J. Great Lakes Res. 21(4):670-679 Intemat. Assoc. Great Lakes Res., 1995

NOTE

Ecological Interactions Between Bythotrephes cederstroemi and Leptodora kindtii and the Implications for Replacement in Lake Michigan

Donn K. Branstrator* Department of Biology Natural Science Building University of Michigan Ann Arbor, Michigan 48109

ABSTRACT. The zooplankton predator, Leptodora kindtii, declined in abundance at an offshore reference station in Lake Michigan in the mid-1980s following the invasion of another zooplankton predator, Bythotrephes cederstroemi. Both predators feed largely on daphnid prey and it was observed that densities of three species declined in abundance at the reference station following the Bythotrephes invasion. Circumstantial evidence would suggest that the native predator, Leptodora, was competitively suppressed by Bythotrephes. However, results of laboratory experiments presented here show that Bythotrephes will readily attack and eat Leptodora when the predators are maintained under concentrated densities, even if alternative prey are available for the Bythotrephes. There was no evidence in these experiments that Leptodora attacked or ate Bythotrephes. These results imply that by Bythotrephes on Leptodora may alternatively account for the collapse of Leptodora in offshore Lake Michigan. In this note I discuss evidence in support of competition and predation as alternative hypotheses to explain the pattern of species replacement, Bythotrephes for Leptodora, observed in Lake Michigan. The existing data are not definitive and tempt further inquiry. INDEX WORDS: Bythotrephes, Leptodora, Lake Michigan, invading species.

INTRODUCTION Bythotrephes cederstroemi invaded the Laurentian Great Lakes during the 1980s. Its time of arrival and dispersal pattern were particularly well documented for Lake Michigan (Lehman 1987, Evans 1988) where shortly following its invasion several contemporaneous changes occurred in the composition of the lake's native community suggesting that the predator was altering the community (Lehman and Caceres 1993). Among these changes was a strikingly quick decline in the abundance of a native zooplankton predator, Leptodora kindtii. During the two years preceding the Bythotrephes invasion, Leptodora had achieved summer abundances >3,000 individuals m-2 in offshore regions of Lake Michigan (Table 1). But in regions where Bythotrephes subsequently established populations, Leptodora declined in abundance. Lehman (1991) used 128 independent zooplankton samples collected from inshore and offshore regions of Lake Michigan to demonstrate that an inverse relationship in the abundance of Bythotrephes and Leptodora occurred lake-wide from 1986 to 1990. Garton et al. (1990) reported that a similar pattern of species replacement, Bythotrephes for Leptodora, had developed in the western basin of Lake Erie following the invasion of Bythotrephes there.

*Current address: The University of Chicago Department of Ecology and Evolution 1101 East 57th Street Chicago, Illinois 60637

These spatial and temporal patterns of species replacement are suggestive of strong, biotic interaction between Bythotrephes and Leptodora, however, the relative importance of alternative hypotheses remains unresolved. It seems worthwhile to investigate both compe- tition and predation as possible mechanisms that explain the pattern. The invader, Bytho- trephes, occupies a dietary niche similar to that of Leptodora in that both predators prefer small cladoceran prey (Mordukhai-Boltovskaia 1958, Edmondson and Litt 1987, Branstrator and Lehman 1991). The predators will therefore likely compete for food, especially for small daphnid instars which appear to be the selected prey of both species. Bythotrephes is also substantially larger than Leptodora, weighing up to 10 times more as adult instars (Lehman and Caceres 1993) and may therefore also prey upon Leptodora. In this report I present experimental evidence that Bythotrephes will capture and eat Leptodora in the laboratory. I also apply an encounter rate model developed by Gerritsen and Strickler (1977) to estimate rates of encounter between Leptodora and Bythotrephes in Lake Michigan. I then discuss the available evidence in support of both the predation and competition hypotheses. Finally, I address a third hypothesis that planktivory increased in Lake Michigan and caused the decline in Leptodora abundance. This third hypothesis is particularly relevant since Sprules et al. (1990) have argued that fish planktivory was the cause of other contemporaneous changes in the zooplankton community, particularly the decline in daphnid abundances, that followed the Bythotrephes invasion.

METHODS

Predation Experiments All zooplankton used in the experiments were collected with a conical net of 130-µm mesh towed obliquely through the water column. Leptodora were collected from Third Sister Lake, Ann Arbor, Michigan, and Bythotrephes were collected from offshore Lake Michigan (43°N, 86°40'W). The logistics of sampling two lakes did not permit both predator species to be collected simultaneously. Consequently, the Leptodora and Bythotrephes were starved for different amounts of time before experimentation. A reciprocal starvation schedule was employed in the 1-L jar experiments in order to control for the effects of starvation. Two experimental designs, using different sized containers, were used to test whether Bythotrephes will capture and eat Leptodora. Experiments were conducted in either polystyrene tissue culture vessels (Costar) with 10 mL Lake Michigan water, or glass jars with 1 L lake water, 50% Third Sister Lake and 50% Lake Michigan by volume. All water was pre-screened (50-µm mesh). Experimental vessels were kept in a walk-in environmental chamber at the University of Michigan and maintained at 16°C and light <2 µЕ m-2 s-1. The experimental temperature was chosen based on measured thermal tolerances of the two species (Garton et al. 1990). The low light levels were maintained in order to simulate more natural lake conditions. One experiment was conducted in the 10-mL vessels which had one control and three treatment series, with 60 replicates per series. Treatment replicates each received 1 adult Leptodora and 1 Bythotrephes. The Bythotrephes were either instar-1 or -2 (both pre-adult), or instar-3 (adult). The three instars cover the full range of body size achieved by Bythotrephes in Lake Michigan (Yurista 1992). Control replicates each received 1 Leptodora and no Bythotrephes. The were observed several times daily under an Olympus dissecting microscope with darkfield illumination and their conditions were noted. The Leptodora and Bythotrephes were pre-starved for 3 h and 16 h, respectively. Two experiments, designated A and B, were conducted in the 1-L jars. Each experiment had 1 control and 1 treatment series, with 20 replicates per series. The treatment jars each received 1 Leptodora, 1 Bythotrephes of instar-3, and 30 to 40 Daphnia (non-helmeted D. galeata mendotae from Third Sister Lake and juvenile D. pulicaria from Lake Michigan). The control jars each received 1 Leptodora, no Bythotrephes, and the same complement of Daphnia as that added to each treatment jar. The condition of the predators was recorded several times over 3 days by briefly holding the jar to light. Leptodora and Bythotrephes were pre-starved in experiment A for 10 h and 35 h and in experiment В for 24 h and 0 h, respectively. It was possible to distinguish between Leptodora that had been killed or eaten by Bythotrephes and those that had died with no apparent injury. Leptodora were recorded as killed by Bythotrephes if 1) Leptodora was not found in the container, 2) Leptodora was partially eaten, or 3) Bythotrephes was seen eating Leptodora.

Body Size Estimates I hypothesized that predator body size would affect the experimental results. Although both species reach body lengths of 10 to 15 mm, large Bythotrephes females can weigh fully 10 times more than Leptodora of comparable length. Therefore, both lengths and dry weights of the animals were estimated. For the experiment conducted in 10-mL vessels, the mean body size of Leptodora was estimated with specimens taken from the same collection used to set up the experiment. These Leptodora were randomly selected when other specimens were being moved to the vessels for the start of the experiment. They were measured live from the center of the eye to the base of the bifurcation in the tailspine. Their dry weights in ug were estimated as

where L is Leptodora length (mm) (Lehman and Caceres 1993). For experiments conducted in 1-L jars, the mean Leptodora length was estimated from Leptodora still alive in the control jars at the end of each experiment. The Leptodora were preserved individually in 4% sugar-Formalin and later measured for total length as described above. A correction factor was applied to the length measurements to account for body shrinkage caused by the preservative:

where LL is live length and PL is preserved length for Leptodora preserved in 4% sugar- Formalin >55 d. Length measurements, corrected for shrinkage, were converted to µg dry weight (Eq. 1). The average weight of Bythotrephes instars can vary dramatically during the summer in Lake Michigan (Burkhardt 1994). For example, the average weight of instar-3 females ranged from approximately 100 µg to 630 µg during July to September, 1990. This variation in body weight has been shown to correlate significantly with epilimnetic water temperature (Burk- hardt 1994). The relationship between temperature and weight complicates the application of general regression equations for weight at length that do not account for water temperature (Berg and Garton 1988, Garton et al. 1990). The estimation of dry weights for the Bythotrephes used in the 10-mL vessel experiment was simplified here because the animals were collected at the same location in Lake Michigan and within 1 d of those specimens that Burkhardt (1994) used for measurement. For the 1-L jar experiments, however, Bytho- trephes weights were estimated from temperature data and a regression relating temperature to Bythotrephes dry weight (Burkhardt 1994). The estimated mean length ± SD and dry weight ± SD of animals used in the 10-mL vessel experiment are: Leptodora (9.0 ±1.2 mm, 98 ± 30 µg), and weights only for Bythotrephes instar-1 (125 µg), instar-2 (400 µg), and instar-3 (630 µg) (Burkhardt 1994 Fig. 2). Estimated sizes for the 1-L jar experiments are as follows: experiment A, Leptodora (9.1 ± 1.0 mm, 99 ± 24 µg), Bythotrephes instar-3 weight only (600 µg); experiment B, Leptodora (8.1 ± 0.9 mm, 74 ± 20 µg), Bythotrephes instar-3 weight only (400 µg).

Encounter Rate Model Because I was interested in describing the outcome of a predatory interaction between Leptodora and Bythotrephes, and not measuring an encounter rate or predation rate, I intentionally used high densities of Leptodora and Bythotrephes in the experiments. Natural densities of Leptodora in Lake Michigan and other lakes are generally less than 0.1 L-1, fully 10 times lower than the density in my experiments (Balcer et al. 1984, Branstrator and Lehman 1991). In comparison, natural abundances of Bythotrephes may be even less than Leptodora. In Lake Michigan, Bythotrephes reaches an average summer density of about 0.01 L-1, 100 times lower than the density in my experiments (Table 1, Lehman 1991). It therefore remains questionable whether Bythotrephes encounters Leptodora frequently enough at natural densities to be an important source of mortality. To investigate this I estimated probable encounter rates (E) between the species using the model of Gerritsen and Strickler (1977):

where Rb is the encounter radius of Bythotrephes (a distance within which it can detect other animals), Nl is Leptodora density, and sb and sl are the mean swimming speeds of Bythotrephes and Leptodora, respectively. The total rate of encounter between all individuals of each species population (T) was computed as

where Nb is the density of Bythotrephes. The encounter radius is a function of many variables that measures an animal's ability to sense mechanical disturbance in its surroundings (Gerritsen and Strickler 1977). For Leptodora, precise measurements of the encounter radius have been made with video photography and show that prey must enter a distance within 2.5 mm from the front edge of the feeding appendages to ensure a successful prey capture (Browman et al. 1989). An encounter radius has not been reported for Bythotrephes, therefore I used a range of encounter radii (Rb = 1 to 5 mm) based on the length dimensions of Bythotrephes' core body (Burkhardt 1994). A constant mean swimming -1 speed of 15 mm s was applied for both Bythotrephes (sb) and Leptodora (sl) (Browman et -1 al. 1989, Zozulya 1978), and a constant density of 0.01 Bythotrephes L (Nb) was used. I let -1 Leptodora density (Nh) range between 0.01 to 0.3 individuals L (Fig. 1).

RESULTS AND DISCUSSION Recent annual and seasonal trends in the composition of the zooplankton community in Lake Michigan indicate that Bythotrephes replaced Leptodora as the dominant cladoceran predator in offshore areas during the mid-1980s (Table 1, Lehman 1991). This occurred as part of a larger transition in the offshore zooplankton community that included changes in the herbivorous cladoceran community and rotifers as well (Lehman 1991, Branstrator and Lehman 1991, Sandgren and Lehman 1990). Because several changes occurred contemporaneously in the zooplankton community, identifying the most important cause for the collapse of Leptodora is not straight forward. At this time, there are still too few data to distinguish between the importance of predation by Bythotrephes on Leptodora versus competition between the predators for food as the key cause of the Leptodora decline. There is reasonable evidence to conclude that increased fish predation on Leptodora, a third hypothesis, was probably not an important cause of its decline. Here I lay out the evidence for these three hypotheses in hope that this report will stimulate further study of this problem.

Predation by Bythotrephes The results of laboratory experiments reported here indicate that Bythotrephes clearly had a negative effect on the survival of Leptodora (Tables 2, 3). Tables 2 and 3 report only the number of trials where Bythotrephes killed Leptodora because there was never any evidence that suggested Leptodora killed Bythotrephes in these experiments. The results were explored statistically with a Chi-Square Homogeneity test (alpha = 0.05) that compared numbers of live to dead Leptodora in the treatment versus the control series. Results of the 10-mL vessel experiment were statistically compared at 42 h. At this time there were significantly fewer Leptodora surviving in each treatment series compared to the control series. At 42 h there was also a significant effect among the treatment series which indicated that Leptodora that were paired with larger Bythotrephes survived a shorter time, on average, in comparison to the Leptodora that were paired with smaller Bythotrephes. Results of the 1-L jar experiments were statistically compared at 49 h (experiment A) and at 40 h (experiment B). At these times, only the results of experiment В showed a significant departure from a null model indicating that Leptodora survival was lowered in the presence of Bythotrephes in experiment В only. Leptodora was pre-starved longer in experiment В (24 h) in comparison to 10 h in experiment A which may have lowered the survival success of Leptodora in experiment B. In general, the experiments demonstrate that Bythotrephes attack and eat Leptodora. The results are fully consistent with body size and morphological differences between Leptodora and Bythotrephes that would suggest, when considered together, that Bythotrephes should win a predatory encounter between the species. Bythotrephes can weigh up to 10 times more than a Leptodora of similar body length. In my experiments, the average dry weight of Bythotrephes ranged from 125 µg to 630 µg and the average dry weight of Leptodora ranged from 74 µg to 99 µg. These ranges encompass typical average weights of the predators in natural populations. In Lake Michigan, Bythotrephes range in length from 7 mm (instar-1) to 11 mm (in-star-3) and in dry weight from 100 µg to 500 µg individual-1 (Burkhardt 1994). Leptodora range in length from 2 mm at birth to as long as 13 mm or more as adult females. These lengths correspond to dry weights of 3 µg to 230 µg individual-1 which are substantially lighter than the weights of Bythotrephes of comparable length. The importance of body size as a determinant of the one-sided outcome of these experiments is partly substantiated by results of the 10-mL vessel experiment that demonstrate a significant negative relationship between Leptodora survivorship and the size of the Bythotrephes instar (Table 2). Whether or not body size is an important factor in the field, where the predators are not forced to interact, remains to be tested. In and Leptodora coexist with a much smaller species of the spined cladoceran, Bythotrephes longimanus (Patalas and Patalas 1966, De Bernardi 1974, Kogan 1975, Pejler 1975, Craig 1978, Guma'a 1978, Naesje et al. 1978, Ketelaars and Lambert 1993). The core body and caudal spine length of B. longimanus are comparatively much smaller than B. cederstroemi. Data from those stud- ies offer circumstantial evidence that the strength of predatory interactions between Bythotrephes and Leptodora depends on the relative size of the two species. Only the larger species of Bythotrephes, B. cederstroemi, has invaded North American lakes at this time.

Leptodora and Bythotrephes are also morphologically quite different and this could also account for the one-sided outcome of the experiments. The exoskeleton of Leptodora appears to be comparatively weaker than that of Bythotrephes, evidenced by the fact that Leptodora become flaccid when removed from an aqueous support medium. Leptodora are therefore not well adapted to escape tactile predators, such as Bythotrephes, that grasp and shred prey. On the other hand, the core body of Bythotrephes maintains form, as daphnids do, when removed from water. Its large body size probably also offers substantial protection against tactile predators. In light of this, Bythotrephes can still be highly cannibalistic when placed under high densities (personal observation).

Finally, we know that the size of prey eaten by Leptodora is largely determined by the dimensions of its trap basket (Herzig and Auer 1990) which consists of five pairs of opposing legs used to capture prey. The maximum diameter of the trap basket is approximately 1/10 the total length of Leptodora (Herzig and Auer 1990). Therefore, even the largest Leptodora, which can reach 18 mm length, has a trap basket <2.0 mm diameter which is too small to hold the core body of even most instar-1 Bythotrephes (Burkhardt 1994). This trophic con- straint may account for the fact that Leptodora did not eat Bythotrephes in any of the experiments. Studies have shown that Bythotrephes prefer small cladoceran prey in the size range of juvenile Daph-nia (Mordukhai-Boltovskaia 1958, De Bernardi 1974) but it is not known what factors set the upper size threshold of ingestible prey. Experiments here clearly demonstrate that Bythotrephes will eat Leptodora that are upwards of 10 mm length. The laboratory and field data suggest that Bythotrephes could be an important predator on Leptodora. But whether or not the experimental results are an artifact of high predator densities remains to be tested. The 1-L jars substantially raised the densities of the two predators and likely affected their natural swimming behaviors due to edge effects of the confining walls, a result that has been reported for predatory (O'Brien 1988). Given that potential biases in encounter rates may have occurred in the experiments, their results were not used to estimate encounter rates in the jars. Because both Bythotrephes and Leptodora achieve relatively low densities in lakes, and therefore might be expected to experience few random encounters on a daily basis, the importance of predation under natural conditions is potentially unimportant. The model of Gerritsen and Strickler (1977) showed that Leptodora and Bythotrephes should experience in the range of 5 to 400 encounters m-3 d-1 (Fig. 1). This range reflects the bounds on the two parameters that were permitted to vary, encounter radius (Rb) and Leptodora density (Nl), as well as the constants assigned to the other parameters. Other factors to consider that would affect physical interaction rates are the spatial overlap between populations of Leptodora and Bythotrephes and the encounter success rate which would measure the frequency of encounters between the two species that resulted in a successful predatory attack. Values of both additional parameters would probably lower the potential predatory effect that Bythotrephes has on Leptodora. For example, in Lake Michigan it was estimated that Bythotrephes and Leptodora populations overlapped on average by 57% during day and 70% during night from 1987 to 1990 (Lehman and Caceres 1993). My analysis of encounter rates suggests that Bythotrephes could be an important predator on Leptodora when the predators reach high densities, as was the case in the experiments, but that this interaction is likely to be minor under average conditions that these animals experience in lakes.

Food Resource Competition Between Bythotrephes and Leptodora Bythotrephes and Leptodora occupy similar dietary niches in that both are predators on small cladocerans and rotifers (Mordukhai-Boltovskaia 1958, Edmondson and Litt 1987, Branstrator and Lehman 1991). They both reproduce parthenogenically and can double their population sizes in days to weeks when temperatures and food resources are favorable (Table 1). The potential for rapid growth permits them to respond quickly to changing re- source conditions; so rapidly that they reduce their prey to densities that limit their own growth. For example, from 1987 to 1990 the estimated consumptive demand by Bythotrephes in Lake Michigan exceeded the replacement production by their cladoceran prey for approximately 1 to 2 months every summer (Lehman and Caceres 1993). It should thus be anticipated that these predators will compete for limiting resources in the Great Lakes. Contemporaneous with the invasion of Bythotrephes in Lake Michigan were also major changes in the daphnid community. At the reference station, the daphnids were reduced to essentially 1 species, Daphnia galeata mendotae, from three species (D. galeata mendotae, D. pulicaria, and D. retrocurva) that dominated during the 2 years of 1985 and 1986 preceding the invasion. During the summer of 1987 there was also a marked decrease in daphnid biomass over previous years' summer densities (Lehman and Caceres 1993). These changes in the daphnid community have been ascribed to the predatory impact of Bythotrephes (Lehman 1991). The fact that Leptodora declined during the same period provides circumstantial evidence that reduced daphnid abundance, and especially the absence of the smallest species, D. retrocurva, led to poor success of Leptodora at that time in the offshore waters of the lake. Since the collapse of Leptodora in 1987, however, many of Leptodora's major prey species including Daphnia galeata mendotae, longirostris, Conochilus unicornis, and copepods have reached densities as high as they achieved in 1985 and 1986 (Sandgren and Lehman 1990, Branstrator and Lehman 1991, Lehman and Caceres 1993). Nonetheless, Leptodora has remained sparse, often undetectable with 1-m-diameter nets. The only enduring changes in the herbivorous cladoceran community have been reduced densities of the large-bodied daphnid, D. pulicaria, and the small-bodied species, D. retrocurva. The paucity of the smaller species, D. retrocurva, may be a key reason why Leptodora has not since re-established populations in the offshore region of the lake. The collapse of Leptodora has not been observed at an inshore reference station (20-m depth) in Lake Michigan (Lehman and Caceres 1993). Nonetheless, Bythotrephes has remained sparse at this inshore station and the small daphnid species, D. retrocurva, has been abundant. Because the separate effects of predation by Bythotrephes and competition cannot be separated by studying the inshore zooplankton community, it offers little additional insight to the question at hand. The success of Leptodora inshore implies a fundamental difference between the inshore and offshore food webs of Lake Michigan. Strong vertebrate planktivory has been a consistent feature of the inshore region of Lake Michigan (Evans and Jude 1986) which most likely accounts for the poor success of large zooplankton, such as Bythotrephes, and increased success of inconspicuous species such as Leptodora, and small-bodied species such as D. retrocurva, in this region of the lake.

Predation by Fish The changes in the daphnid community in offshore Lake Michigan that occurred coincident with the Bythotrephes invasion have been interpreted by Sprules et al. (1990) as the result of increased fish predation, particularly by (Alosa pseudo-harengus). By analogy, the loss of Leptodora from the offshore zooplankton community during this same period could also be interpreted as a response to increased fish predation. Nonetheless, for the following two reasons I suggest that it is unlikely fish predation was a major factor in the collapse of Leptodora. First, Lehman (1991) has shown that the size structure of the daphnid community in offshore Lake Michigan changed in a way indicative of invertebrate predation, not vertebrate (fish) predation. The smallest of the three species, Daphnia retrocurva, was eliminated first from the community, which is consistent with a hypothesis of invertebrate planktivory. In contrast, a hypothesis of vertebrate planktivory would predict that the largest species should be eliminated first which did not happen. In fact, some of the largest individuals of Daphnia pulicaria, >3 mm length, still persisted after D. retrocurva was eliminated. Second, there were no associated changes in other zooplankton taxa suggestive of increased fish predation. In particular, the densities of large-bodied copepods that should be most susceptible to fish planktivory, including Limnocalanus macrurus and Epischura la- custris, remained relatively stable during the period 1985 through 1988 (Branstrator and Lehman 1991). Lake Michigan was the site of a classic study by Wells (1970) on the effects of fish planktivory on zooplankton size structure. Wells (1970) found that increased alewife densities led to the collapse of three cladoceran species (Leptodora kindtii, Daphnia galeata mendotae, and Daphnia retrocurva), three calanoid species (Limnocalanus macrurus, Epischura lacustris, and si-cilis), and the cyclopoid copepod Mesocyclops edax between 1954 and 1966. The effects of fish planktivory during that period were clearly discriminatory based on size. However, the current situation is quite different in that Leptodora has been eliminated yet other large, conspicuous taxa continue to persist. This current pattern is different from historical changes that have been ascribed to the impacts of fish planktivory.

Species Replacement Bythotrephes has replaced Leptodora as the dominant cladoceran predator at the reference station in Lake Michigan (Table 1). The data reviewed in this report suggest that predation by Bythotrephes on Leptodora and competition for food between the predators both remain plausible mechanisms that could account for this pattern. Data suggest that fish planktivory was not an important cause for the loss of Leptodora. Neither of the two remaining mecha- nisms, however, has been tested with experiments that employ natural densities of the predators. The effect of Bythotrephes predation could be a laboratory artifact seen only at unnaturally high densities. Moreover, estimates of natural encounter rates between the two predators are so low as to cast doubt on the importance of this mechanism in the field. Nonetheless, Leptodora has remained absent from the offshore zooplankton community despite the resurgence of prey species including Daphnia galeata mendotae which has achieved densities as high and higher than historical densities predating the Bythotrephes invasion (Lehman and Caceres 1993). This observation suggests that prey densities may not be the only factor preventing the re-establishment of Leptodora in the offshore community. Eventually other factors, such as the supply of resting eggs in sediments, may affect the success of vernal Leptodora populations. However, Leptodora continue to be successful in the nearshore region which could provide a source population to offshore waters. A report by Garton et al. (1990) showed that densities of Leptodora also collapsed in Lake Erie in 1987 and 1988 following the population maxima of Bythotrephes each year, yet the factors responsible there are still unresolved. Garton et al. (1990) were not able to address resource competition in their study because data on cladoceran prey abundances were not collected. Instead, they ascribed the collapse of Leptodora in Lake Erie to its sensitivity to declining water temperatures. They showed experimentally that Bythotrephes and Leptodora have different thermal tolerances, with Leptodora being less tolerant than Bythotrephes to low temperatures. Data for Lake Michigan indicate that water temperature does not appear to play a role in the poor success of Leptodora at the reference station. As shown in Table 1, Leptodora have remained virtually absent since 1987, yet the water temperature pattern from May to September has remained relatively constant on an interannual basis since at least 1985 (Lehman and Caceres 1993). The population dynamics of Leptodora in Lake Michigan are more consistent with changes in biotic interactions, either competition or predation, that may have occurred after the appearance of Bythotrephes in 1986. The collapse of Leptodora in both Lake Michigan and Lake Erie, coincident with the invasion of Bythotrephes, appears to be more than spurious correlations. Lehman (1991) argued that Bythotrephes cederstroemi has had profound negative effects on daphnid populations in Lake Michigan through predation on the small instar stages. His conclusions are consistent with De Bernardi (1974) who found that Bythotrephes longimanus is an important predator on small daphnid instars in Lago Maggiore. Together, their data suggest that Bythotrephes could feasibly be an important competitor with Leptodora for daphnid prey. Sprules et al. (1990) have taken issue with Lehman (1988) and predict that Bythotrephes cederstroemi is unlikely to have a major effect as a predator on daphnid populations in the Great Lakes. Understanding the one or more mechanisms that account for the replacement of Bythotrephes for Leptodora will help resolve the issue concerning the predatory effects of Bythotrephes. If observed reductions in daphnid abundances in Lake Michigan can account for the loss of Leptodora, the result would imply that predation effects imposed by Bythotrephes on daphnids can be large enough to competitively suppress other species, namely Leptodora. If it is demonstrated that predation by Bythotrephes on Leptodora can be significant at natural densities, the result would be important in itself; the nature and importance of direct predatory interactions between carnivorous invertebrates, other than copepods, appear to have been overlooked until recently (McNaught 1993). It is my hope that this report motivates further research on the Bythotrephes-Leptodora interaction. The opportunities for study should increase as Bythotrephes begin to colonize small inland lakes of (Yan et al. 1992). These invasion events represent more opportunities to document the effect of this predator on plankton communities and particularly on Leptodora. Smaller lakes that Bythotrephes invade should also be more amenable than the Laurentian Great Lakes to enclosure-type experiments in which natural densities of the predators can be more easily simulated in situ. An experiment at this mesocosm level should be instrumental in distinguishing the relative importance of competition versus predation in the apparent replacement of Bythotrephes for Leptodora in Lake Michigan and other North American lakes.

ACKNOWLEDGMENTS I thank John T. Lehman for providing me with plankton samples and the captain and crew of the RV Laurentian for assistance with field sampling. I also thank J.T. Lehman and two anonymous reviewers for comments on the manuscript. This research was supported by NSF grants ОСЕ 84-15970, ОСЕ 86-13880, and ОСЕ 87-16187 to J.T. Lehman and by an Okkelberg Fellowship from the Department of Biology and grant support from the Rackham Graduate School, University of Michigan, to D. K. Branstrator.