Investigating mesh‐based approximation methods for the normalization constant in the log Gaussian Cox process likelihood

Jullum, Martin

doi:10.1002/sta4.285

Cited by 4 publications

(6 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The thinning parameters,

p_{0}

and

α

, were also estimated with low bias (−0.4% and −1.2% relative bias, respectively). Hyperparameters of the spatial random effect,

τ

and

κ

, had larger biases across simulations (−37.0% for

τ

and 14.8% for

κ

), which may be a result of the generic spatial mesh used for analyses (Dambly et al., 2023; Jullum, 2020).…”

Section: Resultsmentioning

confidence: 99%

“…We used a non-customized mesh across simulations, which likely contributed to biases. For specific applications, such as our monarch case study (Supporting Information S3), tuning the spatial mesh to the system at hand can minimize such biases (Dambly et al, 2023;Jullum, 2020;Krainski et al, 2018). Krainski et al (2018) provides in-depth instruction on how to build triangulated meshes using the R-INLA package while Dambly et al (2023) documents tradeoffs encountered when specifying a spatial mesh.…”

Section: Discussionmentioning

confidence: 99%

“…for ), which may be a result of the generic spatial mesh used for analyses (Dambly et al, 2023;Jullum, 2020).…”

Section: Simulation Studymentioning

confidence: 99%

See 2 more Smart Citations

Overcoming data gaps using integrated models to estimate migratory species' dynamics during cryptic periods of the annual cycle

Farr,

Zylstra,

Ries

et al. 2024

Methods Ecol Evol

View full text Add to dashboard Cite

Environmental and anthropogenic factors affect the population dynamics of migratory species throughout their annual cycles. However, identifying the spatiotemporal drivers of migratory species' abundances is difficult because of extensive gaps in monitoring data. The collection of unstructured opportunistic data by volunteer (citizen science) networks provides a solution to address data gaps for locations and time periods during which structured, design‐based data are difficult or impossible to collect. To estimate population abundance and distribution at broad spatiotemporal extents, we developed an integrated model that incorporates unstructured data during time periods and spatial locations when structured data are unavailable. We validated our approach through simulations and then applied the framework to the eastern North American migratory population of monarch butterflies during their spring breeding period in eastern Texas. Spring climate conditions have been identified as a key driver of monarch population sizes during subsequent summer and winter periods. However, low monarch densities during the spring combined with very few design‐based surveys in the region have limited the ability to isolate effects of spring weather variables on monarchs. Simulation results confirmed the ability of our integrated model to accurately and precisely estimate abundance indices and the effects of covariates during locations and time periods in which structured sampling are lacking. In our case study, we combined opportunistic monarch observations during the spring migration and breeding period with structured data from the summer Midwestern breeding grounds. Our model revealed a nonstationary relationship between weather conditions and local monarch abundance during the spring, driven by spatially varying vegetation and temperature conditions. Data for widespread and migratory species are often fragmented across multiple monitoring programs, potentially requiring the use of both structured and unstructured data sources to obtain complete geographic coverage. Our integrated model can estimate population abundance at broad spatiotemporal extents despite structured data gaps during the annual cycle by leveraging opportunistic data.

show abstract

“…The thinning parameters,

p_{0}

and

α

, were also estimated with low bias (−0.4% and −1.2% relative bias, respectively). Hyperparameters of the spatial random effect,

τ

and

κ

, had larger biases across simulations (−37.0% for

τ

and 14.8% for

κ

), which may be a result of the generic spatial mesh used for analyses (Dambly et al., 2023; Jullum, 2020).…”

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Overcoming data gaps using integrated models to estimate migratory species' dynamics during cryptic periods of the annual cycle

Farr,

Zylstra,

Ries

et al. 2024

Methods Ecol Evol

View full text Add to dashboard Cite

show abstract

“…The weights were calculated based on the integrals of the piecewise linear basis functions used to represent the spatially discretized GF model. This generated non‐zero weights for all triangle vertices of triangles that intersect the corresponding availability disk (Figure 1), computed via dense deterministic sampling within each triangle (Jullum, 2020). This step was automatically performed with help of the inlabru package.…”

Section: Methodsmentioning

confidence: 99%

Accounting for unobserved spatial variation in step selection analyses of animal movement via spatial random effects

Arce Guillen,

Lindgren,

Muff

et al. 2023

Methods Ecol Evol

View full text Add to dashboard Cite

Step selection analysis (SSA) is a common framework for understanding animal movement and resource selection using telemetry data. Such data are, however, inherently autocorrelated in space, a complication that could impact SSA‐based inference if left unaddressed. Accounting for spatial correlation is standard statistical practice when analysing spatial data, and its importance is increasingly recognized in ecological models (e.g. species distribution models). Nonetheless, no framework yet exists to account for such correlation when analysing animal movement using SSA. Here, we extend the popular method integrated step selection analysis (iSSA) by including a Gaussian field (GF) in the linear predictor to account for spatial correlation. For this, we use the Bayesian framework R‐INLA and the stochastic partial differential equations (SPDE) technique. We show through a simulation study that our method provides accurate fixed effects estimates, quantifies their uncertainty well and improves the predictions. In addition, we demonstrate the practical utility of our method by applying it to three wolverine (Gulo gulo) tracks. Our method solves the problems of assuming spatially independent residuals in the SSA framework. In addition, it offers new possibilities for making long‐term predictions of habitat usage.

show abstract

“…This generated non-zero weights for all triangle vertices of triangles that intersect the corresponding availability disk (Fig. 1), computed via dense deterministic sampling within each triangle (Jullum 2020). This step was automatically performed with help of the inlabru package.…”

Section: Model Fittingmentioning

confidence: 99%

Accounting for unobserved spatial variation in step selection analyses of animal movement via spatial random effects

Guillen

Lindgren

Muff

et al. 2023

Preprint

View full text Add to dashboard Cite

1. Step selection analysis (SSA) is a common framework for understanding animal movement and resource selection using telemetry data. Such data are, however, inherently autocorrelated in space, a complication that could impact SSA-based inference if left unaddressed. Accounting for spatial correlation is standard statistical practice when analyzing spatial data, and its importance is increasingly recognized in ecological models (e.g., species distribution models). Nonetheless, no framework yet exists to account for such correlation when analyzing animal movement using SSA. 2. Here, we extend the popular method Integrated Step Selection Analysis (iSSA) by including a Gaussian Field (GF) in the linear predictor to account for spatial correlation. For this, we use the Bayesian framework R-INLA and the Stochastic Partial Differential Equations (SPDE) technique. 3. We show through a simulation study that our method provides unbiased fixed effects estimates, quantifies their uncertainty well and improves the predictions. In addition, we demonstrate the practical utility of our method by applying it to three wolverine (Gulo gulo) tracks. 4. Our method solves the problems of assuming spatially independent locations in the SSA framework. In addition, it offers new possibilities for making long-term predictions of habitat usage.

show abstract

Investigating mesh‐based approximation methods for the normalization constant in the log Gaussian Cox process likelihood

Cited by 4 publications

References 21 publications

Overcoming data gaps using integrated models to estimate migratory species' dynamics during cryptic periods of the annual cycle

Overcoming data gaps using integrated models to estimate migratory species' dynamics during cryptic periods of the annual cycle

Accounting for unobserved spatial variation in step selection analyses of animal movement via spatial random effects

Accounting for unobserved spatial variation in step selection analyses of animal movement via spatial random effects

Contact Info

Product

Resources

About