Search for: All records

Abstract BackgroundAnimal movement data are increasingly used to make ecological inferences, as well as to inform conservation and management actions. While advanced statistical methods to estimate behavioral states from these datasets have become widely available, the large number to choose from may make it difficult for practitioners to decide which method best addresses their needs. To guide decisions, we compared the behavioral state estimates and inferences from three methods (movement persistence models [MPM], hidden Markov models [HMM], and mixed-membership method for movement [M4]) when analyzing animal telemetry data. Tracks of post-breeding adult male green sea turtles (Chelonia mydas) were treated as an empirical example for this method comparison. The effect of temporal scale on behavioral state estimates was also investigated (at 1, 4, and 8 h time steps). ResultsThe HMM and M4 models produced relatively similar behavioral state estimates (compared to the MPM) and estimated anywhere from three to five states depending on the time interval of the tracks and the method used. Likewise, for all three methods, sampling movement at coarser time scales smoothed estimates of behavioral transitions. Additionally, the selection of movement metrics for analysis by the HMM and M4 also appeared to be a critical decision regarding state estimation and interpretation. At the longest time step (8 h), all three models were able to distinguish area-restricted search (ARS) behavior from migratory behavior, with greater nuance estimated by the HMM and M4 methods. By comparison, the MPM was the only model that was able to identify fine-scale behavioral patterns when analyzing the shortest time step (1 h). Moreover, the analysis of tracks with short time steps via MPM identified likely periods of resting during long-distance migration, which had only previously been hypothesized in green turtles. ConclusionsWhile there is no single best method to estimate behavioral states, our findings demonstrate that results can vary widely among different statistical methods and that model assumptions should be thoroughly checked during the model fitting process to reduce any potential biases. Thus, practitioners should carefully consider which methods best address their needs while also accounting for the inherent properties of their telemetry dataset. 

Abstract Species distribution models (SDMs) have become increasingly popular for making ecological inferences, as well as predictions to inform conservation and management. In predictive modeling, practitioners often use correlative SDMs that only evaluate a single spatial scale and do not account for differences in life stages. These modeling decisions may limit the performance of SDMs beyond the study region or sampling period. Given the increasing desire to develop transferable SDMs, a robust framework is necessary that can account for known challenges of model transferability. Here, we propose a comparative framework to develop transferable SDMs, which was tested using satellite telemetry data from green turtles (Chelonia mydas). This framework is characterized by a set of steps comparing among different models based on (1) model algorithm (e.g., generalized linear model vs. Gaussian process regression) and formulation (e.g., correlative model vs. hybrid model), (2) spatial scale, and (3) accounting for life stage. SDMs were fitted as resource selection functions and trained on data from the Gulf of Mexico with bathymetric depth, net primary productivity, and sea surface temperature as covariates. Independent validation datasets from Brazil and Qatar were used to assess model transferability. A correlative SDM using a hierarchical Gaussian process regression (HGPR) algorithm exhibited greater transferability than a hybrid SDM using HGPR, as well as correlative and hybrid forms of hierarchical generalized linear models. Additionally, models that evaluated habitat selection at the finest spatial scale and that did not account for life stage proved to be the most transferable in this study. The comparative framework presented here may be applied to a variety of species, ecological datasets (e.g., presence‐only, presence‐absence, mark‐recapture), and modeling frameworks (e.g., resource selection functions, step selection functions, occupancy models) to generate transferable predictions of species–habitat associations. We expect that SDM predictions resulting from this comparative framework will be more informative management tools and may be used to more accurately assess climate change impacts on a wide array of taxa.