Skip to main content
European Commission logo print header

PRedicting Individual fish response as a Measure of Environmental change

Final Report Summary - PRIME (Predicting individual fish response as a measure of environmental change)

Human activity, including water abstraction, flow regulation and species introductions, impacts freshwater fish populations. In addition, the influence of climate change will also become increasingly important, as shifts in temperature and precipitation will directly impact habitats. Yet, our ability to determine how fish populations may respond to such environmental changes is somewhat limited, despite ecologists frequently being requested to provide policy makers and managers with predictions under new environmental conditions. Impacts of environmental change on fish populations are traditionally assessed through modelling of physical habitat and habitat selection modelling. The problem of these habitat association models is that there is often no way of knowing whether the relationships on which the models are based will be the same under the new circumstances for which predictions are required. To gain confidence in predictions beyond the previous empirical range, models must operate on basic principles that will still apply in new environmental conditions. Habitat association models also rarely consider the ecological processes underlying individual distribution and growth, e.g. optimal habitat choice, and do not predict the effects of environmental change on survival, reproduction or dispersal of individuals, which are processes that, in combination, determine the overall population functioning.

The environmental changes that aquatic organisms will have to face in the future as a result of global changes are likely to generate an increase in water temperature as well as a change of flow patterns. Allied to habitat alteration resulting from environmental changes are changes in the abundance of prey and competitors, with competition being a key process influencing the response of animal populations to environmental change. Underpinning research identified simple rules that determined how individual fish responded to changes in the abundance of their prey and competitors, but the lack of such understanding had previously been a bottleneck in model development. Therefore, individual-based models (IBMs) were developed to overcome these problems. Their key feature was that they were based on the assumption that individuals within animal populations always behaved to maximise their own chances of survival and reproduction, thus maximising their fitness, no matter how much the environment changed. The decisions made by model animals were based on optimal foraging and game theory which were thought to provide a reliable basis for prediction.

The overall aim of PRIME was to develop, calibrate and validate IBMs to test the applicability of this approach for several European freshwater fishes and to predict how fish populations would respond to various environmental change scenarios, with subsequent demonstration of the practical application to conservation management. Consequently, the model species were selected to represent different components of global change and different conservation contexts. Namely, the model species were the northern pike, esox lucius l., used in the context of stocking strategies and three riverine species, namely the common dace, leuciscus leuciscus l., used in the context of the introduction of novel pathogen and two salmonid species, the brown trout salmo trutta and the Atlantic salmon s. Salar, used in the context of climate change and biological invasions.

Model parameterisation, calibration and validation were made using both an existing database for the pike IBM and new field-collected data for dace, trout and salmon, relating to habitat use, growth rates and trophic ecology of the model species in the study sites. The study reaches, a temporary flooded grassland for pike IBM and a chalk-river stretch, were divided into a two-dimensional grid of uniform patches, each representing an area assumed to be of regular habitat characteristics, e.g. depth, velocity and food availability. The river field study was carried out for 117 days starting in mid-June 2008 until early October 2008. This survey period was divided into two study periods, i.e. a 'normal flow' period for approximately 99 days followed by an artificially 'decreased flow' period for 18 days. The data collected during the 'normal flow' period were used to parameterise and calibrate the IBMs and the data collected during the 'decreased flow' period were used to validate the IBMs' predictive ability for conditions outside of those for which it was parameterised and calibrated, i.e. for conditions under which traditional habitat models were unreliable.

A fundamental aspect of PRIME was to understand the relationship between individual fish distribution and growth. Consequently, data collection consisted of monitoring these two parameters at individual level using a combination of innovative tools. Individuals belonging to the model species were captured and individually tagged using passive integrated transponder (PIT) tags. Subsequently, individuals' habitat use was quantified by tracking tagged individuals using a portable antenna while growth was measured on consecutive recaptures. In addition, stable isotope analyses, namely carbon and nitrogen, were used to provide baseline information on the trophic ecology of these individuals and coexisting prey and predators and to quantify inter and intra-specific variability in foraging ecology.

At the beginning of simulations IBMs were parameterised with an initial population of fish, each with an initial body-mass drawn from field observations. Each individual was allocated a dominance score according to its mass rank in the population which determined its ability to compete for foraging locations. During simulations, each fish aimed to gain mass at the maximum rate possible in the study reach and to avoid predation by choosing which patch to occupy and whether or not to feed. In our model, patch-choice decisions depended on a trade-off between the net rate of energy gain, expressed as energy assimilated from food minus energy expended, and predation risk in a patch. The rate at which energy was assimilated depended on the density of prey items, the quantity of prey fish were able to capture and their energetic content. The rate at which energy was expended was based on bioenergetic models and depended on individual size, temperature and activity level.

Generally, we observed that IBMs accurately predicted a number of empirically observed patterns. Sensitivity analyses showed that model predictions were relatively insensitive to parameters that were not measured from the real system, e.g. bioenergetic ones. These findings demonstrated that IBMs previously applied to one group of species could be transferred efficiently to another group to address fundamental and applied ecological issues. Indeed, our simulations showed that fish populations usually demonstrated non-linear and somewhat unexpected responses to various scenarios of environmental change. For instance, the pike IBM was used to predict the effects of realistic alternative management, i.e. stocking scenarios, where variable stocking densities, locations, dates and water depths at stocking locations were simulated. The IBM predicted that juvenile pike mortality could be minimised by early stocking into the deepest parts of the study as it minimised the risk of predation while maximising the time that the pike could feed profitably on zooplankton. Moreover, the dace IBM allowed us to identify the ecological mechanisms involved in the observed decrease of fitness in individuals parasitised by a non-native pathogen. The IBM was then used to predict the potential effects of parasite introduction on a naive host population located at the edge of the current European distribution of the parasite under different prevalence and parasite load scenarios. The simulations predicted that juveniles were more impacted than adults by different parasite prevalence and load, suggesting a selective mortality induced by the parasite that could have strong effects on population functioning. By the end of the project work was undertaken to calibrate and validate the salmonids' IBM for its subsequent use in predicting the potential effects of different climate change and biological invasion scenarios.

In conclusion, by using an innovative approach coupling integrative field surveys and individual-based modelling, PRIME deepened our understanding of fish responses to different components of global change and provided a decision support tool for wildlife management.