Limits...
Using stochastic gradient boosting to infer stopover habitat selection and distribution of Hooded Cranes Grus monacha during spring migration in Lindian, Northeast China.

Cai T, Huettmann F, Guo Y - PLoS ONE (2014)

Bottom Line: Our field work in 2013 using systematic ground-truthing confirmed that this prediction was accurate.Based on this study, we suggest that Lindian plays an important role for migratory birds and that cultivation practices should be adjusted locally.Furthermore, public education programs to promote the concept of the harmonious coexistence of humans and cranes can help successfully protect this species in the long term and eventually lead to its delisting by the IUCN.

View Article: PubMed Central - PubMed

Affiliation: College of Nature Conservation, Beijing Forestry University, Beijing, China.

ABSTRACT
The Hooded Crane (Grus monacha) is a globally vulnerable species, and habitat loss is the primary cause of its decline. To date, little is known regarding the specific habitat needs, and stopover habitat selection in particular, of the Hooded Crane. In this study we used stochastic gradient boosting (TreeNet) to develop three specific habitat selection models for roosting, daytime resting, and feeding site selection. In addition, we used a geographic information system (GIS) combined with TreeNet to develop a species distribution model. We also generated a digital map of the relative occurrence index (ROI) of this species at daytime resting sites in the study area. Our study indicated that the water depth, distance to village, coverage of deciduous leaves, open water area, and density of plants were the major predictors of roosting site selection. For daytime resting site selection, the distance to wetland, distance to farmland, and distance to road were the primary predictors. For feeding site selection, the distance to road, quantity of food, plant coverage, distance to village, plant density, distance to wetland, and distance to river were contributing factors, and the distance to road and quantity of food were the most important predictors. The predictive map showed that there were two consistent multi-year daytime resting sites in our study area. Our field work in 2013 using systematic ground-truthing confirmed that this prediction was accurate. Based on this study, we suggest that Lindian plays an important role for migratory birds and that cultivation practices should be adjusted locally. Furthermore, public education programs to promote the concept of the harmonious coexistence of humans and cranes can help successfully protect this species in the long term and eventually lead to its delisting by the IUCN.

Show MeSH

Related in: MedlinePlus

Partial dependence plots for the predictor variables employed in the roosting site selection model; (A) water depth (cm), (B) distance to village (m), and (C) coverage of deciduous leaves (%).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3935961&req=5

pone-0089913-g004: Partial dependence plots for the predictor variables employed in the roosting site selection model; (A) water depth (cm), (B) distance to village (m), and (C) coverage of deciduous leaves (%).

Mentions: After the data were technically cleaned and formatted, we ran small sample sizes in TreeNet to generate a total of 249 optimal trees from 1000 possible trees. The ROC curve indicated the model discrepancy (accuracy). We then integrated the area under the curve (AUC) to assess the model performance or predictive power [37]. From the result of model, we found that the roosting site selection model achieved a high AUC value (learning/training = 0.98/0.80), indicating a relatively high discrimination and prediction accuracy. For each predictor variable, TreeNet also provided a relative importance score (Table 2), which describes the individual contribution of each predictor in explaining the response variable in the tree models [34]. The model showed that the five most important predictors were the water depth, distance to village, coverage of deciduous leaves, open water area, and density of plants. However, two variables (food quantity and type of food) did not contribute to the model. In addition, we obtained the response curves of these predictors. The response curve shows the relative index as a function of the predictor in the context of the multivariate model: a positive partial dependence indicates preference, and a negative partial dependence indicates avoidance. Detailed data on the predictors indicate that the Hooded Cranes selected the roosting site where water was between 2 to 32 cm deep (Figure 4A), where the distance to the village was greater than 2,760 m (Figure 4B), and where the coverage of deciduous leaves was greater than 48% (Figure 4C).


Using stochastic gradient boosting to infer stopover habitat selection and distribution of Hooded Cranes Grus monacha during spring migration in Lindian, Northeast China.

Cai T, Huettmann F, Guo Y - PLoS ONE (2014)

Partial dependence plots for the predictor variables employed in the roosting site selection model; (A) water depth (cm), (B) distance to village (m), and (C) coverage of deciduous leaves (%).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3935961&req=5

pone-0089913-g004: Partial dependence plots for the predictor variables employed in the roosting site selection model; (A) water depth (cm), (B) distance to village (m), and (C) coverage of deciduous leaves (%).
Mentions: After the data were technically cleaned and formatted, we ran small sample sizes in TreeNet to generate a total of 249 optimal trees from 1000 possible trees. The ROC curve indicated the model discrepancy (accuracy). We then integrated the area under the curve (AUC) to assess the model performance or predictive power [37]. From the result of model, we found that the roosting site selection model achieved a high AUC value (learning/training = 0.98/0.80), indicating a relatively high discrimination and prediction accuracy. For each predictor variable, TreeNet also provided a relative importance score (Table 2), which describes the individual contribution of each predictor in explaining the response variable in the tree models [34]. The model showed that the five most important predictors were the water depth, distance to village, coverage of deciduous leaves, open water area, and density of plants. However, two variables (food quantity and type of food) did not contribute to the model. In addition, we obtained the response curves of these predictors. The response curve shows the relative index as a function of the predictor in the context of the multivariate model: a positive partial dependence indicates preference, and a negative partial dependence indicates avoidance. Detailed data on the predictors indicate that the Hooded Cranes selected the roosting site where water was between 2 to 32 cm deep (Figure 4A), where the distance to the village was greater than 2,760 m (Figure 4B), and where the coverage of deciduous leaves was greater than 48% (Figure 4C).

Bottom Line: Our field work in 2013 using systematic ground-truthing confirmed that this prediction was accurate.Based on this study, we suggest that Lindian plays an important role for migratory birds and that cultivation practices should be adjusted locally.Furthermore, public education programs to promote the concept of the harmonious coexistence of humans and cranes can help successfully protect this species in the long term and eventually lead to its delisting by the IUCN.

View Article: PubMed Central - PubMed

Affiliation: College of Nature Conservation, Beijing Forestry University, Beijing, China.

ABSTRACT
The Hooded Crane (Grus monacha) is a globally vulnerable species, and habitat loss is the primary cause of its decline. To date, little is known regarding the specific habitat needs, and stopover habitat selection in particular, of the Hooded Crane. In this study we used stochastic gradient boosting (TreeNet) to develop three specific habitat selection models for roosting, daytime resting, and feeding site selection. In addition, we used a geographic information system (GIS) combined with TreeNet to develop a species distribution model. We also generated a digital map of the relative occurrence index (ROI) of this species at daytime resting sites in the study area. Our study indicated that the water depth, distance to village, coverage of deciduous leaves, open water area, and density of plants were the major predictors of roosting site selection. For daytime resting site selection, the distance to wetland, distance to farmland, and distance to road were the primary predictors. For feeding site selection, the distance to road, quantity of food, plant coverage, distance to village, plant density, distance to wetland, and distance to river were contributing factors, and the distance to road and quantity of food were the most important predictors. The predictive map showed that there were two consistent multi-year daytime resting sites in our study area. Our field work in 2013 using systematic ground-truthing confirmed that this prediction was accurate. Based on this study, we suggest that Lindian plays an important role for migratory birds and that cultivation practices should be adjusted locally. Furthermore, public education programs to promote the concept of the harmonious coexistence of humans and cranes can help successfully protect this species in the long term and eventually lead to its delisting by the IUCN.

Show MeSH
Related in: MedlinePlus