Connecting the dots in databases of endangered species: a Bayesian hierarchical imputation strategy for missing Peary caribou (Rangifer tarandus pearyi) population data
Autor: | Dong-Kyun Kim, George B. Arhonditsis, Samarth Kaluskar, E. Agnes Blukacz-Richards, C. A. Johnson |
---|---|
Rok vydání: | 2020 |
Předmět: |
0106 biological sciences
education.field_of_study geography.geographical_feature_category biology Ecology 010604 marine biology & hydrobiology Ecological Modeling Population Endangered species biology.organism_classification Species at Risk Act 010603 evolutionary biology 01 natural sciences Overexploitation Geography Peary caribou Arctic Archipelago Imputation (statistics) education Ecology Evolution Behavior and Systematics |
Zdroj: | Ecological Complexity. 43:100846 |
ISSN: | 1476-945X |
Popis: | Peary caribou (Rangifer tarandus pearyi) is listed as Endangered under Canada's Species at Risk Act. Studying rare and endangered species can be particularly challenging due to the constraints posed by incomplete datasets owing to poor weather conditions, lack of technology, organizational deficiencies, and high survey costs in remote areas. A defensible way to remedy data gaps and thus improve the robustness of any given modelling exercise is through imputation schemes, which are typically selected on the basis of missing data patterns. In this study, we develop a regression-based imputation method aiming to identify the linkages among available population records in time and space. One of the fundamental assumptions of our imputation model is that there is a subset of primary islands that act as the core areas from where the Peary caribou populations migrate to secondary or satellite islands. Specifically, we delineated six distinct geographic clusters (Banks, Axel Heiberg, Bathurst, Boothia, Melville and Mackenzie King) across the Canadian Arctic Archipelago, and within each cluster Peary caribou seasonal inter-island movements take place in any given year. Parameterized with Bayesian inference techniques, our imputation modelling strategy has a flexible structure that can accommodate non-monotonic spatiotemporal patterns, and was thus able to capture more than 65% of the variability in our dataset. The two islands (Banks and Northwest Victoria) of the Banks island complex exhibited an average decline rate of 6% per year over the past four decades, which collectively reflects the dramatic population decrease from the early 1970s until the late 1990s, as well as the distinct recovery after the early 2000s. Similar “wax-and-wane” cycles characterize the Peary caribou population patterns on Melville and Bathurst island complexes. Our analysis provides evidence of positive rate of change of the population trends on Axel Heiberg and Ellesmere Islands, which likely stems from pockets on those islands, where favorable climatic/geomorphological conditions, and rich vegetation prevail. In stark contrast, overharvesting, higher predation, adverse climatic conditions, and human-induced forage shortage have led to a dramatic decline, nearing their extirpation, on the Boothia island complex. |
Databáze: | OpenAIRE |
Externí odkaz: |