In the Baltic Sea, increased dominance of ephemeral and bloom-forming algae is presently attributed to increased nutrient loads. Simultaneously, coastal predatory fish are in strong decline. Using field data from nine areas covering a 700-km coastline, we examined whether formation of macroalgal blooms could be linked to the composition of the fish community. We then tested whether predator or nutrient availability could explain the field patterns in two small-scale field experiments, by comparing joint effects on algal net production from nutrient enrichment with agricultural fertilizer and exclusion of larger predatory fish with cages. We also manipulated the presence of invertebrate grazers. The abundance of piscivorous fish had a strong negative correlation with the large-scale distribution of bloom-forming macroalgae. Areas with depleted top-predator communities displayed massive increases in their prey, small-bodied fish, and high covers of ephemeral algae. Combining the results from the two experiments showed that excluding larger piscivorous fish: (1) increased the abundance of small-bodied predatory fish; (2) changed the size distribution of the dominating grazers, decreasing the smaller gastropod scrapers; and (3) increased the net production of ephemeral macroalgae. Effects of removing top predators and nutrient enrichment were similar and additive, together increasing the abundance of ephemeral algae many times. Predator effects depended on invertebrate grazers; in the absence of invertebrates there were no significant effects of predator exclusion on algal production. Our results provide strong support for regional declines of larger predatory fish in the Baltic Sea promoting algal production by decreasing invertebrate grazer control. This highlights the importance of trophic interactions for ecosystem responses to eutrophication. The view emerges that to achieve management goals for water quality we need to consider the interplay between top-down and bottom-up processes in future ecosystem management of marine resources.

In this study, the combined influence of vegetation density and water turbidity on habitat utilization of a prey fish, 0þ year perch Perca fluviatilis, under predation risk (pike, Esox lucius) was investigated. The vegetated habitat was overall preferred over the open habitat in the presence of a predator. The level of turbidity, and to a lesser extent vegetation density, however, influenced the response of 0þ year perch. The use of the vegetated habitat was lower in very turbid than in clear and turbid conditions, suggesting reduced antipredator behaviour in very turbid water. The effect of vegetation density on antipredator behaviour was only present in clear water, where the use of a structural refuge decreased with increasing vegetation density. No such effect was observed in turbid and very turbid water. The results showed that the structuring role of vegetation or habitat complexity may diminish with increased turbidity. The observed masking effect of turbidity suggests that predator-prey interactions in vegetated habitats are more complex than what has generally been thought. # 2004 The Fisheries Society of the British Isles

Substrate spawning fish are believed to be selective in their choice of spawning habitat, yet few studies have shown the relative importance of different characteristics in terms of habitat quality. We used an extensive and detailed dataset to identify the factors that govern both large-scale (10 3 to 10 5 m) and local-scale (10 1 to 10 2 m) selection by a substrate-spawning fish, the Eurasian perch Perca fluviatilis L. Distribution of spawning habitat was strongly dependent on habitat characteristics defined by substrate, wave exposure, temperature and depth. The most important predictor was the type of spawning substrate, which generally consisted of different types of vegetation. Substrates providing rigidity and structural complexity were preferred, despite abundant presence of other substrate types. Shallow depth and sheltered areas were also selected habitat characteristics. The response to temperature was scale-dependent, with a stronger selection expressed at the local scale. The specific selectivity suggests that spawning patterns can be successfully modelled with sufficient detail using only a few fundamental environmental variables. Wave exposure and depth are readily available for large-scale spatial predictions, while temperature and substrate require further development in most coastal areas. The high specificity of the characteristics determining habitat quality suggests that it should be possible to apply this modelling approach for identification and conservation of spawning habitats of Eurasian perch and other substrate-spawning fishes in coastal waters.

Summary1. Eutrophication is severely affecting species distributions and ecosystem functioning in coastal areas. Targets for eutrophication reduction have been set in the Baltic Sea Action Plan (BSAP) using Secchi depth, a measure of water transparency, as the main status indicator. Despite the high economic costs involved, the potential effects of this political decision on key species and habitats have not been assessed. 2. In a case study including species central to coastal ecosystem functioning, we modelled the effects of changing Secchi depth on the distribution of bladderwrack Fucus vesiculosus and eelgrass Zostera marina vegetation as well as recruitment areas of the main predatory fish species, perch Perca fluviatilis and pikeperch Sander lucioperca. Specifically, we explored the effects of changing Secchi depth on species distributions under a set of scenarios based on the BSAP, using three fundamentally different modelling techniques: maximum entropy, generalized additive and random forest modelling. 3. Improved Secchi depth (reduced eutrophication) was predicted to cause a substantial increase in the distribution of bladderwrack, while the distribution of eelgrass remained largely unaffected. For the fish, a large increase in perch recruitment areas was predicted and a concurrent decrease in recruitment areas of pikeperch. These changes are likely to have effects on biodiversity and ecosystem services. 4. The three modelling methods exposed differences in the quantitative predictions for species with a weaker coupling to Secchi depth. Qualitatively, however, the results were consistent for all species. 5. Synthesis and applications. We show how ecological effects of environmental policies can be evaluated in an explicit spatial context using species distribution modelling. The modelspecific responses to changes in eutrophication status emphasize the importance of using ensemble modelling for exploring how species distributions may respond to alternative management regimes. A pronounced difference in response between species suggests that eutrophication mitigation will have consequences for ecosystem functioning, and thus ecosystem goods and services, by inducing changes in the simple food webs of the Baltic Sea. These model predictions form a basis for spatially explicit cost-benefit estimates under different scenarios, providing valuable information for both decision-makers and the wider society.

No abstract