No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Novel methods to select environmental variables in MaxEnt: A case study using invasive crayfish
... distribution pattern of the target species 17,18 . Therefore, to compensate for the uncertainty of individual methods, it may be necessary to adopt an ensemble approach that combines two or more variable selection methods [19][20][21][22] . ...
... In this study, although the difference was not significant, the accuracy and TSS of the ensemble model on the test dataset were slightly better compared to the individual models, while the F1 score was the second highest, following the BIO-COR model. This indicates that using different variable selection methods in an ensemble of models can ensure robustness and reliability in predicting potential distributions 17,18 . In addition to performance metrics, the ensemble model can identify regions commonly suggested by individual models, thereby minimizing uncertainty in defining areas at risk for pest outbreaks 22 . ...
The vine weevil (Otiorhynchus sulcatus) is a polyphagous pest that affects various economically important crops, but its potential distribution has not been studied. This research developed multiple species distribution models (SDMs) using different variable selection methods, including correlation, biological considerations, and principal component analysis, and integrated them into an ensemble model to predict the pest’s distribution under climate change. The MaxEnt algorithm was used to develop the models, showing robust performance with raw bioclimatic variables (TSS 0.34–0.37, F1 score 0.60–0.67), while lower performance and different distribution patterns were observed with reconstructed variables (TSS 0.13, F1 score 0.48). The vine weevil was predicted to be primarily distributed in North America and Europe, with the highest invasion risk in Far East Asia and northern India. Climate change could shift its habitat northward, particularly in areas where it currently occurs, and human activities may help spread the pest to new regions. This study offers a potential distribution map to aid in monitoring and controlling the vine weevil, emphasizing the importance of variable selection methods in predictive modeling.
... Their use has recently increased due to the improved availability of species occurrence data and environmental variables accessible through various online platforms (e.g., Worldclim data, Global Biodiversity Information Facilities-GBIF). Thus, various algorithms have been developed to analyze these data, with the Maximum Entropy (MaxEnt) algorithm being the most widely used due to its better predictive ability (Phillips et al., 2006;Zeng et al., 2016;Phillips et al., 2017). The adoption of these widely used SDM tools for the assessment of climate change impact on species distribution has tremendously improved ecological research, conservation efforts, and resource management (Pacifici et al., 2015;Zurell et al., 2020). ...
... Earlier studies showed that MaxEnt models performed better than other models when the sample sizes were small (Schmidt et al. 2020;Zeng et al. 2016;Wei et al. 2018). There are still certain drawbacks to the D. odorifera model, even if it performed well. ...
The influence of global climate change on endangered species is of growing concern, especially for rosewood species that are in urgent need of protection and restoration. Ecological niche models are commonly used to evaluate probable species’ distribution under climate change and contribute to decision-making to define efficient management strategies. A model was developed to forecast which habitat was most likely appropriate for the Dalbergia odorifera . We screened the main climatic variables that describe the current geographic distribution of the species based on maximum entropy modelling (Maxent). We subsequently assessed its potential future distribution under moderate (RCP 2.6 ) and severe (RCP 8.5 ) climate change scenarios for the years 2050 and 2070. The precipitation ranges of the wettest month and the warmest quarter are the primary limiting factors for the current distribution of D. odorifera among the climatic predictors. Climate change will be expected to have beneficial effects on the distribution range of D. odorifera . In conclusion, the main limits for the distribution of D. odorifera are determined by the level of precipitation and human activities. The results of this study indicate that the coasts of southern China and Chongqing will play a key role in the protection and restoration of D. odorifera in the future.
... In order to improve the performance of the MaxEnt model, this study uses the "kuenm" R package to optimize the two parameters of feature combination (FC) and regularization multiplier(RM) to help reduce model overfitting and complexity, thereby significantly improving the model's performance. prediction accuracy (Zeng et al., 2016;Radosavljevic and Anderson, 2014). From all 1,160 model species selected from the four native tree species, the optimal model with a missing rate of less than 5% and a minimum delta AICc value was screened. ...
Introduction
The target valleys along the Jinsha, Nujiang, Lancang, and Yuanjiang Rivers exhibit acute human-land conflicts and ecosystem vulnerability. Predicting the distribution of potential suitable habitats for native tree species in Yunnan Province provides basin-scale insights for the management of ecosystems in dry and hot valleys, thereby advancing restoration planning in dry-hot valleys.
Methods
This study investigates native tree species suitability in Yunnan’s dry-hot valleys using an integrated MaxEnt-InVEST modeling framework.
Results
Temperature and precipitation emerged as dominant bioclimatic controls, with optimal species occurrence (1 000–2 500 m) showing negative elevation correlation. Four native tree species (Osteomeles schwerinae, Phyllanthus emblica, Quercus francetii and Sapindus delavayi) displayed fragmented suitable areas along mountainous riparian zones, while habitat quality hotspots mainly covered non-urbanized regions, avoiding central urban clusters and northeastern/southeastern karst zones. The coupled model demonstrated significantly improved accuracy compared to the standalone MaxEnt by incorporating land-use impacts, with Yuanmou County case analysis confirming the enhanced predictive capability through actual distribution patterns. Spatial prioritization identified core planting clusters in central/southeastern valleys, though fragmented by agricultural encroachment.
Discussion
This methodology provides a cost-effective solution for vegetation restoration planning in ecologically fragile dry-hot ecosystems. The research results can provide scientific support for the restoration of degraded ecosystems in dry-hot valleys of Yunnan Province, the national Afforestation program and soil and water conservation projects.
... Multicollinearity usually occurs among the predictor variables in ENMs. In addition, reducing the number of variables has been proven to improve the accuracy and transferability of the model and mitigate the risk of overfitting [27,28]. In order to minimize the collinearity and retain valuable variables, a stepwise removal process was adopted by pre-running the model. ...
In the context of global warming, there is an increasing risk of the emergence and re-emergence of vector-borne diseases (VBDs). As one of the most important vectors, Culex tritaeniorhynchus can carry and transmit numerous human and animal infectious pathogens. To better understand the current distribution and possible future dynamics of Cx. tritaeniorhynchus in China, an ecological niche modeling approach (MaxEnt) was adopted to model its current and future habitat suitability. The most comprehensive dataset (1100 occurrence records) in China to date was established for model training. Multiple global climate models (GCMs) and climate change scenarios were introduced into the model to counter the uncertainties of future climate change. Based on the model prediction, Cx. tritaeniorhynchus currently exhibits high habitat suitability in southern, central, and coastal regions of China. It is projected that its suitable niche will experience continuous expansion, and the core distribution is anticipated to shift northward in the future 21st century (by the 2050s, 2070s and 2090s). Several environmental variables that reflect temperature, precipitation, and land-use conditions were considered to have a significant influence on the distribution of Cx. tritaeniorhynchus, among which annual mean temperature and urban land contribute the most to the model. Our study conducted a quantitative analysis of the shift and expansion of the future habitats of Cx. tritaeniorhynchus, providing references for vector monitoring and the prevention and control of VBDs.
... (3) vegetation variables including land use/land cover (LULC; Chen et al., 2020), proximity to the managed forest (i.e., forests with disturbances since 2000), the natural forest (i.e., forests without disturbances since 2000) and the cropland; and (4) anthropogenic variables including proximity to roads and proximity to villages. Then, we used Pearson correlation coefficients to reduce the multicollinearity and improve model interpretability and performance since they can identify redundant variables, quantify the relationship between predicted and observed data, and support accurate, standardized comparisons across models (Elith et al., 2006;Zeng et al., 2016). ...
Interdisciplinary efforts are fundamental for achieving successful conservation translocations. However, behavioral information is usually lacking to guide conservation translocations for social animals. This is particularly significant for the conservation of endangered Asian elephants. Therefore, by tracing the long-term behavioral logbook records in the southern central part of Myanmar, our study highlighted that younger semi-captive elephants (male ≤21 years old; female ≤42 years old) were identified as suitable candidates for translocations since they were more easily accepted by the wild population, with fewer fighting events and higher mingling probability. Furthermore, we recorded 136 present data combining field surveys and collection from literature , and we identified 4349.69 km 2 of suitable habitat in this region located around 10 km away from the villages, closer to managed forests and water. This study integrated ecological and behavioral information to support reinforcement conservation for Asian elephants in Southeast Asia, where most of the semi-captive elephants are distributed. These insights could guide more effective reinforcement projects by considering age and sex for improved success in integration. Additionally, our study emphasizes the importance of habitats near human-dominated areas, which are preferred by elephants, offering practical implications for habitat management and human-elephant conflict mitigation efforts. Further research efforts from the behavioral perspectives, such as using camera trappings or video recordings, are encouraged to facilitate social animal conservation.
... (3) vegetation variables including land use/land cover (LULC; Chen et al., 2020), proximity to the managed forest (i.e., forests with disturbances since 2000), the natural forest (i.e., forests without disturbances since 2000) and the cropland; and (4) anthropogenic variables including proximity to roads and proximity to villages. Then, we used Pearson correlation coefficients to reduce the multicollinearity and improve model interpretability and performance since they can identify redundant variables, quantify the relationship between predicted and observed data, and support accurate, standardized comparisons across models (Elith et al., 2006;Zeng et al., 2016). ...
Interdisciplinary efforts are fundamental for achieving successful conservation translocations. However, behavioral information is usually lacking to guide conservation translocations for social animals. This is particularly significant for the conservation of endangered Asian elephants. Therefore, by tracing the long‐term behavioral logbook records in the southern central part of Myanmar, our study highlighted that younger semi‐captive elephants (male ≤21 years old; female ≤42 years old) were identified as suitable candidates for translocations since they were more easily accepted by the wild population, with fewer fighting events and higher mingling probability. Furthermore, we recorded 136 present data combining field surveys and collection from literature, and we identified 4349.69 km² of suitable habitat in this region located around 10 km away from the villages, closer to managed forests and water. This study integrated ecological and behavioral information to support reinforcement conservation for Asian elephants in Southeast Asia, where most of the semi‐captive elephants are distributed. These insights could guide more effective reinforcement projects by considering age and sex for improved success in integration. Additionally, our study emphasizes the importance of habitats near human‐dominated areas, which are preferred by elephants, offering practical implications for habitat management and human‐elephant conflict mitigation efforts. Further research efforts from the behavioral perspectives, such as using camera trappings or video recordings, are encouraged to facilitate social animal conservation.
... We used a combination of multicollinearity ) and a priori predictor selection based on ecological knowledge (Zeng et al. 2016) to determine the final list of predictors. Once all predictors determined to have ecological relevance were retrieved, we used pair-wise Pearson correlations to remove highly correlated predictors (Kuemmerle et al. 2011), |r| > 0.50. ...
When introduced species invade new environments, they often overlap with native species currently occupying those spaces, either spatially through suitable habitat or environmentally through their realized niches. The goal of this research is to determine the overlap between native New England cottontail (Sylvilagus transitionalis) and introduced eastern cottontail (Sylvilagus floridanus) to identify potential areas of invasion by the eastern cottontail and potential areas of refuge for the New England cottontail from the eastern cottontail (Connecticut, USA). Using presence data from a regional, standardized monitoring protocol, we developed habitat suitability models using Maxent and conducted niche overlap analyses using environmental principal component analysis. We used several covariates that reflected proximity to habitat characteristics, such as young forest, shrubland, and understory, as well as proximity to threats, such as development. We also included topographic and climatic covariates. We used the Guidos software to categorize the spatial arrangement of young forest, shrubland, and understory vegetation. We found that the overlap in both niches and suitable habitat was high for two species. Only areas of low precipitation and high elevation shifted niches in favor of the New England cottontail. We also found that habitat suitability for the New England cottontail was higher when patches of mature forest without understory were within complexes of young forest, shrubland, and mature forest with understory. Increasing habitat heterogeneity could improve the habitat suitability of existing patches or create new patches for New England cottontail. However, habitat management alone is likely not enough to discourage eastern cottontail; thus, direct species management, such as removal of eastern cottontail and augmentation of New England cottontail populations, should be explored.
... Models were created using systematic sets of environmental variables that avoided using directly correlated variables (e.g., river distance to the river's inlet and river distance to the river's outlet). The initial model included all the variables in the systematic set after which the variable with the least importance in the MaxEnt jackknife output was eliminated until only two variables remained (da Silva et al., 2022;Yiwen et al., 2016). All MaxEnt models created were assessed using the area under the receiver operating characteristic curve (AUC), a probability measure that the model is able to correctly differentiate random locations from species' detection locations (Merow et al., 2013;Phillips et al., 2006;Razgour et al., 2011). ...
Native freshwater gastropods are a highly diverse and imperiled group of mollusks in North America and are influenced by a growing number of problematic invasive species. Consequently, there has been an increased need for understanding aquatic gastropod assemblages throughout North America to implement conservation and management strategies. In the Laurentian Great Lakes, gastropod surveys have been sparse, and most surveys have focused on invasive species. To investigate gastropod assemblages in two large connecting rivers of the Great Lakes, the Detroit and St. Clair rivers, benthic surveys were conducted in 2019 and 2021. Sites in the Detroit River (n = 56) and the St. Clair River (n = 51) were surveyed using petite PONAR grabs from which gastropod shells were identified and quantified to family or a group of two combined families. In both the Detroit and St. Clair rivers, the gastropod family Pleuroceridae (37 % and 56 % total composition, respectively) and combined families Amnicolidae + Hydrobiidae (42 % and 23 % total composition, respectively) contributed the most to overall gastropod composition. Invasive Potamopyrgus antipodarum shells were identified at 4 (7 %) Detroit River sites and 10 (20 %) St. Clair River sites and represent the first documented occurrence in the Detroit River. Although this study was limited to quantifying densities based on shells and cannot assume live-collected snail densities, these results provide a baseline knowledge of the gastropod assemblages and habitat use in these two large river systems which can be used to implement conservation and management strategies.
... Most of these studies use species distribution models (SDMs). These models examine the relationship between species occurrence and environmental factors to predict the current and future species distribution (Zeng et al., 2016, Phillips et al., 2017Barlow et al., 2021). Among SDMs, the maximum entropy algorithm (MaxEnt) is widely used because of its high predictive power (Elith et al., 2006, Phillips et al., 2006. ...
... The principle of maximum entropy is a mathematical method used to infer unknown information from limited known information. Its core principle is to preserve all uncertainties, meaning that, in the absence of subjective assumptions, the probability distribution with the maximum entropy value minimizes risk [50]. Assuming the probability distribution of a discrete random variable x is P(x), the entropy is calculated using the following formula: ...
Chinese forests, particularly the coniferous forest ecosystems represented by pines, play a crucial role in the global carbon cycle, significantly contributing to mitigating climate change, regulating regional climates, and maintaining ecological balance. However, pine wilt disease (PWD), caused by the pine wood nematode (PWN), has become a major threat to forest carbon stocks in China. This study evaluates the impact of PWN invasion on forest carbon stocks in China using multi-source data and an optimized MaxEnt model, and the study analyzes this invasion’s spread trends and potential risk areas. The results show that the high-suitability area for PWN has expanded from 68,000 km² in 2002 to 184,000 km² in 2021, with the spread of PWN accelerating, especially under warm and humid climate conditions and due to human activities. China’s forest carbon stocks increased from 111.34 billion tons of carbon (tC) to 168.05 billion tC, but the carbon risk due to PWN invasion also increased from 87 million tC to 99 million tC, highlighting the ongoing threat to the carbon storage capacity. The study further reveals significant differences in tree species’ sensitivity to PWN, with highly sensitive species such as Masson’s pine and black pine mainly concentrated in the southeastern coastal regions, while less sensitive species such as white pine and larch show stronger resistance in the northern and southwestern areas. This finding highlights the vulnerability of high-sensitivity tree species to PWN, especially in high-risk areas such as Guangdong, Guangxi, and Guizhou, where urgent and effective control measures are needed to reduce carbon stock losses. To address this challenge, the study recommends strengthening monitoring in high-risk areas and proposes specific measures to improve forest management and policy interventions, including promoting cross-regional joint control, enhancing early warning systems, and utilizing biological control measures, while encouraging local governments and communities to actively participate. By strengthening collaboration and implementing control measures, the health and sustainable development of forest ecosystems can be ensured, safeguarding the forests’ important role in climate regulation and carbon sequestration and contributing to global climate change mitigation.
... The same applies to the selection of variables. It has been shown that automated variable selection yields superior outcomes compared to modeling approaches without such functionalities [27,41]. If variable selection is conducted by an automated algorithm, it means that the modeler inputs all relevant variables into the model. ...
Background
Ticks are the primary vectors of numerous zoonotic pathogens, transmitting more pathogens than any other blood-feeding arthropod. In the northern hemisphere, tick-borne disease cases in humans, such as Lyme borreliosis and tick-borne encephalitis, have risen in recent years, and are a significant burden on public healthcare systems. The spread of these diseases is further reinforced by climate change, which leads to expanding tick habitats. Switzerland is among the countries in which tick-borne diseases are a major public health concern, with increasing incidence rates reported in recent years.
Methods
In response to these challenges, the “Tick Prevention” app was developed by the Zurich University of Applied Sciences and operated by A&K Strategy Ltd. in Switzerland. The app allows for the collection of large amounts of data on tick attachment to humans through a citizen science approach. In this study, citizen science data were utilized to map tick attachment to humans in Switzerland at a 100 m spatial resolution, on a monthly basis, for the years 2015 to 2021. The maps were created using a state-of-the-art modeling approach with the software extension spatialMaxent, which accounts for spatial autocorrelation when creating Maxent models.
Results
Our results consist of 84 maps displaying the risk of tick attachments to humans in Switzerland, with the model showing good overall performance, with median values ranging from 0.82 in 2018 to 0.92 in 2017 and 2021 and convincing spatial distribution, verified by tick experts for Switzerland. Our study reveals that tick attachment to humans is particularly high at the edges of settlement areas, especially in sparsely built-up suburban regions with green spaces, while it is lower in densely urbanized areas. Additionally, forested areas near cities also show increased risk levels.
Conclusions
This mapping aims to guide public health interventions to reduce human exposure to ticks and to inform the resource planning of healthcare facilities. Our findings suggest that citizen science data can be valuable for modeling and mapping tick attachment risk, indicating the potential of citizen science data for use in epidemiological surveillance and public healthcare planning.
Graphical Abstract
... To improve the accuracy of model predictions, we adjusted two parameters: the regularization multiplier (RM) and feature combination (FC) [35,36]. First, the distribution point data and related environmental variables for T. helena were selected. ...
Butterflies are highly sensitive to climate change, and Troides helena, as an endangered butterfly species, is also affected by these changes. To enhance the conservation of T. helena and effectively plan its protected areas, it is crucial to understand the potential impacts of climate change on its distribution. This study utilized a MaxEnt model in combination with ArcGIS technology to predict the global potential suitable habitats of T. helena under current and future climate conditions, using the species’ distribution data and relevant environmental variables. The results indicated that the MaxEnt model provided a good prediction accuracy for the distribution of T. helena. Under the current climate scenario, the species is primarily distributed in tropical regions, with high suitability areas concentrated in tropical rainforest climates. In future climate scenarios, the suitable habitat areas for T. helena in medium and high suitability categories generally show an expansion trend, which increases over time. Especially under the SSP5-8.5 scenario, by the 2090s, the area of high suitability for T. helena is projected to increase by 42.85%. The analysis of key environmental factors revealed that precipitation of the wettest quarter (Bio16) was the most significant environmental factor affecting the distribution of T. helena. The species has high demands for precipitation and temperature and can adapt to future climate warming. This study is valuable for identifying the optimal conservation areas for T. helena and provides a reference for future conservation efforts.
... The generated resulting model creates a so-called bioclimatic envelope illustrating the potential species niche ( Elith et al., 2011 ). This niche model is used to examine the geographic space and highlights areas where the studied species is likely to occur under the given predicted environmental factors and scenarios ( Van Echelpoel et al., 2015 ;Zeng et al., 2016 ). Many different algorithms are used in species distribution modelling depending of each particular situation, and the predicted distributions can vary between algorithms ( Elith et al., 2011 ;Marmion et al., 2009 ). ...
Reptile fauna should be considered a conservation objective, especially in respect of the impacts of climate change on their distribution and range’s dynamics. Investigating the environmental drivers of reptile species richness and identifying their suitable habitats is a fundamental prerequisite to setting efficient long-term conservation
measures. This study focused on geographical patterns and estimations of species richness for herpetofauna widely spread Z. vivipara, N. natrix, V. berus, A. colchica , and protected in Latvia C. austriaca, E. orbicularis , L. agilis inhabiting northern (model territory Latvia) and southern (model territory Ukraine) part of their European range. The ultimate goal was to designate a conservation network that will meet long-term goals for survival of the target species in the context of climate change. We used stacked species distribution models for creating maps depicting the distribution of species richness under current and future (by 2050) climates for marginal reptile populations. Using cluster analysis, we showed that this herpeto-complex can be divided into “widespread species ” and “forest
species ”. For all forest species we predicted a climate-driven reduction in their distribution range both North (Latvia) and South (Ukraine). The most vulnerable populations of “forest species ” tend to be located in the South of their range, as a consequence of northward shifts by 2050. By 2050 the greatest reduction in range is predicted for currently widely spread Z. vivipara (by 1.4 times) and V. berus (by 2.2 times). In terms of designing an effective protected-area network, these results permit to identify priority conservation areas where the full ensemble of selected reptile species can be found, and confirms the relevance of abiotic multi-factor GIS-modelling for achieving this goal.
... Feature classes are transformations that can be applied to each predictor variable by the model, for example, linear and quadratic transformations, while the regularization multiplier is a penalty to avoid overfitting (Merow et al., 2013). Fifteen combinations were tested using the corrected Akaike's Information Criterion (AICc) as a selection criterion (R-package ENMeval; Kass et al., 2021) (Table 2A; Zeng et al., 2016). Following the recommendations of Merow et al. (2013), we included all combinations of the feature classes L, LQ and LQH (with L linear, Q quadratic and H hinge) and the regularization multipliers 1, 2, 4, 8 and 32. ...
The withdrawal of the United Kingdom from the European Union will likely result in reduced fishing grounds for the Belgian fishing fleet. This fleet now targets demersal fish, but there used to be a tradition of catching Atlantic herring (Clupea harengus). After the stock collapse of Atlantic herring in the 1970s, fishing on herring by the Belgian fleet did not recover and herring quotas are now exchanged with the Netherlands and Germany. To assess the feasibility of reintroducing herring fisheries for the Belgian fishing fleet, our study created spatiotemporal species distribution models for Atlantic herring in the Northeast Atlantic Ocean, focusing results on the Belgian Part of the North Sea. In total 30078 occurrence records were derived and processed to fit species-environmental relationships with temperature, salinity, seabed characteristics and plankton concentration using Maximum entropy (Maxent) models. The Area Under the Curve of the Receiver Operating Characteristic plot (AUC) and the True Skill Statistic (TSS) were used to assess model fit. Models performed well (AUC > 0.7 and TSS > 0.6). While a broad spatiotemporal distribution of Atlantic herring in the Northeast Atlantic Ocean was inferred, regional differences show that herring habitat is most suitable during winter months in the Belgian Part of the North Sea for both adult and larval herring (habitat suitability index > 75%). This regional trend in the Belgian Part of the North Sea was negatively correlated (R = -0.8) with the North Atlantic Oscillation (NAO). We anticipate that these findings will provide valuable insights for policymakers to implement sustainable fisheries management practices.
... First, they are pre-selected based on the deep knowledge on the biology and other characteristics of the studied taxa. Second approach is automatic removal of the least contributing variables (e.g., Zeng, Low & Yeo, 2016). The second approach was chosen, because the biology is poorly known for all three species. ...
Insects are poikilothermic organisms and temperature increase usually accelerates their development rates, population and distribution area growth. Therefore, it is assumed that global warming can be beneficial for the pests and other widespread species at least in the relatively cool temperate zones. However, climate change’s effect on the widespread species in the Palearctic remains poorly studied. This work was performed on three plant bug species (Insecta: Heteroptera: Miridae), at present inhabiting Europe and Asia. Liocoris tripustulatus is known from the Western Palearctic, Lygocoris pabulinus occupies the territories from Western Europe to South Asia, Lygus punctatus is distributed from Northern Europe to the Far East. In this paper, it is tested whether temperature rise is positively connected with the area of preferred climatic conditions for those species, and explores the particular climatic variables which can be limiting for the distribution of those species. Maxent software was used for the environmental niche modeling and to find the variables with significant contribution to the climatic models for the studied species. Based on those models, areas with preferred climatic conditions over different periods were calculated in QGIS. Principal component analysis and logistic regression were performed to find the variables highly contributing to the differences between the species. The results contradict the assumption that temperature growth alone can be a predictor for the widespread species and pest distribution range change. All species differ in suitable climatic conditions and their area dynamics in time, and the temperature affects each species differently. Only Liocoris tripustulatus might significantly expand its distribution area by 2070 due to the climate change. The areas in Asia and above the polar circle will be more suitable by that time for all three species than now. However, conditions in Europe might be less suitable for Lygocoris pabulinus and Lygus punctatus in the future. Both, temperature and precipitation variables, can be important for shaping distribution of Liocoris tripustulatus and Lygocoris pabulinus . Mean annual temperature and temperature in winter, most probably, limit the distribution of at least Liocoris tripsutulatus and Lygus punctatus , but changes in this variable affect those two species differently.
... Comparison of multiple SDMs is fundamental to selecting the models, or the combination of models (i.e., ensemble modeling), permitting minimal prediction error (Guisan et al., 2017). Most studies compare more than one model to minimize error and get the best result (Brown et al., 2014;Zeng et al., 2016). Some studies advise using presence-absence models, while others suggest presence-only models. ...
Geographical distribution and diversity patterns of bird species are influenced by climate change. The Rouget's rail (Rougetius rougetii) is a ground‐dwelling endemic bird species distributed in Ethiopia and Eritrea. It is a near‐threatened species menaced by habitat loss, one of the main causes of population declines for bird species. The increasing effects of climate change may further threaten the species’ survival. So far, the spatial distribution of this species is not fully documented. With this study, we develop current potential suitable habitat and predict the future habitat shift of R. rougetii based on environmental data such as bioclimatic variables, population density, vegetation cover, and elevation using 10 algorithms. We evaluated the importance of environmental factors in shaping the bird's distribution and how it shifts under climate change scenarios. We used 182 records of R. rougetii from Ethiopia and nine bioclimatic, population density, vegetation cover, and elevation variables to run the 10 model algorithms. Among 10 algorithms, eight were selected for ensembling models according to their predictive abilities. The current suitable habitats for R. rougetii were predicted to cover an area of about 82,000 km² despite being highly fragmented. The model suggested that temperature seasonality (bio4), elevation, and mean daily air temperatures of the driest quarter (bio9) contributed the most to delimiting suitable areas for this species. R. rougetii is sensitive to climate change associated with elevation, which leads shrinking distribution of suitable areas. The projected spatial and temporal pattern of habitat loss of R. rougetii suggests the importance of climate change mitigation and implementing long‐term conservation and management strategies for this threatened endemic bird species.
... Because Maxent is a machine learning algorithm, (strongly) correlating variables have no effect on performance when predictions are only made for the current time period 73 . However, there is a risk of overparameterisation, and suitability maps based on ecologically relevant variables are more accurate than maps based on arbitrary environmental variables [74][75][76] . The percentage contribution of each environmental variable was therefore retrieved from the results of the initial models and used to select the most influential variables. ...
Mosquitoes are important vectors of disease pathogens and multiple species are undergoing geographical shifts due to global changes. As such, there is a growing need for accurate distribution predictions. Ecological niche modelling (ENM) is an effective tool to assess mosquito distribution patterns and link these to underlying environmental preferences. Typically, macroclimatic variables are used as primary predictors of mosquito distributions. However, they likely undervalue local conditions and intraspecific variation in environmental preferences. This is problematic, as mosquito control takes place at the local scale. Utilising high-resolution (10 × 10 m) Maxent ENMs on the island of Bonaire as model system, we explore the influence of local environmental variables on mosquito distributions. Our results show a distinct set of environmental variables shape distribution patterns across ecologically-distinct species, with urban variables strongly associated with introduced species like Aedesaegypti and Culexquinquefasciatus, while native species show habitat preferences for either mangroves, forests, or ephemeral water habitats. These findings underscore the importance of distinct local environmental factors in shaping distributions of different mosquitoes, even on a small island. As such, these findings warrant further studies aimed at predicting high-resolution mosquito distributions, opening avenues for preventative management of vector-borne disease risks amidst ongoing global change and ecosystem degradation.
... These models estimate species distribution patterns using digital environmental datasets and species occurrence records, which helps with planning for species conservation (Zhang, Xu, Capinha, Weterings, & Gao, 2019). Among the various SDMs available, the Maximum Entropy (MaxEnt) algorithm reigns supreme due to its superior predictive performance (Hijmans, Phillips, Leathwick, & Elith, 2017;Zeng, Low, & Yeo, 2016). Based on the 19 BIOCLIM variables from the WorldClim database, MaxEnt has been frequently used to simulate species distribution of numerous endangered species, including Pterocarpus erinaeus (Biaou et al., 2023), Faidherbia albida (Delile) A. Chev. ...
Climate change is likely to affect the distribution of species worldwide. Understanding how these changes affect species distribution is important for planning conservation strategies and sustainable management methods. Adansonia digitata L. is of major ecological and socio-economic importance in Malawi, South Africa, but is highly threatened in its habitat. This work aims to investigate the effect of climate change on the ecological niche of Baobab and to find suitable habitats for its conservation and cultivation in Malawi. The distribution of this species was modeled using a maximum entropy algorithm (MaxEnt) based on 21 environmental variables and the occurrence of 480 species. Habitat prioritization was performed using Zonation software. Our results show that the variable contributing most significantly was the warmest month (47%), followed by isotherms (13.9%) and precipitation of the coldest quarter (8.6%). Under the current model, 1.17% of Malawi’s territory is highly favorable for baobab development. A slight increase of 0.09% and 0.38% in highly favorable zones is predicted by 2055 under scenarios SSP370 and SSP 585, respectively. Southern Malawi and parts of the Central region should be prioritized in baobab reforestation policies to optimize conservation and value chain sustainability for baobab. Under the current model, 1.17% of Malawi will be highly favorable for baobab. A slight increase of 0.09 % and 0.38 % in highly favorable zones is predicted by 2055 under scenarios SSP 370 and SSP 585, respectively. Priority areas (98-100%) for conservation and cultivation of Baobab was a male located in the Southern region (34.51%) and central (7.62%), in contrast to the Northern region (0.21%). Our results suggest that climate change causes the reduction and shift of suitable habitats for species along a south-north gradient. These findings highlight the urgent need to incorporate climate change projections into conservation plans. Identifying and prioritizing suitable habitats in the southern and central regions is crucial for effective conservation and sustainability.
... Ecological predictors, WorldClim's bioclimatic variables, are well documented for being highly multicollinear (Arif et al., 2007). Some studies address this issue by removing multicollinear variables (Cobos, Peterson, Barve, et al., 2019; entirely while other studies suggest using a priori-determined selection method (Zeng et al., 2016). A purely algorithmic-based selection could select variables that result in a predictively accurate model with variables that are realistically uninfluential (Smith & Santos, 2020). ...
Hemp (Cannabis sativa L.) has historically played a vital role in agriculture across the globe. Feral and wild populations have served as genetic resources for breeding, conservation, and adaptation to changing environmental conditions. However, feral populations of Cannabis, specifically in the Midwestern United States, remain poorly understood. This study aims to characterize the abiotic tolerances of these populations, estimate suitable areas, identify regions at risk of abiotic suitability change, and highlight the utility of ecological niche models (ENMs) in germplasm conservation. The Maxent algorithm was used to construct a series of ENMs. Validation metrics and MOP (Mobility‐oriented Parity) analysis were used to assess extrapolation risk and model performance. We also projected the final projected under current and future climate scenarios (2021–2040 and 2061–2080) to assess how abiotic suitability changes with time. Climate change scenarios indicated an expansion of suitable habitat, with priority areas for germplasm collection in Indiana, Illinois, Kansas, Missouri, and Nebraska. This study demonstrates the application of ENMs for characterizing feral Cannabis populations and highlights their value in germplasm conservation and breeding efforts. Populations of feral C. sativa in the Midwest are of high interest, and future research should focus on utilizing tools to aid the collection of materials for the characterization of genetic diversity and adaptation to a changing climate.
... Our approach to niche modelling broadly draws from the concepts of information theory (IT), wherein multiple models are constructed, using only relevant predictors, with the specific purpose of testing a certain hypothesis (see Hegyi & Garamszegi, 2011;Whittingham et al., 2006;Zeng et al., 2016). We used this approach in a maximumentropy framework to quantify the influences of topographical, canopy, temperature and precipitation variables in limiting the distribution of the Brown Hornbill to the south of the Brahmaputra River. ...
Aim
In continuous environments, niche limits of species often determine their distribution limits. However, when these limits spatially coincide with a perceived dispersal barrier, the determinants of species' ranges may be confounded. We investigate the distribution pattern of the Brown Hornbill (Aves: Bucerotidae), which spans significant riverine barriers, but stops south of the Brahmaputra River. Considering its preference for low‐elevation evergreen forests, we posit that the lack of sufficient habitats north of the Brahmaputra prevents dispersal of the Brown Hornbill, and not the river itself.
Location
The Brahmaputra valley and the Indo‐Burma hotspot.
Taxa
Anorrhinus austeni, Aceros nipalensis.
Methods
We analysed citizen‐science occurrence data on hornbill presence against a suite of climatic, canopy‐specific and topographical predictors to model the environmental niche of the Brown Hornbill. We used presence‐only maximum entropy modelling in an information theoretic framework, in conjunction with constructing binary logistic regression models using presence and pseudoabsence data. We compared niche models of the Brown Hornbill and the Rufous‐necked hornbill, a close relative with a similar distribution, but which has spanned the Brahmaputra River.
Results
Despite the presence of suitable wet‐evergreen forests north of the Brahmaputra River, the hilly terrains in the region act as a biogeographic barrier for the Brown Hornbill, which prefers lowland evergreen forests. Further, highly suitable regions for the Brown and the Rufous‐necked Hornbills precisely delineate low and high‐elevation evergreen forests respectively, indicating that these birds are separated along an elevational axis.
Main Conclusions
The Brahmaputra River lies at the cusp of two subtly different environmental regimes. It may therefore serve as the niche limit for certain organisms and not as a physical obstacle to their dispersal. Our study implicitly predicts how widespread deforestation prevalent in the lowland evergreen forests of this region adversely impacts the distributions of organisms that depend on them, such as the Brown Hornbill. Further, our study proffers an approach to ascertain determinants of species distributions in a hypothesis testing framework.
... The same applies to variable selection; it has been shown that using too many variables can lead to bias in the results (Irving et al., 2020;Wang et al., 2016;Williams et al., 2012). Moreover, better results were achieved when variables were selected automatically rather than manually by the modeler (Meyer et al., 2019;Yiwen et al., 2016). Additionally, variable selection must be based on a spatial validation method to generate reliable results (Bald et al., 2023;Meyer et al., 2018Meyer et al., , 2019. ...
In response to the pressing challenges of the ongoing biodiversity crisis, the protection of endangered species and their habitats, as well as the monitoring of invasive species are crucial. Habitat suitability modeling (HSM) is often treated as the silver bullet to address these challenges, commonly relying on generic variables sourced from widely available datasets. However, for species with high habitat requirements, or for modeling the suitability of habitats within the geographic range of a species, variables at a coarse level of detail may fall short. Consequently, there is potential value in considering the incorporation of more targeted data, which may extend beyond readily available land cover and climate datasets. In this study, we investigate the impact of incorporating targeted land cover variables (specifically tree species composition) and vertical structure information (derived from LiDAR data) on HSM outcomes for three forest specialist bat species (Barbastella barbastellus, Myotis bechsteinii, and Plecotus auritus) in Rhineland‐Palatinate, Germany, compared to commonly utilized environmental variables, such as generic land‐cover classifications (e.g., Corine Land Cover) and climate variables (e.g., Bioclim). The integration of targeted variables enhanced the performance of habitat suitability models for all three bat species. Furthermore, our results showed a high difference in the distribution maps that resulted from using different levels of detail in environmental variables. This underscores the importance of making the effort to generate the appropriate variables, rather than simply relying on commonly used ones, and the necessity of exercising caution when using habitat models as a tool to inform conservation strategies and spatial planning efforts.
... Comparison of multiple species distribution models is fundamental to selecting the models, or the combination of models (i.e., ensemble modelling), permitting minimal prediction error (Guisan et al., 2017). Most studies compare more than one model to minimize error and get the best result (Brown et al., 2014;Zeng et al., 2016). Some studies advise using presence-absence models, while others suggest presence-only models. ...
Geographical distribution and diversity patterns of bird species are influenced by climate change. The Rouget's rail (Rougetius rougetii) is a ground-dwelling endemic bird species distributed in Ethiopia and Eritrea. It is a near-threatened species menaced by habitat loss, one of the main causes of population declines for bird species. The increasing effects of climate change may further threaten the species' survival. So far, the spatial distribution of this species is not fully documented. With this study, we develop current potential suitable habitat and predict the future habitat shift of R. rougetii based on environmental data such as bioclimatic variables, population density, vegetation cover, and elevation using ten algorithms. We evaluated the importance of environmental factors in shaping the bird's distribution and how it shifts under climate change scenarios. We used 182 records of R. rougetii from Ethiopia and nine bioclimatic, population density, vegetation cover, and elevation variables to run the 10 model algorithms. Among 10 algorithms, eight were selected for ensembling models according to their predictive abilities. The current suitable habitats for R. rougetii were predicted to cover an area of about 82,000 km² despite being highly fragmented. The model suggested that temperature seasonality (bio4), elevation, and mean daily air temperatures of the driest quarter (bio9) contributed the most to delimiting suitable areas for this species. R. rougetii is sensitive to climate change associated with elevation, leading to a large, shrinking distribution of suitable areas. The projected spatial and temporal pattern of habitat loss of R. rougetii suggests the importance of climate change mitigation and implementing long-term conservation and management strategies for this threatened endemic bird species.
... Predictor variables were tested for multicollinearity using the variance inflation factor (VIF). After assessing the VIF for all variables, a stepwise analysis was conducted to remove variables with VIF > 5 (see Zeng et al. 2016;Norberto et al. 2023 for examples of stepwise variable selection). During the stepwise process, the variable with the highest VIF was eliminated, and the VIF of all remaining variables was recalculated until no variables remained with a VIF > 5. ...
Effective management of invasive species often relies upon early detection and rapid implementation of management actions. Unfortunately, early detection can be hampered in coastal regions where sites may be challenging to access and survey or resources are limited. Conventionally considered a freshwater plant with a near-global invaded range, yellow flag iris (Iris pseudacorus) is sufficiently tolerant of salinity to disperse through seawater and successfully invade coastal habitats. Clarifying the characteristics that influence coastal habitat suitability for I. pseudacorus and identifying locations at risk of invasion is critical for monitoring and control efforts. We developed a Maxent species distribution model to inform the risk of coastal invasion by I. pseudacorus in southern British Columbia (Canada). Among the variables assessed, precipitation during the driest quarter was the most important contributor in predicting I. pseudacorus’s coastal habitat suitability (40.6% variable contribution [VC]). Urban areas (24.9% VC), south-facing aspects (6.2% VC) and locations at elevations (7.9% VC) at or above the mean high tide extent were also considered suitable for establishment. Though proximity to perennial freshwater sources (7% VC) was also predicted to provide suitable habitat, close distance to these features was not found to be essential. Regionally, invasion hotspots were predicted in coastal habitats around Tofino/Ucluelet (western Vancouver Island), the Fraser River delta (metro Vancouver), and among the southern Gulf Islands and southern Vancouver Island (capital region). Land managers can use these variables of interest and geographic locations to inform early detection monitoring efforts that help attenuate the emerging threat of I. pseudacorus in coastal ecosystems.
... With the advancement of machine learning techniques, prediction models using different algorithms and quantum approaches have been developed (Guisan & Zimmermann, 2000;Brito et al., 2009;Elith & Leathwick, 2009;Özkan, 2016;Özdemir, 2018;Wei et al., 2018). One of the most widely used machine learning techniques in plant species modeling is MaxEnt, which uses the maximum entropy algorithm (Zeng et al., 2016;Koch et al., 2017;Xu et al., 2019). MaxEnt uses digital climate data and spatial data to create a probability distribution map showing the most suitable habitats for species. ...
In this study, present and future distributions of stone pine due to climate changes were modeled with MaxEnt. CNRM ESM2-1 climate model and bioclimatic variables obtained from the WorldClim database were used as climate models. As climate scenarios, SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5 climate change scenarios and 2041–2060 and 2081–2100 periods were used. Pearson Correlation analysis was performed to prevent high correlation in bioclimatic variables and the multicollinearity problem was eliminated by reducing 19 bioclimatic variables to 9 variables. The contribution of bioclimatic variables to the model was determined by the Jackknife test. To determine the spatial and locational differences between the present and future potential distributions estimated for the species, an analysis of change was conducted. According to the findings of the study, our model has a very high predictive power and the Jackknife test results, the bioclimatic variables BIO19, BIO6, and BIO4 contribute the most to the model. Our prediction model predicts that the distribution area of stone pine will decrease, shifting northward and towards higher altitudes. We believe that this will lead to increased risk of forest fires, loss of ecosystem services, and reduced income from stone pine. For these reasons, benefit from stone pine need to take into account the effects of climate change in their land use planning and give importance to climate change adaptation efforts. These maps, created with current and future predictions of potential habitat distribution, can be use in afforestation, ecological restoration, rural development, conservation, and all kinds of land use studies.
... The aggregation of the four models was employed to define the "Brazil" column, effectively depicting locations where a minimum of one species could potentially be located (Rouget et al. 2004;Fonseca et al. 2006;West et al. 2016), crayfish (Yiwen et al. 2016) and birds (Silva et al. 2018;Strubbe and Matthysen 2014). Almost all studies have relied on the concept of niche conservatism, which is the tendency of current species to retain the ecological characteristics of their ancestral lineage (Wiens and Graham 2005). ...
Global pet trade demand has led to the introduction of large constrictor snakes into new environments either intentionally or accidentally. Brazil has the third-highest reptile species diversity globally, with snakes representing the predominant reptilian group, including 12 constrictor species. The potential for competition and predation between invasive and native snakes underscores the need for comprehensive assessment of the risks faced by endemic herpetofauna. This study aimed to identify potential areas for the establishment of invasive Python and assess their impact on native constrictors in Brazil. Environmental Niche Models were employed to predict suitable habitats for invasive pythons and the four endemic Brazilian snake species. By overlapping Python spp. records with those of endemic serpents, this study sought to understand the resource availability for potentially invasive species and the vulnerability of native species to Python invasion. These results highlight Python sebae and Python bivittatus as potentially invasive species that threaten native constrictors. Conversely, Eunectes murinus, with its semi-aquatic behavior, exhibited lower vulnerability. Endemic serpents, including Boa constrictor, Corallus hortulanus, and Epicrates cenchria, were identified as being highly susceptible to potential competition from invasive pythons. These findings emphasize the importance of understanding the potential ecological impacts of introducing invasive species into native ecosystems.
... We performed a correlation analysis of environmental variables using ENMTools [16]. When the absolute value of the correlation coefficient between two variables surpassed 0.7, we excluded the less influential variable, as indicated by Jackknife test results [17,18]. ...
The research on the significant toxic weed Oxytropis glabra, which adversely affects the grazing industry and the ecological integrity of natural grasslands in the arid and semi-arid regions of northern China, aims to delineate its potential distribution amidst changing climate conditions. This analysis involves both current conditions (1970–2000) and future projections (2050s and 2070s) under four climate scenarios using an R-optimized MaxEnt model. The results indicate that the distribution of O. glabra was primarily influenced by the temperature of the coldest quarter (bio11, ranging from −12.04 to −0.07 °C), precipitation of the coldest quarter (bio19, 0 to 15.17 mm), and precipitation of the warmest quarter (bio18, 0 to 269.50 mm). Currently, the weed predominantly occupies parts of Xinjiang, Inner Mongolia, Gansu, Qinghai, Ningxia, and Tibet. Projections indicate that, across four future climate scenarios, the area of suitable habitats for O. glabra is expected to expand and shift toward higher latitudes and elevations. The research provides valuable information and a theoretical foundation for the management of O. glabra, alongside advancing grassland ecological research and grazing practices.
... The prioritization of environmental factors controls the complexity of models used for climate sensitivity evaluations [56]. Therefore, the minimization of collinearity [57], sample bias [58], and the improvement of variable selection [59] are desirable. ...
The high mountain forests on Earth are characterized by sharp environmental heterogeneity, high species endemism, and unique phenotypic adaptations. Yet, global warming is jeopardizing this ecosystem as no other as some forests may have nowhere to go beyond the mountain’s summits. One of the most human-endangered high-elevation ecosystems is the Polylepis forest. Despite its vulnerability, forecasting climate reactions in this distinctive high mountain forest type remains a formidable challenge. Therefore, in this study, we modeled climate change responses of high-elevation allopatric Polylepis sericea and P. quadrijuga (Rosaceae) in the northern Andes. The analysis took into account VIF-prioritized bioclimatic variables for near-to-present and future (2081–2100 MPI-ESM1-2-HR) conditions. Model selection was carried out following the AUC scores of 12 MaxEnt and six machine learning algorithms. Predictive power reached 0.97 for MaxEnt’s model fc.H_rm.1 (H-1). Results indicate a moderate vulnerability of P. sericea, with a 29% loss of area in the trailing edge, due to climate change for the period 2081–2100. On the other hand, P. quadrijuga is likely to experience even larger distribution losses, up to 99%, for the same period. The findings of this study suggest that P. quadrijuga, as compared to P. sericea, exhibits a restricted ability to maintain the corresponding habitat requirements in the face of climatic change. Higher niche specialization of P. quadrijuga in the environmentally heterogeneous Eastern Cordillera contrasts with the more generalist nature of P. sericea in the topographically less complex Central Cordillera. In addition to climate change, this trend may be exacerbated by the detrimental effects of agriculture, mining, and an expanding rural population, which represent significant human-driven pressures yet to be modeled in the northern Andean highlands. Yet, based on previous studies, the historical population dynamics during the past glacial cycles suggests that range shifts could play a more significant role at deeper time scales than previously forecasted in the species’ reaction to climate change. Additionally, Polylepis forests may be able to endure at the current locations via local adaptation and plasticity, major drivers of the phenotypic variation in long-lived trees, counteracting the vulnerability scenario under a niche conservatism hypothesis. Ultimately, the modeling procedure employed in this study provides insights into the potential effects of climate change on Polylepis forests 70 years from now. Oncoming studies may consider alternative responses inherent to the gene pool of the species and the interaction with edaphic and biotic agents. We advocate for the application of comparable estimations in other high-elevation tree communities found at the tree line.
... All eight environmental variables were retained, while only four bioclimatic variables (i.e., Bio4, Bio15, Bio24, Bio25) remained after conducting multicollinearity testing. During the model formation process, the number of predictor variables was reduced(Breiner et al., 2015) by iteratively removing the least contributing variables to mitigate overfitting(Zeng et al., 2016). This was achieved through scrutinisation of the variable importance scores to identify those with minimal impact, and they were designated for removal. ...
Honey bees play a vital role in providing essential ecosystem services and contributing to global agriculture. However, the potential effect of climate change on honey bee distribution is still not well understood. This study aims to identify the most influential bioclimatic and environmental variables, assess their impact on honey bee distribution, and predict future distribution. An ensemble modelling approach using the biomod2 package in R was employed to develop three models: a climate‐only model, an environment‐only model, and a combined climate and environment model. By utilising bioclimatic data (radiation of the wettest and driest quarters and temperature seasonality) from 1990 to 2009, combined with observed honey bee presence and pseudo absence data, this model predicted suitable locations for honey bee apiaries for two future time spans: 2020–2039 and 2060–2079. The climate‐only model exhibited a true skill statistic (TSS) value of 0.85, underscoring the pivotal role of radiation and temperature seasonality in shaping honey bee distribution. The environment‐only model, incorporating proximity to floral resources, foliage projective cover, and elevation, demonstrated strong predictive performance, with a TSS of 0.88, emphasising the significance of environmental variables in determining habitat suitability for honey bees. The combined model had a higher TSS of 0.96, indicating that the combination of climate and environmental variables enhances the model's performance. By the 2020–2039 period, approximately 88% of highly suitable habitats for honey bees are projected to transition from their current state to become moderate (14.84%) to marginally suitable (13.46%) areas. Predictions for the 2060–2079 period reveal a concerning trend: 100% of highly suitable land transitions into moderately (0.54%), marginally (17.56%), or not suitable areas (81.9%) for honey bees. These results emphasise the critical need for targeted conservation efforts and the implementation of policies aimed at safeguarding honey bees and the vital apiary industry.
... A total of 500 iterations and 10,000 background points were specified. We employed the jackknife test to assess the contribution of each environmental variable [30][31][32][33]. ...
Phytolacca americana, introduced to China in the 20th century for its medicinal properties, has posed a significant ecological and agricultural challenge. Its prolific fruit production, high reproductive coefficient, adaptability, and toxic roots and fruits have led to the formation of monoculture communities, reducing native species diversity and posing threats to agriculture, human and animal health, and local ecosystems. Understanding its potential distribution patterns at a regional scale and its response to climate change is essential for effective monitoring, management, and control. In this study, we utilized the Maxent model to simulate potential habitat areas of P. americana across three timeframes (current, 2050s, and 2070s) under three climate change scenarios (SSP126, SSP245, and SSP585). Leveraging data from 556 P. americana sites across China, we employed ROC curves to assess the prediction accuracy. Our findings highlight key environmental factors influencing P. americana’s geographical distribution, including the driest month’s precipitation, the coldest month’s minimum temperature, the wettest month’s precipitation, isothermality, and temperature annual range. Under current climate conditions, P. americana potentially inhabits 280.26 × 10⁴ km² in China, with a concentration in 27 provinces and cities within the Yangtze River basin and its southern regions. While future climate change scenarios do not drastically alter the total suitable area, the proportions of high and low-suitability areas decrease over time, shifting towards moderate suitability. Specifically, in the SSP126 scenario, the centroid of the predicted suitable area shifts northeastward and then southwestward. In contrast, in the SSP245 and SSP585 scenarios, the centroid shifts northward.
... It is therefore recommended to explore and experiment with Maxent hyperparameters (Merow et al. 2013;Morales et al. 2017;Wiltshire and Tanner 2020) to yield more parsimonious (i.e., simpler, better fitted, and more accurate) SDMs. Although proven to aid in addressing the challenges imposed by hyperdimensional low-resolution predictors and limited occurrence records on model accuracy, model optimization is rarely performed due to its laborious and time-consuming nature (Phillips and Dudík 2008;Zeng et al. 2016). Developing a pipeline for model optimization addresses this issue; however, existing literature only caters to species with sufficient occurrence records (see Zhang et al. 2019;Li et al. 2020;Tang et al. 2021;Yan et al. 2021;Tesfamariam et al. 2022). ...
The upsurge in technical and epidemiological research employing Maximum Entropy (Maxent) establishes this machine-learning algorithm for species distribution modeling (SDM). Although Maxent robustly and accurately predicts the potential distribution of various species in different environments, data quality and varying hyperparameters influence its predictions. Optimizing hyperparameters can compensate for the rigidity of data quality. Addressing this caveat of Maxent, a bipartite approach (tuning and fine-tuning) in increasing model parsimony was developed to optimize the pipeline for range prediction of vectors with limited occurrence records in the Philippines. Tuned models reveal the influence of predictor collinearity on model accuracy, with a Pearson correlation threshold of 0.7 yielding the highest Area Under the Receiving Operator Characteristic Curve (AUC) score, analogous to popularly used methods in SDM. Fine-tuned models show that, contrary to the conventional pipeline, ΔAICc values approaching but not equal to zero produce a combination of hyperparameters (feature classes and regularization multiplier) leading to higher AUC scores. Fine-tuned models are more parsimonious and portray wider distributions than the a priori models generated using the default Maxent settings. This study integrates the best approaches to advance the conventional pipeline for Maxent modeling, substantiating the call for intensive surveying of vectors in a data-poor and high-burden country.
... The ENM effectiveness depends on correctly identifying the most complete set of environmental variables constituting the vector space dimensions [33]. Environmental variable selection for ENMs is frequently conducted through statistical analysis [41][42][43][44][45] or machine learning models [46][47][48][49][50][51]. Generally, an ENM uses statistical analysis, machine learning, or expert-defined rules to estimate a function relating an ecological entity (e.g. a species, community, or ecosystem) with specific environmental conditions defined on a set of environmental variables. ...
Integrated Environmental Assessment systems and ecosystem models study the links between anthropogenic and climatic pressures on marine ecosystems and help understand how to manage the effects of the unsustainable exploitation of ocean resources. However, these models have long implementation times, data and model interoperability issues and require heterogeneous competencies. Therefore, they would benefit from simplification, automatisation, and enhanced integrability of the underlying models. Artificial Intelligence can help overcome several limitations by speeding up the modelling of crucial functional parts, e.g. estimating the environmental conditions fostering a species’ persistence and proliferation in an area (the species’ ecological niche) and, consequently, its geographical distribution. This paper presents a full-automatic workflow to estimate species’ distributions through statistical and machine learning models. It embeds four ecological niche models with complementary approaches, i.e. Artificial Neural Networks, Maximum Entropy, Support Vector Machines, and AquaMaps. It automatically estimates the optimal model parametrisations and decision thresholds to distinguish between suitable- and unsuitable-habitat locations and combines the models within one ensemble model. Finally, it combines several ensemble models to produce a species richness map (biodiversity index). The software is open-source, Open Science compliant, and available as a Web Processing Service-standardised cloud computing service that enhances efficiency, integrability, cross-domain reusability, and experimental reproduction and repetition. We first assess workflow stability and sensitivity and then demonstrate effectiveness by producing a biodiversity index for the Mediterranean based on ∼1500 species data. Moreover, we predict the spread of the invasive Siganus rivulatus in the Mediterranean and its current and future overlap with the native Sarpa salpa under different climate change scenarios.
... However, to avoid the generation of misleading results, meticulous optimization is required. By adjusting the combination of feature class parameters (FC) and the regularization multiplier (RM), the model can be customized according to the specific circumstances of the species, thereby enhancing the model's accuracy and flexibility [66,[69][70][71]. ...
Simple Summary
Biodiversity conservation under global climate change is a global challenge. Many wildlife species are altering their distributions to adapt to the changes in their living environments caused by climate change. Amphibians are the most vulnerable group among vertebrates in the context of global climate change, and salamanders are one of the most vulnerable groups among amphibians. The primary threat to salamanders is habitat loss. There have been multiple studies on the potential distribution prediction of umbrella species (umbrella species are those species whose conservation is expected to provide protection for a wide range of co-occurring species due to their extensive habitat requirements), such as the Chinese giant salamander (Andrias davidianus). However, research on rare small salamanders is urgently needed for their conservation under climate change conditions. This study employed an optimized MaxEnt model to predict and analyze the potential distribution and trends of three rare salamanders from Chongqing, China, based on data collected from field surveys, museum collections, and the existing literature. This study assesses the impact of various environmental factors in the context of climate change on three salamanders species with different habitat preferences in Chongqing. It offers implications for the conservation of these salamanders.
Abstract
The Wushan Salamander (Liua shihi), Jinfo Salamander (Pseudohynobius jinfo), and Wenxian Knobby Salamander (Tylototriton wenxianensis) are rare national Class II protected wild animals in China. We performed MaxEnt modeling to predict and analyze the potential distribution and trends of these species in Chongqing under current and future climate conditions. Species distribution data were primarily obtained from field surveys, supplemented by museum collections and the existing literature. These efforts yielded 636 records, including 43 for P. jinfo, 23 for T. wenxianensis, and 570 for L. shihi. Duplicate records within the same 100 m × 100 m grid cell were removed using ENMTools, resulting in 10, 12, and 58 valid distribution points for P. jinfo, T. wenxianensis, and L. shihi, respectively. The optimization of feature class parameters (FC) and the regularization multiplier (RM) were applied using R package “ENMeval 2.0” to establish the optimal model with MaxEnt. The refined models were applied to simulate the suitable distribution areas for the three species. The results indicate that the current suitable habitat area for L. shihi accounted for 9.72% of the whole region of the Chongqing municipality. It is projected that, by 2050, the proportion of suitable habitat will increase to 12.54% but will decrease to 11.98% by 2070 and further decline to 8.80% by 2090. The current suitable habitat area for P. jinfo accounted for 1.08% of the whole region of the Chongqing municipality, which is expected to decrease to 0.31%% by 2050, 0.20% by 2070, and 0.07% by 2090. The current suitable habitat area for T. wenxianensis accounted for 0.81% of the whole region of the Chongqing municipality, which is anticipated to decrease to 0.37% by 2050, 0.21% by 2070, and 0.06% by 2090. Human disturbance, climate variables, and habitat characteristics are the primary factors influencing the distribution of three salamander species in Chongqing. The proximity to roads significantly impacts L. shihi, while climate conditions mainly affect P. jinfo, and the distance to water sources is crucial for T. wenxianensis. The following suggestions were made based on key variables identified for each species: (1) For L. shihi, it is imperative to minimize human disturbances and preserve areas without roads and the existing vegetation within nature reserves to ensure their continued existence. (2) For P. jinfo, the conservation of high-altitude habitats is of utmost importance, along with the reduction in disturbances caused by roads to maintain the species’ ecological niche. (3) For T. wenxianensis, the protection of aquatic habitats is crucial. Additionally, efforts to mitigate the impacts of road construction and enhance public awareness are essential for the preservation of this species and the connectivity of its habitats.
... We performed the multicollinearity test for the selected environmental variables and avoided the environmental variables with correlation coefficients >0.8 and variance inflation factor (VIF) >5, which helped to prevent our model overfitting. Finally, 16 variables were retained as predictor variables in habitat suitability modeling for HMD as suggested byZeng et al. (2016). ...
Himalayan Musk deer, Moschus chrysogaster is widely distributed but one of the least studied species in Nepal. In this study, we compiled a total of 429 current presence points of direct observation of the species, pellets droppings, and hoofmarks based on field-based surveys during 2018-2021 and periodic data held by the Department of National Park and Wildlife Conservation. We developed the species distribution model using an ensemble modeling approach. We used a combination of bioclimatic, anthropogenic, topographic, and vegetation-related variables to predict the current suitable habitat for Himalayan Musk deer in Nepal. A total of 16 predictor variables were used for habitat suitability modeling after the multicollinearity test. The study shows that the 6973.76 km 2 (5%) area of Nepal is highly suitable and 8387.11 km 2 (6%) is moderately suitable for HMD. The distribution of HMD shows mainly by precipitation seasonality, precipitation of the warmest quarter, temperature ranges, distance to water bodies, anthropogenic variables, and land use and land cover change (LULC). The probability of occurrence is less in habitats with low forest cover. The response curves indicate that the probability of occurrence of HMD decreases with an increase in precipitation seasonality and remains constant with an increase in precipitation of the warmest quarter. Thus, the fortune of the species distribution will be limited by anthropogenic factors like poaching, hunting, habitat fragmentation and habitat degradation , and long-term forces of climate change. K E Y W O R D S anthropogenic variables, climate change impacts, habitat suitability modeling, Himalaya, predictive performance, species distribution models
... In the future, more factors affecting species distribution should be considered. Finally, although the MaxEnt model is widely used, its accuracy is still not 100%, although recently more research has been done to improve model accuracy [63,64]. There are many factors affecting species distribution, which are mutual and dynamic, so more factors should be added to predict species distribution in the future. ...
Comprehending invasive alien species’ potential habitat area and spread trend are of great importance for effective prevention and control strategy and prevention of spread. However, previous studies have mainly been based on large regional scales (national or global level). Research on the smaller regional scale of ecologically fragile karst makes the prevention and control measures more feasible and targeted. For invasive Solidago canadensis, based on an MaxEnt model and ArcGIS, we determined its current and future potential distributions. The main drivers of S. canadensis distribution were precipitation changes and human activities. S. canadensis occurrence probability initially increased, and then decreased with increased precipitation variables, and increased rapidly initially, then gradually with increasing human footprint. Under current climate conditions, S. canadensis suitable area is 8.13×104 km2, with the highly suitable area concentrated in Guiyang, east of Bijie, Zunyi, Anshun and Duyun. Under climate conditions of the 2050s, the suitable area drops slightly to 8.00×104 km2. Under climate conditions of the 2070s, the suitable area expands to 8.31×104 km2. And move toward the south within the study
area. Based on the modelling and space optimization software ZONATION key monitoring area covers 79,857 km2, including a primary monitoring area mainly distributed in Guiyang, east of Bijie, northeast of Anshun and northwest of Duyun, a secondary monitoring area mainly outside the primary monitoring area, and a third-level monitoring area widely distributed in Zunyi, Tongren, Duyun, west of Kaili and Anshun and east of Bijie. Linking our results with this specie’s invasive power, we thus recommended to increase the prevention and control sites in the first-level monitoring area, and continue to pay attention to the risk of the southward spread of this species.
... The spatial resolution of the data used in this study was 2.5 arcminute (approximately 4.5 km 2 ). In order to improve the accuracy of the model, it is necessary to exclude environmental variables with small contributions (Banerjee et al., 2019;Jiang et al., 2019;Zeng et al., 2016). Thus, we used the jackknife test in MaxEnt v3.4.4 (https:// biodi versi tyinf ormat ics. ...
Graphium sarpedon is a significant foliar pest of Laurel plants in China. In this study, the MaxEnt model was used to investigate the distribution of G. sarpedon and predict its potential distribution areas in China in the future (2050s and 2090s) based on three Shared Socioeconomic Pathways (SSP1‐2.6, SSP2‐4.5, and SSP5‐8.5), and key environmental variables affecting its distribution were identified. The results showed that under the current climatic conditions, the suitable distribution areas of G. sarpedon were 92.17°–134.96° E and 18.04°–33.61° N, including Yangtze Plain (Middle and Lower), Pearl River Delta, Yangtze River Delta, and Lingnan areas. Under the future climate conditions, the total suitable distribution area of G. sarpedon decreased, but the area of medium suitable area increased. The study identified 11 key environmental variables affecting the distribution of G. sarpedon , the most critical of which was Precipitation of Warmest Quarter (bio18) and precipitation in April, May, June, and September (prec4, prec5, prec6, and prec9). This study is beneficial for monitoring and preventing the possible changes of G. sarpedon and provides theoretical references for its prevention and control.
... In addition to modeling the habitat suitability for the Persian squirrel, we harnessed the predictive capabilities inherent in SDMs to assess the forest mortality risk. Among various SDMs, maximum entropy (MaxEnt) stands out as the most widely employed and recognized model due to its flexibility and ease of implementation [47][48][49][50]. Nonetheless, given the diversity in model performance [51], our approach initially involved applying 15 distinct models (Table 2) using the sdm and dismo packages in the R statistical software version 4.3.1 [52,53]. ...
The Zagros forests in Iran are currently experiencing an exacerbation of climate-induced mortality, placing the Persian squirrel, a keystone species reliant on these ecosystems, in jeopardy. Addressing this imminent threat, our research employed a spatial prioritization methodology, integrating assessments of habitat suitability and mortality risk. Utilizing a weighted ensemble approach, incorporating the strengths of diverse models and expert rules, we discerned that approximately 62% of surveyed forests are at risk, with 7% classified as high risk and 17% as very high risk. Notably, 83% of the forests exhibited varying degrees of habitat suitability, with 11% and 12% demonstrating high and very high suitability, respectively. Employing a conservation prioritization framework, we systematically categorized habitats into priority classes, with 8%, 17%, 29%, and 46% assigned to very high, high, moderate, and low conservation priority classes, respectively. Significantly, areas classified as very high priority demand immediate restoration efforts due to ongoing mortality, while other priority classes underscore the importance of protection and prevention in unaffected habitats. Acknowledging the irreversible nature of current climatic conditions in the Zagros forests, our strategic emphasis aligns with conservation triage principles, prioritizing the preservation of intact habitats yet to succumb to irreversible mortality.
Climate change, driven by both natural and human influences, presents serious threats to biodiversity and the distribution of species over the coming decades. To examine the effects of climate change on species distribution and habitat suitability, researchers have widely employed Species Distribution Models (SDMs). This study specifically investigated Zataria multiflora, a valuable medicinal plant found in Iranian rangelands. The aim was to identify the critical factors influencing the distribution of Z. multiflora and assess how climate change may affect its range. The researchers utilized the Maximum Entropy Model (MaxEnt), incorporating eight environmental variables: maximum temperature of the warmest month, minimum temperature of the coldest month, annual temperature range, precipitation of the driest quarter, precipitation of the coldest quarter, depth to bedrock, slope, and solar radiation. The results indicated that solar radiation, slope, precipitation during the warmest quarter (bio17), and maximum temperature of the warmest month (bio5) were the most significant factors affecting the species' distribution. Projections suggest a potential decrease in the species' range by 43.98%, 51.60%, 54.20%, and 68.10% under RCPs 4.5-2050, 8.5-2050, 4.5-2070, and 8.5-2070, respectively. These insights can assist decision-makers in developing effective conservation strategies to protect and sustainably manage Z. multiflora in light of global climate change. Given the expected habitat loss, in-situ conservation efforts should be prioritized for the protection of Z. multiflora.
Understanding the effects of climate change on species distribution patterns is fundamental to biogeography and evolutionary ecology, especially for socio-economically important agroforestry species such as Parkia biglobosa R.Br. ex Benth and Vitellaria paradoxa CF Gaertn. However, gaps remain regarding the evolution of ecological niches of these widespread species. This study examines the effects of climate change on P. biglobosa and V. paradoxa, two key species of agroforestry parklands in Benin. By integrating intraspecific variation into niche modeling, three homogeneous subgroups (North, Center, and South) were identified and analyzed under current and future climate conditions. Maximum Entropy Modeling (MaxEnt), was used, chosen for its predictive power and flexibility in correcting for data bias. Climate projections were based on common socio-economic scenarios (SSP2-4.5 and SSP5-8.5), representing a moderate mitigation scenario and a high emissions scenario, respectively. The results showed significant climate change effects across subgroups for both species, mediated by temperature and humidity. Under the SSP2-4.5 scenario, the northern subgroup of P. biglobosa exhibited habitat shifts towards the central and southern subgroup regions, while the central and southern subgroups showed predictive habitat loss. For V. paradoxa, this scenario predicted significant habitat loss for the northern and central subgroups, while the southern subgroup demonstrated habitat expansion along altitudinal gradients. The more severe SSP-5–8.5 scenario predicted extensive habitat loss for both species. The Sudano-Guinean and Guineo-Congo zones remain climatically suitable for both species under future projections. The most pronounced interspecific niche overlap occurred between the northern and southern subgroups, indicating considerable ecological similarity. These results highlight the need for climate-adaptive management strategies in agroforestry systems with particular emphasis on subgroups facing imminent habitat loss. Future research directions should explore adaptive management approaches while maintaining essential ecosystem services and supporting local livelihoods.
Cacopsylla chinensis is an oligophagous pest and has become one of the main pests that cause yield loss in commercial pear orchards in China. Predicting the impact of climate change on the distribution range of C. chinensis is crucial for its effectively preventing and managing. In this study, we collected 102 geographic distribution information of C. chinensis with 8 selected crucial environmental variables to simulate its potential suitable habitats. On this basis, the parameter-optimized maximum entropy model was utilized to predict the potential effect of future climate variation on its distribution, considering various socio-economic pathway scenarios and 3 Earth system models. The findings showed that the current total potential suitable area for C. chinensis was 578.29 × 104 km2, which accounts for 60.24% of China’s territory. In the total area, the suitability areas of low, medium, and high were 308.21 × 104 km2, 118.50 × 104 km2, and 151.58 × 104 km2, respectively. Among them, the high suitability areas are mainly distributed in Anhui, Beijing, Chongqing, Hebei, Henan, Hubei, Jiangsu, Liaoning, Shandong, Shanxi, Shaanxi, Sichuan, and Tianjin. Furthermore, our predictions suggest that the potentially suitable areas for this pest will increase by 8.49–35.02% under various future climate change conditions in China. The findings will be propitious to understand the linkage between C. chinensis niches and the relevant environment. It also provides valuable insights for developing future pest management strategies.
Insect biogeography is poorly documented globally, particularly in the tropics. Recent intensive research in tropical Asia, combined with increasingly available records from citizen science, provides an opportunity to map the distributions of tropical Asian butterflies. We compiled a dataset of 724,247 occurrences of 3,591 tropical Asian butterfly species by aggregating records from GBIF (651,285 records), published literature (21,271), published databases (37,695), and unpublished data (13,993). Here, we present this dataset and single-species distribution maps of 1,520 species. Using these maps, along with records of the 2,071 remaining species, we identified areas of limited sampling (e.g., the Philippines, Myanmar, and New Guinea) and predicted areas of high diversity (Peninsular Malaysia and Borneo). This dataset can be leveraged for a range of studies on Asian and tropical butterflies, including 1) species biogeography, 2) sampling prioritization to fill gaps, 3) biodiversity hotspot mapping, and 4) conservation evaluation and planning. We encourage the continued development of this dataset and the associated code as a tool for the conservation of tropical Asian insects.
The Himalayan region is home to the endangered and endemic medicinal plant, Meconopsis aculeata. Habitat degradation due to overexploitation and unscientific collection of the species in recent years has caused the restricted distribution range of species across its range. Therefore, the present study was conducted in the Hindu Kush Himalayan region (4.5 million km2 area) to study population ecology and identify suitable habitats for the habitat distribution modelling under projected climate change to monitor and restore the dwindling populations. Ten sites/populations (20 × 20 m plot) of M. aculeata have been studied. East aspect (5 sites) and rocky habitat (04 sites) represented the maximum number of sites. The density of M. aculeata has a significant negative correlation with slope and available potassium in soil. The MaxEnt algorithm, species occurrence data and environmental variables were used in the current study. CMIP6 project's MIROC6 and BCC-CSM2-MR climate change models were used to identify the potential distribution area for the future under the SSP245 and SSP370 scenarios for the period 2081–2100. The results showed the satisfactory quality of models based on the True-Skill-Statistic (0.838 ±0.149), Area-Under-Curve values (0.958±0.038), and the Receiver-Operating-Characteristic curves. MaxEnt predicted that 1,78,993 km2 (4.10 %) area is favourable for the occurrence of the species. The precipitation of the driest quarter, precipitation of the coldest quarter and elevation were the most influential variables for habitat suitability prediction. The distribution area with an increase in greenhouse gases shows a decrease in the period 2081–2100 in both SSPs and GCMs. The study predicted highly suitable areas that would play a vital role in in situ conservation and improving the status of species. The predicted distribution maps of the species would help policymakers and decision-makers in developing suitable management plans, which will ultimately help in the conservation of species.
Flood is a widespread, recurring, and damaging natural hazard across the globe. In the Philippines, yearly heavy rains, monsoons, and typhoons cause flooding that affects millions of Filipinos. Mapping flooding hazards, especially in remote areas of the country, can provide essential inputs to manage and reduce the risk and impact of flooding on different infrastructures, properties, and people. Statistical and machine learning approaches have provided a faster way of producing accurate flood hazard maps without numerous data input and computational requirements from physically based hydrological models. This study uses different geo-environmental variables and maximum entropy, a machine learning approach, to map flood hazards in Marinduque, Philippines. Results showed that elevation, soil texture, drainage density, slope, and annual rainfall are the main predictors of flooding in the province. Also, approximately 4512 hectares in Marinduque were under moderate to very high flooding hazards.
Gonimbrasia belina , known as the mopane worm, is a large edible caterpillar in tropical and subtropical regions. However, little is known about the bioecology of this species as influenced by its host trees. This study evaluated the importance of different potential host trees in understanding mopane worms’ behaviour and spatial distribution. To assess their relative importance, the study compared models incorporating various mopane worm host trees and predictor variables. Using the species distribution modelling (SDM) package in R, an ensemble of random forest (RF), support vector machine (SVM), and boosted regression tree (BRT) algorithms were used to assess the spatial extent of mopane worm distribution in Southern Africa. Four host tree-based scenarios were developed to assess their contribution to the relative distribution of the mopane worm i.e. (1) by excluding all the potential host trees as explanatory variables and considering only the environmental variables, (2) focusing on the primary host tree, Colophospermum mopane as an explanatory variable together with the other environmental variables, (3) incorporating all the host trees, including C. mopane and (4) examining all other host trees excluding C. mopane . Results demonstrated that incorporating all host trees enhanced the models’ predictive abilities (mean AUC = 0.87) underscoring the significant impact of the alternative host trees on the mopane worm distribution patterns beyond just the C. mopane . This study highlights the significance of host trees in predicting the behaviour and distribution of mopane worm populations, providing valuable insights and decision-making for mopane worm use as an alternative protein source, conservation efforts, and land management practices.
Maximum entropy modeling (Maxent) is a widely used algorithm for predicting species distributions across space and time. Properly assessing the uncertainty in such predictions is non-trivial and requires validation with independent datasets. Notably, model complexity (number of model parameters) remains a major concern in relation to overfitting and, hence, transferability of Maxent models. An emerging approach is to validate the cross-temporal transferability of model predictions using paleoecological data. In this study, we assess the effect of model complexity on the performance of Maxent projections across time using two European plant species (Alnus glutinosa (L.) Gaertn. and Corylus avellana L.) with an extensive late Quaternary fossil record in Spain as a study case. We fit 110 models with different levels of complexity under present time and tested model performance using AUC (area under the receiver operating characteristic curve) and AICc (corrected Akaike Information Criterion) through the standard procedure of randomly partitioning current occurrence data. We then compared these results to an independent validation by projecting the models to mid-Holocene (6000 years before present) climatic conditions in Spain to assess their ability to predict fossil pollen presence–absence and abundance. We find that calibrating Maxent models with default settings result in the generation of overly complex models. While model performance increased with model complexity when predicting current distributions , it was higher with intermediate complexity when predicting mid-Holocene distributions. Hence, models of intermediate complexity resulted in the best trade-off to predict species distributions across time. Reliable temporal model transferability is especially relevant for forecasting species distributions under future climate change. Consequently, species-specific model tuning should be used to find the best modeling settings to control for complexity, notably with paleoecological data to independently validate model projections. For cross-temporal projections of species distributions for which paleoecological data is not available, models of intermediate complexity should be selected.
Global land-cover data are widely used in regional and global models because land cover influences land atmosphere exchanges of water, energy, momentum, and carbon. Many models use data of maximum green vegetation fraction (MGVF) to describe vegetation abundance. MGVF products have been created in the past using different methods, but their validation with ground sites is difficult. Furthermore, uncertainty is introduced because many products use a single year of satellite data. In this study, a global 1-km MGVF product is developed on the basis of a "climatology" of data of Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index and land-cover type, which removes biases associated with unusual greenness and inaccurate land-cover classification for individual years. MGVF shows maximum annual variability from 2001 to 2012 for intermediate values of average MGVF, and the standard deviation of MGVF normalized by its mean value decreases nearly monotonically as MGVF increases. In addition, there are substantial differences between this climatology and MGVF data from the MODIS Continuous Fields (CF) Collection 3, which is currently used in the Community Land Model. Although the CF data only use 2001 MODIS data, many of these differences cannot be explained by usage of different years of data. In particular, MGVF as based on CF data is usually higher than that based on the MODIS climatology from this paper. It is difficult to judge which product is more realistic because of a lack of ground truth, but this new MGVF product is more consistent than the CF data with the MODIS leaf area index product (which is also used to describe vegetation abundance in models).
The genetic variation among nine populations of the red swamp crayfish Procambarus clarkii was examined using partial sequences of the mitochondrial COI gene. Three populations (Illinois and Louisiana, United States and northern Coahuila, Mexico) represented the native range and six populations came from areas where the species has been introduced (central Coahuila, southern Nuevo León, Durango, Chihuahua and Chiapas, Mexico, and Cartago, Costa Rica). A 689 bp fragment was amplified from 37 samples. Uncorrected genetic distances among sequences were p = 0 to 0.02031 and 12 haplotypes were found. A phylogenetic reconstruction shows that the three populations from the native range remain very similar to each other and some introduced populations can be directly associated to one of them. The populations from Nuevo León, central Coahuila and Costa Rica were the most divergent ones. Overall the genetic variation found in P. clarkii in both native and introduced populations is low.
Recent studies have demonstrated a need for increased rigour in building and evaluating ecological niche models ( ENM s) based on presence‐only occurrence data. Two major goals are to balance goodness‐of‐fit with model complexity (e.g. by ‘tuning’ model settings) and to evaluate models with spatially independent data. These issues are especially critical for data sets suffering from sampling bias, and for studies that require transferring models across space or time (e.g. responses to climate change or spread of invasive species). Efficient implementation of procedures to accomplish these goals, however, requires automation.
We developed ENM eval , an R package that: (i) creates data sets for k ‐fold cross‐validation using one of several methods for partitioning occurrence data (including options for spatially independent partitions), (ii) builds a series of candidate models using Maxent with a variety of user‐defined settings and (iii) provides multiple evaluation metrics to aid in selecting optimal model settings. The six methods for partitioning data are n −1 jackknife, random k ‐folds ( = bins), user‐specified folds and three methods of masked geographically structured folds. ENM eval quantifies six evaluation metrics: the area under the curve of the receiver‐operating characteristic plot for test localities ( AUC TEST ), the difference between training and testing AUC ( AUC DIFF ), two different threshold‐based omission rates for test localities and the Akaike information criterion corrected for small sample sizes ( AIC c).
We demonstrate ENM eval by tuning model settings for eight tree species of the genus Coccoloba in Puerto Rico based on AIC c. Evaluation metrics varied substantially across model settings, and models selected with AIC c differed from default ones.
In summary, ENMeval facilitates the production of better ENM s and should promote future methodological research on many outstanding issues.
The Australian redclaw crayfish, Cherax quadricarinatus, was introduced to Israel from the U.S.A. in the early 1990s by the Department of Fisheries, Ministry of Agriculture, with no assessment of its potential to establish feral populations. It has been raised in agricultural research and academic facilities and in commercial farms. A warning against its introduction to the temperate part of the country has been ignored. We report the first appearance of the redclaw in the Sea of Galilee, and discuss possible vectors.
Crop damages by wildlife is a frequent form of human-wildlife conflict. Identifying areas where the risk of crop damages is highest is pivotal to set up preventive measures and reduce conflict. Species distribution models are routinely used to predict species distribution in response of environmental changes. The aim of this paper was assessing whether species distribution models can allow to identify the areas most at risk of crop damages, helping to set up management strategies aimed at the mitigation of human-wildlife conflicts. We obtained data on wild boar Sus scrofa damages to crops in the Alta Murgia National Park,
Southern Italy, and related them to landscape features, to identify areas where the risk of wild boar damages is highest. We used MaxEnt to build species distribution models. We identified the
spatial scale at which landscape mostly affects the distribution damages, and optimized the regularization parameter of models, through an information-theoretic approach based on AIC. Wild
boar damages quickly increased in the period 2007-2011; cereals and legumes were the crops more affected. Large areas of the park have a high risk of wild boar damages. The risk of damage
s was related to low cover of urban areas or olive grows, intermediate values of forest cover, and high values of shrubland cover within a 2-km radius. Temporally independent validation data
demonstrated that models can successfully predict damages in the future. Species distribution models can accurately identify the areas most at risk of wildlife damages, as
models calibrated on data collected during only a subset of years correctly predicted damages in the subsequent year
MAXENT is now a common species distribution modeling (SDM) tool used by conservation practitioners for predicting the distribution of a species from a set of records and environmental predictors. However, datasets of species occurrence used to train the model are often biased in the geographical space because of unequal sampling effort across the study area. This bias may be a source of strong inaccuracy in the resulting model and could lead to incorrect predictions. Although a number of sampling bias correction methods have been proposed, there is no consensual guideline to account for it. We compared here the performance of five methods of bias correction on three datasets of species occurrence: one "virtual" derived from a land cover map, and two actual datasets for a turtle (Chrysemys picta) and a salamander (Plethodon cylindraceus). We subjected these datasets to four types of sampling biases corresponding to potential types of empirical biases. We applied five correction methods to the biased samples and compared the outputs of distribution models to unbiased datasets to assess the overall correction performance of each method. The results revealed that the ability of methods to correct the initial sampling bias varied greatly depending on bias type, bias intensity and species. However, the simple systematic sampling of records consistently ranked among the best performing across the range of conditions tested, whereas other methods performed more poorly in most cases. The strong effect of initial conditions on correction performance highlights the need for further research to develop a step-by-step guideline to account for sampling bias. However, this method seems to be the most efficient in correcting sampling bias and should be advised in most cases.
Dengue fever is a mosquito-borne disease that affects more than 2.5 billion people worldwide. Here, we used the dataset of municipality infestation level from the Brazilian Health Ministry with the aim of building vector distribution models to identify epidemiological hotspots.
Maxent software was used to predict the environmental suitability of the vector under current and 2050 climatic conditions. We built potential risk maps for current and future epidemiological scenarios in order to provide data for vector control planning.
The results showed that the current epidemiological status is critical in the coastal region, with 80% of the population in risk areas and 30% in epidemiological outbreak areas. Our results also suggest that the area covered by the vector distribution in Brazil will decrease in future projections in the north, but will spread to the south.
The results may provide useful information for health agencies and policymakers in focusing efforts in epidemiological hotspots. Therefore, understanding the niche distribution dynamics of Aedes aegypti is an important step towards public health planning for vector control.
AimEcological niche models are increasingly being used to aid in predicting the effects of future climate change on species distributions. Complex models that show high predictive performance on current distribution data may do a poor job of predicting new data due to overfitting. In addition, model performance is often evaluated using techniques that are sensitive to spatial sampling bias. Here, we explore the effects of model complexity and spatial sampling bias on niche models for 90 vertebrate taxa of conservation concern. LocationCalifornia, USA. Methods
We used Akaike information criterion (AICc) to select variables and tune Maxent's built-in regularization parameter (β) to constrain model complexity. In addition, we incorporated several estimates of spatial sampling bias based on interpolations of target group data. Ensemble forecasts were developed for future conditions from two emission scenarios and three climate change models for the year 2050. ResultsReducing the number of predictors and tuning β resulted in a reduction in the number of parameters in models built with sample sizes greater than approximately 10 occurrence points. Reducing the number of predictors had a substantially higher impact on the relative prioritization of different grid cells than did increasing regularization. There was little difference in prioritization of habitat when comparing models built using different spatial sampling bias estimates. Over half of the taxa were predicted to experience >80% reductions in environmental suitability in currently occupied cells, and this pattern was consistent across taxonomic groups. Main Conclusions
Our results demonstrate that reducing the number of correlated predictor variables tends to decrease the breadth of models, while tuning regularization using AICc tends to increase it. These two strategies may provide a reasonable bracketing strategy for assessing climate change impacts.
The American bullfrog (Lithobates catesbeianus) is a native species from eastern North America that was introduced to Ecuador in 1985. We built two models in Maxent, (1) one model with native records and, (2) one model with native and invasive records, to provide an approximate map of the potential geographical distribution for this species in Ecuador. Both models showed significant differences in the prediction of suitable areas, model 2 being the most consistent in relation to occurrence records. Here, we present the invasive potential of the American bullfrog to occupy a wide variety of environments such as Amazonia, if human activities lead to an accidental or induced introduction. Furthermore, this study is the first survey about the distribution of the American bullfrog in Ecuador, thus identifying some susceptible areas where conservation efforts should be focused to prevent new settlements and uncontrolled breeding of this species.
Background/Question/Methods
Maxent, one of the most commonly used methods for inferring species distributions and environmental tolerances from occurrence data, allows users to fit models of arbitrary complexity. Model complexity is typically constrained via a process known as L1 regularization, but at present little guidance is available for setting the appropriate level of regularization, and the effects of inappropriately complex or simple models are largely unknown. In this study, we demonstrate the use of information criterion approaches to setting regularization in Maxent, and compare models selected using information criteria to models selected using other criteria that are common in the literature. We evaluate model performance using occurrence data generated from a known “true” initial Maxent model, using several different metrics for model quality and transferability.
Results/Conclusions
We demonstrate that models that are inappropriately complex or inappropriately simple show reduced ability to infer habitat quality, reduced ability to infer the relative importance of variables in constraining species’ distributions, and reduced transferability to other time periods. We also measure the relative effectiveness of different model selection criteria, and demonstrate that information criteria may offer significant advantages over the AUC-based methods commonly used in the literature.
Whether a species niche is conserved or shifts across space and time is a question of heightened interest in ecology and evolution. Considerable scientific inquiry into this topic has used invasive species to evaluate conservatism of the Grinnellian climatic niche while generally neglecting the Eltonian functional niche. By contrast, we report here on the first simultaneous reciprocal comparison of both the Grinnellian and Eltonian niches for the globally invasive signal crayfish Pacifastacus leniusculus between its native (Pacific Northwest of North America) and non-native ranges (Japan). Using multivariate statistics and the Maxent machine-learning algorithm, we found strong evidence for a climatic niche shift between geographic regions for P. leniusculus. Pacifastacus leniusculus shifted from warmer temperatures with strong precipitation seasonality in western North America to cooler temperatures with less precipitation seasonality in Japan. However, analysis of stable isotopes of carbon and nitrogen revealed conservatism of the functional niche of P. leniusculus between multiple lakes in the native and non-native ranges. We found that trophic position and niche width of P. leniusculus were equivalent between regions, and that niche attributes including reliance on autochthonous or allochthonous energy sources and ontogenetic shifts were comparable regardless of origin. Our finding of functional niche conservatism contrasts with the results of some recent studies evaluating the Grinnellian niche for invasive species, as well as our own climatic niche comparison, and raises the question of whether Grinnellian or Eltonian niches are more prone to shifts or conservatism. We hypothesize that the Grinnellian niche may be more labile than the Eltonian niche in general, but argue that resolving this question will require more reciprocal comparisons of Eltonian niches to keep pace with the recent increase in Grinnellian niche studies.
Aim In response to a recent paper suggesting the failure of ecological niche models to predict between native and introduced distributional areas of fire ants ( Solenopsis invicta ), we sought to assess methodological causes of this failure.
Location Ecological niche models were developed on the species’ native distributional area in South America, and projected globally.
Methods We developed ecological niche models based on six different environmental data sets, and compared their respective abilities to anticipate the North American invasive distributional area of the species.
Results We show that models based on the ‘bioclimatic variables’ of the WorldClim data set indeed fail to predict the full invasive potential of the species, but that models based on four other data sets could predict this potential correctly.
Main conclusions The difference in predictive abilities appears to centre on the complexity of the environmental variables involved. These results emphasize important influences of environmental data sets on the generality and ability of ecological niche models to anticipate novel phenomena, and offer a simpler explanation for the lack of predictive ability among native and invaded distributional areas than that of niche shifts.
Aim
Advancement in ecological methods predicting species distributions is a crucial precondition for deriving sound management actions. Maximum entropy (MaxEnt) models are a popular tool to predict species distributions, as they are considered able to cope well with sparse, irregularly sampled data and minor location errors. Although a fundamental assumption of MaxEnt is that the entire area of interest has been systematically sampled, in practice, MaxEnt models are usually built from occurrence records that are spatially biased towards better‐surveyed areas. Two common, yet not compared, strategies to cope with uneven sampling effort are spatial filtering of occurrence data and background manipulation using environmental data with the same spatial bias as occurrence data. We tested these strategies using simulated data and a recently collated dataset on Malay civet V iverra tangalunga in Borneo.
Location
Borneo, Southeast Asia.
Methods
We collated 504 occurrence records of Malay civets from Borneo of which 291 records were from 2001 to 2011 and used them in the MaxEnt analysis (baseline scenario) together with 25 environmental input variables. We simulated datasets for two virtual species (similar to a range‐restricted highland and a lowland species) using the same number of records for model building. As occurrence records were biased towards north‐eastern Borneo, we investigated the efficacy of spatial filtering versus background manipulation to reduce overprediction or underprediction in specific areas.
Results
Spatial filtering minimized omission errors (false negatives) and commission errors (false positives). We recommend that when sample size is insufficient to allow spatial filtering, manipulation of the background dataset is preferable to not correcting for sampling bias, although predictions were comparatively weak and commission errors increased.
Main Conclusions
We conclude that a substantial improvement in the quality of model predictions can be achieved if uneven sampling effort is taken into account, thereby improving the efficacy of species conservation planning.
Invasive species are one of the most severe threats to biodiversity, and an ability to predict the extent of potential invasions can help conservation strategies. Species distribution models ( SDMs ) have been widely used to project the potential range of invasive species. These models assume that species retain their niche properties during invasion (niche conservatism), although this assumption is seldom verified.
We gathered occurrence records for the crayfish P rocambarus clarkii from the U.S.A. and Mexico (native + invasive ranges) and from the Iberian P eninsula (invasive) to test for niche conservatism across continents using niche overlap metrics (Schoener's D ). To test for differences in the climate space occupied by the species on the different continents, we performed two principal component analyses ( PCA s) on the environmental data extracted from occurrence records: first, separately for each occurrence data set (i.e. each continent) and secondly, using the pooled data. Subsequently, we projected the model to South America, where this species has the potential to become invasive.
Schoener's D showed high overlap (0.68) between the two regions (the A mericas and I beria), and there was no difference between the regions in both PCA s. The crayfish has conserved its niche across continents, and therefore, our model projection to South America may accurately demonstrate where invasion is most likely to occur.
Large parts of South America are apparently suitable, mainly A rgentina, C hile, P araguay, U ruguay and southern B razil. This result is of great concern since this invasive species can spread quickly in suitable areas. Stronger laws and regulations should be made to protect native biodiversity and agricultural land. Our approach could be replicated for the study of invasions by other species where extensive data on the potentially invaded areas are available.
Impacts of nonindigenous crayfishes on ecosystem services exemplify the mixture of positive and negative effects of intentionally introduced species. Global introductions for aquaculture and ornamental purposes have begun to homogenize naturally disjunct global distributions of crayfish families. Negative impacts include the loss of provisioning (e.g., reductions in edible native species, reproductive interference or hybridization with native crayfishes), regulatory (e.g., lethal disease spread, increased costs to agriculture and water management), supporting (e.g., large changes in ecological communities), and cultural (e.g., loss of festivals celebrating native crayfish) services. Where quantification of impacts exists (e.g., Procambarus clarkii and Pacifastacus leniusculus in Europe), regulations now prohibit introduction and spread of crayfishes, indicating that losses of ecosystem services have outweighed gains. Recent research advances such as predicting invasiveness, predicting spread, improved det...
In the present study, the life history and diet of the highly successful
North American invader Orconectes immunis was
assessed for the first time in its introduced European range. In 2007,
O. immunis population dynamics were monitored
in a typical backwater habitat using unbaited funnel traps, and its
life history was analysed using Von Bertalanffy’s growth function.
Juveniles hatched as early as March and may attain sexual maturity at
the end of their first summer. The adult population moulted up to four
times during the summer months, with the non-breeding form (II)
lasting for a remarkably short time period. The high growth rate of
O. immunis was combined with a short longevity,
which was estimated at 2.5 years. The fecundity ranged from 119 to 495
pleopodal eggs. The stomach contents were dominated by detritus,
followed by macroinvertebrates and macrophytes, and no ontogenetic
shift in diet was observed. The ability to prey on a wide array of
invertebrate taxa presumably supports the sustained high growth rate
of O. immunis. The presented data provide evidence
that O. immunis exhibits a strongly r-selected life
history and omnivorous feeding habits. These ecological properties
have often been linked to successful invaders and enhance the
invasiveness of O. immunis.
Aim Techniques that predict species potential distributions by combining observed occurrence records with environmental variables show much potential for application across a range of biogeographical analyses. Some of the most promising applications relate to species for which occurrence records are scarce, due to cryptic habits, locally restricted distributions or low sampling effort. However, the minimum sample sizes required to yield useful predictions remain difficult to determine. Here we developed and tested a novel jackknife validation approach to assess the ability to predict species occurrence when fewer than 25 occurrence records are available.
Location Madagascar.
Methods Models were developed and evaluated for 13 species of secretive leaf‐tailed geckos ( Uroplatus spp.) that are endemic to Madagascar, for which available sample sizes range from 4 to 23 occurrence localities (at 1 km ² grid resolution). Predictions were based on 20 environmental data layers and were generated using two modelling approaches: a method based on the principle of maximum entropy (Maxent) and a genetic algorithm (GARP).
Results We found high success rates and statistical significance in jackknife tests with sample sizes as low as five when the Maxent model was applied. Results for GARP at very low sample sizes (less than c. 10) were less good. When sample sizes were experimentally reduced for those species with the most records, variability among predictions using different combinations of localities demonstrated that models were greatly influenced by exactly which observations were included.
Main conclusions We emphasize that models developed using this approach with small sample sizes should be interpreted as identifying regions that have similar environmental conditions to where the species is known to occur, and not as predicting actual limits to the range of a species. The jackknife validation approach proposed here enables assessment of the predictive ability of models built using very small sample sizes, although use of this test with larger sample sizes may lead to overoptimistic estimates of predictive power. Our analyses demonstrate that geographical predictions developed from small numbers of occurrence records may be of great value, for example in targeting field surveys to accelerate the discovery of unknown populations and species.
We present software that facilitates quantitative comparisons of environmental niche models (ENMs). Our software quantifies similarity of ENMs generated using the program Maxent and uses randomization tests to compare observed similarity to that expected under different null hypotheses. ENMTools is available online free of charge from <http://purl.oclc.org/enmtools>.
1. Species are shifting their ranges at an unprecedented rate through human transportation and environmental change. Correlative species distribution models (SDMs) are frequently applied for predicting potential future distributions of range-shifting species, despite these models’ assumptions that species are at equilibrium with the environments used to train (fit) the models, and that the training data are representative of conditions to which the models are predicted. Here we explore modelling approaches that aim to minimize extrapolation errors and assess predictions against prior biological knowledge. Our aim was to promote methods appropriate to range-shifting species.
Prediction of species’ distributions is central to diverse applications in ecology, evolution and conservation science. There is increasing electronic access to vast sets of occurrence records in museums and herbaria, yet little effective guidance on how best to use this information in the context of numerous approaches for modelling distributions. To meet this need, we compared 16 modelling methods over 226 species from 6 regions of the world, creating the most comprehensive set of model comparisons to date. We used presence-only data to fit models, and independent presence-absence data to evaluate the predictions. Along with well-established modelling methods such as generalised additive models and GARP and BIOCLIM, we explored methods that either have been developed recently or have rarely been applied to modelling species’ distributions. These include machine-learning methods and community models, both of which have features that may make them particularly well suited to noisy or sparse information, as is typical of species’ occurrence data. Presence-only data were effective for modelling species’ distributions for many species and regions. The novel methods consistently outperformed more established methods. The results of our analysis are promising for the use of data from museums and herbaria, especially as methods suited to the noise inherent in such data improve.
Orconectes rusticus currently is undergoing an explosive range expansion in the midwestern U.S.A., but information on the potentially important effects of water temperature on the species' biology is lacking. The thermal ecology of O. rusticus in southwestern Ohio, U.S.A., was examined by determining 1) the effects of four water temperatures (16, 20, 25, and 29C) on survival and growth of juveniles, 2) the responses of juveniles and adults to a thermal gradient (7–27 C), and 3) the thermal tolerances (critical thermal maximum, CTMax, and critical thermal minimum, CTMin) of free-living, field-acclimatized juveniles and adults on a biweekly basis throughout the summer. Month-long growth experiments predicted maximum growth rates of juveniles at water temperatures between 26 and 28C, but greatest survival between 20 and 22 C. Laboratory-acclimated (22 C) adults and field-acclimatized (2.5 C) juveniles both had an acute preferred temperature of 22 C. CTMaxs and CTMins of juveniles were 0.5–2.6 C higher than those of adults throughout the summer, suggesting that juveniles were exposed to water temperatures 1.5–6.8 C warmer than those of adults. Juvenile and adult O. rusticus prefer habitats where water temperatures favor maximum survival, but they usually are not found together in the same habitat; adults apparently displace the juveniles into warmer habitats. Warmer temperatures can decrease survival of juveniles but improve their growth rates, leading to enhanced fecundity and competitive ability. The past and future success of O. rusticus in expanding its range may depend, in part, on the species ability to adjust to new thermal environments occupied by other species of crayfish.
Introduced in Italy in the 1980s for aquaculture enterprises, the red swamp crayfish, Procambarus clarkii, has invaded many water bodies, giving rise to breeding populations that now threaten freshwater ecosystems. An understanding of the spatial behaviour of this crayfish could be the baseline for future research aimed at control and management. Following the same pattern as other freshwater decapods, P. clarkii, studied in an irrigation ditch system in Florence, Tuscany, displayed stationary phases (several marked individuals were recaptured in the area) interposed with nomadic bursts of movement (many marked crayfish disappeared). With a few exceptions, nocturnal activity prevailed in almost all the seasons in two different analysed habitats (irrigation ditches and the Massaciuccoli Lake, Lucca). However, in the laboratory, locomotion occurred mostly at daytime. Although still prevailing, P. clarkii's nocturnal activity appeared puzzling, because some of its major predators are nocturnal and this species is mostly herbivorous. The intercalation between stationary and wandering phases leaves open further promising studies on social structures in the field.
This tutorial gives a basic introduction to use of the Maxent program for maximum entropy modelling of species’ geographic distributions, written by Steven Phillips, Miro Dudik, and Rob Schapire, with support from AT&T Labs-Research, Princeton University, and the Center for Biodiversity and Conservation, American Museum of Natural History.
AimTo explore the performance of phylogenetic diversity metrics and of the novel categorical analysis of neo- and palaeo-endemism (CANAPE) using a dataset of Australian native Asteraceae and in particular to compare the results at two taxonomic ranks: genus and species.LocationAustralia.Methods
We used specimen data from Australia's Virtual Herbarium to produce species and genus distribution models with Maxent, and reconstructed a genus-level phylogeny. Spatial analyses were conducted at a 100 km × 100 km scale. Randomization tests were employed to identify cells with significantly high or low values of phylogenetic diversity (PD), and CANAPE was used to identify significant hotspots of neo- and palaeo-endemism.ResultsSignificantly high PD values were found scattered along the northern and north-eastern coast, whereas significantly low PD values characterized the arid interior. CANAPE signalled hotspots of neo-endemism in the mountainous south-east of Australia and hotspots of palaeo-endemism in the tropical north. Patterns were similar between genus- and species-level analyses, although the latter inferred more cells with significant values.Main conclusionsPD and CANAPE generally provided results for Australian Asteraceae consistent with expectations based on previous studies. This is further evidence for their utility in formulating and testing hypotheses about phylogenetic and biogeographical processes. The strength of the results is, however, partly dependent on the taxonomic scale of the analysis, a fact that has to be taken into account in the design and interpretation of future studies.
Red swamp crayfish (Procambarus clarkii) and signal crayfish (Pacifastacus leniusculus) are two invasive freshwater species with a worldwide distribution. The objective of this work was to investigate how the two species move and use space in an area of recent coexistence. Simultaneously, we test the use of new tools and indices to describe their movement patterns. To accomplish this we performed a radio-tracking program within a river-type habitat during two different periods (September/October 2010 and June/July 2013). We used spatial analysis tools to map crayfish radio-location data with and without accounting for the curvature of the river. To assess the consistency of the direction of movement and of the distances traveled by crayfish, two indices were developed. To assess the habitat preferences of each species we applied Ivlev's Electivity Index and the Standardized Forage Ratio. Movement of P. clarkii and P. leniusculus differed. The average detected movement was 8.8 m day−1 for P. clarkii and 17.5 m day−1 for P. leniusculus. However, crayfish behavior ranged from almost complete immobility – sometimes during several days – to large movements, in half a day, up to a maximum of 255 m for P. clarkii and 461 m for P. leniusculus. The proportion of upstream or downstream movements was independent of the species and both species displayed no preference for either direction. The indices of consistency of movement showed a large interindividual variation. Species and period (2010 or 2013) affected the mean daily distance traveled, maximum observed distance from location of release and percentage of observations under vegetation cover. The Ivlev's Electivity Index and the Standardized Forage Ratio presented similar results. P. clarkii showed a preference for pool areas with riparian vegetation cover while P. leniusculus preferred riffle and pool areas with riparian vegetation cover. Our work provided new and valuable data for modeling the active dispersal of these two problematic invaders in a context of coexistence.
A wide range of modelling algorithms is used by ecologists, conservation practitioners, and others to predict species ranges from point locality data. Unfortunately, the amount of data available is limited for many taxa and regions, making it essential to quantify the sensitivity of these algorithms to sample size. This is the first study to address this need by rigorously evaluating a broad suite of algorithms with independent presence-absence data from multiple species and regions. We evaluated predictions from 12 algorithms for 46 species (from six different regions of the world) at three sample sizes (100, 30, and 10 records). We used data from natural history collections to run the models, and evaluated the quality of model predictions with area under the receiver operating characteristic curve (AUC). With decreasing sample size, model accuracy decreased and variability increased across species and between models. Novel modelling methods that incorporate both interactions between predictor variables and complex response shapes (i.e. GBM, MARS-INT, BRUTO) performed better than most methods at large sample sizes but not at the smallest sample sizes. Other algorithms were much less sensitive to sample size, including an algorithm based on maximum entropy (MAXENT) that had among the best predictive power across all sample sizes. Relative to other algorithms, a distance metric algorithm (DOMAIN) and a genetic algorithm (OM-GARP) had intermediate performance at the largest sample size and among the best performance at the lowest sample size. No algorithm predicted consistently well with small sample size (n < 30) and this should encourage highly conservative use of predictions based on small sample size and restrict their use to exploratory modelling.
The MaxEnt software package is one of the most popular tools for species distribution and environmental niche modeling, with over 1000 published applications since 2006. Its popularity is likely for two reasons: 1) MaxEnt typically outperforms other methods based on predictive accuracy and 2) the software is particularly easy to use. MaxEnt users must make a number of decisions about how they should select their input data and choose from a wide variety of settings in the software package to build models from these data. The underlying basis for making these decisions is unclear in many studies, and default settings are apparently chosen, even though alternative settings are often more appropriate. In this paper, we provide a detailed explanation of how MaxEnt works and a prospectus on modeling options to enable users to make informed decisions when preparing data, choosing settings and interpreting output. We explain how the choice of background samples reflects prior assumptions, how nonlinear functions of environmental variables (features) are created and selected, how to account for environmentally biased sampling, the interpretation of the various types of model output and the challenges for model evaluation. We demonstrate MaxEnt’s calculations using both simplified simulated data and occurrence data from South Africa on species of the flowering plant family Proteaceae. Throughout, we show how MaxEnt’s outputs vary in response to different settings to highlight the need for making biologically motivated modeling decisions.
Aim
Species distribution models are an invaluable tool for anticipating the potential range of invasive species. These models often improve when both native and non‐native occurrences are available for model development and validation. Therefore, how might ecologists anticipate the potential distributions for emerging invasive species that lack any or abundant non‐native range occurrences? Here, we evaluate the recent suggestion of transferring niche shifts from well‐established ‘avatar’ invaders to emerging invaders by testing if ensemble niche shifts from a group of globally invasive plants improve model predictions when each of these species is iteratively treated as an ‘emerging’ invader.
Location
Global.
Methods
We built species distribution models using Mahalanobis distance and four climatic predictors (maximum and minimum temperature and precipitation) for 26 invasive terrestrial plants from an Australian priority list of weeds. Models using only native range occurrences for each species were modified with avatar niche shifts from the remaining ensemble of 25 species based on both a typical (median) niche shift and a large (extreme) niche shift (or niche expansion). Native range and both median and extreme avatar models were then compared with total range models (developed with both native and non‐native occurrences) for performance by measures of discrimination and an approximation of calibration.
Results
Avatar niche shifts reduced errors of omission for known non‐native occurrences relative to native range models, with a trade‐off of increased errors of commission of lesser magnitude. Further, our approximation of model calibration measured relative to total range models improved with avatar niche shifts. Differences between native range and avatar models were most pronounced for the larger ‘extreme’ avatar niche shifts (or expansion) based on increased niche size and decreased (towards 0) covariance among climatic axes.
Main conclusions
We suggest that researchers and managers evaluating risk of invasion of their jurisdiction by emerging data‐poor invaders modify native range models with observed avatar niche shifts from ensembles of well‐studied invaders. Alternative implementations of the avatar invader concept are discussed and research needs for methodological improvements proposed. Despite these opportunities for improved implementation of avatar niche shifts, ample evidence now supports that researchers should expect models based on only native ranges to underestimate or misrepresent the total range for data‐poor emerging invaders. Avatar niche shifts (and specifically expansion) from well‐studied species offer a precautionary means to anticipate the extent to which native range models may underestimate total ranges.
Aim
Models of species niches and distributions have become invaluable to biogeographers over the past decade, yet several outstanding methodological issues remain. Here we address three critical ones: selecting appropriate evaluation data, detecting overfitting, and tuning program settings to approximate optimal model complexity. We integrate solutions to these issues for Maxent models, using the Caribbean spiny pocket mouse, H eteromys anomalus , as an example.
Location
N orth‐western S outh A merica.
Methods
We partitioned data into calibration and evaluation datasets via three variations of k ‐fold cross‐validation: randomly partitioned, geographically structured and masked geographically structured (which restricts background data to regions corresponding to calibration localities). Then, we carried out tuning experiments by varying the level of regularization, which controls model complexity. Finally, we gauged performance by quantifying discriminatory ability and overfitting, as well as via visual inspections of maps of the predictions in geography.
Results
Performance varied among data‐partitioning approaches and among regularization multipliers. The randomly partitioned approach inflated estimates of model performance and the geographically structured approach showed high overfitting. In contrast, the masked geographically structured approach allowed selection of high‐performing models based on all criteria. Discriminatory ability showed a slight peak in performance around the default regularization multiplier. However, regularization levels two to four times higher than the default yielded substantially lower overfitting. Visual inspection of maps of model predictions coincided with the quantitative evaluations.
Main conclusions
Species‐specific tuning of model parameters can improve the performance of Maxent models. Further, accurate estimates of model performance and overfitting depend on using independent evaluation data. These strategies for model evaluation may be useful for other modelling methods as well.
The tiny island of South Bimini contains 4 species of lizards of he genus Anolis, a number surpassed only on the 4 largest islands of the Greater Antilles and on 2 very large and nearby satellite islands. These species are syntopic with respect to a two-dimensional area of the ground but divide the habitat according to perch height and perch diameter: sagrei is partly terrestrial but occurs more often on small and large low perches; distichus prefers the trunks and large branches of medium to large trees; angusticipes inhabits small twigs, especially at great heights; and carolinensis is found mostly on leaves or on the adjacent twigs and brances. The size classes of the species are staggered in such a way that the interspecific classes which overlap most in habitat overlap least in prey size. Similarities in prey size and prey taxa for classes of the same species are somewhat greater than those expected on the basis of habitat and morphology alone. The distribution of the species among the vegetation communities of Bimini can be explained on the basis of perch height and diameter preference. Within the same species, the larger lizards usually eat larger food, fewer items, and in sagrei more fruit; and they have a greater average range of food size per digestive tract. One species (distichus) is extremely myrmecophagous: about 75-90% of its food items are ants. In 3 of the 4 species, subadult males take more food and average smaller prey then females of the same head length. That species (distichus) which takes the smallest food items and whose classes overlap the most in habitat preference with those of other species is least dimorphic is size between the sexes. It is suggested that such small, nondimorphic species are best suited for insinuation into complex faunas, whereas larger, dimorphic forms are best for the colonization of empty areas. The usefulness of various measures of "overlap" and "specialization" is evaluated for this lizard association.
Photoperiod and temperature control ovarian growth and maturation in the crayfish Orconectes virilis, and increased water temperature induces egg laying in spring.Laboratory experiments with O. virilis from Alberta revealed that although ovarian growth will occur in warm water and darkness, or cold water and long-day photoperiod, subsequent exposure to spring temperature and photoperiod will not induce egg laying when the ovary has matured under these conditions. Complete ovarian maturation requires 4–5 months of low temperature and constant darkness.
We tested for differences in maximum daily consumption (C-max) and respiration (R) rates at four temperatures (18, 22, 26, and 30degreesC) for crayfishes with varying distributions in Missouri, U.S.A. Five species of crayfish were studied: Orconectes eupunctus (Williams, 1952), found only in the Eleven Point River downstream of Greer Spring, O. hylas (Faxon, 1890), found only in the Black and St. Francis River drainages in the Ozarks, O. luteus (Creaser, 1933) and O. punctimanus (Creaser, 1933), which are widespread through the Ozarks, and O. virilis (Hagen, 1870), which has a broad distribution in Missouri but is most abundant in the northern agricultural region and in the transitional area along the northern and western border of the Ozarks. For species restricted to the Ozarks, C-max increased from 18 to 22degreesC and stabilized or declined at the two warmest temperatures, while C-max peaked at 26degreesC for O. virilis before declining at 30degreesC. Significant interspecific differences in C-max were found at several temperatures. C-max for O. virilis was significantly greater than C-max for one or more Ozark crayfishes at 18, 22, and especially at 26degreesC. Respiration rates (R) increased with temperature for all species, but few differences in R were observed between species. Results suggest that O. virilis has a potential bioenergetic advantage over Ozark crayfishes, particularly at temperatures near 26degreesC. Our findings hint that warming of Ozark streams due to changes in climate or land use could yield growth conditions which are more favorable for introduced species such as O. virilis and possibly O. rusticus compared to native species, resulting in increased probability of shifts in crayfish community composition.
Aim The assumption of equilibrium between organisms and their environment is a standard working postulate in species distribution models (SDMs). However, this assumption is typically violated in models of biological invasions where range expansions are highly constrained by dispersal and colonization processes. Here, we examined how stage of invasion affects the extent to which occurrence data represent the ecological niche of organisms and, in turn, influences spatial prediction of species’ potential distributions.
Location Six ecoregions in western Oregon, USA.
Methods We compiled occurrence data from 697 field plots collected over a 9-year period (2001–09) of monitoring the spread of invasive forest pathogen Phytophthora ramorum. Using these data, we applied ecological-niche factor analysis to calibrate models of potential distribution across different years of colonization. We accounted for natural variation and uncertainties in model evaluation by further investigating three hypothetical scenarios of varying equilibrium in a simulated virtual species, for which the ‘true’ potential distribution was known.
Results We confirm our hypothesis that SDMs calibrated in early stages of invasion are less accurate than models calibrated under scenarios closer to equilibrium. SDMs that are developed in early stages of invasion tend to underpredict the potential range compared to models that are built in later stages of invasion.
Main conclusions A full environmental niche of invasive species cannot be effectively captured with data from a realized distribution that is restricted by processes preventing full occupancy of suitable habitats. If SDMs are to be used effectively in conservation and management, stage of invasion needs to be considered to avoid underestimation of habitats at risk of invasion.
MaxEnt is a program for modelling species distributions from presence-only species records. This paper is written for ecologists and describes the MaxEnt model from a statistical perspective, making explicit links between the structure of the model, decisions required in producing a modelled distribution, and knowledge about the species and the data that might affect those decisions. To begin we discuss the characteristics of presence-only data, highlighting implications for modelling distributions. We particularly focus on the problems of sample bias and lack of information on species prevalence. The keystone of the paper is a new statistical explanation of MaxEnt which shows that the model minimizes the relative entropy between two probability densities (one estimated from the presence data and one, from the landscape) defined in covariate space. For many users, this viewpoint is likely to be a more accessible way to understand the model than previous ones that rely on machine learning concepts. We then step through a detailed explanation of MaxEnt describing key components (e.g. covariates and features, and definition of the landscape extent), the mechanics of model fitting (e.g. feature selection, constraints and regularization) and outputs. Using case studies for a Banksia species native to south-west Australia and a riverine fish, we fit models and interpret them, exploring why certain choices affect the result and what this means. The fish example illustrates use of the model with vector data for linear river segments rather than raster (gridded) data. Appropriate treatments for survey bias, unprojected data, locally restricted species, and predicting to environments outside the range of the training data are demonstrated, and new capabilities discussed. Online appendices include additional details of the model and the mathematical links between previous explanations and this one, example code and data, and further information on the case studies.
We compared predictive success in two common algorithms for modeling species’ ecological niches, GARP and Maxent, in a situation that challenged the algorithms to be general – that is, to be able to predict the species’ distributions in broad unsampled regions, here termed transferability. The results were strikingly different between the two algorithms – Maxent models reconstructed the overall distributions of the species at low thresholds, but higher predictive levels of Maxent predictions reflected overfitting to the input data; GARP models, on the other hand, succeeded in anticipating most of the species’ distributional potential, at the cost of increased (apparent, at least) commission error. Receiver operating characteristic (ROC) tests were weak in discerning models able to predict into broad unsampled areas from those that were not. Such transferability is clearly a novel challenge for modeling algorithms, and requires different qualities than does predicting within densely sampled landscapes – in this case, Maxent was transferable only at very low thresholds, and biases and gaps in input data may frequently affect results based on higher Maxent thresholds, requiring careful interpretation of model results.
The presence of the yabbie Cherax destructor in a number of wild aquatic systems in the Pilbara and Southwest Coast Drainage Divisions of Western Australia is documented. This is of great concern as all native freshwater crayfishes in Western Australia are endemic and restricted to the southwest, while the Pilbara Division has no native species. An introduced population of C. destructor was sampled monthly from the Hutt River (Pilbara Drainage Division) for determination of life-history and reproductive biology in a wild aquatic system in Western Australia for the first time. Proliferation in that system was attributed to specific traits including: attaining first maturity at the end of its first year of life; a protracted spawning period (July–January); relatively high mean ovarian fecundity of 210.2 (9.24 S.E.); and a rapid growth rate (curvature parameter K=0.78 and asymptotic orbital carapace length OCL∞=51.25mm ascertained from a seasonal von Bertalanffy growth curve) that was comparable to the larger sympatric marron Cherax cainii in this system. The life-history characteristics of C. destructor in the Hutt River were typical of many other invasive crayfish species and it has the potential to impact the unique aquatic ecosystems and the endemic freshwater crayfish species of the region.
In a rather provocative article, Parker et al. (1999) claimed that, up to then, little scientific attention had been placed on developing either theoretical or operational
generalizations about the impact of invasive species. Specifically, the authors lamented the lack of a general framework in
which to discuss “what impact is, or how we decide that the non-indigenous species exceeds that of another, or how we decide
that the impact of a particular species is greater in one place than in another” (Parker et al. 1999, p. 4). Today, this scenario seems to have changed for several freshwater non-indigenous species (NIS), e.g. the zebra
mussel Dreissena polymorpha (Pallas) (e.g. Karatayev et al. 2002, Ricciardi 2003), but it has remained practically unaltered for other widely diffused bioinvaders that have, however,
attracted much scientific attention in these latest years, such as freshwater crayfish.