TY - JOUR
T1 - Investigating spatial non-stationary environmental effects on the distribution of giant pandas in the Qinling Mountains, China
AU - Ye, Xinping
AU - Yu, Xiaoping
AU - Wang, Tiejun
PY - 2020/3
Y1 - 2020/3
N2 - Analyses of species distribution have commonly been performed with global regression models by assuming species-environment interactions are spatially stationary. However, environmental variables are often spatially heterogeneous and their effects on species distribution may vary across space. Here we employed a geographically weighted logistic regression (logistic GWR) to investigate environmental effects on the distribution of giant pandas (Ailuropoda melanoleuca) in the Qinling Mountains of China. Outputs from the logistic GWR were compared with those derived from a global logistic regression in predicting panda distribution. A k-means cluster analysis was used to identify distinct zones of panda-environment relationships. We found that logistic GWR outperformed global logistic regression in terms of goodness-of-fit and predictive accuracy. Results from the logistic GWR model clearly showed both the strength and direction of the environmental effects on panda distribution changed spatially and formed distinct subareas with particular panda-environment relationships. The findings emphasize the importance of considering spatial non-stationarity in studying ecological relationships between organisms and their environments, especially for threatened species such as the giant panda with small populations in highly fragmented habitats.
AB - Analyses of species distribution have commonly been performed with global regression models by assuming species-environment interactions are spatially stationary. However, environmental variables are often spatially heterogeneous and their effects on species distribution may vary across space. Here we employed a geographically weighted logistic regression (logistic GWR) to investigate environmental effects on the distribution of giant pandas (Ailuropoda melanoleuca) in the Qinling Mountains of China. Outputs from the logistic GWR were compared with those derived from a global logistic regression in predicting panda distribution. A k-means cluster analysis was used to identify distinct zones of panda-environment relationships. We found that logistic GWR outperformed global logistic regression in terms of goodness-of-fit and predictive accuracy. Results from the logistic GWR model clearly showed both the strength and direction of the environmental effects on panda distribution changed spatially and formed distinct subareas with particular panda-environment relationships. The findings emphasize the importance of considering spatial non-stationarity in studying ecological relationships between organisms and their environments, especially for threatened species such as the giant panda with small populations in highly fragmented habitats.
KW - ITC-ISI-JOURNAL-ARTICLE
KW - ITC-GOLD
UR - https://ezproxy2.utwente.nl/login?url=https://library.itc.utwente.nl/login/2020/isi/wang_inv.pdf
U2 - 10.1016/j.gecco.2019.e00894
DO - 10.1016/j.gecco.2019.e00894
M3 - Article
VL - 21
SP - 1
EP - 13
JO - Global Ecology and Conservation
JF - Global Ecology and Conservation
SN - 2351-9894
M1 - e00894
ER -