The predictive performance and stability of six species distribution models

Decreasing brown bear (Ursus arctos) habitat due to climate change in Central Asia and the Asian Highlands

Around the world, climate change has impacted many species. In this study, we used bioclimatic variables and biophysical layers of Central Asia and the Asian Highlands combined with presence data of brown bear (Ursus arctos) to understand their current distribution and predict their future distribution under the current rate of climate change. Our bioclimatic model showed that the current suitable habitat of brown bear encompasses 3,430,493 km2 in the study area, the majority of which (>65%) located in China. Our analyses demonstrated that suitable habitat will be reduced by 11% (378,861.30 km2) across Central Asia and the Asian Highlands by 2,050 due to climate change, predominantly (>90%) due to the changes in temperature and precipitation. The spatially averaged mean annual temperature of brown bear habitat is currently -1.2°C and predicted to increase to 1.6°C by 2,050. Mean annual precipitation in brown bear habitats is predicted to increase by 13% (from 406 to 459 mm) by 2,050. Such changes in two critical climatic variables may significantly affect the brown bear distribution, ethological repertoires, and physiological processes, which may increase their risk of extirpation in some areas. Approximately 32% (1,124,330 km2) of the total suitable habitat falls within protected areas, which was predicted to reduce to 1,103,912 km2 (1.8% loss) by 2,050. Future loss of suitable habitats inside the protected areas may force brown bears to move outside the protected areas thereby increasing their risk of mortality. Therefore, more protected areas should be established in the suitable brown bear habitats in future to sustain populations in this region. Furthermore, development of corridors is needed to connect habitats between protected areas of different countries in Central Asia. Such practices will facilitate climate migration and connectivity among populations and movement between and within countries.

Species distribution models and empirical test: Comparing predictions with well-understood geographical distribution of Bothrops alternatus in Argentina

Species distribution models (SDMs) estimate the geographical distribution of species although with several limitations due to sources of inaccuracy and biases. Empirical tests arose as the most important steps in scientific knowledge to assess the efficiency of model predictions, which are poorly rigorous in SDMs. A good approach to the empirical distribution (ED) of a species can be obtained from comprehensive empirical knowledge, that is, well-understood distributions gathered from large amount of data generated with appropriate spatial and temporal samples coverage. The aims of this study were to (a) compare different SDMs predictions with an ED; and (b) evaluate if fuzzy global matching (FGM) could be used as an index to compare SDMs predictions and ED. Six algorithms with 5 and 20 variables were used to assess their accuracy in predicting the ED of the venomous snake Bothrops alternatus (Viperidae). Its entire distribution is known, thanks to thorough field surveys across Argentina, with 1,767 records. ED was compared with SDMs predictions using Map Comparison Kit. SDMs predictions showed important biases in all methods used, from 70% sub-estimation to 40% over-estimation of ED. BIOCLIM predicted ≈31% of B. alternatus ED. DOMAIN predicted 99% of ED, but over-estimated 40% of the area. GLM with five variables calculated 75% of ED, while Genetic Algorithm for Rule-set Prediction showed ≈60% of ED; the last two presenting overpredictions in areas with favorable climatic conditions but not inhabited by the species. MaxEnt and RF were the only methods to detect isolated populations in the southern distribution of B. alternatus. Although SDMs proved useful in making predictions about species distribution, predictions need validation with expert maps knowledge and ED. Moreover, FGM showed a good performance as an index with values similar to True Skill Statistic, so that it could be used to relate ED and SDMs predictions.

Invasive alien plant species dynamics in the Himalayan region under climate change

Climate change will impact the dynamics of invasive alien plant species (IAPS). However, the ability of IAPS under changing climate to invade mountain ecosystems, particularly the Himalayan region, is less known. This study investigates the current and future habitat of five IAPS of the Himalayan region using MaxEnt and two representative concentration pathways (RCPs). Two invasive species, Ageratum conyzoides and Parthenium hysterophorus, will lose overall suitable area by 2070, while Ageratina adenophora, Chromolaena odorata and Lantana camara will gain suitable areas and all of them will retain most of the current habitat as stable. The southern Himalayan foothills will mostly conserve species ecological niches, while suitability of all the five species will decrease with increasing elevation. Such invasion dynamics in the Himalayan region could have impacts on numerous ecosystems and their biota, ecosystem services and human well-being. Trans-boundary response strategies suitable to the local context of the region could buffer some of the likely invasion impacts.

What Can We Learn from an Early Test on the Adaptation of Silver Fir Populations to Marginal Environments?

In order to determine the adaptive potential of silver fir in the southeast of Poland, the stability of the height of its five-year-old progeny was analyzed. The study was conducted in two different population groups in a total of four environments, including one ecologically marginal environment. The linear mixed model was used to evaluate the differentiation of populations in terms of height growth. The genotype and genotype-by-environment interaction biplot (GGE) were used to verify the stability of height. The climate of populations origin, in relation to actual fir distribution in Poland, was verified based on principal components analysis (PCA) of bioclimatic parameters. The highest total variability was explained by the genotype-environment interaction effect (GE) (54.50%), while the genotype effect (G) explained 41.27% and only 4.23% was explained by the site effect. The result of height growth variations revealed the Komańcza site as the most representative among study sites, while the Lesko site characterized the highest discriminating ability. The progeny occurring in climatic conditions most different from the average testing conditions showed a heterogeneous growth reaction, only adapting to the marginal environment, while the progeny of the second population in this region as well as the northernmost one was characterized by a mean but stable growth. The westernmost population revealed maladaptation. The assessment of the adaptability of silver fir depends on the broad spectrum of test conditions considering the ecologically marginal environments.

Climatic influence on anthrax suitability in warming northern latitudes

Climate change is impacting ecosystem structure and function, with potentially drastic downstream effects on human and animal health. Emerging zoonotic diseases are expected to be particularly vulnerable to climate and biodiversity disturbance. Anthrax is an archetypal zoonosis that manifests its most significant burden on vulnerable pastoralist communities. The current study sought to investigate the influence of temperature increases on geographic anthrax suitability in the temperate, boreal, and arctic North, where observed climate impact has been rapid. This study also explored the influence of climate relative to more traditional factors, such as livestock distribution, ungulate biodiversity, and soil-water balance, in demarcating risk. Machine learning was used to model anthrax suitability in northern latitudes. The model identified climate, livestock density and wild ungulate species richness as the most influential features in predicting suitability. These findings highlight the significance of warming temperatures for anthrax ecology in northern latitudes, and suggest potential mitigating effects of interventions targeting megafauna biodiversity conservation in grassland ecosystems, and animal health promotion among small to midsize livestock herds.

Using geomorphological variables to predict the spatial distribution of plant species in agricultural drainage networks

To optimize ecosystem services provided by agricultural drainage networks (ditches) in headwater catchments, we need to manage the spatial distribution of plant species living in these networks. Geomorphological variables have been shown to be important predictors of plant distribution in other ecosystems because they control the water regime, the sediment deposition rates and the sun exposure in the ditches. Whether such variables may be used to predict plant distribution in agricultural drainage networks is unknown. We collected presence and absence data for 10 herbaceous plant species in a subset of a network of drainage ditches (35 km long) within a Mediterranean agricultural catchment. We simulated their spatial distribution with GLM and Maxent model using geomorphological variables and distance to natural lands and roads. Models were validated using k-fold cross-validation. We then compared the mean Area Under the Curve (AUC) values obtained for each model and other metrics issued from the confusion matrices between observed and predicted variables. Based on the results of all metrics, the models were efficient at predicting the distribution of seven species out of ten, confirming the relevance of geomorphological variables and distance to natural lands and roads to explain the occurrence of plant species in this Mediterranean catchment. In particular, the importance of the landscape geomorphological variables, ie the importance of the geomorphological features encompassing a broad environment around the ditch, has been highlighted. This suggests that agro-ecological measures for managing ecosystem services provided by ditch plants should focus on the control of the hydrological and sedimentological connectivity at the catchment scale. For example, the density of the ditch network could be modified or the spatial distribution of vegetative filter strips used for sediment trapping could be optimized. In addition, the vegetative filter strips could constitute new seed bank sources for species that are affected by the distance to natural lands and roads.

Using worldwide edaphic data to model plant species niches: An assessment at a continental extent

Ecological niche modeling (ENM) is a broadly used tool in different fields of plant ecology. Despite the importance of edaphic conditions in determining the niche of terrestrial plant species, edaphic data have rarely been included in ENMs of plant species perhaps because such data are not available for many regions. Recently, edaphic data has been made available at a global scale allowing its potential inclusion and evaluation on ENM performance for plant species. Here, we take advantage of such data and address the following main questions: What is the influence of distinct predictor variables (e.g. climatic vs edaphic) on different ENM algorithms? and what is the relationship between the performance of different predictors and geographic characteristics of species? We used 125 plant species distributed over the Neotropical region to explore the effect on ENMs of using edaphic data available from the SoilGrids database and its combination with climatic data from the CHELSA database. In addition, we related these different predictor variables to geographic characteristics of the target species and different ENM algorithms. The use of different predictors (climatic, edaphic, and both) significantly affected model performance and spatial complexity of the predictions. We showed that the use of global edaphic plus climatic variables generates ENMs with similar or better accuracy compared to those constructed only with climate variables. Moreover, the performance of models considering these different predictors, separately or jointly, was related to geographic properties of species records, such as number and distribution range. The large geographic extent, the variability of environments and the different species' geographical characteristics considered here allowed us to demonstrate that global edaphic data adds useful information for plant ENMs. This is particularly valuable for studies of species that are distributed in regions where more detailed information on soil properties is poor or does not even exist.

Mapping the Potential Global Range of the Brown Marmorated Stink Bug, Halyomorpha halys, with Particular Reference to New Zealand

Originating from Asia, the brown marmorated stink bug (BMSB) is a significant pest of horticultural/agricultural crops, grapes, woody ornamental and herbaceous plants, and is also a nuisance to people, due to its overwintering behavior in human habitation. The global range of this pest is steadily increasing and previous predictions of environmental suitability have shown New Zealand to be highly suitable. Due to the economic value of horticultural and agricultural industries to the New Zealand economy, it is vital to understand the range of potential risk within the country. Global and New Zealand potential suitability for BMSB was modeled using three algorithms and the resulting predictions ensembled to predict the potential range under current climatic conditions and under trajectories of future low (Representative Concentration Pathways, RCP, 2.6) and high (RCP 8.5) greenhouse gas emissions for both 2050 and 2070. Under current conditions, models showed a high global suitability within latitudes 25°–50° N, southern South America, southeast and southwest regions of Australia and large areas of New Zealand. Modeling the effect of climate change on BMSB range in New Zealand resulted in a southerly range shift over time, particularly with high emissions trajectory. Currently, BMSB is not established in New Zealand and it is vital that this remains the case.

Full-coverage spatial distribution of epibenthic communities in the south-eastern North Sea in relation to habitat characteristics and fishing effort

Full-coverage spatial data of occurrence and a detailed description of habitat requirements of epibenthic communities are needed in many conservation and management contexts. In the North Sea the focus has so far been on small benthic infauna, whereas structure and ecosystem functions of larger epifaunal communities have been largely ignored. This study provides a comprehensive analysis of epibenthic community structure in the south-eastern North Sea, including a detailed inventory of species, diversity and spatially contiguous distribution of communities. Data from nearly 400 stations were compiled for the study, enabling us to describe epibenthic community structure at unprecedented spatial resolution. Eight distinct epibenthic communities were found in the south-eastern North Sea by using multivariate analysis. Distribution modelling with eight environmental variables (bottom temperature and salinity, temperature differences between summer and winter, mud content of sediments, maximum bottom shear stress, stratification, water depth and annual primary production) and one human pressure (fishing effort) was used to extrapolate probable spatial distributions and to identify associated habitat characteristics of the communities in the south-eastern North Sea. Three large epibenthic communities "Coast", "Oysterground" and "Tail End" reflect a gradual habitat change from the coast towards offshore regions, expressed in gradients of bottom salinity, seasonal temperature differences and stratification as the dominant environmental factors. Five smaller communities ("Amrum Bank", "Frisian Front", "Deeps", "Dogger Bank" and "Dogger Slope") outline specific habitats in the south-eastern North Sea. The "Dogger Slope" community has not been recognized before, but has a predicted spatial extent of 7118 km². Due to the high occurrence of long-lived, sessile species such as sponges this community is very sensitive to demersal fishing.

Using niche-modelling and species-specific cost analyses to determine a multispecies corridor in a fragmented landscape

Misiones, Argentina, contains the largest remaining tract of Upper Paraná Atlantic Forest ecoregion; however, ~50% of native forest is unprotected and located in a mosaic of plantations, agriculture, and pastures. Existing protected areas are becoming increasingly isolated due to ongoing habitat modification. These factors, combined with lower than expected regional carnivore densities, emphasize the need to understand the effect of fragmentation on animal movement and connectivity between protected areas. Using detection dogs and genetic analyses of scat, we collected data on jaguars (Panthera onca), pumas (Puma concolor), ocelots (Leopardus pardalis), oncillas (Leopardus tigrinus), and bush dogs (Speothos venaticus) across habitats that varied in vegetation, disturbance, human proximity, and protective status. With MaxEnt we evaluated habitat use, habitat suitability, and potential species richness for the five carnivores across northern-central Misiones, Argentina. Through a multifaceted cost analysis that included unique requirements of each carnivore and varying degrees of overlap among them, we determined the optimal location for primary/secondary corridors that would link the northern-central zones of the Green Corridor in Misiones and identified areas within these corridors needing priority management. A secondary analysis, comparing these multispecies corridors with the jaguar’s unique requirements, demonstrated that this multispecies approach balanced the preferences of all five species and effectively captured areas required by this highly restricted and endangered carnivore. We emphasize the potential importance of expanding beyond a single umbrella or focal species when developing biological corridors that aim to capture the varied ecological requirements of coexisting species and ecological processes across the landscape. Detection dogs and genetic analyses of scat allow data on multiple species to be collected efficiently across multiple habitat types independent of the degree of legal protection. These data used with multifocal GIS analyses balance the varying degree of overlap and unique properties among them allowing for comprehensive conservation strategies to be developed relatively rapidly. Our comprehensive approach serves as a model to other regions faced with habitat loss and lack of data. The five carnivores focused on in our study have wide ranges, so the results from this study can be expanded and combined with surrounding countries, with analyses at the species or community level.

Modeling the ecologic niche of plague in sylvan and domestic animal hosts to delineate sources of human exposure in the western United States

Plague has been established in the western United States (US) since 1900 following the West Coast introduction of commensal rodents infected with Yersinia pestis via early industrial shipping. Over the last century, plague ecology has transitioned through cycles of widespread human transmission, urban domestic transmission among commensal rodents, and ultimately settled into the predominantly sylvan foci that remain today where it is maintained alternatively by enzootic and epizootic transmission. While zoonotic transmission to humans is much less common in modern times, significant plague risk remains in parts of the western US. Moreover, risk to some threatened species that are part of the epizootic cycle can be quite substantive. This investigation attempted to predict the risk of plague across the western US by modeling the ecologic niche of plague in sylvan and domestic animals identified between 2000 and 2015. A Maxent machine learning algorithm was used to predict this niche based on climate, altitude, land cover, and the presence of an important enzootic species, Peromyscus maniculatus. This model demonstrated good predictive ability (AUC = 86%) and identified areas of high risk in central Colorado, north-central New Mexico, and southwestern and northeastern California. The presence of P. maniculatus, altitude, precipitation during the driest and wettest quarters, and distance to artificial surfaces, all contributed substantively to maximizing the gain function. These findings add to the known landscape epidemiology and infection ecology of plague in the western US and may suggest locations of particular risk to be targeted for wild and domestic animal intervention.