: a Bayesian perspective. Genetics, 180(2), 977- 993. doi: 10.1534/genetics.108.092221
Frankham, R., Ballou, J. D., Eldridge, M. D., Lacy, R. C., Ralls, K., Dudash, M. R., & Fenster, C. B. (2011). Predicting the probability of outbreeding depression. Conservation Biology, 25(3), 465-475. doi: 10.1111/j.1523-1739.2011.01662.x Franklin, I. R. (1980). Evolutionary change in small populations. In M.E. Soule & B.A. Wilcox (Eds.), Conservation biology: an evolutionary-ecological perspective (pp. 135–149).

Sunderland, Massachusetts: Sinauer Associates
Fuentes‐ Pardo, A. P., & Ruzzante, D. E. (2017). Whole‐ genome sequencing approaches for conservation biology: Advantages, limitations and practical recommendations. Molecular ecology, 26(20), 5369-5406. doi: 10.1111/mec.14264
García-Dorado, A., & Caballero, A. (2021). Neutral genetic diversity as a useful tool for conservation biology. Conservation Genetics, 1-5. doi: 10.1007/s10592-021-01384-9 Gonzalez‐ Borrajo, N., López‐ Bao, J. V., & Palomares, F. (2017). Spatial ecology of jaguars, pumas, and ocelots: a review of the state of knowledge. Mammal review, 47(1), 62-75. doi: 10.1111/mam.12081
Goudet, J. (2005). Hierfstat, a package for R to compute and test hierarchical F‐ statistics. Molecular Ecology Notes, 5(1), 184-186. doi: 10.1111/j.1471- 8286.2004.00828.x
Accepted Article
This article is protected by copyright. All rights reserved Guerisoli, M. D. L. M., Luengos Vidal, E., Caruso, N., Giordano, A. J., & Lucherini, M. (2021).

Puma–livestock conflicts in the Americas: A review of the evidence. Mammal Review, 51(2), 228-246. doi: 10.1111/mam.12224
Gustafson, K. D., Gagne, R. B., Vickers, T. W., Riley, S. P., Wilmers, C. C., Bleich, V. C., ... & Ernest, H. B. (2019). Genetic source–sink dynamics among naturally structured and anthropogenically fragmented puma populations. Conservation Genetics, 20(2), 215-227. doi: 10.1007/s10592-018-1125-0
Gustafson, K. D., Vickers, T. W., Boyce, W. M., & Ernest, H. B. (2017). A single migrant enhances the genetic diversity of an inbred puma population. Royal Society open science, 4(5), 170115. doi: 10.1098/rsos.170115
Hawley, J. E., Rego, P. W., Wydeven, A. P., Schwartz, M. K., Viner, T. C., Kays, R., ... & Jenks, J. A. (2016). Long-distance dispersal of a subadult male cougar from South Dakota to Connecticut documented with DNA evidence. Journal of Mammalogy, 97(5), 1435-1440. doi: 10.1093/jmammal/gyw088
Hedrick, P. (2005). ‘Genetic restoration:’ a more comprehensive perspective than ‘genetic rescue’. Trends in Ecology & Evolution, 20(3), 109. doi: 10.1016/j.tree.2005.01.006 Herrero, S. (1970). Man and the grizzly bear (present, past, but future?). BioScience, 1148-1153. doi: 10.2307/1295334
Hsu, T. C., Rearden, H. H., & Luquette, G. F. (1963). Karyological studies of nine species of Felidae. The American Naturalist, 97(895), 225-234. doi: jstor.org/stable/2458703 Huffmeyer, A. A., Sikich, J. A., Vickers, T. W., Riley, S. P. D., & Wayne, R. K. (2021). First reproductive signs of inbreeding depression in Southern California male mountain lions (Puma concolor). Theriogenology, In press
Jombart, T. (2008). adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics, 24(11), 1403-1405. doi: 10.1093/bioinformatics/btn129 Jordan, N., Boody, G., Broussard, W., Glover, J. D., Keeney, D., McCown, B. H., ... & Wyse, D. (2007). Sustainable development of the agricultural bio-economy. Science 316(5831), 1570. doi: 10.1126/science.1141700
Kamvar, Z. N., Tabima, J. F., & Grünwald, N. J. (2014). Poppr: an R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ, 2, Accepted Article
e281. doi: 10.7717/peerj.281
This article is protected by copyright. All rights reserved Kiesecker, J. M., Evans, J. S., Fargione, J., Doherty, K., Foresman, K. R., Kunz, T. H., ... & Niemuth, N. D. (2011). Win-win for wind and wildlife: a vision to facilitate sustainable development. PLoS One, 6(4), e17566. doi: 10.1371/journal.pone.0017566 Kyriazis, C. C., Wayne, R. K., & Lohmueller, K. E. (2021). Strongly deleterious mutations are a primary determinant of extinction risk due to inbreeding depression. Evolution letters, 5(1), 33-47. doi: 10.1002/evl3.209
Li, H., & Durbin, R. (2009). Fast and accurate short read alignment with Burrows–Wheeler transform. bioinformatics, 25(14), 1754-1760. doi: 10.1093/bioinformatics/btp324 Li, H., Handsaker, B., Wysoker, A., Fennell, T., Ruan, J., Homer, N., ... & Durbin, R. (2009).

The sequence alignment/map format and SAMtools. Bioinformatics, 25(16), 2078-2079. doi: 10.1093/bioinformatics/btp352
McRae, B. H., Beier, P., Dewald, L. E., Huynh, L. Y., & Keim, P. (2005). Habitat barriers limit gene flow and illuminate historical events in a wide‐ ranging carnivore, the American puma. Molecular Ecology, 14(7), 1965-1977. doi: 10.1111/j.1365-294x.2005.02571.x Miller, S. M., Druce, D. J., Dalton, D. L., Harper, C. K., Kotze, A., Packer, C., ... & Bloomer, P. (2020). Genetic rescue of an isolated African lion population. Conservation Genetics, 21(1), 41-53.

Mills, L. S., & Allendorf, F. W. (1996). The one‐ migrant‐ per‐ generation rule in conservation and management. Conservation biology, 10(6), 1509-1518. doi: 10.1046/j.1523- 1739.1996.10061509.x
Narum, S. R., & Hess, J. E. (2011). Comparison of FST outlier tests for SNP loci under selection. Molecular ecology resources, 11, 184-194. doi: 10.1111/j.1755- 0998.2011.02987.x
Nei, M. (1987). Molecular evolutionary genetics. Columbia, NY: Columbia university press.

Newbold, T., Hudson, L. N., Arnell, A. P., Contu, S., De Palma, A., Ferrier, S., ... & Purvis, A. (2016). Has land use pushed terrestrial biodiversity beyond the planetary boundary? A global assessment. Science, 353(6296), 288-291. doi: 10.1126 / science.aaf2201 Perez, I., Geffen, E., & Mokady, O. (2006). Critically endangered Arabian leopards Panthera pardus nimr in Israel: estimating population parameters using molecular scatology. Oryx, Accepted Article
40(3), 295-301.

This article is protected by copyright. All rights reserved Peterson, B. K., Weber, J. N., Kay, E. H., Fisher, H. S., & Hoekstra, H. E. (2012). Double digest RADseq: an inexpensive method for de novo SNP discovery and genotyping in model and non-model species. PloS one, 7(5), e37135. doi: 10.1371/journal.pone.0037135 Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M. A., Bender, D., ... & Sham, P.

C. (2007). PLINK: a tool set for whole-genome association and population-based linkage analyses. The American journal of human genetics, 81(3), 559-575. doi: 10.1086/519795 R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Ralls, K., Ballou, J. D., Dudash, M. R., Eldridge, M. D., Fenster, C. B., Lacy, R. C., ... & Frankham, R. (2018). Call for a paradigm shift in the genetic management of fragmented populations. Conservation Letters, 11(2), e12412. doi: 10.1111/conl.12412 Ralls, K., Sunnucks, P., Lacy, R. C., & Frankham, R. (2020). Genetic rescue: a critique of the evidence supports maximizing genetic diversity rather than minimizing the introduction of putatively harmful genetic variation. Biological Conservation, 251, 108784. doi:

10.1016/j.biocon.2020.108784
Riley, S. P., Serieys, L. E., Pollinger, J. P., Sikich, J. A., Dalbeck, L., Wayne, R. K., & Ernest, H.

B. (2014). Individual behaviors dominate the dynamics of an urban mountain lion population isolated by roads. Current Biology, 24(17), 1989-1994. doi:

10.1016/j.cub.2014.07.029
Roelke, M. E., Martenson, J. S., & O'Brien, S. J. (1993). The consequences of demographic reduction and genetic depletion in the endangered Florida panther. Current Biology, 3(6), 340-350. doi: 10.1016/0960-9822(93)90197-V
Saha, D., & Paterson, R. G. (2008). Local government efforts to promote the “Three Es” of sustainable development: survey in medium to large cities in the United States. Journal of Planning Education and Research, 28(1), 21-37. doi: 10.1177/0739456X08321803 Saremi, N. F., Supple, M. A., Byrne, A., Cahill, J. A., Coutinho, L. L., Dalén, L., ... & Shapiro, B. (2019). Puma genomes from North and South America provide insights into the genomic consequences of inbreeding. Nature communications, 10(1), 1-10. doi:

10.1038/s41467-019-12741-1
Schmidt, R. H. (1991). Gray wolves in California: their presence and absence. California Fish Accepted Article
and Game, 77(2), 79.

This article is protected by copyright. All rights reserved Suraci, J. P., Nickel, B. A., & Wilmers, C. C. (2020). Fine-scale movement decisions by a large carnivore inform conservation planning in human-dominated landscapes. Landscape Ecology, 35(7), 1635-1649. doi: 10.1007/s10980-020-01052-2 Sweanor, L. L., Logan, K. A., & Hornocker, M. G. (2000). Cougar dispersal patterns, metapopulation dynamics, and conservation. Conservation Biology, 14(3), 798-808. doi:

10.1046/j.1523-1739.2000.99079.x
Tallmon, D. A., Luikart, G., & Waples, R. S. (2004). The alluring simplicity and complex reality of genetic rescue. Trends in Ecology & Evolution, 19(9), 489-496. doi:

10.1016/j.tree.2004.07.003
Torres, S. G., Mansfield, T. M., Foley, J. E., Lupo, T., & Brinkhaus, A. (1996). Mountain lion and human activity in California: testing speculations. Wildlife Society Bulletin, 24(3), 451-460. doi: jstor.org/stable/3783326
Trumbo, D. R., Salerno, P. E., Logan, K. A., Alldredge, M. W., Gagne, R. B., Kozakiewicz, C.

P., ... & Funk, W. C. (2019). Urbanization impacts apex predator gene flow but not genetic diversity across an urban‐ rural divide. Molecular ecology, 28(22), 4926-4940. doi: 10.1111/mec.15261
United States Census Bureau. 2019. State population totals and components of change: 2010– 2019. Retrieved from https://www.census.gov/data/tables/timeseries/demo/popest/2010s-state-total.html Vickers, T. W., Sanchez, J. N., Johnson, C. K., Morrison, S. A., Botta, R., Smith, T., ... & Boyce, W. M. (2015). Survival and mortality of pumas (Puma concolor) in a fragmented, urbanizing landscape. PloS one, 10(7), e0131490. doi: 10.1371/journal.pone.0131490 Wang, J. (2004). Application of the one‐ migrant‐ per‐ generation rule to conservation and management. Conservation Biology, 18(2), 332-343. doi: 10.1111/j.1523- 1739.2004.00440.x
Waples, R. K., Larson, W. A., & Waples, R. S. (2016). Estimating contemporary effective population size in non-model species using linkage disequilibrium across thousands of loci. Heredity, 117(4), 233-240. doi: 10.1038/hdy.2016.60 Weir, B. S., & Cockerham, C. C. (1984). Estimating F-statistics for the analysis of population Accepted Article
structure. evolution, 1358-1370. doi: jstor.org/stable/2408641 This article is protected by copyright. All rights reserved Yap, T., Cummings, B., & Rose, J. P. (2019). A petition to list the Southern California/Central Coast evolutionarily significant unit (ESU) of mountain lions as threatened under the California Endangered Species Act (CESA). Center for Biological Diversity, Tucson, Arizona, USA and the Mountain Lion Foundation, Sacramento, California, USA.

Zeller, K. A., Vickers, T. W., Ernest, H. B., & Boyce, W. M. (2017). Multi-level, multi-scale resource selection functions and resistance surfaces for conservation planning: Pumas as a case study. PLoS One, 12(6), e0179570. doi: 10.1371/journal.pone.0179570

DATA ACCESSIBILITY
Individual genotype data and associated location data are provided on Open Science Framework (doi: XXXXXX to be made public upon acceptance for publication; temporary anonymous link for peer review: https://osf.io/huf4k/?view_only=c723d0623c5c441a98513af06ed3357d).

Accepted Article
This article is protected by copyright. All rights reserved Table 1. Heat map of genetic diversity statistics for K = 4 broad-scale and K = 10 nested finescale genetic clusters, including sample size (N), observed heterozygosity (HO); gene diversity (Hs), proportion of polymorphic loci out of 16,285 (Poly), allelic richness corrected for sample size (Ar), private alleles (Ap), and effective population size (Ne). Values for Ne are not presented for the K = 4 Sierra Nevada (SN), Southern Coast (SC), Central Coast (CC), or North Coast (NC) because of model assumption violations. There were no private alleles at K = 10, including Klamath–Cascades (KC), Western Sierra Nevada (WSN), Eastern Sierra Nevada (ESN), Eastern Peninsular (EP), San Gabriel–San Bernardino (SGSB), Santa Ana (SA), Central Coast Central (CC-C), Central Coast South (CC-S), Central Coast North (CC-N), and North Coast (NC). Heat map colors bound the minimum (white) and maximum (darkest grey) values within rows.

Genetic diversity Genetic Cluster
K = 4 SN SC CC NC
N 193 96 79 33
HO 0.31 0.26 0.24 0.26
Hs 0.34 0.29 0.28 0.27
Poly 0.93 0.79 0.77 0.78
Ar 1.79 1.69 1.67 1.67
Ap 37 34 17 0
K = 10 KC WSN ESN EP SGSB SA CC-C CC-S CC-N NC
N 53 110 27 66 13 25 27 17 35 28
HO 0.32 0.31 0.31 0.27 0.29 0.23 0.27 0.24 0.22 0.25
Hs 0.33 0.33 0.33 0.29 0.30 0.24 0.29 0.27 0.24 0.26
Poly 0.91 0.90 0.87 0.78 0.78 0.64 0.77 0.70 0.63 0.74
Ar 1.33 1.33 1.33 1.29 1.30 1.24 1.29 1.27 1.24 1.26
Ne 28.9 54.4 42.2 14.8 2.3 3.5 26.9 4.1 19.0 14.1
Accepted Article
This article is protected by copyright. All rights reserved Table 2. Heat map of mean pairwise genetic distance values for the broad-scale K = 4 and finescale K = 10 genetic clusters. Weir and Cockerham FST is presented below the diagonal and Nei’s FST is presented above the diagonal (WC\Nei). All pairwise FST estimates were significant (p < 0.001) based on a bootstrapping analysis using hierfstat::boot.ppfst. Sierra Nevada (SN), Southern Coast (SC), Central Coast (CC), North Coast (NC), Klamath–Cascades (KC), Western Sierra Nevada (WSN), Eastern Sierra Nevada (ESN), Eastern Peninsular (EP), San Gabriel–San Bernardino (SGSB), Santa Ana (SA), Central Coast Central (CC-C), Central Coast South (CCS), Central Coast North (CC-N). Heat map colors bound the minimum (white) and maximum (darkest grey) values either below or above the diagonals.

WC\Nei FST Genetic Cluster
K=4 SN SC CC NC
SN - 0.133 0.124 0.100
SC 0.129 - 0.173 0.198
CC 0.120 0.173 - 0.156
NC 0.094 0.196 0.156 -
K=10 KC WSN ESN EP SGSB SA CC-C CC-S CC-N NC
KC - 0.022 0.041 0.141 0.109 0.215 0.117 0.146 0.183 0.093 WSN 0.022 - 0.045 0.149 0.111 0.222 0.121 0.147 0.188 0.126 ESN 0.041 0.045 - 0.163 0.116 0.226 0.168 0.189 0.233 0.183 EP 0.141 0.146 0.166 - 0.130 0.100 0.164 0.196 0.231 0.214 SGSB 0.105 0.106 0.113 0.132 - 0.212 0.140 0.163 0.210 0.205 SA 0.202 0.203 0.221 0.095 0.217 - 0.254 0.287 0.319 0.301 CC-C 0.114 0.116 0.168 0.163 0.141 0.251 - 0.060 0.098 0.164 CC-S 0.137 0.136 0.183 0.192 0.164 0.289 0.059 - 0.148 0.202 CC-N 0.178 0.176 0.237 0.227 0.221 0.320 0.100 0.152 - 0.229 NC 0.090 0.118 0.183 0.211 0.210 0.300 0.164 0.203 0.230 - Accepted Article
This article is protected by copyright. All rights reserved Figure 1. Location of 401 sampled pumas used in analyses, including (A) sample distribution Accepted Article
across California, (B) geography of the mountain ranges surrounding the Los Angeles and San This article is protected by copyright. All rights reserved Diego regions, and (C) inset map showing the location of California in the United States of America.

Figure 2. Principal component analysis (PCA) of 401 pumas at 16,285 SNPs reveals four genetic clusters. (A) The colorplot (R package adegenet) of the PCA represents colors corresponding to a combination of the first 3 eigenvectors. The inset plot shows the proportion of the variance explained by shaded PC eigenvectors 1–3 compared to other eigenvectors. The color values are plotted at sample locations to demonstrate geographic structure. Colorplots of (B) PC1 and PC2 and (C) PC1 and PC3 resolved the 4 broad-scale genetic clusters.

Accepted Article
This article is protected by copyright. All rights reserved Figure 3. Interpolated ancestry proportions from tess3r, demonstrating the geographic distribution of biologically meaningful genetic clusters (K) ranging from 2–10. The “best” iterations of each K, based on cross-entropy score, is presented (shaded circles of inset plot in Accepted Article
K=2 panel). Root mean square error is also presented (inset plot in K=3 panel). Both tess3r and This article is protected by copyright. All rights reserved the PCA (Fig. 2) support K=4 and therefore the genetic clusters are labeled. At K=10, nested genetic clusters are labeled consistent with previous microsatellite data (Gustafson et al. 2019).

For visualization, at each K, the genetic cluster that emerges is labeled. In alphabetical order, acronyms include Central Coast Central (CC-C), Central Coast North (CC-N), Central Coast South (CC-S), Eastern Peninsular Range (EP), Eastern Sierra Nevada (ESN), Klamath–Cascades (KC), North Coast (NC), Santa Ana (SA), San Gabriel–San Bernardino (SGSB), and Western Sierra Nevada (WSN).

Figure 4. Correlation of SNPs with genomic distance, ranging from hundreds to 8 million nucleotides in distance. Based on pairwise estimates from 16,285 SNPs, linkage decay is presented for all 401 pumas sampled in California (All), from the K = 4 broad-scale genetic Accepted Article
clusters (A: North Coast, NC; Southern Coast, SC; Central Coast, CC; Sierra Nevada, SN), and This article is protected by copyright. All rights reserved from the K = 10 fine-scale genetic clusters (B–D). The nested and finer-scale clusters are presented within their corresponding broad-scale group. The NC is presented only in the first panel because it did not exhibit substructure. (B) Eastern Sierra Nevada (ESN), Klamath– Cascades (KC) and Western Sierra Nevada (WSN) are nested within SN. (C) Central Coast South (CC-S), Central Coast North (CC-N), and Central Coast Central (CC-C) are nested within CC. (D) San Gabriel–San Bernardino (SGSB), Santa Ana (SA), and Eastern Peninsular Range (EP) are nested within SC. In each figure, the dashed line represents the broadest-scale designation within the group.

Accepted Article