A Hierarchical View of Ecological Encounters

arXiv:2611.13466 · q-bio.PE, physics.bio-ph · Submitted 2026-11-10 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.

Ines: We're looking at the paper: "A Hierarchical View of Ecological Encounters".

Marcus: Here's the gist of the paper. Encounters between individuals, or between individuals and entities like vehicles, are central to understanding a host of topics in ecology and related fields Here.

Ines: The authors propose an encounters–interactions–outcomes hierarchical framework implemented as a hierarchical Bayesian model to organize core concepts across diverse ecological scenarios Abstract.

Marcus: This framework allows for tight connections with multi-level ecological data relevant to estimating encounter rates and their downstream impacts on interactions and outcomes Abstract.

Ines: Conceptual Foundation of the Framework The authors conceptualize ecological encounters and their consequences as an encounters–interactions–outcomes hierarchy that can integrate movement-derived empirical data Introduction.

Title and authors: Ines: So we’re looking at this paper today called A Hierarchical View of Ecological Encounters. It’s got a pretty big ambition, trying to connect the individual level stuff with the population level consequences.

Marcus: Yeah, it’s about building a framework that organizes those connections into something manageable, moving from encounters to interactions and finally to outcomes. It tries to take all this messy data—movement data, interaction counts—and tie it together statistically using a hierarchical Bayesian model.

Yuki: What I find interesting is how they define the core concepts. They aren't just looking at random meetings; they’re defining an encounter as a discrete event where two things are temporarily co-located in space and time, and an interaction as that co-location actually causing a change in one of the parties involved.

Ines: Right, so we have encounters, then interactions that follow those encounters, and then outcomes that result from those interactions. That structure is what makes the model work for linking movement data to demographic changes.

Marcus: Exactly. And they’re tackling some real headaches in quantifying these things right now, like how different people use different terms for the same thing or how traditional models just give you relative intensities instead of precise encounter rates needed for dynamic modeling.

Yuki: And that leads into the challenges they identify, like the difficulty in scaling up from individual movement to population-level consequences and fitting those traditional mass action principles when we really need things like non-equilibrium descriptions.

Ines: The authors propose a hierarchical Bayesian model structured as a directed acyclic graph where encounter statistics, interaction statistics, and outcome statistics are all coupled together in this way.

Marcus: They’re setting up two levels of hierarchy here: one level for interactions based on encounters and another level for outcomes based on those interactions. It’s a lot to handle statistically when you're dealing with empirical data from movement tracking.

Yuki: I wonder how well this structure actually holds up when we have imperfect bio-logging data, which is what they’re aiming to use here, since the observation process itself isn't perfect.

Ines: That brings us to their suggested improvements for the model. They propose using an inhomogeneous Poisson point process for encounter detection and a binomial-mixture model for the observation process to estimate that true but unobservable number of encounters at a site, which is called rhoE.

Marcus: That sounds like they’re trying to bridge the gap between what we actually see in our surveys and what the underlying process is doing, quantifying that detection rate rhoE.

Title and authors: Yuki: And then they want interaction models that assign probabilities to different interaction types given an encounter, allowing for covariates like position or time of day to influence those outcome probabilities.

Ines: That lets us test if external factors like weather or prey availability can actually change the likelihood of different interactions arising from an encounter. It’s a big step toward making the model predictive rather than just descriptive.

Marcus: And they also build in outcome set models that allow for the possibility that different interactions can sometimes lead to the same outcomes, which helps handle those complex consequences where outcomes might happen either immediately or after some time delay.

Yuki: The deterministic mapping matrices, like Theta, are used to specify the relationship between encounters, interactions, and outcomes by defining which outcome sets arise from specific interactions. That’s a neat way to ensure logical dependencies among the outcome elements.

Ines: It’s about making sure that when you map an interaction to an outcome set using that matrix, the model respects those rules of consequence.

Marcus: So, it’s not just counting things anymore; it’s about modeling the flow and transformation between these three levels of activity using this coupled statistical structure.

Yuki: When we look at scaling up these individual-based stochastic models into deterministic descriptions, they suggest using systems of advection diffusion equations to model the time-evolving probability density of each organism's location.

Ines: That approach can be computationally heavy, but it offers a way to model movement terms that depend on interactions with other organisms in a more continuous setting.

Marcus: Another scaling method they suggest is treating the spatial distribution as a point pattern and using coupled differential equations to model mean density and spatial correlations, which replaces the mass action law with an explicit description of those spatial correlations.

Yuki: I have to point out their limitation here, though, that these scaling approaches assume that encounters are independent and can’t capture encounter-mediated behavioral changes.

Ines: That’s a crucial caveat; they admit the model doesn't fully capture how interactions themselves drive movement patterns or other behaviors.

Marcus: So, even with this sophisticated hierarchical Bayesian framework, the researchers flag that we still need to figure out those non-independent movements and behavioral feedback loops.

Yuki: And looking forward, a major open question they highlight is inferring missing elements when you only have some data available, like estimating movement paths from occurrence records.

Ines: That’s a practical hurdle; if you don't have the full path data for every single observation, how do you fill in those gaps statistically?

Title and authors: Marcus: Another thing they want to predict is whether this framework can move beyond just assessing when and where encounters happened to actually predicting future encounters and outcomes.

Yuki: And they also question determining the best contexts for reliably assessing interaction durations, especially in scenarios like disease transmission where that timing matters a lot.

Ines: So, the main implication is that we have a more robust statistical tool to estimate rates at all three levels of this hierarchy simultaneously, which is better than looking at any single level in isolation.

Marcus: It gives us a way to link the fine-scale movement observations to broader population dynamics by modeling those intermediate interaction steps explicitly.

Yuki: It’s a powerful tool for linking spatial ecology with demographic change, but we need to be careful about what the data actually supports when inferring those higher-level consequences.

Ines: To wrap up, the paper A Hierarchical View of Ecological Encounters lays out a way to integrate encounter rates, interaction probabilities, and outcome sets into a single statistical structure.

Marcus: It uses an inhomogeneous Poisson point process and a binomial-mixture model to handle imperfect observation data when estimating the true number of encounters at a site.

Yuki: The paper suggests that by modeling the relationship between encounters, interactions, and outcomes through these coupled statistics, we can better understand the consequences of these events on populations.

Ines: It’s a framework for connecting individual-level movement biology with population-level demographic change in a statistically rigorous way.

Marcus: The authors show how to use deterministic mapping matrices to define which outcome sets result from specific interactions, which is helpful for understanding the logical flow of consequences.

Yuki: We need to remember that while the structure is strong, they admit that scaling up requires assumptions about encounter independence that might not hold in reality.

Ines: That’s a fair point; the paper stops short of fully modeling those complex non-independent movement patterns and behavioral changes yet.

Marcus: So, while this framework offers a path to estimating rates across all three levels, the next step seems to be testing how well it performs when we incorporate those more complex dynamic processes.

Yuki: It opens up avenues for multidisciplinary work, linking stochastic processes with movement ecology to study things like roadkill rates or disease transmission dynamics.

Ines: That’s the real utility there—it provides a common language for researchers working on very different ecological problems.

Marcus: So, A Hierarchical View of Ecological Encounters is a solid statistical structure that helps us organize the complexity of encounters into three interconnected tiers, even if we still have some big questions about dynamic behavior and data gaps.

The paper's summary: Ines: So, A Hierarchical View of Ecological Encounters takes all this messy data—movement tracking, interaction counts—and organizes it into a clear structure: encounters lead to interactions, and those interactions lead to outcomes.

Marcus: Right, so we’re talking about moving beyond just counting how often things meet in the wild and trying to build a statistical machine that predicts what happens next based on those meetings.

Ines: Basically, they’ve put this whole process into a hierarchical Bayesian model represented as a directed acyclic graph where all three levels—encounters, interactions, and outcomes—are statistically coupled together.

Marcus: It’s about taking the raw numbers from those different tiers and using them to estimate the system-level dynamics that happen across the entire ecosystem.

Ines: The authors are really trying to show how this structure allows you to determine the likelihood of different outcomes by aggregating information across all three levels at once.

Marcus: That’s a big deal because it tries to solve that problem where you only know some of the data and need a way to infer what the rest is doing, like figuring out what's actually happening at a site when you only have survey counts.

Ines: They propose using an inhomogeneous Poisson point process for detecting those encounters, and then a binomial mixture model for observing them imperfectly, which lets them estimate that true but unobservable number of encounters at a specific location.

Marcus: So they’re trying to quantify the detection rate, which they call rhoE, so you know how good your survey data is telling you about the actual encounter frequency in the field.

Ines: And then they build interaction models that assign probabilities to different types of interactions given that an encounter happened, allowing for things like position or time of day to influence those results.

Marcus: That means if we have an encounter, we can ask what’s likely to happen next based on where the animals are and when it is, which helps us predict how those interaction probabilities change with external factors like weather or food availability.

Ines: They also model the outcomes by allowing different interactions to sometimes lead to the same final result, which is important because consequences don't always happen immediately; they can have a time delay.

Marcus: And they use deterministic mapping matrices, like Theta, to precisely define which outcome sets arise from specific interactions, making sure the logical flow between encounters and final results is respected in the model.

Ines: So what this means for us is that we get a way to link individual-level movement biology directly to population-level demographic change in a statistically rigorous way across all these tiers.

Marcus: It’s a powerful tool for connecting those fine-scale movement observations to broader population dynamics, but the authors flag that scaling up these individual-based models requires assumptions about encounter independence that might not hold true in reality.

Ines: That’s a crucial caveat; they admit the model doesn't fully capture those complex non-independent movement patterns and behavioral changes yet, which is where the real open questions lie.

Marcus: So while this framework gives us a better way to estimate rates across all three levels, the next step seems to be testing how well it performs when we incorporate those more complex dynamic processes that they admit are missing.

The paper's improvements: Ines: So, A Hierarchical View of Ecological Encounters suggests several ways to improve this framework for real use.

Marcus: I mean, they aren't just stopping there with the initial model setup; they’re suggesting specific statistical tweaks to make it work better with real-world data.

Ines: They are proposing using an inhomogeneous Poisson point process for encounter detection and a binomial mixture model for observing those encounters, which helps them estimate that true but unobservable number of encounters at a site, as we talked about before.

Marcus: That’s good because it directly addresses the imperfect bio-logging data issue, giving us a better handle on the survey counts by quantifying that detection rate.

Ines: Plus, they suggest building interaction models that allow for covariates like position or time of day to influence those outcome probabilities, so we can test how external factors actually change those probabilities.

Marcus: I agree; being able to see how prey availability or weather might shift the likelihood of a certain interaction is really useful for making predictions about population behavior.

Ines: And they also look at outcome set models that let different interactions sometimes lead to the same final result, which helps handle those scenarios where consequences happen after some time delay.

Marcus: That deals with the complexity of consequences, where things aren't always direct; it’s about modeling those temporal relationships more accurately than just looking at immediate results.

Ines: The authors also suggest using deterministic mapping matrices, like Theta, to precisely define the relationship between encounters and outcomes by specifying which outcome sets arise from which interactions.

Marcus: That gives us a clear rule for how to map those three levels together logically so the model stays consistent across all its components.

Ines: So what this means is that they’re giving us a more granular set of tools—better observation models, covariate-driven interaction probabilities, and clearer rules for outcome mapping—to handle real-world ecological complexity.

Marcus: It shifts the focus from just building a static structure to creating an adaptive system that can incorporate richer information about the environment influencing every step.

Ines: This moves us closer to being able to truly predict future encounters and outcomes, rather than just describing what has already happened.

Marcus: And that brings us back to those open questions we mentioned earlier, like inferring missing elements when you only have some data available, which these improvements should help tackle better.

Ines: So the goal is a model that doesn't just fit the data, but one that can reliably estimate rates on all three levels of this hierarchy simultaneously.

Conclusion: Ines: So, to wrap up, A Hierarchical View of Ecological Encounters gives us a statistical way to connect individual movement data all the way up to population consequences through encounters and interactions.

Marcus: It’s basically a framework for building a model where you can estimate rates on all three levels—encounters, interactions, and outcomes—all at the same time.

Ines: The main implication is that we have a more robust statistical tool for linking fine-scale movement biology to population dynamics by modeling those intermediate interaction steps explicitly.

Marcus: It shows how you can use tools like an inhomogeneous Poisson point process and a binomial mixture model to handle imperfect observation data when estimating the true number of encounters at a site.

Yuki: From a population perspective, this structure is useful because it lets us see how individual encounters translate into broader demographic changes across the whole species history.

Ines: It’s about connecting that movement ecology directly to population genetics, showing how spatial events influence the wider gene pool dynamics.

Marcus: We have to remember though that they admit scaling up these models requires assumptions about encounter independence that might not hold true in reality, so we can't take those results as completely final without more testing.

Yuki: And I’d add that while this framework is strong for linking movement and demographics, the authors still have some open questions about inferring missing elements when you only have partial data.

Ines: Exactly; they need to figure out how to fill in those gaps when we don't have the full path data for every single observation.

Marcus: It’s a solid structure for organizing this kind of complex ecological problem, but it definitely opens up the door for testing those more complicated dynamic processes moving forward.

Yuki: The work on A Hierarchical View of Ecological Encounters gives us a better way to think about how spatial events shape evolution and population trends across time.

William F. Fagan, Mark A. Lewis, Justin M. Calabrese, Joanna Mills Flemming, Marron McConnell, Christian Che-Castaldo, Rafael Menezes, Eduardo H. Colombo, Sarah C., Natasha Ellison-Neary, Claire D. Fagan, Luca Giuggioli, Manuela Gonz´alez-Su´arez, Andrew Hein, Abigail King, Michael Noonan, Anudeep Surendran, Brian R. Tsuru, Rebecca Tyson, Ruth E. Baker, Benjamin Garcia de Figueiredo, Christen H. Fleming, Eliezer Gurarie, Kezia Manlove, Jorge F.S. Menezes, Inˆes Silva, Gandhimohan M. Viswanathan

Department of Biology, University of Maryland, College Park, MD, USA · Departments of Mathematics and Statistics and of Biology, University of Victoria, Victoria, BC, Canada · Center for Advanced Systems Understanding (CASUS), Helmholtz-Zentrum Dresden-Rossendorf · Department of Mathematics and Statistics, Dalhousie University · U.S. Geological Survey Wisconsin Cooperative Wildlife Research Unit Department of Forest and Wildlife Ecology University of Wisconsin-Madison · ICTP South American Institute for Fundamental Research and Instituto de Física Teórica Universidade Estadual Paulista Sao Paulo Brazil · Department of Migration, Max Planck Institute of Animal Behavior Radolfzell Germany Department of Biology, University of Konstanz Konstanz Germany · Department of Wildlife Fisheries and Aquaculture Mississippi State University Mississippi State MS USA · School of Engineering Mathematics and Technology University of Bristol Bristol UK · Ecology and Evolutionary Biology School of Biological Sciences University of Reading Reading UK · Department of Computational Biology Cornell University Ithaca NY USA · Okanagan Institute for Biodiversity Resilience, and Ecosystem Services The University of British Columbia Okanagan Kelowna BC Canada · Department of Computer Science Mathematics Physics, and Statistics The University of British Columbia Okanagan Kelowna BC Canada · Mathematical Institute University of Oxford Oxford UK · Lewis-Sigler Institute for Integrative Genomics Princeton University Princeton NJ USA · Department of Biology University of Central Florida Orlando FL USA · Department of Environmental Biology SUNY College of Environmental Science and Forestry Syracuse NY USA · Department of Wildland Resources and Ecology Center Utah State University Logan UT USA · Feline Research Group Mamirau´a Institute for Sustainable Development Tef´e AM Brazil Department of Ecology Institute of Biosciences University de São Paulo Sao Paulo SP Brazil · Department of Physics Universidade Federal do Rio Grande do Norte Natal RN Brazil

q-bio.PE, physics.bio-ph

Submitted: 2026-11-10

Updated: 2026-11-10

Comments: 39 pages, 4 figures

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 87/100

The gist: Encounters between individuals, or between individuals and entities like vehicles, are central to understanding a host of topics in ecology and related fields Here.

Key concepts

Encounter
A discrete event where two entities are temporarily co-located in a specific space during a certain time interval. This is the foundational unit used to start the hierarchical analysis.
Interaction
A change that occurs in one or both parties involved because of an encounter. This links the initial encounter to subsequent effects, such as movement changes or behavioral shifts.
Outcome
The consequence that follows an interaction. Outcomes are the final results of an interaction, which can be immediate or delayed, and are what the model aims to determine likelihoods for.

Summary

Encounters between individuals, or between individuals and entities like vehicles, are central to understanding a host of topics in ecology and related fields Here. p. 1 The authors propose an encounters–interactions–outcomes hierarchical framework implemented as a hierarchical Bayesian model to organize core concepts across diverse ecological scenarios Abstract. p. 1 This framework allows for tight connections with multi-level ecological data relevant to estimating encounter rates and their downstream impacts on interactions and outcomes Abstract. p. 1

Conceptual Foundation of the Framework

The authors conceptualize ecological encounters and their consequences as an encounters–interactions–outcomes hierarchy that can integrate movement-derived empirical data Introduction. This framework builds from the dyadic framework introduced in Gurarie and Ovaskainen (2013) where encounters are two-step processes conditional on spatial co-location or proximity Page 5. p. 3 They define an encounter as a discrete event where an individual animal and another entity are temporarily co-located within a particular region of space and during a particular interval of time Page 5. p. 4 An interaction occurs when the co-location underlying an encounter leads to a change in one or more of the parties involved Page 5. p. 4 Finally, an outcome is the consequence of an interaction that has occurred following an encounter Page 5. p. 4

Challenges in Quantifying Encounters

The study identifies four critical challenges associated with studying encounters Page 5. These challenges include differences in terminology hindering tool sharing, difficulties in collecting direct encounter data, methodological limitations in estimating encounters from movement data, and scaling up to investigate population-level consequences Page 5. p. 5 Furthermore, traditional models often assume encounters follow mass action principles which yield only relative encounter intensities rather than precise rates needed for dynamic models Page 4. p. 3 Addressing this mismatch requires new analytical and statistical methods potentially including non-equilibrium descriptions Page 4. p. 3

Modeling the Hierarchical Structure

The authors propose a hierarchical Bayesian model represented as a directed acyclic graph (DAG) where encounter statistics, interaction statistics, and outcome statistics are coupled Page 32. The model consists of two levels of hierarchy: one for interactions based on encounters and another for outcomes based on interactions Page 12. p. 9 The encounter component models the stochastic process, using a simple (homogeneous) Poisson point process model where C E is the true but unobservable number of encounters at a site and λ is the mean number of encounters Page 13. p. 5 The interaction model associates probabilities with different possible interaction types given that an encounter occurs Page 14. p. 4

Implementation and Future Directions

The implementation guide suggests defining parameters such as the encounter region for a focal individual based on its sensory radius or a specific stretch of road Page 13. The model allows for the estimation of parameters at each tier, such as λ and ρE from encounter counts, p and ρI from interaction counts, and Π from outcome counts Page 18. p. 1 Future open questions involve inferring missing elements when only some data are available, predicting future encounters and interactions, determining the best contexts for assessing interaction durations, and evaluating encounter dynamics for non-independent movements Page 19. The framework offers options for multidisciplinary work linking stochastic processes and movement ecology to understand disease transmission or roadkill rates Page 20.

Glossary of Key Terms

The paper defines key terms such as Encounter, Interaction, Outcome, and First Passage Time in Table 1 Page 32. These definitions clarify the relationship between encounters, interactions, and outcomes within the EIO framework Page 5. p. 4 Terms like Encounter rate describe the average frequency with which an individual animal and another entity are co-located within the encounter region and time window Page 32. p. 4 The notation in Table 2 specifies symbols used in the hierarchical encounters–interactions–outcomes model Page 32. p. 3

Specific Model Application

The example for the coyote involves defining K=3 mutually exclusive interactions and L=4 ecologically relevant outcomes Page 18. The deterministic mapping from outcome sets to outcomes is defined by the matrix Θ Page 37. p. 4 A specific probability, such as the probability of displacement, is recovered via equation (20) which aggregates displacement through altercation and territorial signaling Page 18. The observation models specify how latent counts are observed imperfectly through site-level surveys Page 37. p. 7 The final probability of displacement is recovered via equation (20) as ϕ2 = ˜π1 + ˜π2 = (π11 + π21) p1 + π22 p2

Scaling and System Dynamics

Scaling individual-based stochastic models in terms of deterministic descriptions of their statistical properties can reduce computational time dramatically Page 10. One scaling approach uses a system of advection–diffusion equations to model the time-evolving probability density of each organism’s location, with movement terms that can depend on interactions with other organisms Page 10. p. 9 A second approach treats the spatial distribution as a point pattern and models mean density and spatial correlations using coupled differential equations Page 10. p. 9 These methods can refine classical population-level models by replacing the law of mass action with an explicit description of spatial correlations Page 7. p. 9 However, these approaches assume that encounters are independent and cannot capture encounter-mediated behavioral changes Page 8. p. 3

Open Questions for Research

Major open questions involve inferring missing elements when only some data are available, such as estimating movement paths from occurrence records Page 18. Another question is whether component data can move beyond assessing when, where, and why encounters have occurred into predicting future encounters and outcomes Page 20. p. 6 Additionally, researchers need to determine the best contexts for reliable assessments of interaction durations in disease transmission scenarios Page 37. The paper concludes by suggesting initial case studies should leverage different data sources to estimate rates on all three levels of the hierarchy Page 20.

Acknowledgments and References

The authors acknowledge various institutions and provide an extensive list of references supporting the work Page 28. The references cover topics ranging from mathematical concepts like 'first passage time' to empirical studies on roadkill and marine mammal collisions Page 27. The work is supported by various academic departments and research institutes Page 5. p. 0 The paper is identified as arXiv:2611. p. 013466v1 [q-bio. p. 0 p. 0PE] 10 Nov 2026

--- Page 30 ---

The paper is identified as arXiv:2611.13466v1 [q-bio.PE] 10 Nov 2026

--- Page 31 ---

Table 1: Glossary of key terms. Terms defined here are italicized when they appear in the definitions of other terms

--- Page 32 ---

Table 2: Notation used in the hierarchical encounters–interactions–outcomes model

--- Page 36 ---

Figure 4: A generalized hierarchical Bayesian model for the encounters–interactions–outcomes framework represented as a directed acyclic graph (DAG)

--- Page 18 ---

The end result is the ability to determine the likelihood of different outcomes

--- Page 20 ---

Looking Forward

--- Page 26 ---

Andersen, A. and D¨olger, J. (2019). Planktonic encounter rates with non-spherical encounter zones. p. 1 Journal of The Royal Society Interface, 16(156):20190398

--- Page 37 ---

Box 1. Disease transmission encounters and interactions Disease transmission processes are inherently three-component processes involving an infective donor, a susceptible recipient, and the infectious agent itself

--- Page 38 ---

Box 2. Encounters, interactions, and outcomes involving two organisms Different categories of encounter-mediated interactions between two organisms have distinct implications for modeling and statistical analyses

--- Page 39 ---

Box 3. Encounters, interactions, and outcomes involving animals and infrastructure Human infrastructure is present in nearly all parts of the world and continues to expand, impacting biodiversity

--- Page 17 ---

A Brief Implementation Guide This section outlines key practical steps necessary to implement the HBM model in a realistic scenario

--- Page 21 ---

Calsbeek, R., Alonzo, S. H., Zamudio, K., and Sinervo, B. (2002). Sexual selection and alternative mating behaviours generate demographic stochasticity in small populations

--- Page 24 ---

Gurarie, E. and Ovaskainen, O. p. 3 (2013). Towards a general formalization of encounter rates in ecology

--- Page 28 ---

Moorcroft, P. R. and Lewis, M. p. 5 A. (2006). Mechanistic home range analysis

--- Page 39 ---

de Jonge et al., 2022. The impacts of linear infrastructure on terrestrial vertebrate populations: A trait-based approach

--- Page 11 ---

Ellison, N., Hatchwell, B. J., Biddiscombe, S. J., Napper, C. J., and Potts, J. p. 6 R., 2020. Mechanistic home range analysis reveals drivers of space use patterns for a non-territorial passerine

--- Page 19 ---

Nisi et al., 2024. Ship collision risk threatens whales across the world’s oceans

--- Page 25 ---

Kays, R., Davidson, S. C., Berger, M., Bohrer, G., Fiedler, W., Flack, A., Hirt, J., Hahn, C.

--- Page 24 ---

Garcia de Figueiredo et al. 2024. Ergodicity shapes inference in biological reactions driven by a latent trajectory

--- Page 26 ---

Menezes et al., 2025. The range-resident logistic model: A new framework to formalise the population-dynamics consequences of range residency

--- Page 28 ---

Grilo et al. 2021. Conservation threats from roadkill in the global road network

--- Page 39 ---

Arantes et al. 2022. Functional responses of fisheries to hydropower dams in the Amazonian floodplain of the Madeira River

--- Page 38 ---

Ben´ıtez-L´opez et al. 2010. The impacts of roads and other infrastructure on mammal and bird populations: A meta-analysis

--- Page 29 ---

Strandburg-Peshkin et al. 2015. Shared decision-making drives collective movement in wild baboons

--- Page 30 ---

Moorcroft and Lewis, M. A., 2006. Mechanistic home range analysis

--- Page 27 ---

Singh, J. A., Donadio, E., Bidder, O. R., Pauli, J. N., Sheriff, M. J., Perrig, P. L., and Middleton, A. D., 2020

--- Page 31 ---

Long and Nelson, T. A., 2013

--- Page 28 ---

Smith, J. E. and Pinter-Wollman, N. p. 3, 2021

--- Page 29 ---

Kays et al., 2024. The Movebank system for studying global animal movement and demography

--- Page 30 ---

Kenkre, V. M. N. and Giuggioli, L. p. 3, 2021

--- Page 25 ---

Jorge and Martinez-Garcia, R., 2024

--- Page 36 ---

Fagan et al., 2024. Intraspecific encounters can lead to reduced range overlap

--- Page 19 ---

Ruprecht et al. 2021. Variable strategies to solve risk–reward tradeoffs in carnivore communities

--- Page 38 ---

Smith et al., 2020. Where and when to hunt? Decomposing predation success of an ambush carnivore

--- Page 25 ---

Jorge and Martinez-Garcia, R.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Develop a hierarchical Bayesian model that integrates encounter rates, interaction probabilities, and outcome sets to estimate system-level dynamics from imperfect bio-logging data. This improved AI system can determine the likelihood of different outcomes by aggregating information across three levels: encounters, interactions, and outcomes.

  2. Implement models that use an inhomogeneous Poisson point process (iPPP) for encounter detection where the observation process is a binomial-mixture model to estimate an encounter detection rate rhoE. This allows the system to quantify the true but unobservable number of encounters at a site from observed survey data.

  3. Create interaction models that associate probabilities with different interaction types given an encounter, allowing for covariates such as positions, environmental factors, and time of day to influence outcomes. This enables prediction of how probabilities of different interactions arising from an encounter are affected by external variables like prey availability or weather conditions.

  4. Build outcome set models that allow for the possibility that different interactions can sometimes lead to the same outcomes, facilitating the handling of complex consequences where outcomes can occur either as an immediate consequence of the interaction or with some time delay.

  5. Use deterministic mapping matrices, such as Theta, to specify the relationship between encounters, interactions, and outcomes by defining which outcome sets arise from specific interactions. This ensures that the model respects logical interdependencies among outcome elements and allows for multiple outcome sets to share a single consequence.

Sources

Related papers