Estimation of the nesting season of marine turtles from incomplete data: statistical adjustment of a sinusoidal function
Abstract
Because of logistical and financial constraints, nest counts of marine turtles are often limited in time and space. To overcome this difficulty, we developed a numerical model that fits the seasonal pattern of marine turtles nesting from complete or fragmented datasets. The duration of the main nesting season, the position and amplitude of its maximum as well as the residual number of nests, outside of the main season are obtained numerically by a least square adjustment. For the seven complete time series at our disposal (Dermochelys coriacea and Lepidochelys olivacea turtles, coast of French Guiana), the model reproduces the seasonal pattern with a correlation of r≥0.97. When applied on a fragmented dataset, the model accuracy depends on the duration and on the temporal distribution of the monitoring (effort equally distributed during the entire season or concentrated on a part of it only). As a result of this study, we clearly advocate a strategy of monitoring distributed all over the nesting season. Following this recommendation, the model estimates the annual number of nests with a median error lower than 10% when considering only 50 days of monitoring.