As part of the Collaborative Research Grant “Determining Factors Affecting Distribution and Population Variability of the Ice-obligate Weddell Seal”, we proposed to quantify the spatial distribution of the species and to determine environmental factors that explain its distribution around Antarctica at multiple scales (regional to continental). To do this, we used the Digital Globe Inc.’s Tomnod platform to expose images of the Antarctic coast to hundreds of thousands of citizen scientists. Tomnod is a crowd-sourcing platform that assists users in searching for and tagging “features” (e.g., seals) on satellite images. Specifically for Point Blue, our goals included the development of statistical analyses to: 1) determine presence/absence of seal populations and factors that influence occupancy during summers of 2010 and 2011 in the Ross Sea; 2) estimate the numbers of seals in images based on the results of the crowd counts, accounting for tagger error and other sources of miscounts; 3) describe the most suitable habitat for breeding Weddell seals around the entirety of the Antarctic coastline, comparing environmental factors between locations; and 4) identify the seal “hot spots” – areas where populations thrive, as well as areas where populations are declining or sporadically present.

We successfully accomplished our goals: we were able to investigate the habitat characteristics that determine seal presence, and to estimate the number of seals at multiple spatial locations based on citizen scientist crowd counts. To date, because the very difficult locations where seals are found, most of our knowledge on Weddell seals comes from a single location and a limited number of seals. Therefore, our results represent significant advances in the study of the ecology of Weddell seals, as we are now able to detect and count them anywhere in the continent. Relevant outcomes of our work thus far include a peer-reviewed publication on the determination of the preferred habitats for Weddell seals in the Ross Sea. Another outcome is an R-language based statistical package (available through the open collaboration platform GitHub) that provides information about fast-ice coverage at different spatial resolutions for the entire continent, as well as other important information (bathymetry, proximity to penguin colonies, etc.). Additionally, we completed the analyses and are working on a second publication with the methodology to estimate the number of seals based on crowd counts, including the estimation of error in these counts, and an adjustment to match the image-based counts to ground counts. This adjustment is needed because at the time images are taken a percentage of the seals will be diving under the ice and would not be counted.  Notable results include evidence suggesting some degree of competition for food resources with Adélie penguins, and convergence in habitat preferences with Emperor penguins. We further noted the importance of stable fast-ice, which is absolutely necessary but not the sole determinant of seal presence.  Our analyses strongly evidence that one type of satellite sensor produced imagery that was prone to larger counting error. Lastly, it is a well-known fact from ground counts that seals are more actively diving (thus fewer of them on-ice for counting) during some hours of the day. We were able to detect this pattern using the crowd counts, evidencing the effectiveness of our methodology.  We are still working on determining the seal “hot spots” around the continent for a third publication.

In summary, our work resulted in a new methodology to estimate the numbers and habitat preferences of Weddell seals, new insights on what constitutes good seal habitat, and suggestive evidence of biological factors affecting seal presence. These are all new and ground-breaking results, considering that up to this point the study of Weddell seals was limited to just one location in the entire continent where access to seal colonies was possible. Our results are in the process of, or have been submitted to peer-reviewed journals. More generally, the statistical analyses we developed have broader applicability to wildlife counts conducted by crowds of citizen scientists, as they are independent of our specific methodology.

The broader impacts of our work pertain to the proper management of Antarctic toothfish fisheries. Toothfish is a critical food resource for seals recovering from the substantial energy demands of lactation. A fishery is operating in the Ross Sea with no knowledge of its impact on Weddell seals. Under the mandate of the Commission for the Conservation of Antarctic Marine Living Resources (CCAMLR), the fishery must operate with no detrimental effect to the ecosystem. Our methodology will permit the development of a timeline of spatial changes in seal populations. With these data and information about the location and intensity of toothfish extractions, it will be possible to ascertain to some degree of confidence the level of impacts of the fishery on seal populations, and therefore whether it is operating within the CCAMLR mandate.

Last Modified: 05/16/2019
Modified by: Leonardo A Salas