Species richness and diversity
We observed a different response in species richness and biodiversity for the various taxonomic groups. For most of the variables included, species richness and diversity showed a unimodal relationship with TP ( Figs. 1 and 2) and were also variably related to lake area and/or depth. A different pattern was found for species richness of zooplankton and submerged macrophytes, which declined monotonically with TP.
Numerous studies have shown that species richness of fish in lakes, in accordance with the theory of island biography and habitat diversity ( MacArthur & Wilson, 1967), is strongly linked to lake area ( Magnuson, 1976; Browne, 1981; Keller & Crisman, 1990; Bachman et al., 1996 ). We found fish species diversity to be independent of lake area, which may, however, be ascribed to the small variation in lake area characterizing our study lakes ( Table 2). How the species richness and diversity of the fish community change along a trophic gradient is not clear. While several studies indicate a decline in species richness with increasing eutrophication (for a review see Larkin & Northcote, 1969; Lee, Jones & Jones, 1991), others have not recorded any changes ( Bachman et al., 1996 ; Eckmann & Rösch, 1998). In our study of mainly shallow lakes, species richness and the three selected diversity indices were all unimodally related to TP. Depth added significantly to species richness and diversity when based on abundance, which might be explained by a greater number of available niches in deep lakes.
Table 2. Correlation analyses of morphometric and nutrient data based on loge-transformed data
|TP||ns||−0.36 *||0.64 ***|
|Surface area||–||0.49 **||ns|
|Mean depth||0.48 **||–||ns|
Most studies have reported an increase in microcrustacean zooplankton species richness with lake size (e.g. Patalas, 1972; Fryer, 1985; Dodson, 1991, 1992). Others have found a closer correlation with lake depth ( Keller & Conlon, 1994), which may be explained by the larger heterogeneity ( Keller & Conlon, 1994) and perhaps also the lower fish predation pressure in deeper lakes ( Keller & Conlon, 1994; Jeppesen et al., 1997 ). In our study of mainly shallow lakes, however, lake depth contributed significantly only to the variation in the species richness of rotifers, but not of cladocerans or copepods. The higher sensitivity of rotifers to lake depth may reflect the fact that rotifer species, unlike most microcrustaceans, are adapted to life in the often oxygen-poor hypolimnion ( Hofmann, 1985), implying that the abundance of species is probably higher in stratified than in non-stratified lakes. We found that the species richness of all selected zooplankton taxa decreased monotonically with TP and was not independent of lake size. This contrasts with the findings of Dodson (1991, 1992) who, by compiling data from 32 European and 66 North American lakes, found that species richness in both sets of lakes increased with lake size. In addition, it was unimodally related to phytoplankton production, which in other studies has been shown to be linked with TP ( OECD, 1982). Furthermore, Dodson (1992) found that species richness was higher in areas rich in lakes. Our study includes only a few oligotrophic lakes so we cannot exclude the possibility that a unimodal pattern would emerge if more nutrient-poor lakes had been included. Another explanation of the high species richness in the lower TP classes, however, could be that the abundance of submerged macrophytes increases habitat heterogeneity and then, presumably, the species richness. In shallow mesotrophic lakes in which the area and volume occupied by submerged macrophytes may be high, the presence of macrophytes may have a great impact on overall structural complexity. This contrasts with deep lakes in which the macrophytes are restricted to near-shore areas. In contrast to species richness, all three selected diversity indices were unimodally and positively related to TP. The higher species richness in the lower TP classes does, therefore, not result in higher diversity.
For phytoplankton, both species richness and the Shannon–Wiener Index were unimodally related to TP, whereas the importance of lake depth and lake area varied: diversity increasing with lake area and species richness with lake depth. As for zooplankton, different patterns may be found in the literature ( Margalef, 1978). For example, studies of the historical development in species richness and diversity of diatoms in Lake Washington also revealed a unimodal relationship with increased eutrophication, while Margalef (1980) provided evidence of higher diversity in oligotrophic than in eutrophic lakes and marine waters. The Shannon–Wiener Index for phytoplankton, based on abundance, varies from 1.0 to 4.5 bits, but typically falls in the range 2.4–2.6 ( Harris, 1986). Our data are at the lower end of this range, which may reflect an overall lower diversity in lakes (and shallow coastal estuaries) compared with the open ocean ( Margalef, 1978, 1980).
Cross-analysis of data from 641 lakes revealed that lake area contributed most significantly to the variation in macrophyte species richness in Scandinavian lakes ( Rørslett, 1991). Residual analyses showed low species richness in lakes with low pH and a unimodal relationship to eutrophication, with mesotrophic–eutrophic lakes supporting more species than oligotrophic and hypertrophic lakes. In accordance with the latter, case studies of cultural eutrophication of mesotrophic lakes have typically shown a loss in species richness ( Ozimek & Kowalczewski, 1984; Kowalczewski & Ozimek, 1993, Sand-Jensen, 1997). We also found a significant decrease in macrophyte species richness in our study. The low contribution of oligotrophic lakes to our dataset may explain why we, unlike Rørslett (1991), found a monitonical decline with increasing TP. Hence, a recent study of Danish lakes including more oligotrophic lakes showed a tendency towards a unimodal relationship between species richness and TP (Vestergaard & Sand-Jensen, unpublished). Compared with the other taxa in our study, the percentage decline in species richness with increasing TP was particularly high for submerged macrophytes. This may reflect the concurrent significant decrease in maximum depth distribution and macrophyte-covered area with eutrophication and, accordingly, a loss of habitat heterogeneity for the plants ( Fig. 3). For floating-leaved plants, which are less sensitive to increased turbidity, we observed a unimodal relationship with TP.
The general unimodal response in species richness and/or diversity for the different taxonomic groups to increasing TP or trophic state is to be expected ( Dodson, 1992): only a few species occur in distilled water and hypertrophic conditions lead to loss of species due either to competitive exclusion ( Tilman, 1982) or adverse conditions (e.g. high pH, Hansen, Christensen & Sortkjær, 1991; and low oxygen in the hypolimnion during summer or under ice during winter, Magnuson, 1976). Various other factors also play a role, however, and may obscure clear-cut relationships. We have already mentioned morphometry, physical structure and biological complexity in general. Variation in top-down control also plays a role, and it is generally believed that high predation or grazing pressure result in loss of diversity of prey organisms ( Paine, 1969), though the effect seems to depend on nutrient state ( Proulx & Mazumder, 1998). Increased fish predation may have contributed to the decline in species richness and diversity of zooplankton at high TP, but it does not explain the decline in the same variables for phytoplankton as the grazing pressure on phytoplankton was low at high TP. The variation in disturbance of various kinds may also be important and it is generally accepted that species richness and diversity are highest at intermediate disturbances ( Connell, 1978).
The increase in CPUEw and the decline in the percentage of piscivores with increasing TP agree with the results from other studies in both temperate and subtropical regions ( Bays & Crisman, 1983; Hanson & Peters, 1984; Persson et al., 1988 ; Quiros, 1990; Bachman et al., 1996 ). The change in the fish community from dominance by perch and pike in mesotrophic lakes to exclusive dominance by cyprinids in eutrophic lakes is well-documented in other studies of North European lakes ( Svärdson, 1976; Leach et al., 1977 ; Persson, 1983; Persson et al., 1988 ). The superiority of roach in eutrophic lakes is attributed to a high potential growth rate along with a higher predation efficiency on cladocerans ( Persson, 1983) and an ability to exploit smaller zooplankton prey ( Stenson, 1979; Lessmark, 1983). Moreover, cyprinids are omnivorous, while large perch are piscivores. Finally, the loss of habitat complexity with the disappearance of submerged macrophytes, and thus increased turbidity, disfavours percids and, furthermore, augments intra-specific competition among them ( Persson et al., 1988 ).
The enhanced dominance of cyprinids was accompanied by a decline in average size of both cyprinids and perch. The decline probably reflects enhanced competition for food, which in turn may be mediated by reduced predation of piscivores on young fish. Thus, the fraction of large piscivorous individuals among the perch population declined. In addition, the average weight of pike increased considerably, probably as a result of enhanced cannibalism due to reduced structural complexity (less macrophytes) and, consequently, the loss of refuges ( Grimm & Backx, 1990). As large pike control small prey fish less efficiently than small pike ( Grimm & Backx, 1990), the incapability to control young planktivores decreased.
It is generally believed that ruffe biomass increases with increasing TP ( Biro, 1977; Hartman & Nümann, 1977; Persson, 1983; Bergman, 1991). Our study covered a larger TP gradient than in previous work, however, and showed a significant unimodal relationship peaking in class 3 (0.1–0.2 mg P L−1). The ruffe is benthivorous ( Johnsen, 1965; Bergman, 1991) and a competitor to young benthivorous perch. It has been argued that an increase in ruffe abundance with increasing TP reflects the fact that the foraging ability of ruffe (unlike that of perch) is largely independent of light ( Bergman, 1991). Thus, ruffe may forage effectively in turbid eutrophic lakes ( Bergman, 1991; Bergman & Greenberg, 1994). However, in eutrophic lakes ruffe faces other competitors. The observed decline at high TP is probably caused by the increase in the abundance of bream ( Fig. 3). The bream is supposedly a superior competitor in these lakes, both because it is an efficient predator on benthic invertebrates and because it, unlike ruffe, feeds efficiently in the pelagic and therefore may alternate between the benthic and pelagic feeding mode. That ruffe is an inferior competitor at high cyprinid density is evident from several biomanipulation experiments showing high abundance of ruffe during the first 1–2 years after cyprinid removal, followed by a major decline when young perch, after having initially taken advantage of the ‘empty niche’ in the pelagic, reached the benthivorous state (E. Jeppesen and M. Søndergaard, unpublished). The enhanced importance of pike–perch in eutrophic lakes is well-known from other European lakes ( Svärdson, 1976) and in accordance with their adaptations (high light sensitivity of their eyes, Ali, Ryder & Anctil, 1977) for foraging efficiently in turbid water.
Confirming most other studies, zooplankton biomass increased with increasing TP ( McCauley & Kalff, 1981; Hanson & Peters, 1984). In contrast to a number of studies ( Brooks, 1969; Bays & Crisman, 1983), however, we did not find any differences in the proportion of major taxa to biomass. Rotifers typically constituted 10–15% of biomass in all TP classes, this being consistent with results from oligo-mesotrophic Norwegian lakes ( Hessen, Faafeng & Andersen, 1995). The results from the Danish and Norwegian lakes suggest that changes in the contribution of rotifers are generally not to be expected over a substantial TP gradient (3–1000 μg P L−1) in north temperate lakes, which contradicts studies from subtropical lakes, in which an increase in rotifer:zooplankton biomass was found with increasing trophic state ( Bays & Crisman, 1983). Some studies have found an increasing share of cladocerans with TP at the expense of copepods ( Patalas, 1972; Rognerud & Kjellberg, 1984; Straile & Geller, 1998). In our study, the contribution of cladocerans and copepods to total biomass in the five TP classes ranged from 50 to 58% and 34 to 42%, respectively, and the contribution of cladocerans tended to decrease and copepods to increase slightly, though not significantly, with increasing TP. Major changes occurred in the contribution of calanoids and cyclopoids to copepod biomass, however, as well as in the contribution and mean individual weight of large cladocerans. A shift from calanoids to cyclopoids with increasing trophic state has been suggested by several authors ( Gliwicz, 1969; Patalas, 1972; Bays & Crisman, 1983) and has been related to changes in food size spectra ( Pace, 1986) or predation ( Hessen et al., 1995 ; Jeppesen et al., 1997 ). Owing to their ability to make evasive jumps, cyclopoids are often less vulnerable to fish predation than calanoids ( Winfield et al., 1983 ), although there are exceptions ( Brooks, 1969). Moreover, juvenile cyclopoids might be superior to calanoids at high food concentrations ( Santer & Van den Bosch, 1994), and increased predation on juvenile calanoids by adult cyclopoids may further reduce calanoid abundance ( Straile & Geller, 1998). Several authors record an increased share of small cladocerans with increasing trophic state (e.g. McNaught, 1975; Patalas, 1972). This is often attributed to enhanced fish predation ( Brooks, 1969; Lyche, 1990). Accordingly, we found an increase in the contribution of Daphnia spp. among cladocerans and of cladoceran mean body weight after being released from fish predation after fish removal and fish kills ( Fig. 9). World-wide, biomanipulation experiments in eutrophic lakes have shown similar results provided that fish biomass was substantially reduced (e.g. Benndorf, 1990; Hansson et al., 1998 ).
The changes in the zooplankton community structure and biomass appeared to cascade down to phytoplankton. Assuming that during summer cladocerans ingest organic carbon corresponding to 100% of their biomass per day (compared with copepods 50% per day and rotifers 200% per day) ( Hansen et al., 1992 ) and assuming that they exclusively feed on phytoplankton, then grazing amounted to 59% of phytoplankton biomass per day in class 1 and 16–19% per day in classes 4 and 5. The latter is so low that zooplankton probably have little effect on phytoplankton growth, while in class 1 and 2 lakes it seems probable that zooplankton apply a considerable grazing pressure to phytoplankton. In addition, the observed reduction in the individual weight of cladocerans and Daphnia spp. presumably further adversely affected the grazing capacity of phytoplankton, because small-bodied cladocerans are less efficient grazers on large-sized phytoplankton than the large-bodied forms ( Gliwicz, 1977, 1990). How the alteration in fish community structure and biomass, mediated by the changes in trophic state, affects herbivory on phytoplankton has been the subject of extensive discussion ( DeMelo, France & McQueen, 1992; Carpenter & Kitchell, 1992). Our data, as well as those of Leibold (1990) and Sarnelle (1992), suggest a more significant effect in eutrophic lakes than in mesotrophic lakes.
Several authors have argued that filamentous cyanobacteria prevent grazer control by zooplankton in eutrophic lakes, which may explain a decrease in herbivory from mesotrophic to eutrophic lakes ( Elser & Goldman, 1990; Carney & Elser, 1990). Supporting this view, controlled laboratory experiments have shown that dense cyanobacterial assemblages can prevent the growth of large cladocerans (e.g. Lampert, 1981; Dawidovicz, Gliwicz & Gulati, 1988; Gliwicz, 1990). Furthermore, in field experiments Daphnia spp. have occasionally failed to respond to minor, and in a few cases even to major, reductions in planktivorous fish populations (e.g. Van Donk et al., 1990 ; Riemann et al., 1990 ; Moss, Stansfield & Irvine, 1991; DeMelo, France & McQueen, 1992). The results from Danish lakes indicate, however, that the role of cyanobacteria in the decline in herbivory from mesotrophic to eutrophic lakes is considerably less significant than that of fish. First, the zooplankton:phytoplankton ratio and the size of cladocerans are low not only in eutrophic lakes dominated by cyanobacteria but also in the most nutrient-rich lakes dominated by edible green algae. Second, a reduction in fish predation on zooplankton, due either to the removal of cyprinids or to fish kills, caused a major increase in the abundance and average size of cladocerans and a decrease in phytoplankton biomass in eutrophic lakes, irrespective of whether cyanobacteria or chlorophytes dominated previously ( Jeppesen et al., 1997 ; Søndergaard, Jeppesen & Jensen, 1998; Jeppesen et al., 1999 ; authors’ unpublished data). Finally, whereas in Danish lakes CPUEw contributed highly significantly to the variation in the zooplankton:phytoplankton ratio and cladoceran mean size, the percentage contribution of cyanobacteria to the total biomass of phytoplankton did not ( Jeppesen et al., 1997 ).
The strong evidence of a decrease in herbivory, mediated by planktivorous fish from mesotrophic to hypertrophic lakes implies that the cascading effect of fish manipulation on phytoplankton in shallow lakes will be greatest under hypertrophic conditions, though the benefit of such a manipulation may be transient as planktivorous fish are highly successful in such lakes ( Fig. 3; Kitchell et al., 1977 ; Leach et al., 1977 ; Persson et al., 1988 ; Jeppesen et al., 1990 ). Therefore, a return to a turbid state and to low zooplankton grazing seems likely, even though it may be somewhat delayed in shallow lakes with an extensive growth of submerged macrophytes ( Meijer et al., 1994 ). The data in Fig. 3 suggest that the likelihood of obtaining high zooplankton grazing in shallow lakes increases markedly when summer mean TP drops below about 0.1 mg P L−1. Therefore, the prospects of long-term success of biomanipulation in shallow lakes are most probably greatest below this threshold ( Jeppesen et al., 1990 ). The results shown in Figs. 3–8 may leave the impression that the shift from clear to turbid conditions occurred gradually with increasing TP. However, several case studies of shallow lakes have shown a step-wise shift and, within certain nutrient regimes (typically 0.5–0.15 mg P L−1; Jeppesen et al., 1990 ), lakes may shift between the two states ( Moss, 1990; Jeppesen et al., 1990 ; Scheffer, 1990), this being the case in Danish lakes ( Jeppesen et al., 1990 ). Lake-specific differences in nutrient thresholds, and the fact that we also included somewhat deep lakes, may explain the lack of step-wise shifts in the present study.
In summary, species richness and diversity of most of the selected taxonomic groupings were unimodally related to TP in the nutrient range from 0.02 to 1.0 mg P L−1 and, moreover, often related positively to mean depth and/or lake area. Only the species richness of zooplankton and submerged macrophytes deviated by showing a monotonical decline with increasing TP. Major changes occurred in the community structure and size distribution of fish, which apparently cascaded to the lower trophic levels. The share of piscivores declined markedly and, moreover, the size of the predatory fish changed towards those size classes less likely to control cyprinids (larger pike and smaller perch). Enhanced predation appeared to lead to a marked reduction of the zooplankton:phytoplankton biomass ratio and the grazing pressure upon phytoplankton. The latter may have been exacerbated by a decline in the size of grazing zooplankton.