Can thermal metabolic physiology information predict fish behaviour observed on Baited Remote Underwater Video?

Abstract

Globally, climate change is placing fish populations under pressure through multiple environmental stressors including ocean warming. Fish may respond to these stressors through physiological acclimation, altering their behaviour or shifting their distributions. However, since the physiology and behaviour of fishes is linked, understanding this relationship is critical for improving our understanding of the likely response of fishes to a rapidly changing climate. Up to now there is limited information on the link between physiological attributes and the wild behaviour of fishes, primarily because of methodological difficulties of observing behaviour in aquatic environments. However, Baited Remote Underwater Videos (BRUVs) allows us to observe fish behaviour in the wild and with the increase in use and the advancement of BRUV systems provides a technique to better understand the wild behaviour of fishes. When this information is combined with appropriate metabolic physiology experiments, it may provide information on the link between physiology and wild behaviour. This study aimed to improve our understanding of the link between metabolic physiology and the wild behaviour of a linefish species, the fransmadam, Boopsiodea inornata by combining thermal metabolic information with wild behaviour observed using BRUVs. A total of 45 live adult and subadult B. inornata were collected from an area (Cape St Francis) that is heavily exploited by the hook and line fishery. Fish were transported to the NRF SAIAB Aquatic Ecophysiology Research Platform (AERP) laboratory at the Department of Ichthyology and Fisheries Science (DIFS), Rhodes University, and acclimated for at least two months at 16 °C. Standard metabolic rate (SMR), maximum metabolic rate (MMR) and aerobic scope (AS) of B. inornata was quantified using flow-through respirometry techniques at 8 °C, 12 °C, 16 °C, 20 °C, and 24 °C. The relationship between the metabolic rates and temperature were modelled using a second order polynomial relationship and a population-level aerobic scope curve was developed for the species. Behavioural information was collected from existing stereo-BRUV videos collected by the NRF SAIAB Marine Remote Imagery Platform (MARIP) in Algoa Bay, Cape Recife and Cape St Francis. These videos were available at temperatures between 10 °C and 18 °C. The videos were analysed using EventMeasure software and the “MaxN” (maximum number of individuals observed within a single frame), “time to arrive” (first time for individual to appear within a single frame once BRUV on seafloor) and “time to feed” (time for first individual to feed on BRUV bait canister within a single frame) was recorded. The relationships between fish behaviour and temperature were modelled using a GAM with the Negative Binomial family for “MaxN”, and a GLM with Tweedie family for “time to arrive” and “time to feed”. The SMR of B. inornata ranged from 0.58 to 3.86 O2mg.min-1.kg-1 across test temperatures, with increased variability at the higher temperatures. Temperature had no significant effect on the polynomial relationship of SMR (p-value = 0.20, R2 = 0.58) but was a significant predictor for the linear relationship (p < 0.01). MMR ranged from 1.64 O2mg.min-1.kg-1 (SD = ± 0.41) at 8 °C to 7.39 O2mg.min-1.kg-1 (SD = ± 1.33) at 20 °C across test temperatures with increased variability at the higher temperatures. Temperature had a significant effect on MMR for both linear (p < 0.01) and polynomial relationship (p < 0.01, R2 = 0.77). The resultant AS ranged from 0.39 O2mg.min-1.kg-1 (SD = ± 0.47) at 8 °C to 4.51 O2mg.min-1.kg-1 (SD = ± 1.00) at 20 °C. Temperature had a significant effect on AS for both the linear (p < 0.01) and polynomial relationship (p < 0.01, R2 = 0.41). The findings suggest an optimal peak AS at a temperature of 18.5 °C, which coincided with a relatively low SMR and a peak in MMR, and declines in metabolic performance below 12 °C and above 20 °C. “MaxN” ranged from three to 110 individuals. Highest variability in “MaxN” was observed between 16 °C and 18 °C. Temperature did not have a significant effect on “MaxN” (p = 0.22, R2 = 0.10). “Time to arrive” ranged from 2.25 minutes (17.5 °C) to 38.69 minutes (11.4 °C). Temperature (p < 0.01) and “MaxN” (p < 0.01) both had a negatively significant effect. The interaction effect between “MaxN” and temperature revealed a positive significant effect on “time to arrive” (p < 0.01, R2 = 0.33). “Time to feed” ranged from 3.58 minutes (13.6 °C) to 50.92 minutes (11.74 °C). Temperature (p < 0.01) and “MaxN” (p < 0.01) both had a significant effect on “time to feed”. In addition, there was a significant relationship between “time to feed” and the interaction between “MaxN” and temperature (p < 0.01, R2 = 0.71). The findings suggest an increase in “MaxN” and decrease in “time to arrive” and “time to feed” with highest “MaxN”, fastest “time to arrive” and “time to feed” around 17 °C, whereafter “MaxN” decreases and “time to arrive” and “time to feed” increases slightly until 18 °C. The peak physiological performance (AS) at 18.5 °C closely corresponded to the peak in “MaxN” and fastest “time to arrive” and “time to feed” around 17 °C. Similarly, the low AS at 8 °C corresponded to the lowest mean “MaxN”, and the highest times for the “time to arrive” and “time to feed”. The findings of this study indicate that there may be a link between the metabolic physiology and behaviour of fish in the wild at average and cold temperatures and suggests that aerobic scope (AS) may be informative for understanding responses to environmental change in the wild. It also highlights the potential of archived BRUVs for understanding the wild response of fishes to environmental change. However, this study was limited by the availability of BRUVs at warmer temperatures and by a lack of information on the dissolved oxygen concentration in the wild. Therefore, future research should endeavour to collect this information. While this approach is valuable, it represents only one of several methods (including bioenergetic models, growth experiments, and survival analyses) for assessing thermal performance, and which may provide complementary insights into fitness, population persistence, and productivity. Future research should integrate these multiple approaches, alongside physiological and environmental data such as dissolved oxygen availability, to generate robust performance curves that capture both individual and population-level responses to changing thermal environments.