MaxEnt's parameter configuration and small samples: are we paying attention to recommendations? A systematic review

Long-distance influence of the Rhône River plume on the marine benthic ecosystem: Integrating descriptive ecology and predictive modelling

The Gulf of Lions (GoL) is among the most productive areas of the Mediterranean Sea, with the Rhône River contributing with as much as 90% of the liquid and solid materials (including anthropogenic chemicals) reaching the area. In this paper, we assessed whether classical descriptive ecology and MaxEnt predictive species distribution modelling were able to provide complementary information when analysing the long-distance influence of the river discharges on the GoL benthic ecosystem. Samples were collected in August 2014 from 12 stations covering the sedimentary plain of the deep submarine delta, from the Gulf of Fos to Gruissan. Sediments were mostly muddy with a high organic carbon and low P and N contents first decreasing and then increasing from east to west. The same pattern occurred for chlorophyll-a, particulate organic carbon and sea surface temperature, and was overall correlated with metal and pollutant contents derived from agricultural, port, urban and industrial sources driven by Rhône outputs. We observed a typical deltaic succession in the benthos, showing a relatively low diversity and including polychaetes (Sternaspis scutata) and holothurians (Oestergrenia digitata) known to be indicators of high sedimentation rates. Overall, benthos showed an inversed pattern regarding environmental variables, an evident consequence of the Rhône River influence. The suitability of some species was either positively or negatively correlated with some of the environmental variables, producing species-specific predicted distribution patterns, with the highest amount of information allowing to predict distributions being mainly provided by organic pollutants. Even with a limited number of available samples, our integrated approach reveals to be a very robust tool to highlight hidden patterns and contributes to improve our knowledge on how river-mediated anthropogenic discharges may influence biodiversity distribution and functional patterns in marine benthic ecosystems.

Environmental impact assessment of development projects improved by merging species distribution and habitat connectivity modelling

Environmental impact assessment (EIA) is performed to limit potential impacts of development projects on species and ecosystem functions. However, the methods related to EIA actually pay little attention to the landscape-scale effects of development projects on biodiversity. In this study we proposed a methodological framework to more properly address the landscape-scale impacts of a new stadium project in Lyon (France) on two representative mammal species exemplary for the endemic fauna, the red squirrel and the Eurasian badger. Our approach combined species distribution model using Maxent and landscape functional connectivity model using Graphab at two spatial scales to assess habitat connectivity before and after development project implementation. The development project had a negative impact on landscape connectivity: overall habitat connectivity (PC index) decreased by -6.8% and -1.8% and the number of graph components increased by +60.0% and +17.6% for the red squirrel and the European badger respectively, because some links that formerly connected habitat patches were cut by the development project. Changes affecting landscape structure and composition emphasized the need to implement appropriate avoidance and reduction measures. Our methodology provides a useful tool both for EIA studies at each step of the way to support decision-making in landscape conservation planning. The method could be also developed in the design phase to compare the effectiveness of different avoidance or mitigation measures and resize them if necessary to maximize habitat connectivity.

Improving Species Distribution Modelling of freshwater invasive species for management applications

Freshwater ecosystems rank among the most endangered ecosystems in the world and are under increasing threat from aquatic invasive species (AIS). Understanding the range expansion of AIS is key for mitigating their impacts. Most approaches rely on Species Distribution Models (SDMs) to predict the expansion of AIS, using mainly environmental variables, yet ignore the role of human activities in favouring the introduction and range expansion of AIS. In this study, we use five SDM algorithms (independently and in ensemble) and two accuracy measures (TSS, AUC), combined with a null modelling approach, to assess the predictive performance of the models and to quantify which predictors (environmental and anthropogenic from the native and introduced regions) best explain the distribution of nine freshwater invasive species (including fish, arthropods, molluscs, amphibians and reptiles) in a large island (Great Britain), and which species characteristics affect model performance. Our results show that the distribution of invasive species is difficult to predict by SDMs, even in cases when TSS and AUC model accuracy values are high. Our study strongly advocates the use of null models for testing SDMs performance and the inclusion of information from the native area and a variety of both human-related and environmental predictors for a more accurate modelling of the range expansion of AIS. Otherwise, models that only include climatic variables, or rely only on standard accuracy measures or a single algorithm, might result in mismanagement of AIS.

One-class land-cover classification using MaxEnt: the effect of modelling parameterization on classification accuracy

Multiple-class land-cover classification approaches can be inefficient when the main goal is to classify only one or a few classes. Under this scenario one-class classification algorithms could be a more efficient alternative. Currently there are several algorithms that can fulfil this task, with MaxEnt being one of the most promising. However, there is scarce information regarding parametrization for performing land-cover classification using MaxEnt. In this study we aimed to understand how MaxEnt parameterization affects the classification accuracy of four different land-covers (i.e., built-up, irrigated grass, evergreen trees and deciduous trees) in the city of Santiago de Chile. We also evaluated if MaxEnt manual parameterization outperforms classification results obtained when using MaxEnt default parameters setting. To accomplish our objectives, we generated a set of 25,344 classification maps (i.e., 6,336 for each assessed land-cover), which are based on all the potential combination of 12 different classes of features restrictions, four regularization multipliers, four different sample sizes, three training/testing proportions, and 11 thresholds for generating the binary maps. Our results showed that with a good parameterization, MaxEnt can effectively classify different land covers with kappa values ranging from 0.68 for deciduous trees to 0.89 for irrigated grass. However, the accuracy of classification results is highly influenced by the type of land-cover being classified. Simpler models produced good classification outcomes for homogenous land-covers, but not for heterogeneous covers, where complex models provided better outcomes. In general, manual parameterization improves the accuracy of classification results, but this improvement will depend on the threshold used to generate the binary map. In fact, threshold selection showed to be the most relevant factor impacting the accuracy of the four land-cover classification. The number of sampling points for training the model also has a positive effect on classification results. However, this effect followed a logarithmic distribution, showing an improvement of kappa values when increasing the sampling from 40 to 60 points, but showing only a marginal effect if more than 60 sampling points are used. In light of these results, we suggest testing different parametrization and thresholds until satisfactory kappa or other accuracy metrics values are achieved. Our results highlight the huge potential that MaxEnt has a as a tool for one-class classification, but a good understanding of the software settings and model parameterization is needed to obtain reliable results.

Using the past to contextualize anthropogenic impacts on the present and future distribution of an endemic Caribbean mammal

Island species are difficult to conserve because they face the synergy of climate change, invasive species, deforestation, and increasing human population densities in areas where land mass is shrinking. The Caribbean island of Hispaniola presents particular challenges because of geopolitical complexities that span 2 countries and hinder coordinated management of species across the island. We employed species distribution modeling to evaluate the impacts of climatic change and anthropogenic activities on the distribution of an endemic mammal of conservation concern, the Hispaniolan solenodon (Solenodon paradoxus). We aggregated occurrence points for this poorly known species for the Last Glacial Maximum (LGM) and the present (1975-2016) based on museum collections, online biodiversity databases, and new field surveys. We quantified degree of overlap between periods and scenarios with Schoener's D. Through a conservation paleobiology lens, we found that over time humans played an increasing role in shaping the distribution of S. paradoxus, thus, providing a foundation for developing conservation strategies on appropriate spatiotemporal scales. Human population density was the single most important predictor of S. paradoxus occurrence. Densities >166 people/km² corresponded to a near-zero probability of occurrence. Models that accounted for climate but not anthropogenic variables falsely identified suitable habitat in Haiti, where on-the-ground surveys confirm habitat is unavailable. Climate-only models also significantly overestimated the potential for habitat connectivity between isolated populations. Our work highlights that alternative fates for S. paradoxus in the Anthropocene exist across the political border between the Dominican Republic and Haiti due to the fundamentally different economic and political realities of each country. Relationships in the fossil record confirm that Hispaniola's sociopolitical boundary is not biologically significant but instead represents one imposed on the island's fauna in the past 500 years by colonial activity. Our approach reveals how a paleontological perspective can contribute to concrete management insights.

Machine learning of large-scale spatial distributions of wild turkeys with high-dimensional environmental data

Species distribution modeling often involves high-dimensional environmental data. Large amounts of data and multicollinearity among covariates impose challenges to statistical models in variable selection for reliable inferences of the effects of environmental factors on the spatial distribution of species. Few studies have evaluated and compared the performance of multiple machine learning (ML) models in handling multicollinearity. Here, we assessed the effectiveness of removal of correlated covariates and regularization to cope with multicollinearity in ML models for habitat suitability. Three machine learning algorithms maximum entropy (MaxEnt), random forests (RFs), and support vector machines (SVMs) were applied to the original data (OD) of 27 landscape variables, reduced data (RD) with 14 highly correlated covariates being removed, and 15 principal components (PC) of the OD accounting for 90% of the original variability. The performance of the three ML models was measured with the area under the curve and continuous Boyce index. We collected 663 nonduplicated presence locations of Eastern wild turkeys (Meleagris gallopavo silvestris) across the state of Mississippi, United States. Of the total locations, 453 locations separated by a distance of ≥2 km were used to train the three ML algorithms on the OD, RD, and PC data, respectively. The remaining 210 locations were used to validate the trained ML models to measure ML performance. Three ML models had excellent performance on the RD and PC data. MaxEnt and SVMs had good performance on the OD data, indicating the adequacy of regularization of the default setting for multicollinearity. Weak learning of RFs through bagging appeared to alleviate multicollinearity and resulted in excellent performance on the OD data. Regularization of ML algorithms may help exploratory studies of the effects of environmental factors on the spatial distribution and habitat suitability of wildlife.

Throwing light on dark diversity of vascular plants in China: predicting the distribution of dark and threatened species under global climate change

Background

As global climate change accelerates, ecologists and conservationists are increasingly investigating changes in biodiversity and predicting species distribution based on species observed at sites, but rarely consider those plant species that could potentially inhabit but are absent from these areas (i.e., the dark diversity and its distribution). Here, we estimated the dark diversity of vascular plants in China and picked up threatened dark species from the result, and applied maximum entropy (MaxEnt) model to project current and future distributions of those dark species in their potential regions (those regions that have these dark species).

Methods

We used the Beals probability index to estimate dark diversity in China based on available species distribution information and explored which environmental variables had significant impacts on dark diversity by incorporating bioclimatic data into the random forest (RF) model. We collected occurrence data of threatened dark species (Eucommia ulmoides, Liriodendron chinense, Phoebe bournei, Fagus longipetiolata, Amentotaxus argotaenia, and Cathaya argyrophylla) and related bioclimatic information that can be used to predict their distributions. In addition, we used MaxEnt modeling to project their distributions in suitable areas under future (2050 and 2070) climate change scenarios.

Results

We found that every study region’s dark diversity was lower than its observed species richness. In these areas, their numbers of dark species are ranging from 0 to 215, with a generally increasing trend from western regions to the east. RF results showed that temperature variables had a more significant effect on dark diversity than those associated with precipitation. The results of MaxEnt modeling showed that most threatened dark species were climatically suitable in their potential regions from current to 2070.

Discussions

The results of this study provide the first ever dark diversity patterns concentrated in China, even though it was estimated at the provincial scale. A combination of dark diversity and MaxEnt modeling is an effective way to shed light on the species that make up the dark diversity, such as projecting the distribution of specific dark species under global climate change. Besides, the combination of dark diversity and species distribution models (SDMs) may also be of value for ex situ conservation, ecological restoration, and species invasion prevention in the future.

Mapping wader biodiversity along the East Asian-Australasian flyway

Background and goal

The study is conducted to facilitate conservation of migratory wader species along the East Asian-Australasian Flyway, particularly to 1) Identify hotspots of wader species richness along the flyway and effectively map how these might change between breeding, non-breeding and migratory phases; 2) Determine if the existing network of protected areas (PA) is sufficient to effectively conserve wader biodiversity hotspots along the EAAF; 3) Assess how species distribution models can provide complementary distribution estimates to existing BirdLife range maps.

Methods

We use a species distribution modelling (SDM) approach (MaxEnt) to develop temporally explicit individual range maps of 57 migratory wader species across their annual cycle, including breeding, non-breeding and migratory phases, which in turn provide the first biodiversity hotspot map of migratory waders along the EAAF for each of these phases. We assess the protected area coverage during each migration period, and analyse the dominant environmental drivers of distributions for each period. Additionally, we compare model hotspots to those existing range maps of the same species obtained from the BirdLife Internationals’ database.

Results

Our model results indicate an overall higher and a spatially different species richness pattern compared to that derived from a wader biodiversity hotspot map based on BirdLife range maps. Field observation records from the eBird database for our 57 study species confirm many of the hotspots revealed by model outputs (especially within the Yellow Sea coastal region), suggesting that current richness of the EAAF may have been underestimated and certain hotspots overlooked. Less than 10% of the terrestrial zones area (inland and coastal) which support waders are protected and, only 5% of areas with the highest 10% species richness is protected.

Main conclusions

The study results suggest the need for new areas for migratory wader research and conservation priorities including Yellow Sea region and Russian far-East. It also suggests a need to increase the coverage and percentage of current PA network to achieve Aichi Target 11 for Flyway countries, including giving stronger consideration to the temporal dynamics of wader migration.

Identifying long-term stable refugia for relict plant species in East Asia

Today East Asia harbors many “relict” plant species whose ranges were much larger during the Paleogene-Neogene and earlier. The ecological and climatic conditions suitable for these relict species have not been identified. Here, we map the abundance and distribution patterns of relict species, showing high abundance in the humid subtropical/warm-temperate forest regions. We further use Ecological Niche Modeling to show that these patterns align with maps of climate refugia, and we predict species’ chances of persistence given the future climatic changes expected for East Asia. By 2070, potentially suitable areas with high richness of relict species will decrease, although the areas as a whole will probably expand. We identify areas in southwestern China and northern Vietnam as long-term climatically stable refugia likely to preserve ancient lineages, highlighting areas that could be prioritized for conservation of such species.

East Asia contains “relict” plant species that persist under narrow climatic conditions after once having wider distributions. Here, using distribution records coupled with ecological niche models, the authors identify long-term stable refugia possessing past, current and future climatic suitability favoring ancient plant lineages.

Mapping Fishing Activities and Suitable Fishing Grounds Using Nighttime Satellite Images and Maximum Entropy Modelling

Fisheries surveys over broad spatial areas are crucial in defining and delineating appropriate fisheries management areas. Yet accurate mapping and tracking of fishing activities remain largely restricted to developed countries with sufficient resources to use automated identification systems and vessel monitoring systems. For many countries, the spatial extent and boundaries of fishing grounds are not completely known. We used satellite images at night to detect fishing grounds in the Philippines for fishing gears that use powerful lights to attract coastal pelagic fishes. We used nightly boat detection data, extracted by U.S. NOAA from the Visible Infrared Imaging Radiometer Suite (VIIRS), for the Philippines from 2012 to 2016, covering 1713 nights, to examine spatio-temporal patterns of fishing activities in the country. Using density-based clustering, we identified 134 core fishing areas (CFAs) ranging in size from 6 to 23,215 km2 within the Philippines’ contiguous maritime zone. The CFAs had different seasonal patterns and range of intensities in total light output, possibly reflecting differences in multi-gear and multi-species signatures of fishing activities in each fishing ground. Using maximum entropy modeling, we identified bathymetry and chlorophyll as the main environmental predictors of spatial occurrence of these CFAs when analyzed together, highlighting the multi-gear nature of the CFAs. Applications of the model to specific CFAs identified different environmental drivers of fishing distribution, coinciding with known oceanographic associations for a CFA’s dominant target species. This case study highlights nighttime satellite images as a useful source of spatial fishing effort information for fisheries, especially in Southeast Asia.

High-Resolution Ecological Niche Modeling of Ixodes scapularis Ticks Based on Passive Surveillance Data at the Northern Frontier of Lyme Disease Emergence in North America

BACKGROUND

Lyme disease (LD) is a bacterial infection transmitted by the black-legged tick (Ixodes scapularis) in eastern North America. It is an emerging disease in Canada due to the expanding range of its tick vector. Environmental risk maps for LD, based on the distribution of the black-legged tick, have focused on coarse determinants such as climate. However, climatic factors vary little within individual health units, the level at which local public health decision-making takes place. We hypothesize that high-resolution environmental data and routinely collected passive surveillance data can be used to develop valid models for tick occurrence and provide insight into ecological processes affecting tick presence at fine scales.

METHODS

We used a maximum entropy algorithm (MaxEnt) to build a habitat suitability model for I. scapularis in Ottawa, Ontario, Canada using georeferenced occurrence points from passive surveillance data collected between 2013 and 2016 and high-resolution land cover and elevation data. We evaluated our model using an independent tick presence/absence dataset collected through active surveillance at 17 field sites during the summer of 2017.

RESULTS

Our model showed a good ability to discriminate positive sites from negative sites for tick presence (AUC = 0.878 ± 0.019, classification accuracy = 0.835 ± 0.020). Heavily forested suburban and rural areas in the west and southwest of Ottawa had higher predicted suitability than the more agricultural eastern areas.

CONCLUSIONS

This study demonstrates the value of passive surveillance data to model local-scale environmental risk for the tick vector of LD at sites of interest to public health. Given the rising incidence of LD and other emerging vector-borne diseases in Canada, our findings support the ongoing collection of these data and collaboration with researchers to provide a timely and accurate portrait of evolving public health risk.

Potential global distribution of Diabrotica species and the risks for agricultural production

BACKGROUND

Despite efforts in the last few decades to prevent biological invasions, agricultural pests continue to spread as a result of globalization and international trade. This study was conducted to identify suitable areas for the occurrence of four Diabrotica species and to assess the potential impact of these species in a scenario of invasion followed by spread throughout the estimated suitable regions.

RESULTS

Our findings reveal that a large proportion of the suitable areas for Diabrotica species overlap with cultivated areas. Niche analyses also demonstrated that these species occupy a small proportion of the suitable habitats available to them, indicating that, if new areas are invaded, there is a risk of spread throughout adjacent regions.

CONCLUSION

Most of the suitable areas for Diabrotica species overlap with highly productive agricultural areas, suggesting that a potential spread of these species may cause economic loss. Our study provides a valuable contribution to the development of tools aiming to predict the potential spread of these species throughout the world. © 2018 Society of Chemical Industry.

The disjunct pattern of the Neotropical harvestman Discocyrtus dilatatus (Gonyleptidae) explained by climate-driven range shifts in the Quaternary: Paleodistributional and molecular evidence

The disjunct distribution of the harvestman Discocyrtus dilatatus (Opiliones, Gonyleptidae) is used as a case study to test the hypothesis of a trans-Chaco Pleistocene paleobridge during range expansion stages. This would have temporarily connected humid regions (‘Mesopotamia’ in northeastern Argentina, and the ‘Yungas’ in the northwest, NWA) in the subtropical and temperate South American lowlands. The present study combines two independent approaches: paleodistributional reconstruction, using the Species Distribution Modeling method MaxEnt and projection onto Quaternary paleoclimates (6 kya, 21 kya, 130 kya), and phylogeographic analyses based on the cytochrome oxidase subunit I molecular marker. Models predict a maximal shrinkage during the warm Last Interglacial (130 kya), and the rise of the hypothesized paleobridge in the Last Glacial Maximum (21 kya), revealing that cold-dry stages (not warm-humid ones, as supposed) enabled the range expansion of this species. The disjunction was formed in the mid-Holocene (6 kya) and is intensified under current conditions. The median-joining network shows that NWA haplotypes are peripherally related to different Mesopotamian lineages; haplotypes from Santa Fe and Córdoba Provinces consistently occupy central positions in the network. According to the dated phylogeny, Mesopotamia-NWA expansion events would have occurred in the last glacial period, in many cases closely associated to the Last Glacial Maximum, with most divergence events occurring shortly thereafter. Only two (out of nine) NWA haplotypes are shared with Mesopotamian localities. A single, presumably relictual NWA haplotype was found to have diverged much earlier, suggesting an ancient expansion event not recoverable by the paleodistributional models. Different measures of sequence statistics, genetic diversity, population structure and history of demographic changes are provided. This research offers the first available evidence for the historical origin of NWA disjunct populations of a Mesopotamian harvestman.