Date:
2018/06/13
Time:
18:00
Room:
K305 Alvar
Remote Sensing for biodiversity studies of very high spatial resolution
(Oral)
, , ,
SEE PEER REVIEW
Remote Sensing (RS) is currently one of most important tool for Earth Observation. Many biodiversity and conservation studies depend on RS imagery and techniques as main source of environmental data. However, data are often limited to the available satellite imagery. In the case of local studies, satellite imagery frequently lacks adequate spatial or temporal resolution. Drones can provide valuable data with very high spatial resolution. However, when study areas are very small, even drones are not a good solution. Here, we present a study case where we modelled the distribution of several individuals of lizards using a simple camera attached to a stick and a matrix of temperature/humidity data-loggers to obtain several environmental layers with very high resolution.
Fieldwork was performed near Porto (Portugal) during May-June 2016 in a mesocosm of 20×20 m. We captured 25 adults (15 males and 10 females) of the Iberian lizard Podarcis bocagei and marked each individual with a unique combination of three non-toxic colour dyes. A researcher walked randomly through the mesocosm looking for lizards, several times per day throughout the whole period of daily activity. Lizards’ positions were recorded with a Trimble GPS receptor (~10 cm error). We built an orthophoto map with a spatial resolution of 2 cm from a set of 1152 photos captured by a Canon compact camera fastened to a stick. Photos were processed with Agisoft Photoscan 1.2.0. We created a digital elevation model with a pixel resolution of 2 cm using 3016 accurate altitude points obtained with a RTK (Real Time Kinematic)-GPS and a triangulated irregular network. We classified the orthophoto with a supervised maximum likelihood algorithm in ArcGIS, using four different classes (refuges, vegetation, bare soil and organic soil). We recreated Bioclim variables by combining data from 27 temperature and 23 for humidity/temperature Maxim’s iButton dataloggers. For each individual we also calculated the distance to males and to females, excluding the focal lizard. From the set of 22 variables, we selected 11 variables with a correlation lower than 0.6. We calculated realised niche models for each individual and for all individuals together with the correlative method for presence-only data Maxent.
All models obtained an AUC higher than 0.8. The most important variables were related to distances to males or to females and to climate (isothermality, minimum temperature and humidity), organic soil and vegetation. Our very high spatial resolution models provided information about the differential space use by each individual lizard. Correlative models can identify the most suitable areas inside the home range, similarly to core areas estimated from kernel algorithms. Overall, RS tools provided high quality data on both animal presence and its environmental context which allowed better interpretation of the spatial patterns in this species outperforming other methods.