Predictive performance of presence‐only species distribution models: a benchmark study with reproducible code

Ecological Monographs, Journal Year: 2021, Volume and Issue: 92(1)

Published: Oct. 8, 2021

Abstract Species distribution modeling (SDM) is widely used in ecology and conservation. Currently, the most available data for SDM are species presence‐only records (available through digital databases). There have been many studies comparing performance of alternative algorithms data. Among these, a 2006 paper from Elith colleagues has particularly influential field, partly because they several novel methods (at time) on global set that included independent presence–absence model evaluation. Since its publication, some further developed new ones emerged. In this paper, we explore patterns predictive across methods, by reanalyzing same (225 six different regions) using updated knowledge practices. We apply well‐established such as generalized additive models MaxEnt, alongside others received attention more recently, including regularized regressions, point‐process weighted random forests, XGBoost, support vector machines, ensemble framework biomod. All use include background samples (a sample environments landscape) fitting. impacts weights presence points introduce ways evaluating fitted to these data, area under precision‐recall gain curve, focusing rank results. find way matters. The top method was an tuned individual models. contrast, ensembles built biomod with default parameters performed no better than single moderate performing Similarly, second forest parameterized deal (contrasted relatively few records), which substantially outperformed other implementations. that, general, nonparametric techniques capability controlling complexity traditional regression MaxEnt boosted trees still among code working examples provided make study fully reproducible.

Language: Английский

---

A comparison between Ensemble and MaxEnt species distribution modelling approaches for conservation: A case study with Egyptian medicinal plants

Ecological Informatics, Journal Year: 2020, Volume and Issue: 60, P. 101150 - 101150

Published: Sept. 3, 2020

Language: Английский

---

The recent past and promising future for data integration methods to estimate species’ distributions

Methods in Ecology and Evolution, Journal Year: 2019, Volume and Issue: 10(1), P. 22 - 37

Published: Jan. 1, 2019

Abstract With the advance of methods for estimating species distribution models has come an interest in how to best combine datasets improve estimates distributions. This spurred development data integration that simultaneously harness information from multiple while dealing with specific strengths and weaknesses each dataset. We outline general principles have guided review recent developments field. then key areas allow a more framework integrating provide suggestions improving sampling design validation integrated models. Key advances been using point‐process thinking estimators developed different types. Extending this new types will further our inferences, as well relaxing assumptions about parameters are jointly estimated. These along better use regarding effort spatial autocorrelation inferences. Recent form strong foundation implementation Wider adoption can inferences distributions dynamic processes lead distributional shifts.

Language: Английский

---

A practical guide for combining data to model species distributions

Ecology, Journal Year: 2019, Volume and Issue: 100(6)

Published: March 30, 2019

Understanding and accurately modeling species distributions lies at the heart of many problems in ecology, evolution, conservation. Multiple sources data are increasingly available for distributions, such as from citizen science programs, atlases, museums, planned surveys. Yet reliably combining can be challenging because vary considerably their design, gradients covered, potential sampling biases. We review, synthesize, illustrate recent developments multiple distribution modeling. identify five ways which typically combined distributions. These approaches ability to accommodate bias, uncertainty when quantifying environmental relationships models. Many challenges solved through prudent use integrated models: models that simultaneously combine different on locations quantify explaining distribution. these using survey 24 birds coupled with opportunistically collected eBird southeastern United States. This example illustrates some benefits integration, increased precision relationships, greater predictive accuracy, accounting sample bias. it also vastly methodologies amounts data. provide one solution this challenge weighted joint likelihoods. Weighted likelihoods a means emphasize based criteria (e.g., size), we find weighting improves predictions all considered. conclude by providing practical guidance

Language: Английский

---

Integrated species distribution models: combining presence‐background data and site‐occupancy data with imperfect detection

Methods in Ecology and Evolution, Journal Year: 2017, Volume and Issue: 8(4), P. 420 - 430

Published: April 1, 2017

Summary Two main sources of data for species distribution models (SDMs) are site‐occupancy (SO) from planned surveys, and presence‐background (PB) opportunistic surveys other sources. SO give high quality about presences absences the in a particular area. However, due to their cost, they often cover smaller area relative PB data, usually not representative geographic range species. In contrast, is plentiful, covers larger area, but less reliable lack information on absences, characterised by biased sampling. Here we present new approach modelling that integrates these two types. We have used an inhomogeneous Poisson point process as basis constructing integrated SDM fits both simultaneously. It first implementation Integrated SO–PB Model which uses repeated survey occupancy also incorporates detection probability. The Model's performance was evaluated, using simulated compared approaches or alone. found be superior, improving predictions spatial distributions, even when sparse collected limited effective environmental covariates were significantly correlated. Our method demonstrated with real Yellow‐bellied glider ( Petaurus australis ) south‐eastern Australia, predictive again superior. known produce estimates abundance. small sample size datasets results poor out‐of‐sample predictions. combine sources, providing superior abundance either source Unlike conventional SDMs restrictive scale‐dependence predictions, our based model has no such scale‐dependency. may at any spatial‐scale while still maintaining underlying relationship between

Language: Английский

---

The basis function approach for modeling autocorrelation in ecological data

Ecology, Journal Year: 2016, Volume and Issue: 98(3), P. 632 - 646

Published: Dec. 9, 2016

Abstract Analyzing ecological data often requires modeling the autocorrelation created by spatial and temporal processes. Many seemingly disparate statistical methods used to account for can be expressed as regression models that include basis functions. Basis functions also enable ecologists modify a wide range of existing in order autocorrelation, which improve inference predictive accuracy. Furthermore, understanding properties is essential evaluating fit or time‐series models, detecting hidden form collinearity, analyzing large sets. We present important concepts related illustrate several tools techniques use when data.

Language: Английский

---

Species distribution modeling: a statistical review with focus in spatio-temporal issues

Stochastic Environmental Research and Risk Assessment, Journal Year: 2018, Volume and Issue: 32(11), P. 3227 - 3244

Published: April 19, 2018

Language: Английский

---

Point process models for spatio-temporal distance sampling data from a large-scale survey of blue whales

The Annals of Applied Statistics, Journal Year: 2017, Volume and Issue: 11(4)

Published: Dec. 1, 2017

Distance sampling is a widely used method for estimating wildlife population abundance. The fact that conventional distance methods are partly design-based constrains the spatial resolution at which animal density can be estimated using these methods. Estimates usually obtained survey stratum level. For an endangered species such as blue whale, it desirable to estimate and abundance finer scale than stratum. Temporal variation in structure also important. We formulate process generating data thinned point propose model-based inference log-Gaussian Cox process. adopts flexible stochastic partial differential equation (SPDE) approach model not accounted by explanatory variables, integrated nested Laplace approximation (INLA) Bayesian inference. It allows simultaneous fitting of detection models permits prediction arbitrarily fine scale. whale Eastern Tropical Pacific Ocean from thirteen shipboard surveys conducted over 22 years. find higher associated with colder sea surface temperatures space, although there some positive association between mean annual temperature, our estimates consistent no trend across Our analysis indicates substantial spatially structured explained available covariates.

Language: Английский

---

Species-Habitat Associations: Spatial data, predictive models, and ecological insights

Published: Dec. 1, 2020

Ecologists develop species-habitat association (SHA) models to understand where species occur, why they are there and else might be. This knowledge can be used designate protected areas, estimate anthropogenic impacts on living organisms assess risks from invasive or disease spill-over wildlife humans. Here, we describe the state of art in SHA models, looking beyond apparent correlations between positions their local environment. We highlight importance ecological mechanisms, synthesize diverse modelling frameworks motivate development new analytical methods. Above all, aim synthetic, bringing together several apparently disconnected pieces theory, taxonomy, spatiotemporal scales, mathematical statistical technique our field. The first edition this ebook reviews ecology associations, mechanistic interpretation existing empirical shared foundations that help us draw scientific insights field data. It will interest graduate students professionals for an introduction literature SHAs, practitioners seeking analyse data animal movements distributions quantitative ecologists contribute methods addressing limitations current incarnations models.

Language: Английский

---

Forest tree species distribution for Europe 2000–2020: mapping potential and realized distributions using spatiotemporal machine learning

PeerJ, Journal Year: 2022, Volume and Issue: 10, P. e13728 - e13728

Published: July 25, 2022

This article describes a data-driven framework based on spatiotemporal machine learning to produce distribution maps for 16 tree species ( Abies alba Mill., Castanea sativa Corylus avellana L., Fagus sylvatica Olea europaea Picea abies L. H. Karst., Pinus halepensis nigra J. F. Arnold, pinea sylvestris Prunus avium Quercus cerris ilex robur suber and Salix caprea L.) at high spatial resolution (30 m). Tree occurrence data total of three million points was used train different algorithms: random forest, gradient-boosted trees, generalized linear models, k-nearest neighbors, CART an artificial neural network. A stack 305 coarse covariates representing spectral reflectance, biophysical conditions biotic competition as predictors realized distributions, while potential modelled with environmental only. Logloss computing time were select the best algorithms tune ensemble model stacking logistic regressor meta-learner. An trained each species: probability uncertainty produced using window 4 years six per species, distributions only one map produced. Results cross validation show that consistently outperformed or performed good individual in both tasks, models achieving higher predictive performances (TSS = 0.898, R 2 logloss 0.857) than ones average 0.874, 0.839). Ensemble Q. achieved 0.968, 0.952) 0.959, 0.949) distribution, P. 0.731, 0.785, 0.585, 0.670, respectively, distribution) 0.658, 0.686, 0.623, 0.664) worst. Importance predictor variables differed across green band summer Normalized Difference Vegetation Index (NDVI) fall diffuse irradiation precipitation driest quarter (BIO17) being most frequent important distribution. On average, fine-resolution (250 m) +6.5%, +7.5%). The shows how combining continuous consistent Earth Observation series state art can be derive dynamic maps. predictions quantify temporal trends forest degradation composition change.

Language: Английский

---