A comparison of approaches for modelling the occurrence of marine animals

Approaches for modelling the distribution of animals in relation to their environment can be divided into two basic types, those which use records of absence as well as records of presence and those which use only presence records. For terrestrial species, presence–absence approaches have been found to produce models with greater predictive ability than presence-only approaches. This study compared the predictive ability of both approaches for a marine animal, the harbour porpoise (Phoceoena phocoena). Using data on the occurrence of harbour porpoises in the Sea of Hebrides, Scotland, the predictive abilities of one presence–absence approach (generalised linear modelling—GLM) and three presence-only approaches (Principal component analysis—PCA, ecological niche factor analysis—ENFA and genetic algorithm for rule-set prediction—GARP) were compared. When the predictive ability of the models was assessed using receiver operating characteristic (ROC) plots, the presence–absence approach (GLM) was found to have the greatest predictive ability. However, all approaches were found to produce models that predicted occurrence significantly better than a random model and the GLM model did not perform significantly better than ENFA and GARP. The PCA had a significantly lower predictive ability than GLM but not the other approaches. In addition, all models predicted a similar spatial distribution. Therefore, while models constructed using presence–absence approaches are likely to provide the best understanding of species distribution within a surveyed area, presence-only models can perform almost as well. However, careful consideration of the potential limitations and biases in the data, especially with regards to representativeness, is needed if the results of presence-only models are to be used for conservation and/or management purposes. Guest editor: V. D. Valavanis Essential Habitat Mapping in the Mediterranean     

Mathematical contributions to link biota with environment

Hutchinson's pioneering work on the niche concept, dating from 1957, inspired the development of many ecological models. The first proposals, BIOCLIM and HABITAT, were simple geometric approximations to the shape of the niche. Despite their simplicity, they combine two features that make them adequate for the purpose of exploring the niche: they fit a predefined shape to the empirical data; and produce binary or ordinal predictions rather than continuous predictions. Thus, both explicitly delineate a precise boundary for the niche. However, the two methods present some limitations: BIOCLIM assumes that the variables are independent in their action on the species; and HABITAT, although not having that limitation, only delineates the boundaries of the niches without distinguishing levels of suitability for the species. We propose, discuss and illustrate: (1) the use of depth functions to identify regions with distinct suitability inside the niche; and (2) a general framework to assess overlap of the niches of two species, which can be applied to predictions from models that decompose the niche into a finite number of measurable regions.     

An evaluation of a GARP model as an approach to predicting the spatial distribution of non-vagile invertebrate species

One of the primary goals of any systematic, taxonomic or biodiversity study is the characterization of species distributions. While museum collection data are important for ascertaining distributional ranges, they are often biased or incomplete. The Genetic Algorithm for Rule-set Prediction (GARP) is an ecological niche modelling method based on a genetic algorithm that has been argued to provide an accurate assessment of the spatial distribution of organisms that have dispersal capabilities. The primary objective of this study is to evaluate the accuracy of a GARP model to predict the spatial distribution of a non-invasive, non-vagile invertebrate whose full distributional range was unknown. A GARP predictive model based on seven environmental parameters and 42 locations known from historical museum records for species of the trapdoor spider genus Promyrmekiaphila was produced and subsequently used as a guide for ground truthing the model. The GARP model was neither a significant nor an accurate predictor of spider localities and was outperformed by more simplistic BIOCLIM and GLM models. The isolated nature of Promyrmekiaphila populations mandates that environmental layers and their respective resolutions are carefully chosen for model production. Our results strongly indicate that, for modelling the spatial distribution of low vagility organisms, one should employ a modelling method whose results are more conducive to interpretation than models produced by a 'black box' algorithm such as GARP.     

Novel methods improve prediction of species' distributions from occurrence data

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.     

Presence-absence versus presence-only modelling methods for predicting bird habitat suitability

Habitat suitability models can be generated using methods requiring information on species presence or species presence and absence. Knowledge of the predictive performance of such methods becomes a critical issue to establish their optimal scope of application for mapping current species distributions under different constraints. Here, we use breeding bird atlas data in Catalonia as a working example and attempt to analyse the relative performance of two methods: the Ecological Niche factor Analysis (ENFA) using presence data only and Generalised Linear Models (GLM) using presence/absence data. Models were run on a set of forest species with similar habitat requirements, but with varying occurrence rates (prevalence) and niche positions (marginality). Our results support the idea that GLM predictions are more accurate than those obtained with ENFA. This was particularly true when species were using available habitats proportionally to their suitability, making absence data reliable and useful to enhance model calibration. Species marginality in niche space was also correlated to predictive accuracy, i.e. species with less restricted ecological requirements were modelled less accurately than species with more restricted requirements. This pattern was irrespective of the method employed. Models for wide‐ranging and tolerant species were more sensitive to absence data, suggesting that presence/absence methods may be particularly important for predicting distributions of this type of species. We conclude that modellers should consider that species ecological characteristics are critical in determining the accuracy of models and that it is difficult to predict generalist species distributions accurately and this is independent of the method used. Being based on distinct approaches regarding adjustment to data and data quality, habitat distribution modelling methods cover different application areas, making it difficult to identify one that should be universally applicable. Our results suggest however, that if absence data is available, methods using this information should be preferably used in most situations.     

Predicting the distribution of non‐vagile taxa: a response to McNyset and Blackburn (2006) and re‐evaluation of Stockman et al. (2006)

This paper addresses the issues raised by McNyset and Blackburn (2006 ) in their response to Stockman et al. (2006 ). Re‐evaluation of our published GARP analyses by McNyset and Blackburn showed that a much improved ecological niche model is obtained for predicting the distribution of the trapdoor spider genus Promyrmekiaphila in central/northern California. The improved niche model results in a substantially reduced omission error rate and a predictive model comparable to models obtained using other methods (GLM and BIOCLIM). However, the improved GARP models have a high commission error rate (> 0.75); consequently, the inferences regarding difficulties in modelling non‐vagile taxa drawn by Stockman et al. remain valid. Finally, we discuss other relatively minor criticisms of our study raised by McNyset and Blackburn and issues related to the peer review of our original paper.     

Predicting the potential distribution of invasive ring-necked parakeets Psittacula krameri in northern Belgium using an ecological niche modelling approach

The threat to biodiversity due to invasive alien species is considered second only to that of habitat loss. Given the large number of species that are currently invading ecosystems all over the world, we need to distinguish invaders with minor effects from those with large effects in order to prioritize management efforts. Ecological niche models can be used to predict the potential distribution of an invasive species from occurrence records and environmental data layers. We used the Ecological Niche Factor Analysis (ENFA), a presence-only predictive modelling approach, to describe the invasive ring-necked parakeets’ realized niche and to identify areas suitable for the parakeet in northern Belgium. ENFA proved to be a robust and reliable modelling technique, able to gauge the ecological requirements of an invasive species without the need to include historical information on the starting point of the invasion. ENFA shows that the parakeets tend to occupy relatively rare habitats compared to the main environmental conditions in northern Belgium, although they show some tolerance for environmental conditions inside parks and forests. The general distribution of the ring-necked parakeet is governed primarily by the amount of older forest patches, parks and built-up area in the landscape—reflecting the parakeets’ need for suitable nesting cavities and its reliance upon urban areas to forage. Our resulting habitat suitability maps show that the parakeets have ample room to further increase their range in northern Belgium. Our results indicate some concern for increased competition between parakeets and the nuthatches, native cavity nesters known to suffer from competition with parakeets, as some regions known as nuthatch strongholds are highly likely to be invaded by the parakeets.     

Predicting distributions of Mexican birds using ecological niche modelling methods

We tested the utility of the modelling program Genetic Algorithm for Rule-set Prediction (GARP) for modelling ecological niches to make accurate predictions of geographical distributions for 25 bird species across Mexico. Specimen-based point-occurrence data were entered into the algorithm in the form of geographical coordinates, and related to digitized maps of environmental variables, including mean annual precipitation, elevation, mean annual temperature, and potential vegetation. Two Mexican states were used as test areas by withholding their points from model construction; these points were later overlaid on predictions to measure model performance. Statistically, most models (7890%) were significantly more powerful than random models in predicting occurrences in test states; model failures were most often due to low sample size for testing, rather than an inability to model distributions of particular species. The success of this test indicates that ecological niche modelling approaches such as GARP provide a promising tool for exploring a broad range of questions in ecology, biogeography and conservation.     

Assessment of Predictive Habitat Models for Bighorn Sheep in California's Peninsular Ranges

ABSTRACT We developed predictive habitat models for a bighorn sheep (Ovis Canadensis) population in the Peninsular Ranges of southern California, USA, using 2 Geographic Information System modeling techniques, Ecological Niche Factor Analysis (ENFA) and Genetic Algorithm for Rule-set Production (GARP). We used >16,000 Global Positioning System locations from 34 animals in 5 subpopulations to develop and test ENFA and GARP models, and we then compared these models to each other and to the expert-based model presented in the United States Fish and Wildlife Service's Recovery Plan for this population. Based on a suite of evaluation methods, we found both ENFA and GARP to provide useful predictions of habitat; however, models developed with GARP appeared to have higher predictive power. Habitat delineations resulting from GARP models were similar to the expert-based model, affirming that the expert-based model provided a useful delineation of bighorn sheep habitat in the Peninsular Ranges. In addition, all 3 models identified continuous bighorn sheep habitat from the northern to southern extent of our study area, indicating that the Recovery Plan's recommendation of maintaining habitat connectivity throughout the range is an appropriate goal.     

LIVES: a new habitat modelling technique for predicting the distribution of species’ occurrences using presence-only data based on limiting factor theory

Predictive modelling techniques using presence-only data have attracted increasing attention because they can provide information on species distributions and their potential habitat for conservation and ecosystem management. However, the existing predictive modelling techniques have several limitations. Here, we propose a novel predictive modelling technique, Limiting Variable and Environmental Suitability (LIVES), for predicting the distributions and potential habitats of species using presence-only data. It is based on limiting factor theory, which postulates that the occurrence of a species is only determined by the factor that most limits its distribution. LIVES predicts the suitability of a candidate grid cell for a species in terms of limiting environmental factor. It also predicts the most limiting factor or the potential limiting factor at the grid cell. The environmental factors can be climatic, geological, biological and any other relevant environmental factors, whether quantitative or qualitative. The predicted habitats consist of the current distribution of the species and the potentially suitable areas for the species where there is currently no record of occurrence. We also compare several properties of LIVES and other predictive modelling techniques. On the basis of 1,000 simulations, the average predictions of LIVES are more accurate than the two other commonly used modelling techniques (BIOCLIM and DOMAIN) for presence-only data.     

Generalized models vs. classification tree analysis: Predicting spatial distributions of plant species at different scales

Abstract. Statistical models of the realized niche of species are increasingly used, but systematic comparisons of alternative methods are still limited. In particular, only few studies have explored the effect of scale in model outputs. In this paper, we investigate the predictive ability of three statistical methods (generalized linear models, generalized additive models and classification tree analysis) using species distribution data at three scales: fine (Catalonia), intermediate (Portugal) and coarse (Europe). Four Mediterranean tree species were modelled for comparison. Variables selected by models were relatively consistent across scales and the predictive accuracy of models varied only slightly. However, there were slight differences in the performance of methods. Classification tree analysis had a lower accuracy than the generalized methods, especially at finer scales. The performance of generalized linear models also increased with scale. At the fine scale GLM with linear terms showed better accuracy than GLM with quadratic and polynomial terms. This is probably because distributions at finer scales represent a linear sub‐sample of entire realized niches of species. In contrast to GLM, the performance of GAM was constant across scales being more data‐oriented. The predictive accuracy of GAM was always at least equal to other techniques, suggesting that this modelling approach is more robust to variations of scale because it can deal with any response shape.     

Modelling habitat-suitability using museum collections: an example with three sympatric Apodemus species from the Alps

Aim, Location Although the alpine mouse Apodemus alpicola has been given species status since 1989, no distribution map has ever been constructed for this endemic alpine rodent in Switzerland. Based on redetermined museum material and using the Ecological-Niche Factor Analysis (ENFA), habitat-suitability maps were computed for A. alpicola , and also for the co-occurring A. flavicollis and A. sylvaticus .
Methods In the particular case of habitat suitability models, classical approaches (GLMs, GAMs, discriminant analysis, etc.) generally require presence and absence data. The presence records provided by museums can clearly give useful information about species distribution and ecology and have already been used for knowledge-based mapping. In this paper, we apply the ENFA which requires only presence data, to build a habitat-suitability map of three species of Apodemus on the basis of museum skull collections.
Results Interspecific niche comparisons showed that A. alpicola is very specialized concerning habitat selection, meaning that its habitat differs unequivocally from the average conditions in Switzerland, while both A. flavicollis and A. sylvaticus could be considered as 'generalists' in the study area.
Main conclusions Although an adequate sampling design is the best way to collect ecological data for predictive modelling, this is a time and money consuming process and there are cases where time is simply not available, as for instance with endangered species conservation. On the other hand, museums, herbariums and other similar institutions are treasuring huge presence data sets. By applying the ENFA to such data it is possible to rapidly construct a habitat suitability model. The ENFA method not only provides two key measurements regarding the niche of a species (i.e. marginality and specialization), but also has ecological meaning, and allows the scientist to compare directly the niches of different species.     

蒙古黄芪潜在分布区预测的多模型比较

根据蒙古黄芪（Astragalus membranaceus（Fisch.）Bge.var.mongholicus（Bge.）Hsiao）123个样本点数据和19个环境数据,采用4种生态位模型对蒙古黄芪在中国的潜在适生区进行综合分析,并采用受试者工作特征曲线ROC和Kappa统计量,比较不同模型的预测效果。结果显示：4个模型预测精度良好,一致性显著。AUC值均达到0.8以上,Kappa值均达到0.6以上;其中DOMAIN模型的AUC值和Kappa值均最大,说明该模型的预测精度最佳,预测结果最稳定。潜在适生区的预测结果发现,GARP模型预测的最适宜区范围最广;MAXENT和BIOCLIM模型预测结果较为相似;DOMAIN模型预测结果比较分散。4个模型预测结果均表明西北一带可以作为蒙古黄芪栽培引种的主要产区。蒙古黄芪潜在适生区主要分布于中国北纬33°以北地区;最适宜区主要分布于甘肃、宁夏、陕西、山西、河北和内蒙古等地区。     

Using ecological niche modelling to evaluate niche stability in deep time

Aim To investigate relative niche stability in species responses to various types of environmental pressure (biotic and abiotic) on geological time‐scales using the fossil record. Location The case study focuses on Late Ordovician articulate brachiopods of the Cincinnati Arch in eastern North America. Methods Species niches were modelled for a suite of fossil brachiopod species based on five environmental variables inferred from sedimentary parameters using GARP and Maxent . Niche stability was assessed by comparison of (1) the degree of overlap of species distribution models developed for a time‐slice and those generated by projecting niche models of the previous time‐slice onto environmental layers of a second time‐slice using GARP and Maxent , (2) Schoener’s D statistic, and (3) the similarity of the contribution of each environmental parameter within Maxent niche models between adjacent time‐slices. Results Late Ordovician brachiopod species conserved their niches with high fidelity during intervals of gradual environmental change but responded to inter‐basinal species invasions through niche evolution. Both native and invasive species exhibited similar levels of niche evolution in the invasion and post‐invasion intervals. Niche evolution was related mostly to decreased variance within the former ecological niche parameters rather than to shifts to new ecospace. Main conclusions Although the species examined exhibited morphological stasis during the study interval, high levels of niche conservatism were observed only during intervals of gradual environmental change. Rapid environmental change, notably inter‐basinal species invasions, resulted in high levels of niche evolution among the focal taxa. Both native and invasive species responded with similar levels of niche evolution during the invasion interval and subsequent environmental reorganization. The assumption of complete niche conservatism frequently employed in ecological niche modelling (ENM) analyses to forecast or hindcast species geographical distributions is more likely to be accurate for climate change studies than for invasive species analyses over geological time‐scales.     

An assessment of the effect of data partitioning on the performance of modelling algorithms for habitat suitability for ticks

A comparison of the performance of five modelling methods using presence/absence (generalized additive models, discriminant analysis) or presence-only (genetic algorithm for rule-set prediction, ecological niche factor analysis, Gower distance) data for modelling the distribution of the tick species Boophilus decoloratus (Koch, 1844) (Acarina: Ixodidae) at a continental scale (Africa) using climate data was conducted. This work explicitly addressed the usefulness of clustering using the normalized difference vegetation index (NDVI) to split original records and build partial models for each region (cluster) as a method of improving model performance. Models without clustering have a consistently lower performance (as measured by sensitivity and area under the curve [AUC]), although presence/absence models perform better than presence-only models. Two cluster-related variables, namely, prevalence (commonness of tick records in the cluster) and marginality (the relative position of the climate niche occupied by the tick in relation to that available in the cluster) greatly affect the performance of each model (P < 0.05). Both sensitivity and AUC are better for NDVI-derived clusters where the tick is more prevalent or its marginality is low. However, the total size of the cluster or its fragmentation (measured by Shannon's evenness index) did not affect the performance of models. Models derived separately for each cluster produced the best output but resulted in a patchy distribution of predicted occurrence. The use of such a method together with weighting procedures based on prevalence and marginality as derived from populations at each cluster produced a slightly lower predictive performance but a better estimation of the continental distribution of the tick. Therefore, cluster-derived models are able to effectively capture restricting conditions for different tick populations at a regional level. It is concluded that data partitioning is a powerful method with which to describe the climate niche of populations of a tick species, as adapted to local conditions. The use of this methodology greatly improves the performance of climate suitability models.     

Evaluating models to assess the distribution of Buxus balearica in southern Spain

Question: Which is the best model to predict the habitat distribution of Buxus balearica Lam. in southern Spain? Location: Málaga and Granada, Spain, across an area of 38 180 km². Methods: Prediction models based on 17 environmental variables were tested. Six methods were compared: multivariate adaptive regression spline (MARS), maximum entropy approach to modelling species' distributions (Maxent), two generic algorithms based on environmental metrics dissimilarity (BIOCLIM and DOMAIN), Genetic Algorithm for Rule‐set Prediction (GARP), and supervised learning methods based on generalized linear classifiers (support vector machines, SVMs). To test the predictive power of the models we used the Kappa index. Results: Maxent most accurately predicted the habitat distribution of B. balearica, followed by MARS models. The other models tested yielded lower accuracy values. A comparison of the predictive power of the models revealed that climate variables made the highest contributions among the environmental variables studied. The variables that made the lowest contributions were the insolation models. To examine the sensitivity of the models to a reduction in the number of variables, a test showed that accuracy of over 0.90 was maintained by applying just three climatic variables (spring rainfall, mean temperature of the warmest month, and mean temperature of the coldest month). Maps derived from the algorithms of all models tested coincided well with the known distribution of the species. Conclusions: Model habitat prediction is a preliminary step towards highlighting areas of high habitat suitability of B. balearica. These data support the results of previous research, which show that MaxEnt is the best technique for modelling species distributions with small sample sizes.     

The effect of sample size and species characteristics on performance of different species distribution modeling methods

Species distribution models should provide conservation practioners with estimates of the spatial distributions of species requiring attention. These species are often rare and have limited known occurrences, posing challenges for creating accurate species distribution models. We tested four modeling methods (Bioclim, Domain, GARP, and Maxent) across 18 species with different levels of ecological specialization using six different sample size treatments and three different evaluation measures. Our assessment revealed that Maxent was the most capable of the four modeling methods in producing useful results with sample sizes as small as 5, 10 and 25 occurrences. The other methods compensated reasonably well (Domain and GARP) to poorly (Bioclim) when presented with datasets of small sample sizes. We show that multiple evaluation measures are necessary to determine accuracy of models produced with presence-only data. Further, we found that accuracy of models is greater for species with small geographic ranges and limited environmental tolerance, ecological characteristics of many rare species. Our results indicate that reasonable models can be made for some rare species, a result that should encourage conservationists to add distribution modeling to their toolbox.     

Using n‐dimensional hypervolumes for species distribution modelling: A response to Qiao et al. ()

Hypervolume approaches are used to quantify functional diversity and quantify environmental niches for species distribution modelling. Recently, Qiao et al. ( 2016 ) criticized our geometrical kernel density estimation (KDE) method for measuring hypervolumes. They used a simulation analysis to argue that the method yields high error rates and makes biased estimates of fundamental niches. Here, we show that (a) KDE output depends in useful ways on dataset size and bias, (b) other species distribution modelling methods make equally stringent but different assumptions about dataset bias, (c) simulation results presented by Qiao et al. ( 2016 ) were incorrect, with revised analyses showing performance comparable to other methods, and (d) hypervolume methods are more general than KDE and have other benefits for niche modelling. As a result, our KDE method remains a promising tool for species distribution modelling.     

应用多个生态位模型预测福寿螺在中国的潜在适生区

高危性外来入侵种福寿螺严重危害我国的农业生产、生态系统完整性和人体健康.为制定有效的防控策略提供科学依据,本研究通过选取最适的生态位模型以预测福寿螺在我国的潜在适生区.结合福寿螺在我国的337条分布记录和年均温、年降水量等19个生物气候变量数据,本文采用MaxEnt、GARP、BIOCLIM和DOMAIN等4种生态位模型分别模拟预测了福寿螺在我国的潜在适生区,并利用受试者工作特征曲线(ROC)和Kappa统计量分析比较不同模型的预测效果.结果表明: 4种模型均能较好地模拟福寿螺在我国的分布,其中MaxEnt模型的模拟准确度最高(受试者工作特征曲线下的面积AUC=0.955±0.004,Kappa=0.845±0.017),其次是GARP和DOMAIN,准确度相对较小的是BIOCLIM,但其平均AUC也达0.898±0.017,平均Kappa值为0.771±0.025.MaxEnt模型的预测结果显示,福寿螺的潜在适生区主要分布在30° N以南地区,但其中也有部分地区地处30°N以北.适生区面积占国土面积的13.2%,广东、广西、湖南、重庆、浙江和福建沿海地区具有高度潜在入侵风险.本研究可以为福寿螺的科学防控提供参考,并且对大尺度上外来水生生物的适生区预测具有一定的借鉴意义.     

Does the interpolation accuracy of species distribution models come at the expense of transferability?

Model transferability (extrapolative accuracy) is one important feature in species distribution models, required in several ecological and conservation biological applications. This study uses 10 modelling techniques and nationwide data on both (1) species distribution of birds, butterflies, and plants and (2) climate and land cover in Finland to investigate whether good interpolative prediction accuracy for models comes at the expense of transferability – i.e. markedly worse performance in new areas. Models’ interpolation and extrapolation performance was primarily assessed using AUC (the area under the curve of a receiver characteristic plot) and Kappa statistics, with supplementary comparisons examining model sensitivity and specificity values. Our AUC and Kappa results show that extrapolation to new areas is a greater challenge for all included modelling techniques than simple filling of gaps in a well‐sampled area, but there are also differences among the techniques in the degree of transferability. Among the machine‐learning modelling techniques, MAXENT, generalized boosting methods (GBM), and artificial neural networks (ANN) showed good transferability while the performance of GARP and random forest (RF) decreased notably in extrapolation. Among the regression‐based methods, generalized additive models (GAM) and generalized linear models (GLM) showed good transferability. A desirable combination of good prediction accuracy and good transferability was evident for three modelling techniques: MAXENT, GBM, and GAM. However, examination of model sensitivity and specificity revealed that model types may differ in their tendencies to either increased over‐prediction of presences or absences in extrapolation, and some of the methods show contrasting changes in sensitivity vs specificity (e.g. ANN and GARP). Among the three species groups, the best transferability was seen with birds, followed closely by butterflies, whereas reliable extrapolation for plant species distribution models appears to be a major challenge at least at this scale. Overall, detailed knowledge of the behaviour of different techniques in various study settings and with different species groups is of utmost importance in predictive modelling.