1. The effect of total nitrogen (TN) and phosphorus (TP) loading on trophic structure and water clarity was studied during summer in 24 field enclosures fixed in, and kept open to, the sediment in a shallow lake. The experiment involved a control treatment and five treatments to which nutrients were added: (i) high phosphorus, (ii) moderate nitrogen, (iii) high nitrogen, (iv) high phosphorus and moderate nitrogen and (v) high phosphorus and high nitrogen. To reduce zooplankton grazers, 1 + fish (Perca fluviatilis L.) were stocked in all enclosures at a density of 3.7 individuals m )2 . 2. With the addition of phosphorus, chlorophyll a and the total biovolume of phytoplankton rose significantly at moderate and high nitrogen. Cyanobacteria or chlorophytes dominated in all enclosures to which we added phosphorus as well as in the high nitrogen treatment, while cryptophytes dominated in the moderate nitrogen enclosures and the controls. 3. At the end of the experiment, the biomass of the submerged macrophytes Elodea canadensis and Potamogeton sp. was significantly lower in the dual treatments (TN, TP) than in single nutrient treatments and controls and the water clarity declined. The shift to a turbid state with low plant coverage occurred at TN >2 mg N L )1 and TP >0.13-0.2 mg P L )1 . These results concur with a survey of Danish shallow lakes, showing that high macrophyte coverage occurred only when summer mean TN was below 2 mg N L )1 , irrespective of the concentration of TP, which ranged between 0.03 and 1.2 mg P L )1 . 4. Zooplankton biomass and the zooplankton : phytoplankton biomass ratio, and probably also the grazing pressure on phytoplankton, remained overall low in all treatments, reflecting the high fish abundance chosen for the experiment. We saw no response to nutrition addition in total zooplankton biomass, indicating that the loss of plants and a shift to the turbid state did not result from changes in zooplankton grazing. Shading by phytoplankton and periphyton was probably the key factor. 5. Nitrogen may play a far more important role than previously appreciated in the loss of submerged macrophytes at increased nutrient loading and for the delay in the reestablishment of the nutrient loading reduction. We cannot yet specify, however, a

1. Zooplankton use macrophytes as day-time refuge areas when trying to escape from pelagic predators. But macrophytes can also host a diverse and abundant macroinvertebrate assemblage and zooplankton are also likely to face predacious macroinvertebrates once they enter the littoral zone. This study aimed to elucidate the role of macroinvertebrates in determining the refuge capacity of macrophytes. 2. We conducted a field enclosure experiment using plastic bags and complementary laboratory feeding trials to test how macroinvertebrates counteract the benefits to zooplankton of the macrophyte refuge. The field experiment consisted of three treatments with different macroinvertebrate assemblages: without predators (WP), low abundance and diversity (LAD) and high abundance and diversity of predators (HAD -which represents lake conditions). 3. Populations of Diaphanosoma brachyurum, Bosmina huaronensis and Moina micrura (Cladocera) and of both male and female Notodiaptomus incompositus (Copepoda, Calanoida) declined (by nearly 80%) in the presence of HAD in comparison to WP and LAD treatments. 4. Feeding trials revealed that Buenoa sp. (backswimmer), adults of Palaemonetes argentinus (grass shrimp) and Cyanallagma interruptum (damselfly) had a significant negative impact on cladocerans (D. brachyurum, B. huaronensis) and the calanoid copepod population (males, females and copepodites). These predators showed a strong predation effect ranging from 75% to 100% reductions of zooplankton populations. 5. The refuge effect offered by macrophytes to zooplankton depends on and is balanced by the predacious macroinvertebrate assemblage that plants host. The risk of confronting littoral predators is high and macroinvertebrate presence can turn the macrophytes into risky areas for zooplankton.

Summary 1. The alternative state theory claims that shallow lakes may have either clear water, and be dominated by submerged macrophytes, or turbid water and be dominated by phytoplankton. Most evidence for this theory comes from studies in temperate or boreal regions of Europe. Because of differences in the strength of trophic interactions, such as in the pressure of zooplankton grazing on phytoplankton, this influential theory might not apply elsewhere. 2. Here, we test the theory for South American lakes, combining field data and Landsat satellite data. We studied the frequency distribution of primary producers and water transparency, looking for potential bimodality separating clear and turbid lakes. A bimodal distribution might be observed if there are indeed alternative states, although would not itself be sufficient evidence for the theory. Possible shifts between alternative states were analysed by comparing satellite data from 1987 to 2005. 3. In our field data, there was a bimodal pattern in phytoplankton abundance and possibly in the abundance of submerged macrophytes, but not in water transparency. Analyses of the larger satellite data set revealed bimodality in lake transparency in 2005, but less so in 1987. In 1987, the lakes were generally clearer, and the transition to higher turbidity was more gradual than in 2005. The stronger bimodality in the more recent data, and the overall lower transparency, could have been caused by an increase in fertiliser use and subsequent eutrophication but also by differences in hydrology. Further, 1987 was much wetter than 2005, which could have caused dilution of suspended particles, leading to clearer water. 4. While a bimodal distribution in the abundance of primary producers and water clarity is not decisive evidence for or against the theory of alternative states, our data clearly fail to refute it.

Summary 1. The zooplankton often undergoes diel horizontal migration (DHM) from the open water to the littoral of shallow lakes, thus avoiding predators in the former. This behaviour has functional impacts within the lake, as it enhances zooplankton survival, increases their control of phytoplankton and tends to stabilise the clear water state. However, most of the evidence supporting this migration pattern comes from cold north temperate lakes, and more evidence from tropical and subtropical areas, as well as from southern temperate areas, is needed. 2. We conducted a field study of the diel horizontal and vertical migration of zooplankton, and the horizontal distribution of potential predatory macroinvertebrates and fish, over two consecutive days in the summer in a temperate lake in the southern hemisphere. We took zooplankton samples at two depths, at three sampling stations (inside beds of aquatic macrophytes, at their edge and in open water) along three transects running from the centre of a bed of Ceratophyllum demersum to open water. At each sampling station, we also took samples of macroinvertebrates and fish and measured physical and chemical environmental variables. 3. Zooplankton (pelagic cladocerans, calanoid copepods and rotifers) avoided the shore, probably because of the greater risk from predators there. Larger and more vulnerable cladocerans, such as Diaphanosoma brachyurum and Moina micrura, were two to four times more abundant in open water than at the edge of or inside beds of macrophytes, respectively, by both day and night. Less vulnerable zooplankton [i.e. of medium body size (Ceriodaphnia dubia) or with the ability to swim fast (calanoid copepods)] were distributed evenly between open water and the edge of the plant beds. Small zooplankton, Bosmina huaronensis and pelagic rotifers, showed an even distribution among the three sampling stations. Accordingly, no DHM of zooplankton occurred, although larger organisms migrated vertically inside C. demersum stands. 4. Macrophytes contained high densities of predatory macroinvertebrates and fish. The predator assemblage, composed of large‐bodied macroinvertebrates (including odonates and shrimps) and small littoral fish, was permanently associated with submerged macrophytes. None of these groups moved outside the plant beds or changed their population structure (fish) over the diel cycle. 5. Submerged macrophyte beds do not represent a refuge for zooplankton in lakes where predators are numerous among the plants, implying a weaker top‐down control of phytoplankton biomass by zooplankton and, consequently, a more turbid lake. The effectiveness of macrophytes as a refuge for zooplankton depends on the associated assemblage of predatory macroinvertebrates and fish among the plants.