Most rotifer taxa are believed to have powerful passive dispersal strategies (Dumont 1983; Pennak 1978) though knowledge of the extent of biogeographic distribution for particular species is modified as understanding of a confusing taxonomy (created by the great degree of parthenogenesis and extensive morphological plasticity in some species) changes (Dumont 1983; Sudzuki 1964; Ruttner-Kolisko 1993). Most pelagic species appear to have fairly wide physiological tolerances (Gannon and Stemberger 1978). In spite of these wide tolerances and powerful dispersal strategies, planktonic rotifer communities in lakes are often strongly dominated by a single species (Green 1993).

Relative density and dominance by particular rotifer taxa is related in part to lake trophic state (Gannon and Stemberger 1978) and trophic state indicators like total phosphorous (Siegfried et al. 1989; Berzins and Bertilsson 1989) and conductivity (Berzins and Bertilsson 1989). Though boundaries between trophic states are not sharply defined, lake trophy is a holistic indicator of physical and chemical conditions (Wetzel 1975), of the productivity, composition (Hutchinson 1967; Wetzel 1975), and size structure (Gliwicz 1969) of the phytoplankton, and of bacterial production (Geller and Muller 1981; Wetzel 1975). The structure of the microorganismic community (planktonic algae, bacteria, and protists) is linked to lake trophic state (Pejler 1983; Stockner and Shortreed 1989; Stockner 1987; Hutchinson 1967), and individual rotifer species have different microorganismic food preferences and food size requirements (Pourriot 1977; Arndt 1993) because of the connections between feeding (mechanisms, requirements, and efficiency) and the microorganismic community (available food size and type). For example, Brachionid rotifers such as Keratella and Kellicottia species are relatively unselective microfiltrators which feed on picoplankton (0.2 - 2 um) and small nannoplankton (2- 20 um; Pourriot 1977; Arndt 1993; Moore 1978). Small-bodied rotifers like the above mentioned taxa also have lower food concentration thresholds than larger rotifers (Stemberger and Gilbert 1987a), thereby allowing population growth when food sources are relatively less dense. Because microbial systems in ultraoligo- and oligotrophic lakes tend to be dominated by pico- and nannoplankton (Stockner and Shortreed 1989; Stockner 1987; Pejler 1983), Brachionids may dominate rotifer communities in these unproductive lakes (Pejler 1983). Conversely, dominant rotifer taxa in lakes of greater productivity are more likely to include grasping feeders or predators, such as Polyarthra or Synchaeta species, which feed on larger microorganisms (Pourriot 1977). This is likely since lakes of greater productivity also are more likely to contain larger microorganisms (Stockner 1987; Pejler 1983). Therefore, physico-chemical conditions (and lake trophic state) operate through microorganismic food resources to structure the rotifer communities.

Besides characteristics of the physico-chemical environment and food resources, rotifer communities are structured by the presence and density of crustacean zooplankton which may act as predators, competitors, or both. Daphnids, as obligate filter feeders, utilize food size fractions which are within the same range as that utilized by microfiltrating rotifers: that is, in the 0.2 - 20 um range, depending on species (Gilbert 1985; Geller and Muller 1981). Daphnids may also utilize much larger foods (Gilbert 1985), and appear to be competitively superior to most rotifers under all but very productive conditions (Gilbert 1985).

Diaptomid copepods may either filter feed or exhibit predatory behavior (Zaret 1980; Williamson 1987). While filter feeding allows them to exploit algae (Zaret 1980; Moore 1978), they have been reported to feed on relatively larger microbes and microzooplankton including rotifers. While Diaptomids generally reject loricated rotifers (hard-bodied taxa like Brachionids) as food (Williamson 1987), they prey on soft-bodied rotifer taxa including Polyarthra and Synchaeta species (Williamson 1987, Anderson 1980).

Density of particular rotifer taxa may also be affected by vertebrate predators. Large or colonial rotifers like Asplanchna or Conochilus may be directly preyed upon by fish (Stenson 1982). Changes in density of some rotifers may also occur indirectly because of their interactions with crustacean zooplankton (Stenson 1982), the larger of which may be preyed upon by fish (Anderson 1972).

The objectives of this study were to determine the distribution and relative density of rotifers in mountain lakes in North Cascades National Park Service Complex, and to relate their distribution and density to physico-chemical conditions, the presence of crustacean zooplankton, and the presence of vertebrate predators. Few studies of rotifers conducted in North America have focused on western mountain systems (Larson et al. 1994; Neill 1984; Anderson 1980).

Lake depth was determined with a handheld sonar gun along transects parallel and perpendicular to the long axis of a lake. Surface area was estimated by digitizing 7.5 min USGS topographical maps. Elevation was determined from these maps. Each time a lake was sampled, water temperature was measured at 1 m intervals from the lake surface to the bottom over the deepest point in the lake using an Omega B71A digital thermo-couple. Water temperature at 1 m below the surface was recorded at each sampling occasion. Water samples were taken from a depth of 1 m over the deepest point in a lake using a 1.5 L Van Dorn bottle. Frozen filtered and unfiltered water samples were transported to the Forest Sciences Laboratory at Oregon State University for chemical analysis (total Kjeldahl nitrogen (TKN); total phosphorous; ortho phosphorous; nitrate; ammonia; PH; alkalinity; conductivity).

Vertical tows for zooplankton (crustaceans and rotifers) were taken from 66 lakes between 1989 and 1993 (Table 1). Sampling occurred only between the period after ice-out and before the return of inclement fall weather (approximately mid-June through mid-September). Samples were taken with a 20 cm diameter number 25 (64 urn mesh) zooplankton net. The net was lowered either to within one meter of the bottom of the lake, or to at least 10 meters in deep lakes, and towed upward at a rate of 0.5 meters per second. Three replicate vertical tows were taken near the deepest spot in the lake, except for lakes sampled in 1989 when only one vertical tow was taken. In the field, samples were preserved in 5% neutral sugar formalin (Haney and Hall 1973). In the lab, samples were split using a Folsom plankton splitter. The split portion was poured into a settling chamber and left to settle for 24 hours. Organisms were counted using an inverted microscope at 100X magnification.

Mean densities of each taxon at each sampling occasion (referred to as 'sample means') were calculated from the three replicate vertical tows (except for 1989 data). These data were used to calculate the proportional abundance of each taxon among all lakes. A single mean for each taxon in each lake was calculated by averaging data within each year (if the lake had been sampled more than once in a particular year), then averaging again over all years (if the lake had been sampled more than one year). These data are referred to as 'overall means.'

Hierarchical agglomerative cluster analysis was used to group lakes according to overall means of proportional density of rotifer taxa. Cluster analysis was performed only on lakes where overall mean rotifer density was > 1.0 per liter (30 lakes). The resemblance measure selected was relative Euclidean distance, and the fusion strategy was group average. Means of physical and chemical parameters for each cluster of lakes and means of absolute density of groups of crustacean zooplankton were calculated. These groups were: all adult Cladocerans (primarily Daphnia rosea and Holopedium gibberum, but also others; see Liss et al. 1995); adult Daphnids (primarily Daphnia rosea, but also D. longiremis, D. middendorffiana, and D. schodleri); adult large-bodied Diaptomids (primarily Diaptomus kenai, but also D. arcticus and D. leptopus); and adult small-bodied Diaptomids (primarily Diaptomus tyrrelli, but also D. lintoni). Detrended correspondence analysis was also run on species proportional abundance data from these 30 lakes, and the resulting axis scores for each lake were correlated, through use of Pearson product-moment correlations, with chemical and physical data, and with proportional density of the major crustacean zooplankton.

Scatter plots and chi-square contingency table analyses were used to examine relationships between density of rotifers and crustacean zooplankton groups. Data used for these analyses were sample means from the 30 lakes described above. A cumulative frequency histogram was plotted for the density (number per liter) of loricated rotifers (all taxa combined, excluding Keratella armadura). The curve of this histogram first climbed steeply for each whole number increase. Fourty-four percent of all samples had less than one loricated rotifer per liter. An additional twelve percent of the samples had between one and two rotifers per liter. The curve of the histogram then tapered off. Samples which had densities above 5.0 rotifers per liter (31% of the samples) were more evenly distributed. Each whole number increment above 5.0 per liter accounted for 2% or less of the total samples. The distribution of densities of adult Cladocerans and adult small Diaptomids showed similar trends. Therefore, for chi square analyses, density categories were formed depending on whether the sample mean of that species (or group) was greater than or less than 5 individuals per liter. For large Diaptomids however, density categories were greater than zero and equal to zero individuals per liter. These categorizations resulted in 2 x 2 tables. Resulting chi squares were then corrected for continuity since the distribution of chi-square in a 2 x 2 table is discreet (Snedecor and Cochran 1967).

In order to examine potential effects of vertebrate predation on rotifer groups, 27 east slope lakes were grouped according to types of vertebrate predators present (Table 2). These 27 lakes were all those (except for 2 west-slope lakes) for which adequate information existed on presence and density of vertebrate predators. Trout (cutthroat, Onchorhynchus clarkii, and rainbow, O. mykiss) were categorized as non- reproducing or reproducing in each lake. Lakes with reproducing fish generally had a more complex age and size structure, and much higher densities of fish than did lakes with non-reproducing fish (Liss et al. 1995). Other east-slope lakes had salamanders (Ambystoma macrodactylum) as their only vertebrate predators (Liss et al. 1995), and a few lakes had no vertebrate predators.

An overall mean was calculated for each lake for the following rotifer groups or species: illoricated rotifers (soft-bodied and generally larger taxa including Polyarthra, Synchaeta, Asplanchna, and Asplanchnopus species); loricated rotifers (hard-bodied taxa including all Brachionids except Keratella cochlearis, as well as Colurella, Lecane, Lepadella, Monostyla, and Trichotria species); Conochilus unicornis; and Keratella cochlearis. The latter taxon is a form distinct from Keratella cochlearis faluta, but it has no form name (R. Stemberger, personal communication). Kruskal-Wallis one-way analyses were run on each of these rotifer groups or species where the classification factor or rank was vertebrate predation category.

Many rotifer taxa were relatively widely distributed. Twenty of the 41 taxa (49%) occurred in more than 10% of the sampled lakes (Table 3), and 12 of these taxa (29%) occurred in more than 25% of sampled lakes. Kellicottia longispina was the most ubiquitous rotifer, occurring in over 90% of sampled lakes. Conochilus unicornis, the second most common species, was found in over 70% of lakes.

Hierarchical agglomerative cluster analysis grouped lakes according to their dominant taxa (Figure 1). Dominance by single taxa was generally high.

Figure 1. Dandrogram from a hierarchical agglomerative cluster analysis of lakes which had mean rotifer densities > 1.0 per liter (30 lakes).

Cluster One

Cluster one represented the most common type of rotifer community among lakes in the complex which had overall mean rotifer densities greater than 1.0 per liter (Table 4). The dominant taxon in these lakes was Conochilus unicornis, which had a mean proportional density of 83%. Other taxa occurred at lower proportions. Lakes in this cluster were generally lower elevation lakes and had the highest mean alkalinity and conductivity, pH, and temperature of all clusters (Table 5). These lakes had relatively high mean densities of adult Cladocerans, and large-bodied and small-bodied Diaptomids (Table 6).

+ Includes primarily Daphnia rosea and Holopedium gibberum, but also others (see Liss et al. 1994).
++ Includes primarily Daphnia rosea, but also D. longiremis, D. middendorffiana, and D. schodleri.
@ Includes primarily Diaptomus kenai, but also D. arcticus, and D. leptopus.
@@ Includes primarily Diaptomus tyrrelli, but also D. lintoni.

Cluster Two

Lakes falling within cluster two were characterized by high proportions of Kellicottia longispina (Table 4), though Collotheca mutabilis and Conochilus unicornis were also relatively abundant, composing on average 13% and 9% of the rotifer communities in these lakes. Lakes in cluster two were deep and had relatively low concentrations of total nitrogen (TKN) and total phosphorus (Total P; Table 5). Although Kellicottia longispina was dominant only in lakes in this cluster, it was present in all other clusters except cluster four (Table 4). Crustacean zooplankton communities in lakes in this cluster were characterized by low densities of adult Daphnids (and Cladocerans in general), relatively high densities of large Diaptomids, and moderately high densities of small Diaptomids (Table 6).

Cluster Three

Lakes in cluster three were dominated by Keratella cochlearis faluta and Polyarthra species (Table 4). Other taxa occurred at very low proportions. Relative to other clusters, lakes in this group tended to lie at higher elevations, and had cooler temperatures though they were also relatively small and shallow. They had intermediate levels of total nitrogen, alkalinity, and conductivity (Table 5). These lakes were characterized by low densities of adult Cladocerans and small and large Diaptomids (Table 6).

Cluster Four

Lakes in cluster four were characterized by very high dominance of Keratella cochlearis (Table 4), with low densities of other taxa, Lakes in this cluster had the highest mean total nitrogen and total phosphorous concentration of all clusters, though alkalinity and conductivity were intermediate (Table 5). These lakes were relatively small and shallow and had relatively high temperatures. Crustacean communities in these lakes were characterized by very high densities of both adult Cladocerans, primarily Daphnids, and adult small Diaptomids, primarily D. tyrrelli, but had no large Diaptomids (Table 6).

Cluster Five

The two lakes in this cluster were characterized by high dominance of Keratella hiemalis (Table 4), with low proportions of other taxa. These lakes had relatively low concentrations of total nitrogen and total phosphorous, low pH, alkalinity, conductivity, and relatively low temperature (Table 5). These lakes also had very low densities of adult Cladocerans, moderate densities of large Diaptomids, and low densities of small Diaptomids (Table 6).

Cluster Six

The one lake in cluster six was characterized by high dominance of Collotheca mutabilis although Ascomorpha ecaudis composed 12% of the total rotifer community (Table 4). This was a relatively large deep lake with low alkalinity and conductivity (Table 5). Diaptomids occurred in the lake at extremely low densities (Table 6). Adult Cladocerans reached moderate densities, but almost all were Holopedium gibberum.

These results were roughly consistent with those obtained with detrended correspondence analysis (DCA) run on species proportional abundance data. Lakes grouped on the first two DCA axes according to their dominant taxa (Figure 2). Statistically significant positive correlations (Pearson product-moment correlation coefficients) were found between the first ordination axis scores and total nitrogen (r = 0.657, p <0.01), total phosphorous (r = 0.529, p <0.01), and temperature (r = 0.402, p = 0.03), indicating a tendency of lakes with high axis one scores to have relatively high values of these variables. The second ordination axis correlated positively with elevation (r = .387, p = 0.04), and orthophosphate (Ortho-P; r = 0.397, p = 0.03), and correlated negatively with temperature (r = -0.404, p = 0.03).

Figure 2. Graph of axis scores for each lake from detrended correspondence analysis. Axis one scores correlated positively with total Kjeldahl nitrogen (TKN), total phosphorous (Total P), temperature, and proportional abundance of adult Daphnids using Pearson product-moment correlation coefficients. Axis one scores correlated negatively with proportional abundance of large Diaptomids and Holopedium gibberum. Axis two scores correlated positively with elevation and ortho-phosphorous (Ortho-P), but correlated negatively with temperature. Lakes are numbered according to cluster.

Proportional abundance of all Cladocerans was not significantly correlated with first axis scores. This may be partially explained by the fact that the two dominant Cladoceran taxa which are commonly found in these lakes, Daphnids (primarily Daphnia rosea) and Holopedium gibberum, tended to reach high densities in lakes with different physical and chemical characteristics (Liss et al. 1995). While proportional abundance of Daphnids was positively correlated with axis one (r = 0.373, p = 0.05), proportional abundance of Holopedium gibberum was negatively correlated (r = -0.477, p <0.01). Though proportional abundance of all Cladocerans was not significantly correlated with first axis scores, absolute density of all Cladocerans was significantly correlated (r = 0.711, p < 0.01), probably because Daphnids reached much higher absolute density (maximum overall mean = 17.3 per liter) than Holopedium gibberum (maximum overall mean = 2.2 per liter). Negative correlations were found between first axis scores and proportional abundance of large Diaptomids (r = -0.436, p = 0.02). Proportional abundance of small Diaptomids was not significantly correlated with axis one (p > 0.10), though absolute abundance was significantly correlated (r = 0.438, p = 0.016).

Physical and chemical variables and crustacean zooplankton densities that were significantly correlated with ordination axes showed the same general tendencies with groups of lakes in DCA as in cluster analysis. That is, lakes where Keratella cochlearis dominated (cluster 4) had high axis one scores, high total nitrogen and total phosphorous, tended to have high temperatures, high proportional densities of Daphnids, but no large Diaptomids. Lakes in which Kellicottia longispina dominated (cluster 2) had low first axis scores, low total nitrogen and phosphorous, low proportional densities of Daphnids (but relatively high proportional densities of Holopedium gibberum), and high proportional densities of large Diaptomids. On axis two, lakes in which Keratella cochlearis faluta and Polyarthra dominated (cluster 3) occurred at relatively high elevation and had low temperature. Lakes in which Conochilus unicornis dominated (cluster 1) occurred at relatively low elevation and had high temperatures.

Densities of some rotifers or rotifer groups were related to densities of crustacean zooplankton groups. The density of Kellicottia longispina was related to the density of all adult Cladocerans (chi-square, p = 0.025; Figure 3a). When Kellicottia density was low (less than 5 per liter), the number of samples where Cladoceran density was high (greater than 5 per liter) was greater than expected under a random distribution. With high density of Kellicottia, the number of samples with high Cladoceran density was less than expected. A negative relationship is thus suggested (Figure 3a). This corresponds with the relationship between all loricated rotifers (excluding Keratella cochlearis) and adult Cladocerans (chi-square, p = 0.01; Figure 3b).

Figure 3. Plot densities (number per liter) of adult Cladocerans against densities of: a) Kellicottia longispina; b) and all loricated rotifers combined. Data used are sample means.

The density category of Keratella cochlearis was also related to the density of Cladocerans (chi-square, p = 0.01). However, with high densities of Keratella cochlearis, the number of samples with high densities of Cladocerans was greater than expected. This suggests a positive relationship. In fact, Keratella cochlearis occurred at very high densities (447.0 per liter in Dagger Lake) in lakes where the density of adult Cladocerans was quite high (21.9 per liter). This was the only loricated rotifer which appeared to have a positive relationship to Cladoceran density. Density of adult Diaptomids (either all Diaptomids combined, or large and small Diaptomids tested separately) had no statistically significant effect on loricated rotifers.

Presence of Polyarthra spp. was related to the absence of large Diaptomids (D. kenai and D. arcticus; chi-square, p = 0.05). The two taxa rarely co-occurred. However, density of Polyarthra was not related to density of either small Diaptomids (p > 0.10) or adult Cladocerans (p > 0.10) based on the chi-square statistic.

There was no evidence that density of Conochilus unicornis was related to density of either Diaptomids (p > 0.10; all taxa combined, or large and small Diaptomids tested separately) or Cladocerans (p > .10) based on the chi-square statistic. Conochilus reached very high densities (>100 per liter) with high densities of either Diaptomids or Cladocerans.

Kruskal-Wallis analyses were used to examine potential effects of vertebrate predation on rotifer groups or species. There were statistically significant differences in rotifer densities between vertebrate predation categories only for Conochilus unicornis( p = 0.03; Table 7). When lakes with no vertebrate predators were omitted from this analysis, there were no significant differences between the remaining three vertebrate predator categories (p >10). Though there are large differences between the means of vertebrate predation categories for several of the rotifer groups, variability within groups is high (Table 7).

Many of the rotifer taxa found in North Cascade lakes are relatively widely distributed (Table 3). This finding is consistent with the view that, world-wide, rotifers are generally widely distributed (Dumont 1983), and that they generally have powerful passive dispersal strategies (Pennak 1978). It also indicates that the rotifer species pool for each lake is large.

In spite of these factors, dominance by single rotifer taxa is generally high in lakes where overall mean rotifer densities are > 1.0 per liter (Table 4). This is consistent with findings by Green (1993) that lakes generally have high single-species dominance.

Dominance of rotifers related to physico-chemical conditions and crustacean zooplankton.

Dominance by a particular taxon may be a function of the interaction between physico-chemical conditions and crustacean zooplankton densities. The dominant taxon in 13 lakes (cluster 1) was Conochilus unicornis (Table 4). Dominance by this taxon may be related, in part, to its food preference for bacteria and detritus (Pourriot 1977), both of which may be more abundant in relatively more productive systems (Geller and Muller 1981; Wetzel 1975). Lakes in cluster one had the highest mean alkalinity and conductivity (Table 5), the latter being one indicator of greater productivity (Berzins and Bertilsson 1989). Lakes in cluster one also had moderately high mean densities of both adult Cladocerans and Diaptomids (Table 6). Perhaps Conochilus unicornis can occur at high densities in lakes where densities of Diaptomids are high because the mucus sheath and the coloniality of Conochilus unicornis protect it from predation by crustacean zooplankton (Stemberger and Gilbert 1 987b).

The dominant taxon in 16 of the 30 lakes (clusters 2, 3, 4, and 5) was a small-sized Brachionid rotifer (Kellicottia longispina or a Keratella species; Table 4). In general, Brachionidae are relatively unselective microfiltrators which feed on bacteria, detritus, smaller flagellates and ciliates in the 0.2 to 20 um size range (Pourriot 1977; Arndt 1993; Moore 1978). The microbial systems in ultraoligotrophic and oligotrophic lakes tend to be dominated by plankton in this size range (pico- and nanno-plankton; Pejler 1983; Stockner and Shortreed 1989; Stockner 1987), perhaps because of the ability of these small plankton to efficiently uptake nutrients and grow rapidly at low nutrient concentrations (Stockner 1987).

Although there is variation between clusters of lakes in nutrient levels, alkalinity, and conductivity (Table 5), on a worldwide scale, lakes in the North Cascades National Park Service Complex generally have low nutrients, alkalinity, and conductivity. In fact, the most productive lakes examined would only be classified as oligo-mesotrophic (Liss et al. 1991). Small, microfiltrating rotifers may dominate rotifer communities in many of these lakes because they can feed efficiently on the small microorganisms (Pourriot 1977; Arndt 1993; Moore 1978) which generally occur in ultraoligo- and oligotrophic systems (Pejler 1983; Stockner and Shortreed 1989; Stockner 1987). These microfiltrating rotifer taxa are also relatively small-bodied (100-200 um excluding posterior spines; Stemberger 1979), and tend to have low threshold food levels (Stemberger and Gilbert 1987a), a factor that would work to their advantage in relatively unproductive lakes. In fact, the rotifer communities in ultraoligo- and oligotrophic mountain and subarctic lakes in western North America are generally dominated by taxa such as Kellicottia longispina and Keratella cochlearis, at least for part of the year (Neill 1984; Anderson 1980; Moore 1978).

Many of the lake clusters dominated by loricated rotifers had relatively high mean densities of Diaptomids (clusters 1,2,4; Table 6). Because they are generally rejected by Diaptomids as food (Williamson 1987), loricated rotifers may occur at high densities in lakes with high Diaptomid densities. The negative relationship suggested between most loricated rotifer taxa and Cladocerans (Figure 3a,b) is consistent with the view that Daphnids specifically, and perhaps medium- to large-bodied Cladocerans in general, are competitively superior to microfiltrating rotifers (Gilbert 1985). In fact, in the clusters dominated by Kellicottia longispina, Keratella cochlearis faluta, or Keratella hiemalis (clusters 2,3,5), densities of Daphnids and total Cladocerans were relatively low (Table 6).

The loricate Keratella cochlearis occurred at high densities in east-slope lakes where densities of both Cladocerans and small Diaptomids were high (cluster 4; Table 6). There is no clear explanation for this co-occurrence with Cladocerans, though one possibility is that Keratella cochlearis may utilize different food sources, or a different portion of the food size spectrum, than either Daphnids or the other dominant loricated rotifers. The potential for greater productivity of these lakes (reflected by higher total nitrogen, total phosphorous, and conductivity) may allow much higher densities of bacterial food resources (Geller and Muller 1981). The positive relationship between Keratella cochlearis and small Diaptomids is likely related to the fact that small Diaptomids, specifically Diaptomus tyrrelli, reach high densities only in lakes with higher levels of total Kjeldahl nitrogen and phosphorous (Liss et al. 1995).

Rotifers which are grasping feeders or predators (Polyarthra and Synchaeta species) were a significant component of only a few lakes (cluster 3). This may be related, in part, to the generally low productivity of the study lakes. Large microorganisms (the food sources of grasping and predacious rotifer taxa) are not likely to be abundant in ultraoligo- and oligotrophic lakes (Pejler 1983; Stockner 1987). These predacious rotifer taxa are also soft-bodied, and another reason for their general lack of dominance may be the apparent strongly negative effects of predacious large Diaptomids on Polyarthra and other illoricates (Anderson 1980; Williamson 1987) coupled with the general ubiquity of large Diaptomids in these lakes (Liss et al. 1995). Only one cluster of lakes had Polyarthra as a subdominant (cluster 3; Table 4), and these lakes had low mean densities of large Diaptomids (Table 6). Only clusters 4 and 6 had lower densities of large Diaptomids. The one lake in cluster six was relatively unproductive (deep, with low conductivity, and intermediate levels of total nitrogen and total phosphorous; Table 5) perhaps preventing the presence of grasping feeders. Lakes in cluster four all had very high densities of Cladocerans (>20 per liter) as well as high densities of small Diaptomids (>5 per liter). In spite of the lack of statistical significance of the relationship between Polyarthra and either Cladocerans or small Diaptomids, the latter two may have some negative effect on Polyarthra through competition when either crustacean occurs at very high density.

Dominance of rotifers related to categories of vertebrate predation

Lack of any statistically significance relationship between rotifer taxa and vertebrate predators may be related to the small sample sizes and very high standard error of the means. This high standard error could be caused, in part, by sampling lakes too rarely to be assured of determining high points of population density of dominant taxa. However, high standard errors may also result because a classification based only on category of vertebrate predation does not adequately capture the complexity of factors which structure presence and dominance of specific rotifer taxa. Dominance by specific taxa is not only related to density of crustacean zooplankton groups which may be differentially affected by vertebrate predators, but is also likely related to a complex interaction of physical and chemical parameters (nutrient levels, alkalinity, conductivity, temperature regime, lake size and depth) and of the microorganismic community.