Balancing transferability and complexity of species distribution models for rare species conservation

Helmstetter, Nolan A.; Conway, Courtney J.; Stevens, Bryan S.; Goldberg, Amanda R.

doi:10.1111/ddi.13174

Cited by 39 publications

(26 citation statements)

References 50 publications

(125 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Full details on the parameter settings for each algorithm are outlined in the Supplementary Material. We limited model complexity in all four algorithms because this is necessary when the primary goal is to use SDMs for predictive transferability in space (Helmstetter et al . 2020).…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Priority conservation areas and a global population estimate for the Critically Endangered Philippine Eagle derived from modelled range metrics using remote sensing habitat characteristics

Sutton

Ibañez

Salvador³

et al. 2021

Preprint

View full text Add to dashboard Cite

Many range-restricted taxa are currently experiencing severe population declines yet lack fundamental biological information regarding distribution and population size. Establishing baseline estimates for both these key biological parameters is however critical for directing long-term monitoring and conservation planning for at-risk range-restricted species. The International Union for the Conservation of Nature (IUCN) Red List uses three spatial range metrics that define species distributions and inform extinction risk assessments: extent of occurrence (EOO), area of occupancy (AOO) and area of habitat (AOH). However, calculating all three metrics using standard IUCN approaches relies on a geographically representative sample of locations, which for rare species is often spatially biased. Here, we apply model-based interpolation using an ensemble Species Distribution Model (SDM), correlating occurrences with remote-sensing derived environmental covariates, to calculate IUCN range metrics and a global population estimate for the Critically Endangered Philippine Eagle (Pithecophaga jefferyi). Our ensemble-averaged SDM had high predictive accuracy and was able to identify key areas of Philippine Eagle habitat across the species global range. We estimated an AOH = 49,426 km2 and from this metric calculated a maximum EOO = 609,697 km2 and a minimum EOO = 273,794 km2, with an AOO = 54,695 occupied cells. Based on inferred habitat from the AOH metric and territorial habitat area from home range estimates, we provide an updated global population estimate of 677 breeding pairs (range: 549-772 pairs), or 1354 mature individuals, across the entire Philippine Eagle range. We demonstrate that even when occurrence sampling is geographically biased, robust habitat models can be built which enable quantification of IUCN range metrics and a baseline population size estimate. In the absence of adequate location data for many rare and threatened taxa, our method is a promising spatial modelling tool with widespread applications, in particular for island endemics facing high extinction risk.

show abstract

Section: Methodsmentioning

confidence: 99%

“…2017). Elastic net logistic regression imposes a regularization penalty on the model coefficients, shrinking towards zero the coefficients of covariates that contribute the least to the model, reducing model complexity (Zou & Hastie 2005; Gastón & García-Viñas 2011; Helmstetter et al . 2020).…”

Section: Supplementary Materialsmentioning

confidence: 99%

Priority conservation areas and a global population estimate for the Critically Endangered Philippine Eagle derived from modelled range metrics using remote sensing habitat characteristics

Sutton

Ibañez

Salvador³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…2017). Elastic net logistic regression imposes a penalty (known as regularization) to the model shrinking the coefficients of variables that contribute the least towards zero (or exactly zero), to reduce model complexity (Gastón & García-Viñas 2011; Helmstetter et al . 2020).…”

Section: Methodsmentioning

confidence: 99%

“…Fithian & Hastie 2013), in the R packages glmnet (Friedman et al 2010) and maxnet (Phillips et al 2017). Elastic net logistic regression imposes a penalty (known as regularization) to the model shrinking the coefficients of variables that contribute the least towards zero (or exactly zero), to reduce model complexity (Gastón & García-Viñas 2011;Helmstetter et al 2020). An elastic net is used to perform automatic Optimal-model selection was based on Akaike's Information Criterion (Akaike 1974) corrected for small sample sizes (AICc; Hurvich & Tsai 1989), to determine the most parsimonious model from two model parameters: regularization multiplier (β) and feature classes (Warren & Seifert 2011).…”

Section: Habitat Suitability Modelmentioning

confidence: 99%

Distribution and habitat use of the Madagascar Peregrine Falcon: first estimates for area of habitat and population size

Sutton

Roland²,

Thorstrom

et al. 2021

Preprint

View full text Add to dashboard Cite

Accurately demarcating species distributions has long been at the core of ecology. Yet our understanding of the factors limiting species range limits is incomplete, especially for tropical species in the Global South. Human-driven threats to the survival of many taxa are increasing, particularly habitat loss and climate change. Identifying distributional range limits of at-risk and data-limited species using Species Distribution Models (SDMs) can thus inform spatial conservation planning to mitigate these threats. The Madagascar Peregrine Falcon (Falco peregrinus radama) is the resident sub-species of the Peregrine Falcon complex distributed across Madagascar, Mayotte, and the Comoros Islands. Currently, there are significant knowledge gaps regarding its distribution, habitat preferences and population size. Here, we use point process regression models and ordination to identify Madagascar Peregrine Falcon environmental range limits and propose a population size estimate based on inferred habitat. From our models, the core range of the Madagascar Peregrine Falcon extends across the central upland plateau of Madagascar with a patchier range across coastal and low-elevation areas. Range-wide habitat use indicated that the Madagascar Peregrine Falcon prefers areas of high elevation and aridity, coupled with high vegetation heterogeneity and > 95 % herbaceous landcover, but generally avoids areas of > 30 % cultivated land and > 10 % mosaic forest. Based on inferred high-class habitat, we estimate this habitat area could potentially support a population size ranging between 150-300 pairs. Following International Union for Conservation of Nature Red List guidelines, we recommend this sub-species be classed as Vulnerable, due to its small population size. Despite its potentially large range, the Madagascar Peregrine has specialized habitat requirements and would benefit from targeted conservation measures based on spatial models in order to maintain viable populations.

show abstract

“…Ecologically, both species are semi-colonial and patchily distributed, representing classic examples of metapopulation structure whereby dispersal among populations is uncommon and tends to occur in a 'stepping stone' manner (Yensen 1991;Yensen & Sherman 1997;USFWS 2003). NIDGS live in open meadows, grassy scabs and small rocky outcroppings at an elevation of 1100 to 2300 m within coniferous forests of central Idaho (Burak 2011;Goldberg, Conway, Mack, et al 2020), and they persist within only a small fraction of their former range likely due to habitat loss and reduced population connectivity, mostly as a result of forest encroachment (Sherman & Runge 2002;Suronen & Newingham 2013;Yensen & Dyni 2020;Helmstetter et al 2021). SIDGS live in sagebrush steppe and rolling hill slopes at an elevation of 630 to 1400 m in southwestern Idaho, and are currently threatened by urban and agricultural development, as well as the spreading of invasive annual plants (USFWS 2000;Lohr et al 2013).…”

Section: Introductionmentioning

confidence: 99%

The role of neutral and adaptive genomic variation in population diversification and speciation in two ground squirrel species of conservation concern

Barbosa

Andrews

Goldberg

et al. 2021

Preprint

View full text Add to dashboard Cite

Understanding the neutral (demographic) and adaptive processes leading to the differentiation of species and populations is a critical component of evolutionary and conservation biology. In this context, recently diverged taxa represent a unique opportunity to study the process of genetic differentiation. Northern and southern Idaho ground squirrels (Urocitellus brunneus – NIDGS, and U. endemicus - SIDGS, respectively) are a recently diverged pair of sister species that have undergone dramatic declines in the last 50 years and are currently found in metapopulations across restricted spatial areas with distinct environmental pressures. Here we genotyped single-nucleotide polymorphisms (SNPs) from buccal swabs with restriction site-associated DNA sequencing (RADseq). With these data we evaluated neutral genetic structure at both the inter- and intra-specific level, and identified putatively adaptive SNPs using population structure outlier and genotype-environment association (GEA) analyses. At the interspecific level, we found a clear separation between NIDGS and SIDGS, and evidence for adaptive differentiation relating to differences in hibernation. At the intraspecific level, we identified 3 Evolutionarily Significant Units for NIDGS and 2 for SIDGS plus multiple Management and Adaptive Units. Elevation appears to be the main driver of adaptive differentiation in NIDGS, while neutral variation patterns match and extend that identified in previous studies using microsatellite markers. For SIDGS, neutral substructure generally reflected the effect of natural geographic barriers, while adaptive variation reflected differences in land cover and temperature. These results clearly highlight the roles of neutral and adaptive processes for understanding species and population differentiation, which can have important conservation implications in threatened species.

show abstract

Balancing transferability and complexity of species distribution models for rare species conservation

Abstract: This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Cited by 39 publications

References 50 publications

Priority conservation areas and a global population estimate for the Critically Endangered Philippine Eagle derived from modelled range metrics using remote sensing habitat characteristics

Priority conservation areas and a global population estimate for the Critically Endangered Philippine Eagle derived from modelled range metrics using remote sensing habitat characteristics

Distribution and habitat use of the Madagascar Peregrine Falcon: first estimates for area of habitat and population size

The role of neutral and adaptive genomic variation in population diversification and speciation in two ground squirrel species of conservation concern

Contact Info

Product

Resources

About