Predicting the distribution of Stipa purpurea across the Tibetan Plateau via the MaxEnt model

Abstract

BackgroundThe ecosystems across Tibetan Plateau are changing rapidly under the influence of climate warming, which has caused substantial changes in spatial and temporal environmental patterns. Stipa purpurea, as a dominant herbsage resource in alpine steppe, has a great influence on animal husbandry in the Tibetan Plateau. Global warming has been forecasted to continue in the future (2050s, 2070s), questioning the future distribution of S. purpurea and its response to climate change. The maximum entropy (MaxEnt) modeling, due to its multiple advantages (e.g. uses presence-only data, performs well with incomplete data, and requires small sample sizes and gaps), has been used to understand species environment relationships and predict species distributions across locations that have not been sampled.ResultsAnnual mean temperature, annual precipitation, temperature seasonality, altitude, and precipitation during the driest month, significantly affected the distribution of S. purpurea. Only 0.70% of the Tibetan Plateau area included a very highly suitable habitat (habitat suitability [HS] = 0.8–1.0). Highly suitable habitat (HS = 0.6–0.8), moderately suitable habitat (HS = 0.4–0.6), and unsuitable habitat (HS = 0.2–0.4) occupied 6.20, 14.30 and 22.40% of the Tibetan Plateau area, respectively, and the majority (56.40%) of the Tibetan Plateau area constituted a highly unsuitable habitat (HS = 0–0.2). In addition, the response curves of species ecological suitability simulated by generalized additive model nearly corresponded with the response curves generated by the MaxEnt model.ConclusionsAt a temporal scale, the habitat suitability of S. purpurea tends to increase from the 1990s to 2050s, but decline from the 2050s to 2070s. At a spatial scale, the future distribution of S. purpurea will not exhibit sweeping changes and will remain in the central and southeastern regions of the Tibetan Plateau. These results benefit the local animal husbandry and provide evidence for establishing reasonable management practices.Electronic supplementary materialThe online version of this article (10.1186/s12898-018-0165-0) contains supplementary material, which is available to authorized users.

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

“…For those models that needed presence/absence data, we generated 1,000 random absence records outside the distribution area (ED) (Tognelli, Roig‐junent, Marvaldi, Flores, & Lobo, ), where the evidence in the last 100 years showed that B. alternatus does not presently occur but rather became true absences, which are based on reliable field evidence of nonoccurrence (Figure c) (Saupe et al., ). Moreover, several studies obtained good performance using pseudo‐absence/absence data outside a predefined region based on a minimum distance to the presence (Barbet‐Massin et al., ; Lobo, Jiménez‐Valverde, & Hortal, ). This way of generated absence records is recommended when using classification and machine‐learning techniques (Barbet‐Massin et al., ).…”

“…We obtained the ROC curve, that is, AUC index, which represents the probability that the model correctly predicted the observed presences and absences and varies from 0 to 1, 1 being perfect discrimination and 0.5 to 0 implying a discrimination worse than random Elith et al, 2006). One of the greatest advantages of the ROC curve (AUC) is that it is threshold independent (Lobo, Jiménez-Valverde, & Real, 2008); however, its use and efficiency has been widely criticized (Jiménez-Valverde, 2012;Lobo et al, 2008), although it continues to be used in the literature (e.g., Ma & Sun, 2018;Taylor, Papeş, & Long, 2018). Other metrics have been proposed to evaluate SDMs (see Hijmans, 2012;Phillips & Elith, 2010), despite this, no measure has succeeded in replacing AUC, which is still being used in more than 80% of SDMs studies (Fourcade et al, 2018).…”

Identifying climate refugia and its potential impact on Tibetan brown bear (Ursus arctos pruinosus) in Sanjiangyuan National Park, China

“…MaxEnt model is considered one of the most efficient tools to predict species distribution with presence-only data, leading to its widespread use (Aryal eT al., 2016;Gomes et al, 2018;Lamsal, Kumar, Aryal, & Atreya, 2018;Ma & Sun, 2018;Phillips, Anderson, & Schapire, 2006). The parameters of MaxEnt model were set to: 25% for random test percentage and 1 regularization multiplier.…”

Associations Between Habitat Quality And Body Size In The Carpathian Land SnailVestia turgida: Species Distribution Model Selection And Assessment Of Performance

“…Suitability indices describe the relationship between habitat suitability score and a given environmental variable of a target species. Habitat suitability is a way to predict the suitability of habitat at a certain location for a given species or group of species based on their observed affinity for particular environmental conditions (Yi et al, 2016;Ma, Sun, 2018). SDMs therefore have been widely used for predicting distributions of species in terrestrial, freshwater and marine environments, and across taxa from many biological groups (Elith, Leathwick, 2009), with increasing numbers of publications each year (Robinson et al, 2011;Brotons, 2014).…”