No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
A statistical explanation of MAXENT for ecologists
Diversity and Distributions
Abstract
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.
... As a result, SDMs based on presence-only data, such as MaxEnt, have gained popularity in recent years [29]. MaxEnt applies the principle of maximum entropy to model the relationship between presence-only data and environmental variables, helping to estimate species' ecological niches and their potential geographic ranges [30]. Changes in population size and species diversity can lead to trophic cascades, causing potential changes in land cover [31,32]. ...
... Since dogs are generalists and can survive in extreme environments, climatic variables were omitted from the feral dog model. A correlation analysis was conducted among environmental variables, and variables with a high correlation coefficient (r > 0.7) were discarded [30]. Different models were compared using ENMTools [36], varying some of the algorithm parameters including the regularisation parameter [37] along with the environmental variable functions. ...
... Factors such as sample size, environmental variable accuracy, and species' ecological traits can affect model validity [30,45]. Due to limited access to certain island regions, our sampling was uneven, with more reports from northern and central areas than from the southern region. ...
Feral dogs (Canis familiaris) are an emerging threat to biodiversity on Navarino Island, Chile, where they have become apex predators in the absence of natural carnivores. This study evaluated the spatial distribution of feral dogs and their impacts on native species, including guanacos (Lama guanicoe) , upland geese (Chloephaga picta) , and flightless steamer ducks (Tachyeres pteneres) . Presence-only data collected during two field expeditions were analysed using species distribution models (MaxEnt) to predict habitat suitability for feral dogs and guanacos. Habitat connectivity analyses identified at least two potentially isolated feral dog populations. Using generalised linear and non-linear models, we assessed the ecological impacts of feral dogs, finding significant habitat overlap with guanacos, particularly in central areas of the island. This overlap corresponded to a reduced likelihood of guanaco occurrence, suggesting behavioural adaptations to disturbance and predation pressure. Upland geese exhibited a negative association with both actual and predicted feral dog presence, while flightless steamer ducks appeared unaffected. Furthermore, NDVI changes observed on Navarino Island compared to Torres del Paine, where no invasive species are present, are likely linked to the spread of invasive species and the decline of guanacos in the Magellanic forest, highlighting the cascading ecological consequences of feral dog invasion. Our findings emphasise the urgent need for feral dog management to protect vulnerable species and maintain ecosystem health, particularly in fragile environments where invasive predators can have disproportionate impacts.
Graphical Abstract
Highlights
Invasive species presence leads to environmental changes and potential tipping points
Feral dogs pose major harm as an invasive species
Endemic wildlife adapts their behaviour such as habitat use to avoid dog predation
Species diversity and population stability are higher in areas with no feral dogs
Dogs in general should be removed from the wild
... Among SDMs, the maximum entropy model (MaxEnt) has been recognized as one of the best performing methods compared to others when dealing with presence-only data and has been extensively used to assess ecological requirements, environmental responses, and habitat suitability of species. MaxEnt is esteemed as a powerful algorithm for predicting species distributions, particularly in scenarios with few occurrence data such as endemic species, typically representing restricted distribution patterns (Elith et al. 2006;Elith et al. 2011;Phillips et al. 2006). This model illustrates the relationships between species distributions and environmental conditions across space and time, considering it among the most powerful tools for identifying appropriate habitats and predicting potential distributions of species (Ahmadi et al. 2023;Damaneh et al. 2022;Khajoei Nasab et al. 2022;Hosseini et al. 2024;Zeraatkar and KhajoeiNasab 2024). ...
... Afterwards, collinearity among environmental variables was tested by Pearson's correlation coefficient (r). If two variables were highly correlated (r > 0.70), one of them was excluded in order to avoid co-linearity (Elith et al. 2011). Finally, guided by our observations on the ecological needs of each species during our field studies, available literature (K. ...
... The implementation of the MaxEnt model, as outlined by Phillips et al. (2006), was carried out using MaxEnt v3.4.4 k (Phillips et al. 2006). This algorithm has been recognized as a high-performance approach for predicting species distributions, particularly when working with limited sample sizes (Elith et al. 2006;Elith et al. 2011;Pearson et al. 2007). We employed 25% of species occurrence records for model testing and 75% for model calibration. ...
Woody plants offer valuable services to ecosystems, including providing useful products, stabilizing ecosystems, and mitigating climate and pollution effects. However, they face significant abiotic and biotic stresses, with climate change being the most critical challenge. It is essential to understand that reducing populations of woody species, particularly those found only in a specific area, can have severe and irreversible effects on the entire ecosystem. Therefore, exploring the potential influence of climate change on the distribution of endemic woody species is an appealing subject for conservation researchers. This study investigates how climate change affects the distribution of three endemic species of woody plants in the genus Colutea in Iran. The MaxEnt model was used to analyze the data, and the results showed that the model was effective for predicting the impact of climate change on the plants (AUC ≥ 0.9). The distribution of C. persica was significantly affected by solar radiation, Precipitation of Wettest Month, sand, and silt content. C. porphyrogamma's distribution was impacted by Mean Temperature of Coldest Quarter, Precipitation of Driest Month, and Cation Exchange Capacity, while C. triphylla was most affected by Precipitation Seasonality, Precipitation of Driest Quarter, and Isothermality. According to the findings, the distribution of these species is expected to decrease in the 2050s and 2070s due to climate change, based on the RCP4.5 and RCP8.5 climate scenarios. These findings can be useful for developing strategies to manage the impacts of climate change on these species.
... Toshkent: Noshir. 10 [12], chunki ular yashash joyining mosligini bashorat qilishda bioiqlim oʻzgaruvchilarga tayanadi [25] va turlarning uchrash joylari haqidagi ma'lumotlardan foydalanadi. Ushbu modellar koʻpincha fazoviy nuqta (koordinata) sifatida ifodalanadi, bunda jarayon intensivligi atrofdagi omillarga bog'liq boʻladi [10]. ...
... Dalillar global isish yashash joylarining mosligiga sezilarli ta'sir etishini koʻrsatmoqda [24]. Iqlim oʻzgarishi ta'sirida turlarning yashash hududi albatta oʻzgaradi [12]. Vaqt va fazoga asoslangan modellar iqlim oʻzgarishlari jarayonida monitoring tizimlarini erta ogohlantirish signallari sifatida yaratish imkonini beradi [20]. ...
... Haqiqiy sharoitlarda oʻqitish munosabati natijasida bu model oʻzgaruvchan ekologik omillar sharoitida turlarning tarqalishini ham baholash uchun ishlatilishi mumkin [29]. Natijada, ushbu model global miqyosda turlarning tarqalish sohasi kengayishi yoki torayishini baholaydi [12,26]. ...
Markaziy Osiyoda Elymus turkumining ahamiyati yuqori boʻlsa-da, ushbu turkum turlarining yashash joylarini aniqlashda yuqori aniqlikdagi kompyuter modellaridan foydalanilgan tadqiqotlar juda kam. Bizning maqsadimiz Elymus turkumining Oʻrta Osiyo uchun endem hisoblangan turlarining fazoviy tarqalishi kelajakdagi iqlim sharoitlarida qanday oʻzgarishini aniqlash boʻlib, ushbu tadqiqot ishida turkumning hozirgi va kelajakdagi tarqalishini baholash uchun bioiqlimiy modellashtirish usulidan foydalanildi.
... Spatially explicit habitat models also result in greater survey efficiency when compared with simple or stratified random sampling over large areas (Guisan et al. 2006). Many modeling strategies have been implemented to predict species occurrences, including generalized linear models ("GLMs"; presence and predictor variables (Elith et al. 2011). The predictions of individual models can be highly variable, and the choice of modeling method is known to impact model outcomes and accuracy (Araújo and New 2007). ...
... Maxent modeling has also received scrutiny because it does not produce estimates of the probability of presence, but rather estimates an index of habitat suitability for each raster cell between 0 and 1 (Elith et al. 2011;Royle et al. 2012). Maxent also uses a complementary log-log ("cloglog") link function by default, which assumes that a species' presence or absence at nearby sites is independent (Phillips et al. 2017). ...
... e.g.,Carlos-Júnior et al. 2020), maximum entropy modeling ("Maxent";Phillips et al. 2006, Elith et al. 2011, and random forests ("RF";Breiman 2001, Hengl et al. 2018, Valavi et al. 2021). Maxent is one of the most widely-used SDM techniques because it can be used to model presence-only data(Phillips et al. 2006), is robust to the small sample sizes typical of rare species surveys (Kaky et al. 2020), and can model non-linear relationships between species ...
The bridle shiner (Notropis bifrenatus) is a small minnow species native to the eastern United States and southeastern Canada. The species is declining dramatically throughout most of its native range and has legal protection or concern status in thirteen states and two Canadian provinces. In Maine, the bridle shiner is listed as a Species of Special Concern and considered a Species of Greatest Conservation Need, partially because we lack a basic understanding of their status and distribution within the state. Bridle shiners have historically been found in southern and western Maine in densely vegetated, shallow habitats along the shorelines of streams and ponds. Surveys performed at sites where the shiners were once abundant have yielded very few or none of these fish. This project informed the Maine Department of Inland Fisheries & Wildlife on the status of the species in Maine and provides a foundation for future long-term monitoring of bridle shiner populations in the State.
We used a combination of both direct capture techniques and environmental DNA (eDNA) to locate bridle shiners. eDNA is increasingly being used to detect rare aquatic species such as bridle shiners because it is both highly sensitive and less invasive than direct capture. We designed a single-species primer-probe assay to detect bridle shiner DNA, then surveyed 32 sites with a record of historic bridle shiner occurrence. In addition to collecting eDNA samples (2021-2022), we surveyed 29 sites using traditional seine netting techniques in 2021. In 2022, we used a preliminary habitat suitability model to select 46 locations with unknown bridle shiner presence to survey with eDNA. To refine eDNA methodology, we assessed trends in eDNA detection probability across seasons and compared DNA detection between three filter pore sizes. We rediscovered bridle shiner populations at 11 of 32 historically occupied sites and documented bridle shiners in four additional waterbodies. We determined that eDNA surveys were most effective in early or midsummer, and that larger filter pore sizes are a viable option for surveying bridle shiners.
Species distribution modeling (SDM) statistically associates species occurrence data with environmental variables to evaluate habitat suitability. We used an ensemble species distribution modeling (SDM) approach to identify both the current and historic range of the bridle shiner within Maine and New Hampshire. We also investigated how local habitat characteristics influenced bridle shiner presence using generalized linear models. Both historic site surveys and ensemble SDMs suggest that there has been a substantial loss of historic bridle shiner habitat in Maine (-62%) and New Hampshire (-46%). At the landscape scale, we found significant effects of forest type, catchment position, soil composition, elevation, and slope on bridle shiners. Within a site, bridle shiners were associated with areas that had a higher proportion of complex-leaved submerged aquatic vegetation and a lower proportion of persistent emergent and floating vegetation. We determined that both eDNA and seine net surveys are viable options for monitoring bridle shiners in Maine, and that such survey strategies can be used with species distribution models to focus future surveys and to identify areas of possible conservation, reintroduction, or restoration actions.
... La idoneidad se refiere a la medida en que un área geográfica proporciona las condiciones ambientales óptimas para el crecimiento y desarrollo de una especie o cultivo (Elith et al., 2011). En el contexto del café, la idoneidad climática se evalúa considerando factores como temperatura, precipitación y altitud, que influyen directamente en la productividad y calidad del cultivo . ...
... (Phillips et al., 2006). Se seleccionaron características lineales, cuadráticas, de producto y de bisagra, además de las características automáticas, para permitir una representación flexible de las relaciones entre las variables ambientales y la idoneidad del hábitat (Elith et al., 2011). Esta configuración permitió al modelo capturar relaciones complejas y no lineales entre las variables predictoras y la distribución de las especies. ...
... El análisis Jackknife se utilizó para evaluar la importancia relativa de cada variable ambiental, para identificar los factores de mayor importancia para la distribución de las especies de café (Phillips et al., 2006). Además, la generación de curvas de respuesta facilita la interpretación del efecto de cada variable en predicción del modelo y proporciona perspectivas ecológicas valiosas sobre los requerimientos de hábitat de las especies (Elith et al., 2011). ...
El sector cafetalero es fundamental en los ámbitos económico, social y ambiental de las regiones productoras de café en México. Sin embargo, el cambio climático está alterando las condiciones agroclimáticas idóneas para el cultivo de Coffea arabica y C. canephora, amenazando la producción futura. Dada la variabilidad regional de estos impactos, es necesario identificar cómo el cambio climático afectará la idoneidad productiva en regiones específicas. Estudios enfocados en la distribución de especies aportan herramientas clave para dimensionar los impactos y emitir recomendaciones orientadas en la adaptación al cambio climático. El objetivo de este trabajo es evaluar el impacto potencial del cambio climático en la producción de Coffea arabica y C. canephora, bajo el enfoque de distribución de especies en la región nororiental de Puebla, las montañas en Veracruz y la región mazateca de Oaxaca. El estudio se realizó bajo el enfoque de distribución de especies, la modelación se desarrolló mediante el software MaxEnt para proyectar la distribución potencial de las especies. Se evaluaron escenarios climáticos futuros SSP126 y SSP585 para el período 2041-2060. Los resultados sugieren cambios significativos en la distribución para ambas especies. Para la especie C. arabica, se observa una tendencia general de reducción en áreas óptimas, con variaciones regionales notables. Para C. canephora se observa una respuesta diferente, con algunas áreas con tendencia al aumento de la distribución de esta especie. Las variables más influyentes en la distribución de ambas especies fueron la precipitación, la elevación y la velocidad del viento. Si las condiciones agroclimáticas futuras favorecen la expansión de C. canephora como sugieren las proyecciones para algunas regiones, podría plantearse una transición parcial hacia esta especie como estrategia de adaptación. El estudio concluye que el cambio climático reducirá significativamente la distribución de C. arabica y favorecerá moderadamente a C. canephora, con impactos regionales diferenciados. Es crucial ajustar estrategias de manejo según cada región para asegurar la sostenibilidad de la producción cafetalera.
... When scientists recorded the influence of the environmental variables on the development and distribution of plant species (Crawley, 1997;Hageer et al., 2017), they used qualitative estimations but also accurate numerical techniques (Rosenzweig, 1995;Rockwood, 2006) and developed models that have strong predictive capabilities for describing species distribution in their natural range Reese et al., 2005;Elith & Leathwick, 2009). Following species inventory from an area and recording the spatial characteristics or/and environmental factors from that area, for estimating their complex relationship, it has been developed species distribution models (SDMs) (Elith et al, 2011;Wisz et al., 2013). Species distribution models for plants and animals (called also bioclimatic or environmental niche modelling) progressed hugely for usage in: forecasting anthropogenic effects on patterns of biodiversity at different spatial scales Reese et al., 2005), providing estimates of the spatial distributions of rare species with limited occurrence for their conservation management (Hernandez et al., 2006;Sharma et al., 2018) or with very small populations (Liu et al., 2024) estimating species responses to climate change (Austin, 2007;Hosseini et al., 2024), estimating the potential for species populations to occur in areas not previously surveyed (Hernandez et al., 2008;Rhoden et al., 2017;Ma & Sun, 2018;Bao et al., 2022). ...
... Species distribution models for plants and animals (called also bioclimatic or environmental niche modelling) progressed hugely for usage in: forecasting anthropogenic effects on patterns of biodiversity at different spatial scales Reese et al., 2005), providing estimates of the spatial distributions of rare species with limited occurrence for their conservation management (Hernandez et al., 2006;Sharma et al., 2018) or with very small populations (Liu et al., 2024) estimating species responses to climate change (Austin, 2007;Hosseini et al., 2024), estimating the potential for species populations to occur in areas not previously surveyed (Hernandez et al., 2008;Rhoden et al., 2017;Ma & Sun, 2018;Bao et al., 2022). In the recent years many methods are available in ecology, conservation biology, biogeography (Franklin, 2013), being distinctive throughout the data used (Elith et al, 2011). Some database might contain accurate data but other data are limited in presenceabsent and occurrence of the species (Merow et al., 2013). ...
... We used the MaxEnt model (Phillips et al. 2006;Elith et al. 2011) as implemented in the dismo package (Hijmans et al. 2023) to model suitable habitats for Sesamum. The MaxEnt model is the most suitable package used for SDMs (Phillips et al. 2017;Soberón et al. 2017;Urbina-Cardona et al. 2019;Ahmadi et al. 2023) and was developed in the context of chosen contemporary environmental layers. ...
... The high AUC values for predicting suitable habitats for Sesamum species align with previous studies that have highlighted the utility of the Maxent model in ecological niche modeling and species distribution predictions (Elith et al. 2011;Lorestani et al. 2022;Tesfamariam et al. 2022;Valavi et al. 2022;Ahmadi et al. 2023). Key environmental variables were identified as significant contributors to the habitat suitability for the studied species. ...
Plants with restricted distributions and small population sizes are particularly vulnerable to climate change. Sesamum species are ideal for species distribution modeling due to their ecological sensitivity, agricultural and economic importance, and wide geographic range, providing insights for conservation and policy. Global. Sesamum. We applied the maximum entropy (MaxEnt) model to assess the global ecological niche breadth of Sesamum species and examine how bioclimatic and soil variables influence their future (2080) distribution. We identified key environmental drivers and projected species‐specific range shifts under changing climatic conditions. MaxEnt models effectively predicted suitable habitats, with climate variables playing a dominant role. Precipitation of the wettest month (BIO13) was particularly influential for S. abbreviatum, S. alatum, and S. angustifolium, while temperature variables (BIO7, BIO11) were also key. Elevation moderately impacted S. angolense, while soil factors such as pH (S. abbreviatum) and clay content (S. angolense) exhibited species‐specific effects. Principal component analysis revealed variation in niche breadth, with S. indicum and S. schinzianum occupying broader ecological ranges, whereas S. saxicola and S. abbreviatum were more restricted. Future projections suggest 46.4% of the species will experience range contractions, with S. schinzianum facing the most significant decline. Conversely, 39.3% of the species, including S. imperatricis and S. abbreviatum, are expected to expand their ranges. Phylogenetic analyses indicate a random distribution of niche breadth and extinction risk across the genus. Our findings highlight the susceptibility of Sesamum species to climate change, emphasizing the need for urgent conservation actions. Prioritizing vulnerable species such as S. forbesii and S. sesamoides, alongside habitat restoration and long‐term monitoring, is crucial to mitigate population declines and prevent extinction.
... SDMs are important in assessing the potential impacts of climate change on species' mos because they can help ecologists and policymakers understand how species respond to climate change and provide a scientific basis for developing effective conservation measures (Hadi Ahmad et al. 2023, Zhang et al. 2024c. Maximum Entropy (MaxEnt) model is one of the main methods used in ecological niche models (ENMs) to construct predictive models by maximizing entropy, ie selecting the most homogeneous and unbiased data to simulate the potential distribution of a species in the presence of incomplete data (Guillera-Arroita et al. 2014, Elith et al. 2021, Iannella et al. 2021. MaxEnt models have been widely used in biodiversity conservation planning, invasive species management, climate change impact assessment, pest monitoring, and other related fields, especially in pest management, it can effectively predict the potential distribution areas of pests. ...
... The growth, development, and reproduction of insects are closely tied to climatic variables, with temperature and precipitation acting as key drivers of their distribution and survival (Zhang et al. 1993, Keena et al. 2021, Li et al. 2022. Previous studies have demonstrated that Bio2, Bio4, Bio14, and Bio15 collectively contributed 49.4% to the distribution of M. alternatus (Elith et al. 2021, Gao et al. 2023. Similarly, Bio1 and Bio18 were found to account for 68.2% of the distribution of M. saltuarius (Estay et al. 2014). ...
Monochamus sutor, an important phytophagous pest, is a known vector insect of Bursaphelenchus mucronatus in addition to feeding directly on trees. Although B. mucronatus causes relatively minor damage in European and Asian forests, its threat to coniferous forests is similar to that of Bursaphelenchus xylophilus. Given that B. xylophilus evolved into a destructive pathogen after its introduction into Asia, B. mucronatus may also pose a potential threat to North American coniferous forests. Therefore, we assessed the potential global distributions areas of M. sutor and their relative dynamics under different climate scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) in the current (i. considering only bioclimatic factors; ii. including anthropogenic factors) and in the future (2050s and 2070s) using an optimized Maximum Entropy ecological niche model. The mean area under the curve value of the optimized model was greater than 0.86 and the true skill statistic value was greater than 0.79. Potentially suitable habitat for M. sutor is driven by a combination of temperature (Bio1 and Bio2), precipitation (Bio14, Bio15, and Bio18), and human activities. In the current period, suitable areas are concentrated in Europe, East Asia, and North America, and are smaller in the presence of anthropogenic disturbance than in the presence of bioclimatic factors alone. At the same time, under future climate scenarios, the potential range of M. sutor will always expand more than contract, with a projected increase of 1,329.02 to 1,798.23 × 104 km2 compared to the current time period, especially spread toward Canada and the United States of America in North America. The present study provides important insights into the potential risks of M. sutor, which is important to help guide decision-making in pest control as well as forest conservation.
... The most common solution to dealing with autocorrelation when research questions are framed around larger scale patterns of space use is to subsample the data to the point that autocorrelation becomes negligible (Northrup et al. 2013). Then estimation can proceed with conventional methods for estimating RSFs and SDMs (Manly et al. 2007;Aarts et al. 2012;Renner and Warton 2013;Renner et al. 2015;Elith et al. 2011;Hooten et al. 2017). However, doing so discards valuable ecological information and comes at the cost of statistical power to detect potentially important spatial patterns (Hooten et al. 2014). ...
... While many researchers are perhaps more accustomed to using an use-availability design and logistic regression procedure for estimating in an RSF (Manly et al. 2007), or MaxEnt or similar in the SDM case (Elith et al. 2011;Merow et al. 2013), it is well established that these methods estimate equivalent models as to the IPP methods (Warton and Shepherd 2010;Renner and Warton 2013;Fithian and Hastie 2013;Fieberg et al. 2021;Matthiopoulos et al. 2023). Nonetheless, all of these methods involve the critical assumption that animal locations are independent from one another, which in many applications might necessitate discarding data to meet that assumption (Northrup et al. 2013). ...
Aim
Recent methodological advances for studying how animals move and use space with telemetry data have focused on fine‐scale, more mechanistic inference. However, in many cases, researchers and managers remain interested in larger scale questions regarding species distribution and habitat use across study areas, landscapes, or seasonal ranges. Point processes offer a unified framework for many methods applied in studies of species distribution and resource selection; however, challenges remain in terms of dealing with temporal autocorrelation common in many types of telemetry data collected from animal locations.
Innovation
Space–time point processes (STPPs) have a unique property, in that marginalising time offers a connection between individual animal movement and broader point processes, yet this property has seen little attention in both statistical and applied research. In this paper, we first present some of the details of this marginalisation property and methods for applying marginalised STPPs (mSTTPs) to autocorrelated telemetry data and then apply a mSTTP in a case study on the summer space use and habitat selection of female caribou (Rangifer tarandus) in Denali National Park and Preserve, Alaska.
Main Conclusions
The case study demonstrated that an mSTPP approach can improve inference over other commonly used methods in terms of its ability to account for temporal autocorrelation and offers greater precision in parameter estimates and improved predictions of space use. As this method fits conveniently into the existing point process frameworks, it offers a practical solution to dealing with temporal autocorrelation inherent to many types of telemetry data when research questions center around broader scale patterns of animal habitat selection and space use.
... SDMs are particularly valuable for identifying areas that are critical for conservation. Different modeling techniques such as maximum entropy (MaxEnt), generalized linear models (GLM) and random forests (RF) are used to build these predictive models (Elith et al. 2011). The accuracy of SDM depends on the quality of the input data and the selection of appropriate modeling techniques and environmental variables. ...
... MaxEnt is a machine-learning method used for species distribution modeling, which estimates the probability distribution of a species' presence based on environmental constraints while making the fewest assumptions about the unknown factors (Phillips, 2005). It is particularly effective for predicting species distributions with limited occurrence data and has become a widely used tool in ecological niche modeling (Elith et al. 2011). The 19 selected environmental variables and species occurrence records of C. exarillata were loaded into MaxEnt. ...
Accurate prediction of habitat suitability is crucial for species of conservation importance. Predictive distribution models play a key role in conservation by identifying current and future suitable habitats. Cullenia exarillata A. Robyns is an endemic and keystone tree species of the tropical wet evergreen forests of the Western Ghats of India. This study used a species distribution model to predict the current and future distribution of C. exarillata. Various environmental variables and the MaxEnt model were used to assess the current potential distribution and shifts within different shared socioeconomic pathways. The findings illustrate the potential reduction of the species ecological niche in certain landscapes of Karnataka, Kerala and Tamil Nadu under future climate change scenarios. The receiver operating characteristic area under the curve was used to evaluate the accuracy of the model. The Jackknife test was used to assess the significance of environmental factors. This study highlights the importance of targeted conservation and habitat management strategies for the conservation of C. exarillata. This spatial approach can be applied to other species facing similar threats, making it an essential tool for broader conservation efforts.
... With SDMs, the fundamental niche of a species can be modeled by relating species occurrence records in the form of presence-only, presence-absence, or pseudo-absence to environmental predictors of the background area to estimate habitat suitability across a study area (Franklin 2010;Guisan and Thuiller 2005). Presence-only models use species occurrences commonly curated by natural history museums or online species occurrence databases wherein corresponding data on species absence is unavailable (Elith et al. 2011;Pearson 2007;Phillips et al. 2006). A widely used SDM method implemented in Maxent (Maximum entropy method; Phillips et al. 2017) is advantageous for analyses using presence-only data because it allows for tuning model inputs and settings to alleviate bias and optimize predictive performance (Kramer-Schadt et al. 2013;Merow et al. 2013;Phillips et al. 2009). ...
... Because there is inherent uncertainty associated with projecting models into novel environmental conditions (Barbosa et al. 2016;Elith and Leathwick 2009;Elith et al. 2011), we explored the ecological underpinnings of any incongruence between in situ and projected models on either side of the Mississippi River. Specifically, we asked two different-but inherently interrelated-questions regarding differences in geographic space (G-space) and environmental space (Espace; Peterson et al. 2012 (Elith et al. 2010). ...
Freshwater fishes are among the most threatened taxa in the world. A major challenge for the conservation and management of threatened fishes is scarce information regarding life history, habitat requirements, and the drivers of declines. Species distribution models (SDMs) that leverage existing occurrence records and geospatial data have aided in addressing these challenges. We used SDMs to better understand large‐scale distributional patterns of Bluntface Shiner (Cyprinella camura; BFS), a minnow facing declines across its range. We modeled the potential distribution of BFS based on natural, abiotic factors and existing occurrence records to identify landscape‐scale factors underlying their distribution. We also compared environmental conditions between their disjunct ranges east and west of the Mississippi River and examined model transferability when projecting models into opposing ranges. Our analyses revealed a naturally fragmented distribution both east and west of the Mississippi River, but populations to the east of the Mississippi River occupy streams with broadscale environmental conditions that differ from those to the west. Models projected across the Mississippi River did not reflect the contemporary range of BFS, underscoring differences in occupied niches on either side of the biogeographic divide and emphasizing the need for caution when projecting SDMs to novel ranges. Our results provide a baseline to gauge range loss of BFS, highlight areas of high suitability for conservation, and identify locations where further sampling or research may be warranted.
... Occurrence records that fell outside the 400 km buffer of the species polygon maps were removed, following Ficetola et al. (2014). Then, all datasets were combined and the 'CoordinateCleaner' package in R was used to eliminate records entry errors in a standardized or reproducible way and to decrease sampling bias and ensure data accuracy and reliability (see Boria et al. 2014;Dormann et al. 2007;Elith et al. 2010). Subsequently, the 'spThin' package in R (Aiello-Lammens et al. 2015) was employed to reduce spatial autocorrelation by filtering occurrences within a 5 arc minutes × 5 arc minutes area (Stolar & Nielsen 2014;Supplementary Material Fig. S2). ...
... Recently different studies have conducted quantitative analyses to detect sampling bias using the sampbias R package (Chauvier et al. 2021;Zizka et al. 2021), that is specific to the sampling effort issue in the field. However, here, quantitative analyses were conducted to assess sampling bias, like 'CoordinateCleaner' and 'spThin' packages (Aiello-Lammens et al. 2015;Boria et al. 2014;Dormann et al. 2007;Elith et al. 2010). Therefore, studies that provide data on the physiological responses of animals in a climate change context can limit inaccuracies when predicting species' invasion risks (Guisan et al. 2012;Estrada et al. 2016;Jiménez et al. 2019). ...
Biological invasions represent one of the main anthropogenic drivers of global change with a substantial impact on biodiversity. Traditional studies predict invasion risk based on the correlation between species’ distribution and environmental factors, with little attention to the potential contribution of physiological factors. In this study, we incorporated temperature-dependent sex determination (TSD) and sex-ratio data into species distribution models (SDMs) to assess the current and future suitable habitats for the world’s worst invasive reptile species, the pond slider turtle (Trachemys scripta). First, occurrence records of T. scripta from online databases and published scientific literature were identified. Then, climatic variables representing current (1976–2013) and future (2060–2080) climate scenarios were extracted and combined with sex-ratio records to create hybrid-SDMs with which to assess the current and future suitable habitats for T. scripta. It was found that T. scripta has potential suitable habitat in 136 countries at present. Under the four climate change scenarios (ssp126, ssp245, ssp370 and ssp585) that were modeled, the distribution of T. scripta is predicted to decrease in 78–93 countries but increase in the northern hemisphere. This confirms that there is a greater likelihood that this species will increase in more developed countries. Incorporating the thermal dependence of sex ratio into hybrid-SDMs can be an important addition to detect the invasion risk of TSD species and to develop region-specific invasion management strategies to prevent and/or control invasive species such as T. scripta.
... This is achieved by using current data on species occurrence together with the environmental factors that influence their spatial distribution [48]. The MaxEnt is recognized as one of the most accurate and high-performance modelling methods for predicting species distribution [19], especially when presenceonly occurrence data are available and in cases of species with limited records of occurrence data [27]. The MaxEnt approach was chosen for this study due to its reliability, accuracy, and ability to predict species' future habitat suitability under climate change scenarios. ...
... The MaxEnt approach was chosen for this study due to its reliability, accuracy, and ability to predict species' future habitat suitability under climate change scenarios. It uses species presence-only data, both quantitative and qualitative environmental variables, produces a spatially explicit map, allows repeated runs of the model for robustness, and can be used for planning conservation measures [19]. ...
Climate change is expected to significantly alter and modify the ecological conditions of plant distribution and growth, particularly in the Mediterranean Basin, which is considered one of the hot spots for global warming. Measuring and modeling the response (sensitivity) of wild plants to current and future climate is critical to predicting future biodiversity and ecological values. Arbutus pavarii Pamp. (family Ericaceae) is a narrow endemic Libyan medicinal plant and one of the Red List species according to the IUCN that faces the threats of extinction due to habitat deterioration, overuse, and low reproductive rates. In this study, the species distribution model (SDM) approach was used to model and forecast range shifts in Arbutus pavarii under current and future climate change scenarios at various Shared Socio-economic Pathways SSP1-2.6 (lowest emission scenario) and SSP5-8.5 (highest emission scenario) for the years 2050s and 2070s. The modeling results indicate that the current highly suitable areas of the plant will decrease in the future compared to the low and moderate ones. The distribution range of A. pavarii will increase under lower emission scenarios (SSP1-2.6, 2050s) by 1.12% but under higher emission scenarios (SSP5-8.5, 2070s), the suitability of the habitat will decrease by 1.39%. Given the low reproductive fitness and the anticipated rise in air temperature, A. pavarii is likely to encounter greater challenges in its natural existence and dispersal. Lands with high elevation and precipitation are suitable for its future distribution. We recommend further ecophysiological and tree-ring studies on this species to investigate its growth-climate relationship and performance under drought conditions. The in-situ conservation of A. pavarii as well as its cultivation in the projected high and moderate habitats are recommended. Local community engagement may be beneficial in any conservation program for this species.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12862-025-02370-2.
... MaxEnt (Phillips et al. 2006) is a machine learning method accessible through 'maxnet' in R (Phillips et al. 2017). MaxEnt estimates probability distributions of a species across a user-defined landscape (Elith et al. 2011;Phillips and Dudík 2008), and automatically handles imbalanced-biassed occurrence datasets and collinearity amongst predictor variables (Ahmadi et al. 2023;De Marco Júnior and Nóbrega 2018;Feng et al. 2019). ...
... Species-specific variable importance underlies the observed differences in predicted spatial distributions. The A. hyacinthus complex required fewer topographic variables than the other species, notably omitting depth and SVF, with some evidence from cross-validations that these models were less transferable to novel locations (Elith et al. 2011). Rather than fine resolution topographic influences, this tabular species might be more affected by broader-scale geographic, oceanographic or climatic variables, in line with its high sensitivity to storm damage, coral bleaching, and disease (Ortiz et al. 2021;Pratchett et al. 2015). ...
Effective biodiversity conservation requires knowledge of species' distributions across large areas, yet prevalence data for marine sessile species is scarce, with traditional variables often unavailable at appropriate temporal and spatial resolutions. As marine organism distributions generally depend on terrain heterogeneity, topographic variables derived from digital elevation models (DEMs) can be useful proxies in ecological modelling, given appropriate spatial resolutions. Here, we use three reef‐building Acropora coral species across the Great Barrier Reef, Australia, in a case study to (1) assess high‐resolution bathymetry DEM sources for accuracy, (2) harness their derived topographic variables for regional coral species distribution models (SDMs), and (3) develop a transferable framework to produce, select and integrate multi‐resolution variables into marine spatial models. For this, we obtained and processed three distinct bathymetric digital depth models that we treat as DEMs, which are available across the GBR extent: (i) Allen Coral Atlas (ACA) at 10 m, (ii) DeepReef at 30 m and (iii) DeepReef at 100 m. We generalised the three DEMs to multiple nested spatial resolutions (15 m–120 m) and derived the same eight topographic variables to assess SDM sensitivity to bathymetry source and spatial resolution. The ACA and DeepReef DEMs shared similar vertical accuracies, each producing topographic variables relevant to marine SDMs. Slope and vector ruggedness measure (VRM), capturing hydrodynamic movement and shelter or exposure, were the most relevant variables in SDMs of all three species. Interestingly, variables at the finest resolution (15 m) were not always the most relevant for producing accurate coral SDMs, with optimal resolutions between 15 and 60 m depending on the variable type and species. Using multi‐resolution topographic variables in SDMs provided nuanced insights into the multiscale drivers of regional coral distributions. Drawing from this case study, we provide a practical and transferable framework to facilitate the adoption of multiscale SDMs for better‐informed conservation and management planning.
... We created 10,000 random pseudo-absences obtained from all coastal grid cells. We used a Maxent modeling approach, a robust machine-learning algorithm successfully applied to implement SDMs (Elith et al. 2011;Phillips et al. 2017). The model fit was evaluated using the Area Under the Curve (AUC) of the Receiver Operating Characteristic Curve, where values close to 1 indicate a perfect fit. ...
Human activities have been transporting caprellid amphipods (or “skeleton shrimps”) across the oceans for many decades. As a result, some caprellid amphipods now are among the most widespread non-indigenous species in many different coastal regions of the world. The global spread of these species is still ongoing in some cases, such as that of the successful invader Caprella mutica Schurin, 1935. Here, we report on the arrival of C. mutica in South America and modelled its environmental niche based on its current global distribution in order to evaluate future expansion risks. The species distribution model confirmed high occupancy probabilities for already invaded areas of Europe and North America with generally lower probabilities in the southern hemisphere and mean sea surface temperature as best predictor. Further, the model suggested that our discovery of C. mutica in northern Chile was made in a region that is less favorable for this species, while occupancy probabilities increased further south. Given the invasion history of C. mutica in other marine regions of the world and the more favorable oceanographic conditions, a further spread of this invader southwards along the South American Pacific coast seems very likely.
... MaxEnt is amongst the most widely used SDM approaches. The advantage of using MaxEnt is that it can be trained on presence-only data and works well with low-size samples [9,10]. However, the employment of ensemble models is recommended over relying on a single modeling approach to evaluate the role of climatic changes in causing changes in species geographic extent [11,12]. ...
As climate change accelerates, it may significantly alter species distributions and endanger many species. The use of species distribution modeling (SDM) has become increasingly vital for assessing the likely effects of climatic changes on biodiversity. This approach is especially relevant as our understanding of environmental shifts and their ecological implications deepens. SDMs are frequently employed to forecast future shifts in species’ geographic ranges, estimate extinction risks, evaluate the effectiveness of existing conservation areas, and prioritize conservation efforts. The urgency of these assessments is highlighted by the fact that the Mediterranean area is heating up 20% quicker than the universal average. Given that species have varying ecological tolerances and attributes, their biological responses to environmental changes are likely to differ significantly. This study aimed to assess the potential future distribution of three native Mediterranean species— Thymelaea hirsuta (L.) Endl., Ononis vaginalis Vahl, and Limoniastrum monopetalum (L.) Boiss.—under two GCMs of HadGEM3-GC31-LL and IPSL-CM6A-LR for the periods of 2060s and 2080s and two Shared Socioeconomic Pathway (SSP 1-2.6 and SSP5-8.5), comparing the use of MaxEnt and ensemble modelling techniques in predicting the impact of future climatic changes on these species’ distribution. The results indicated that there are high similarities and agreement between MaxEnt and the ensemble models’ outputs. The two modelling techniques exhibited excellent fits and performance. The distribution range of T. hirsuta and O. vaginalis will expand and migrate to the northwest direction of the Mediterranean coast of Egypt, while L. monopetalum will contract. The insights gained from species distribution modeling could guide future conservation efforts and promote the sustainable use of the studied species in the arid coastal environments of the Mediterranean region.
Clinical trial number Not applicable.
... However, these individual models pose risks such as overfitting, computational complexity, and interpretability challenges. For example, RF models are demanding and hard to interpret [21,22], while MaxEnt's reliance on presence-only data can cause biased predictions [23]. ANNs are "black box" models [24], and models like GBM, GLM, GAM, MARS, and CTA also risk overfitting and are difficult to interpret [25][26][27][28]. ...
Invasive Alien Plants (IAPs) pose significant risk to ecology and socio-economic settings globally. A better understanding of potential spread and its consequences is crucial for developing effective IAPs management strategies. This study aims to assess current and future distribution under climate change scenarios of highly invasive weed, Ageratina adenophora in Nepal integrating bioclimatic, topographic and anthropogenic factors. Multiple data sources were used to collect the different data and employ ensemble modeling using 11 key environmental predictors to understand the current and future distribution. At present, A. adenophora covers about 34,194. sq.km. area of Nepal. The results revealed that A. adenophora is widely spread in the Middle Mountains (24,910 sq.km.) but faces habitat reduction in this region including lower elevation like Chure (−13.23 sq.km.) under future climate change scenarios. In contrast, the habitat is expected to expand significantly in higher elevation regions including High Himalayan (+32.54 sq.km.) and High Mountain (+5898.75 sq.km). However, this expansion is accompanied by the total habitat reduction by 30–50 % (−10,744 sq.km.) in the 2050–2070 time period. The minimum temperature of the coldest month, proximity to roads, better soil organic matter and topographic factor were identified as the major drivers of its spread in Nepal. As this study used nationwide samples, ensuring diverse environmental conditions were captured, we believe the findings are more accurate and applicable on a national scale. Observing these trends, we emphasize immediate and effective mitigation measures which include- early detection, regular monitoring, prioritizing high-risk zone and development of region-specific integrated management plan to minimize the ecological and socio-economic impact of IAPS. Moreover, we recommend that future studies should incorporate land-use changes and socio-economic factors for a better understanding of invasion dynamics.
... FC refers to the mathematical transformation of the environmental variables into predictors: linear (L), quadratic (Q), product (P), and hinge (H), as well as their combinations. RM penalizes model complexity and reduced the number of parameters (Elith et al. 2011). Thus, we tested five FC combinations (L, LQ, H, LQH, and LQHP) and RM values from 0 to 5 with a step of 0.5. ...
Human-induced changes, such as climate variability and escalating anthropogenic pressures, profoundly impact species distribution, dispersal, and competitive interactions worldwide. In the Neotropical region, the expansion of rice cultivation under these conditions has facilitated the spread of species within the rice stink bug complex, notably Oebalus species. Among these species, Oebalus pugnax prevails in the United States (US) and O. poecilus in Neotropical America co-occurring with O. ypsilongriseus, which was recently introduced into the US. This study explores the determinants and predictive variables influencing the occurrence and overlap of rice stink bugs, utilizing maximum entropy species distribution modeling to project their potential distribution. Precipitation and temperature were identified as pivotal factors shaping the ecological niche of O. pugnax in the US, while sensitivity to dry spells appears instrumental in the niche specialization of Neotropical species such as O. ypsilongriseus and O. poecilus. Notably, O. pugnax shows potential for establishment in South America, whereas O. ypsilongriseus and O. poecilus exhibit potential to establish in the southeastern US. Prospects that may aggravate concerns with their potential economic losses under the current trends on global climate changes. Although bioclimate-based distribution modeling provides valuable insights into habitat suitability and species distribution, future research should focus on delineating thermal and humidity thresholds for their development, as well as elucidating interspecific relationships. These endeavors are essential for enhancing our comprehension of their distribution dynamics and assisting in the design of effective pest management strategies, taking advantage of the biological peculiarities and susceptibility to different control methods by each individual species.
... MaxEnt predicts the likelihood of the presence of a species at a specific geographic location by comparing the probability densities of environmental variables at known occurrence points with those at randomly selected points within a user-defined model background (Elith et al., 2011). The spatial scale of the model background and the approach used to select pseudo-absence points are considered as factors that can significantly influence model outcomes (Chefaoui & Lobo, 2008;VanDerWal et al., 2009;Radosavljevic & Anderson, 2014;Martin et al., 2020). ...
... (Hijmans et al., 2024). FWS occurrence data were partitioned into training data (80% of occurrence points) and test data (20% of occurrence points) (kfold=5) (Phillips et al., 2004;Phillips et al., 2006;Elith et al., 2011). Selected model performance metrics such as Receiving Operator Characteristic (ROC) computing for Area Under Curve (AUC), Cohen's kappa-statistic (k), and sensitivity were obtained for each SDM. ...
An invasion of Ceroplastes floridensis Comstock, Florida wax scale (FWS), which is not known to occur in the Philippines but was recently found infesting mangoes in Davao del Sur, is under strict pest surveillance. This study aims to determine the potential distribution of FWS in mainland Mindanao using environmental, topographic, and ecological factors under different climate scenarios: historical, RCP 4.5, and RCP 8.5 at moderate emissions. Concerning infestation foci, FWS distribution was predicted to spread south-westerly based on the constructed SDMs (AUC = 0.988–0.995) across climate projections. Average annual temperature, elevation, and the presence of seedling sources were identified as the most influential factors. This study emphasizes the importance of containment measures for FWS from infestation foci (i.e., infested mango farms) and other potential sources of infested plant propagules to prevent further spread to other mango-growing areas in Mindanao and the rest of the archipelago.
... SDMs also incorporate the physiological responses and adaptation mechanisms of species [11]. Common algorithms used in species distribution modelling include the maximum entropy model (MaxEnt) [12], random forest (RF) [13], generalized additive model (GAM) [14] generalized linear model (GLM) [15], among others. The biomod2 package in R is widely adopted for species distribution modelling [16,17]. ...
Global warming has influenced phenological shifts and community degradation in alpine herbaceous plants. However, the current distribution and future shifts of these plants and their environmental driving factors under climate change are not well known. Here, we focus on Lepidium meyenii , a medicinal herb native to the high-altitude regions of the Andes Mountains in South America, which may face habitat contraction and potential population decline in the future due to changing environmental conditions. We integrate species distribution data with a random forest model to simulate the potential habitats of L. meyenii across time and identify their key environmental drivers. We find that the most significant environmental variables that impact the distribution of L. meyenii include elevation, temperature annual range (bio7), and mean diurnal range (bio2). Since the Last Glacial Maximum, the western Andes in South America has consistently offered suitable habitat for L. meyenii . In contrast, habitat suitability in the Tibetan Plateau (TP) has varied over time, and in the futurer the TP may become a potential area for biodiversity conservation. This study will enhance our understanding of the distribution patterns of alpine herbaceous plants in response to climate change, and contributes to future biodiversity conservation efforts, the establishment of protected areas, and the sustainable management of medicinal plants as biological resources.
... was employed to evaluate Anhui musk deer's current and future habitat suitability. MaxEnt is regarded as an efficient and powerful tool for habitat suitability assessment using presence data (Elith et al. 2011). Following the default settings of Max-Ent, 10,000 background points were used to represent the available environmental conditions across the study area. ...
Context
Restricted to the densely populated regions of Eastern China, musk deer populations have dramatically declined over the past several centuries, primarily due to extensive poaching and significant habitat loss. In recent years, the Anhui musk deer has further faced an extremely high risk of extinction as a result of climate change and escalating human disturbances. However, research on the Anhui musk deer based on extensive field survey data remains limited, thereby constraining the development of effective conservation strategies.
Objectives
This study aims to evaluate the current and future habitat suitability of the Anhui musk deer, assess the vulnerability of these habitats to climate change, and delineate potential refuges that could support the population amidst climatic shifts. Additionally, we seek to identify ecological corridors that would facilitate dispersal among habitat patches under changing climate conditions.
Methods
Occurrence locations (N = 101) of the Anhui musk deer were collected using large-scale infrared camera monitoring data from 2013 to 2023. The MaxEnt model was employed to predict suitable habitat and assess the potential impacts of climate change. A gap analysis was then conducted to identify deficiencies in current nature reserves, and potential climate refuges were determined based on areas of stable habitat. Subsequently, connectivity analysis was performed to identify potential ecological corridors between habitat patches.
Results
The Anhui musk deer is highly sensitive to temperature and human disturbance. The total area of suitable habitat for Anhui musk deer is 1188.90 km² currently, with only 11 habitat patches larger than 5 km² (median patch size = 79.05 km²; and mean = 19.84 km²). Future projections suggest a continuous decline in suitable habitat, with areas decreasing to 1,052.45 km² by 2050, 841.82 km² by 2070, and 798.35 km² by 2090. Currently, nature reserves encompass only 38.52% of the suitable habitat. Climate refuges for the Anhui musk deer cover 586.24 km², but 44.38% of these areas remain unprotected. The construction of 14 potential ecological corridors connecting isolated populations to core populations could mitigate the adverse effects of climate change and protect the Anhui musk deer population.
Conclusions
Our findings indicated that the habitat area of Anhui musk deer will continue to decline under future climate conditions, and current conservation efforts are insufficient. The remaining Anhui musk deer populations in the Dabie Mountains should be considered as regional populations and managed meticulously to ensure their long-term survival under human disturbances and climate warming.
... A species distribution model was developed using the MaxEnt 3.4.1 software [33], suitable for presence-only data. Approximately 30-50 presence points were used to maximize the predictive power [34,35]. ...
Beavers (Castor fiber L.) are recognized as keystone ecological engineers who shape freshwater ecosystems by modifying hydrology, sediment dynamics, and biodiversity. Although beaver populations have recovered across Europe, including Romania, understanding the environmental factors driving their dam distribution remains limited. This study aimed to (i) characterize the physical and compositional features of beaver dams in the Râul Negru basin, Romania, (ii) model the environmental variables influencing the dam distribution using MaxEnt, and (iii) evaluate the implications for broader conservation strategies. Over a five-year survey covering 353.7 km of watercourses, 135 beaver families were identified, with an estimated population of 320–512 individuals. The dam dimensions showed strong correlations with the river slope, channel width, and wetness index. Predictive models based on LIDAR data achieved over 90% accuracy, outperforming SRTM-based models. The results reveal that topographic wetness, flow accumulation, and valley morphology are the strongest predictors of dam presence. These findings contribute to proactive beaver management strategies, highlighting areas of potential future expansion and offering data-driven guidance for balancing ecosystem restoration with human land use, contributing to the development of conservation strategies that balance ecosystem engineering by beavers with human land-use needs in Romania and across Europe.
... This selection is based on the model's superior performance with small sample sizes compared to other modeling methods (Elith et al., 2006;Pearson et al., 2007). MaxEnt, which is based on the principle of Maximum Entropy, utilizes presence-only data to predict species distribution, while aiming to estimate a probability distribution of species occurrence that aligns as closely as possible with uniformity but is still subject to environmental constraints (Elith et al., 2011). The MaxEnt model inherently includes variable interactions and can manage both continuous and categorical predictor variables. ...
... We executed the MaxEnt program with 10 replicates to evaluate the averaged results; the convergence threshold was set to 10 −5 , the maximum number of background points was equal to 10,000, a logistic output format, and "ASC" as the output file type (Merow et al. 2013). The logistic output applies a probabilistic transformation to convert raw model predictions into a standardised suitability index (0-1 scale) (Elith et al. 2011). This provides an ecologically interpretable estimate of S. alterniflora invasion potential, particularly critical for effectively communicating spatially explicit risks to stakeholders and policymakers responsible for coastal ecosystem management (Ficetola et al. 2010;Guo et al. 2013). ...
Aim: Functional traits that affect plant performance and adaptation strategies are critical for shaping the distribution of invasive plants under global change. However, quantitative assessments of functional traits on spatial niche predictions are still lacking. In this study, we aimed to understand how functional traits alter the distribution of invasive plants across native and introduced ranges. Location: Eastern USA (66-106W, 24-47N); Eastern China (104-125E, 18-43N). Taxon: Spartina alterniflora Loisel. Methods: We developed a new integrated framework using structural equations and ecological niche models to determine the causal relationships among plant functional traits in the field, presence distribution data, and environmental variables. This framework was used to predict the latitudinal distribution and spatial dynamics of Spartina alterniflora Loisel across its native (USA) and introduced (China) ranges. Results: We found that functional traits were variable and remarkably altered the distribution of S. alterniflora, especially in mid-and low-latitudes of China. Furthermore, leaf, vegetative, and sexual reproductive traits had different effects on the distribution of S. alterniflora, with approximately −2% to 15% and 10% to 40% of the distribution areas influenced by functional traits in the native and introduced ranges, respectively. Notably, sexual reproductive traits affected plant distribution more than leaf and vegetative traits. Additionally, hump-shaped relationships were observed between habitat suitability and most of the functional traits, thus demonstrating that moderately suitable areas had better plant performance. Main Conclusions: These findings suggest that plant functional traits influence the prediction of species distribution and thus need to be accounted for when performing niche modelling of invasive species. Furthermore, we suggest that global change may threaten the habitat in the native range but will improve the spatial niche of species in the introduced range. This study highlights the focus areas for conservation and prevention efforts.
... In this study, we developed a method for estimating wild boar habitat distribution using a MaxEnt model, based on geographic coordinates of observation points and publicly available environmental data, with the aim of enabling its application to other regions. The MaxEnt model, which handles presence-only data, has significant limitations due to sample selection bias [20]. To address this, we corrected for reporting bias in wild boar observations caused by geographic factors, thereby providing accurate habitat distribution predictions and contributing to disease control. ...
Control of infectious diseases in wildlife is often considered challenging due to the limited availability of information. Some infectious diseases in wildlife can also affect livestock, posing significant problems for the animal farming industry. In Japan, classical swine fever (CSF) reemerged in September 2018. Given the availability of commercial vaccines, control measures mainly involve the vaccination of domestic pigs and the distribution of oral vaccines to wild boars. Despite these efforts, the disease continues to spread, primarily due to wild boars. This transmission is further exacerbated by Japan’s challenging geography—about 66% forested—making many areas difficult to access and leading to spatial bias in surveillance. As a result, the epidemic situation cannot be fully understood, limiting the effectiveness of control measures. This study estimated wild boar distribution using a species distribution model (SDM) that incorporates geographic bias correction. Two maximum entropy (MaxEnt) models—a standard model and a reporting bias‐corrected model—were developed using wild boar observation data from Aichi Prefecture. Both models demonstrated excellent prediction accuracy (area under the curve [AUC] of 0.946 and 0.946, sensitivity of 0.868 and 0.943, and specificity of 0.999 and 0.991), with the most influential variables identified in a similar order (solar radiation in November, followed by elevation, precipitation during the wettest quarter, and solar radiation in August). While both models identified high‐probability areas in the east, the bias‐corrected model also revealed expanded high‐probability zones in the northeast. During the epidemic phases, protecting farms takes priority; however, in eradication phases, control measures must also target wild boar habitats in forested areas. By using open‐access environmental data, this modeling approach can be applied to other regions. Accurate estimation of wild boar distribution can contribute to improving wildlife disease surveillance and optimizing oral vaccine delivery strategies.
... Ecological niche models (ENM) are statistical models that use environmental data to predict the distribution of a species (such as Lu. shannoni), based on ecological suitability [33][34][35][36]. These models can be utilized for a variety of reasons, such as when a species' distribution is not well-defined and surveillance data are lacking. ...
Background
In the Americas, sand flies of the Lutzomyia genus are the vectors of pathogens of human and animal health significance. Lutzomyia shannoni is suspected to transmit vesicular stomatitis virus, along with Leishmania mexicana and Leishmania infantum (causative agents of leishmaniases). Despite the suspected vector potential of Lu. shannoni, significant knowledge gaps remain, including how ongoing climate changes could facilitate their range expansion. The objectives of this study were to predict the current and future ecological suitability of regions across North America for Lu. shannoni and to identify variables driving ecological suitability.
Methods
Occurrence records were obtained from the Global Biodiversity Information Facility, Disease Vectors Database, the National Museum of Natural History (Smithsonian Institution) and published literature on Lu. shannoni surveillance and capture. Historical climate data from 1991–2020, along with projection data for Shared Socioeconomic Pathways 2–4.5 and 3–7.0 were obtained. An additional terrestrial ecoregions layer was applied. The ecological niche model was created using maximum entropy (MaxEnt) algorithms to identify regions which currently are or may become ecologically suitable for Lu. shannoni.
Results
Currently, regions in eastern, western and southern Mexico, along with the Midwest, southeastern and eastern regions of the USA are ecologically suitable for Lu. shannoni. In the future, ecological suitability for Lu. shannoni is expected to increase slightly in the northeastern regions of the USA and in Atlantic Canada, and to decrease in the southeastern reaches of Mexico. Degree-days below 0 °C (spring and autumn), precipitation as snow (summer and winter), terrestrial ecoregions, number of frost-free days (summer), Hargreaves climatic moisture deficit (summer), degree-days above 5 °C (autumn) and Hogg’s climatic moisture index (summer) were all identified as predictors of ecological suitability.
Conclusions
The findings from this study identified climate and environmental variables driving the ecological suitability of regions for Lu. shannoni and can be used to inform public health professionals of high-risk regions for exposure at present and into the future.
Graphical abstract
... 1. Often also referred to as 'MaxEnt,' cf., e.g., Elith et al. (2011);Soley-Guardia et al. (2024). ...
This study presents the first suitability map for identifying Roman-era archaeological features in Northern Noricum (modern Lower Austria/AUT) from ca. 50 – 500 CE. It supports archaeological site prediction and assesses the vulnerability of heritage sites to natural hazards, contributing to Cultural Heritage and Resource Management. The 1161 km² area of interest includes the municipium Aelium Cetium (Sankt Pölten) and the forts Arelape (Pöchlarn), Favianis (Mautern an der Donau), and Augustianis (Traismauer), now part of the UNESCO World Heritage site „Danube Limes.“ Based on 1184 features from 551 findspots grouped into 129 sites, a machine learning-based Archaeological Predictive Model was developed using Maximum Entropy (Maxent), integrating environmental and agency-related factors. Area Under the Curve (AUC) scores indicate strong predictive performance (training AUC: 0.809; test AUC: 0.795). This was further supported by a case study confirming Roman rural settlements in high-suitability zones. Comparison with a 2021 flood event additionally demonstrated the model’s potential to assess climate-related risks to archaeological heritage.
... Hasil uji jackknife didapatkan bahwa varibel jarak dari satwa mangsa mencapai nilai AUC tertinggi yakni lebih dari 0,85 (Gambar 8). Nilai AUC dapat berkurang apabila menghilangkan variabel yang berkontribusi tinggi terhadap hasil pemodelan (Elith, Phillips, Hastie, Dudík, Chee, & Yates, 2011). Nilai AUC dari uji jackknife ini menunjukkan bahwa variabel jarak dari satwa mangsa merupakan variabel lingkungan yang dianggap penting terhadap model prediksi hanya dengan variabel tersebut (digunakan secara independen) dan memiliki informasi yang penting jika digunakan sendiri. ...
Macan tutul jawa (Panthera pardus melas) merupakan predator puncak yang berstatus dilindungi, kritis (critically endangered), dan salah satu dari 25 satwa prioritas nasional. Data dan informasi terkait penggunaan habitat dan pemodelan distribusi spasial macan tutul jawa masih sedikit. Tujuan dari penelitian ini adalah mengidentifikasi penggunaan habitat macan tutul jawa, serta membuat pemodelan distribusi spasial macan tutul jawa. Data penelitian ini dikumpulkan melalui pemasangan camera trap, metode eksplorasi, dan pengumpulan variabel lingkungan dengan metode Maximum Entropy. Analisis dilakukan menggunakan Relative Use Index, pemodelan Maximum Entropy, dan Relative Abundance Index. Berdasarkan penelitian, penggunaan habitat oleh macan tutul jawa dengan nilai paling tinggi terdapat pada tipe habitat berupa hutan alam primer, ketinggian 751-1.250 mdpl, dan kemiringan lahan curam (26-40%), dengan sedikit gangguan manusia. Variabel lingkungan yang paling berpengaruh adalah jarak dari satwa mangsa, lereng, dan ketinggian. Penelitian ini menghasilkan implikasi pengelolaan, yaitu 1) area yang dinilai sesuai bagi macan tutul jawa dapat dipertahankan keberadaannya, sedangkan area yang kurang sesuai dapat dilakukan perbaikan habitat; 2) integrasi antara kawasan konservasi dengan kawasan hutan disekitarnya penting untuk dilakukan agar populasi macan tutul jawa dapat tertampung dengan baik, 3) mitigasi konflik juga penting untuk dilakukan melalui kerja sama yang melibatkan seluruh pemangku kepentingan.
... Otros, en cambio, utilizan únicamente información de presencia, como Bioclim, Domain y MaxEnt (Rodríguez, 2011), entre otros. Diversas investigaciones destacan a MaxEnt como uno de los enfoques más eficaces en el modelado de nicho ecológico (Toranza, 2011;Elith et al., 2006;Hernández et al., 2006). Sus aplicaciones abarcan la identificación de nuevas áreas de distribución (Pearson et al., 2007), la predicción de invasiones biológicas (Ward, 2007), el diseño de planes de conservación y la evaluación de los posibles impactos del cambio climático (Levinsky et al., 2007). ...
El cambio climático es uno de los principales factores que afectan la distribución y viabilidad de numerosas especies a nivel global. Inia geoffrensis, el delfín rosado del Amazonas, es una especie clave y un bioindicador crítico de los ecosistemas fluviales, cuya distribución podría verse gravemente afectada por el cambio climático. En este estudio, se aplicó el modelado de nicho ecológico utilizando el algoritmo MaxEnt para evaluar la distribución espacial actual y futura de Inia geoffrensis en la provincia de Sucumbíos, Ecuador. Se utilizaron 65 registros de presencia y siete variables bioclimáticas actuales y proyectadas para el año 2070 bajo dos escenarios de cambio climático: SSP 245 (moderado) y SSP 585 (extremo). Los resultados indican que el hábitat adecuado de Inia geoffrensis en la región, actualmente estimado en 2110 km², se reducirá en un 72.7% bajo el escenario SSP 245 y desaparecerá completamente bajo el escenario SSP 585. En los límites políticos de la provincia de Sucumbios, el área de idoneidad alta para el presente es de 1780.64 km² y en el futuro, escenario SSP 245 tendrá una reducción del 69.67 %. La modelación reveló que la temperatura mínima del mes más frío y la estacionalidad de la temperatura son los principales factores que determinan la distribución de la especie. Estas proyecciones resaltan la vulnerabilidad del delfín rosado frente al cambio climático y la necesidad urgente de implementar estrategias de conservación.
... We used Maxent (a) for specifically delineating suitable breeding habitat using only climatic variables relevant to the breeding season and (b) for training three other Maxent models with specific occurrence data and variables relevant to all seasons for a general analysis of the possible importance of environmental variables (see next section). For a concise mathematical definition of Maxent, discussion of its application to species distribution modelling and testing of the approach see Phillips et al. (2006) and Elith et al. (2011). For modelling suitable breeding habitat we used the records made during spring and early summer only (February-June) as we assume birds are more specific at selecting their breeding habitats which are of utmost importance for population viability. ...
This study highlights the importance of incorporating temporal landscape dynamics in landslide susceptibility assessments (LSAs). Two models are compared: one integrates multi-temporal data, while the other relies solely on present conditions. Artificial Digital Elevation Models (ADEMs) for 1960, 1980, 2000, and 2020 were generated using a landscape manipulation tool to simulate phases of land degradation and rehabilitation, thereby enabling the assessment of susceptibility over time. Landslide occurrences were simulated to increase over time—primarily as a result of anthropogenic changes, such as deforestation and land use alterations—with partial stabilization following conservation efforts. Both models achieved identical AUC values of 0.97, but the blind model misclassified stable areas and missed historically unstable regions. While conventional performance metrics such as ROC curves provide insights into model accuracy, they fail to detect misclassifications arising from temporal landscape changes, leading to overestimation in some areas and underestimation in others, especially in evolving environments. This study demonstrates that neglecting temporal landscape evolution leads to flawed susceptibility maps, potentially misguiding hazard mitigation efforts. To improve LSA accuracy, the study advocates for integrating multi-temporal thematic maps and adopting performance metrics that assess temporal robustness. It emphasizes the need for a shift from a static-to-hazard paradigm to a temporally evolved susceptibility-to-hazard framework for more accurate hazard and risk predictions.
Tea [Camellia sinensis (L.) O. Kuntze], one of the most widely consumed beverages globally, is cultivated on over5.27 million ha of land, is a source of income for low-income families. However, the effect of future emissionscenarios on the suitability of areas for tea production is poorly understood, and any adverse outcome underclimate change scenarios can affect millions of households and the tea industry. We assess the current and futurehabitat suitability of tea across the globe under two shared socio-economic pathways (SSPs = SSP2.4-5 andSSP5.8-5) for 2050 and 2070 and determine the key factors that influence tea distribution. We employed anensemble of five species distribution models to assess the tea distribution. Overall, the highly suitable areas fortea production will experience some decline in 11 out of the top 20 tea-producing countries by 2050 and 2070under both SSP2-4.5 (moderate emission scenarios) and SSP5-8.5 (high emission scenarios) relative to thecurrent conditions. However, marginally suitable areas will increase by 39 % by 2050 and 72 % by 2070 underthe high-emission scenario. Some tea-producing countries such as Argentina, Malawi and Tanzania are projectedto experience a 60-78 % reduction of their highly suitable area for tea, while Iran, Rwanda, Thailand, and Turkeywill witness an 11-27 % increase by 2050. Precipitation of the warmest quarter, mean temperature of the driestquarter, soil pH, isothermality, and mean temperature of the wettest quarter are the key variables influencing thecurrent distribution of tea. Here, we provide projections of changes in the highly suitable area for the top 20 tea-producing countries in the world by 2050, which we hope will serve as baseline information for decision-makersto make evidence-informed choices. Since the present study was performed at a coarse scale, it is stronglyrecommended that future studies produce fine-scale projections with regional climate and ground-truth teaproduction data.
Aim
Invasive species pose a major threat to biodiversity and ecosystem functioning. Understanding how invasive species change their niche during invasion is fundamental for predicting their potential invasion risk in novel areas and environments. We here study the fall armyworm, FAW (Spodoptera frugiperda), one of the world's most devastating invasive pests due to its severe impact on cereal crops, posing a serious threat to agriculture. We first investigated climatic niche changes and tested for niche conservatism between native and invasive FAW ranges. We then modelled the potential FAW distribution to assess its invasion risk on a global scale, focusing on Europe and its agricultural areas.
Location
Global, with a focus on Europe.
Methods
We analysed niche overlap and niche dynamics to quantify climatic niche changes between native and invasive FAW ranges. To predict potential climatically suitable areas for FAW invasion, we used species distribution models (SDMs) calibrated on native, invasive and global (both native and invasive) occurrence datasets.
Results
Our results did not support niche conservatism between native and invasive FAW ranges. We observed a climatic niche shift and expansion in the invasive range towards warmer and drier areas. The highest potential climatic suitability for FAW invasion was identified in parts of Africa, India, China, Australia and in large areas of southwestern, central and southeastern Europe.
Main Conclusions
Given the observed niche shift and its strong migratory capacity, FAW is likely to further expand its range and establish itself in southern Europe, with potential seasonal migrations to more temperate regions driven by increased temperatures associated with ongoing climate change. Our study confirms that FAW poses a significant threat to global and European biodiversity and agriculture, highlighting the urgent need for coordinated monitoring efforts and the implementation of adaptive management strategies, including sustainable ones, to protect agricultural systems worldwide.
Climate change is transforming marine ecosystems, opening new fishing possibilities for large‐scale trawling vessels in the Arctic. This study investigates the potential for new fishing grounds to emerge in West Greenland. We employed a maximum entropy model to predict fishing suitability based on climatological and time‐invariant variables alongside public fishing vessel data. The model, validated with high accuracy, identified maximum depth, ice thickness and ice concentration as the most important predictors of fishing suitability. Results indicate a 6.2% increase in whole domain suitable fishing grounds from the 2010s to the 2040s, and a 11.4% increase from the 2010s to the 2090s. This change is driven by decreasing ice coverage, allowing extended access to the productive shelf edge. Increased fishing suitability could enhance fishing opportunities leading to increased economic benefits. However, the prolonged fishing season also raises concerns about overexploitation, ecological sustainability and sediment dispersion. Earlier access would enable trawling over carbon‐rich sediments for extended periods, leading to a resuspension of sediment‐bound carbon. Our findings highlight the need for adaptive management strategies to balance economic gains with the protection of marine ecosystems in the face of ongoing climate changes.
Undocumented and unnamed biodiversity is difficult to conserve effectively. We encountered several Kurixalus (Anura: Rhacophoridae) individuals of an unknown species identity in Fujian, China, a province with no previous records of Kurixalus. This genus of frogs, of which many new species have been described in recent years, has been the subject of much historical taxonomic confusion, largely due to conserved morphology. We sequenced two mitochondrial and one nuclear genes (1748 total bp) for three individuals from Fujian and reconstructed their phylogeny including all known lineages of Kurixalus from mainland China, and other East Asia lineages when data was available. Both Bayesian Inference and Maximum Likelihood trees consistently showed that the individuals from Fujian were most closely related to K. inexpectatus, which previously was only known from northern Zhejiang, over 830 km north of this new location. Our results help clarify the identity of the unknown Kurixalus in Fujian and the distribution of Kurixalus, particularly regarding the distribution of K. hainanus, which has previously been confused with other members of the K. odontotarsus species complex. Additionally, we created ecological niche models using Maxent for the clade including both K. inexpectatus and its sister species K. idiootocus to guide future survey efforts.
The family Pycnothelidae comprises medium to small-sized spiders that construct silk tubes under stones or logs or dig burrows in open environments. In Argentina, Acanthogonatus centralis Goloboff, 1995 and Stenoterommata platensis Holmberg, 1881 are two species commonly found in central region, associated to the mountainous grasslands or to the La Plata River Basin. The Tandilia mountainous grassland is an extended belt located southeastern Buenos Aires province, Argentina. Although some pycnothelid spiders are distributed close to this mountainous system, they are not recorded in Tandilia. Through niche modeling we attempt to evaluate the influence of bioclimatic variables in shaping the distribution of these species and explore their absence in the mountainous belt under past, present and future climatic conditions. We found that Tandilia is a non-suitable area for the establishment of the populations of these species with the exception of Stenoterommata platensis under a future scenario of climatic change.
KEYWORDS:
SDM; geography; Tandilia; Acanthogonatus; Stenoterommata
Climate change threatens global biodiversity and could drive many vertebrate species to extinction by the end of this century. In this study, we assessed how climate change may affect the distribution and conservation status of the endemic lizard Liolaemus maldonadae in the Chilean Mediterranean hotspot. We modeled the species' potential distribution using the maximum entropy algorithm and climate projections under both pessimistic and optimistic scenarios. We applied IUCN Red List criterion B2 to evaluate its conservation status. Our models predict that by 2100, the species' climatic niche will shrink by more than 30%—from 12,211.6 to 8,322.7 km2 under the pessimistic scenario, and to 3,693.7 km2 under the optimistic one. We also estimated an Extent of Occurrence of 5,131 km2 and an Area of Occupancy of just 72 km2, which support classifying the species as Endangered. Other human activities, such as mining, could further increase the extinction risk for L. maldonadae.
Cupressus funebris forests grow relatively fast and have a strong natural regeneration ability, showing great potential in carbon sequestration. Global warming has already had a significant impact on its distribution pattern. This study used the Maximum Entropy Model (MaxEnt) and the distribution data of Cupressus funebris communities to explore the contraction and expansion of the adaptive distribution of Cupressus funebris. The research results are as follows: The contemporary adaptive distribution area of Cupressus funebris is mainly located in the southern region of China, and the area of the adaptive distribution accounts for approximately 7.15% of the total land area. The main driving variables affecting the distribution of Cupressus funebris are annual precipitation, the minimum temperature of the coldest month, isothermality, temperature seasonality, carbonate content, and altitude. Among them, climate plays a dominant role in the distribution of this community. Under different carbon emission scenarios in the future, the adaptive distribution areas show an expansion trend, but most of the highly adaptive areas are shrinking and the changes are relatively significant. In the high emission pathway, the distribution area continues to expand in the north while gradually contracting in the southern regions. The community distribution shows a trend of migrating to higher latitudes and altitudes in northern regions, and there are significant non-linear characteristics in altitude migration under the scenario of intensified carbon emissions. This study provides theoretical guidance for the protection and management of Cupressus funebris forests and helps to improve the carbon sequestration capacity of the communities in the context of carbon neutrality.
Background and aims This study focused on modeling the potential distribution of Orchis anatolica, a species belonging to the Orchidaceae family. Methods As a result of field studies conducted between March and June from 2019 to 2022, presence data for this species were obtained from 30 sampling sites. Based on the collected data, the potential distribution of Orchis anatolica, which naturally occurs in the Isparta province, was modeled using the Maximum Entropy (MaxEnt) method. To determine the potential distribution, relationships between the presence data of the target species and 22 uncorrelated environmental variables including bioclimatic variables and base maps produced or digitized via geographic information systems were modeled and mapped using MaxEnt software. Results According to the potential distribution model generated for Orchis anatolica (training dataset AUC: 0.974, test dataset AUC: 0.953), the environmental variables most influencing the species' distribution were identified as site index (bonitet), bio10, age class, bio12, and terrain ruggedness classes. Conclusions This study investigated the relationships among the ecological characteristics, geographic distribution, and habitat features of Orchis anatolica, and produced a potential distribution map for the species in the Isparta region. It provides a valuable foundation for the conservation and sustainable use of Orchis diversity and genetic resources in Isparta province.
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.
How can the shape of biodiversity inform us about cities’ ecoclimatic fitness and guide their development? Can we use species as the harbingers of climatic extremes? Eco-climatically sensitive species carry information about hydroclimatic change in their distribution, fitness, and preferential gradients of habitat suitability. Conversely, environmental features outside of the species’ fitness convey information on potential ecological anomalies in response to extremes to adapt or mitigate, such as through urban parks. Here, to quantify ecosystems’ fitness, we propose a novel computational model to extract multivariate functional ecological networks and their basins, which carry the distributed signature of the compounding hydroclimatic pressures on sentinel species. Specifically, we consider butterflies and their habitat suitability (HS) to infer maximum suitability gradients that are meaningful of potential species networks and flows, with the smallest hydroclimatic resistance across urban landscapes. These flows are compared to the distribution of urban parks to identify parks’ ecological attractiveness, actual and potential connectivity, and park potential to reduce hydroclimatic impacts. The ecosystem fitness index (EFI) is novelly introduced by combining HS and the divergence of the relative species abundance (RSA) from the optimal log-normal Preston plot. In Shenzhen, as a case study, eco-flow networks are found to be spatially very extended, scale-free, and clustering for low HS gradient and EFI areas, where large water bodies act as sources of ecological corridors draining into urban parks. Conversely, parks with higher HS, HS gradients, and EFIs have small-world connectivity non-overlapping with hydrological networks. Diverging patterns of abundance and richness are inferred as increasing and decreasing with HS. HS is largely determined by temperature and precipitation of the coldest quarter and seasonality, which are critical hydrologic variables. Interestingly, a U-shape pattern is found between abundance and diversity, similar to the one in natural ecosystems. Additionally, both abundance and richness are mildly associated with park area according to a power function, unrelated to longitude but linked to the degree of urbanization or park centrality, counterintuitively. The Preston plot’s richness–abundance and abundance-rank patterns were verified to reflect the stationarity or ecological meta-equilibrium with the environment, where both are a reflection of community connectivity. Ecological fitness is grounded on the ecohydrological structure and flows where maximum HS gradients are indicative of the largest eco-changes like climate-driven species flows. These flows, as distributed stress-response functions, inform about the collective eco-fitness of communities, like parks in cities. Flow-based networks can serve as blueprints for designing ecotones that regulate key ecosystem functions, such as temperature and evapotranspiration, while generating cascading ecological benefits across scales. The proposed model, novelly infers HS eco-networks and calculates the EFI, is adaptable to diverse sensitive species and environmental layers, offering a robust tool for precise ecosystem assessment and design.
Keteleeria evelyniana Mast., which is widespread in southwestern China, is valuable for studying under different future climate scenarios to assess potential distribution shifts in response to climate warming. Understanding these changes can provide theoretical support for species conservation, rational utilization, ecological restoration, and management of K. evelyniana habitats. The Maxent model was optimized using the package of ENMeval to adjust the Regularization Multiplier (RM) and Feature Class Combinations (FC) parameters. Utilizing 221 effective distribution points and 33 environmental variables, the potential distribution of K. evelyniana in current and future climate scenarios was predicted, with the key environmental variables analyzed. The model with FC = LQ and RM = 0.5, demonstrated low complexity, minimal overfitting, and high accuracy, achieving an AUC value of 0.946 with a standard deviation of 0.011. Under the current climate conditions, 68% of the suitable areas for K. evelyniana were focused on Yunnan Province, with additional areas in western and southwestern Guizhou, southwestern Sichuan, and the southeastern Xizang Autonomous Region. In various future climate scenarios, the suitable areas for K. evelyniana gradually decreased, with a maximum reduction of 33%. Simultaneously, the centroids of these areas are expected to migrate northward by up to 33 km. Temperature was the dominant factor affecting its distribution (77.8%), whereas the effects of soil variables and altitude were significant. This study clarified the current distribution of K. evelyniana, projected the potential shifts under different future climate scenarios, and identified the main environmental factors affecting the distribution. These findings offer valuable theoretical support for the conservation, ecological restoration, and sustainable use of K. evelyniana.
Digital accessible knowledge (DAK) is of utmost importance for biodiversity conservation. The Global Biodiversity Information Facility (GBIF, www.gbif.org) is a mega data infrastructure with more than three billion and seventy million (3,070,000,000) occurrence records as of ⁰⁴ March 2025. It is by far the largest initiative assembling and sharing DAK to support scientific research, conservation, and sustainable development. We analyzed plant data published at the GBIF site in Africa to highlight the contribution of the continent to the GBIF and thereby highlight data quality issues and data gaps across taxonomic groups and geographic space. We therefore downloaded data from 17th January 2023 from the Plantae kingdom in Africa. They are available at https://doi.org/10.15468/dl.p2n6um. We achieved data treatment and analysis via R, several packages and related functions. Although Africa is home to rich biodiversity with many hotspots, the global data contribution of the continent to the GBIF (61,176,994 as of 17th January 2023) is still extremely low (2.69%). Furthermore, there are large disparities between African countries, with South Africa contributing far more than 50% of the continent’s data alone. The plant data of Africa (9,116,401 occurrence records) accounted for 14.90% of the data of the continent; this underlines enormous gaps between taxonomic groups. We noted important data loss during the process of data cleaning, clearly underlining the limited data quality from the continent; indeed, the data fitness for completeness analysis was only 50.94% of the total data records initially downloaded. Efforts for quality checks before data publication at the GBIF site are still needed across African countries. The Magnoliopsida was the dominant plant class with the highest number of records (71.07%) and the highest number of species (68.36%), followed by Liliopsida, with 22.80% of the records and 19.06% of the species. In geographic space, plant data gaps are also quite large across the continent; data completeness is greater in West Africa, Southern Africa, East Africa, and Madagascar. To account for the non-normal distribution of the data, robust correlation methods and robust mean comparison methods were used. According to the results, accessibility by rivers and roads as well as accessibility to protected areas are limiting factors for data completeness across the continent. The large multidimensional data gaps identified in this study and the important data loss noted during the data cleaning process should be prioritized in future data collection across the continent.
Premise
Many angiosperms have evolved specialized systems that promote pollination by specific taxa. Therefore, plant distributions may be limited by the local abundance of their specialist pollinators. In eastern North America, Lobelia cardinalis is thought to be pollinated solely by Archilochus colubris , the only hummingbird species found in the region. Here we tested the hypothesis that the distribution of a plant species with specialized pollination is controlled by the range and abundance of its specialist pollinator.
Methods
We investigated the importance of A. colubris abundance, sourced from eBird, as a variable in a MaxEnt species distribution model of L. cardinalis using presence data from iNaturalist. We also compared hummingbird abundance between locations of L. cardinalis and congeneric during their respective flowering periods and explored whether the flowering periods of L. cardinalis and congenerics align with the week of peak local hummingbird abundance.
Results
Unexpectedly, MaxEnt modelling did not suggest that A. colubris abundance is a key driver of the species distribution. Lobelia cardinalis habitat suitability was lowest in the absence of A. colubris and increased with increasing abundance, but habitat suitability was also low in regions where hummingbird abundance is highest. Still, hummingbird abundance at L. cardinalis locations was generally higher than most congenerics, and L. cardinalis tended to flower near the week of local peak A. colubris abundance.
Conclusions
While populations of hummingbird‐pollinated plant species may require the local presence of hummingbirds, fine‐scale variation in hummingbird abundance may not strongly influence their spatial distributions.
Various studies have shown that model performance may vary depending on the species being modelled, the study area, or the number of sampled localities, and suggest that it is necessary to assess which model is better for a particular situation. Thus, in this study we evaluate the performance of different techniques for modelling the distribution of Patagonian insects. We applied eight of the most widely used modelling methods (artificial neural networks, BIOCLIM, classification and regression trees, DOMAIN, generalized additive models, GARP, generalized linear models, and Maxent) to the distribution of ten Patagonian insect species. We compared model performance with five accuracy measures. To overcome the problem of not having reliable absence data with which to evaluate model performance, we used randomly selected pseudo-absences located outside of the polygon area defined by taxonomic experts. Our analyses show significant differences among modelling methods depending on the chosen accuracy measure. Maxent performed the best according to four out of the five accuracy measures, although its accuracy did not differ significantly from that obtained with artificial neural networks. When assessed on per species basis, Maxent was also one of the strongest performing methods, particularly for species sampled from a relatively low number of localities. Overall, our study identified four groups of modelling techniques based on model performance. The top-performing group is composed of Maxent and artificial neural networks, followed closely by the DOMAIN technique. The third group includes GARP, GAM, GLM, and CART, and the fourth best performer is the BIOCLIM technique. Although these results may allow obtaining better distributional predictions for reserve selection, it is necessary to be cautious in their use due to the provisional nature of these simulations.
Neglect of ecological knowledge is a limiting factor in the use of statistical modelling to predict species distribution. Three components are needed for statistical modelling, an ecological model concerning the ecological theory to be used or tested, a data model concerning the collection and measurement of the data, and a statistical model concerning the statistical theory and methods used. This component framework is reviewed with emphasis on ecological theory. The expected shape of a species response curve to an environmental gradient is a central assumption on which agreement has yet to be reached. The nature of the environmental predictors whether indirect variables, e.g. latitude that have no physiological impact on plants, or direct variables, e.g. temperature also influence the type of response expected. Straight-line relationships between organisms and environment are often used uncritically. Many users of canonical correlation analysis use linear (straight-line) functions to relate ordination axes to variables such as slope and aspect though this is not a necessary part of the method. Some statisticians have used straight lines for species/environment relationships without testing, when evaluating new statistical procedures. Assumptions used in one component often conflict with those in another component. Statistical models can be used to explore ecological theory. Skewed species response curves predominate contrary to the symmetric unimodal curves assumed by some statistical methods. Improvements in statistical modelling can be achieved based on ecological concepts. Examples include incorporating interspecific competition from dominant species; more proximal predictors based on water balance models and spatial autocorrelation procedures to accommodate non-equilibrium vegetation. # 2002 Elsevier Science B.V. All rights reserved.
The Violet-throated Metaltail Metallura baroni is a high altitude hummingbird endemic to south-central Ecuador currently considered globally 'Endangered'. Here we present the first detailed assessment of its distribution, ecology and conservation. We first used a maximum entropy model (Maxent model) to create a predicted distribution for this species based on very limited species occurrence data. We used this model to guide field surveys for the species between April and October 2006. We found a positive relationship between model values and species presence, indicating that the model was a useful tool to predict species occurrence and guide exploration. In the sites where the metaltail was found we gathered data on its habitat requirements, food resources and behaviour. Our results indicate that Violet-throated Metaltail is restricted to the Western Cordillera of the Andes Mountains in Azuay and Canar provinces of Ecuador, with an area of extent of less than 2,000 km(2). Deep river canyons to the north and south, lack of suitable habitat, and potential interspecific competition in the east may limit the bird's distribution. The species occurred in three distinct habitats, including Polylepis woodland, the upper edge of the montane forest, and in shrubby paramo, but we found no difference in relative abundance among these habitats. The metaltail seems to tolerate moderate human intervention in its habitats as long as some native brushy cover is maintained, We found that Brachyotum sp., Berberis sp., and Barnadesia sp. were important nectar resources. The 'Endangered' status of this species is supported due to its restricted distribution in fragmented habitats which are under increasing human pressures.
Aim To evaluate a suite of species distribution models for their utility as predictors of suitable habitat and as tools for new population discovery of six rare plant species that have both narrow geographical ranges and specialized habitat requirements.
Location The Rattlesnake Creek Terrane (RCT) of the Shasta‐Trinity National Forest in the northern California Coast Range of the United States.
Methods We used occurrence records from 25 years of US Forest Service botanical surveys, environmental and remotely sensed climate data to model the distributions of the target species across the RCT. The models included generalized linear models (GLM), artificial neural networks (ANN), random forests (RF) and maximum entropy (ME). From the results we generated predictive maps that were used to identify areas of high probability occurrence. We made field visits to the top‐ranked sites to search for new populations of the target species.
Results Random forests gave the best results according to area under the curve and Kappa statistics, although ME was in close agreement. While GLM and ANN also gave good results, they were less restrictive and more varied than RF and ME. Cross‐model correlations were the highest for species with the most records and declined with record numbers. Model assessment using a separate dataset confirmed that RF provided the best predictions of appropriate habitat. Use of RF output to prioritize search areas resulted in the discovery of 16 new populations of the target species.
Main conclusions Species distribution models, such as RF and ME, which use presence data and information about the background matrix where species do not occur, may be an effective tool for new population discovery of rare plant species, but there does appear to be a lower threshold in the number of occurrences required to build a good model.
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.
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.
Aim The main aims of this study are: (1) to test if temperature and related parameters are the primary determinants of the regional distribution of macrofungi (as is commonly recognized for plants); (2) to test if the success of modelling fungal distribution patterns depends on species and distribution characteristics; and (3) to explore the potential of using herbarium data for modelling and predicting fungal species’ distributions.
Location The study area, Norway, spans 58–71° N latitude and 4–32° E longitude, and embraces extensive ecological gradients in a small area.
Methods The study is based on 1020 herbarium collections of nine selected species of macrofungi and a set of 75 environmental predictor variables, all recorded in a 5 × 5-km grid covering Norway. Primarily, generalized linear model (GLM; logistic regression) analyses were used to identify the environmental variables that best accounted for the species’ recorded distributions in Norway. Second, Maxent analyses (using variables identified by GLM) were used to produce predictive potential distribution maps for these species.
Results Variables relating to temperature and radiation were most frequently included in the GLMs, and between 24.8% and 59.8% of the variation in single-species occurrence was accounted for. The fraction of variation explained by the GLMs ranged from 41.6% to 59.8% for species with restricted distributions, and from 24.8% to 39.3% for species with widespread/scattered and intermediate distributions. The two-step procedure of GLM followed by Maxent gave predictions with very high values for the area under the curve (0.927–0.997), and maps of potential distribution were generally credible.
Main conclusions We show that temperature is a key factor governing the distribution of macrofungi in Norway, indicating that fungi may respond strongly to global warming. We confirm that modelling success depends partly on species and distribution characteristics, notably on how the distribution relates to the extent of the study area. Our study demonstrates that the combination of GLM and Maxent may be a fruitful approach for biogeography. We conclude that herbarium data improve insight into factors that control the distributions of fungi, of particular value for research on fleshy fungi (mushrooms), which have largely cryptic life cycles.
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.
Despite their economic and environmental impacts, there have been relatively few attempts to model the distribution of invasive
ant species. In this study, the potential distribution of six invasive ant species in New Zealand are modelled using three
fundamentally different methods (BIOCLIM, DOMAIN, MAXENT). Species records were obtained from museum collections in New Zealand.
There was a significant relationship between the length of time an exotic species had been present in New Zealand and its
geographic range. This is the first time such a time lag has been described for exotic ant species, and shows there is a considerable
time lag in their spread. For example, it has taken many species several decades (40–60years) to obtain a distribution of
17–25% of New Zealand regions. For all six species, BIOCLIM performed poorly compared to the other two modelling methods.
BIOCLIM had lower AUC scores and higher omission error, suggesting BIOCLIM models under-predicted the potential distribution
of each species. Omission error was significantly higher between models fitted with all 19 climate variables compared to those
models with fewer climate variables for BIOCLIM, but not DOMAIN or MAXENT. Widespread species had a greater commission error.
A number of regions in New Zealand are predicted to be climatically suitable for the six species modelled, particularly coastal
and lowland areas of both the North and South Islands.
Logistic regression is a statistical tool widely used for predicting species’ potential distributions starting from presence/absence
data and a set of independent variables. However, logistic regression equations compute probability values based not only
on the values of the predictor variables but also on the relative proportion of presences and absences in the dataset, which
does not adequately describe the environmental favourability for or against species presence. A few strategies have been used
to circumvent this, but they usually imply an alteration of the original data or the discarding of potentially valuable information.
We propose a way to obtain from logistic regression an environmental favourability function whose results are not affected
by an uneven proportion of presences and absences. We tested the method on the distribution of virtual species in an imaginary
territory. The favourability models yielded similar values regardless of the variation in the presence/absence ratio. We also
illustrate with the example of the Pyrenean desman’s (Galemys pyrenaicus) distribution in Spain. The favourability model yielded more realistic potential distribution maps than the logistic regression
model. Favourability values can be regarded as the degree of membership of the fuzzy set of sites whose environmental conditions
are favourable to the species, which enables applying the rules of fuzzy logic to distribution modelling. They also allow
for direct comparisons between models for species with different presence/absence ratios in the study area. This makes them
more useful to estimate the conservation value of areas, to design ecological corridors, or to select appropriate areas for
species reintroductions.
Seeking more precise knowledge of avian endemism on the east slope of the Andes in Peru and Bolivia, one of the most diverse faunal regions on Earth, we used distribution models based on locality records and 10–12 uncorrelated environmental variables to map the distributions of 115 species. Both maximum-entropy and deductive models reveal three areas of endemism, broadly supporting previous assessments of endemism in the region but showing much more detail. Regions such as the southwestern Cordillera de Vilcabamba and the Río Mapacho-Yavero valley in Cusco, Peru, and the Cordillera de Apolobamba in western Bolivia support a greater richness of endemic species than has been recognized, a result likely attributable to the ability of predictive models to partially control for biases in survey effort. National-level protected areas cover ≥1,000 km2 of the ranges, or four-fifths of the ranges of species with distributions <1,000 km2, of 77% of the endemic species. However, an analysis of summed irreplaceability, which emphasizes the locations of the most narrowly distributed endemics, showed that only 18% of these critical areas are currently protected. The fine-scale maps of endemic areas are suitable for regional and local-scale conservation planning, activities that can fill current gaps in protection of many species.
We propose a new method for estimation in linear models. The ‘lasso’ minimizes the residual sum of squares subject to the sum of the absolute value of the coefficients being less than a constant. Because of the nature of this constraint it tends to produce some coefficients that are exactly 0 and hence gives interpretable models. Our simulation studies suggest that the lasso enjoys some of the favourable properties of both subset selection and ridge regression. It produces interpretable models like subset selection and exhibits the stability of ridge regression. There is also an interesting relationship with recent work in adaptive function estimation by Donoho and Johnstone. The lasso idea is quite general and can be applied in a variety of statistical models: extensions to generalized regression models and tree‐based models are briefly described.
The relative roles of environment and history in controlling large-scale species distributions are important not only theoretically, but also for forecasting range responses to climatic change. Here, we use atlas data to examine the extent to which 55 tree species fill their climatically determined potential ranges in Europe. Quantifying range filling (R/P) as realized/potential range size ratios using bioclimatic envelope modelling we find mean R/P ¼ 38.3% (±30.3% SD). Many European tree species naturalize extensively outside their native ranges, providing support for interpreting the many low R/Ps as primarily reflecting dispersal limitation. R/P increases strongly with latitudinal range centroid and secondarily with hardiness and decreases weakly with longitudinal range centroid. Hence, European tree species ranges appear strongly controlled by geographical dispersal constraints on post-glacial expansion as well as climate. Consequently, we expect European tree species to show only limited tracking of near-future climate changes.
Aim Maps of species richness are the basis for applied research and conservation planning as well as for theoretical research investigating patterns of richness and the processes shaping these patterns. The method used to create a richness map could influence the results of such studies, but differences between these methods have been insufficiently evaluated. We investigate how different methods of mapping species ranges can influence patterns of richness, at three spatial resolutions. Location California, USA. Methods We created richness maps by overlaying individual species range maps for terrestrial amphibians and reptiles. The methods we used to create ranges included: point-to-grid maps, obtained by overlaying point observations of species occurrences with a grid and determining presence or absence for each cell; expert-drawn maps; and maps obtained through species distribution modelling. We also used a hybrid method that incorporated data from all three methods. We assessed the correlation and similarity of the spatial patterns of richness maps created with each of these four methods at three different resolutions. Results Richness maps created with different methods were more correlated at lower spatial resolutions than at higher resolutions. At all resolutions, point-to-grid richness maps estimated the lowest species richness and those derived from species distribution models the highest. Expert-drawn maps and hybrid maps showed intermediate levels of richness but had different spatial patterns of species richness from those derived with the other methods. Main conclusions Even in relatively well-studied areas such as California, different data sources can lead to rather dissimilar maps of species richness. Evaluating the strengths and weaknesses of different methods for creating a richness map can provide guidance for selecting the approach that is most appropriate for a given application and region.
. Past explanations of the large disjunctions in the distribution of New Zealand's four Nothofagus species have emphasized displacement during glacial cycles followed by slow re-occupation of suitable sites, or the effects of plate tectonics coupled with ecological and/or environmental limitations to further spread. In this study the degree of equilibrium between Nothofagus distribution and environment was compared with that of other widespread tree species by statistical analysis. Generalized additive regression models were used to relate species distribution data to estimates of temperature, solar radiation, soil water deficit, atmospheric humidity, lithology and drainage. For each species, the amount of spatial patterning remaining unexplained by environment was assessed by adding a variable describing species presence/absence on adjacent plots. Results indicate that Nothofagus species occur more frequently in environments suboptimal for tree growth, i.e. having various combinations of cool temperatures, low winter solar radiation, high root-zone water deficit, low humidity, and infertile granitic substrates. Despite these demonstrated preferences, they exhibit substantially more spatial clustering which is unexplained by environment, than most other widespread tree species. Predictions formed from regressions using environment alone confirm that several major Nothofagus disjunctions are not explicable in terms of the environmental factors used in this analysis, but more likely reflect the effects of historic displacement coupled with slowness to invade forest dominated by more rapidly dispersing endomycorrhizal species. The technique used in this study for detecting residual spatial autocorrelation after fitting explanatory variables has potentially wide application in other studies where either regression or ordination techniques are used for analysis of compositional data.
To conserve biodiversity, it is necessary to understand how species are distributed and which aspects of the environment determine distributions. In large parts of the world and for the majority of species, data describing distributions are very scarce. Museums, private collections and the historical literature offer a vast source of information on distributions. Records of the occurrence of species from these sources are increasingly being captured in electronic databases and made available over the internet. These records may be very valuable in conservation efforts. However, there are a number of limitations with museum data. These limitations are dealt with in the first part of this review. Even if the limitations of museum data can be overcome, these data present a far-from-complete picture of the distributions of species. Species distribution models offer a means to extrapolate limited information in order to estimate the distributions of species over large areas. The second part of this paper reviews the challenges of developing species distribution models for use with museum data and describes some of the questions that species distribution models have been used to address. Given the rapidly increasing number of museum records of species occurrence available over the internet, a review of their usefulness in conservation and ecology is timely.
The burrowing nematode (Radopholus similis) is one of the main quarantine pests in China, and the risk of invasion posed by this nematode is becoming more and more serious with regard to the international trade being intensified day by day. It is urgent to analyse the potential geographic distribution of R. similis in China. Genetic algorithm for rule-set prediction modeling system (GARP) and maximum entropy (MAXENT), the two niche models which have been widely used to predict the potential geographic distribution of alien species, were used to predict the distribution of R. similis in China. We also presented a model comparison of the results by both threshold-dependent and threshold-independent evaluations. It has been shown that the two niche models could be used to predict the potential distribution of R. similis reliably. The potential distribution of R. similis should be constricted within the south of China, such as Hainan, Guangdong, Guangxi, Fujian, Yunnan provinces, and Taiwan of China. The MAXENT gives a better prediction than that of GARP. R. similis can be introduced to China by flowers and nursery stock's international shipping. The predicted results indicate that R. similis can occur in south coastal area of China and Yunnan Province, which are the main flower and nursery stock's import-export areas in China. Consequently, a strong quarantine program is needed at the ports of such areas to prevent the pest from being introduced to China.
Logistic regression is an important tool for wildlife habitat-selection studies, but the method frequently has been misapplied due to an inadequate understanding of the logistic model, its interpretation, and the influ- ence of sampling design. Topromote better use of this method, we review its application and interpretation under 3 sampling designs: random, case–control, and use–availability. Logistic regression is appropriate for habitat use–nonuse,studies employing,random,sampling,and can be used to directly model,the conditional,probability of use in such cases. Logistic regression also is appropriate for studies employing case–control sampling designs, but careful attention is required to interpret results correctly. Unless bias can be estimated or probability of use is small for all habitats, results of case–control studies should be interpreted as odds ratios, rather than probability of use orrelative probability of use. When data are gathered under a use–availability design, logistic regression can be used to estimate approximate odds ratios if probability of use is small, at least on average. More generally, howev- er, logistic regression is inappropriatefor modeling habitat selection in use–availability studies. In particular, using logistic regression to fit the exponential,model,of Manly et al. (2002:100) does not guarantee,maximum-likelihood estimates, valid probabilities, or valid likelihoods. We show that the resource selection function (RSF) commonly used for the exponential model is proportional to a logistic discriminant function. Thus, it may be used to rank habitats with respect to probability of use and to identify important habitat characteristics or their surrogates, but itis not guaranteed ,to be ,proportional ,to probability ,of use. Other problems ,associated with the exponential model,also are discussed. We describe,an alternative,model,based on Lancaster,and Imbens (1996) that offers a method for estimating conditional probability of use in use–availability studies. Although promising, this model fails to converge ,to a ,unique ,solution in some ,important ,situations. Further work ,is needed ,to obtain ,a robust method,that is broadly applicable to use–availability studies. JOURNAL OF WILDLIFE MANAGEMENT 68(4):774–789 Key words: bias, case–control, contaminated control, exponential model, habitat modeling, log-binomial model, logis-
The relative roles of environment and history in controlling large-scale species distributions are important not only theoretically, but also for forecasting range responses to climatic change. Here, we use atlas data to examine the extent to which 55 tree species fill their climatically determined potential ranges in Europe. Quantifying range filling (R/P) as realized/potential range size ratios using bioclimatic envelope modelling we find mean R/P = 38.3% (±30.3% SD). Many European tree species naturalize extensively outside their native ranges, providing support for interpreting the many low R/Ps as primarily reflecting dispersal limitation. R/P increases strongly with latitudinal range centroid and secondarily with hardiness and decreases weakly with longitudinal range centroid. Hence, European tree species ranges appear strongly controlled by geographical dispersal constraints on post-glacial expansion as well as climate. Consequently, we expect European tree species to show only limited tracking of near-future climate changes.
Aim The aim of this study is to analyse the distribution pattern of the botanical collecting effort in Amazonia so that it can be accounted for when interpreting phytogeographical patterns such as inferred species ranges. We also develop a mechanistic and transparent method for taking into account the bias in collecting effort when estimating likelihoods of species occurrences.
Location Amazonia, Neotropics.
Methods We utilized electronic data sets of georeferenced herbarium collections (1,063,530 in total). We plotted collecting localities (68,246 in total) on maps overlaid with 1° and 0.5° square grids, and analysed collecting effort using a geographical information system (GIS). We also drew a map of Thiessen polygons, using collecting localities as polygon centres, to visualize collecting density in a scale-independent way. We then created a ‘collecting activity landscape’ in which well-collected areas appear as peaks and poorly studied areas as valleys. We demonstrate how this surface can be utilized when estimating species distributions.
Results The data available to us confirm that botanical collecting activity is still severely biased in Amazonia. The uncollected area represents 43% of the total area of Amazonia, while another 28% is poorly collected and only 2% can be considered relatively well collected. The Thiessen polygon network represents an improvement in the presentation of collecting intensity compared with square grids.
Main conclusions The maps of botanical collecting effort in the Neotropics should be used for visually correcting phytogeographical interpretations. With the help of GIS applications the observed spatial bias in collecting effort can be utilized in estimating the likelihood of occurrence of species in a repeatable manner. These estimates, in turn, can be used for various purposes in basic and applied science as well as in decision-making. The biased collecting effort should, in the long run, be corrected by further field work in unexplored areas, which can be identified with the maps presented here.
Aim We aim to propose validated, spatially explicit hypotheses for the late Quaternary distribution of the Brazilian Atlantic forest, and thereby provide a framework for integrating analyses of species and genetic diversity in the region.
Location The Atlantic forest, stretching along the Brazilian coast.
Methods We model the spatial range of the forest under three climatic scenarios (current climate, 6000 and 21,000 years ago) with BIOCLIM and MAXENT. Historically stable areas or refugia are identified as the set of grid cells for which forest presence is inferred in all models and time projections. To validate inferred refugia, we test whether our models are matched by the current distribution of the forest and by fossil pollen data. We then investigate whether the location of inferred forest refugia is consistent with current patterns of species endemism and existing phylogeographical data.
Results Forest models agree with pollen records and predict a large area of historical forest stability in the central corridor (Bahia), as well as a smaller refuge (Pernambuco) along the Brazilian coast, matching current centres of endemism in multiple taxa and mtDNA diversity patterns in a subset of the species examined. Less historical stability is predicted in coastal areas south of the Doce river, which agrees with most phylogeographical studies in that region. Yet some widely distributed taxa show high endemism in the southern Atlantic forest. This may be due to limitations of the modelling approach, differences in ecology and dispersal capability, historical processes not contemplated by the current study or inadequacy of the available test data sets.
Main conclusions Palaeoclimatic models predict the presence of historical forest refugia in the Atlantic rain forest and suggest spatial variation in persistence of forests through the Pleistocene, predicting patterns of biodiversity in several local taxa. The results point to the need for further studies to document genetic and species endemism in the relatively poorly known and highly impacted areas of Atlantic rain forests of north‐eastern Brazil.
Wildlife surveys often seek to determine the presence or absence of species at sites. Such data may be used in population monitoring, impact assessment, and species– habitat analyses. An implicit assumption of presence/absence surveys is that if a species is not detected in one or more visits to a site, it is absent from that site. However, it is rarely if ever possible to be completely sure that a species is absent, and false negative observation errors may arise when detection probabilities are less than 1. The detectability of species in wildlife surveys is one of the most important sources of uncertainty in de-termining the proportion of a landscape that is occupied by a species. Recent studies emphasize the need to acknowledge and incorporate false negative observation error rates in the analysis of site occupancy data, although a comparative study of the range of available methods for estimating detectability and occupancy is notably absent. The motivation for this study stems from the lack of guidance in the literature about the relative merits of alternative methods for estimating detection probabilities and site occupancy proportions from presence/absence survey data. Six approaches to estimating underlying detection prob-abilities and the proportion of sites occupied from binary observation data are reviewed. These include three parametric methods based on binomial mixtures, one nonparametric approach based on mark–recapture theory, and two approaches based on simplistic as-sumptions about occupancy rates. We compare the performance of each method using simulated data for which the ''true'' underlying detection rate is known. Simulated data were realized from a beta-binomial distribution, incorporating a realistic level of variation in detection rates. Estimation methods varied in their precision and bias. The ''binomial-with-added-zeros'' mixture model, estimated by maximum likelihood, was the least biased estimator of detection probability and, therefore, occupancy rate. We provide an Excel spreadsheet to execute all of the methods reviewed. Stand-alone programs such as PRES-ENCE may be used to estimate all models including the ''binomial with added zeros'' model. Our findings lend support to the use of maximum likelihood methods in estimating site occupancy and detectability rates.
Maps of species' distributions or habitat suitability are required for many aspects of environmental research, resource management and conservation planning. These include biodiversity assessment, reserve design, habitat management and restoration, species and habitat conservation plans and predicting the effects of environmental change on species and ecosystems. The proliferation of methods and uncertainty regarding their effectiveness can be daunting to researchers, resource managers and conservation planners alike. Franklin summarises the methods used in species distribution modeling (also called niche modeling) and presents a framework for spatial prediction of species distributions based on the attributes (space, time, scale) of the data and questions being asked. The framework links theoretical ecological models of species distributions to spatial data on species and environment, and statistical models used for spatial prediction. Providing practical guidelines to students, researchers and practitioners in a broad range of environmental sciences including ecology, geography, conservation biology, and natural resources management.
1. The concept of the ecological niche relates a set of environmental variables to the fitness of species, while habitat suitability models (HSMs) relate environmental variables to the likelihood of occurrence of the species. In spite of this relationship, the concepts are weakly linked in the literature, and there is a strong need for better integration. 2. We selectively reviewed the literature for habitat suitability studies that directly addressed four common facets of niche theory: niche characteristics, niche interactions, community-wide processes and niche evolution. 3. We found that HSMs have mostly contributed to the study of niche characteristics, but the three other themes are gaining impetus. We discuss three issues that emerge from these studies: (i) commonly used environmental variables and their link with ecological niches; (ii) the causes of false absences and false presences in species data, and associated issues; (iii) the three axes of model generalization (interpolation and extrapolation): environmental, spatial and temporal. Finally, we propose a list of 12 recommendations to strengthen the use of HSMs for wildlife management. 4. Synthesis and applications . This selective review provides conservation biologists with a list of pointers to key niche-theory concepts and a wide palette of related HSM studies. It also brings together frameworks that are often separated: theoretical and applied ecology studies; botany, zoology and parasitology; and different HSM frameworks, such as Resource Selection Functions, Species Distribution Modelling, Ecological Niche Modelling, and Gradient Analysis. We hope that integration of all these slices of knowledge will improve the quality and reliability of HSM predictions.
Presence‐only data, for which there is no information on locations where the species is absent, are common in both animal and plant studies. In many situations, these may be the only data available on a species. We need effective ways to use these data to explore species distribution or species use of habitat.
Many analytical approaches have been used to model presence‐only data, some inappropriately. We provide a synthesis and critique of statistical methods currently in use to both estimate and evaluate these models, and discuss the critical importance of study design in models where only presence can be identified
Profile or envelope methods exist to characterize environmental covariates that describe the locations where organisms are found. Predictions from profile approaches are generally coarse, but may be useful when species records, environmental predictors and biological understanding are scarce.
Alternatively, one can build models to contrast environmental attributes associated with known locations with a sample of random landscape locations, termed either ‘pseudo‐absences’ or ‘available’. Great care needs to be taken when selecting random landscape locations, because the way in which they are selected determines the modelling techniques that can be applied.
Regression‐based models can provide predictions of the relative likelihood of occurrence, and in some situations predictions of the probability of occurrence. The logistic model is frequently applied, but can rarely be used directly to estimate these models; instead, case–control or logistic discrimination should be used depending on the sample design.
Cross‐validation can be used to evaluate model performance and to assess how effectively the model reflects a quantity proportional to the probability of occurrence. However, more research is needed to develop a single measure or statistic that summarizes model performance for presence‐only data.
Synthesis and applications. A number of statistical procedures are available to explore patterns in presence‐only data; the choice among them depends on the quality of the presence‐only data. Presence‐only records can provide insight into the vulnerability, historical distribution and conservation status of species. Models developed using these data can inform management. Our caveat is that researchers must be mindful of study design and the biases inherent in presence data, and be cautious in the interpretation of model predictions.
The fraction of sampling units in a landscape where a target species is present (occupancy) is an extensively used concept in ecology. Yet in many applications the species will not always be detected in a sampling unit even when present, resulting in biased estimates of occupancy. Given that sampling units are surveyed repeatedly within a relatively short timeframe, a number of similar methods have now been developed to provide unbiased occupancy estimates. However, practical guidance on the efficient design of occupancy studies has been lacking.
In this paper we comment on a number of general issues related to designing occupancy studies, including the need for clear objectives that are explicitly linked to science or management, selection of sampling units, timing of repeat surveys and allocation of survey effort. Advice on the number of repeat surveys per sampling unit is considered in terms of the variance of the occupancy estimator, for three possible study designs.
We recommend that sampling units should be surveyed a minimum of three times when detection probability is high (> 0·5 survey ⁻¹ ), unless a removal design is used.
We found that an optimal removal design will generally be the most efficient, but we suggest it may be less robust to assumption violations than a standard design.
Our results suggest that for a rare species it is more efficient to survey more sampling units less intensively, while for a common species fewer sampling units should be surveyed more intensively.
Synthesis and applications . Reliable inferences can only result from quality data. To make the best use of logistical resources, study objectives must be clearly defined; sampling units must be selected, and repeated surveys timed appropriately; and a sufficient number of repeated surveys must be conducted. Failure to do so may compromise the integrity of the study. The guidance given here on study design issues is particularly applicable to studies of species occurrence and distribution, habitat selection and modelling, metapopulation studies and monitoring programmes.
Species presence/absence surveys are commonly used in monitoring programs, metapopulation studies and habitat modelling, yet they can never be used to confirm that a species is absent from a location. Was the species there but not detected, or was the species genuinely absent? Not accounting for imperfect detection of the species leads to misleading conclusions about the status of the population under study. Here some recent modelling developments are reviewed that explicitly allow for the detection process, enabling unbiased estimation of occupancy, colonization and local extinction probabilities. The methods are illustrated with a simple analysis of presence/absence data collected on larvae and metamorphs of tiger salamander (Ambystoma tigrinum) in 2000 and 2001 from Minnesota farm ponds, which highlights that misleading conclusions can result from naive analyses that do not explicitly account for imperfect detection.
Despite some populations of European wildcat Felis silvestris in central Europe are stable or increasing, the Iberian subpopulation is in decline and is listed as ‘vulnerable’. In Portugal, little is known about wildcat populations, making conservation policies extremely difficult to define. Furthermore, the secretive behaviour of these mammals, along with low population densities, make data collection complicated. Thus, it is crucial to develop efficient analytical tools to interpret existing data for this species. In this study, we determine the home-range size and environmental factors related to wildcat spatial ecology in a Mediterranean ecosystem using a combined analysis of habitat selection and maximum entropy (Maxent) modelling. Simultaneously, we test the feasibility of using radio-tracking locations to construct an ecologically meaningful distribution model. Six wildcats were captured and tracked. The average home-range size (MCP95) was 2.28 km2 for females and 13.71 km2 for one male. The Maxent model built from radio-tracking locations indicated that the abundance of the European rabbit Oryctolagus cuniculus and limited human disturbance were the most important correlates of wildcat presence. Habitat selection analysis revealed that wildcats tend to use scrubland areas significantly more than expected by chance. A mosaic of scrublands and agricultural areas, with a higher proportion of the former, benefits wildcat presence in the study area; however, species distribution is mainly constrained by availability of prey and resting sites. The Maxent model validation with camera-trapping data indicated that highly adequate model performance. This technique may prove useful for recovering small radio-tracking datasets as it provides a new alternative for handling data and maximizing the ecological information on a target population, which can then be used for conservation planning.
Accurate modeling of geographic distributions of species is crucial to various applications in ecology and conservation. The best performing techniques often require some parameter tuning, which may be prohibitively time-consuming to do separately for each species, or unreliable for small or biased datasets. Additionally, even with the abundance of good quality data, users interested in the application of species models need not have the statistical knowledge required for detailed tuning. In such cases, it is desirable to use ‘‘default settings’’, tuned and validated on diverse datasets. Maxent is a recently introduced modeling technique, achieving high predictive accuracy and enjoying several additional attractive properties. The performance of Maxent is influenced by a moderate number of parameters. The first contribution of this paper is the empirical tuning of these parameters. Since many datasets lack information about species absence, we present a tuning method that uses presence-only data. We evaluate our method on independently collected high-quality presenceabsence data. In addition to tuning, we introduce several concepts that improve the predictive accuracy and running time of Maxent. We introduce ‘‘hinge features’ ’ that model more complex relationships in the training data; we describe a new logistic output format that gives an estimate of probability of presence; finally we explore ‘‘background sampling’’ strategies that cope with sample selection bias and decrease model-building time. Our evaluation, based on a diverse dataset of 226 species from 6 regions, shows: 1) default settings tuned on presence-only data achieve performance which is almost as good as if they had been tuned on the evaluation data itself; 2) hinge features substantially improve model
Aim Maps of species richness are the basis for applied research and conservation planning as well as for theoretical research investigating patterns of richness and the processes shaping these patterns. The method used to create a richness map could influence the results of such studies, but differences between these methods have been insufficiently evaluated. We investigate how different methods of mapping species ranges can influence patterns of richness, at three spatial resolutions.
Location California, USA.
Methods We created richness maps by overlaying individual species range maps for terrestrial amphibians and reptiles. The methods we used to create ranges included: point-to-grid maps, obtained by overlaying point observations of species occurrences with a grid and determining presence or absence for each cell; expert-drawn maps; and maps obtained through species distribution modelling. We also used a hybrid method that incorporated data from all three methods. We assessed the correlation and similarity of the spatial patterns of richness maps created with each of these four methods at three different resolutions.
Results Richness maps created with different methods were more correlated at lower spatial resolutions than at higher resolutions. At all resolutions, point-to-grid richness maps estimated the lowest species richness and those derived from species distribution models the highest. Expert-drawn maps and hybrid maps showed intermediate levels of richness but had different spatial patterns of species richness from those derived with the other methods.
Main conclusions Even in relatively well-studied areas such as California, different data sources can lead to rather dissimilar maps of species richness. Evaluating the strengths and weaknesses of different methods for creating a richness map can provide guidance for selecting the approach that is most appropriate for a given application and region.
Species distribution models (SDM) are commonly used to obtain hypotheses on either the realized or the potential distribution of species. The reliability and meaning of these hypotheses depends on the kind of absences included in the training data, the variables used as predictors and the methods employed to parameterize the models. Information about the absence of species from certain localities is usually lacking, so pseudo-absences are often incorporated to the training data. We explore the effect of using different kinds of pseudo-absences on SDM results. To do this, we use presence information on Aphodius bonvouloiri, a dung beetle species of well-known distribution. We incorporate different types of pseudo-absences to create different sets of training data that account for absences of methodological (i.e. false absences), contingent and environmental origin. We used these datasets to calibrate SDMs with GAMs as modelling technique and climatic variables as predictors, and compare these results with geographical representations of the potential and realized distribution of the species created independently. Our results confirm the importance of the kind of absences in determining the aspect of species distribution identified through SDM. Estimations of the potential distribution require absences located farther apart in the geographic and/or environmental space than estimations of the realized distribution. Methodological absences produce overall bad models, and absences that are too far from the presence points in either the environmental or the geographic space may not be informative, yielding important overestimations. GLMs and Artificial Neural Networks yielded similar results. Synthetic discrimination measures such as the Area Under the Receiver Characteristic Curve (AUC) must be interpreted with caution, as they can produce misleading comparative results. Instead, the joint examination of ommission and comission errors provides a better understanding of the reliability of SDM results.
Aim Globally, species distribution patterns in the deep sea are poorly resolved, with spatial coverage being sparse for most taxa and true absence data missing. Increasing human impacts on deep‐sea ecosystems mean that reaching a better understanding of such patterns is becoming more urgent. Cold‐water stony corals (Order Scleractinia) form structurally complex habitats (dense thickets or reefs) that can support a diversity of other associated fauna. Despite their widely accepted ecological importance, records of scleractinian corals on seamounts are patchy and simply not available for most of the global ocean. The objective of this paper is to model the global distribution of suitable habitat for stony corals on seamounts.
Location Seamounts worldwide.
Methods We compiled a database containing all accessible records of scleractinian corals on seamounts. Two modelling approaches developed for presence‐only data were used to predict global habitat suitability for seamount scleractinians: maximum entropy modelling (Maxent) and environmental niche factor analysis (ENFA). We generated habitat‐suitability maps and used a cross‐validation process with a threshold‐independent metric to evaluate the performance of the models.
Results Both models performed well in cross‐validation, although the Maxent method consistently outperformed ENFA. Highly suitable habitat for seamount stony corals was predicted to occur at most modelled depths in the North Atlantic, and in a circumglobal strip in the Southern Hemisphere between 20° and 50° S and shallower than around 1500 m. Seamount summits in most other regions appeared much less likely to provide suitable habitat, except for small near‐surface patches. The patterns of habitat suitability largely reflect current biogeographical knowledge. Environmental variables positively associated with high predicted habitat suitability included the aragonite saturation state, and oxygen saturation and concentration. By contrast, low levels of dissolved inorganic carbon, nitrate, phosphate and silicate were associated with high predicted suitability. High correlation among variables made assessing individual drivers difficult.
Main conclusions Our models predict environmental conditions likely to play a role in determining large‐scale scleractinian coral distributions on seamounts, and provide a baseline scenario on a global scale. These results present a first‐order hypothesis that can be tested by further sampling. Given the high vulnerability of cold‐water corals to human impacts, such predictions are crucial tools in developing worldwide conservation and management strategies for seamount ecosystems.
In the paper I give a brief review of the basic idea and some history and then discuss some developments since the original paper on regression shrinkage and selection via the lasso.
Species distribution models (SDMs) are numerical tools that combine observations of species occurrence or abundance with environmental estimates. They are used to gain ecological and evolutionary insights and to predict distributions across landscapes, sometimes requiring extrapolation in space and time. SDMs are now widely used across terrestrial, freshwater, and marine realms. Differences in methods between disciplines reflect both differences in species mobility and in “established use. Model realism and robustness is influenced by selection of relevant predictors and modeling method, consideration of scale, how the interplay between environmental and geographic factors is handled, and the extent of extrapolation. Current linkages between SDM practice and ecological theory are often weak, hindering progress. Remaining challenges include: improvement of methods for modeling presence-only data and for model selection and evaluation; accounting for biotic interactions; and assessing model uncertainty.
Species distribution models (habitat models) relate the occurrence or abundance of a species to environmental and/or geographical predictors that then allow predictions to be mapped across an entire region. These models are used in a range of policy settings such as managing greenhouse gases, biosecurity threats and conservation planning. Prediction errors are almost ubiquitous in habitat models. An understanding of the source, magnitude and pattern of these errors is essential if the models are to be used transparently in decision making.
This study considered the sources of errors in habitat models. It divided them into two main classes, error resulting from data deficiencies and error introduced by the specification of the model. Common and important data errors included missing covariates, and samples of species’ occurrences that were small, biased or lack absences. These affected the types of models that could be developed and the probable errors that would occur. Almost all models had missing covariates, and this introduced significant spatial correlation in the errors of the analysis.
A challenging aspect of modelling is that species’ distributions are affected by processes operating in both environmental and geographical space. We differentiated between global (aspatial) and local (spatial) errors, and discussed how they arise and what can be done to alleviate their effects.
Synthesis and applications. This study brings together statistical and ecological thinking to consider the appropriate techniques for habitat modelling. Ecological theory suggests models capable of defining optima, while allowing for interactions between variables. Statistical considerations, including impacts of data errors, suggest models that deal with multimodality and discontinuity in response surfaces. Models are typically simple approximations of the true probability surface. We suggest the use of flexible regression techniques, and explain what makes such methods superior for ecological modelling. The most robust modelling approaches are likely to be those in which care is taken to match the model with knowledge of ecology, and in which each is allowed to inform the other.
This chapter provides information on species distribution models (SDMs) and their use in conservation prioritization. In it we aim to give readers an understanding of the breadth of methods and applications of SDM, and to equip them to identify and resolve key decisions in creating a robust model. Our primary objectives are to describe the main methodological steps required to fit and evaluate an SDM, and to summarize the most common problems with SDMs in the context of spatial prioritization, pointing to methods for dealing with these problems or alternative approaches that might be more suitable. Note that in terms of scope, we ignore the wider use of SDMs to make inferences about ecological relationships only because this broader application is tangential to the particular focus of this volume.
To provide a context for these objectives we begin in sections 6.1 and 6.2 by outlining what SDMs are, their ecological bases, what they are commonly used for, and the benefits they offer for spatial prioritization. In Section 6.3 we then outline the broad classes of models, and provide a commentary on their relative utility so that a newcomer can understand how to choose a method and set of analyses appropriate for their particular data and prioritization problem. Details on evaluation and on methodological steps and problems in building SDMs are treated in Sections 6.4 and 6.5. We then provide an illustrative example of fitting and evaluating SDMs in Section 6.6. Finally, Section 6.7 includes brief comment on limitations and future directions.
Aim To determine the potential combined effects of climate change and land transformation on the modelled geographic ranges of Banksia .
Location Mediterranean climate South West Australian Floristic Region (SWAFR).
Methods We used the species distribution modelling software M axent to relate current environmental conditions to occurrence data for 18 Banksia species, and subsequently made spatial predictions using two simple dispersal scenarios (zero and universal), for three climate‐severity scenarios at 2070, taking the impacts of land transformation on species’ ranges into account. The species were chosen to reflect the biogeography of Banksia in the SWAFR.
Results Climate‐severity scenario, dispersal scenario, biogeographic distribution and land transformation all influenced the direction and magnitude of the modelled range change responses for the 18 species. The predominant response of species to all climate change scenarios was range contraction, with exceptions for some northern and widespread species. Including land transformation in estimates of modelled geographic range size for the three climate‐severity scenarios generally resulted in smaller gains and larger declines in species ranges across both dispersal scenarios. Including land transformation and assuming zero dispersal resulted, as expected, in the greatest declines in projected range size across all species. Increasing climate change severity greatly increased the risk of decline in the 18 Banksia species, indicating the critical role of mitigating future emissions.
Main conclusions The combined effects of climate change and land transformation may have significant adverse impacts on endemic Proteaceae in the SWAFR, especially under high emissions scenarios and if, as expected, natural migration is limiting. Although these results need cautious interpretation in light of the many assumptions underlying the techniques used, the impacts identified warrant a clear focus on monitoring across species ranges to detect early signs of change, and experiments that determine physiological thresholds for species in order to validate and refine the models.
An important decision in presence-only species distribution modeling is how to select background (or pseudo-absence) localities for model parameterization. The selection of such localities may influence model parameterization and thus, can influence the appropriateness and accuracy of the model prediction when extrapolating the species distribution across time and space. We used 12 species from the Australian Wet Tropics (AWT) to evaluate the relationship between the geographic extent from which pseudo-absences are taken and model performance, and shape and importance of predictor variables using the MAXENT modeling method. Model performance is lower when pseudo-absence points are taken from either a restricted or broad region with respect to species occurrence data than from an intermediate region. Furthermore, variable importance (i.e., contribution to the model) changed such that, models became increasingly simplified, dominated by just two variables, as the area from which pseudo-absence points were drawn increased. Our results suggest that it is important to consider the spatial extent from which pseudo-absence data are taken. We suggest species distribution modeling exercises should begin with exploratory analyses evaluating what extent might provide both the most accurate results and biologically meaningful fit between species occurrence and predictor variables. This is especially important when modeling across space or time—a growing application for species distributional modeling.
The availability of detailed environmental data, together with inexpensive and powerful computers, has fueled a rapid increase in predictive modeling of species environmental requirements and geographic distributions. For some species, detailed presence/absence occurrence data are available, allowing the use of a variety of standard statistical techniques. However, absence data are not available for most species. In this paper, we introduce the use of the maximum entropy method (Maxent) for modeling species geographic distributions with presence-only data. Maxent is a general-purpose machine learning method with a simple and precise mathematical formulation, and it has a number of aspects that make it well-suited for species distribution modeling. In order to investigate the efficacy of the method, here we perform a continental-scale case study using two Neotropical mammals: a lowland species of sloth, Bradypus variegatus, and a small montane murid rodent, Microryzomys minutus. We compared Maxent predictions with those of a commonly used presence-only modeling method, the Genetic Algorithm for Rule-Set Prediction (GARP). We made predictions on 10 random subsets of the occurrence records for both species, and then used the remaining localities for testing. Both algorithms provided reasonable estimates of the species’ range, far superior to the shaded outline maps available in field guides. All models were significantly better than random in both binomial tests of omission and receiver operating characteristic (ROC) analyses. The area under the ROC curve (AUC) was almost always higher for Maxent, indicating better discrimination of suitable versus unsuitable areas for the species. The Maxent modeling approach can be used in its present form for many applications with presence-only datasets, and merits further research and development.
Projections of species’ distribution under global change (climatic and environmental) are of great scientific and societal relevance. They rely on a proper understanding of how environmental drivers determine species occurrence patterns. This understanding is usually derived from an analysis of the species’ present distribution by statistical means (species distribution models). Projections based on species distribution models make several assumptions (such as constancy of limiting factors, no evolutionary adaptation to drivers, global dispersal), some of which are ecologically untenable. Also, methodological issues muddy the waters (e.g. spatial autocorrelation, collinearity of drivers). Here, I review the main shortcomings of species distribution models and species distribution projections, identify limits to their use and open a perspective on how to overcome some current obstacles. As a consequence, I caution biogeographers against making projections too light-heartedly and conservation ecologists and policy makers to be aware that there are several unresolved problems.
Modelling species distributions with presence data from atlases, museum collections and databases is challenging. In this paper, we compare seven procedures to generate pseudo-absence data, which in turn are used to generate GLM-logistic regressed models when reliable absence data are not available. We use pseudo-absences selected randomly or by means of presence-only methods (ENFA and MDE) to model the distribution of a threatened endemic Iberian moth species (Graellsia isabelae). The results show that the pseudo-absence selection method greatly influences the percentage of explained variability, the scores of the accuracy measures and, most importantly, the degree of constraint in the distribution estimated. As we extract pseudo-absences from environmental regions further from the optimum established by presence data, the models generated obtain better accuracy scores, and over-prediction increases. When variables other than environmental ones influence the distribution of the species (i.e., non-equilibrium state) and precise information on absences is non-existent, the random selection of pseudo-absences or their selection from environmental localities similar to those of species presence data generates the most constrained predictive distribution maps, because pseudo-absences can be located within environmentally suitable areas. This study shows that if we do not have reliable absence data, the method of pseudo-absence selection strongly conditions the obtained model, generating different model predictions in the gradient between potential and realized distributions.