Modelling estuarine Crustacea population fluctuations in response to physico-chemical trends

Abstract

Regular samples (generally every 2 wk) of 6 estuarine crustacean species, Carcinus maenas, Liocarcinus holsatus, Crangon crangon, Palaemon longirostris, Palaemon serratus and Gammarus spp. (mainly G. zaddachi ), were taken over a 12 yr period from the cooling water intake screens of West Thurrock power station on the Thames estuary, UK. Additionally, comparative data sets for abiotic variables (freshwater flow, salinity, temperature, dissolved oxygen, pH, suspended solids, total nitrogen) were collected for the same time period. The comprehensive nature of the time series, and accompanying suite of variables, allowed the construction of statistical models for the trends in population abundance of the 6 species using multiple linear regression techniques. Statistically significant models were constructed for C. maenas, C. crangon and Gammarus spp., accurately predicting annual, and longer term, fluctuations in abundance. All models had strong seasonal components, although for C. maenas temperature was the only physico-chemical variable with significant explanatory power. The importance of temperature as a controlling variable for the species was reinforced by the inclusion of an instrumental variable to simulate a threshold temperature for foraging activity. The optimal value was found to be 8°C. C. crangon was found to be positively correlated with dissolved oxygen, but showed a slight decline in abundance over the time period. There was no significant relationship with either salinity or temperature, variables previously suggested as being important. Gammarus spp. abundance had 2 significant explanatory variables (temperature and salinity) but also demonstrated a large decrease in population size with time. L. holsatus and P. serratus are summer-occurring species, so were recorded too infrequently to adequately capture seasonal dynamics. Despite the long time series, no significant model was possible for P longirostris abundance (non-normal residuals), which has been suggested previously as having a strong relationship with salinity. The results of the study provide the first significant multiple linear regression models that accurately predict estuarine crustacean abundance. Whilst these models are useful for helping to understand variability in the Thames, it will be interesting to determine whether populations in other estuaries demonstrate relationships with similar suites of physico-chemical parameters.