A statistical regression approach to estimate zooplankton mortality

It is notoriously difficult to estimate mortality rates for zooplankton populations in the open ocean. In a new paper, Kvile and colleagues demonstrate that mortality estimation can be improved using a statistical regression approach (SRA) that takes into account advection and spatiotemporal trends in recruitment. Using this method on Calanus finmarchicus survey data from the Norwegian Sea–Barents Sea, they find indications of increased mortality for the oldest copepodite stage pair (CIV–CV), possibly reflecting higher predation pressure on larger copepodites.

Figure 1: It is challenging to estimate mortality rates for zooplankton populations in ocean areas dominated by strong currents. Kvile and colleagues used a new statistical regression approach to estimate mortality of Calanus finmarchicus sampled in the Norwegian Sea and the Barents Sea. The dominant surface currents in the area are the Norwegian Atlantic Current (red solid arrows), the Norwegian Coastal Current (green dotted arrows) and Arctic Water Currents (blue dashed arrows). Survey stations are indicated with open circles

Knowledge about zooplankton mortality rates is essential in order to understand variation in top-down control (e.g. predation by planktivorous fish), as well as potential responses of zooplankton populations to environmental change. Estimating mortality rates in oceanic zooplankton populations is however challenged by the difficulty in separating recruitment and mortality from advection. In short, observing a different number of individuals one day compared to the next can reflect recruitment and mortality, but also transport in or out of the area. The vertical life table (VLT) approach (Aksnes & Ohman, 1996) is commonly advocated for its robustness to advection and has been applied to estimate zooplankton mortality rates in numerous studies. However, the method is known to be sensitive to both spatial and temporal trends in recruitment (e.g. non-uniform egg production in space or time).

In a paper recently published in Journal of Plankton Research, Kristina Øie Kvile and colleagues describe a statistical regression approach (SRA) for zooplankton mortality estimation which takes into account both the effects of advection and spatiotemporal trends in recruitment. They compare the performance of the two methods on a dataset of a simulated zooplankton population subject to known mortality levels. They find that the both the VLT and the SRA accurately estimate mortality when zooplankton stage distributions are stable. However, when stage-specific abundances are influenced by trends in recruitment, the SRA performs better than the VLT. The SRA is also robust to the influence of advection.

After confirming the suitability of the SRA for spatiotemporal survey data influenced by trends in recruitment, the authors applied the method on actual survey data of Calanus finmarchicus abundances, collected between 1959 and 1993 in the Norwegian Sea–Barents Sea area. Calanus finmarchicus is the dominant mesozooplankton species in terms of biomass in the North Atlantic, including in the Atlantic waters of the Norwegian and Barents Seas, and constitutes an important food source for both pelagic fish and larvae and juveniles of demersal fish. The mortality rates for C. finmarchicus copepodite stages CI–CV estimated with the SRA were relatively low (0.03–0.07 d^-1), but indicated increased mortality at the older stage-pair (CIV–CV). Older (and larger) copepodites may be subject to higher predation pressure by visual predators. Alternatively, higher estimated mortality of older stages might reflect vertical overwintering migration of stage CV individuals to depths not covered by the survey data. However, since the survey primarily covered the period when C. finmarchicus is active in the upper waters (spring–summer), we can be relatively confident in the mortality estimates, especially for the earlier copepodite stages.

In summary, this study provides both new estimates of mortality rates for a key zooplankton species in the North Atlantic, and a promising tool to estimate mortality from non-uniform, spatiotemporal survey data.

References:

Aksnes, D., & Ohman, M. (1996). A vertical life table approach to zooplankton mortality estimation Limnology and Oceanography, 41 (7), 1461-1469 DOI: 10.4319/lo.1996.41.7.1461

Kvile, K., Stige, L., Prokopchuk, I., & Langangen, Ø. (2016). A statistical regression approach to estimate zooplankton mortality from spatiotemporal survey data Journal of Plankton Research DOI: 10.1093/plankt/fbw028

Tags: Barents Sea, Calanus finmarchicus, mortality, statistical regression approach, vertical life table, Norwegian Sea, zooplankton, top-down control By Kristina Kvile