Predicting Species Distributions Research Articles

Projecting the global potential distribution of nine Rhododendron Subgenus Hymenanthes species under different climate change scenarios

As one of China’s most treasured traditional flowers, Rhododendron Subgen. Hymenanthes is renowned worldwide for its evergreen foliage, vibrant flowers, and significant ornamental, landscaping, and economic value. However, climate change poses a serious threat to its future, leading to population declines and endangerment of some species. Despite the ecological and economic importance of Rhododendron Subgen. Hymenanthes, the future distribution of suitable habitats and the most effective strategies for its conservation and utilization remain unclear. This study employs the MaxEnt model, which is well-known for its reliability in predicting species distribution under changing environmental conditions, to predict the potential global distribution of nine species of Rhododendron Subgen. Hymenanthes. The goal is to provide a solid foundation for their conservation, cultivation management, and breeding. The results indicate that, under future climate scenarios, suitable habitat areas for four species (R. irroratum, R. agastum, R. decorum, and R. arboreum) will significantly decrease, while suitable habitats for the remaining five species (R. delavayi, R. fortunei, R. calophytum, R. simiarum, and R. wardii) will experience slight expansion. Temperature and precipitation are identified as key environmental factors influencing the growth and distribution of these species, affecting their ability to colonize new regions. The migration direction of the expanding regions for all nine species is consistent, with their centroids shifting towards the northwest. These findings provide critical insights for developing targeted conservation strategies, including identifying potential refugia and prioritizing conservation areas under future climate conditions.

Advances and Challenges in Species Ecological Niche Modeling: A Mixed Review

Species distribution modeling (SDM) is a vital tool for ecological and biogeographical research, allowing precise predictions of species distributions based on environmental variables. This study reviews the evolution of SDM techniques from 1985 to 2023, focusing on model development and applications in conservation, climate change adaptation, and invasive species management. We employed a mixed review with bibliometric and systematic element approaches using the Scopus database, analyzing 982 documents from 275 sources. The MaxEnt model emerged as the most frequently used technique, applied in 85% of the studies due to its adaptability and accuracy. Our findings highlight the increasing trend in international collaboration, particularly between China, the United Kingdom, and Brazil. The study reveals a significant annual growth rate of 11.99%, driven by technological advancements and the urgency to address biodiversity loss. Our analysis also shows that while MaxEnt remains dominant, deep learning and other advanced computational techniques are gaining traction, reflecting a shift toward integrating AI in ecological modeling. The results emphasize the importance of global cooperation and the continued evolution of SDM methodologies, projecting further integration of real-time data sources like UAVs and satellite imagery to enhance model precision and applicability in future conservation efforts.

Effects of plant crown characteristics on dispersal kernels of wind-dispersed seeds under low wind speeds

Abstract Plant crown can affect the seed dispersal process. Clarifying the influence of plant crown type on seed dispersal kernel is important for predicting species distribution. However, the impact of different plant crown types on seed dispersal has rarely been tested. To determine the effect of plant crown type on seed dispersal kernel, we conducted wind tunnel experiments in which we investigated seed dispersal average distance, range, kurtosis and skewness of 29 species with different diaspore traits (type of appendage, mass, projected area, shape index, wing loading and terminal velocity) under 3 wind speeds (2, 4 and 6 m•s-1) for seed passing through different crown types (classified based on crown size, branch density and with or without leaves). We fit functions to the seed dispersal distance and found that the Gaussian function had the best fit. Plant crown type had significant effects on dispersal kurtosis and skewness, but not on dispersal average distance and range, of seeds passing through it under low wind speed condition. Width and branch density of plant crown affected dispersal kurtosis, and leaves of plant crown influenced kurtosis and skewness. Increase in wind speed reduced the influence of planting crown on the dispersal kurtosis and skewness. Plant crown had no significant effect on average seed dispersal distance but affected seed dispersal kernel shape, so, the dispersal range of seeds through different plants is invariant while the density of dispersal varies greatly. The information could be helpful for understanding plant metapopulation dynamics.

The global population status and distribution of the Indian Swiftlet: Implications for Conservation

Abstract The current study aims to comprehend the population status and distribution of the Indian Swiftlet (Aerodramus unicolor, family: Apodidae) using primary and secondary data. We acquired the population data from the secondary literature and presence data from open source (GBIF.org). We conducted surveys to document the breeding and foraging locations of the species in the Western Ghats, West Coast and Offshore islands of Maharashtra. After estimating the population, we used the presence data to create a model predicting species distribution in current and future scenarios. The Indian Swiftlet’s current distribution is from Southwest Maharashtra to Kerala and Sri Lanka. In future, the changing climate might restrict it to the southern Western Ghats and some pockets in Sri Lanka. Burnt Island, home to the largest known colony, deserves conservation attention. We recommend population surveys and immediate conservation efforts to ensure the survival of the Indian Swiftlet endemic to India and Sri Lanka.

Can fieldwork driven by predictive species distribution models yield new rare or relevant geographic records? A case study with Neotropical snakes

AbstractUnderstanding species distribution patterns has been a major quest in biodiversity research. Due to their secretive habits and rarity, snakes have been historically underrepresented in assessments of geographic distribution range. In this work, we employ a pipeline for predictive model‐based species sampling, using Neotropical snakes as a model organism. We employ species distribution models based on verified point records for five candidate snake species of probable occurrence to Rio Grande do Sul state, Brazil: Apostolepis dimidiata (Jan 1862), Erythrolamprus aesculapii (Linnaeus 1758), Helicops leopardinus (Schlegel 1837), Lygophis meridionalis (Schenkel 1901), and Micrurus corallinus (Merrem 1820). Based on the resulting models, we conducted fieldwork on areas with higher overlap of suitable ranges and probability of new records. Our study yields a new state record of A. dimidiata to Rio Grande do Sul and highlights the usefulness of species distribution models in eliciting priority areas for faunal assessments.

Introduction to deep learning methods for multi‐species predictions

Abstract Predicting species distributions and entire communities is crucial for ecologists, to enhance our understanding of the drivers behind species distributions and community assembly and to provide quantitative data for conservation efforts. Popular species distribution models use statistical and machine learning methods but face limitations with multi‐species predictions at the community level, hindered by scalability and data imbalance sensitivity. This paper explores the potential of deep learning methods to overcome these challenges and provide more accurate multi‐species predictions. Specifically, we introduced four distinct deep learning models that use site × species community data but differ in their internal structure or on the input environmental data structure: (1) a multi‐layer perceptron (MLP) model for tabular data (e.g. in‐situ/raster climate or soil data), (2) a convolutional neural network (CNN) and (3) a vision transformer (ViT) models tailored for image data (e.g. aerial ortho‐photographs, satellite imagery), and a multimodal model that integrates both tabular and image data. We also show how adapted loss functions can address imbalance issues. We applied these deep learning models to a plant community dataset comprising 130,582 vegetation surveys encompassing 2522 species located in the French Alps. The tabular environmental data consisted of climate, terrain and soil information, while the images were derived from aerial photographs. All models achieved approximately 70% true skill statistics on hold‐out data, demonstrating high predictive capacity for community data, the multimodal model being the best performing one. Additionally, we showcased how interpretability tools can illuminate community structure as seen by deep learning models. Deep learning models offer a broad array of features for predicting entire species communities. They handle imbalance issues and accommodate various data types, from tabular datasets to images, while also being equipped with insightful interpretation tools. The versatility extends to tabular datasets and images, with no clear superiority between the two. The last hidden layers can provide valuable features for modelling other species, and the trained models can be used to support transfer learning to related tasks. The field of ecology now possesses an additional, potent tool in its arsenal that can foster basic and fundamental research.

Genomics-Informed Range Predictions Under Global Warming Reveal Reduced Adaptive Diversity Whilst Buffering Range Shifts for a Marine Snail.

Understanding the genetic basis of local adaptation in thermal performance is useful for predicting species distribution shifts under anthropogenic climate change. Many species are distributed across multiple biogeographic regions, and the uniquely adapted populations in each region may respond to future ocean warming with distinct distribution changes. In the present study, we investigated phylogeographic patterns, thermal sensitivity, and genetic differentiation in the intertidal snail Littorina brevicula along China's coast. Whole-genome sequencing results based on a newly assembled chromosome-level genome revealed two genetic lineages, with a north-south divergence that is linked to the thermal environment. Within each lineage, individuals could be further subdivided into genetic subgroups that differ at key genomic loci underpinning differences in upper heat tolerance. Heat stress drives adaptive divergence across multiple levels of organization, from the individual to the biogeographic level. Taking into account genetic diversity associated with variation in heat tolerance, a physiological species distribution model (pSDM) was applied to predict the distributions of the different genetic subgroups in response to climate change. Both northern and southern lineages were predicted to experience declines in habitat suitability under a 4°C future warming scenario, and that a genotypic subset of snails from the southern lineage may even be driven to extinction. These findings illustrate that even when a species' range is maintained, it can nonetheless experience a significant decrease in adaptive diversity as a result of climate change. The integrated approach presented here, which considered both physiological and adaptive genetic variation at the level of individuals within a biogeographical context, provided new insights into how marine species can respond to global warming.

Using vessels of opportunity for determining important habitats of bottlenose dolphins in Port Phillip Bay, south-eastern Australia.

Understanding species' critical habitat requirements is crucial for effective conservation and management. However, such information can be challenging to obtain, particularly for highly mobile, wide-ranging species such as cetaceans. In the absence of systematic surveys, alternative economically viable methods are needed, such as the use of data collected from platforms of opportunity, and modelling techniques to predict species distribution in un-surveyed areas. The present study used data collected by ecotourism and other vessels of opportunity to investigate important habitats of a small, poorly studied population of bottlenose dolphins in Port Phillip Bay, south-eastern Australia. Using 16 years of dolphin sighting location data, an ensemble habitat suitability model was built from which physical factors influencing dolphin distribution were identified. Results indicated that important habitats were those areas close to shipping channels and coastlines with these factors primarily influencing the variation in the likelihood of dolphin presence. The relatively good performance of the ensemble model suggests that simple presence-background data may be sufficient for predicting the species distribution where sighting data are limited. However, additional data from the center of Port Phillip Bay is required to further support this contention. Important habitat features identified in the study are likely to relate to favorable foraging conditions for dolphins as they are known to provide feeding, breeding, and spawning habitat for a diverse range of fish and cephalopod prey species. The results of the present study highlight the importance of affordable community-based data collection, such as ecotourism vessels, for obtaining information critical for effective management.

Climate change is expected to reduce the potential distribution of Ceiba glaziovii in Caatinga, the largest area of dry tropical forest in South America

Ecological niche modeling is a widely used tool to predict species distribution considering current, past, or future climate change scenarios across different geographic areas. Modeling scenarios allow researchers to assess the impacts of climate change on species distribution and identify priority areas for conservation. This study aimed to model the current and future potential distribution of Ceiba glaziovii under different climate change scenarios in Brazil. The MaxEnt algorithm was used to correlate species occurrence points with bioclimatic variables in current and future climate scenarios. Four General Circulation Models (GCMs) from CMIP6 were employed: BCC-CSM2-MR, CNRM-CM6-1, IPSL-CM6A-LR, and MIROC6, considering optimistic and pessimistic projections. The contribution of variables and model accuracy were assessed using the Jackknife statistical test and the Area Under the Curve (AUC) parameter. AUC values for current and future scenarios demonstrated high accuracy. The bioclimatic variables of precipitation and temperature were the main contributors to determining areas with higher habitat suitability. In the future climate scenario, there was a reduction in areas with good climatic suitability for all four GCMs, considering optimistic and pessimistic projections. Among the areas with high habitat suitability, the IPSL-CM6A-1 model in the optimistic projection showed the smallest reduction, while in the pessimistic scenario, all areas with high suitability disappeared. The species' climatic niche is expected to decrease under all tested climate change scenarios. The central areas of the Caatinga and its transition zones exhibit the highest climatic suitability in current and future scenarios and should be prioritized for the species' conservation.

Predictions of species distributions based only on models estimating future climate change are not reliable

Changes in climate and land use are the most often mentioned factors responsible for the current decline in species diversity. To reduce the effect of these factors, we need reliable predictions of future species distributions. This is usually done by utilizing species distribution models (SDMs) based on expected climate. Here we explore the accuracy of such projections: we use orchid (Orchidaceae) recordings and environmental (mainly climatic) data from the years 1901–1950 in SDMs to predict maps of potential species distributions in 1980–2014. This should enable us to compare the predictions of species distributions in 1980–2014, based on records of species distribution in the years 1901–1950, with real data in the 1980–2014 period. We found that the predictions of the SDMs often differ from reality in this experiment. The results clearly indicate that SDM predictions of future species distributions as a reaction to climate change must be treated with caution.

Unraveling key environmental drivers of spatial variation in plant functional traits: Insights from Dacrydium pectinatum de Laub. in natural communities on Hainan Island, China

Accurate predictions of species distribution and coexistence in response to environmental changes depend on a thorough understanding of how functional traits relate to environmental conditions. However, there is still limited evidence regarding the sources of functional trait variation and the environmental mechanisms that regulate these traits in tropical forests. This study focuses on Dacrydium pectinatum de Laub., a crucial and endangered species native to the tropical mountain forests of Hainan Island, China. We gathered functional trait data from 68 permanent plots of D. pectinatum situated in three regions: Bawangling (163 tree species), Diaoluoshan (127 tree species), and Jianfengling (175 tree species). Nine functional traits were measured at the species level, and their variation patterns and sources were analyzed using linear mixed-effects models and variance decomposition methods. Our findings reveal that environmental pressures cause variations in trait relationships across different site conditions at the species level. Typically, inter-specific trait variation exceeds intra-specific variation, particularly for leaf area (LA) and specific leaf area (SLA). This indicates that trait variability is influenced by how species adapt to and balance specific environmental conditions. Site conditions significantly impact trait variation, especially for SLA, leaf dry matter content (LDMC), and wood density (WD). Additionally, intraspecific variation is notable for certain traits, reflecting individual responses to environmental heterogeneity. At the community level, eight out of nine traits exhibit significant changes in community-weighted mean (CWM) across various site conditions and environmental gradients. For instance, LA's CWM increases with soil fertility, whereas SLA, leaf nitrogen content (LNC), and leaf phosphorus content (LPC) show differences between site conditions. LDMC and WD decrease significantly in high-fertility environments, suggesting a transition from resource conservation to resource acquisition strategies within communities as soil fertility rises. Geographic variables play a crucial role in trait distributions across different community levels, with elevation and soil factors also contributing significantly to trait variation. Overall, our study provides direct evidence of how environmental filtering influences plant functional trait distributions and community assembly. Future research should explore the mechanisms driving these trait-environment relationships and their responses across various ecological contexts.

Shifting trait coordination along a soil‐moisture‐nutrient gradient in tropical forests

Abstract Soil nutrients and water availability are strong drivers of tropical tree species distribution across scales. However, the physiological mechanisms underlying environmental filtering along these gradients remain incompletely understood. Previous studies mostly focused on univariate variation in structural traits, but a more integrative approach combining multiple physiological traits is needed to fully portray species functional strategies. We measured nine leaf functional traits related to trees' resource capture and hydraulic strategies for 552 individuals belonging to 21 tropical tree species across an environmental gradient in Amazonian forests. Our sampling included generalist and specialist species from terra firme (TF) and seasonally flooded (SF) forests. We tested the influence of the topographic wetness index, a proxy for soil moisture and nutrient gradients, on each trait separately and on the trait integration through multivariate indices computed from the eigenvalues of a principal component analysis on the traits of the species. Finally, we evaluated intraspecific trait variability (ITV) for generalists and specialists by calculating the coefficient of variation for each trait. Results showed that (1) the environment had a greater influence on trait syndromes than single trait variation. Moreover, (2) SF specialist species expressed a stronger leaf trait coordination than TF specialist species. Furthermore, (3) the ability of generalist species to occupy a broader range of environments was not reflected by a larger ITV than specialist species but by the capacity to change trait coordination across environments. Our work highlights the need to investigate functional strategies as multidimensional syndromes in physiological trait space to fully understand and predict species distribution along environmental gradients. Read the free Plain Language Summary for this article on the Journal blog.

Modeling of the spatial distribution of species of interest in agriculture for their conservation: case of Punica granatum L.

Abstract Modeling the spatial distribution of species is an important step in biodiversity conservation. The models used can be helpful in predicting the impacts of climate change on the geographical distribution of species and in identifying areas where they are most likely to occur. The purpose of this work was to model the spatial distribution of the pomegranate species (Punica granatum L.) in Morocco according to the principle of maximum entropy (Maxent). This modelling method is widely used in ecology and biogeography because of its ability to work with datasets, and to produce accurate predictions of species distribution. Based on agro-ecological data such as topographical factors and climatic variables and focusing on regions where pomegranate cultivation is significant, these data can be collected at different spatial and temporal scales. They are typically integrated into Geographic Information Systems (GIS) for utilization within the simulation model. The resulting model depicts the potential spatial distribution of pomegranate cultivation throughout Morocco. The model obtained agrees perfectly with the actual distribution of the species in different regions of the country, especially since it is known for its hardiness and its adaptation to variable environmental conditions. Thus, the modeling showed that other geographical areas present favorable conditions for the development of pomegranate cropping. The determination of spatial distribution constitutes a first step to predict possible evolution of the pomegranate cropping according to climate change. The importance of this process for biodiversity conservation lies in several aspects, such as the identification of areas at risk, conservation planning, and the assessment of impacts on ecosystems.

Modeling the Current and Future Distribution of Indianthus virgatus (Roxb.) Suksathan & Borchs.: AMonotypic Plant Endemic to the Western Ghats-Sri Lanka Biodiversity Hotspot.

Species distribution modeling (SDM) is an essential tool in ecology and conservation for predicting species distributions based on species presence/absence data and environmental variables. The present study aimed to understand the distribution pattern and habitat suitability of Indianthus virgatus under current and future climate change scenarios (2050 and 2070) using MaxEnt (3.4.4) and Wallace Ecological Modeling (v2.1.2) tools. The study also intended to identify key environmental predictors of I. virgatus' distribution. Species occurrence data were collected from various sources, including herbarium (online and physical), field surveys, and online databases, yielding 105 unique locations in the Western Ghats (WG) of India and Sri Lanka. We used 19 bioclimatic variables and elevation data sourced from WorldClim for modeling. The MaxEnt and Wallace models showed excellent performance in predicting the distribution of I. virgatus, with area under the curve values of 0.958 (± 0.002) and 0.93, respectively. In MaxEnt modeling, Temperature Seasonality (bio4) was the most significant environmental parameter, followed by the Precipitation of the Coldest Quarter (bio19). In contrast, the Annual Mean Temperature (bio1), Temperature Seasonality (bio4), and Annual Precipitation (bio12) were among the key contributors in Wallace EcoMod. Both the models predicted relatively lesser areas in the species' distribution range as highly suitable habitats (HSH) in India and Sri Lanka. We found divergent trends in predicting I. virgatus distributions using MaxEnt and Wallace EcoMod, particularly for future projections. Nevertheless, both models predicted significant habitat loss under future climate change scenarios, especially under RCP85, with varying degrees of suitability across India and Sri Lanka. Overall, our findings on expected habitat loss under future climate change scenarios highlight the importance of conserving I. virgatus, which has already been declared critically endangered (CR) in Sri Lanka.

Predicting species distributions in the open ocean with convolutional neural networks

Predicting species distributions in the open ocean with convolutional neural networks

Application of species distribution models in predicting the distribution of marine macrobenthos.

Species distribution models (SDMs) are valuable tools in predicting species distribution ranges and the suitable habitats, which are based on environmental conditions and species distribution data. These models encompass correlative models, mechanistic models, and mechanistic-correlative models. In the field of marine science, SDMs have been extensively used for predicting the spatial distribution patterns of various marine organisms including fish, mammals, algae, et al. However, the application of SDMs in predicting the distribution of macrobenthos remains scarce. Understanding the distribution of macrobenthos, the integral components of marine ecosystems, has significant implications for ecological conservation and resource management. We reviewed common methodologies employed in SDMs and presented case studies using different models to predict the distribution patterns of marine macrobenthos. Further, we emphasized the use of correlative and mechanistic models to analyze the impact of climate change on the spatial distribution of marine macrobenthos. Finally, we discussed the challenges and prospects associated with SDMs. With the advances in remote sensing technology and modeling techniques, SDMs are becoming increasingly pivotal in marine ecological research, which could offer a robust scientific foundation for addressing climate change and preserving marine biodiversity.

The role of species interactions in shaping the geographic pattern of ungulate abundance across African savannah

Macroecologists traditionally emphasized the role of environmental variables for predicting species distribution and abundance at large scale. While biotic factors have been increasingly recognized as important at macroecological scales, producing valuable biotic variables remains challenging and rarely tested. Capitalizing on the wealth of population density estimates available for African savannah ungulates, here we modeled species average population density at 100 × 100 km as a function of both environmental variables and proxies of biotic interactions (competition and predation) and estimated their relative contribution. We fitted a linear mixed effect model on 1043 population density estimates for 63 species of ungulates using Bayesian inference and estimated the percentage of total variance explained by environmental, anthropogenic, and biotic interactions variables. Environmental and anthropogenic variables were the main drivers of ungulate population density, with NDVI, Distance to permanent water bodies and Human population density showing the highest contribution to the variance. Nonetheless, biotic interactions altogether contributed to a quarter of the variance explained, with predation and competition having a negative effect on species density. Despite the limitations of modelling biotic interactions in macroecological studies, proxies of biotic interactions can enhance our understanding of biological patterns at broad spatial scales, uncovering novel predictors as well as enhancing the predictive power of large-scale models.

Should I stay or move? Quantifying landscape of fear to enhance environmental management of road networks in a highly transformed landscape

The development and expansion of road networks pose considerable threats to natural habitats and wildlife, fostering a landscape of fear. In addition to direct mortality caused by road collisions, road construction and maintenance often result in habitat fragmentation and loss, impeding animal movement and gene flow between populations. Mountain ungulates are already confined to fragmented habitat patches and roads can cause substantial disturbances to their critical ecological processes, such as dispersal and migration. In this study, we employed two key mountain ungulates, the wild goat (Capra aegagrus) and mouflon (Ovis gmelini), as functional models to examine how road networks impact the quantity and connectivity of natural habitats in southwestern Iran, where extensive road construction has led to significant landscape changes. We used the MaxEnt method to predict species distribution, the circuit theory to evaluate habitat connectivity, and the Spatial Road Disturbance Index (SPROADI) to assess road impacts. During the modeling process, we selected eleven important variables and employed a model parametrization strategy to identify the optimal configuration for the MaxEnt model. For SPROADI index we used three sub-indices, including traffic intensity, vicinity impact, and fragmentation grade. We then integrated the results of these analyses to identify areas with the most significant environmental impacts of roads on the coherency of the natural habitats. The findings indicate that suitable habitats for wild goats are widely distributed across the study area, while suitable habitats for mouflon are primarily concentrated in the northeastern region. Conservation gap analysis revealed that only 8% of wild goat habitats and 7% of mouflon habitats are covered by protected areas (PAs). The SPROADI map highlighted that 23% of the study area is negatively influenced by road networks. Moreover, 30.4% of highest-probability corridors for mouflon, and 25.7% for wild goat, were highly vulnerable to the impacts of roads. Our combined approach enabled us to quantitatively assess species-specific vulnerability to the impacts of heavy road networks. This study emphasizes the urgent need to address the negative effects of road networks on wildlife habitats and connectivity corridors. Our approach effectively identifies sensitive areas, which can help inform mitigation strategies and support more effective conservation planning in significantly transformed landscapes.

Neglecting biodiversity baselines in longitudinal river connectivity restoration impacts priority setting

River habitats are fragmented by barriers which impede the movement and dispersal of aquatic organisms. Restoring habitat connectivity is a primary objective of nature conservation plans with multiple efforts to strategically restore connectivity at local, regional, and global scales. However, current approaches to prioritize connectivity restoration do not typically consider how barriers spatially fragment species' populations. Additionally, we lack knowledge on biodiversity baselines to predict which species would find suitable habitat after restoring connectivity. In this paper, we asked how neglecting these biodiversity baselines in river barrier removals impacts priority setting for conservation planning. We applied a novel modelling approach combining predictions of species distributions with network connectivity models to prioritize conservation actions in rivers of the Rhine-Aare system in Switzerland. Our results show that the high number and density of barriers has reduced structural and functional connectivity across representative catchments within the system. We show that fragmentation decreases habitat suitability for species and that using expected distributions as biodiversity baselines significantly affects priority settings for connectivity restorations compared to species-agnostic metrics based on river length. This indicates that priorities for barrier removals are ranked higher within the expected distributions of species to maximize functional connectivity while barriers in unsuitable regions are given lower importance scores. Our work highlights that the joint consideration of existing barriers and species past and current distributions are critical for restoration plans to ensure the recovery and persistence of riverine fish diversity.

Machine learning ensemble model prediction of northward shift in potato cyst nematodes (Globodera rostochiensis and G. pallida) distribution under climate change conditions

Potato cyst nematodes (PCNs) are a significant threat to potato production, having caused substantial damage in many countries. Predicting the future distribution of PCN species is crucial to implementing effective biosecurity strategies, especially given the impact of climate change on pest species invasion and distribution. Machine learning (ML), specifically ensemble models, has emerged as a powerful tool in predicting species distributions due to its ability to learn and make predictions based on complex data sets. Thus, this research utilised advanced machine learning techniques to predict the distribution of PCN species under climate change conditions, providing the initial element for invasion risk assessment. We first used Global Climate Models to generate homogeneous climate predictors to mitigate the variation among predictors. Then, five machine learning models were employed to build two groups of ensembles, single-algorithm ensembles (ESA) and multi-algorithm ensembles (EMA), and compared their performances. In this research, the EMA did not always perform better than the ESA, and the ESA of Artificial Neural Network gave the highest performance while being cost-effective. Prediction results indicated that the distribution range of PCNs would shift northward with a decrease in tropical zones and an increase in northern latitudes. However, the total area of suitable regions will not change significantly, occupying 16–20% of the total land surface (18% under current conditions). This research alerts policymakers and practitioners to the risk of PCNs’ incursion into new regions. Additionally, this ML process offers the capability to track changes in the distribution of various species and provides scientifically grounded evidence for formulating long-term biosecurity plans for their control.