Université Laval is the official host and sponsor of the 2021 EURING Analytical meeting.

Cover photo credits:

Snowy Owl (Bubo scandiacus), G. Gauthier

Tree Swallow (Tachycineta bicolor), M. Bélisle

Greater Snow Goose (Anser caerulescens atlanticus), G. Gauthier

Ruby-throated Hummingbird (Archilochus colubris), D. Greene

Common Eider (Somateria mollissima), J.-F. Giroux

Piping Plover (Charadrius melodus), Creative Commons

Bicknell’s Thrush (Catharus bicknelli), P. DeMontminy and J.-F. Rousseau

Northern Saw-whet Owl (Aegolius acadicus), J. Gagnon

This is the program of the 2021 EURING Analytical Meeting held in Quebec City, Canada. Contents

Bienvenue à Québec (Welcome to Quebec city)4 Meeting format ...... 4 Organizing committee ...... 4 General information ...... 5

Timetable 6 Monday 31 May 2021 ...... 6 Tuesday 1 June 2021 ...... 7 Thursday 3 June 2021 ...... 7 Friday 4 June 2021 ...... 8 Monday 7 June 2021 ...... 8 Tuesday 8 June 2021 ...... 9 Thursday 10 June 2021 ...... 10 Friday 11 June 2021 ...... 10

List of Abstracts – Talks 12 Session 1. Building on the Eurasian-African Migration Atlas: towards robust quantitative analyses of avian movements ...... 12 Session 2: Spatially-explicit capture-mark-recapture analysis ...... 20 Session 3: Integral projection modelling ...... 25 Session 4: Animal movement ...... 30 Honored speaker: Pertti Saurola ...... 36 Session 5: Data integration and population analysis I ...... 38 Session 6: Population management ...... 45 Session 7: Survival estimation ...... 50 Session 8: Data integration and population analysis II ...... 56

List of Posters 61 Poster Session (Tuesday 8 June 2021) ...... 61

3 Bienvenue à Québec (Welcome to Quebec city)

Meeting format

We are looking forward to seeing you at the EURING 2021 meeting hosted by Université Laval in Québec city from May 31st to June 11th. The meeting is completely online. To avoid online strain, we have spread the eight half-day sessions over a 2-week period. We encourage participants to attend the meeting live, although recordings of the sessions will be available on the meeting platform.

Organizing committee

Committee chairs Marc J. Mazerolle (local chair) Fränzi Korner-Nievergelt and Duane R. Diefenbach (scientific chairs)

Maintainer of conference website and mailing list Evan Cooch

Local collaborators Clara Casabona i Amat Aurore Fayard Joëlle Spooner Barbara Vuillaume Marc Bélisle Gilles Gauthier Louis-Paul Rivest Heidi Turcot

Session chairs Stephen Baillie (Migration) J. Andrew Royle and Richard Chandler (Spatially-explicit capture-mark-recapture analysis) Emily Simmonds, Floriane Plard, and Erik Sandvig (Integral projection modelling) Roland Kays and Frances Buderman (Animal movement) Sarah Converse and Elise Zipkin (Data integration and population analysis) Mike C. Runge and Anna M. Tucker (Population management) Eleni Matechou and Fernando Colchero (Survival estimation)

4 General information

All sessions will be held online using Zoom within the EURING 2021 meeting platform on Slack. Each participant will receive an invitation from the EURING2021 Slack site at the email they used to register for the meeting. The Slack website will open one week before the start of the meeting.

Participants are encouraged to install Zoom on their computer before the meeting. Zoom can be downloaded freely from https://zoom.us/download for GNU/Linux, Mac, or Windows.

Talks will be recorded and available for viewing directly on the Slack site.

The poster session will be held on Tuesday 8 June 2021 12:50–14:50 on the meeting platform. Each poster will have a dedicated breakout room where you can interact with each poster presenter.

5 Timetable

All times are in Quebec time (Eastern Daylight Time, UTC-4:00). If unsure about the conversion to your own time zone, check https://www.thetimezoneconverter.com/.

KL: Keynote Lecture, HS: Honored Speaker, CT: Contributed Talk.

Monday 31 May 2021

10:00–10:15 Welcome and details on meeting logistics Building on the Eurasian-African Migration Atlas: towards robust quantitative analyses of avian movements Funding analyses of over a century of F. Spina bird ringing data: challenges, strategies 10:15–10:30 KL and potential for the profile and applied use of innovative analytical approaches The Eurasian African bird migration atlas S. Franks and S. Baillie 10:30–11:15 KL – a starting point for a more quantitative understanding of avian movements 11:15–11:25 Break Session chair: S. Baillie Patterns of migratory connectivity in the N. Fattorini European-African migratory system as 11:25–11:45 CT inferred by ring recovery analysis of the EURING database J. Hostetler Estimating migratory connectivity 11:45–12:05 CT strength and pattern across data types 12:05–12:15 Break Disentangling survival, migratory S. Schirmer 12:15–12:35 CT connectivity and observation in continuous space Quantifying individual variation in P. Acker migration versus residence within and 12:35–12:55 CT among year: a dynamic-finite-mixture multi-state capture-recapture model R. Ambrosini Assessing the timing of migration 12:55–13:15 CT progression based on ring recoveries 13:15–13:25 Break 13:25–14:25 Interaction time

6 Tuesday 1 June 2021

Spatially-explicit capture-mark-recapture analysis Spatial capture-recapture and a general R. Chandler 10:00–11:00 KL framework for individual based models of population dynamics 11:00–11:10 Break Session chairs: J. A. Royle and R. Chandler P. van Dam-Bates A marked Poisson process for latent id 11:10–11:30 CT spatial capture-recapture models H. Gaya Modeling species interactions using 11:30–11:50 CT spatial capture recapture 11:50–12:00 Break Spatial capture-recapture density B. Augustine 12:00–12:20 CT estimation using acoustic recording units without individual identification Secondary forest, secondary role: B. Rodrigues-do-Amaral 12:20–12:40 CT density and home range size variation of understory birds in a forest mosaic 12:40–12:50 Break 12:50–13:50 Interaction time

Thursday 3 June 2021

Integral projection modelling E. Simmonds Using IPMs to study the response of 10:00–11:00 KL populations to a changing environment 11:00–11:10 Break Session chairs: E. Simmonds, F. Plard, and E. Sandvig The demographic causes of population A. Allen 11:10–11:30 CT change vary across four decades in a long-lived shorebird Integral projection modeling for Weddell seals: evaluating multi-decadal K. Macdonald 11:30–11:50 CT population dynamics and the importance of individual heterogeneity to population stability 11:50–12:00 Break Y.-L. Fung Process-dependent integral projection 12:00–12:20 CT models

7 Benefits and costs of male and female B. Chamiot-Clerc subordinates on the population 12:20–12:40 CT dynamics of a cooperative breeding species 12:40–12:50 Break 12:50–13:50 Interaction time

Friday 4 June 2021

Animal movement R. Kays Big data empowers a new era of 10:00–11:00 KL comparative animal movement analyses 11:00–11:10 Break Session chairs: R. Kays and F. Buderman E. Eisenhauer Modeling yearly patterns in golden eagle 11:10–11:30 CT movement G. Fandos Empirical natal and breeding dispersal 11:30–11:50 CT kernels for European birds 11:50–12:00 Break H. McCaslin Hierarchical computing for hierarchical 12:00–12:20 CT models in ecology F. Buderman Guidance on inferring behavioural states 12:20–12:40 CT from animal location data 12:40–12:50 Break 12:50–13:50 Interaction time

Monday 7 June 2021

Honored speaker P. Saurola Life behind the numbers: 55 years 10:00–11:00 HS among the owls. Why? 11:00–11:10 Break Data integration and population analysis I E. Zipkin Using data integration to evaluate 11:10–12:10 KL ecological processes across scales 12:10–12:20 Break Session chairs: S. Converse and E. Zipkin A semi-spatial integrated population L. Petracca 12:20–12:40 CT model to assess population dynamics of a recolonizing species

8 How accurate are integrated population M. Paquet models in estimating the contribution of 12:40–13:00 CT immigration to changes in population growth? 13:00–13:10 Break Using multispecies integrated M. Queroue population model to understand 13:10–13:30 CT interspecific competition: a case study on Great Tits and Blue Tits Beyond bespoke: Standardized C. Nater integrated population models reveal 13:30–13:50 CT drivers of population dynamics of migratory birds across latitudes Integrated population model for J. Sanderlin Ferruginous Hawk and Golden Eagle in 13:50–14:10 CT Wyoming, USA, to assist with a long-term, state-wide monitoring plan 14:10–14:20 Break 14:20–15:20 Interaction time

Tuesday 8 June 2021

Population management M. C. Runge and A. M. Tucker A decision analytical framework for 10:00–11:00 KL guiding demographic estimation 11:00–11:10 Break Session chairs: M. C. Runge and A. M. Tucker C. Yackulic Integrating understanding to inform 11:10–11:30 CT management, research and monitoring C. McGowan Data analysis and modeling for 11:30–11:50 CT endangered species listing decisions 11:50–12:00 Break Hierarchical models to link occupancy M. Eaton and habitat dynamics to management 12:00–12:20 CT decisions: Recovering endangered Florida Scrub and Scrub-Jays Integrated simulations of demography, S. Canessa 12:20–12:40 CT monitoring and decisions for conservation of endangered species 12:40–12:50 Break 12:50–14:50 Poster session and interactions with poster presenters

9 Thursday 10 June 2021

Survival estimation E. Matechou Models for survival: from φ to φ and 10:00–11:00 KL ta c everything in between 11:00–11:10 Break Session chairs: E. Matechou and F. Colchero R. King Large data and individual heterogeneity 11:10–11:30 CT models: when worlds collide Laplace approximations for R. Herliansyah 11:30–11:50 CT capture-recapture models in the presence of individual heterogeneity 11:50–12:00 Break Misidentification errors in reencounters E. Rakhimberdiev result in biased estimates of survival 12:00–12:20 CT from CJS models: evidence and a possible solution using the robust design Using multievent recovery and auxiliary W. Kendall 12:20–12:40 CT resighting data to improve inference from multistate mark-recapture studies 12:40–12:50 Break 12:50–13:50 Interaction time

Friday 11 June 2021

Data integration and population analysis II Session chairs: S. Converse and E. Zipkin Integrating tracking and resight data C. Rushing enables unbiased inferences about 11:10–11:30 CT migratory connectivity and winter range survival from archival tags Spatial integrated models foster V. Lauret complementarity between monitoring 11:30–11:50 CT programs in producing large-scale ecological indicators 11:50–12:00 Break Integrating data types to reveal avian S. Saunders 12:00–12:20 CT migration patterns across the Western Hemisphere Integrating demographic and A. Gamble 12:20–12:40 CT epidemiological data improves dispersal quantification 12:40–12:50 Break

10 12:50–13:50 General assembly, Student prizes, and Closing remarks

11 List of Abstracts – Talks

Session 1. Building on the Eurasian-African Migration Atlas: towards robust quantitative analyses of avian movements

Funding analyses of over a century of bird ringing data: challenges, strategies and potential for the profile and applied use of innovative analytical approaches KL

Spina, F., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Italy

Baillie, S., British Trust for Ornithology, UK

Cutting edge modelling and innovative analytical approaches to study the contents of large-scale and long-term international biological datasets may sometimes take place only after long years of enduring efforts aimed to funding biostatisticians in performing such analyses. Almost two decades of attempts to support the production of the Eurasian-African Migration Atlas are a concrete example of such challenges. This contribution will offer a quick review of the different attempts made by representatives of the EURING constituency to find adequate resources to analyse the largest existing dataset, spanning over a century, of recoveries of birds marked with metal rings. We will show how hard it has been to seek resources in different arenas, from purely scientific international environment, to private foundations, to international institutions. Finally, the solution has been offered by a pledge offered by a single country. In proposing analytical approaches and the foreseen deliverables, we will show the importance of compromising between possible themes and priorities from a perspective of basic biological research vs research targeted also to produce results able to offer a scientific baseline to the implementation of wildlife conservation legislation. In the case of this atlas, the structure of the project and the connected modules devoted to specific aspects and issues will produce outputs of direct relevance and use for various international legislation and Multilateral Environmental Agreements (like the UN Convention on Migratory Species CMS and its related agreements and MoUs (e.g., AEWA, Raptors, Illegal Killing of Birds), the EU Wild Birds Directive, the Ramsar Convention, the UN Framework Convention on Climate Change). Initiatives like this atlas are interesting cases to reflect on how innovative analytical tools deriving from the community of biostatisticians involved in the EURING Analytical Meetings may gain high profile in supporting a scientifically robust implementation of international environmental legislation and the ensuing political decisions in prioritising actions and allocating resources for Nature conservation.

12 The Eurasian African bird migration atlas – a starting point for a more quantitative understanding of avian movements KL

Franks, S. E., British Trust for Ornithology, UK

Baillie, S., British Trust for Ornithology, UK

In Europe over 25 Migration Atlas volumes and large numbers of research papers document the movements of birds ringed in individual countries. The Eurasian African Bird Migration Atlas, to be published in Spring 2022, will present a unique on-line analysis of movements for populations of some 300 bird species that occur in Europe at some point in their annual cycle. The Atlas draws on over 24 million ringing and re-encounter records held in the EURING Databank together with selected tracking data from Movebank. The Atlas species accounts will include recovery maps showing overall connectivity patterns by ringing region, connectivity by month and by region, annual movements and major recovery causes together with maps of tracking data, summary info-graphics showing the types of data used and written species texts designed to integrate with the web presentation.

We briefly outline how this huge dataset has been collated and analysed, the strengths of the data in terms of spatial and temporal coverage and its weaknesses, most conspicuously the lack of com- prehensive information on numbers marked and a linked inability to formally account for variation in reporting rates. We show how techniques such as visualisations based on different recovery causes facilitate an assessment of the likely impacts of such biases. We compare our visualisations with some limited examples where quantitative analyses have been undertaken and for selected populations with patterns from tracking data which are much less affected by biases in encounter rates. The Migration Atlas project also includes quantitative analyses of migration timing, migratory connectivity, long-term changes in migration behaviour and illegal killing which are covered by separate talks within this symposium. We provide an outline of this analytical work and how it complements the core analyses of migration routes.

The various types of tracking data that are available in increasing quantity and with constantly improving precision and sample sizes are a critically important source of information on movements. However ringing and recovery data continue to offer a powerful and complementary source of information on movements, including opportunities to quantify long-term changes that are so important in relation to issues such as climate and habitat change. We review some promising opportunities for analyses of such data based on recent statistical developments. These include models that formally quantify reporting rates and those that incorporate data on occurrence and abundance. Models that simulate the decisions made by migrating birds can also provide important insights. We anticipate substantial advances in our understanding of bird movements over the next decade and that analyses of ringing data will continue to make a very substantial contribution.

13 Patterns of migratory connectivity in the European-African migratory system as in- ferred by ring recovery analysis of the EURING database

Fattorini, N., Department of Environmental Science and Policy, University of Milan, Italy

Spina F., Istituto Superiore per la Ricerca Ambientale, Italy

Rubolini, D., Department of Environmental Science and Policy, University of Milan, Italy

Ambrosini, R., Department of Environmental Science and Policy, University of Milan, Italy

Migratory connectivity, i.e. the degree by which individuals stay together while moving to different geographical areas where they spend different phases of their annual life cycle, has been investigated through the analysis of ring recoveries of 140 bird species in the framework of the CMS/EURING European-African Bird Migration Atlas.

We filtered more than 12 million ringing encounters collected over more than a century and including short- and long-distance migrants. After a robust data selection based on spatiotemporal masking and condition-specific criteria, we retained individual records in both breeding and non-breeding grounds. To quantify migratory connectivity, we calculated the Mantel correlation coefficient between ortho- dromic distance matrices and assessed significance though a permutation test and 95% confidence interval through bootstrap resampling. In case of significant connectivity, we implemented a k-means cluster analysis to test the origin of such connectivity, assessing the degree of clustering.

Migratory connectivity showed a high interspecific variability, but most species exhibited significant connectivity. Generally, connectivity was due to clustering, whereas few species showed connectivity from pattern transference. We found that the strength and pattern of migratory connectivity were influenced by both ecological and geographical predictors, as well as by phylogenetic relatedness among species. Finally, sensitivity analysis performed for each species showed that our method was robust even for small samples (i.e., at least 30 encounters), and that it was not affected by encounter condition (alive recaptures or dead recoveries). Measuring migratory connectivity may shed light on ecological factors underlying avian migration, as well as assisting in bird conservation and management at the population level. Our work demonstrates that ringing encounters can provide an excellent tool to perform analyses of migratory connectivity.

14 Estimating migratory connectivity strength and pattern across data types

Hostetler, J. A., Division of Migratory Bird Management, US Fish and Wildlife Service, MD, USA

Cohen, E. B., University of Maryland Center for Environmental Science, MD, USA

Hallworth, M. T., Cary Institute for Ecosystem Studies, NY, USA; Vermont Center for Ecostudies, VT, USA

Rushing, C. S., Department of Wildland Resources and the Ecology Center, Utah State University, UT, USA

Contina, A., Department of Integrative Biology, University of Colorado Denver, CO, USA

Bossu, C., Department of Biology, Colorado State University, CO, USA

Ruegg, K., Department of Biology, Colorado State University, CO, USA

Understanding migratory connectivity, or how migratory populations are distributed during two or more phases of their annual cycle, is critically important information for effective conservation and management of birds and other wildlife. The strength of migratory connectivity (MC) is a key metric quantifying how separate/intermixed breeding populations are in other parts of the annual cycle. We developed methods for estimating the pattern and strength of migratory connectivity from a) ring return data using a non-Markovian multistate mark-recovery model; and b) genoscape data from collected feathers using a bootstrap approach. We developed methods to integrate genoscape, stable- isotope, light-level geolocator, and GPS data collected from the same or different individual animals to allow for more precise estimation of a species’ migratory connectivity pattern and strength, that incorporate the inherent uncertainty associated with each data type. We implement these methods in the R package MigConnectivity and use multiple types of data collected across the range of the Painted Bunting (Passerina ciris) to estimate the strength and pattern of migratory connectivity. As these data become increasingly available for many species, we discuss how current and future methods for estimating metrics of migratory connectivity will allow for a more complete picture of wildlife migration ecology.

15 Disentangling survival, migratory connectivity and observation in continuous space

Schirmer, S., Department of Mathematics and Computer Science, University of Greifswald, Germany

Korner-Nievergelt, F., Swiss Ornithological Institute, Switzerland von Rönn J. A. C., Swiss Ornithological Institute, Switzerland

Liebscher, V., Department of Mathematics and Computer Science, University of Greifswald, Germany

Survival and movement in space is essential for the evolution and demography of migratory animals. They repeat their migratory journeys over large timespans and spaces in which survival and its spatial heterogeneity play an important role. Both, use of space and survival, significantly affects the observed pattern of animal distribution, e.g., in a map of dead recoveries. The spatial point pattern is a product of where animals go, where they die and if they are found. Drawing conclusions on one of these processes needs the knowledge of the others and therefore mathematical methods to distinguish between them. Annual survival, migratory connectivity and recovery probability can be disentangled for a few discrete spatial areas by a model based on the division coefficient and the multinomial reencounter model. This approach mainly uses characteristics of the multinomial distribution. Moreover, an analogous formulation in continuous space provides maps of continuous spatial annual survival and migratory connectivity using mixed binomial thinned point processes. As a major drawback in continuous space, recovery probability must be constant. Finally, we found a method relaxing this assumption by combining the advantages of the two models. Therefore, we piecewise optimize the match of continuous to discrete migratory connectivity estimates using penalized M-splines. This expands the application possibilities enormously. In our talk, we provide a short outline of the two basic models and how they are joined in the combined approach. Simulated and real-world data illustrate the application of the method.

16 Quantifying individual variation in migration versus residence within and among year: a dynamic-finite-mixture multi-state capture-recapture model

Acker, P., Centre for Biodiversity Dynamics, NTNU, Norway; School of Biological Sciences, University of Aberdeen, UK

Daunt, F., Centre for Ecology & Hydrology, Bush Estate, UK

Burthe, S. J., Centre for Ecology & Hydrology, Bush Estate, UK

Reid, J. M., Centre for Biodiversity Dynamics, NTNU, Norway; School of Biological Sciences, University of Aberdeen, UK

Seasonal migration is a key trait that underpins spatio-seasonal population dynamics and distributions, and allows individuals to escape from temporarily hostile environments. Global changes are now alter- ing patterns of spatio-seasonal environmental variability, including increasing frequencies of extreme climatic events (‘ECEs’). But, the potential for changing migration to mediate rapid eco-evolutionary responses remains largely unknown. In “partially migratory” populations, different individuals express migrant and resident phenotypes during the same non-breeding season. Further, individuals may undertake different tactics, such that they are resident or migrant throughout the non-breeding season, or express a mix of both phenotypes given late migration and/or early return to residence. Moreover, from one year to the next, individuals may either repeat the same annual migratory tactic or change tactic, and tactics may be subject to survival selection. Quantifying within- and between-year individual variation in migration versus residence, and associations with components of fitness, is thus central to understanding long-term spatio-seasonal population dynamics in the context of environmental change. These objectives can be achieved using multi-year year-round resightings of marked individu- als within partially migratory populations, but advanced capture-recapture models are required to infer tactic distribution within years, transition probabilities between years, and associated survival probabilities. Accordingly, we devised a novel full-annual-cycle multi-state model that includes a dynamic finite mixture of annual tactics along the residence-migration continuum, and accounts for spatio-temporal heterogeneity in re-encounter probabilities. We applied this model to 11 years of large-scale ring-resighting data from sympatric-breeding European shags (Phalacrocorax aristotelis), encompassing three winters with ECEs. We demonstrate that the different annual migratory tactics (i.e. full-winter residence, mixed residence-migration, and full-winter migration) coexist in non-negligible proportions in the population, and are typically highly repeatable from one year to the next. However, this repeatability is tactic- and year-dependent. Full-winter residents and full-winter migrants were typically more repeatable than mixed resident-migrants. Further, full-winter residents and migrants preferentially switched to mixed residence-migration, while mixed resident-migrants were similarly likely to switch to full-winter migration or residence. There was also considerable among-year and sex-specific variation in relationships between migratory tactic and survival probability, with strongly contrasting effects in years with ECEs. Such patterns of within-year and among-year variability in the expression of migration versus residence, and associated survival selection, are consistent with threshold-trait dynamics and highlight the potential for flexible seasonal movements to shape complex eco-evolutionary dynamics.

17 Assessing the timing of migration progression based on ring recoveries

Ambrosini, R., Department of Environmental Science and Policy, University of Milan, Italy

Fattorini, N., Department of Environmental Science and Policy, University of Milan, Italy

Rubolini, D., Department of Environmental Science and Policy, University of Milan, Italy

Spina, F., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Italy

Serra, L., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Italy

Imperio, S., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Italy

Bairlein, F., Institute of Avian Research, Vogelwarte Helgoland, Germany

For modelling the temporal progression of migration in the framework of the CMS/EURING European- African Bird Migration Atlas using ring recoveries, we built out of a previously published approach that used conditional autoregressive models applied to ringing data, which was slightly modified to account for partial migrants. The study area was divided into cells of different size according to a procedure that, starting from cells of 2 x 2 degrees (latitude x longitude), produced cells of different size, inversely proportional to encounter density and including at least 20 encounters in at least 10 different days of the year (minimum sample size). The number of encounters (either ringing or recovery data) per calendar date and cell was then calculated, independently of the year of encounter and transformed into the cumulative proportion of encounters at each date and cell, which was finally modelled using a binomial Generalized Linear Mixed Model (GLMM) with a complementary log-log link function and an exponential spatial covariance structure. Partial migrants, i.e. species that are present in winter at least in some parts of the study area, should generate encounters also during stationary periods. We therefore produced a heuristic procedure that identified the day when the curve that estimates the cumulated proportion of birds started growing more than what was expected based on the rate of encounters during the stationary period. This procedure was based on the assumption that the probability of encountering individuals should increase more when they start moving than when they are stationary, and this would reflect in a sudden increase in the encounter probability estimated by the model. This procedure thus estimated the day when a given proportion of individuals has arrived into a cell, after having accounted for the encounters generated by local birds. Estimated dates were then spatially interpolated using a grid with 2 x 2 degree cells and the resulting interpolated map was downscaled to obtain the expected values at cells of one degree of size using the bilinear method. The procedure includes a set of parameters whose values were chosen arbitrarily. To assess the robustness of the method, we re-run the analyses with different values of all parameters, obtaining 27 final maps that were then averaged. A map of the standard error of predicted dates was also produced to assess the uncertainty of the model due to arbitrary parameter choice, which was always very low, confirming the robustness of the procedure. Our procedure thus allows robust, spatially

18 explicit estimates of the date of onset of bird migration based on ring recoveries and may thus have wide-ranging management applications. For instance, they may have significant applied value for the implementation of international environmental legislation like the EU Wild Birds Directive and offer a novel contribution to the seasonality of harvesting of migratory populations shared across the Member States of the EU.

19 Session 2: Spatially-explicit capture-mark-recapture analysis

Spatial capture-recapture and a general framework for individual-based models of population dynamics KL

Chandler, R. B., Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Gaya, H., Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

During their early development, capture-recapture methods were often thought of as tools for es- timating demographic parameters, which might later be plugged into a population model. Recent developments such as integrated population models have illustrated a more unified approach wherein the population model—often a matrix population model—is fitted to capture-recapture and count data. Fitting population models to data, rather than viewing estimation and modeling as distinct endeavors, greatly facilitates the proper accounting of variation and uncertainty in the ecological and observation processes. Although matrix population models, and more recently integral projection models, have become the workhorses of population ecology, these models struggle to describe spatial variation and individual heterogeneity in vital rates because they either aggregate individuals into a small number of discrete classes or describe continuous variation in a small number of traits. Moreover, fitting these models to capture-recapture data involves a type of misalignment because capture-recapture data are inherently individual-level information, whereas matrix population models and integral projection models do not retain individual identity over time. Individual-based (i.e., agent-based) population models offer a more realistic description of how population-level patterns emerge from individual-level variation. However, these models are often dismissed as being too ad hoc to allow for general insights, and they are rarely fitted to data. In this talk, we discuss a general framework for individual-based models of population dynamics based on spatio-temporal point processes, and we describe how spatial capture-recapture methods make it possible to fit this class of individual-based population models to data. The approach, which builds upon previous work on open-population spatial capture-recapture models, will be demonstrated using avian mist-netting data from the Appalachian Mountains where we are studying the dynamics of populations near the southern edge of their breeding ranges.

20 A marked Poisson process for latent ID spatial capture-recapture models van Dam-Bates, P., School of Mathematics and Statistics, University of St Andrews, Scotland

Borchers, D., School of Mathematics and Statistics, University of St Andrews, Scotland

Papathomas, M., School of Mathematics and Statistics, University of St Andrews, Scotland

Stevenson, B., Department of Statistics, University of Auckland, New Zealand

Fewster, R., Department of Statistics, University of Auckland, New Zealand

The use of digital technology for spatial capture-recapture (SCR) surveys introduces new analytical challenges. One of the most significant is how to estimate animal density in the absence of observed animal identities. For camera traps, the location of the traps at which animals are detected and animal features such as pelage, can help to inform ID. Similarly, for acoustic surveys, the locations at which a call is heard contains information on where the call originated. The current methods for latent ID SCR use this information to tease apart animal ID by assuming that observations made similarly in space are likely to be from the same animal. We formulate SCR as a marked Poisson process with the mark distribution comprising a mixture of distributions that depend on latent animal activity centres. This is a generalization of the unmarked SCR model of Chandler and Royle (2013), and it allows us to incorporate into SCR partial identity information such as is provided by pelage.

21 Modeling species interactions using spatial capture recapture

Gaya, H., Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Chandler, R. B., Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Patterns of species co-occurrence are often important for understanding species-level processes such as species diversity, population dynamics or niche overlap. Using co-occurrence data to infer the relative roles of habitat characteristics and interspecific competition on species distributions can be difficult because competition arises as an individual-level process that can occur at fine spatial scales. To quantify the drivers of species co-occurrence, we present a two-species spatial capture-recapture model that includes a Markov point process in which an individual’s location is dependent upon both abiotic covariates and the locations of individuals of other species. We applied the model to data on two ecologically similar songbird species — Hooded Warbler and Black-throated Blue Warbler — that segregate over a climate gradient in the Appalachian Mountains of North Carolina. In spite of many ecological similarities between the two species, we found minimal evidence of competition between the Hooded Warbler and Black-throated Blue Warbler at our study sites. Rather, spatial variation in density of the two species was much better explained by climate variables. Unlike previous statistical models that attempt to infer competition from species-level co-occurrence data, the framework proposed here allows for inference at the individual-level and can be used to assess the spatial scale of biotic interactions.

22 Spatial capture-recapture density estimation using acoustic recording units with- out individual identification

Augustine, B. C., U.S. Geological Survey, John Wesley Powell Center, Cornell Department of Natural Resources, Ithaca, NY, USA

Royle, J. A., U.S. Geological Survey, Patuxent Wildlife Research Center, MD, USA

Chandler, R. B., Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Fuller, A. K., U.S. Geological Survey, New York Cooperative Fish and Wildlife Research Unit, Department of Natural Resources, Cornell University, NY, USA

Autonomous recoding units (ARUs) are increasingly being used to survey species that are difficult to monitor using more established methods such as camera traps, but exhibit vocalizations that can be reliably classified to the species-level. To date, the majority of bioacoustic surveys are analyzed as species occupancy data because of the lack of individual identities typically required for density estimation using capture-recapture models. However, density estimates are generally preferred due to the greater ecological significance of population density compared to occupancy. Spatial capture- recapture is a promising framework for estimating density from ARU data if the lack of individual identities can be satisfactorily accommodated. To date, the lack of individual identities has been addressed by first estimating the call density of the species of interest and then using an independent estimate of the species’ individual call rate to convert call density to individual density. This approach is limited by the fact that individual call rates can vary across space and time, making it difficult to know whether the call rate estimate is appropriate for the ARU data. Further, current approaches accommodate the fact that individuals call multiple times, but not that they may move while calling. We present a spatial capture-recapture model for ARU data with no individually-identifying information that allows for density estimation for species that move while calling and that estimates the individual call rate from the ARU data in hand. The model uses a Thomas point process, treating individuals as “parents” and their calls as “children”, describing the situation where an individual’s call locations are concentrated around the individual activity center. Unlike a typical Thomas point process, the children (call) locations are not observed, but must be localized using a detection or acoustic attenuation function. Further, the number of undetected children must be estimated in addition to the number of parents (individuals). In order to estimate both the number of individuals and calls along with the call rate, we use data augmentation for both individuals and calls. Using this approach, we demonstrate that density estimation using SCR-based methods for bioacoustics detections is possible without individual identities and supplementary information about sound attenuation or call rates and movement parameters, though supplementary information for these parameters can be included via informative priors. Further, call features that correlate with individual identity (e.g., spectrograph features) can be used as partial identity covariates to further improve estimation. Finally, this estimation framework may be applied to other capture-recapture problems such as area searches with imperfect detection and density estimation of group-living species whose group membership is not known when detected.

23 Secondary forest, secondary role: density and home range size variation of under- story birds in a forest mosaic

Amaral, B. R., Department of Ecosystem Sciences and Management, Pennsylvania State University, PA, USA

Miller, D. A. W., Department of Ecosystem Sciences and Management, Pennsylvania State University, PA, USA

Ferraz, G., Department of Ecology, Federal University of Rio Grande do Sul, Brazil

In the neotropics, second-growth forests are common in the landscape and exceed the area of old- growth forests in most countries. Although these areas can serve as habitat for wildlife, major differ- ences between the two forest types may limit the potential of second-growth forests to sustain viable populations and offset species loss. However, much less is known about tropical bird communities than temperate bird communities. Extending our understanding of how these communities requires robust tools for studying the demography of large numbers of species. Here, we use spatially explicit capture-recapture (SECR) models and mist-net data to investigate how birds from the Amazon terra firme forest, which occur both in old- and second-growth forests, use space differently according to forest type. Using six years of mark and recapture data from banded understory insectivores at the Biological Dynamics of Forest Fragments Project, in Brazil, we estimated changes in population density and home range size of twenty-one species of birds. We had a total of 7,161 captures of 4,279 individuals, with fifty-six percent of the captures in old-growth forest. We observed higher densities and smaller home range sizes in old-growth forest for most species. From the species that have second- growth forest effect coefficient estimates that did not overlap zero, sixty-five percent have lower density (average-1.6) in second-growth forest, and all have smaller home range areas (average-1.7) in old-growth forest. Even though birds were able to increase their home range to satisfy resource intake, the negative effect of secondary forest on density might suggest that birds have a preference for old-growth, or that fitness benefits sustain larger populations in this habitat, or both. We were able to use mist-net data and SECR models to demonstrate that space use in a tropical rainforest of several elusive understory insectivores is altered by forest age. Understanding the ecological value of each component of the disturbance mosaic of Amazonian landscapes is crucial to ensure the persistence of these bird populations. Our results demonstrate the potential of spatially replicated mark-recapture studies using mist-netting and SECR analytical methods to address crucial questions in these highly diverse bird communities.

24 Session 3: Integral projection modelling

Using IPMs to study the response of populations to a changing environment KL

Simmonds, E., Department of Mathematical Sciences, Norwegian University of Science and Technology, Norway

We are currently in a period of rapid environmental change, with populations across the globe ex- periencing alterations to the environments they experience. These changes have consequences for population dynamics and demography. Predicting how individual populations might respond has been a challenge, but one that can be (and has been) addressed with integral projection models. This plenary will synthesize the theoretical and applied advances in integral projection models over the past two decades, which have given new insights into how populations (often birds) could respond to environmental changes. The talk will finish with looking to the future and how integral project models can be used to best predict the fate of our biological populations.

25 The demographic causes of population change vary across four decades in a long- lived shorebird

Allen, A. M., Department of Animal Ecology, Netherlands Institute for Ecology (NIOO-KNAW), The Netherlands

Jongejans, E., Radboud University, Department of Animal Ecology and Physiology, The Netherlands

Van de Pol, M., Department of Animal Ecology, Netherlands Institute for Ecology (NIOO-KNAW), The Netherlands

Ens, B. J., Sovon Dutch Centre for Field Ornithology, The Netherlands

Frauendorf, M., Department of Animal Ecology, Netherlands Institute for Ecology (NIOO-KNAW), The Netherlands

Van der Sluijs, M., Department of Animal Ecology, Netherlands Institute for Ecology (NIOO-KNAW), The Netherlands

De Kroon, H., Radboud University, Department of Experimental Plant Ecology, The Netherlands

Understanding which factors cause populations to decline begins with identifying which parts of the life cycle, and which vital rates, have changed over time. However, in a world where humans are altering the environment both rapidly and in different ways, the demographic causes of decline may also vary over time. Identifying temporal variation in demographic causes is crucial to assure that conservation actions target current and not past drivers of decline. However, this has rarely been studied as it requires long time series. Here we investigate how the demography of a long-lived bird (the Eurasian Oystercatcher Haematopus ostralegus) has changed in the past four decades, resulting in a shift from stable numbers to strong declines (down to -7% per year), and recently back to a modest decline. Since individuals of this species are likely to respond differently to environmental change, we captured individual heterogeneity through three state variables: age, breeding status and lay date (using Integral Projection Models). Lay date explained significant levels of variation in reproduction, with a parabolic relationship of maximal productivity near the average lay date. Variation in reproduction had the greatest influence on the changes in population growth, with poor nest success explaining much of the decline in the last two decades. However, the demographic causes of decline have also been in flux over the last three decades: hatchling survival was also low in the 2000s but improved in the last decade, whilst adult survival was constant in the first three decades but declined in recent years. Our results are in line with previous studies, i.e. nest success is the key demographic variable and remains low today, however we also identify how improvements in other vital rates are buffering the potential severity of decline nowadays. The dynamic nature of the threat landscape is further supported by the finding that the average individual no longer has the highest performance in the population. Our results thus emphasise how individual heterogeneity in vital rates can play an important role in modulating population growth rates. Understanding population declines in the current era requires disentangling demographic mechanisms, individual variability, and how they change over time.

26 Integral projection modeling for Weddell seals: evaluating multi-decadal popula- tion dynamics and the importance of individual heterogeneity to population sta- bility

MacDonald, K. R., Ecology Department, Montana State University, USA

Paterson, J. T., Ecology Department, Montana State University, USA

Rotella, J. J., Ecology Department, Montana State University, USA

We have used hierarchical, multi-state models and over 30 years of data for a large sample of known- age female Weddell seals to estimate age- and breeding-state-specific survival and reproductive rates for recruited females. The analysis also provided estimates of individual heterogeneity and annual variation in the vital rates. Results indicate that rates of reproduction vary strongly with age and year and have notable levels of among-individual heterogeneity. Results also provide evidence for costs of reproduction to both survival and future reproduction such that interesting life-history trade-offs may exist. Integral projection models that incorporate the results of the multi-state modeling and employ a novel formulation to incorporate individual heterogeneity in age- and state-specific reproductive rates (along with previously published rates of recruitment and juvenile survival) are used to evaluate the population’s dynamics. Results provide information on multi-decadal rates of population growth for the southern-most population of mammal and evaluate the importance of individual heterogeneity to the overall population’s ability to sustain itself. Both ecological and evolutionary aspects of the findings will be presented, and results from integral projection modeling are contrasted with results from simpler matrix models that ignore individual heterogeneity.

27 Process-dependent integral projection models

Fung, Y.L., School of Mathematics, University of Edinburgh, Scotland

Newman, K., Biomathematics and Statistics Scotland, Scotland

King, R., School of Mathematics, University of Edinburgh, Scotland

Predictions of population dynamics that ignore possible correlations between demographic processes such as somatic growth or reproduction may be relatively biased or imprecise compared to predictions accounting for such existing correlations. However, commonly used integral projection models (IPMs) typically assume (conditional) independence across these demographic processes. Here, we propose three categories of methods that explicitly incorporate between-process correlations in IPMs, while retaining the tractability of models. Driven by biological interpretations of the correlation structure, these categories of models induce distinctive between-process correlations. We compare our process- dependent methods with conventional IPMs on simulations and a case study of Soay sheep (Ovis aries), focusing on the impact of estimates of the asymptotic log population growth rate. Our results show that estimates of the asymptotic log population growth rate may become biased when the correlations between demographic processes are ignored. This bias may be either positive or negative depending on the form of dependence applied. Also, the magnitude of the bias may be substantial in some scenarios. In particular, we suggest the use of process-dependent IPMs when the study species is believed to carry strong between-process correlation, or the target statistics is not asymptotic log population growth rate.

28 Benefits and costs of male and female subordinates on the population dynamics of a cooperative breeding species

Chamiot-Clerc, B., Biométrie et Biologie Evolutive, Université Claude Bernard (Lyon I), France

Cohas, A., Biométrie et Biologie Evolutive, Université Claude Bernard (Lyon I), France

Plard, F., Biométrie et Biologie Evolutive, Université Claude Bernard (Lyon I), France

Cooperative breeding has often evolved in unpredictable and variable environments. Benefits of cooperative breeding generally include an increase of reproductive success of dominants through helping behaviors of subordinates. Helpers can also act as climatic buffers by reducing the temporal variance of reproductive success. While benefits on annual reproductive success are easily quantified, long-term effect of cooperative breeding on population dynamics are rarely estimated. Indeed both direct and indirect effects of helpers on population dynamics need to be quantified. Moreover, these benefits are trade-off by cost associated with group living including individuals competition for access to food and reproduction that are rarely quantified. In this study, we analysed the long-term effects of male and female subordinates on the population dynamics of a social and cooperative breeding species, the Alpine marmot. In this species living in a variable mountain environment, only subordinate males enhance first year survival but the role of female subordinates is still unknown. We used 25 years of data on a population of Alpine marmot to build an integral projection model and quantified the long-term benefits and costs of male and females subordinates on population growth rate and family vital rates.

29 Session 4: Animal movement

Big data empowers a new era of comparative animal movement analyses KL

Kays, R., Smithsonian Tropical Research Institute, Republic of Panama; North Carolina Museum of Natural Sciences, NC, USA; Department of Forestry and Environmental Resources, North Carolina State University, NC, USA

Hirsch, B., Smithsonian Tropical Research Institute, Republic of Panama; College of Science and Engi- neering, James Cook University, Australia

Caillaud, D., Department of Anthropology, University of California, Davis, CA, USA

Mares, R., Smithsonian Tropical Research Institute, Republic of Panama; Department of Anthropology, University of California, Davis, CA, USA

Alavi, S., Department of Biology, University of Konstanz, Germany; Center for the Advanced Study of Collective Behavior, University of Konstanz, Germany

Worsøe Havmøller, R., Department for the Ecology of Animal Societies, Max Planck Institute of Animal Behavior, Germany; Department of Biology, University of Konstanz, Germany; Center for the Advanced Study of Collective Behavior, University of Konstanz, Germany

Crofoot, M., Smithsonian Tropical Research Institute, Republic of Panama; Department of Anthropology, University of California, Davis, CA, USA; Department for the Ecology of Animal Societies, Max Planck Institute of Animal Behavior, Germany; Department of Biology, University of Konstanz, Germany; Center for the Advanced Study of Collective Behavior, University of Konstanz, Germany

While the comparative method is one of the most powerful approaches in biology its use in move- ment ecology has been limited. Recently, three factors have changed this. First, the flood of data from increasingly small GPS tracking devices now allows us to document animal movements with unprecedented detail, on a growing diversity of species. Second, rapidly developing analytics allow us to richly describe animal behaviors from these data. Finally, live data streams, archives and sharing tools enable joint analyses needed for true comparative studies (i.e. not just meta analyses of results from previous studies). To support this new field we propose a new Multi Scale Movement Syndrome (MSMS) framework for describing and comparing animal movements. MSMS incorporates four hi- erarchical scales of animal movement: 1) fine-scale movement steps which accumulate into 2) daily paths which then, over weeks or months, form a 3) home range. Finally, 4) the lifetime track of an individual consists of a multiple ranges connected by dispersal or migration events. We suggest a series of metrics to describe patterns of movement at each of these scales and use the first three scales of this framework to compare the movement of 47 animals from four species of sympatric frugivorous mammals. While only subtle differences exist between species in their step-level movements, they cluster into three distinct movement syndromes in both path- and range-level analyses. Differences in feeding ecology were a better predictor of movement patterns than a species’ locomotory or sensory

30 adaptations. Given the role these species play as seed disperser, these movement syndromes have important ecosystem implications. This multiscale approach provides a hierarchical framework for comparing animal movement and could be combined with analogous resource selection functions at the same scales to connect movement process with emergent patterns of space use.

31 Modeling yearly patterns in Golden Eagle movement

Eisenhauer, E., Department of Statistics, Penn State University, PA, USA

Hanks, E., Department of Statistics, Penn State University, PA, USA

Murphy, R., Eagle Environmental, Inc., NM, USA

We model yearly patterns in Golden Eagle movement using two competing approaches: a latent-state model which we define differently for migratory, dispersal, and residential individuals, and a varying coefficient model. While latent-state models are more common in the existing animal movement literature, varying coefficient models have various benefits including the flexibility to fit a wide range of movement strategies without the need for major model adjustments. We illustrate the comparison by separately fitting movement paths for three individuals with both a latent-state and varying coefficient model. Simulations from these models illustrate the ability of varying coefficient models to better describe Golden Eagle movement behavior.

32 Empirical natal and breeding dispersal kernels for European birds

Fandos, G., Institute for Biochemistry and Biology, University of Potsdam, Germany; Geography De- partment, Humboldt Universität zu Berlin, Germany

Talluto, M., Department of Ecology, University of Innsbruck, Austria

Fiedler, W., Max Planck Institute of Animal Behavior, Germany; Department of Biology, University of Konstanz, Germany

Robinson, R., British Trust for Ornithology, UK

Thorup, K., Center for Macroecology, Evolution and Climate, Globe Institute, University of Copenhagen, Denmark

Zurell, D., Institute for Biochemistry and Biology, University of Potsdam, Germany; Geography Depart- ment, Humboldt Universität zu Berlin, Germany

Dispersal is a key life-history trait essential for most species to ensure connectivity and gene flow between populations and facilitate viability in their ever-changing natural environment. However, dispersal has proved difficult to quantify, and as a result, there is a lack of empirical data describing animal dispersal patterns. Here we aim to provide empirical dispersal parameters for general, natal (before first breeding) and breeding dispersal (between subsequent breeding attempts) for European breeding birds (239 species). We have standardised and analysed an extensive volunteer-based birdring- recoveries database in Europe (EURING) by accounting for the different reporting scheme thresholds and excluding potentially migratory movements. Then, we fitted four widely used probability density functions in a Bayesian framework to compare and provide the best statistical description of the natal and breeding dispersal kernel for each bird species. We provide breeding and natal dispersal kernels for most European breeding bird species encompassing a large diversity of life-history traits. Our dispersal estimates offer new insights into selecting appropriate dispersal kernels for birds and provide new avenues to improve our understanding of the mechanisms and rules underlying dispersal events.

33 Hierarchical computing for hierarchical models in ecology

McCaslin, H. M., Department of Fish, Wildlife, and Conservation Biology, Colorado State University, CO, USA

Feuka, A. B., Department of Fish, Wildlife, and Conservation Biology, Colorado State University, CO, USA

Hooten, M. B., U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, CO, USA; Department of Fish, Wildlife, and Conservation Biology, Colorado State University, CO, USA; Department of Statistics, Colorado State University, CO, USA

In recent years, movement ecology has experienced rapid advances in tracking technology, resulting in the ability to collect telemetry data for more individuals across a wider range of species. This has improved our ability to learn about movements of animals, specify mechanistic movement models, and synthesize learning about movement processes at population-levels using high-resolution data from many individuals. However, increases in the amounts of data we can collect and the complexity of models we can specify present computational challenges because fitting large models, like hierarchical movement models and other Bayesian models, can be prohibitively time-consuming to implement. We present a recursive Bayesian computing method that can be used to fit Bayesian models efficiently in sequential MCMC stages to ease computation and streamline hierarchical inference. We show how transformation between the stages can be used to create unsupervised MCMC algorithms and improve interpretability of parameters. We applied our method to model seasonal movements of a migratory species to obtain inference about individual- and population-level migratory characteristics, without needing to tune the algorithm for every individual, speeding computation and enabling parallel computing across individuals. Furthermore, we describe how our method enables us to assimilate data from additional individuals as they become available, without requiring us to fit the entire model again. Transformation-assisted recursive Bayesian computing is an accessible method for addressing computational limitations of complex hierarchical models, and is well-suited for fitting a variety of ecological models, including multispecies models and hierarchical models spanning large spatial and temporal scales. By reducing the computational barriers to these models, our approach enables the implementation of new statistical models that take full advantage of modern data sources, like high-resolution telemetry data, to advance our understanding of complex ecological phenomena.

34 Guidance on inferring behavioral states from animal location data

Buderman, F. E., Department of Ecosystem Science and Management, Pennsylvania State University, PA, USA

Gingery, T. M., Pennsylvania Cooperative Fish and Wildlife Research Unit, Pennsylvania State University, PA, USA

Diefenbach, D. R., U.S. Geological Survey, Pennsylvania Cooperative Fish and Wildlife Research Unit, Pennsylvania State University, PA, USA

Gigliotti, L. C., Department of Environmental Science, Policy, and Management, University of California Berkeley, CA, USA

Begley-Miller, D., Teatown Lake Reservation, NY, USA

McDill, M. M., Department of Ecosystem Science and Management, Pennsylvania State University, PA, USA

Wallingford, B. D., Pennsylvania Game Commission, PA, USA

Rosenberry, C. S., Pennsylvania Game Commission, PA, USA

Drohan, P. D., Department of Ecosystem Science and Management, Pennsylvania State University, PA, USA

We are increasingly able to obtain fine-scale long-term animal movement data across multiple indi- viduals. With the advent of modern tracking methods, we have also developed a suite of statistical methods that can be used to model metrics that describe space-use and movement. From an ecological standpoint, the goal is often to link a particular set of metrics or space-use patterns to unobserved, or latent, behaviors. Quantifying these behaviors can, in turn, be used to increase our understanding of events that are difficult to observe in the wild and incorporate these events into management and conservation plans. However, these latent behavioral states, and where and when a model suggests they occur, are rarely validated against independent observations. We discuss multiple case studies, centered on ungulates in Pennsylvania, in which various statistical methods for animal space-use and movement can result in erroneous inference on behavioral states. Our goal is to highlight the statistical underpinnings of commonly used space-use and movement models, how the statistical foundations are related to the obtainable inference, and caution against over-interpreting inferred latent states without auxiliary data. We suggest that researchers plan for data-collection that can help confirm statistical identification of behavioral states.

35 Honored speaker: Pertti Saurola

Professor Pertti Saurola’s passion for nature, science and conservation is today a burning flame in many scientists, politicians and landowners.

After studying biology, Pertti was a researcher and assistant lecturer at the Department of Animal Ecology and Morphology, University of Helsinki (1963–1973). He was then appointed Head of the Finnish Bird Ringing Centre, Curator at the Finnish Museum of Natural History at the University of Helsinki. Since his retirement in 2001, he has continued intensive research on Ural Owls, Tawny Owls and Ospreys that he started in 1965. It was not easy to contact Pertti to write this biography because every spring he and his wife, Hemuli, spend 7 days a week in the forest searching for owls.

Pertti attaches great importance to collaboration across borders and disciplines. He actively fostered the exchange of ideas while serving as President (1981–1995) and Board Member (1977–1999) of the EURING (Union of European Bird Ringing Centres), Representative of Finland (1982–) of the IOC (International Ornithological Committee) and Member of the Council (2001–2005) of the EOU (European Ornitholog- ical Union). Pertti was one of the founders of the EURING Technical Meeting and he is the only person who has participated in every meeting.

Through his open mindedness, willingness to share his data and his desire to understand ecological processes he encouraged ecologists and statisticians to collaborate to develop better statistical models. He has published more than 270 publications. He won The Finnish State Award for Public Information with two books: Sääksi (The Osprey) together with Juhani Koivu in 1987 and The Finnish Bird Ringing Atlas I & II together with six other authors in 2016. In addition, the book Suomen Pöllöt (Finnish Owls), edited and partly written by Pertti, was nominated for The Tieto-Finlandia Award in 1996. In 2016, Pertti received The Champion of Owls Award and Marsh Award for International Ornithology.

Since childhood, Pertti has been a keen birdwatcher. He obtained his ringing permit at the age of 15. Nature conservation, in general, and protection of birds of prey and owls, specifically, has been important to him through his life. When he was 27, he started a long-term population study on owls, which continues today. In his articles and interviews, Pertti has strongly criticized one-track modern forestry, “clear-cutting and transforming of Finnish forests into pulp and toilet paper”, which has destroyed the landscape and dramatically decreased the biodiversity. Pertti has tried to convince forest landowners to retain old trees and to adopt an ecologically-focused approach to forest management. As Head of the Finnish Bird Ringing Scheme, he strongly encouraged ringers to study and protect birds of prey and started the Project Pandion (1971), Raptor Grid (1982), and Raptor Questionnaire (1986) monitoring projects — all were based on the voluntary work of ringers. He has been involved in the monitoring and conservation of all birds of prey in Finland. Pertti’s talent is to turn his enthusiasm for owls and birds of prey into knowledge that can be applied to practical management actions for bird conservation.

36 Beside his passion for biology, Pertti holds a special place in his heart for classical music. He “was born” in the Finnish National Opera, where his mother was the leading dramatic soprano. On the basis of his genetic makeup, Pertti has had the privilege to sing as a baritone soloist in professional male and mixed choirs on the stages of concert halls in more than 25 countries around the world, such as Carnegie Hall and Lincoln Centre in New York and Massey Hall in Toronto, and in Argentina, Brazil, China, Japan, and Russia.

Life behind the numbers: 55 years among the owls. – Why?

Saurola, P., Finnish Museum of Natural History, University of Helsinki, Finland

The first reason why I have been interested in owls, particularly in the Ural Owl (Strix uralensis), has been conservation. In Finland, the Ural Owl suffers from the lack of ideal natural nest-sites — large cavities and chimney-like stumps of big trees — which do not exist in commercial forests. In 1965, I started to provide nest-boxes for hole-nesting owls to compensate the losses caused by forestry.

The second reason why I have studied the Ural Owl has been science. The Ural Owl is an ideal species for population studies. The breeding female can be easily and safely trapped any time during the breeding season, even before egg laying. Trapping males is more difficult, but possible with experience and stamina. Because both sexes are strongly philopatric, it is possible to gather relevant and reliable long-term data of the same individuals. The Ural Owl is a generalist feeder and feeds on a wide variety of vertebrates, ranging from frogs and shrews to mammals and birds weighing up to several hundred grams. However, in Finland, the population dynamics of the Ural Owl has been highly dependent on 3–4-year cycles of microtines. My main goal has been to find out how different population parameters, such as age at first breeding, onset of egg-laying, clutch size, brood size, survival of different age-classes, recruitment, natal and breeding dispersal and lifetime reproduction, vary in relation to fluctuating environments. For comparison, I have collected similar data on the close relative, the Tawny Owl (Strix aluco), a newcomer from the south.

The third reason to spend much of my lifetime with Ural Owls has been the privilege to have an opportunity to be legally in close contact with different individuals of such a fascinating species, which defends its offspring with a kamikaze-like fearlessness and fierceness. Over the decades, some individuals have become my “old friends” and even been given anthropomorphic names like Mama of Yltiö and Papa of Hyypiö. For the general public, the most interesting part of my study has been information about “divorces” and polygamy based on the real lives of owls.

My presentation will be a mixture of memories through the decades and selected facts of the life of the Ural Owl and Tawny Owl.

37 Session 5: Data integration and population analysis I

Using data integration to evaluate ecological processes across scales KL

Zipkin, E., Department of Integrative Biology, Michigan State University, MI, USA

Emerging data integration approaches, such as integrated population models (IPMs) and integrated distribution models (IDMs), incorporate multiple data sources within a unified analytical framework. Such approaches are exceptionally valuable for research conducted at broad extents or across multiple scales as it is rarely possible to estimate all ecological parameters of interest using only a single data source. Multiple data sources can inform various components of the study system that operate at different spatial or temporal scales, providing unique or complementary information on biological patterns and/or processes. Multi-scaled studies increasingly use data integration techniques to improve precision of parameter estimates, account for multiple sources of uncertainty, estimate parameters for which no explicit data exists, and produce predictions of future ecosystem states and processes across space and time. As a result of these advantages, data integration has become a powerful approach for expanding the spatiotemporal coverage of research. I will highlight these benefits by showcasing research case studies. I will also present some of the key ongoing challenges of integrated modeling that are exacerbated in broad and multi-scale research such as data scale mismatches, unbalanced data, sampling biases, and model development and assessment. Use of data integration techniques has increased rapidly in recent years. Given the inferential value of such approaches, we should expect sustained development and wider application across ecological disciplines.

38 A semi-spatial integrated population model to assess population dynamics of a re- colonizing species

Petracca, L. S., Washington Cooperative Fish and Wildlife Research Unit and School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Gardner, B., School of Environmental and Forest Sciences, University of Washington, WA, USA

Maletzke, B. T., Washington Department of Fish and Wildlife, WA, USA

Bassing, S. B., School of Environmental and Forest Sciences, University of Washington, WA, USA

Long, R. A., Woodland Park Zoo, WA, USA

Ransom, J. I., National Park Service, North Cascades National Park Service Complex, WA, USA

Shipley, L. A., School of the Environment, Washington State University, WA, USA

Thornton, D. H., School of the Environment, Washington State University, WA, USA

Converse, S. J., U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit, School of Environmental and Forest Sciences and School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Recolonizing species exhibit unique population dynamics, namely dispersal to and colonization of new areas, that are important to understand from a conservation and management perspective. Integrated population models (IPMs) have proven useful for making inference about population dynamics by integrating multiple data streams, including data relevant to population state and demographic rates. More recently, spatially explicit integrated population models (SIPMs) have leveraged the power of spatial capture-recapture, resulting in a spatially explicit model of population dynamics. SIPMs, however, require information on the spatial observation process to correctly model spatially explicit data. In a recolonizing population of wolves in Washington, USA, we were lacking data on the spatial observation process but wanted to leverage the power of SIPMs to describe the recolonization process, which is critical to recovery. We used GPS collar, camera trap, and count data collected from this population to develop a semi-spatial integrated population model, in which non-spatial data on survival and reproduction are integrated into a semi-spatial model comprising [1] territory size estimated from telemetry data, [2] probabilities of dispersal specific to month, age, and pack size estimated from telemetry data, [3] least-cost movement paths between territories of origin and potential new wolf territories (estimated from telemetry data and a second-order resource selection function [RSF]), [4] a Bernoulli process by which a wolf can remain in a potential territory based on an underlying occupancy model (estimated using camera trap data), and [5] territory-specific count data. Our semi-spatial IPM can be used to assess population dynamics with a spatial component and determine how management strategies can affect population dynamics and recovery.

39 How accurate are integrated population models in estimating the contribution of immigration to changes in population growth?

Paquet, M., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Knape, J., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Arlt, D., Department of Ecology, Swedish Species Information Centre, Swedish University of Agricultural Sciences, Sweden

Forslund, P., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Pärt, T., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Flagstad, Ø., Norwegian Institute for Nature Research, Norway

Jones, C., Mauritian Wildlife Foundation, Mauritius; Durrell Wildlife Conservation Trust, UK

Nicoll, M., Institute of Zoology, Zoological Society of London, UK

Norris, K., Natural History Museum, London, UK

Pemberton, J., Institute of Evolutionary Biology, School of Biological Sciences, University of Edinburgh, UK

Sand, H., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Svensson, L., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Tatayah, V., Mauritian Wildlife Foundation, Mauritius

Wabakken, P., Faculty of Applied Ecology and Agricultural Sciences, Inland Norway University of Applied Sciences, Norway

Wikenros, C., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Åkesson, M., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Low, M., Department of Ecology, Swedish University of Agricultural Sciences, Sweden

Estimating the contribution of demographic parameters to changes in population growth is essential for understanding how populations are fluctuating. Integrated Population Models (IPMs) offer the possibility to estimate contributions of additional demographic parameters, for which no data have been explicitly collected: often immigration. Such parameters are often found to be the main driver of population growth. Accuracy in the estimation of their temporal variation, and consequently their

40 contribution to changes in population growth rate, has however not been investigated. To quantify the magnitude and cause of potential biases when estimating the contribution of immigration using IPMs, we use both empirical and simulated data. We used data from wild populations with known immigration (Soay Sheep Ovis aries and Mauritius Kestrel Falco punctatus with zero and Scandinavian Wolf Canis lupus with near-zero immigration), implementing IPMs that estimate the effect of immigration. We also used simulated data from controlled scenarios to examine the origin of bias and its magnitude depending on IPM parametrization, the level of temporal variation in immigration, and sample size. IPMs strongly overestimated the contribution of immigration to changes in population growth in the wild populations, where the true number of immigrants was zero or near-zero (depending on IPM formulation: Sheep 4.5-32.9%, Kestrel 19.7-97.8%, Wolf 89.7-99.6%), and in scenarios when immigration was simulated with zero temporal variation. Although the estimation of immigration in the simulation study became more accurate with increasing temporal variation and sample size, it was often not possible to distinguish between an accurate estimation from data with high temporal variation versus an overestimation from data with low temporal variation. To minimise the risk of overestimating the contribution of immigration (or any additional parameter) in IPMs, we recommend to: (i) look for evidence of variation in immigration before investigating its contribution to population growth, (ii) model a simulated dataset for comparison to the real dataset, and (iii) use explicit data on immigration where possible. We provide details on what evidence to look for and how to run a simulation study to help researchers identify when estimates from additional parameters are likely to be accurate.

41 Using multispecies integrated population model to understand interspecific com- petition: A case study on Great Tits and Blue Tits

Queroue, M., Centre d’Ecologie Fonctionnelle et Evolutive (CEFE), Univ. Montpellier; Univ. Paul Valéry Montpellier 3, France

Henry, P. Y., Centre de Recherches sur la Biologie des Populations d’Oiseaux (CRBPO), Sorbonne Univer- sités, France

Barraquand, F., Institute of Mathematics of Bordeaux, CNRS, France

Gimenez, O., Centre d’Ecologie Fonctionnelle et Evolutive (CEFE), Univ. Montpellier; Univ. Paul Valéry Montpellier 3, France

Understanding how the effects of interspecific and intraspecific competition can affect the demographic parameters of species is challenging due to the complex community dynamics of interacting species. For reliable inference, there is a need to integrate information over several biological levels (individuals - populations – communities). Integrated Population Models (IPMs) combine population counts and capture–recapture data to infer demographic parameters and population dynamics. IPMs have recently been extended to the community level with the development of community IPMs to fit multispecies nonlinear matrix models to multiple data sources. In this study, we showcase this recent method with a case study on great tits (Parus major) and blue tits (Cyanistes caeruleus), two bird species that compete for food resources and nesting cavities. We combine capture-recapture data from the French Constant bird banding Effort Sites (CES) scheme and population counts from the French Breeding Bird Surveys (BBS) within a multispecies IPM, over the period 2001-2019. We estimate species-specific demographic parameters and quantify fluctuations in population size for both species simultaneously while accommodating heterogeneity in habitat conditions (deciduous forest vs. other habitats). Through multispecies IPMs, we estimate explicit relationships between population size of one species and its effect on demographic parameters of the other species (survival, breeding parameters). This new community IPM framework allows improving our understanding of the effects of inter- and intra-specific interactions and habitat on avian population dynamics.

42 Beyond bespoke: standardized integrated population models reveal drivers of pop- ulation dynamics of migratory birds across latitudes

Burgess, M. D., RSPB Centre for Conservation Science, UK; PiedFly.Net, UK; Centre for Research in Animal Behaviour, University of Exeter, UK

Robinson, R., British Trust for Ornithology, UK

Nater, C. R., Centre for Biodiversity Dynamics (CBD), Norwegian University for Science and Technology, Norway

Identifying key drivers of population dynamics for migratory species is challenging because they are exposed to a variety of different environments throughout their annual cycle. Nonetheless, such knowledge is often crucial for guiding conservation efforts to prevent and reverse decline of species of conservation concern. Many species of migratory birds fall into this category and the recent rise of integrated data analysis has given unique insights into the drivers underlying dynamics of single populations. Conclusions from such analyses are, however, often variable and difficult to compare, and standardized analyses of multiple populations may be necessary to obtain insights relevant at a range-wide scale. We developed an integrated population model that is general enough to be applied to a variety of species of (migratory) hole-nesting birds and can be used for multi-population analyses. Our implementation of the model focused on both accessibility and efficiency by (1) linking it directly to the standardized data format provided by the SPI-Birds data hub, a growing database for long-term individual based studies, and by (2) capitalizing on the versatility of NIMBLE. By fitting the model to mark- recapture and nest box survey data from seven populations of pied flycatchers (Ficedula hypoleuca) breeding at different latitudes encompassing the UK distribution, we show substantial variation in both averages of and environmental impacts on key demographic parameters across populations. Both population size and annual survival covaried among the different breeding populations, indicating that environmental factors during the migration and non-breeding phase may be influential. We put this into perspective with a detailed analysis of the relative importance of different demographic components to realized population growth rate using transient life table response experiments (LTREs). We thus showcase how integrated population models, which are usually tailored to one specific study population, can be generalized to provide a multi-population perspective of demographic drivers across large spatial scales, and highlight the importance of facilitating their use by linking them to relevant databases.

43 Integrated population model for Ferruginous Hawk and Golden Eagle in Wyoming, USA, to assist with a long-term, state-wide monitoring plan

Sanderlin, J. S., USDA Forest Service, Rocky Mountain Research Station, AZ, USA

Wallace, Z. P., Wyoming Natural Diversity Database, University of Wyoming, WY, USA

Olson, L., USDA Forest Service, Rocky Mountain Research Station, MT, USA

Squires, J., USDA Forest Service, Rocky Mountain Research Station, MT, USA

Ferruginous hawk (Buteo regalis) and golden eagle (Aquila chrysaetos) have been identified as Species of Greatest Conservation Need in Wyoming, USA. Wyoming has extensive, high-quality habitat for both species, as well as high pressure for conventional (oil and gas) and renewable (wind) energy development in this habitat. Mitigation and compliance monitoring for energy development in habitat exists but does not cover a large scale covering the entire state. The main objective was to develop a model that could estimate trends in demographic rates for both species across the species’ ranges within Wyoming. We developed an integrated population model (IPM) for ferruginous hawk and golden eagle that incorporates several data sources (detection/non-detection data for territory occupancy and nest surveys for productivity) using dynamic N-occupancy models. The state-space model incorporated multiple age classes and breeding states to obtain demographic trend estimates of the following parameters: abundance, occupancy, survival, and recruitment. We conducted a simulation study using information from the literature to explore sensitivity in parameters (namely, initial abundance, survival, annual productivity, detection probability of breeding territories) with the power to detect trends with variable effort (quantified by number of sites, number of years, and number of territory occupancy sessions) using an IPM. A secondary objective was to determine the information gain in parameters of interest from combining all data sources versus using a single data source of territory-occupancy data only. Using the IPM, there were estimates of latent variables available that otherwise would not be possible using only territory occupancy data. This information is informative for designing a long-term, state-wide monitoring plan and illustrating trade-offs with precision gain and effort if additional data sources could be incorporated into future monitoring designs.

44 Session 6: Population management

A decision analytical framework for guiding demographic estimation KL

Runge, M., U.S. Geological Survey Eastern Ecological Science Center at the Patuxent Research Refuge, Laurel, MD, USA

Tucker, A., U.S. Geological Survey Iowa Cooperative Fish and Wildlife Research Unit, IA, USA

The development of methods to estimate wildlife occupancy, abundance, movements, and demo- graphic parameters was originally motivated by the need for information to support conservation and management decisions. The field of statistical ecology is now broad and multifaceted, but here we will explore the use of analytical methods in applied contexts. We seek to understand how information flows from data to analyses to decisions, and indeed, how information can flow the other way to influence study design. The field of decision analysis provides a framework for embedding estimation in the decision process and ensuring the applicability of results.

45 Integrating understanding to inform management, research and monitoring

Yackulic, C. B., U.S. Geological Survey, Southwest Biological Science Center, AZ, USA

Runge, M., U.S. Geological Survey, Patuxent Wildlife Research Center, Laurel, MD, USA

Management of complex ecological systems is based on a partial and fragmented understanding of system dynamics. While decisions are sometimes informed solely by integration of data collected at appropriate spatial scales within the system of interest, such data are not always available or may omit important understanding derived from other sources. Sources that include field studies at finer spatial extents, detailed lab experiments, and studies in similar systems. Even when systems are well understood within the range of observed conditions, climate change and invasions present novel conditions and managers may pursue novel management actions. Given partial and fragmented understanding and uncertainty associated with novel conditions, forecasting system responses to inform management decisions often requires formalizing understanding through expert elicitation and explicit choices regarding model structure. Here, we discuss efforts to develop forecasting models that integrate diverse forms of understanding to inform management decisions in the Colorado river in its Grand Canyon segment. Modelling in the Colorado river system was focused on interactions among three fish species, including two invasive species and one federally-listed, endemic species. Drivers of population dynamics (including interspecific interactions) are relatively well understood for one, long- established, invasive species (rainbow trout) and the endemic species (humpback chub). Drivers of the second, rapidly expanding, invasive species (brown trout) population and its potential impacts on the endemic species are poorly understood. Managers were faced with decisions regarding the appropriate management actions in response to the emerging threat posed by an expanding population of brown trout. Parameterizing models to forecast system dynamics required outlining a suite of competing hypotheses involving drivers of recent population increases, estimating parameter associated with different hypotheses by integrating mark-recapture and count data in a multistate model using priors on survival derived from other systems, and eliciting expert opinions on the impacts of various potential management actions. Structured decision analysis suggested that the most effective management action depended on which hypothesis was assumed to describe underlying population dynamics. Results of the structured decision analysis have helped guide ongoing research and monitoring over the last three years by identifying critical uncertainties in our understanding.

46 Data analysis and modeling for endangered species listing decisions

McGowan, C. P., U.S. Geological Survey, Florida Cooperative Fish and Wildlife Research Unit, University of Florida, FL, USA

Beisler, W., U.S. Fish and Wildlife Service, National Conservation Training Center, WV, USA

Angeli, N., U.S. Virgin Islands Division of Fish and Wildlife, U.S. Virgin Islands, USA

Rivenbark, E., U.S. Fish and Wildlife Service, Ecological Services, GA, USA

The US Fish and Wildlife Service (USFWS) has initiated a re-envisioned approach for providing decision makers with the best available science and synthesis of that information, called the Species Status Assessment (SSA), for endangered species decision making. The SSA report is a descriptive document that provides decision makers with an assessment of the current and predicted future status of a species. These analyses support all manner of decisions under the US Endangered Species Act, such as listing, reclassification, and recovery planning but have been widely applied to listing determinations. Novel scientific analysis and predictive modeling in SSAs could be an important part of rooting conservation decisions in current data and cutting edge analytical and modeling techniques. Ideally assessing current status uses analysis of available data to estimate demographic parameters such as abundance, trend, survival and the future condition analysis uses the results of that analysis to make predictions about population and species trajectories. Here, we describe an analysis of available presence absence data to assess the current condition of Eastern Black Rail across its range in a dynamic occupancy analysis and how we used the results of that analysis to develop a site occupancy projection model where the model parameters (initial occupancy, site persistence, colonization) were linked to environmental covariates, such as land management and land cover change (sea-level rise, development, etc.). Occupancy probability and site colonization were low in all analysis units, and site persistence was also low, suggesting low resiliency and redundancy currently. Extinction probability was high for all analysis units in all simulated scenarios except one with significant effort to preserve existing habitat, suggesting low future resiliency and redundancy. With the results of these data analyses and predictive models, the USFWS concluded that protections of the Endangered Species Act were warranted for this subspecies.

47 Hierarchical models to link occupancy and habitat dynamics to management deci- sions: recovering endangered Florida Scrub and Scrub-Jays

Eaton, M. J., U.S. Geological Survey, Southeast Climate Adaptation Science Center, NC, USA

Breininger, D. R., NASA, Ecological Monitoring Program, Kennedy Space Center, FL, USA

Nichols, J. D., U.S. Geological Survey, Patuxent Wildlife Research Center, MD, USA

Fackler, P. L., North Carolina State University, NC, USA

We describe the design and application of integrated, hierarchical models developed for an adaptive management program to restore an imperiled ecosystem and recover declining populations of an endemic habitat specialist. In collaboration with a group of land managers, we designed a set of models to estimate management-mediated transition rates among vegetation classes in two dominant scrub communities, and territory-level occupancy dynamics (colonization and extinction probabilities) of scrub-jays as a function of habitat conditions. Estimated vegetation and occupancy parameters provide the system transition for a Markov Decision Process to identifying optimal, state-dependent management policies for any combination of habitat community, vegetation state, occupancy status and the conditions of neighboring territories. Using a decision analytic framework, we worked with managers to establish objectives, identify state-variable conditions and management alternatives, and elicit expert opinion for the initial parameterization of the vegetation model. We treated expert knowledge as pseudo-observations for Dirichlet priors and then updated transition posteriors annually as management activities were implemented. This approach to jointly model habitat and species population dynamics is likely to be of interest from a management perspective by offering a direct link between available management actions and the desired species response. Developing a collaborative framework from the ground up increases our understanding of complex ecological relationships while engaging and explicitly informing management and conservation efforts.

48 Integrated simulations of demography, monitoring and decisions for conservation of endangered species

Canessa, S., Institute of Zoology, Zoological Society London, England

Adaptive management (AM) remains underused and poorly implemented in the conservation of endangered species. An important barrier is that AM for recovery plans must balance conceptual rigor, technical feasibility, and biological and management realism. As plans become more complex, this balance becomes both more necessary and more challenging to achieve. I will present an intuitive approach to AM based on simulations that integrate three components: (1) a real ecological process, (2) a monitoring and learning process, and (3) a decision process. This simulation-based approach to AM is essentially a combination of demographic modelling, analysis of monitoring data, and decision analysis. It integrates processes and tools that many recovery groups already use separately. Therefore, it can be more realistic and easier to interpret and embed in recovery group activities than formal optimization approaches. I will illustrate the application of this approach to two real-world case studies. The first was a small-scale headstarting program for yellow-bellied toads (Bombina variegata) in Italy between 2015 and 2019. Experimental releases were used to choose the most cost-effective method, while seeking to minimize risks to harvested source populations. Although ultimately AM could not guarantee success, it helped decision-makers learn about key uncertainties while following a precautionary principle. The second case study (ongoing) is the design of a 20-year program for hihi (Notiomystis cincta) in Aotearoa-New Zealand. Learning and management must be integrated to optimize supplementary feeding and translocations across multiple sites. This simulation combines a simple population model, a Bayesian state-space model and several complex decision rules to update knowledge and adjust management. In both case studies, simulations allowed managers to evaluate learning and management over spatial, temporal, or management scales that would otherwise be too large for optimization and too complex for intuition. The simulation-based approach can complement formal optimization methods and improve AM uptake, particularly for conservation programs that require greater realism and complexity.

49 Session 7: Survival estimation

Models for survival: from φta to φfc and everything in between KL

Matechou, E., School of Mathematics, Statistics and Actuarial Science, University of Kent, UK

In this talk, I will discuss different assumptions and constraints that are frequently employed when modelling survival probabilities using ecological data and demonstrate them using a number of case studies. I will also present some recent work on Bayesian nonparametric models for the distribution of survival or length-of-stay in ecological studies.

50 Large data and individual heterogeneity models: when worlds collide

King, R., School of Mathematics, University of Edinburgh, UK

Sarzo, B., Foundation for the Promotion of Health and Biomedical Research of Valencia Region, Spain

Elvira, V., School of Mathematics, University of Edinburgh, UK

Individual random effect (or hierarchical) models are commonly used to describe individual heterogene- ity on demographic parameters of interest, most notably survival probabilities. These models typically lead to a likelihood that is expressible only as an analytically intractable integral. Common techniques for fitting such models to data include, for example, the use of numerical approximations for the integral, or a Bayesian data augmentation approach. However, as the size of the data set increases (i.e. the number of individuals increases) and/or the model increases in complexity, these computational tools may become computationally infeasible. We present an efficient Bayesian model-fitting approach, whereby we initially sample from the posterior distribution of a smaller sub-sample of the data, before correcting this sub-sample posterior distribution to obtain estimates of the posterior distribution of the full dataset by means of an importance sampling approach. We consider several practical issues, such us the sub-sampling mechanism, computational efficiencies and combining multiple sub-sampling estimates. The approach is applied to a real dataset of guillemots where approximately 30,000 indi- viduals are ringed and recorded within the given study period. A standard Bayesian data augmentation approach is computationally infeasible (on the order of > 1 week); whereas our approach can provide reliable estimates in a substantially shorter period of time (on the order of hours).

51 Laplace approximations for capture-recapture models in the presences of individ- ual heterogeneity

Herliansyah, R., School of Mathematics, University of Edinburgh, UK; School of Statistics, Institut Teknologi Kalimantan, Indonesia

King, R., School of Mathematics, University of Edinburgh, Edinburgh, UK

King, S., School of Mathematics, University of Edinburgh, Edinburgh, UK

Capture-recapture studies are common for collecting data on wildlife populations. The populations in such studies are often subject to different forms of heterogeneity that may influence their associated demographic rates. We focus on the most challenging of these relating to individual heterogeneity. In particular we consider continuous time-varying individual covariates (such as weight or parasite load). In the most general case the associated likelihood is not available in closed form but only expressible as an analytically intractable integral. In particular, the integral is specified over the unknown individual co- variate values (if an individual is not observed its associated covariate value is also unknown). Previous approaches to dealing with these issues include numerical integration and Bayesian data augmentation techniques. However, as the number of individuals observed and/or capture occasions increases, these methods can become computationally expensive. We propose a new and efficient approach that approximates the analytically intractable integral in the likelihood via a Laplace approximation. We conduct a simulation study to assess the performance of the Laplace approximation for open capture-recapture models before considering real datasets (open population with individual covariates) and compare both the accuracy and computational resources with alternative approaches.

52 Misidentification errors in reencounters result in biased estimates of survival from CJS models: evidence and a possible solution using the robust design

Rakhimberdiev, E., Conservation Ecology Group, Groningen Institute for Evolutionary Life Sciences (GELIFES), University of Groningen, The Netherlands; Department of Vertebrate Zoology, Lomonosov Moscow State University, Russia

Karagicheva, J., NIOZ Royal Netherlands Institute for Sea Research, Department of Coastal Systems, The Netherlands; Utrecht University, The Netherlands

Saveliev, A., Institute of Ecology and Environmental Science, Kazan Federal University, Russia

Loonstra, A. H. J., Conservation Ecology Group, Groningen Institute for Evolutionary Life Sciences (GELIFES), University of Groningen, The Netherlands

Verhoeven, M. A., Conservation Ecology Group, Groningen Institute for Evolutionary Life Sciences (GELIFES), University of Groningen, The Netherlands

Hooijmeijer, J. C. E. W., Conservation Ecology Group, Groningen Institute for Evolutionary Life Sciences (GELIFES), University of Groningen, The Netherlands

Schaub, M., Swiss Ornithological Institute, Switzerland

Piersma, T., NIOZ Royal Netherlands Institute for Sea Research, Department of Coastal Systems, The Netherlands; Utrecht University, The Netherlands; Conservation Ecology Group, Groningen Institute for Evolutionary Life Sciences (GELIFES), University of Groningen, The Netherlands

Misidentification of marked individual animals due to tag misreads is unavoidable in most studies on wild populations. Models commonly used for the estimation of survival from observations of unique identifiers ignore this potential source of bias. With a simulation study we show that misidentification causes a systematic bias in estimates of survival obtained from the Cormack-Jolly-Seber (CJS) model: survival is positively biased, the bias increases with decreasing true survival and survival spuriously declines over time. We developed an extended robust design capture mark-resight (RDM) model that includes an estimation of correct identification to get unbiased survival estimates when resighting histories contain misidentification errors. The model assumes that resightings occur repeatedly within a season, which is in practice often the case when resightings from color-marked individuals are collected. We implemented the RDM model in a state-space formulation and also in an approximate, but computationally faster model (RDMa) in JAGS and evaluated their performances by using simulated and real capture-resight data on black-tailed godwits Limosa l. limosa. For a range of degrees of misidentification errors the novel RDM model provides unbiased and accurate estimates of survival, recapture/resighting and correct-identification probabilities. The RDMa model performed well for large datasets (> 25 individuals), with high resighting (> 0.3) and low misidentification (< 0.3) probabilities. For the field data on black-tailed godwits, the RDMa model estimated a probability of correct identification, Θ, of 0.92. The unbiased estimates of survival produced by RDMa model were lower than estimates by CJS: the adult survival was estimated at 0.94 vs 0.97 and juvenile at 0.29 vs 0.86. Given that

53 misidentification errors are likely to be common in particular in resighting data, we conclude that survival estimates from many studies obtained from such data from CJS models are likely to be incorrect. The bias becomes larger for higher probabilities of individual misidentification and inevitably increases as datasets become longer. Based on our results we recommend the use of the RDM model to provide unbiased parameter estimates.

54 Using multievent recovery and auxillary resighting data to improve inference from multistate mark-recapture studies

Kendall W. L., U.S. Geological Survey, Colorado Cooperative Fish and Wildlife Research Unit, Colorado State University, CO, USA

White, G. C., Department of Fish, Wildlife, and Conservation Biology, Colorado State University, CO, USA

Langtimm, C. A., U.S. Geological Survey, Wetland and Aquatic Research Center, FL, USA

Multistate mark-recapture models, and more generally multievent models, have proven very useful in modeling wildlife populations where individuals transition among states (geographic, life history, disease, etc.) during their lifetime. In some cases, in addition to captures collected systematically in well-defined study areas and study periods, additional sightings of marked individuals are possible in other areas or times. These sightings could be incidental, such as the case where fishermen or birders submit photos of foraging albatross where leg bands can be read. These sightings could also be strategic, such as in manatee studies where observers focus on warm water sites in winter, but also seek manatees in other areas and during other times of year. These auxiliary sightings could provide information just on survival, or in some cases could inform which state (e.g., manatee wintering area) an individual occupies. We extend current mark-recapture methodology to incorporate dead recoveries and auxiliary sightings, both generic and state-specific, into multistate models, allowing for uncertainty about state assignment for recovered, observed or captured individuals, and accounting properly for the potentially continuous collection of resighting and recovery data (sensu Barker 1997). We apply the model to manatee monitoring data where individuals move among three areas of the Gulf Coast from winter to winter, and are also sighted in additional areas and seasons. We test hypotheses about spatially explicit effects of severe cold weather or red tide events on survival and movement among areas. We demonstrate this modeling approach using maximum likelihood estimation via program MARK.

55 Session 8: Data integration and population analysis II

Integrating tracking and resight data enables unbiased inferences about migratory connectivity and winter range survival from archival tags

Rushing, C. S., Department of Wildland Resources and the Ecology Center, Utah State University, UT, USA

Van Tatenhove, A. M., Department of Wildland Resources and the Ecology Center, Utah State University, UT, USA

Sharp, A., Department of Wildland Resources and the Ecology Center, Utah State University, UT, USA

Ruiz-Gutierrez, V., Cornell Lab of Ornithology, Cornell University, NY, USA

Freeman, M. C., U.S. Geological Survey, Patuxent Wildlife Research Center, Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Sykes, P. W., U.S. Geological Survey, Patuxent Wildlife Research Center, Warnell School of Forestry and Natural Resources, University of Georgia, GA, USA

Sillet, T. S., Smithsonian Migratory Bird Center, National Zoological Park, Washington, DC, USA

Archival geolocators have transformed the study of small, migratory organisms but analysis of data from these devices requires bias correction because tags are only recovered from individuals that survive and are re-captured at their tagging location. We show that integrating geolocator recovery data and mark-resight data enables unbiased estimates of both migratory connectivity between breeding and non-breeding populations and region-specific survival probabilities for wintering locations. Using simulations, we first demonstrate that an integrated Bayesian model returns unbiased estimates of transition probabilities between seasonal ranges. We also used simulations to determine how different sampling designs influence estimability of transition probabilities. We then parameterized the model with tracking data and mark-resight data from declining Painted Bunting (Passerina ciris) populations breeding in the eastern United States, hypothesized to be threatened by the illegal pet trade in parts of their non-breeding range. Consistent with this hypothesis, we found that male buntings wintering in Cuba were 20% less likely to return to the breeding grounds than birds wintering elsewhere in their range. Improving inferences from archival tags through proper data collection and further development of integrated models will advance our understanding of full annual cycle ecology of migratory species.

56 Spatial integrated models foster complementarity between monitoring programs in producing large-scale ecological indicators

Lauret, V., CEFE, Univ. Montpellier; Univ. Paul Valéry Montpellier 3, France

Labach, H., MIRACETI, Connaissance et conservation des cétacés, France

Turek, D., Williams College, Department of Mathematics & Statistics, MA, USA

Authier, M., PELAGIS, La Rochelle Université, France

Gimenez, O., CEFE, Univ. Montpellier; Univ. Paul Valéry Montpellier 3, France

Obtaining relevant information about large-scale population dynamics from a single monitoring pro- gram is challenging, and often several sources of data, possibly heterogeneous, need to be integrated. In this context, spatial integrated models combine multiple data types into a single analysis to quantify population dynamics of a targeted population. Using available information at different spatial or temporal scales, spatial integrated models have the potential to produce detailed ecological estimates that would be difficult to obtain if data were analyzed separately. So far, these models are available for open populations to estimate demographic parameters (survival, recruitment), therefore requiring data collected in long-term monitoring programs. However, we often need to quantify population abundance and density in closed populations. Adapting the method developed by Chandler et al. (2018), we showcase the implementation of spatial integrated models to closed populations. We analyzed spatial capture-recapture data together with distance-sampling data to estimate abundance and density. Focusing on the Mediterranean bottlenose dolphins (Tursiops truncatus) as a case study, we combined 21,464 km of at-sea photo-identification surveys collecting spatial capture-recapture data with 24,624 km of aerial line-transect following a distance-sampling protocol. We compared the performances of the spatial integrated model, with that of the distance sampling model, and the spatial capture-recapture model separated. We discussed the benefits of using a spatial integrated model in the context of the assessment of French Mediterranean bottlenose dolphin conservation status to inform continental scale public policies. Spatial integrated models are widely applicable and relevant to conservation research and biodiversity assessment at large spatial scales.

Reference:

Chandler RB, Hepinstall-Cymerman J, Merker S, Abernathy-Conners H, Cooper RJ. 2018. Characterizing spatio-temporal variation in survival and recruitment with integrated population models. The Auk 135:409–426.

57 Integrating data types to reveal avian migration patterns across the Western Hemi- sphere

Meehan, T. D., National Audubon Society, NY, USA

Saunders, S. P., National Audubon Society, NY, USA

DeLuca, W. V., National Audubon Society, NY, USA

Michel, N. L., National Audubon Society, NY, USA

Grand, J., National Audubon Society, NY, USA

Deppe, J. L., National Audubon Society, NY, USA

Jimenez, M., National Audubon Society, NY, USA

Knight, E., National Audubon Society, NY, USA

Seavy, N., National Audubon Society, NY, USA

Smith, M., National Audubon Society, NY, USA

Taylor, L., National Audubon Society, NY, USA

Witko, C., National Audubon Society, NY, USA

Wilsey, C. B., National Audubon Society, NY, USA

For many avian species, migration patterns remain largely undescribed, especially across hemispheric extents. Advancements in tracking technologies and high-resolution species distribution models (i.e., eBird Status and Trends products (eBird ST)) have helped address several research needs, but there have been no attempts to date to integrate these independent data sources and leverage the unique strengths of each for describing migratory bird movements. To advance our understanding of bird migration at relevant spatial and temporal extents, we developed a novel method to integrate individual- based seasonal movement information (i.e., tracking and band re-encounter data) with temporally dynamic eBird ST occurrence using least-cost paths (LCPs). We apply our LCP modeling framework to 12 migratory bird species to describe spatial patterns across the Western Hemisphere during pre- and post- breeding migratory periods. We also use generalized additive mixed models (GAMMs) to both evaluate model fit and create a fully integrated prediction index for mapping seasonal migratory patterns for each species. We found that the LCP method of synthesizing tracking/banding data, migratory connectivity information, and eBird ST occurrence improved our understanding (i.e., via GAMM fits) of migratory patterns relative to eBird ST occurrence alone. In particular, LCPs substantially improved model fit for species with overwater movements and those that migrate over land where there are few eBird sightings (e.g., Amazon rainforest in South America) and thus, low predictive ability of eBird ST models.

58 Accounting for spatiotemporal dynamics strongly influences the success of conservation planning for wide-ranging migratory species, and our methodology provides a comprehensive approach for integrating distinct data types to reveal fine-scale patterns of animal movement. Further development and customization of our approach will continue to advance full annual cycle research and conservation efficacy for migratory birds globally.

59 Integrating demographic and epidemiological data improves dispersal quantifica- tion

Gamble, A., Department of Ecology and Evolutionary Biology, University of California Los Angeles, CA, USA

Garnier, R., CEFE, CNRS, University of Montpellier; EPHE, University Paul Valéry Montpellier 3, France

Chambert, T., CEFE, CNRS, University of Montpellier; EPHE, University Paul Valéry Montpellier 3, France

Gimenez, O., CEFE, CNRS, University of Montpellier; EPHE, University Paul Valéry Montpellier 3, France

Boulinier, T., CEFE, CNRS, University of Montpellier; EPHE, University Paul Valéry Montpellier 3, France

Dispersal is an important component of ecological and evolutionary dynamics. Dispersal rates are usually estimated using multi-site capture-recapture, biologging or population genetic approaches, but inference is often limited by small sample sizes and other method-specific limitations. Accurate estimates of dispersal rates are thus still scarce at scales relevant to (meta)population functioning. Epidemiological dynamics are structured in space and may also contain exploitable latent information about host movements, in particular dispersal. For instance, host exposure to vector-borne agents, such as Lyme disease Borrelia or West Nile virus, often exhibit strong spatial structuration, as it directly depends on the distribution of their vectors. Hence, because individual antibody profiles integrate information on infectious agents encountered in the past at various locations, they may also inform on past dispersal events. At the (meta)population scale, seroprevalences (i.e., proportion of individuals presenting antibodies against a specific agent, obtained from unmarked individuals) may also inform on dispersal. Inspired from the integrated population modelling framework, we propose to use this latent information by integrating multi-site capture-recapture data and seroprevalence data to produce joint estimates of dispersal probabilities, together with other eco-epidemiological parameters such as site-specific survival and infection probabilities. We explore the performance of this integrated model using data simulated under different scenarios. We show that including epidemiological data improves the estimation of dispersal probabilities, compared to the model using demographic (capture- recapture) data only. Importantly, this integrated model performs well even when small numbers of individuals are recaptured, highlighting its usefulness in resource-limited situations. We illustrate how this integrated model can be used to quantify dispersal in a seabird (black-legged kittiwakes Rissa tridactyla) population. Finally, we present several additional promising opportunities from combining demographic and epidemiological data to quantify ecological processes, highlighting the usefulness of epidemiological data as ecological markers.

60 List of Posters

Poster Session (Tuesday 8 June 2021)

Assessing the effects of permanent emigration on survival and population viability predictions in a long-lived territorial raptor

Badia-Boher, J.-A., Grup de Biologia de la Conservació, Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Universitat de Barcelona, Spain

Real, J., Grup de Biologia de la Conservació, Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Universitat de Barcelona, Spain

Parès, F., Grup de Biologia de la Conservació, Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Universitat de Barcelona, Spain

Hernández-Matías, A., Grup de Biologia de la Conservació, Departament de Biologia Evolutiva, Ecologia i Ciències Ambientals, Universitat de Barcelona, Spain

Robust estimates of survival are key to evaluate the dynamics and conservation status of populations. Survival is usually estimated using capture-mark-recapture analyses, but most study designs and modelling frameworks cannot distinguish between mortality and emigration from the study area. This can be especially important in territorial raptors, as tagging usually is focused on fledglings that may move far away from their birthplaces to breed and will not return to the study area throughout their lives. This phenomenon — known as permanent emigration — can lead to negatively biased estimates of survival, and consequently, wrong assumptions of the study population’s status, which can ultimately lead to implementing ineffective conservation actions. Here, we took advantage of a long-term intensive ringing and monitoring scheme on the regionally threatened Bonelli’s eagle (Aquila fasciata) conducted in Catalonia. We also considered data of eagles born in Catalonia and recruited in neighbouring populations of the Iberian Peninsula and France, thanks to simultaneous monitoring schemes in these populations. Such a degree of detail in the monitoring of contiguous populations is infrequent in most species and provides an excellent opportunity to estimate permanent emigration. We applied multistate capture-recapture methods to quantify the magnitude of permanent emigration in the population and assess its potential bias on survival estimation. In addition, we developed an individual-based population model to evaluate the potential implications of this bias on population viability predictions. Our results illustrate that omitting permanent emigration may lead to an underestimation of survival and skewed population predictions. Overall, we highlight the need to improve our understanding of permanent emigration in capture-recapture studies and to develop new methods to account for this potential source of bias.

61 Bayesian inference for models of avian primary moult

Boersch-Supan, P. H., British Trust for Ornithology, UK

Hanmer, H., British Trust for Ornithology, UK

Robinson, R. A., British Trust for Ornithology, UK

Biometric data collected during bird ringing or recapture can provide crucial information about mech- anistic drivers of vital rates inferred from mark-recapture analyses. Observations of moult state are commonly collected during bird ringing. They encapsulate information about a crucial part of the avian lifecycle. In most free-living bird populations moult progression and duration in individuals cannot be observed fully. Instead snapshot measurements of (re)captured individuals are typically used to infer these parameters on a population level. As an additional complication, recording of moult in the field may take various forms both in terms of the subset of the population that is sampled and whether moult is recorded as a categorical state, or a (semi-)continuous progression score.

Underhill & Zucchini (1989; Ibis 130:358) proposed a general modelling framework to accommodate many of these features, implemented in the R package moult (Erni et al. 2013; J Stat Soft 52:8). A simpler approach based on the probit GLM was suggested by Rothery & Newton (2002; Ibis 144:526) for use with categorical moult data, allowing for separate variances on start and end dates. We show that both models are special cases of more general categorical regression models and describe a Bayesian inference framework for this class of models. This framework allows the inclusion of hierarchical model structures to accommodate (i) the integration of moult data sets using different modes of recording, (ii) individual heterogeneity in moult timing and progression, and (iii) hierarchical spatial/temporal effects for multi-site/multi-season data sets.

We provide an R package ‘moultmcmc’ which implements fast inference for these models using Hamiltonian Monte Carlo samplers from Stan. We show that full Bayesian inference provides less biased estimates of moult timing and duration and more accurate uncertainty quantification than existing maximum likelihood methods. This allows a better understanding of moult — a key energetic constraint on avian migration, reproduction, and survival — in the context of the full annual cycle.

62 Estimating survival while accounting for partial mark loss in capture-mark-resight surveys

Bratt, A. E., Quantitative Ecology and Resource Management, School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Hostetter, N. J., U.S. Geological Survey, North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University, NC, USA

Converse, S. J., U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit, School of Environmental and Forest Sciences & School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Capture-mark-resight (CMR) methods are an effective way to estimate survival, and are generally less invasive than methods requiring physical recaptures. As a result, they are common in studies of endangered passerines. The marking assumptions of traditional CMR models include: 1) no mark loss; 2) individuals are correctly classified as marked; and 3) marked individuals are completely and correctly identified. In practice, these assumptions are frequently violated. For instance, in long-lived species or in populations that face harsh weather conditions, partial band loss may be common. Failure to account for partial band loss results in the underestimation of uncertainty in survival estimates and may introduce bias. Moreover, it typically results in the loss of valuable data. We introduce a novel hierarchical model that integrates information about individual activity centers, obtained via nest monitoring, to resolve individual identity in a species that exhibits partial band loss. We developed a custom MCMC sampler to sample individual encounter histories from encounter histories of partial band combinations. We perform a simulation study to assess the performance of this approach under a range of survival and detection rates. We apply this model to a case study of endangered Streaked Horned Larks (Eremophila alpestris strigata) in southwestern Washington State, USA. Leveraging all available mark-resight data will better inform population forecasts and management strategy evaluation for this species and other passerines that may lose bands.

63 The effects of data availability on integrated population model performance

Bratt, A. E., Quantitative Ecology and Resource Management, School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Cappello, C., Department of Biology, University of Washington, WA, USA

Duvall, A. J., School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Sipe, H. A., School of Environmental and Forestry Sciences, University of Washington, WA, USA

Warlick, A. J., School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Gardner, B., School of Environmental and Forest Sciences, University of Washington, WA, USA

Converse, S. J., U.S. Geological Survey Washington Cooperative Fish and Wildlife Research Unit, School of Environmental and Forest Sciences & School of Aquatic and Fishery Sciences, University of Washing- ton, WA, USA

Understanding the processes that control wildlife population dynamics is a critical component of conservation and management but can be limited by missing, sparse, or biased data. Integrated population models (IPMs) are an increasingly popular tool in ecology that can help overcome these limitations through the combination of disparate datasets into a single, unified analytical framework that can lead to improved precision, reduced bias, and the estimation of parameters that would otherwise be unidentifiable. However, despite a recent proliferation of ecological applications using IPMs, the performance of these models remains underexplored, particularly in situations where available data are sparse, of poor quality, or when data to inform specific parameters are entirely absent. Here we employ a simulation analysis to evaluate the ability of an IPM to return unbiased demographic rate estimates and detect abundance trends in a hypothetical passerine bird species across various data landscapes. We use relative bias and root mean squared error to compare model performance using an IPM with three datasets (count surveys, mark-recapture, and nest monitoring) versus an IPM with two datasets (with either mark-recapture or nest monitoring data excluded). Additionally, we compare model performance across a range of sampling scenarios — with varying detection probability and survey frequency — and demographic parameter values. We identify situations where IPMs may not lead to improved precision or reduced bias, highlighting the importance of accounting for both life history and sampling design in determining the benefits of IPMs, and providing useful insights into the effects of data quality on our ability to produce accurate parameter estimates. This information is foundational for navigating the trade-offs inherent in designing and implementing field survey programs that minimize costs and effectively inform management decisions aimed at monitoring or recovering species in need of conservation intervention.

64 An analytical approach to improve comparative studies of survival rates

Brusa, J. L., School of Environmental and Forest Sciences, University of Washington, WA, USA

Rotella, J. J., Department of Ecology, Montana State University, MT, USA

Banner K. M., Department of Mathematical Sciences, Montana State University, MT, USA

Hutchins, P. R., MT, USA

Although survival rates are a central component of life-history strategies of large vertebrate species, very few comparative studies investigating life-history traits have included survival rates. The paucity of these studies could have resulted from challenges associated with obtaining a reliable dataset or incorporating information on the probability scale. The few studies that have included survival rates or some metric of survival (e.g., life span) tend to ignore measurement error. This talk focuses on a technique to overcome the challenges of incorporating information on the probability scale and measurement error in comparative analyses. The approach uses Bayesian phylogenetically controlled regression with the flexibility to incorporate uncertainty in estimated survival rates and quantitative life-history traits as well as genetic similarity among species (with uncertainty). As with any comparative analysis, this approach makes several assumptions regarding the generalizability and comparability of empirical data from separate studies. The model is versatile in that it can be applied to any species group of interest and include any life-history traits as covariates. The utility of the model is demonstrated using both simulated and real data from pinnipeds, which is an excellent taxonomic group for comparative analyses. However, survival rate data for pinnipeds are scarce, and we also emphasize the importance of generating high-quality estimates of age-specific survival rates and information on other life-history traits that reasonably characterize a species. Overcoming these challenges will be necessary for addressing important questions related to broader ecological life- history patterns and how survival-reproduction tradeoffs might shape evolutionary histories of extant taxa.

65 Irrigation drives declines in farmland bird populations

Cabodevilla, X., Department of Zoology and Animal Cell Biology, Faculty of Pharmacy, University of the Basque Country (UPV/EHU), Spain

Wright, A. D., Department of Integrative Biology, Michigan State University, MI, USA

Villanua, D., Navarra Environmental Management (GAN-NIK), Spain

Arroyo, B., Instituto de Investigación en Recursos Cinegéticos (IREC) (CSIC-UCLM-JCCM), Spain

Zipkin, E. F., Department of Integrative Biology, Michigan State University, MI, USA

Assessing the effects of agricultural intensification on biodiversity is critical for developing effective management plans for farmland conservation. One factor that has not yet been thoroughly studied is the impact of irrigation on wildlife, despite significant increases in the surface area of irrigated farmlands since the mid-twentieth century. Irrigated farmlands now cover over 300 million hectares worldwide. Here, we apply a hierarchical multi-species occurrence model with a BACI (Before-After Control-Impact) design, to evaluate the impact of irrigation on bird species occurrence patterns. We use a Bayesian approach in JAGS and we account for imperfect detection using the replicate sampling occasions within each breeding season. Within the occurrence process we estimate species-specific intercept based on the site irrigation status. This approach is quite novel and was based on the chapter 16 of Kéry and Royle, 2020. Our study occurs in a 100-km2 area with rainfed agriculture in the Mediterranean region of northern Spain. We analyze a 13-year dataset comprised of the 47 most common bird species in the region using a multi-species hierarchical occurrence model. We examine how the implementation of irrigation altered the local bird community, identifying which species were negatively (or positively) impacted by changes to the local ecosystem. Irrigation had an overall negative impact on the bird community, with occurrence rates of most species (55.3%) decreasing and only a small fraction (10.6%) increasing after the onset of irrigation, leading to an overall reduction in site-level species richness. Irrigation was most detrimental for farmland birds (including steppe birds, which are of high conservation concern), but also for forest birds, shrubland birds, and non-specialist species that occur frequently in rainfed agricultural environments. The fact that only a few species responded positively to irrigation suggests that in the long term irrigation may lead to substantial negative changes within local bird communities, with less diversity and a lack of ecologically-important farmland species. The negative impact of irrigation on bird occurrences is likely due to the loss of nesting and foraging habitat arising from shifts in crops and/or loss of fallow lands. Irrigation schemes should thus be implemented carefully, avoiding areas with high species richness or high densities of endangered species. In cases where irrigation cannot be avoided, promoting diverse agrosystems, avoiding monocultures, and including interspersed rainfed crops and fallow lands may help to mitigate negative effects on local bird communities and their ecosystems.

66 Occupancy dynamics of the eastern population of Sandhill Cranes during the breed- ing period

Casabona I Amat, C., Département des sciences du bois et de la forêt, Université Laval, QC, Canada

Adde, A., Département des sciences du bois et de la forêt, Université Laval, QC, Canada

Lepage, C., Canadian Wildlife Service, Environment and Climate Change Canada, QC, Canada

Darveau, M., Département des sciences du bois et de la forêt, Université Laval, QC, Canada

Mazerolle, M. J., Centre d’étude de la forêt, Département des sciences du bois et de la forêt, Université Laval, QC, Canada

The eastern population of Sandhill Cranes (Antigone canadensis was extirpated from much of their historical range at the beginning of the 20th century. However, the population has been growing since the 1980s and is thought to have expanded its geographic range to Québec, Canada. We investigated the occupancy dynamics of this recent colonization during a 16-year period (2004–2019). To do so, we combined three datasets to increase the spatial coverage and the number of species occurrence records. These data sets consisted of the Canadian Wildlife Service helicopter surveys, the Second Atlas of Breeding Birds of Southern Quebec, and eBird. We hypothesized that habitat colonization is higher in landscapes with high proportions of open wetlands and agricultural landscapes. Results suggest that Sandhill Cranes have completed their colonization of western Québec and only recently started to nest in eastern areas. Areas with high cover of open wetlands were more likely to be occupied by Sandhill Cranes and less likely to become extinct. Colonization probability increased weakly with the cover of agricultural areas. Finally, the detection probability of Sandhill Cranes was highest for helicopter surveys (0.70) than from either Québec breeding bird atlas surveys or eBird. Our results do not support previous concerns about the potential negative impacts of cranes on crops during the breeding period.

67 Practical assessment of spatial capture-recapture

Chan, D., Department of Statistics, University of Auckland, New Zealand

Stevenson, B. C., Department of Statistics, University of Auckland, New Zealand

Spatial capture-recapture (SCR) has been successfully used to analyse data from passive acoustic surveys of terrestrial populations. However, there are not many applications of SCR on passive acoustic surveys of cetacean populations. I will present two case studies that will investigate whether SCR is suitable for passive acoustic surveys of cetacean populations. The first study has data on calls of migratory Eastern North Pacific gray whales. I discuss whether inference from an inhomogeneous density surface estimated with SCR is consistent with other methods and the literature. The second study looks at estimating the detection function of a playback experiment. Here, I will compare SCR’s estimation of a detection function against other methods. In particular, a detection function estimated with both detections and non-detections.

68 Climate and local weather’s influence on population decline in an avian aerial in- sectivore (Tachycineta bicolor): an integrated population model

Cox, A. R., Canadian Wildlife Service, Environment and Climate Change Canada, QC, Canada

Robertson, R. J, Department of Biology, Queen’s University, Kingston, ON, Canada

Roy, C., Canadian Wildlife Service, Environment and Climate Change Canada, QC, Canada

Bonier, F., Department of Biology, Queen’s University, Kingston, ON, Canada

In North America, avian aerial insectivores are declining faster than any other guild of birds. One hypothesis to explain this decline is that climate change causes reduced insect availability through increasing inclement weather, which prevent adults from foraging efficiently for themselves and their nestlings during the breeding season. Unfortunately, in most cases detailed demographic data necessary to test this hypothesis related to declines of aerial insectivores are unavailable. We monitored a population of box-nesting tree swallows (Tachycineta bicolor) in southern Ontario from 1975–2017. The time period covered includes time before and during the population decline for the species, which allows us to use the species as a model for studying decline in avian insectivores. We used monitoring and banding data gathered over 43 years to fit an integrated population model (IPM) to assess annual estimates of breeding productivity, juvenile, second-year (SY), and after-second-year (ASY) survival, SY and ASY immigration, and population growth rate. We included the effects of climate change and local weather conditions in the IPM model to test the linkages between those variables and productivity and survival in the monitored population. Our results indicate the declines observed in the focal population are strongly associated with local weather during the breeding season.

69 A multi-method, dynamic occupancy approach to monitor wildlife disease spread, elimination, and identify risk corridors

Davis, A. J., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Chipman, R. B., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Nelson, K. M., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Kirby, J. D., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Pepin, K. M., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Gilbert, A. T., United States Department of Agriculture, National Wildlife Research Center, CO, USA

Wildlife diseases can pose a significant threat to humans, livestock, and wildlife. Effective disease control requires an integrated strategy of robust surveillance, coordinated management and program monitoring. Wildlife disease surveillance samples are typically collected from a variety of sources including opportunistic samples, public reports, and targeted surveillance, which are each constrained by inferences and biases. We used a multi-method approach to simultaneously examine several approaches to wildlife rabies surveillance and evaluate their relative biases to gain an overall under- standing of detection probability. We analyzed 12 years of raccoon rabies virus (RABV) surveillance data to estimate the detection probabilities from different types of surveillance methods (e.g., strange acting reports, road kills, nuisance animals) and evaluated the probability of raccoon RABV elimination in an enzootic zone with active oral rabies vaccination (ORV) management in progress using a dy- namic occupancy approach adapted for multiple detection methods. We found that a combination of surveillance methods performed better than any single method. Samples from strange acting animals also had high detection probability and was the least biased method, but only comprised ∼ 6% of all samples. By using a multi-method occupancy approach, we were able to capitalize on the advantages of different surveillance methods while reducing the overall bias from any given method. The results from this analysis provide recommendations to managers on how to prioritize surveillance resources and identify areas of rabies freedom where management efforts have been successful.

70 Point count offsets for population sizes of north american landbirds

Edwards, B. P. M., Department of Biology, Carleton University, ON, Canada

Smith, A. C., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Sólymos, P., Boreal Avian Modelling Project, University of Alberta, AB, Canada

Robinson, B., Canadian Wildlife Service, Environment and Climate Change Canada, AB, Canada

Stralberg, D., Canadian Forest Service, Natural Resources Canada, AB, Canada

Grinde, A., Natural Resources Research Institute, University of Minnesota Duluth, MN, USA

Murray, A., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Harmer, T., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Pasher, J., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Gillespie, C., Klamath Bird Observatory, OR, USA

Gahbauer, M., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Niemi, G., Swenson College of Science and Engineering, University of Minnesota Duluth, MN, USA

Iles, D., Canadian Wildlife Service, Environment and Climate Change Canada, ON, Canada

Michel, N., National Audubon Society, OR, USA

Docherty, T., Faculty of Agricultural, Life and Environmental Science, University of Alberta, AB, Canada

Zlonis, E., Wetland Wildlife Populations and Research Group, Minnesota Department of Natural Resources, MN, USA

Bird monitoring in North America over several decades has led to millions of structured and semi- structured bird observations to be freely available in databases. One such survey, the North American Breeding Bird Survey (BBS), has provided the basis for estimates of both trends and population sizes for North American landbirds. However, there are a number of spatial gaps on several levels that exist in the BBS. QPAD is a detectability function that can translate counts of birds from any survey type into estimates of true density, allowing for disparate surveys to be integrated. The integration of multiple data sets can allow these data and spatial gaps to be filled for better estimates of status, trends, and population sizes. Here, we introduce NA-POPS: Point Count Offsets for Population Sizes of North American Landbirds, a large-scale, multi-agency project that was created to curate as many bird count observations as possible to generate an open-source database of detectability functions

71 for all North American landbirds. As of this study, NA-POPS has collected over 6 million data points spanning 246 projects from across North America. This has allowed for the generation of detectability functions for over 300 species of landbirds so far. We describe the methods used to curate these data and generate these detectability functions, as well as describe the open-access nature of the resulting database. We also describe our vision for use cases of the open-source detectability functions to improve population size estimates of North American landbirds.

72 Recent developments in model averaging

Fletcher, D., David Fletcher Consulting Limited, New Zealand

Model averaging has been widely used in statistical ecology for some time now. Recent research has provided a new range of tools for use in model averaging, from both Bayesian and frequentist perspectives. I will provide an overview of these developments and illustrate their use on mark- recapture models. I will also provide a summary of the pitfalls associated with model averaging, and discuss what it can and cannot provide.

73 Using integrated population models to understand drivers of population change over 55 years in two species of finnish owls

Francis, C. M., Canadian Wildlife Service, Environment and Climate Change Canada, National Wildlife Research Centre, ON, Canada

Saurola, P., Finnish Museum of Natural History, University of Helsinki, Helsinki, Finland

Understanding the drivers of population change and how they interact with different vital rates is necessary to predict how populations may change in the future, as well as to evaluate potential conservation implications. Some drivers, such as climate change may affect population dynamics in complex ways. For example, warmer winters could potentially increase over-winter survival but reduce breeding success by impacting prey abundance. We developed integrated population models to examine the impacts of climate and prey abundance on demographic parameters and population dynamics of two species of large owls that breed regularly in Finland: Ural Owl (Strix uralensis) and Tawny Owl (Strix aluco) using data from several different long-term monitoring programs. The Ural Owl has a predominantly northern range, while the Tawny Owl has relatively recently immigrated into Finland from the south; hence, both species may be expected to respond differently to drivers such as climate. Both species have been targeted by Finnish bird-ringers for about 55 years, resulting in mark-recapture-recovery datasets with more than 55,000 ringed Tawny Owls and 70,000 ringed Ural Owls, along with >15,000 and >20,000 respectively, combined live recaptures and dead recoveries. We analysed these data in joint recapture-recovery models, in a Bayesian framework using program MARK, to estimate age-specific survival rates for 4 age classes as well as an index of breeding propensity in relation to population drivers. From a separate data set, annual monitoring of nest boxes provides information on breeding propensity and productivity. Finally, annual population monitoring programs provide estimates of changes in the number of breeding pairs each year. Modelling each dataset separately, we found that all aspects of their demography, as well as overall population size, are affected by prey abundance (stage of the vole cycle), with increased nesting propensity, clutch size and fledging success when microtine rodents were most abundant during the spring and summer, and higher survival of all age classes when rodents were most abundant in winter. Overwinter survival was also influenced by mean winter temperature and snow cover, though to different degrees for each species. By combining data from each of these sources into hierarchical Bayesian integrated population models, we can estimate the relative importance of variation in each of these parameters (age-specific survival, nesting propensity, reproductive success and recruitment) on overall changes in population size.

74 Over a century of intentional killing: preliminary analyses of the contents of the EURING Data Bank reveal historical, geographical and cultural changes in the rela- tionship between man and birds across Europe and Africa

Funghi, C., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Istituto Superiore per la Ricerca Ambientale, Italy

Serra, L., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Istituto Superiore per la Ricerca Ambientale, Italy

Ambrosini, R., Department of Environmental Science and Policy, University of Milan, Italy

Spina, F., Area Avifauna Migratrice, Istituto Superiore per la Protezione e la Ricerca Ambientale (ISPRA), Istituto Superiore per la Ricerca Ambientale, Italy

The analysis of bird killing by man, with specific reference to illegal killing, is a module of the Eurasian- African Bird Migration Atlas project. Quantifying intentional killing, especially when illegal, has become an urgent conservation issue. Alarming estimates came from recent studies on illegal bird killing and taking, which also identified the pivotal need of robust data, allowing countries to set priorities, track trends and monitor effectiveness and compliance of environmental policies. EURING Data Bank is a fundamental resource to analyse and assess which species are most likely to be seriously affected by deliberate killing and which regions should be priorities for conservation action. Here, we present a wide historical, geographical and phenological overview of the EURING Data Bank content on this topic. Over 1,500,000 of dead birds, ∼400 species have been reported over ∼120 years and across ∼150 countries. First, we explored the spatial and temporal changes in condition and circumstances of dead birds in the recovery countries. Then, targeting the circumstance code, we show how death causes reported varied along decades and months of the year. The amount of intentional killing, weighed for the total number of dead birds reported, will be displayed through mean and density maps, both per country and per decade. Finally, we present preliminary results on geographic variation across Europe on the amount of intentional killing before and after the implementation of the Wild Birds Directive (1979/104 EEC, 2009/147 EC), as the key EU bird conservation legislation, through focusing on “always protected” vs “always huntable” species, and those which “became protected” with the Directive. This analysis gives an initial but unprecedent overview of the phenomenon of intentional killing across Europe and Africa, as a proxy of historical, geographical and cultural changes in the relationship between man and birds; it also confirms the unique potential of the EURING Data Bank as a long-term, large-scale repository of biological data, although several biases in reporting rate are being duly considered.

75 Trends in capture-recapture: ecological questions and methods over the past decade

Gimenez, O., CEFE, Univ. Montpellier; Univ. Paul Valéry Montpellier 3, France

McCrea, R., University of Kent, UK

King, R., University of Edinburgh, UK

Studying wildlife populations is challenging because not all individuals can be captured, identified and monitored exhaustively. Capture-recapture (CR) is a powerful framework to quantify demography and population dynamics of animal and plant populations, while explicitly accounting for the issue of imperfect detection.

The past decade has seen an explosion in the developments and applications of CR in statistics, conservation biology, ecology and evolution. We review the scientific CR literature over the 2009–2019 period and analyzed > 5000 papers using bibliometric and textual analyses. In particular, we used topic modelling to identify hot topics.

We discuss recent applications of CR, including the study of life-history trade-offs and senescence, threats to biodiversity like climate change and overexploitation, dominance and parental care, for- aging and anti-predation vigilance, resistance and tolerance to pathogens, overall providing a better understanding of changes in population size and composition and useful insights for management and conservation.

We also review recent CR methods. The last decade has seen much progress in how to mark and recapture animals with non-invasive methods, in particular camera trapping and genetic tagging. We have seen an uptake of state-space and hidden Markov models to estimate i) transitions between states (survival, dispersal, breeding or infection) and ii) the hidden states (disease/hybrid prevalence, sex ratio, home ranges locations). Also, there has been a growing interest in combining CR data with other sources of information on demography using integrated models.

We suggest future research and model development. New technologies produce new and more data with, e.g., drones, e-DNA, PIT tags and bioacoustics. In this context, survey design will remain important to collect data in ways that can inform ecological questions. We anticipate a growing interest in studying species interactions with community ecology and the assimilation of data using continuous CR models. Other avenues of research will be spatial CR models and the combination of CR data with even more sources of information, including data on unmarked animals and telemetry data. We suspect hidden Markov models will become a unifying framework for data integration. Last, cross-fertilization between disciplines will continue, including between ecology and climate science, artificial intelligence with deep learning for signal processing and machine learning for inference, and statistics and computer science for data combination and the analysis of big data.

The data and code are available on GitHub for reproducibility, https://github.com/oliviergimenez/ appendix_capturerecapture_review.

76 A point process approach to accounting for spatial variation in band return rates

Gonnerman, M. B., Department of Wildlife Fisheries and Conservation Biology, University of Maine, ME, USA

Linden, D., NOAA, Greater Atlantic Regional Fisheries Office, Gloucester, MA, USA

Shea, S. A., School of Food and Agriculture, University of Maine, ME, USA

Sullivan, K. M., Maine Department of Inland Fisheries and Wildlife, ME, USA

Kamath, P. L., School of Food and Agriculture, University of Maine, ME, USA

Blomberg, E. J., Department of Wildlife Fisheries and Conservation Biology, University of Maine, ME, USA

Lincoln estimators based on band recovery and harvest data are increasingly being used in Integrated Population Models (IPM) for evaluating population dynamics and guiding management of harvested species. A major assumption shared by all band recovery models is that banding and recoveries are representative of the whole population, and this assumption will be violated if major sources of variation in survival or harvest rates are not incorporated. The resulting bias in component parameters may thereafter affect estimated abundance and population growth. Unfortunately, collecting banding data is a time and resource intensive process, often necessitating a subset of capture sites relative to the spatial distribution of the entire population. Translating estimates from capture sites to a broader landscape scale often involves identifying and quantifying specific sources of variation in parameters, but this can be arduous and ultimately incomplete. Instead of explicitly measuring suspected sources of variation, we can estimate a spatial covariance function to account for how parameters vary across a landscape. We employed a Spatial Predictive Process within a band-recovery model to account for spatial variation in harvest rate of individuals. In addition to controlling for bias associated with unmeasured sources of variation, we were able to extend our scope of inference for harvest rates into areas where we did not deploy bands. We used simulated datasets to test model structure and assumptions and then applied our IPM to data collected from wild turkeys in Maine to assess harvest rates and abundance over time. These methods have broad applicability to band recovery models and should be considered when assessing populations over a large spatial scale.

77 Efficacy of positional and behavioral change-point models to determine ungulate parturition events

Gundermann, K. P., Department of Ecosystem Science and Management, Pennsylvania State University, PA, USA

Diefenbach, D. R., U.S. Geological Survey, Pennsylvania Cooperative Fish and Wildlife Research Unit, Pennsylvania State University, PA, USA

Walter, W. D., U.S. Geological Survey, Pennsylvania Cooperative Fish and Wildlife Research Unit, Pennsylvania State University, PA, USA

Buderman, F. E., Department of Ecosystem Science and Management, Pennsylvania State University, PA, USA

Animal behavior can be difficult, time-consuming, and costly to directly observe in the field. However, there is a rapidly expanding volume of innovative modelling methods that allow researchers to make inference about unobserved animal behaviors from movement data. In the movement literature, hidden Markov models fit to step-lengths and turning angles have become one of the primary ways that researchers identify behavioral states from movement data. However, we can frame behavioral state identification in a number of ways, and the optimal framework will depend on the ecology of the species in question. We propose two frameworks for identifying changes in movement behavior: a behavioral change-point model and a positional change-point model. We apply these models to two ungulate species white-tailed deer (Odocoileus virginianus) and elk (Cervus canadensis nelsoni) and compare the ability of each model to detect parturition events. Although variations of these methods exist in the literature, we propose these models in a holistic Bayesian framework to more easily compare the efficacy of each model across different species. To summarize the ability of these models to estimate the true parturition date, we will calculate the posterior difference between the estimated parturition event and the parturition date determined by vaginal implant transmitters. We will also systematically truncate the simulated data set to determine how much data, and what frequency, is needed after the change occurs for the model to correctly identify the temporal location of the change-point. This analysis will help inform managers as to the frequency at which locations should be obtained if they wish to accurately determine parturition events while maintaining the battery life of the collar. With the ability to accurately determine parturition events, managers will be better equipped to make informed management and conservation decisions for ungulate species of interest.

78 Understanding the dynamics of a reinforced bird population: the North African Houbara Bustard (Chlamydotis undulata) in Morocco

Harris, S. M., Cornell Lab of Ornithology, NY, USA

Robinson, O., Cornell Lab of Ornithology, NY, USA

Hingrat, Y., Reneco International Wildlife Consultants LLC

Ruiz-Gutierrez, V., Cornell Lab of Ornithology, NY, USA

Integrated population models (IPMs) are powerful tools for estimating population processes and iden- tifying the demographic drivers of population trends. As such, IPMs have high potential in evaluating the success of conservation actions and guiding management decisions. This potential is especially high in the context of species reintroductions, where estimation of vital rates for released individuals, particularly their survival, is crucial. Here, we explore this potential with the North African Houbara Bustard (Chlamydotis undulata undulata) in Morocco. As part of a large-scale conservation effort, more than 135,000 captive-reared houbara have been translocated over the past two decades to reinforce the declining wild population in Morocco. Using over 15 years of telemetry data on captive-bred and wild-born houbara, we developed a Bayesian multi-state capture-recapture model as a sub-model for an IPM, used to evaluate the progress of the reinforcement programme. We included covariates to assess individual and temporal variation in survival among captive-bred and wild-born individuals. Mean annual survival rates of captive-bred houbara were similar to that of wild-born birds, but sub- stantial temporal variation in survival was relatively asynchronous among the two groups. Age-specific survival rates also varied, increasing during the early years of life for captive-bred birds, while survival of wild-born individuals was relatively constant beyond the first year of age. These estimates highlight the importance of quantifying individual and temporal variation in demographic rates when assessing the demographic performance of captive-bred-released and wild-born individuals. Preliminary results from the IPM suggest that annual variation in survival plays an important role in driving the dynamics of the population. Ultimately, this IPM will aid the evaluation of current and future management decisions for this reinforcement programme.

79 Estimating abundance of an endangered whale accounting for spatial sampling bias

Miller, M. W., Institute of Marine Biology, University of Hawaii, HI, USA

Bradford, A. L., NOAA Pacific Island Fisheries Science Center, HI, USA

Franklin, E. C., Institute of Marine Biology, University of Hawaii, HI, USA

Baird, R. W., Cascadia Research Collective, WA, USA

Oleson, E., NOAA Pacific Island Fisheries Science Center, HI, USA

The Main Hawaiian Islands (MHI) insular population of False Killer Whales (Pseudorca crassidens) is listed as endangered under the U.S. Endangered Species Act. A recent analysis estimated that a minimum of 167 (95% CI=128–218) individuals remained in this population in 2015. Photo-identification encounters were analyzed with an open capture-recapture model. Most photographs were taken on the leeward sides of the MHI where survey conditions were optimal. However, satellite telemetry indicates high use areas for this population are largely on windward sides of the MHI.

We develop multi-state Jolly-Seber models that incorporate availability estimated from satellite teleme- try in an effort to estimate abundance correcting for spatial sampling bias. Movement in and out of the observable state are estimated outside of the Jolly-Seber framework using multi-state models. A resulting derived estimate of availability is hardcoded into the multi-state Jolly-Seber likelihood. Simu- lation is used here to explore performance of these models while varying true abundance, detection probability, availability and model structure. Initial results suggest such models may estimate true abundance across a range of parameter values and as such may be useful in accounting for spatial sampling bias.”

80 Spatially-autocorrelated detection probability in spatial capture-recapture

Moqanaki, E. M., Faculty of Environmental Sciences and Natural Resource Management, Norwegian University of Life Sciences, Norway

Milleret, C., Faculty of Environmental Sciences and Natural Resource Management, Norwegian Uni- versity of Life Sciences, Norway

Tourani, M., Faculty of Environmental Sciences and Natural Resource Management, Norwegian Uni- versity of Life Sciences, Norway

Dupont, P., Faculty of Environmental Sciences and Natural Resource Management, Norwegian Univer- sity of Life Sciences, Norway

Bischof, R., Faculty of Environmental Sciences and Natural Resource Management, Norwegian Univer- sity of Life Sciences, Norway

Spatial capture-recapture models (SCR) account for spatial heterogeneity in detection probability that may occur because of varying effort during the sampling, animal density and movement, and the underlying habitat. Unknown and, thus, unmodeled spatial heterogeneity in detection probability remains in many studies due to various factors. Consequences of the amount and structure of unmod- eled spatial variation in detection probability in SCR parameter estimates are not well understood. We explored how the magnitude and configuration of unmodeled, spatially-variable detection probability may influence the performance of SCR models. We simulated SCR data sets with different levels of spatially-variable and autocorrelated detection probability that may happen in wildlife monitoring studies. We then fitted a single-session SCR model ignoring this variation to the simulated data to assess the impact of model misspecification on the estimates of population size N and the spatial scale parameter of the half-normal detection function σ in terms of bias, precision, and the coverage probability of 95% confidence intervals. We found that highly-autocorrelated detection probability in SCR, modulated by the magnitude of that variation, can lead to pronounced negative bias (up to 75%), and reduction in precision (249%) and coverage probability of the 95% credible intervals associated with N estimates to as low as 0. Conversely, low levels of spatial autocorrelation in detection probability we tested did not lead to pronounced bias and only caused slight reductions in precision and coverage of N estimates. We observed negligible effects on σ estimates under all scenarios evaluated. We encourage SCR users to consider the impact that spatially-autocorrelated detection probability may have on their parameter estimates. SCR methods to detect and, if required, account for unknown, or partially unknown, spatial variation in detection probability are needed.

81 Leveraging detection information among species in community models of occu- pancy and abundance

Riecke, T. V., Swiss Ornithological Institute, Switzerland

Gibson, D., Warner College of Natural Resources, Colorado State University, CO, USA

Kéry, M., Swiss Ornithological Institute, Switzerland

Schaub, M., Swiss Ornithological Institute, Switzerland

The estimation of abundance and distribution and drivers of change in these parameters is central to the field of ecology. The continued development of hierarchical models that best utilize available information to inform these processes is a key goal of quantitative ecologists. However, much re- mains to be learned about simultaneously modeling the true abundance, presence, and trajectories of ecological communities. Simultaneous modeling of the population dynamics of multiple species provides an interesting mechanism to examine patterns in community processes and, as we emphasize herein, to improve species-specific estimates by leveraging detection information among species. Here we demonstrate a simple, novel approach to share information about observation parameters among species in hierarchical abundance and occupancy models, where we use shared random effects among species to estimate spatiotemporal heterogeneity in detection probability. We demonstrate the efficacy of our modeling approach using simulated abundance data, where we recover our simu- lated parameters using N-mixture models. Further, our approach substantially increases precision in estimates of abundance compared to models that do not share detection information among species. We then expand this model, and apply it to repeated presence-absence data collected on willow (Poecile montanus montanus, Poecile m. salicarius,& Poecile m. rheananus), marsh (Poecile palustris), crested (Lophophanes cristatus), coal (Periparus ater), blue (Cyanistes caeruleus), and great (Parus major) tits (Paridae) breeding across a P. montanus hybrid zone in northern Switzerland (2004–2020). We demonstrate evidence for interspecific competition on population persistence and colonization probabilities, where the presence of marsh tits reduces population persistence (β = −0.484; 95% CRI −1.465, 0.452) and colonization probability (β = −0.926; 95% CRI −1.689, −0.096) of sympatric willow tits, potentially decreasing gene flow among willow tit subspecies. While conceptually simple, our results have interesting implications for the future modeling of population abundance, colonization, persistence, and trajectories in community frameworks. We suggest potential extensions of the models described in this paper, and discuss how leveraging data from multiple species can improve model performance and ecological inference.

82 Non-native earthworms increase population densities of Plethodon cinereus, a com- mon woodland salamander of eastern north america

Scott, T., Département de Biologie, Université de Sherbrooke, QC, Canada

Bradley, R., Département de Biologie, Université de Sherbrooke, QC, Canada

Bourgault, P., Département de Biologie, Université de Sherbrooke, QC, Canada

Bélisle, M., Département de Biologie, Université de Sherbrooke, QC, Canada

Earthworms are newcomers to the province of Québec (Canada), as native species were unable to survive the last glaciation period that ended about 11,000 years ago. Since their reintroduction by Europeans over recent centuries, non-native earthworm species have had substantial impacts on nutrient cycling, soil hydrology, plant diversity and greenhouse gas emissions. However, less is known about the effects of non-native earthworms on higher faunal species, such as soil-dwelling vertebrates. Here, we report on a study that is currently investigating the possible effects of non-native earthworms on populations of Plethodon cinereus, a common woodland salamander. Earthworms may adversely affect P. cinereus by consuming the forest floor, thereby decreasing ground cover and soil moisture, which increases the risk of desiccation for P. cinereus. The consumption of the forest floor may also decrease the abundance of native preys for P. cinereus such as soil invertebrates. However, earthworms may themselves be a nutritious, soft-bodied prey for P.cinereus and abandoned earthworm burrows may also provide useful refugia for P.cinereus reducing predation and increasing overwintering success. Therefore, the objective of our study is to determine how non-native earthworms affect the demographics and diets of P. cinereus In the summer of 2019, we installed 25 maple wood cover boards (25 cm × 40 cm) in each of 38 mature sugar maple (Acer saccharum) forest sites, located in the Eastern Townships region of Québec. The sites had been selected to provide a wide gradient of earthworm population densities, which we quantified using hot mustard solution extractions and surface surveys. In the summer of 2020, we visited each site on six occasions and noted salamander numbers and body sizes under each cover board. We also began tagging individuals using visible implant elastomers and estimating population densities using capture-mark-recapture techniques. Preliminary results suggest that P. cinereus population size and density are positively correlated with earthworm abundance, contrary to a few similar studies performed in the U.S. The body size of salamanders was also positively correlated to earthworm abundance. As expected, forest floor depth was negatively correlated to earthworm abundance. In the summer of 2021, we intend to use gastric lavage techniques to determine the composition of diets at each site. When completed, our study is expected to provide useful information for guiding nature conservation policies in Québec.

83 A review of spatial capture-recapture in ecology

Tourani, M., Faculty of Environmental Sciences and Natural Resource Management, Norwegian Uni- versity of Life Sciences, Norway; Department of Wildlife, Fish, and Conservation Biology, University of California Davis, CA, USA

First described in 2004 by Efford (2004), spatial capture-recapture (SCR) has become popular tool in ecology. Like traditional capture-recapture, SCR methods account for imperfect detection when estimating ecological parameters. In addition, SCR methods use the information inherent in the spatial configuration of individual detections, thereby allowing spatially-explicit estimation of population parameters, such as abundance. Paired with advances in non-invasive survey methods, SCR has been applied to a wide range of species and habitats, allowing for population- and landscape-level inferences with direct consequences for conservation and management. I conduct a systematic review of SCR studies published since the first description of the method and provide an overview of their scope in terms of taxonomic groups targeted, geography, and data collection methods. In addition, I review approaches for analytical implementation and provide an overview of parameters targeted by SCR studies and conclude with current limitations and future directions in SCR methods.

84 Taking into account the probability of detection with two independent counting programmes

Vallecillo, D., Tour du Valat, Institut de recherche pour la conservation des zones humides méditer- ranéennes, France

Authier, M., Tour du Valat, Institut de recherche pour la conservation des zones humides méditer- ranéennes, France

Guillemain, M., Tour du Valat, Institut de recherche pour la conservation des zones humides méditer- ranéennes, France

Bouchard, C., Tour du Valat, Institut de recherche pour la conservation des zones humides méditer- ranéennes, France

Champagnon, J., Tour du Valat, Institut de recherche pour la conservation des zones humides méditer- ranéennes, France

N-mixture models jointly estimate local abundance and detection probability with a metapopulation design and counts without individual identification or distance measurements. Although highly promis- ing and powerful, N-mixture model inference is very sensitive to model assumptions, which can lead to large biases in estimation when even slightly violated. For instance, immigration and emigration across replicate survey occasions are frequent during monitoring (e.g. when disturbance caused by the first survey causes the displacement of individuals to be censused during the second survey, a violation of the population closure assumption). Here, we developed a log-normal state-space model that gathers data from two distinct counting protocols and allows derivation of an index of detection probability. We applied this model to ground and aerial surveys of wintering waterfowl in the Camargue where false-positive counting errors (double-counting and misidentification) are common in those gregarious species, taking into account an observer effect. Results make it possible to compare the counting methods for the three most abundant species.

85 Bayesian genetic mark-recapture methods for estimating seasonal river run size of stock populations

Wang, Y., Department of Statistics and Actuarial Science, University of Waterloo, ON, Canada

Lysy, M., Department of Statistics and Actuarial Science, University of Waterloo, ON, Canada

Béliveau, A., Department of Statistics and Actuarial Science, University of Waterloo, ON, Canada

Genetic mark-recapture (GMR) is a statistical technique used in estimating population size in ecology. By combining estimated relative abundance for all the species of interest from genetic data with counts for some species, GMR provides information about the total population size and the contributions of each species. In this study, we propose a novel Bayesian GMR framework which presents several advantages over current frequentist techniques. First, the Bayesian framework lends itself nicely to incorporating additional sources of data into a single model. Second, the uncertainty in the estimates is easily obtained from the posterior samples, without the need for additional approximations or bootstrapping. In addition, the Bayesian framework can explicitly incorporate the sampling error in the genetic sample within the model specification (the relative proportions in the genetic sample might differ from those in the population). The effectiveness of the new method is investigated via simulation studies and used to estimate the abundance of Sockeye Salmon in the Taku river.

86 Integrated population modeling of an unmarked population to examine environ- mental drivers of demography for an indicator species

Warlick, A. J., School of Aquatic and Fishery Sciences, University of Washington, WA, USA

Wood, F., Guillemot Research Group, Whidbey Audubon Society, WA, USA

Hostetter, N. J., U.S. Geological Survey, North Carolina Cooperative Fish and Wildlife Research Unit, Department of Applied Ecology, North Carolina State University, NC, USA

Converse, S. J., School of Aquatic & Fishery Sciences, School of Environmental & Forest Sciences; U.S. Geological Survey, Washington Cooperative Fish and Wildlife Research Unit, University of Washington, WA, USA

Integrated population models (IPMs) have emerged as an increasingly popular tool for the estimation of vital rates and abundance to inform conservation and management of avian populations. By combining multiple data sources, IPMs can result in improved precision and estimate demographic parameters that are otherwise unidentifiable. These benefits have inspired numerous applications and extensions, but IPMs typically assume that some portion of the population is marked or uniquely identifiable. Here we present a unique IPM for estimating abundance and demographic rates in an unmarked population of Pigeon Guillemots (Cepphus columba, a cryptic nesting seabird of conservation concern in Puget Sound, Washington, USA. Our model is constructed using count survey data and observations of burrow provisioning by attending adult birds and therefore does not require marked individuals. This framework provides the opportunity to examine the strengths and limitations of IPMs when mark-recapture data are not available to explicitly inform the estimation of survival. Additionally, reproductive success within the IPM is estimated using a novel multi-event mark-recapture model that accounts for the uncertainty when nest age, state, and fate are all unknown. This subcomponent model for estimating reproductive output is applicable to other cryptic nesting species whose nests are partially or fully unobservable, providing new insights into a key demographic rate that often drives avian population dynamics. As one of the most abundant nesting seabirds in Puget Sound, Pigeon Guillemots have been deemed a local indicator species. However, despite this designation, local abundance estimates are outdated, demographic rates are unknown, and the effects of environmental variability have not been investigated. Our approach will allow us to better understand the ecology and status of this species in the region by integrating readily available and less invasive community science datasets, highlighting the importance of grassroots wildlife monitoring programs for informing conservation and management decisions.

87 A semi-Markov survival model for ring-recovery data incorporating environmental covariates

Worthington, H., University of St Andrews, UK

King, R., University of Edinburgh, UK

Survival probabilities often include an age-dependence modelling the biological understanding of the species. For example, first-year survival can be relatively low with the probability of survival peaking once maturity is reached. Incorporating full age-dependence, permitting different survival probabilities for each year of life, produces a significant increase in the number of parameters required particularly if time-dependence is also considered. Instead of considering this problem from the perspective of survival from one year to the next, we consider instead the age-structure of the population and the distribution for the age of death. Specifying a parametric model for the age of death, we propose a semi-Markov model for the survival process. We include environmental covariates, such as the number of frost days indicating winter harshness, to allow for additional variability in the survival probabilities whilst retaining interpretability. This approach allows for the incorporation and exploration of a sensible age-structure for the species under study and associated efficient algorithms are available to fit the model to data. We explore this approach through simulation and consider application to real ring-recovery data sets on several bird populations.

88 Thanks to our sponsor and official host of the EURING 2021 meeting, Université Laval.

89