Description of the winter habitat of grassland birds with remote sensors and visual estimation

ABSTRACT Understanding habitat preferences of grassland birds declining is important for their conservation. Currently, the use of remote sensing technology to describe the habitat of grassland birds is a novel tool in Mexico which may allow for more accurate assessments of grassland habitat. High-resolution photographs and a protocol established by Bird Conservancy of the Rockies that uses ocular estimation was used in order to estimate vegetation cover within areas where individual sparrows of the genus Ammodramus were recorded in two sites located in Durango. Forty location points were randomly selected from detections (n = 1881) recorded from the follow-up of 33 individuals of A. bairdii and 23 of A. savannarum, by telemetry. Vegetation metric was obtained and from high-resolution photographs we created an orthomosaic with supervised classification in 4 classes of vegetation cover (%). At each bird location point, the percentage of each vegetation cover class within a 5 m radius area around the point was estimated. We did not find a significant difference between vegetation cover obtained by a high-resolution photographs or ocular estimations (p≤0.05) by species. Both species were found in areas with grass cover similar to those reported using different methods (61.24±4.07%, 62.78±4.24%). These results indicate that the use of remote sensing provide favorable information for the characterization of grassland bird’s habitat.RESUMEN Comprender la preferencia del hábitat de aves de pastizal que declinan sus poblaciones es importante para su conservación. Actualmente el uso de sensores remotos para describir el hábitat de aves en México es reciente y brinda estudios en menor tiempo y costo. El objetivo del trabajo fue describir la cobertura del suelo en áreas con gorriones del género Ammodramus en dos sitios de pastizal de Durango, con el protocolo de Bird Conservancy of the Rockies e imágenes de alta resolución. Se seleccionaron 40 puntos de localización aleatoriamente a partir de detecciones (n=1881) registradas del seguimiento de 33 individuos de A. bairdii y 23 de A. savannarum, mediante telemetría. Se obtuvo una métrica de vegetación y un ortomosaico con clasificación supervisada en 4 clases, donde se insertaron los puntos de localización, realizando un buffer de 5 metros de radio y generando porcentajes de clases. Ambas técnicas se correlacionan y demuestran ser útiles para cuantificar las variables observadas (p≤0.05); no existen diferencias significativas entre especies, ya que usan lugares con cobertura de pasto similares a los reportados en otros estudios (61.24±4.07%, 62.78±4.24%). Estos resultados muestran que la tecnología geoespacial tiene gran potencial para la descripción del hábitat de aves de pastizal