Fig 1.
The overall research process of this study.
Fig 2.
Geographical location and elevation of the Nakdong River Basin in Korea (base map data source: https://gadm.org/).
(A) The location of the Korean Peninsula in Northeast Asia. (B) Digital elevation model (DEM) of the Nakdong River Basin and its rivers and mainstreams, with occurrence sites (n = 321) and genetic data locations (n = 8) of nutrias indicated as open black circles and closed red squares, respectively. The shuttle radar topography mission (SRTM) DEM with 30-m resolution was provided by the USGS Earth Explorer platform (https://earthexplorer.usgs.gov/). The genetic data location coordinates were obtained from the study by Kim et al. [28]. River and stream data were derived from the OpenStreetMap database (http://www.openstreetmap.org).
Table 1.
Bioclimatic and environmental variables used in ensemble species distribution modeling.
Fig 3.
Response curves for the most important predictor, the minimum temperature of the coldest month for the five species distribution models for nutrias (producing ΔAUC values ≥ 0. 05).
MaxEnt = Maximum Entropy; RF = Random Forest; BRT = Boosted Regression Trees; MARS = Multivariate Adaptive Regression Spline; GLM = Generalized Linear Model.
Fig 4.
Response curves for the second important predictor, the mean temperature of the warmest quarter for three of the five species distribution models for nutrias (producing ΔAUC values ≥ 0. 05).
RF = Random Forest; BRT = Boosted Regression Trees; MARS = Multivariate Adaptive Regression Spline.
Fig 5.
Response curves for the third important predictor, the precipitation of the driest quarter for four of the five species distribution models for nutrias (producing ΔAUC values ≥ 0. 05).
MaxEnt = Maximum Entropy; RF = Random Forest; MARS = Multivariate Adaptive Regression Spline; GLM = Generalized Linear Model.
Fig 6.
Response curves for the fourth important predictor, distance from water bodies for two of the five species distribution models for nutrias (producing ΔAUC values ≥ 0. 05).
BRT = Boosted Regression Trees; GLM = Generalized Linear Model.
Table 2.
Relative variable importance for each species distribution model using an increase in AUC values (ΔAUC).
Fig 7.
Ensemble map for the probability of nutria occurrence across the Nakdong River Basin in Korea (A) using binary predictions of five species distribution models (base map data source: https://gadm.org/). Different colors indicate the level of habitat suitability consensus among the predictions of five species distribution models. For example, red indicates that all five models predicted the area as suitable, while white indicates that all five models predicted the area as unsuitable. (B) Inlet map showing a digital elevation model (DEM) of the Nakdong River Basin and its mainstreams (base map data source: https://gadm.org/). The shuttle radar topography mission (SRTM) DEM with 30-m resolution was provided by the USGS Earth Explorer platform (https://earthexplorer.usgs.gov/). Stream data was derived from the OpenStreetMap database (http://www.openstreetmap.org).
Fig 8.
Mantel correlations between FST and (A) least-cost distance and (B) resistance distance calculated for eight nutria populations in the Nakdong River Basin.
Fig 9.
Least-cost paths connecting pairs of eight nutria populations in the Nakdong River Basin (base map data source: https://gadm.org/).
The middle and downstream area was enlarged. The genetic data location coordinates were obtained from the study by Kim et al. [28]. River and stream data were derived from the OpenStreetMap database (http://www.openstreetmap.org).
Fig 10.
Relationship between the genetic diversity of nutria populations and their habitat connectivity.
(A) Pearson’s r coefficients between the expected heterozygosity (He) of eight nutria populations and their habitat current flow betweenness centrality (CFBC) measured at each cost distance threshold of 5,000 to 25,000 with intervals of 5,000 in the Nakdong River Basin. Closed circles indicate significant correlations (p < 0.05). The strongest correlation was observed at a cost distance of 5,000. (B) Scatter plot and fit line between He and CFBC measured at the cost distance threshold of 5,000.
Fig 11.
Connectivity level (current flow betweenness centrality, CFBC) of habitat nodes for genetic connectivity of nutria populations (base map data source: https://gadm.org/).
(A) CFBC of habitat nodes measured at a cost distance threshold of 5,000 within the middle and downstream regions of the Nakdong River Basin. Circles represent habitat nodes. Red and yellow circles indicate priority habitat nodes of high connectivity for invasive species management. (B) Enlarged view of a key control area including high-priority habitat nodes along the Nakdong River and its major adjacent tributaries. Links connecting habitat nodes are also shown. River and stream data were derived from the OpenStreetMap database (http://www.openstreetmap.org).