Molluscan faunal turnover in the Plio-Pleistocene of the tropical western Atlantic has been attributed to drops in temperature or primary productivity, but these competing hypotheses have not been assessed ecologically. To test these alternatives, we compiled data on changing molluscan life habits and trophic composition over 12 million years derived from 463 newly made collections from the southwestern Caribbean. Shelf ecosystems have altered markedly in trophic structure since the Late Pliocene. Predatory gastropods and suspension-feeding bivalves declined significantly in abundance, but not in diversity, and reef-dwellers became common. By contrast, all other ecological life habits remained remarkably stable. Food-web changes strongly support the hypothesis that declining regional nutrient supply had an increasing impact on regional macroecology, culminating in a faunal turnover.

The European Water Framework Directive establishes the need to define stream type-specific reference conditions to identify ''high ecological status''. Methods for selecting reference sites using a priori criteria have been proposed by many authors. A review of these criteria revealed that the most relevant criteria for streams and rivers were those related to riparian vegetation, diffuse and point sources of pollution, river morphology and hydrological conditions and regulation. In this work, we propose 20 criteria that reflect the characteristics of Mediterranean streams and their most frequent disturbances for the selection of reference sites in Mediterranean streams in Spain. We studied 162 sites located in 33 Mediterranean basins belonging to five stream types. Of the locations, 57% were selected as a priori reference sites by having applied the proposed criteria. Reference sites were identified for all stream types except for ''large watercourses'' which includes the lower reaches of some rivers in this study area. This a priori selection of reference sites was subjected to validation using the macroinvertebrate community by applying of an IBMWP threshold, which is considered to be an indicator of undisturbed sites in Mediterranean streams. This approach determined that whole of this selection (100%) could be considered valid reference sites. Furthermore, we identified differences in the reference conditions for each stream type on the basis of macroinvertebrate assemblage composition.

1. According to the guidelines of the European Water Framework Directive, assessment of the ecological quality of streams and rivers should be based on ecotype-specific reference conditions. Here, we assess two approaches for establishing a typology for Mediterranean streams: a top-down approach using environmental variables and bottom-up approach using macroinvertebrate assemblages. 2. Classification of 162 sites using environmental variables resulted in five ecotypes: (i) temporary streams; (ii) evaporite calcareous streams at medium altitude; (iii) siliceous headwater streams at high altitude; (iv) calcareous headwater streams at medium to high altitude and (v) large watercourses. 3. Macroinvertebrate communities of minimally disturbed sites (n ¼ 105), grouped using UPGMA (unweighted pair-group method using arithmetic averages) on Bray-Curtis similarities, were used to validate four of the five ecotypes obtained using environmental variables; ecotype 5, large watercourses, was not included as this group had no reference sites. 4. Analysis of similarities (ANOSIM A NO S IM ) showed that macroinvertebrate assemblage composition differed among three of the four ecotypes, resulting in differences between the bottom-up and top-down classification approaches. Siliceous streams were clearly different from the other three ecotypes, evaporite and calcareous ecotypes did not show large differences in macroinvertebrate assemblages and temporary streams formed a very heterogeneous group because of large variability in salinity and hydrology. 5. This study showed that stream classification schemes based on environmental variables need to be validated using biological variables. Furthermore, our findings indicate that special attention should be given to the classification of temporary streams.

The effects of grazing by stream invertebrates on algal biomass and spatial heterogeneity were tested experimentally in flow-through microcosms with natural substrates (rocks). One experiment tested the effects of fixed densities of three species of grazers (the caddisfly Allomyia sp. and two mayflies, Epeorus deceptivus and Baetis bicaudatus) on periphyton. Baetis was tested with and without chemical cues from fish predators, which reduced grazer foraging activity to levels similar to the less mobile mayfly (Epeorus). Mean algal biomass (chlorophyll a; chl a) was reduced in grazer treatments compared to ungrazed controls, but there were no differences among grazer treatments. Algal heterogeneity (Morisita index) increased with grazer mobility, with the highest heterogeneity occurring in the Baetis-no fish treatment (most mobile grazer) and the lowest in the caddisfly treatment (most sedentary grazer). A second experiment used a three factorial design, and tested whether initial resource distribution (homogeneous vs. heterogeneous), Baetis density (high vs. low) and fish odor (present vs. absent) affected grazer impact on algal resources. Abundances of Baetis and chl a on individual rocks were recorded to explore the mechanisms responsible for the observed distributions of algae. Initial resource heterogeneity was maintained despite being subjected to grazing. Mean chl a was highest in controls, as in experiment I, and effects of Baetis on algal biomass increased with grazer density. There were no fish effects on algal biomass and no effects of grazer density or fish on algal heterogeneity. At the scale of individual rocks Baetis was unselective when food was homogeneously distributed, but chose high-food rocks when it was heterogeneously distributed. Results of these mechanistic experiments showed that Baetis can track resources at the scale of single rocks; and at moderate densities mobile grazers could potentially maintain periphyton distributions observed in natural streams.