Categorizing the geometry of animal diel movement patterns with examples from high-resolution barn owl tracking

• Published in: Movement ecology Vol. 11; no. 1; p. 15
• Main Authors: Luisa Vissat, Ludovica, Cain, Shlomo, Toledo, Sivan, Spiegel, Orr, Getz, Wayne M
• Format: Journal Article
• Language: English
• Published: England BioMed Central Ltd 21-03-2023; BioMed Central; BMC
• Subjects: Age factors; Algorithms; Analysis; Animals; Annual cycles; ATLAS tracking system; Biotelemetry; Canonical activity modes (CAMs); Classification; Cluster analysis; Clustering; Dependent variables; Diel activity; Diel activity routines (DARs); Ecology; Elongation; Frequency dependence; Global positioning systems; GPS; Hierarchical movement path segmentation; Land use; Localization; Methodology; Methods; Movement; Owls; Populations; Principal components analysis; Relocation; Routines; Sex; Sex differences; Spatial data; Statistical models; Tyto alba; Variation; Israel; Canonical activity modes (CAMs); ATLAS tracking system; Hierarchical movement path segmentation; Biotelemetry; cluster analysis; Diel activity routines (DARs)
• ISSN: 2051-3933; 2051-3933
• DOI: 10.1186/s40462-023-00367-4

Movement is central to understanding the ecology of animals. The most robustly definable segments of an individual's lifetime track are its diel activity routines (DARs). This robustness is due to fixed start and end points set by a 24-h clock that depends on the individual's quotidian schedule. An analysis of day-to-day variation in the DARs of individuals, their comparisons among individuals, and the questions that can be asked, particularly in the context of lunar and annual cycles, depends on the relocation frequency and spatial accuracy of movement data. Here we present methods for categorizing the geometry of DARs for high frequency (seconds to minutes) movement data.
Our method involves an initial categorization of DARs using data pooled across all individuals. We approached this categorization using a Ward clustering algorithm that employs four scalar "whole-path metrics" of trajectory geometry: 1. net displacement (distance between start and end points), 2. maximum displacement from start point, 3. maximum diameter, and 4. maximum width. We illustrate the general approach using reverse-GPS data obtained from 44 barn owls, Tyto alba, in north-eastern Israel. We conducted a principle components analysis (PCA) to obtain a factor, PC1, that essentially captures the scale of movement. We then used a generalized linear mixed model with PC1 as the dependent variable to assess the effects of age and sex on movement.
We clustered 6230 individual DARs into 7 categories representing different shapes and scale of the owls nightly routines. Five categories based on size and elongation were classified as closed (i.e. returning to the same roost), one as partially open (returning to a nearby roost) and one as fully open (leaving for another region). Our PCA revealed that the DAR scale factor, PC1, accounted for 86.5% of the existing variation. It also showed that PC2 captures the openness of the DAR and accounted for another 8.4% of the variation. We also constructed spatio-temporal distributions of DAR types for individuals and groups of individuals aggregated by age, sex, and seasonal quadrimester, as well as identify some idiosyncratic behavior of individuals within family groups in relation to location. Finally, we showed in two ways that DARs were significantly larger in young than adults and in males than females.
Our study offers a new method for using high-frequency movement data to classify animal diel movement routines. Insights into the types and distributions of the geometric shape and size of DARs in populations may well prove to be more invaluable for predicting the space-use response of individuals and populations to climate and land-use changes than other currently used movement track methods of analysis.