Modeling meadow distribution for conservation action in arid and semi-arid Patagonia, Argentina

Meadows are critical in arid and semi-arid Patagonia because of their importance for regional biodiversity. Despite this, little information on the spatial distribution of meadows is available, which hampers conservation planning. We modeled the spatial distribution of meadows across arid and semi-arid Patagonia, Argentina, and investigated conservation status of those areas predicted to contain meadows. We used high-resolution imagery available in Google Earth Software to visually estimate presence and absence of meadows. We used these observations and 11 socio-environmental predictor. variables to model the distribution of meadows using generalized linear, additive, boosting, and random forest models, as the basis for a final mean ensemble technique. The final ensemble model improved accuracy over any of the single models, with an accuracy (area under the curve of the receiver-operating characteristic plot) of 0.97. Based on the final ensemble model, only 0.14% of predicted meadows occur inside current International Union for Conservation of Nature level I, II or III protected areas. Our final ensemble model was an accurate representation of the distribution of. meadows in Patagonia and indicates they are severely under-represented within protected areas. This first regional map of meadow distribution across Argentinean Patagonia represents key information for planning actions to conserve this critical habitat. (C) 2013 Elsevier Ltd. All rights reserved.