Zooplankton in five lake regions once inundated by Lake Agassiz were compared with zooplankton of 10 other lake regions and in 5 Great Lakes in south—central Canada. The highest resemblance to a hypothetical composite of Lake Agassiz plankton, from 94 to 100 percent, was found in lakes of the Northern Manitoba, Northern Saskatchewan and Upper Mackenzie regions. Eighty nine to 94 percent of zooplankton from the Laurentian Great Lakes came from the Agassiz Lake fauna. The only other species found in the Great Lakes in 1969 were Eurytemora affinis, and Eubosmina coregoni, newcomers to the North American fauna from Europe.
The number of species shared with the hypothetical fauna of Lake Agassiz declines with the distance from this glacial lake. From 91 to 93 percent of Lake Agassiz species were found in Manitoba Western, Southern Manitoba and Alberta, 85 to 86 percent in New Brunswick and Newfoundland, 80 to 84 percent in Nova Scotia and Southern British Columbia and about 74 percent in the lakes of the Mackenzie Delta region.
Lake Agassiz appeared to be a very efficient south-north dispersion route. Of the 35 species originating from the Mississippi refugium, 24 species penetrated as far as the Upper Mackenzie region and 16 species reached the Mackenzie Delta. The north to south dispersion route was not as effective. Of the 15 species originating from the Beringia refugium, seven species reached the Upper Mackenzie area and 5 species penetrated only to Northern Manitoba and Northern Saskatchewan. The fact that they did reach so far south could be explained by two different hypotheses. In the first, Lake Agassiz did not extend as far north as suggested by Teller or Fisher and Smith but rather the northern boundary corresponded to that indicated by Elson, and Northern Manitoba and Northern Saskatchewan were part of Lake McConnell. The second hypothesis states that the most northern portion of Lake Agassiz according to the boundary suggested by Teller or by Fisher and Smith served in the very late stage as a wide connecting channel with Lake McConnell and allowed some Beringia species to penetrate south, but no further than to Reindeer Lake, or perhaps to Southern Indian Lake.
Lake Winnipeg is the largest remnant of glacial Lake Agassiz. During the 4,500 year existence of Lake Agassiz, a characteristic fauna must have evolved consisting of the species immigrating from connected refugia and other glacial lakes. After the lake drained to the Tyrrell Sea, thousands of newly created lakes including Lake Winnipeg inherited fauna from their great predecessor. Although some changes may have occurred during the following 8,000 years in individual lakes due to modification of morphology, or dispersion of some species by waterfowl, it is assumed that most of the fauna in the lake region remained the same. Much of the history of dispersal of animals must be inferred from knowledge of their present distribution (Stewart and Lindsay, 1983). To throw some light on the mutual relationship between the plankton faunas of Lake Winnipeg and Lake Agassiz to that in other lake regions is the main theme of this work.
The objectives of this paper are: 1) to review the history of Lake Agassiz and to relate it to the present distribution of zooplankton in the region; 2) to review the crustacean zooplankton assemblage presently existing in the former Lake Agassiz basin and to compare it with the zooplankton composition in other regions of Canada; 3) to discuss the northern extent of Lake Agassiz and its connection with glacial Lake McConnell based on the distribution patterns of some crustacean species; and 4) to discuss the role Lake Agassiz played in the dispersion of planktonic fauna throughout central and northern Canada.
The formation of glacial Lake Agassiz and its dynamic history has been described in detail by Elson (1983), Klassen (1983), Teller and Clayton (1983) and Stewart and Lindsey (1983). Pielou (1991) provided a summary of events that led to the formation and later drainage of the lake.
Lake Agassiz came into existence about 12,000 B.P. and lasted until about 7500 B.P. Its shape, size and location changed continually before it finally drained, leaving its largest remnants modern Lakes Winnipeg, Manitoba and Winnipegosis. Agassiz was the largest glacial lake in North America. It innundated an area totaling 950,000 km2 (Figure 1, based on Teller et al., 1983), although at no single time did the lake occupy this entire area. At its greatest extent it covered about 350,000 km2.
Another giant glacial lake, Lake McConnell, lasted from 11,800 to 8,300 years B.P. and occupied parts of the modern Great Bear, Great Slave and Athabasca Lake basins. The total extent of all phases of the lake was 240,000 km2 (Smith, 1994). It reached its greatest extent at about 10,500 B.P. Drainages from Lakes McConnell and Agassiz changed directions several times during their histories. Lake Agassiz initially drained south into the Mississippi valley. At about 10,700 B.P. the ice front receded, opening up a spillway into modern Lake Superior; Lake Agassiz's waters then flowed to the east and the lake level dropped. A temporary re-advance of ice about 10,000 B.P. dammed the eastern outlet and the lake level rose until it again overflowed southward to the Mississippi. The latter direction of flow is questioned by Fisher and Smith (1994) and Rempel and Smith (1998), who suggest that about 9900 B.P. Lake Agassiz drained rapidly north-west through the Clearwater River—lower Athabasca River spillway to Lake McConnell and that this flow direction persisted for 400 years. About 9,500 B.P. the ice on Lake Superior wasted away and another catastrophic flood occurred, draining waters from Lake Agassiz to Lake Superior via Lake Nipigon.
About 8,000 B.P., Lake Agassiz was probably at its maximum extent in Northern Manitoba and is believed by some to have been connected with Lake McConnell. At about this time the lake had formed an eastward outflow to Lake Barlow-Ojibway, the last of the giant glacial lakes that occupied parts of the present James Bay and Ottawa River drainages (Elson, 1967). The final drainage of Lake Agassiz and inundation of the Hudson Bay Lowland by the Tyrrell Sea occurred about 7,500 years ago.
The chronology of the connections between McConnell and Agassiz lakes is not well established. Also geological interpretations of the northern boundary of Lake Agassiz have been subject to significant changes during the last decades (Elson, 1983).
Upham (1890) shows the northern extent of Lake Agassiz to about Cross Lake (55° N. lat.) and then west to The Pas and the Saskatchewan River. The boundary of the northern Lake Agassiz according to the Glacial Map of North America (Geological Society of America, 1945) expands the northern border to Southern Indian Lake then south to The Pas, close to the borders as described by Upham (1890). Elson (1967) retracted the borders about 100 km south of Southern Indian Lake but expanded it further west to include Lake La Ronge. Teller et al. (1983) expanded the borders further north to about 59° latitude to include Reindeer Lake. Smith (1994) proposed a further north-west extension of the area inundated by Lake Agassiz by an area of about 70,000 km2 reaching as far as to the Clearwater River drainage of Lake Athabasca. It can be expected that the general limits of Lake Agassiz, as now understood, will be revised in the future as more detailed mapping is done (Elson, 1983). The tentative character of these limits is also stressed by Dredge (1983).
Material and methods
The data used in this paper constitute part of a long-term project on zoogeographical distribution of planktonic crustaceans in lakes of Canada (Patalas et al., 1994). The samples were collected during a 35 year period from 1950 to 1986; in most regions, between 1969 and 1973 (Table 1). Twenty to 50 lakes were selected in areas of relatively homogenous geology and climate. The lakes represented a wide range of surface areas, depths, water transparencies and so on. Samples from about 400 lakes representing most of the geological and climatic zones of Canada were collected by the author and staff of the Freshwater Institute in Winnipeg. Up to 500 lakes were sampled in cooperation with other agencies: The University of Manitoba, Dept. of Zoology; The National Water Research Institute, Burlington, ON; The University of British Columbia, Dept. of Zoology; The Biological Station, Nanaimo, B.C. Most lakes in northern Canada (north of 60° latitude) were sampled from fixed-wing aircraft, while those in southern Canada that were accessible by road, were sampled from boats.
The sampling was aimed at obtaining mid-summer, pelagic plankton only. During the summer period the plankton was rich in species and, it was assumed, included most of the species that could be found during the remaining open water season. In most of the smaller lakes a single station, usually near the center of the lake, close to its maximum depth, was sampled during the daylight hours. This sampling scheme allowed for an account of 60 to 90% of species present in the lake (Patalas and Salki, 1993). The missing species were usually very rare and represented no more than 1% of the total number of individuals. Station networks in large lakes were more dense and usually consisted of 30 to 50 stations per lake. The relationship between the species richness and the number of lakes (samples) was visible when fewer than 20 lakes were sampled. Patalas (1990a) demonstrated that in a sample of more than 20 lakes per region, at least 90% of species could be found. In regions considered in the present paper, the average number of lakes per region was 46 (22–108), that is, much above the critical value of 20 lakes.
A vertical haul with a Wisconsin plankton net (20-cm in diameter fitted with 72-um mesh net) from near the bottom to the surface and in deeper lakes from at least 50 m to the surface, was used as a standard procedure, although nets of different size were occasionally applied. Samples were preserved usually in 4% formalin. A standard subsample procedure was applied with at least 200 individuals counted. All larger animals such as Leptodora, Limnocalanus and Senecella were counted in the whole sample or larger subsample. Taxonomy used by Brooks (1957), Wilson (1959), Yeatman (1959), Rylov (1948) and Deevey and Deevey (1971) was generally applied. More detailed information on the plankton of Lake Winnipeg can be found in (Patalas 1981, 1990a), Patalas and Salki (1992) and Salki and Patalas (1993).
In order to define how many species in each of the regions are common with the hypothetical Lake Agassiz plankton, a simple ratio, C/R, was constructed (Figure 2) where C = the number of species in the region, common with Lake Agassiz and R = the total number of species in the region. Each of the Laurentian Great Lakes was treated as a separate region. A C/R ratio of 1.0 would indicate that the fauna would be entirely composed of species that were present in Lake Agassiz.
The 214 lakes situated in the area once inundated by Lake Agassiz were grouped into five geographic regions: Northern Saskatchewan,(54°40′–57°00′ N lat. and 102°30′–105°20′ W long.), Northern Manitoba (54°30′–59°30′ N lat. and 95°00′–101°50′ W long.), Lake Winnipeg (50°00′–53°50′ N lat. and 96°15′–99°15′ W long.), Eastern Manitoba (50°00′–52° 00′ N lat. and 93°00′–96°00′ W long.) and Northwestern Ontario (Experimental Lakes Area)(49°30′–49°45′ N lat. and 93°30′–94°00′ W long.).
As indicated earlier, Lake Agassiz at no time inundated the entire area schematically presented on the enclosed maps. For the purpose of interpreting the planktonic fauna in this paper, the 4,500 years of history of Lake Agassiz was subdivided in two major phases: The early phase expanded until 9,500 B.P. (Emerson Phase, Fenton et al., 1983). The lake reached to about the northern end of the present Lake Winnipeg and drained to Lake Superior via Lake Nipigon during the latter part of the phase. The later phase of Lake Agassiz (Phase 4, Klassen, 1983) followed the event (around 8,000 B.P.) when the lake fell below the level of the eastern outlets (to Lake Superior) and receded north of 53° N latitude.
Results and discussion
The comparison of planktonic fauna of Lake Winnipeg and lakes formed from Lake Agassiz with the fauna of other regions of southern and central Canada
The frequency of occurrence of 70 species in 15 lake regions, of southern and central Canada, in five Laurentian Great Lakes and a composite hypothetical ‘Lake Agassiz’ fauna is presented in Table 1. Currently existing lakes can be viewed as remnants of former glacial Lake Agassiz, hence the species recorded in this area could also be considered as the original inhabitants of this glacial lake. On this assumption, a hypothetical composite ‘Lake Agassiz’ plankton fauna was reconstructed as the set of species found in the 214 lakes sampled. In total, 45 species were recorded in the area.
All species found in samples from 45 lakes of Northwestern Ontario (the Experimental Lake Area), 72 lakes of Eastern Manitoba and 48 stations of Lake Winnipeg were considered as the hypothetical species composition of early Lake Agassiz. Similarly, the reconstruction of the planktonic fauna of later Lake Agassiz was based on the present composition of 57 lakes of Northern Manitoba and 39 (out of 53) lakes of Northern Saskatchewan that were situated within the Lake Agassiz range.
A C/R ratio of 1.0 was found in Great Slave Lake as well as in lakes of the Central Ontario region. means that the fauna of these regions are entirely composed of species that were present in Lake Agassiz. Present Northern Saskatchewan lakes also have almost identical fauna (98%) with the Lake Agassiz with the exception of one diaptomid species (Diaptomus forbesi), found in the part of Saskatchewan outside the Agassiz Lake basin. About 94% of species in the Upper Mackenzie region overlap with those in Lake Agassiz (exceptions are Cyclops capillatus and Cyclops nanus. Laurentian Great Lakes plankton (Patalas, 1975, based on samples from 1969) consists of 89 to 94% of the Agassiz Lake fauna. The only other species found in the Great Lakes in 1969, but absent in Lake Agassiz, are Eurytemora affinis and Eubosmina coregoni. These two species are newcomers to the North American fauna (Patalas, unpubl. data). In addition Bythotrephes cederstroemi (longimanus) (not included in Table 1) was recently introduced from Europe to the Laurentian Great Lakes (Bur et al., 1986). Thus, one might say that a century ago, the planktonic fauna of the Laurentian Great Lakes contained exclusively species that were present in Lake Agassiz.
The proportion of shared species in regions outside the Lake Agassiz was very high in lakes of central Canada, generally declining with distance from Lake Agassiz. Ninety to 93% were found in Alberta, Southern Manitoba, Western Manitoba and Southern Quebec; 84 to 86% in Southern British Columbia, Newfoundland and New Brunswick; 80% in Nova Scotia and 74% in the lakes of the Lower Mackenzie region.
It should be noted that as a rule the Great Lakes contain fewer species than most other regions. Despite intensive sampling of the great lakes (from 20 to 60 stations per lake) they have been found to contain only from 16 to 23 species per lake (Table 1). In each of the other regions, an equivalent number of samples (one sample per lake but in 20 to 70 lakes per region) produced higher number of species, from 20 in Newfoundland to 39 in Northern Manitoba. This could suggest that a much wider range of abiotic conditions existed in the smaller lakes of one region than in one great lake, no matter how intensively it was sampled. Understandably, a sample from a single smaller lake produces fewer species, in a range from 1 to 25 per lake. Most commonly, however, lakes in various regions averaged from 5 to 12 species per lake (Patalas, unpubl. data).
Lakes Agassiz and McConnell—an important dispersion route
The present distributions of five planktonic forms provide evidence that glacial Lake McConnell, which occupied area surrounding Great Slave and Great Bear Lakes, was connected with Lake Agassiz, assuming the northern range defined by Teller (1983) or by Fisher and Smith (1994). Cyclops scutifer (Figure 3) is one of the most widely distributed cyclopoid species, present in the northern parts of the continent and the Canadian Arctic Archipelago, in the Yukon and Central British Columbia and in most of Quebec, New Brunswick, Nova Scotia and Newfoundland. It is conspicuously absent from the Laurentian Great Lakes and from most of the area once inundated by glacial Lake Agassiz. It was found in only a few lakes in the Northern Manitoba and Northern Saskatchewan (Reindeer Lake). Diaptomus pribilofensis (Figure 4) shows a similar pattern of distribution: present in the Yukon and central British Columbia but absent in most of the Arctic Islands, Baffin Island and Quebec. Like Cyclops scutifer, it extends to Northern Manitoba and Northern Saskatchewan. Heterocope septentrionalis (Figure 5) is also found in the Yukon, Central British Columbia, the Lower and Upper Mackenzie, Northern Manitoba and Northern Saskatchewan. Eubosmina longispina (Figure 6) is distributed throughout the northern part of the continent, the Yukon and Southern British Columbia, Quebec, New Brunswick, Nova Scotia and Newfoundland. Conspicuously absent from Laurentian Great Lakes and from most of the area once inundated by Lake Agassiz (except of its most northerly portion). Its distribution pattern resembles that of Cyclops scutifer. Daphnia middendorffiana (Figure 7) occurs mainly in the Mackenzie and Keewatin regions, through most of the Arctic Islands. It has expanded south to some small lakes of the Rocky Mountains (Anderson, 1974) and Northern Manitoba.
The distributional patterns of all five species suggest that they came from the Beringia refugium and that glacial Lake McConnell played an important role in their distribution to the south-east. The fact that they did reach to Northern Manitoba and Northern Saskatchewan could be explained by two different hypotheses.
In the first of these hypotheses, Lake Agassiz did not reach as far north as suggested by Teller (1983) or Fisher and Smith (1994) but rather the northern boundary corresponded to that indicated by Elson (1967), about 300 km south; the area around Reindeer Lake was likely part of glacial Lake McConnell. Such a suggestion could be supported by the conclusion of Schreiner (1983) that there is little evidence of Lake Agassiz in the northern part of the Reindeer Lake basin and further north. As well, Elson (1983) admitted that strandlines are poorly developed in the rocky Precambrian terrain in the eastern and northern parts of Lake Agassiz. He expected that the general limits of Lake Agassiz, as then understood, would be revised in the future as more detailed mapping was done. This is supported in recent papers by Fisher and Smith (1994) and Rempel and Smith (1998).
In the second scenario according to the boundary suggested by Teller (1983), the most northern portion of Lake Agassiz served in the very late stage as a wide connecting channel with Lake McConnell and allowed some Beringia species to penetrate south, but no further than to Reindeer Lake or perhaps to Southern Indian Lake (Patalas and Salki, 1984). The chronology of such a connection is not certain. According to the hypothesis by Fisher and Smith (1994), a catastrophic flood occurred about 9900 B.P. and Lake Agassiz waters drained north-west through the Clearwater-Lower Athabasca spillway to Lake McConnell. This north-west flow continued during the following 400 years till 9500 B.P. (Emerson Phase). Such a hypothesis is contrary to the one accepted by others (see Klassen, 1983) that during this phase re-opening of the southern outlet of Lake Agassiz occurred. The Emerson Phase ended with another catastrophic flood and Lake Agassiz drained south-east to Lakes Nipigon and Superior. It seems possible that a temporary reversal of the flow direction from McConnell to Lake Agassiz could have occurred. Zooplankton distribution patterns suggest that the connection occurred in relatively late stage of Lake Agassiz and was of short duration, allowing the Beringia species to penetrate only to the very northern part of Manitoba and Northern Saskatchewan. This would explain the significant difference between the plankton in the early and later stages of Lake Agassiz, consisting of 35 and 41 species respectively. Only 74% of species were common in early and later phases of Lake Agassiz. Most of the difference is due to the fact that the five species from Beringia arrived too late to reach the southern part of the earlier stage of Lake Agassiz, since this part had already emerged above the Lake Agassiz water level and the drainage toward Lake Superior was no longer in existence. Further dispersion south was then not possible.
The distributions of a few other species can serve as evidence of the existence of an efficient mechanism dispersing plankton in the north-west direction. Limnocalanus macrurus (Figure 8) exhibits a very distinct pattern of distribution. It occurs in Quebec, the Laurentian Great Lakes, Eastern Manitoba, Northern Manitoba, Northern Saskatchewan, Lower and Upper Mackenzie, Keewatin, and most of the Arctic Archipelago. It is conspicuously absent west of the Rocky Mountains (Yukon and British Columbia) and east of Quebec and Baffin Island. This distribution indicates that the species did not come from the Beringia refugium but rather survived glaciation in some coastal refugia in the North. There are also good indications that Limnocalanus macrurus survived in a Mississippian refugium, since they are known to occur in central Wisconsin (Marsh 1883, cited from Dadswell, 1974). Lake Agassiz could be then the major dispersion route in both the north and south directions.
Another calanoid, Epischura lacustris (Figure 9), reveals a distributional pattern that suggests it survived in both the Mississippi and Atlantic refugia and it was efficiently dispersed via Lake Agassiz towards the Upper Mackenzie and further north-west to the Lower Mackenzie. Such a distribution would be difficult to explain without a well functioning connection between Agassiz and McConnell Lakes.
The importance of Lake Agassiz as a dispersion route is illustrated schematically on Figure 10. Thirty-five species originating from the Mississippi refugium were found in lakes associated with the early stage of Lake Agassiz (south of latitude 53°). As many as 23 of these species found their way to the Laurentian Great Lakes, when Lake Agassiz drained via Lake Nipigon to Lake Superior and further east.
The later stage of Lake Agassiz (north of latitude 53°) contained 41 species. Thirty one of them originated from the earlier Agassiz stage, 5 species immigrated via glacial Lake McConnell from the Beringia refugium, and another 5 species (whose origin is not certain) could have arrived from the South-West, possibly from the Missouri drainage area.
Twenty four species of the original 35 Mississippi migrants (69%) reached McConnell Lake and 16 species (45%) were dispersed as far as the Lower Mackenzie area.
A parallel process dispersed fauna from the Beringia refugium in the south-east direction; however, its effectiveness was apparently lower. In the area of Lower Mackenzie only 15 species survived glaciation (11 in Beringia and 4 in the Arctic coastal waters). Of these 15 species only 7 species (47%) dispersed southeastward and are found at present in the area innundated by Lake McConnell. The number of species from Beringia had been reduced to 5 (33%) by the time they entered Northern Saskatchewan and Northern Manitoba. Finally, when the connection between later Lake Agassiz and glacial Lake Barlow-Ojibway opened (about 8300 B.P.), only two (13%) Beringia species, Cyclops scutifer and Eubosmina longispina, and 23 species originating from Mississippi refugium, reached lakes situated north of the Laurentian Great Lakes. None of the Beringia species was able to reach the Laurentian Great Lakes. The connection between Lake McConnell and late Lake Agassiz probably opened when connections between Agassiz and Lake Superior were already disrupted.
The very high proportion of fauna originating from the Mississippian refugium which reached the Lower Mackenzie (Fig 10), seems to be consistent with the north-west flow direction hypothesis, proposed by Fisher and Smith (1994) and Rempel and Smith (1998). However, the presence of the Beringia species in northern Manitoba and northern Saskatchewan suggest a temporary reversal of the flow direction.
The very efficient dispersion mechanism of Lake Agassiz resulted in an extremely high diversity of planktonic fauna in northern Manitoba. In this relatively small area, not greater than 1% of the area of Canada, as many as 39 species were recorded. This accounts for 46% of all 84 species found in the whole of Canada and represents the highest species richness of all lake regions in this country (Patalas, unpubl. data).
It is hoped that the comparison of the geographical distributions of planktonic crustaceans in the lakes once inundated by Lake Agassiz (as interpreted by geomorphologists) with other Canadian lakes outside its former range may contribute to a better understanding of the postglacial history of Manitoba as well as the present patterns of zooplankton distribution.