Abstract
Predicting fish responses to modified flow regimes is becoming central to fisheries management. In this study we present an agent-based model (ABM) to predict the growth and distribution of young-of-the-year (YOY) and one-year-old (1+) Atlantic salmon and brown trout in response to flow change during summer. A field study of a real population during both natural and low flow conditions provided the simulation environment and validation patterns. Virtual fish were realistic both in terms of bioenergetics and feeding. We tested alternative movement rules to replicate observed patterns of body mass, growth rates, stretch distribution and patch occupancy patterns. Notably, there was no calibration of the model. Virtual fish prioritising consumption rates before predator avoidance replicated observed growth and distribution patterns better than a purely maximising consumption rule. Stream conditions of low predation and harsh winters provide ecological justification for the selection of this behaviour during summer months. Overall, the model was able to predict distribution and growth patterns well across both natural and low flow regimes. The model can be used to support management of salmonids by predicting population responses to predicted flow impacts and associated habitat change.
Highlights
Individual fish moving in response to reduced flow will be limited by the availability of suitable areas for feeding[20,21]
We parameterised the agents in FishMORPH to have salmonid drift-feeding behaviour and bioenergetics using models obtained from literature
We report the results comparing the growth and distribution from fish in the field study against the same pattern from virtual fish following the different movement rules
Summary
We first describe the field study site and report the data collected from it. We caught a total of 1453 YOY Atlantic salmon, 283 YOY and 81 1+brown trout of which 94, 96, and 79 were marked recaptures respectively Over both flows, mean fish body mass increased and there was greater population variation in SGR during the MLF period (Fig. 2). Bayesian estimation of the probability of the difference between mean growth rates show that observed and virtual fish matched well for YOY brown trout across both flow periods, while YOY Atlantic salmon SGRs matched the NF period but not the MLF period (Fig. 2 and see Supplementary Table S3). Sensitivity analysis showed the model was most sensitive to parameters associated with bioenergetics whilst parameters pertaining to behavioural drift-feeding submodels had smaller effects (see Supplementary Table S4)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
