Set Of Environmental Predictors Research Articles

A new method for predicting hurricane rapid intensification based on co-occurring environmental parameters

Tropical cyclones (TCs) that undergo Rapid Intensification (RI) can pose serious socioeconomic threats and can potentially result in major damaging impacts along coastal areas. Considering the complexity of various physical mechanisms that play a role in RI and its relatively low probability of occurrence, predicting RI remains a major operational challenge. In this study, we propose a simple deterministic binary classification model based on the co-occurrence of environmental parameters (MCE) to predict an RI event. More specifically, the model determines the possibility of RI based on a simple count of the number of environmental predictors deemed favorable and unfavorable. We compare our model results to logistic regression (LR) and decision tree (DT) models, well-trained using the same set of environmental predictors. Results reveal that at an RI threshold of 30 kt, the MCE exhibits a critical success index score of 0.233 which is 14% higher than DT and LR model performances. When tested at multiple RI thresholds, the MCE displays relatively higher skill scores across multiple metrics. By simultaneously evaluating the favorability of predictors, the MCE is able to comparatively reduce the number of false alarms predicted when certain predictors are unfavorable toward RI. Interpreting these model results to gain a physical understanding of how co-occurring environmental parameters can affect RI, we highlight future directions for using models based on the MCE approach to understand and predict TC RI as well as other meteorological extremes.

Evaluating factors affecting the distribution and timing of Pacific Herring Clupea pallasii spawn in British Columbia

Pacific herring Clupea pallasii spawn in nearshore areas in late winter to early spring, but factors influencing the timing and spatial distribution of spawning are not well known. We modeled the temporal and spatial distribution of spawning for 5 herring stocks in British Columbia from egg deposition surveys conducted from 1988-2018 using different sets of environmental predictors and modeling methods. Random forest modeling showed that the timing of spawning in each year was mainly influenced by the number of daylight hours being >10.5, cumulative degree days >100 and salinity at 30.5. The spatial distribution of spawning tended to occur at consistent locations over time. Results showed that the probability of spawning occurring at a transect in a given year was largely determined by the biomass of herring and location of the transect relative to the center of spawning. Environmental factors at individual transects played a much smaller role in determining spawn distribution. There was mixed evidence for spatial expansion of spawning distribution in years of high biomass, with some stocks, such as Haida Gwaii, showing no expansion of the spawning area in years of higher biomass. Since Pacific herring recruitment has been linked to their ability to time larval hatch to spring bloom timing, future warming temperatures may result in earlier herring spawning relative to the spring bloom. This will increase the probability of mismatch with prey, impacting larval herring starvation, growth and perhaps mortality, leading to reductions in recruitment to these important stocks.

Bayesian prediction of wildfire event probability using normalized difference vegetation index data from an Australian forest

Wildfires are impactful natural disasters, creating a significant impact across many rural communities. Predicting wildfire probability provides authorities with invaluable information to take preventive measures at the early stages. This study establishes Bayesian modelling for predicting the wildfire event probability based on a set of environmental predictors and forest vulnerability, represented by the normalized difference vegetation index. Prior information about the impact of these predictors on the likelihood of wildfire is available in the reports on the past major wildfire events. In that sense, the use of prior information in the Bayesian models has the potential to provide accurate predictions for the wildfire probability. Moreover, the relationship between the predictors creates mediating effects on the likelihood of a wildfire event. A multivariate prior distribution in the Bayesian modelling can capture the mediating effects. In this study, Bayesian models with informative and noninformative priors are considered with independent and multivariate prior distributions to utilize the available prior information and handle the mediating effects between the predictors using the normalized difference vegetation index data provided by Google Earth Engine. Nine years of data were gathered across 9841 sampled areas in a forested land of Australia. Modelling results concluded that forest vulnerability is found to be the dominant predictor of wildfire probability. This modelling can help create a Wildfire Warning Index based on climate data and forest vulnerability measurements, enabling preventative actions in high-risk and targeted areas.

Developing species distribution models for critically endangered species using participatory data: The European sturgeon marine habitat suitability

Increasing knowledge on distribution and habitat selection of coastal fish species is essential for the management of marine environments, and can be provided through the use of Species Distribution Models (SDM). Opportunistic and participatory data, such as recreational or commercial bycatch data, can provide precious information at large scale and low cost to estimate fish spatial distributions, but are subject to biases (e.g. lack of absence data, non-random sampling scheme). The European sturgeon (Acipenser sturio) is a critically endangered diadromous fish and its marine habitat preferences are unknown. The goal of this study was to investigate key variables driving A. sturio distribution and predict its current distribution in inshore waters along the European Atlantic coast. We developed two correlative SDM in A. sturio marine habitats based on incidental observations (2012–2021) and two different sets of environmental predictors. In both attempts, six methods which differ according to the biases they aim to reduce (i.e. oversampled areas or environmental conditions, dependence between calibration and validation datasets) and the input dataset used (i.e. random pseudo-absences or target-group pseudo-absences) were applied at various stages of the model building and evaluation process. By comparing potentially biased and bias-corrected predictions, we found similar satisfactory level of evaluation and spatial patterns of habitat suitability. Marine areas predicted as the most suitable were distributed along the French Atlantic coast, the Seine and Somme bays, and at a lesser extent around the Rhine delta and in some coastal areas of the Cantabrian Sea (Northern Spain). We found a strong influence of the bathymetry, temperature, and salinity as well as an influence of the bottom current velocity and of the distance to the source population in explaining A. sturio marine distribution, while the nature of the substrate was not retained by the models. Results of this study could pave the way toward more integrated conservation measures for this critically endangered fish.

Aggregate population-level models informed by genetics predict more suitable habitat than traditional species-level model across the range of a widespread riparian tree.

Identifying and predicting how species ranges will shift in response to climate change is paramount for conservation and restoration. Ecological niche models are the most common method used to estimate potential distributions of species; however, they traditionally omit knowledge of intraspecific variation that can allow populations to respond uniquely to change. Here, we aim to test how population X environment relationships influence predicted suitable geographic distributions by comparing aggregated population-level models with species-level model predictions of suitable habitat within population ranges and across the species’ range. We also test the effect of two variable selection methods on these predictions–both addressing the possibility of local adaptation: Models were built with (a) a common set, and number, of predictors and, (b) a unique combination and number of predictors specific to each group’s training extent. Our study addresses the overarching hypothesis that populations have unique environmental niches, and specifically that (1) species-level models predict more suitable habitat within the ranges of genetic populations than individual models built from those groups, particularly when compared models are built with the same set of environmental predictors; and (2) aggregated genetic population models predict more suitable habitat across the species’ range than the species-level model, an = d this difference will increase when models are trained with individualized predictors. We found the species models predicted more habitat within population ranges for two of three genetic groups regardless of variable selection, and that aggregated population models predicted more habitat than species’ models, but that individualized predictors increased this difference. Our study emphasizes the extent to which changes to model predictions depend on the inclusion of genetic information and on the type and selection of predictors. Results from these modeling decisions can have broad implications for predicting population-level ecological and evolutionary responses to climate change.

Genome-wide association study on blood pressure traits in the Iranian population suggests ZBED9 as a new locus for hypertension

High blood pressure is the heritable risk factor for cardiovascular and kidney diseases. Genome-wide association studies(GWAS) on blood pressure traits increase our understanding of its underlying genetic basis. However, a large proportion of GWAS was conducted in Europeans, and some roadblocks deprive other populations to benefit from their results. Iranians population with a high degree of genomic specificity has not been represented in international databases to date, so to fill the gap, we explored the effects of 652,919 genomic variants on Systolic Blood Pressure (SBP), Diastolic Blood Pressure (DBP), and Hypertension (HTN) in 7694 Iranian adults aged 18 and over from Tehran Cardiometabolic Genetic Study (TCGS). We identified consistent signals on ZBED9 associated with HTN in the genome-wide borderline threshold after adjusting for different sets of environmental predictors. Moreover, strong signals on ABHD17C and suggestive signals on FBN1 were detected for DBP and SBP, respectively, while these signals were not consistent in different GWA analysis. Our finding on ZBED9 was confirmed for all BP traits by linkage analysis in an independent sample. We found significant associations with similar direction of effects and allele frequency of genetic variants on ZBED9 with DBP (genome-wide threshold) and HTN (nominal threshold) in GWAS summary data of UK Biobank. Although there is no strong evidence to support the function of ZBED9 in blood pressure regulation, it provides new insight into the pleiotropic effects of hypertension and other cardiovascular diseases.

Exploring analog-based schemes for aerosol optical depth forecasting with WRF-Chem

We implement and test an analog-based post-processing method to improve short range forecasts of aerosol optical depth (AOD) using the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem). Model postprocessing of AOD is performed using historical analog forecasts and a Kalman Filter (KF). Analog forecasts are selected from WRF-Chem simulations based on a set of environmental predictors (AOD, wind speed, precipitable water, and particulate matter) that exhibit past values similar to the current forecasts. Space-borne AOD from Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard Terra and Aqua satellites corresponding to the analogs are used to build the analog ensemble. This study focuses on a spatial domain covering the AERONET sites in contiguous United States. We use the analog ensemble weighted mean (AN) and Kalman filter analog (KFAN) algorithms, which are both trained using WRF-Chem AOD forecasts for the months of June to August during 2008–2011 and tested during the same months for 2012. Overall, the AOD forecast are more skillful when the forecast errors are corrected using a combination of analogs and Kalman filter in KFAN. This is especially true for the western US where the correlation of AOD with PM2.5, PM10, and surface horizontal wind speed are higher than those for other predictors. In fact, the overall biases in AOD are significantly reduced close to zero, with KFAN AOD being statistically indistinguishable to MODIS. However, both methods show mixed results (albeit still showing overall improvements) in eastern and central U.S., where AOD and its variability are highest. We find that, during the summer, PM is not the only predominant factor driving AOD in these regions, unlike western United States (U.S.) (except New Mexico and Arizona). We note, however, that the quality of the analogs depends on the model's capability to accurately simulate total precipitable water, which in turn influences aerosol sources and sinks.

Ecological correlates of Golden Eagle Aquila chrysaetos breeding occurrence in Sardinia

ABSTRACT Capsule Golden Eagles Aquila chrysaetos in Sardinia are clustered across the main mountain ranges of the island, with a preference for undisturbed and homogeneous inland habitats. Aims To analyse habitat preferences of the Golden Eagle in Sardinia, Italy, at the landscape and home range spatial scales. Methods Landscape scale habitat preferences were analysed using the 10 × 10 km Universal Transverse Mercator grid and the home range scale was based on the spatial distribution of breeding territories. Generalized linear models were fitted with three different sets of environmental predictors (topographic, bioclimatic and land use variables) to analyse the spatial distribution of Golden Eagles with a case–control design. Results Eagles showed a preference for rugged and elevated areas, characterized by a certain degree of humidity and surrounded by areas of forest. The distribution of Golden Eagles on this Mediterranean island was negatively affected by the occurrence of arable farmland and coastal areas, as well as by the effects of habitat fragmentation. Conclusions The results of this study could contribute to future management strategies and conservation projects aimed to protect this species, and may be used to identify the most suitable conservation areas for this and other competing species, such as the Bonelli’s Eagle Aquila fasciata, which is currently the subject of a reintroduction project in Sardinia.

Identifying and applying an optimum set of environmental variables in species distribution models

ABSTRACTPredictive models such as species distribution models (SDMs) are increasingly applied to inform conservation efforts and management decisions. A thoughtful predictor choice, and hence the variable selection process, is a challenge when modeling large communities. Often, variable choice is made for an entire community and not for specific species, resulting in less appropriate predictors for at least some species and affecting model performance and predicted distributions. Using 2 German river catchments as a model system, we investigated (1) application of boosted regression trees (BRTs) as a variable selection procedure to choose the optimal set of environmental predictors and (2) whether model performance is increased by applying custom-made predictor sets to individual species. From a community of 67 benthic macroinvertebrate species, 10 increased in accuracy with the customized predictor set and 10 species decreased in accuracy. Notably, current preference, stream notation, and functional group differed between these species’ groups, which correspond to varied environmental conditions in their known occurrence sites. The species that increased in accuracy showed a preference toward lowland conditions and were far less widespread than the species that decreased in accuracy. We conclude that BRTs are a useful tool for selecting variables for SDMs on large communities. Also, for specialist, rare, or invasive species, determining a species-specific custom-made predictor set may be preferable because species’ preferences may not be representative of the entire study area. Our study describes a structured variable selection approach that can be readily implemented to predict species distributions informative for river management decisions.

Shark conservation hindered by lack of habitat protection

Many of the world’s shark populations are in decline, indicating the need for improved conservation and management. Well managed and appropriately located marine parks and marine protected areas (MPAs) have potential to enhance shark conservation by restricting fisheries and protecting suitable habitat for threatened shark populations. Here, we used shark occurrence records collected by commercial fisheries to determine suitable habitat for pelagic sharks within the Australian continental Exclusive Economic Zone (EEZ), and to quantify the amount of suitable habitat contained within existing MPAs. We developed generalised linear models using proportional occurrences of pelagic sharks for three families: Alopiidae (thresher), Carcharhinidae (requiem), and Lamnidae (mackerel) sharks. We also considered aggregated species from the Lamnidae and Carcharhinidae families (‘combined sharks’ in the models). Using a set of environmental predictors known to affect shark occurrence, including chlorophyll-a concentration, salinity, sea surface temperature, and turbidity, as well as geomorphological, geophysical, and sedimentary parameters, we found that models including sea surface temperature and turbidity were ranked highest in their ability to predict shark distributions. We used these results to predict geographic regions where habitat was most suitable for pelagic sharks within the Australian EEZ, and our results revealed that suitable habitat was limited in no-take zones within MPAs. For all shark groupings, suitable habitats were found mostly at locations exposed to fishing pressure, potentially increasing the vulnerability of the pelagic shark species considered. Our predictive models provide a foundation for future spatial planning and shark management, suggesting that strong fisheries management in addition to MPAs is necessary for pelagic shark conservation.

Various measures of potential evapotranspiration have species-specific impact on species distribution models

The growth and distribution of plant species in water limited environments is often limited by the atmospheric evaporative demands which is measured in terms of potential evapotranspiration (PET). While PET estimated by different methods have been widely used to assess vegetation response to climate change, species distribution models offer unique opportunity to compare their efficiency in predicting habitat suitability of plant species. In this study, we perform the first multi-species comparison of two widely used metrics of PET i.e., Penman-Monteith and Thornthwaite, and show how they result in similar or different on projected distribution of water limited species and potential consequences on their conservation strategies across North Central U.S. To build species distribution models of eight species, we used two sets of environmental predictors which were identical except for the metric of PET (Penman-Monteith vs Thornthwaite) and projected habitat suitability for historical (2005) and future (2099) periods. We found an excellent model performance with no difference under two sets of predictors (AUC = ∼0.93). The relative influence of Thornthwaite PET on habitat prediction was higher than Penman PET for most of the species. We observed that the area of the projected suitable habitat was always higher under Thornthwaite set of predictors than Penman set of predictors (ranges from 25 % to 941 %), with the exception of Pinus contorta for which the reverse was true. In most cases, these differences were non-trivial, indicating that the choice of the PET metric, although both of them are commonly used, can have dramatic consequences on the conservation management decisions. Therefore, the conservation management decisions can be markedly different depending on the choice of the PET metric used for species distribution modeling of water limited species.

The influence of data source and species distribution modelling method on spatial conservation priorities

AbstractAimSpecies distribution models are an important conservation tool; however, performance can vary with factors including data inputs and modelling method. Model outputs are often under‐evaluated for explanatory and predictive capacity. Our aim was to evaluate the capacity of existing data for seven small mammal species to provide useful inferences for management planning.LocationBathurst and Melville (collectively the Tiwi) Islands, Northern Territory, Australia.MethodsWe developed species distribution models (SDMs) with generalized linear models (GLMs) and boosted regression trees (BRTs) using survey data (351 sites) of small mammals, with two sets of environmental predictors: (a) field‐study measurements and (b) available remotely sensed rasters. Predictive capacity of models was evaluated using percentage of deviance explained (%DE) and area under the receiver operating characteristic curve (AUC). We used Marxan to evaluate the influence of different model and data types as input for identifying spatial priorities.ResultsField‐informed SDMs performed well across both modelling methods, with relatively high test AUC values (mean = 0.82, range = 0.64–0.97) and test %DE (mean = 22.5%, range = 3.5%–65.8%). Remotely sensed models performed relatively poorly, with lower test AUC values (mean = 0.7, range = 0.56–0.86) and lower test %DE (mean = 8.9%, range = 0.03%–24.9%). A notable exception was remotely sensed models forMelomys burtoni(AUC = 0.85 & 0.86, %DE = 23.3% & 24.9%, Bathurst and Melville respectively). Marxan site irreplaceability rankings demonstrated low to marginal agreement using field‐informed and remotely sensed inputs (Pearson correlation coefficient = 0.3), and similarly, using GLM and BRT model inputs (0.29).Main conclusionsThe occurrence of small mammals on the Tiwi Islands can be reasonably explained with field‐informed variables, but not with remotely sensed alternatives. Different models lead to different conservation priorities. Our work emphasizes the importance of thoroughly testing SDMs prior to decision‐making.

Climate Shapes the Geographic Distribution of the Blowfly Sarconesia chlorogaster (Diptera: Calliphoridae): An Environmental Niche Modeling Approach.

For all species, abiotic factors directly affect performance, survival and reproduction, and consequently, their geographic distribution. Species distribution models (SDMs) are important tools to predict the influence of abiotic factors in species distributions and has been more applied over the years. However, these models can be built under different algorithms and using different methods to select environmental predictors, which can lead to different results. Five different algorithms and two sets of environmental predictors were compared to predict the geographic distribution of the blowfly Sarconesia chlorogaster (Wiedemann) (Diptera: Calliphoridae). This species has several occurrence points and a considerable amount of biological data available, which makes S. chlorogaster a good model system to compare environmental predictors. Two sets of environmental predictors (mainly derived from temperature and humidity) were built, and the set based on the influence of abiotic variables on the ecophysiology of S. chlorogaster showed better results than the principal component analysis (PCA) approach using 19 climatic variables. We also employed five modeling algorithms-Envelope Score, Mahalanobis Distance, GARP, Support Vector Machines, and Maxent-and the latter two showed the best performances. The results indicate that temperature is the main factor shaping geographic distribution of S. chlorogaster through its effect on fitness. Furthermore, we showed that this species is mainly distributed in south, southeastern, and some northwestern and southwestern sites of South America. In addition, our results also predicted suitable areas in Ecuador and Colombia, countries without previous records.

Assessment of soil organic carbon stocks under future climate and land cover changes in Europe

Soil organic carbon plays an important role in the carbon cycling of terrestrial ecosystems, variations in soil organic carbon stocks are very important for the ecosystem. In this study, a geostatistical model was used for predicting current and future soil organic carbon (SOC) stocks in Europe. The first phase of the study predicts current soil organic carbon content by using stepwise multiple linear regression and ordinary kriging and the second phase of the study projects the soil organic carbon to the near future (2050) by using a set of environmental predictors. We demonstrate here an approach to predict present and future soil organic carbon stocks by using climate, land cover, terrain and soil data and their projections. The covariates were selected for their role in the carbon cycle and their availability for the future model. The regression-kriging as a base model is predicting current SOC stocks in Europe by using a set of covariates and dense SOC measurements coming from LUCAS Soil Database. The base model delivers coefficients for each of the covariates to the future model. The overall model produced soil organic carbon maps which reflect the present and the future predictions (2050) based on climate and land cover projections. The data of the present climate conditions (long-term average (1950–2000)) and the future projections for 2050 were obtained from WorldClim data portal. The future climate projections are the recent climate projections mentioned in the Fifth Assessment IPCC report. These projections were extracted from the global climate models (GCMs) for four representative concentration pathways (RCPs). The results suggest an overall increase in SOC stocks by 2050 in Europe (EU26) under all climate and land cover scenarios, but the extent of the increase varies between the climate model and emissions scenarios.

Continuous Bayesian networks for the estimation of species richness

We propose a new methodology based on continuous Bayesian networks for assessing species richness. Specifically, we applied a restricted structure Bayesian network, known as tree augmented naive Bayes (TAN), regarding a set of environmental continuous predictors. First, we analysed the relationships between the response variable (called the terrestrial vertebrate species richness) and a set of environmental predictors. Second, the learnt model was used to estimate the species richness in Andalusia (Spain) and the results were depicted on a map. In addition to this, the TAN model was compared to three other methods commonly used for regression in terms of their root mean squared error. The experimental results showed that the TAN model not only was competitive from the point of view of accuracy but also managed to deal with the species richness–environment relationship, which is complex from the ecological point of view. The results highlight that landscape heterogeneity, topographical and social variables had a direct relationship with species richness while climatic variables showed more complicated relationships with the response.

Concordance among aquatic communities in a tropical irrigation system

ABSTRACT Different biological communities may exhibit similar spatial and/or temporal distributional patterns, a property termed community concordance. This study was conducted in a tropical irrigation system (Araguaia River floodplain) and aimed to quantify concordance levels between three aquatic communities (zooplankton, benthic macroinvertebrates, and aquatic macrophytes), and between these communities and a set of environmental predictors. To accomplish these goals, we used ordination techniques and Procrustean analysis. There were no relationships between the communities, and only zooplankton community patterns were significantly correlated with environmental predictors. These results indicate that biological surrogacy can be a flawed approach at small spatial scales and highlight the importance of the zooplankton community as a reliable ecological indicator in this type of system. © 2014 Associacao Brasileira de Ciencia Ecologica e Conservacao. Published by Elsevier Editora Ltda.

Determinants of plant community composition of remnant biancane badlands: a hierarchical approach to quantify species-environment relationships

Question: Which environmental variables best explain patterns in the vegetation of biancane badlands? What is the role of spatial scales in structuring the vegetation of biancane badlands within the agricultural matrix? Location: Five biancane badlands in Central Italy (Tuscany). Methods: An object-oriented approach on high-resolution multispectral images was used to classify physiognomic vegetation types in five biancane badlands. Within each badland, data on vascular plant species abundance were collected using a stratified random design. Variation partitioning based on partial redundancy analysis was used to evaluate the contribution of three sets of environmental predictors, recorded at the spatial scales of plot, patch and biancane badland in explaining patterns in plant community composition. Results: Environmental variables included in the final model ‐ electrical conductivity and carbon/nitrogen ratio (plot scale), shape index (patch scale) and area (biancane badland scale) ‐ accounted for 15.5% of the total variation in plant community composition. Soil characteristics measured at the plot level explained the majority of variation. In the smallest badlands, Bromus erectus perennial grasslands were absent, while annual grasslands, linked with harsh soil conditions (i.e. high soil salinity), were not affected by either the surface area of biancane badlands or by the soil nitrogen availability. Conclusions: The identification of the major predictors of patterns in remnant vegetation requires conducting investigations at multiple spatial scale. Management strategies should operate at different spatial scale, preventing any further reduction in the size of existing badlands and relying on habitat- instead of area-focused conservation practices.

Similar performance in central and range-edge populations of a Eurasian steppe grass under different climate and soil pH regimes

The abundant centre hypothesis predicts that changing environmental conditions are detrimental to a species' abundance and performance towards the periphery of its range. We tested these predictions for the perennial grass Stipa capillata, a species that is commonly found in steppes of Asia but is rare at its north-western range edge, in central Europe. We compared 21 populations in dry grassland fragments in central Europe and 20 populations in steppe habitats of Kazakhstan. We studied 15 plant performance traits both in situ and under laboratory and common greenhouse conditions, including local density, plant size and biomass production, seed size, weight and viability. To assess environmental conditions, we assembled data on topography, soil properties and climatic parameters. Using variance components analysis and multivariate methods we analyzed whether plant performance and environmental attributes differed more, as predicted, between the core and peripheral regions or whether they differed more among their subregions or populations. Additionally, we tested whether performance was affected by the same set of environmental predictors in each region. Contrary to our expectations, plant performance traits were virtually unchanged between the range periphery and centre. As expected, macroclimatic conditions showed a significant difference between the two regions (annual mean temperature, annual precipitation). The only other measured environmental variable that differed significantly between the two regions was soil pH, which was lower in core-range populations. Our findings thus do not support the notion of reduced performance at the range edge. Instead, our data lend support to earlier theories of relative habitat constancy, suggesting that peripheral populations can shift to other habitats through plasticity or adaptation.

Predicting the potential distribution of the endangered Przewalski's gazelle

Species distribution modelling can be a powerful tool in species conservation. Przewalski's gazelle Procapra przewalskii is an endangered ungulate and a conservation focus on the Qinghai–Tibetan Plateau. To identify the potential range and provide a conservation base for the species, we used the maximum entropy approach to build a habitat suitability map and took into account: (1) the comparison among three competing models (the full, uncorrelated and pruned models) with different sets of environmental predictors; and (2) scale effects on model spatial output and performance. Elevation, maximum temperature of the warmest month, mean temperature of the wettest and warmest quarter and isothermality were the five most effective predictors. The 11 threshold-determining approaches identified different thresholds. Spatial patterns of ranges predicted with the three models were similar, although the uncorrelated model was outperformed by the other two models. All three models identified regions in the eastern part of the Qinghai–Tibetan Plateau as the most suitable habitat for Przewalski's gazelle. Cross-validation area under the receiver operating characteristic curve (AUC) of the full model decreased slightly as the scale increased; spatial congruence AUC fluctuated with the small range, and the predicted range increased disproportionately. This study identifies areas to find new populations and representative habitats of a rare and endangered species.

Comparing different environmental variables in predictive models of bird distribution

The response variable (often the presence of a species) in predictive habitat models relies on a set of environmental predictors. Among all known environmental predictors, vegetation has the most effect on species abundance and on their habitat preferences, due to the wide range of necessary resources that it provides for the survival of bird species. However, other predictors, in turn, affect bird distribution, and some-times they play a more important role in habitat selection, depending on the natural history and ecological needs of the bird species. In this regard, different analyses have been conducted to predict the distribution, and define habitat suitability (such as discriminant function analysis, General Linear Models, and ANOVA). In this study, all three analytical designs were used to investigate the relationship of seven bird species to the major environmental gradients in the study area, to find out the significance of each of these factors on habitat selection. GIS has been used to prepare spatial distributional data, and to overlay and calculate different aspects of the environmental factors. The results suggest that potential individual habitat patches play a small role compared to the landscape (entire corresponding habitat patches), when considering vegetation. The influence of built-up areas is significant for all the species, and the proximity to the sea shore is very significant for at least one of the species, however, it is not neutral for all other species.