Premium
Sampling interval affects the estimation of movement parameters in four species of A frican snakes
Author(s) -
Alexander G. J.,
Maritz B.
Publication year - 2015
Publication title -
journal of zoology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.915
H-Index - 96
eISSN - 1469-7998
pISSN - 0952-8369
DOI - 10.1111/jzo.12280
Subject(s) - sampling (signal processing) , biology , sampling interval , statistics , limiting , movement (music) , interval (graph theory) , ecology , zoology , mathematics , computer science , mechanical engineering , philosophy , filter (signal processing) , combinatorics , engineering , computer vision , aesthetics
Abstract Movement patterns in animals have been studied extensively, but differences in sampling frequencies produce datasets that are not always directly comparable. Because information is collected as point coordinates, measures of movement generally underestimate actual distances moved, especially when sampling frequency is low. Many animals, including most snakes, move from one location to another in discrete bouts of movement interspersed by relatively long periods of immobility. The impact of sampling interval on measures of movement in such studies remains poorly quantified, severely limiting appropriate contrasts among species, populations or individuals. We collected data during telemetry studies of four A frican snake species ( B itis arietans , B . schneideri , P ython natalensis and G onionotophis capensis ) to assess the impact of sampling frequency on two measures of movement (mean daily displacement, MDD ; moves per month, MM ), which we assessed by subsampling our datasets to simulate different sampling frequencies. For each species, we analysed two subgroups defined by their movement profiles yielding eight experimental species groups. We found significant effects of sampling interval on MDD and MM with the magnitude of the effect ( b ) of sampling interval ranging from −0.025 to −0.138 among species groups, producing errors of up to 10%. For MM , b estimates ranged between −0.313 and 0.032, yielding error estimates of up to 40%. Our study reveals the potential for sampling interval to significantly impact measures of movement. Moreover, this error can vary dramatically among species, sexes or seasons, emphasizing the importance of taking cognizance of sampling interval in movement studies. We propose a method whereby researchers explicitly account for such effects by reporting measures of displacement and movement frequency corrected to a standardized daily sampling interval to account for this error and better facilitate meta‐analysis. This will facilitate a more holistic understanding of the movement ecology of infrequently moving animals.