Synthesis and modelling of zooplankton at pan-Arctic scales

Abstract

Zooplankton are an important link between primary producers and higher trophic levels. They are sensitive indicators of change in the Arctic ecosystem due to their relatively short lifespan. To date, the greatest impediment to detect changes in the Arctic zooplankton community at pan-Arctic scales is the absence of a reference baseline. To contribute to baseline data, I taxonomically analyzed zooplankton samples from the Canada Basin collected during August and September of 2003-2006. Over 50 taxonomic groups were identified, but copepods dominated abundance and biomass. Non-copepod abundance was dominated by larvaceans, while non-copepod biomass was dominated by chaetognaths. I applied multivariate analysis to look at patterns in community similarity, finding a tendency to separate the years sampled. The sample analysis served the larger goal of my research: collation of zooplankton data from online databases, reports, papers, and through scientific cooperation with scientists throughout the Arctic. In total, 13,014 zooplankton samples were assembled, containing over 200,000 individual taxonomic records spanning the period from 1921 to 2012. I also assembled 25 environmental layers for variables of possible influence on zooplankton distribution. Using these data, I employed the Geographic Information System ArcMap, as well as the data mining approaches TreeNet and RandomForests to predict the climatological mean distribution and abundance of seven ecologically and numerically important epipelagic copepod species (Calanus finmarchicus, C. glacialis, C. hyperboreus, Metridia longa, M. pacifica, N. cristatus, and P. glacialis) on a pan-Arctic scale from 60° - 90°N. The model predicted the overall distribution and abundance characteristics of each species well, but it also predicted potential niches for these or sibling species in areas where they are known to be absent. The model correctly associated species advected to the Arctic with corresponding water masses, while Arctic endemic species were more strongly associated with geographic variables. Continued assimilation of new data, plus rescuing and consolidating older datasets, are critical pathways toward both enhancing this baseline, and building the observational time-series necessary for studying changes in the Arctic zooplankton community.