A Robust Prediction Model for Species Distribution Using Bagging Ensembles with Deep Neural Networks

Species distribution models have been used for various purposes, such as conserving species, discovering potential habitats, and obtaining evolutionary insights by predicting species occurrence. Many statistical and machine-learning-based approaches have been proposed to construct effective species distribution models, but with limited success due to spatial biases in presences and imbalanced presence-absences. We propose a novel species distribution model to address these problems based on bootstrap aggregating (bagging) ensembles of deep neural networks (DNNs). We first generate bootstraps considering presence-absence data on spatial balance to alleviate the bias problem. Then we construct DNNs using environmental data from presence and absence locations, and finally combine these into an ensemble model using three voting methods to improve prediction accuracy. Extensive experiments verified the proposed model’s effectiveness for species in South Korea using crowdsourced observations that have spatial biases. The proposed model achieved more accurate and robust prediction results than the current best practice models.

Download Full-text

Community Confounding In Joint Species Distribution Models

10.21203/rs.3.rs-1111644/v1 ◽

2021 ◽

Author(s):

Justin J. Van Ee ◽

Jacob S. Ivan ◽

Mevin B. Hooten

Keyword(s):

Bark Beetle ◽

Species Distribution ◽

Species Distribution Models ◽

Species Distribution Model ◽

Management Strategies ◽

Single Species ◽

Distribution Model ◽

Subalpine Forest ◽

Community Effects ◽

Distribution Models

Abstract Joint species distribution models have become ubiquitous for studying species-habitat relationships and dependence among species. Accounting for community structure often improves predictive power, but can also alter inference on species-habitat relationships. Modulated species-habitat relationships are indicative of community confounding: The situation in which interspecies dependence and habitat effects compete to explain species distributions. We discuss community confounding in a case study of mammalian responses to the Colorado bark beetle epidemic in the subalpine forest by comparing the inference from independent single species distribution models and a joint species distribution model. We present a method for measuring community confounding and develop a restricted version of our hierarchical model that orthogonalizes the habitat and species random effects. Our results indicate that variables associated with the severity and duration of the bark beetle epidemic suffer from community confounding. This implies that mammalian responses to the bark beetle epidemic are governed by interconnected habitat and community effects. Disentangling habitat and community effects can improve our understanding of the ecological system and possible management strategies. We evaluate restricted regression as a method for alleviating community confounding and distinguish it from other inferential methods for confounded models.

Download Full-text

SDMtoolbox 2.0: the next generation Python-based GIS toolkit for landscape genetic, biogeographic and species distribution model analyses

PeerJ ◽

10.7717/peerj.4095 ◽

2017 ◽

Vol 5 ◽

pp. e4095 ◽

Cited By ~ 93

Author(s):

Jason L. Brown ◽

Joseph R. Bennett ◽

Connor M. French

Keyword(s):

Species Distribution ◽

Species Distribution Model ◽

Species Distribution Modeling ◽

Environmental Data ◽

Distribution Model ◽

Data Types ◽

Distribution Models ◽

Distribution Modeling ◽

Modeling Software ◽

Cost Paths

SDMtoolbox 2.0 is a software package for spatial studies of ecology, evolution, and genetics. The release of SDMtoolbox 2.0 allows researchers to use the most current ArcGIS software and MaxEnt software, and reduces the amount of time that would be spent developing common solutions. The central aim of this software is to automate complicated and repetitive spatial analyses in an intuitive graphical user interface. One core tenant facilitates careful parameterization of species distribution models (SDMs) to maximize each model’s discriminatory ability and minimize overfitting. This includes carefully processing of occurrence data, environmental data, and model parameterization. This program directly interfaces with MaxEnt, one of the most powerful and widely used species distribution modeling software programs, although SDMtoolbox 2.0 is not limited to species distribution modeling or restricted to modeling in MaxEnt. Many of the SDM pre- and post-processing tools have ‘universal’ analogs for use with any modeling software. The current version contains a total of 79 scripts that harness the power of ArcGIS for macroecology, landscape genetics, and evolutionary studies. For example, these tools allow for biodiversity quantification (such as species richness or corrected weighted endemism), generation of least-cost paths and corridors among shared haplotypes, assessment of the significance of spatial randomizations, and enforcement of dispersal limitations of SDMs projected into future climates—to only name a few functions contained in SDMtoolbox 2.0. Lastly, dozens of generalized tools exists for batch processing and conversion of GIS data types or formats, which are broadly useful to any ArcMap user.

Download Full-text

Species Distribution Models (SDM) – A Strategic Tool for Predicting Suitable Habitats for Conserving the Target Species

Advances in Geospatial Technologies - Handbook of Research on Geographic Information Systems Applications and Advancements ◽

10.4018/978-1-5225-0937-0.ch017 ◽

2017 ◽

pp. 427-440

Author(s):

Balaguru Balakrishnan ◽

Nagamurugan Nandakumar ◽

Soosairaj Sebastin ◽

Khaleel Ahamed Abdul Kareem

Keyword(s):

Species Distribution ◽

Spatial Data ◽

Environmental Variables ◽

Species Distribution Models ◽

Species Distribution Model ◽

Distribution Model ◽

Suitable Habitat ◽

Target Species ◽

Distribution Models ◽

Suitable Habitats

Conservation of the species in their native landscapes required understanding patterns of spatial distribution of species and their ecological connectivity through Species Distribution Models (SDM) by generation and integration of spatial data from different sources using Geographical Information System (GIS) tools. SDM is an ecological/spatial model which combines datasets and maps of occurrence of target species and their geographical and environmental variables by linking various algorithms together, that has been applied to either identify or predict the regions fulfilling the set conditions. This article is focused on comprehensive review of spatial data requirements, statistical algorithms and softwares used to generate the SDMs. This chapter also includes a case study predicting the suitable habitat distribution of Gnetum ula, an endemic and vulnerable plant species using maximum entropy (MaxEnt) species distribution model for species occurrences with inputs from environmental variables such as bioclimate and elevation.

Download Full-text

Not so Normal Normals: Species Distribution Model Results are Sensitive to Choice of Climate Normals and Model Type

Climate ◽

10.3390/cli7030037 ◽

2019 ◽

Vol 7 (3) ◽

pp. 37 ◽

Cited By ~ 3

Author(s):

Catherine Jarnevich ◽

Nicholas Young

Keyword(s):

Species Distribution ◽

Species Distribution Models ◽

Species Distribution Model ◽

Distribution Model ◽

Model Parameters ◽

Climate Data ◽

Distribution Models ◽

Model Type ◽

Time Periods ◽

The Impact

Species distribution models have many applications in conservation and ecology, and climate data are frequently a key driver of these models. Often, correlative modeling approaches are developed with readily available climate data; however, the impacts of the choice of climate normals is rarely considered. Here, we produced species distribution models for five disparate species using four different modeling algorithms and compared results between two different, but overlapping, climate normals time periods. Although the correlation structure among climate predictors did not change between the time periods, model results were sensitive to both baseline climate period and model method, even with model parameters specifically tuned to a species. Each species and each model type had at least one difference in variable retention or relative ranking with the change in climate time period. Pairwise comparisons of spatial predictions were also different, ranging from a low of 1.6% for climate period differences to a high of 25% for algorithm differences. While uncertainty from model algorithm selection is recognized as an important source of uncertainty, the impact of climate period is not commonly assessed. These uncertainties may affect conservation decisions, especially when projecting to future climates, and should be evaluated during model development.

Download Full-text

Species Distribution Models (SDM) – A Strategic Tool for Predicting Suitable Habitats for Conserving the Target Species

Environmental Information Systems ◽

10.4018/978-1-5225-7033-2.ch023 ◽

2019 ◽

pp. 555-568

Author(s):

Balaguru Balakrishnan ◽

Nagamurugan Nandakumar ◽

Soosairaj Sebastin ◽

Khaleel Ahamed Abdul Kareem

Keyword(s):

Species Distribution ◽

Spatial Data ◽

Environmental Variables ◽

Species Distribution Models ◽

Species Distribution Model ◽

Distribution Model ◽

Suitable Habitat ◽

Target Species ◽

Distribution Models ◽

Suitable Habitats

Download Full-text

When exotic introductions fail: updating invasion beliefs

Biological Invasions ◽

10.1007/s10530-019-02163-x ◽

2019 ◽

Vol 22 (3) ◽

pp. 1097-1107

Author(s):

Daniel K. Heersink ◽

Peter Caley ◽

Dean Paini ◽

Simon C. Barry

Keyword(s):

Species Distribution ◽

Habitat Suitability ◽

Species Distribution Models ◽

Species Distribution Model ◽

Region Of Interest ◽

Distribution Model ◽

Perna Viridis ◽

Prior Belief ◽

Distribution Models ◽

Establishment Probability

AbstractDecisions regarding invasive risk of exotic species are often based on species distribution models projected onto the recipient region of interest. Such projections are essentially a measure of prior belief in the ability of an organism to invade. Whilst many decisions are made on the basis of such projections, it is less clear how such prior belief may be empirically modified on the basis of data, in particular introduction events that haven’t led to establishment. Here, using the Asian green mussel (Perna viridis) as an example, we illustrate how information on failed introduction attempts may be used to continually update our beliefs in the ability of an organism to invade per introduction, and the underlying habitat suitability for establishment. Our results show that the establishment probability of P. viridis per fouled ship visit in the supposedly favourable northern Australian waters are much lower than initially though, and are continuing to decline. A Bayesian interpretation of our results notes the dramatic reduction in our belief of the ability of P. viridis to invade in the light of what we estimate to be 100’s of fouled vessels per year visiting ports without any persistent populations establishing. Under a hypothetico-deductive approach we would reject the null (prior) species distribution model as being useful, and seek to find a better one that can withstand the challenge of data.

Download Full-text

Joint species distribution models of Everglades wading birds to inform restoration planning

PLoS ONE ◽

10.1371/journal.pone.0245973 ◽

2021 ◽

Vol 16 (1) ◽

pp. e0245973

Author(s):

Laura E. D’Acunto ◽

Leonard Pearlstine ◽

Stephanie S. Romañach

Keyword(s):

Species Distribution ◽

Species Distribution Models ◽

Species Distribution Model ◽

Site Occupancy ◽

The United States ◽

Decision Makers ◽

Distribution Model ◽

Distribution Models ◽

Landscape Characteristics ◽

Wading Bird

Restoration of the Florida Everglades, a substantial wetland ecosystem within the United States, is one of the largest ongoing restoration projects in the world. Decision-makers and managers within the Everglades ecosystem rely on ecological models forecasting indicator wildlife response to changes in the management of water flows within the system. One such indicator of ecosystem health, the presence of wading bird communities on the landscape, is currently assessed using three species distribution models that assume perfect detection and report output on different scales that are challenging to compare against one another. We sought to use current advancements in species distribution modeling to improve models of Everglades wading bird distribution. Using a joint species distribution model that accounted for imperfect detection, we modeled the presence of nine species of wading bird simultaneously in response to annual hydrologic conditions and landscape characteristics within the Everglades system. Our resulting model improved upon the previous model in three key ways: 1) the model predicts probability of occupancy for the nine species on a scale of 0–1, making the output more intuitive and easily comparable for managers and decision-makers that must consider the responses of several species simultaneously; 2) through joint species modeling, we were able to consider rarer species within the modeling that otherwise are detected in too few numbers to fit as individual models; and 3) the model explicitly allows detection probability of species to be less than 1 which can reduce bias in the site occupancy estimates. These improvements are essential as Everglades restoration continues and managers require models that consider the impacts of water management on key indicator wildlife such as the wading bird community.

Download Full-text

Very high resolution environmental predictors in species distribution models

Progress in Physical Geography Earth and Environment ◽

10.1177/0309133313512667 ◽

2013 ◽

Vol 38 (1) ◽

pp. 79-96 ◽

Cited By ~ 63

Author(s):

Jean-Nicolas Pradervand ◽

Anne Dubuis ◽

Loïc Pellissier ◽

Antoine Guisan ◽

Christophe Randin

Keyword(s):

High Resolution ◽

Species Distribution ◽

Species Distribution Models ◽

Environmental Data ◽

Swiss Alps ◽

Fine Scale ◽

Distribution Models ◽

Model Quality ◽

Vegetation Height ◽

Very High

Recent advances in remote sensing technologies have facilitated the generation of very high resolution (VHR) environmental data. Exploratory studies suggested that, if used in species distribution models (SDMs), these data should enable modelling species’ micro-habitats and allow improving predictions for fine-scale biodiversity management. In the present study, we tested the influence, in SDMs, of predictors derived from a VHR digital elevation model (DEM) by comparing the predictive power of models for 239 plant species and their assemblages fitted at six different resolutions in the Swiss Alps. We also tested whether changes of the model quality for a species is related to its functional and ecological characteristics. Refining the resolution only contributed to slight improvement of the models for more than half of the examined species, with the best results obtained at 5 m, but no significant improvement was observed, on average, across all species. Contrary to our expectations, we could not consistently correlate the changes in model performance with species characteristics such as vegetation height. Temperature, the most important variable in the SDMs across the different resolutions, did not contribute any substantial improvement. Our results suggest that improving resolution of topographic data only is not sufficient to improve SDM predictions – and therefore local management – compared to previously used resolutions (here 25 and 100 m). More effort should be dedicated now to conduct finer-scale in-situ environmental measurements (e.g. for temperature, moisture, snow) to obtain improved environmental measurements for fine-scale species mapping and management.

Download Full-text

Examining Local Transferability of Predictive Species Distribution Models for Invasive Plants: An Example with Cogongrass (Imperata cylindrica)

Invasive Plant Science and Management ◽

10.1614/ipsm-d-10-00077.1 ◽

2011 ◽

Vol 4 (4) ◽

pp. 390-401 ◽

Cited By ~ 14

Author(s):

Gary N. Ervin ◽

D. Christopher Holly

Keyword(s):

Invasive Species ◽

Species Distribution ◽

Species Distribution Models ◽

Imperata Cylindrica ◽

Environmental Data ◽

Training Data ◽

Distribution Data ◽

Strongly Correlated ◽

Distribution Models ◽

Tree Canopy Cover

AbstractSpecies distribution modeling is a tool that is gaining widespread use in the projection of future distributions of invasive species and has important potential as a tool for monitoring invasive species spread. However, the transferability of models from one area to another has been inadequately investigated. This study aimed to determine the degree to which species distribution models (SDMs) for cogongrass, developed with distribution data from Mississippi (USA), could be applied to a similar area in neighboring Alabama. Cogongrass distribution data collected in Mississippi were used to train an SDM that was then tested for accuracy and transferability with cogongrass distribution data collected by a forest management company in Alabama. Analyses indicated the SDM had a relatively high predictive ability within the region of the training data but had poor transferability to the Alabama data. Analysis of the Alabama data, via independent SDM development, indicated that predicted cogongrass distribution in Alabama was more strongly correlated with soil variables than was the case in Mississippi, where the SDM was most strongly correlated with tree canopy cover. Results suggest that model transferability is influenced strongly by (1) data collection methods, (2) landscape context of the survey data, and (3) variations in qualitative aspects of environmental data used in model development.

Download Full-text

Optimization of the Fuzzy Matter Element Method for Predicting Species Suitability Distribution Based on Environmental Data

Sustainability ◽

10.3390/su10103444 ◽

2018 ◽

Vol 10 (10) ◽

pp. 3444 ◽

Cited By ~ 5

Author(s):

Quanzhong Zhang ◽

Haiyan Wei ◽

Zefang Zhao ◽

Jing Liu ◽

Qiao Ran ◽

...

Keyword(s):

Species Distribution ◽

Habitat Suitability ◽

Species Distribution Model ◽

Environmental Data ◽

Distribution Model ◽

New Approach ◽

Suitability Index ◽

Biological Laws ◽

Experimental Process ◽

Key Variables

Over the years, with the efforts of many researchers, the field of species distribution model (SDM) has been well explored. The model of fuzzy matter elements (FME), which, combined with GIS to predict species distribution, has received extensive attention since its emergence. Based on previous studies, this paper improved FME, extended the scope of the membership degree and habitat suitability index, and explored the unsuitable areas of species. We have enhanced the limitation effect of key variables on species habitats, making the operation of FME more consistent with biological laws. By optimizing the FME, it could avoid the accumulation of predicted errors with multi-variables, and make the predicted results more reasonable. In this study, Gynostemma pentaphyllum (Thunb.) Makino was used as an example. The experimental process used several major environmental variables (climate, soil, and terrain variables) to predict the habitat suitability distribution of G. pentaphyllum in China for its current and future period, which includes the period of 2050s (average for 2041–2060) and 2070s (average for 2061–2080) under representative concentration pathways 4.5 (RCP4.5). The results of the analysis showed that the model performed well with a high accuracy by reducing the redundancy of the environmental data. The study could relieve the reliance on a large database of environmental information and propose a new approach for protecting the G. pentaphyllum in unsuitable areas under climate change.

Download Full-text