A Geomorphic–Trophic Model for Landscape Control of Arctic Lake Food Webs

Lake trophic structure is a result of complex interactions among consumers and their resources. Algal productivity is channeled to higher trophic levels, which in turn structure planktonic consumer food webs and control algal biomass through cascading trophic interactions (Shapiro 1980, Carpenter et al. 1985). Top predators also affect benthic food web structure (Bronmark et al. 1992). However, cascading effects in benthic food webs are confounded by habitat complexity in the littoral zone and accumulation of organic matter in lake sediments (Crowder and Cooper 1982, Gilinsky 1984, Hershey 1985, 1992, Carpenter and Lodge 1986). Thus, benthic-pelagic coupling is not well integrated into the current understanding of lake trophic dynamics. However, because arctic lake food webs are dominated by benthic-pelagic interactions, a model that focuses on planktonic food webs does not describe them well. Lake ecosystems are embedded within the landscape but have rarely been studied in that context. One of the best contexts to examine this question is the Arctic, where geomorphic constraints on fish distribution are especially apparent because the fish community is small and geographically constrained. The benthic and pelagic invertebrate communities are also not species rich, although their trophic diversity is similar to that found in temperate lakes. The arctic landscape therefore provides a unique opportunity to resolve the relationships between geomorphic setting, fish distribution, and lake trophic structure. A unifying conceptual model of the arctic landscape is the geomorphic–trophic hypothesis (Figure 1). This model integrates lake processes into their geomorphic setting by illustrating how the stream network constrains or channels the dispersal of fish and links fish dispersal with fish control of lake food webs; because landscape criteria control the distribution of fishes, and fish control lake trophic structure, the landscape indirectly controls lake trophic structure. Similar principles have been applied to fish distributions in other systems (see Magnuson et al. 1998), and landscape factors are also known to affect hydrologically driven chemical parameters, which may in turn have implications for food web structure (Kratz et al. 1991, 1997). BECAUSE LANDSCAPE CRITERIA CONTROL THE DISTRIBUTION OF FISHES, AND FISH CONTROL LAKE TROPHIC STRUCTURE, THE LANDSCAPE INDIRECTLY CONTROLS LAKE TROPHIC STRUCTURE According to the geomorphic–trophic hypothesis for arctic lake food webs, extant and paleological landscape constraints provide a template resulting in lakes with six different types offish communities (Figure 1): lakes dominated by lake trout (Salvelinus namaycush) but also usually containing grayling (Thymallus arcticus), sculpin (Cottus cognatus), burbot (Lota lota), round whitefish (Prosopium cylindraceum), and, rarely, arctic char (Salvelinus alpinus); lakes containing arctic char and sculpin; lakes with grayling and sculpin; lakes with grayling only throughout the year; lakes with grayling in the summer only; and lakes with no fish. Lakes without fish can occur either because the stream outflow gradient (both extant and paleological) is inaccessible or because the lake is of insufficient depth to support fish and is not connected to a deeper lake (Figure 1). The geomorphic–trophic hypothesis. The geomorphic–trophic hypothesis is a conceptual model presented as a decision tree whereby fish species distribution is determined by geomorphic features of the landscape, including lake depth, lake surface area, and outflow gradient. By quantifying the landscape criteria for each fish species, it is possible to determine fish community structure for a given lake. Fish community structure in turn affects benthic and pelagic invertebrate communities. For arctic lakes, an ability to predict the distribution of lake trout (Figure 2) is of particular interest because when present, they serve as keystone predators. By quantifying the landscape criteria for distribution of fishes, including this keystone species, the geomorphic–trophic hypothesis also serves as a predictive model for lake trophic structure. In this article, we describe the fish communities in the foothills region of the North Slope of Alaska in the vicinity of the Toolik Lake Long-Term Ecological Research (LTER) site. We then discuss the geomorphic–trophic hypothesis of landscape control of fish distribution and illustrate how landscape criteria can be incorporated into a predictive model for distribution of lake trout. Finally, we summarize fish control of trophic structure in arctic lakes and show how an important food web characteristic, linkage complexity, is also influenced by fish distribution and, thus, landscape characteristics. Lake trout (Salvelinus namaycush.) Many of the lake trout in the vicinity of the Toolik Lake Long-Term Ecological Research site in Alaska are relatively small and slow growing, reflecting the bioenergetic constraints imposed by their snail diet. In addition, fishing pressure in the region has shifted the size distribution of lake trout toward smaller individuals (McDonald and Hershey 1989). Fishes of Northern Alaska in the vicinity of Toolik Lake are regionally constrained by the Brooks Range, which isolates watersheds draining into the Arctic Ocean from watersheds draining into the Gulf of Alaska. However, local distribution of fishes is determined by geomorphic barriers established by glacial history and postglacial geological processes that established extant lake and river basins (Hamilton 1982, 1986). These geomorphic constraints resulted in lakes with a small number of discrete fish communities, which in turn determine the characteristics of both benthic and pelagic consumer food webs in the lakes. Lake trout are the dominant fish in most large lakes in the region. They are typically characterized as piscivores (Johnson 1972), but only approximately 12% of the lake trout stomachs we have examined contained fish remains (Merrick et al. 1992). A generalized food web for lakes with lake trout is shown in Figure 3. Lakes with fewer fish species have food webs that are subwebs of the one shown in Figure 3. The major food resource for lake trout in arctic lakes is the large-bodied snail, Lymnaea elodes (O'Brien et al. 1979, Hershey 1990, Merrick et al. 1991, 1992). The biotic community in lakes of the region is strongly influenced by lake trout, either directly, through predation (Hershey 1985, 1990, Merrick et al. 1991, 1992), or indirectly, due to the risk of predation (McDonald and Hershey 1992, McDonald et al. 1992, 1996). The presence of large lake trout forces the other fishes, including juvenile and young-of-year lake trout, into marginal nearshore habitats where predation risk is low but so are food availability and growth rates (Johnson 1972, 1976, McDonald and Hershey 1992, McDonald et al. 1992, 1996). Generalized food web of arctic lakes. Large lakes are dominated by lake trout or, in the absence of trout, by arctic char. Both top predators feed predominantly on snails but also control recruitment and spatial distribution of their own young and young of other fishes that may be present. Lakes that lack both lake trout and char will have food webs that are subwebs of the one shown. Anthropogenic activities directly or indirectly affect lake trout at discrete points during their ontogeny. For example, fishing pressure on lake trout can reduce the density of the population and change its size structure (Johnson 1972, McDonald and Hershey 1989). If the lake trout are harvested before reaching reproductive maturity, loss of adults might be expected to result in lower recruitment of young-of-year lake trout, which would in turn have a long-term effect on the population age structure. However, a whole-lake experiment showed that, in the absence of adult lake trout, small lake trout grow 2-10 times faster than they do in the presence of adult lake trout (Figure 4), presumably because they have better access to food of higher quality. Thus, small lake trout are able to grow rapidly when new food resources become available. Removal of adult lake trout enhances juvenile growth. Increased growth of juvenile lake trout occurred following experimental removal of the adult lake trout population in Lake NE-12. Solid circles show weight at a given age before removal of adult fish. Open diamonds show weight at a given age after the removal of adult fish. Young lake trout grow much faster when adult fish are absent. Where lake trout are absent, their role as top predator is sometimes replaced by arctic char. In the rare cases in which the two fishes co-occur, lake trout appear to have eliminated arctic char or relegated them to a subordinate position (Johnson 1976). Arctic char are typically an anadromous fish that overwinter in lakes (Johnson 1980, Parker and Johnson 1991), but on the Alaskan North Slope they currently occur only in landlocked populations, where they are considered glacial relicts (Reist et al. 1997). In this region, historical river-run populations of arctic char have been replaced by morphologically similar Dolly Varden (Salvelinus malma), a primarily riverine species adapted to warmer environments (Reist et al. 1997). Because arctic char can survive at glacial fronts, it is likely that they were more widely distributed before Illinoian glaciation (more than 40,000 years ago). The other four fishes that may co-occur with lake trout and/or arctic char are trophically distinct. Slimy sculpin are small, benthic-feeding fish that are abundant and widely distributed. In arctic lakes they feed almost exclusively on Hershey 1985). Arctic grayling are primarily when small, but as they grow they also benthic and and (O'Brien et al. 1979, Merrick et al. 1992). appear to be the most widely distributed fish in the this distribution their and which feed on the snail, juvenile and other small, benthic (Merrick et al. 1992), are found only in a lakes in the region, although they occur in the large river systems feed on sculpin and other distribution is to because they are not to However, they do not appear to be as widely distributed as lake trout. The geomorphic–trophic presented as a decision tree in Figure provides a conceptual to the distribution of lake trout and other the landscape criteria embedded within this decision tree that to lake trout is important because lake trout are the dominant fish in the region. geomorphic constraints in the geomorphic–trophic hypothesis fish extant and paleological of stream to lakes through the stream and lake depth and One of the geomorphic–trophic hypothesis is that outflow stream gradient control which species of fish can a with and depth which species can in lakes that are to a given fish. controls species because different fish species have to Lakes with a outflow gradient will sculpin and lake trout but not and will of lake size and depth (Figure 1). in have summer in and lakes, and through to overwinter in lakes et al. They may in the summer in lakes lakes are connected to deeper lakes by lake lake trout do not to into as do and they are not they outflow to lakes. In addition, their is only that of arctic char thus, char are also expected to that are higher than to lake trout. Slimy sculpin lack and are relatively and Thus, they not be able to small and would have to they are not found in arctic that they are in large which do not of the lakes in the Toolik Lake area, lakes and illustrate the role of gradient in the geomorphic–trophic hypothesis because they have and (Figure Lake is approximately from Toolik Lake and in the and Lake is higher in the and into Lake contained grayling but not presumably because the landscape of outflow gradient sculpin to a food web that from that in sculpin lakes. sculpin the landscape gradient to examine the role Lake with its has no fish (Figure Thus, the of Toolik to Lake to Lake an of the effect of gradient on fish species on Slimy sculpin were into Lake in 1992, to in density and on Slimy sculpin were into Lake in 1992, to in density and gradient fish species among lakes. to provides on the of the outflow stream between lakes. In this example, only arctic grayling can the gradient from Toolik Lake has species to Lake and grayling the gradient to Lake major to the glaciation occurred in the Toolik Lake region (Hamilton 1982, 1986). The glaciation is at more than 40,000 years and most likely the Illinoian the glaciation is of age years were more than a glacial also has a role in landscape In this smaller with rapidly their into other and of This affects fish distribution because lakes that a more outflow gradient may have a gradient. of where occurred on the landscape is to a lake to fish. The that a of gradient lakes in the region, the arctic char and to of the paleological landscape setting in the geomorphic–trophic hypothesis. of and of the landscape lakes paleological with riverine systems glacial that would have resulted in for fish. the of the lakes would have become as the riverine due to stream and other geomorphic However, the fish Thus, although extant stream outflow gradient is an important in arctic lakes, as well as in other lake (see Magnuson et al. 1998), the Lakes provides a that gradient is to change geological a that be incorporated into model of fish lakes that are connected to lakes by summer grayling overwinter in lakes, but because lakes are food availability may be higher than in the lakes or where grayling the Lake trout also among lakes, although they are of than are grayling (McDonald et al. their of lakes during the Lake depth and are but they different in the of fish species and the structure of the food lack fish because they In is more than thus, only lakes of depth have habitat for fishes and usually support grayling and However, of the more than arctic lakes that we have (O'Brien et al. 1979, Hershey 1990, et al. 1992, Hershey et al. in lake trout are found only in relatively large (more than and (more than lakes with low extant or paleological gradient. These that a lake size is to provide food and habitat for a lake trout A depth the relatively low for lake trout and McDonald et al. a lake is to provide a during summer we have geomorphic criteria can be to predict the presence or absence of lake trout. We to the of geomorphic features in lake trout distribution at two different spatial to for lake within systems and characteristics within a (Figure In the of the systems within the and systems were lake trout or on the surface of the lake within the The into the Arctic By including extant and historical riverine we of the systems in of the presence or absence of lake trout. In the of the lakes were lake trout or within systems in the to have lake trout. Lake surface to this of lakes with the of lake depth or outflow stream gradient not this tree for landscape control of lake trout The two landscape criteria that affect distribution of lake trout are lake and to large lakes. Lake trout are from lakes that do not occur in current or paleological) containing a large lake (more than If a large lake is or in the the decision tree that lake trout will be in lakes with a surface more than The the presence or absence of lake trout in of the lakes According to for each lake trout are typically found in systems that or a lake is than as well as in lakes is more than This that lake trout distribution is on a landscape the of lake trout in trophic structure, this also the of geomorphic control of food of the lake communities in the Toolik Lake on the geographically constrained species thus, they invertebrate However, the different geomorphic constraints on fish the invertebrate communities as well. We invertebrate communities 2) that in of particular species as well as in of of the species the geomorphic–trophic hypothesis as a decision tree for fish benthic and pelagic community structure can then be 2) on understanding of fish control of trophic structure. communities are determined by fish species constraints on fish in the Toolik Lake Long-Term Ecological Research site result in six different fish community These community types have different for predator control of trophic structure and, different community characteristics. Lakes in the Toolik Lake region that have been well studied serve as communities are determined by fish species constraints on fish in the Toolik Lake Long-Term Ecological Research site result in six different fish community These community types have different for predator control of trophic structure and, different community characteristics. Lakes in the Toolik Lake region that have been well studied serve as Lake trout serve as keystone predators in large lakes of the region, the structure of both benthic and pelagic thus, control of lake trout distribution to landscape control of lake trophic structure. The benthic communities of lakes with lake trout have a higher species and diversity than without fish or grayling and sculpin they also have a of species and Hershey 1992). The and occur in lakes that lake trout and are not found as an in the other lake types and Hershey 1992). sculpin control the and species diversity of on sediments (Hershey 1985, and Hershey 1992), the dominant habitat in lakes in the region. Lake trout in turn determine the distribution of sculpin on In lakes with lake trout, sculpin are more abundant on and at the where they have from lake trout but where is also food with lakes that do not lake trout et al. 1992, McDonald and Hershey 1992). are more abundant in the absence of both lake trout and sculpin than when sculpin but not lake trout are in the absence of they appear to for sculpin in the food Thus, are higher in lakes and than in lakes that have sculpin but not lake lakes containing lake trout have higher than either or sculpin lakes because lake trout sculpin to low density on the sediments and Hershey 1992, et al. 1992). The sculpin are to at low density (Hershey 1985, and Hershey 1992). This is by an experimental of sculpin into Lake Lake contained only which rarely feed on although adult a of their (Merrick et al. 1992). the of Lake of with of the to years after sculpin during which sculpin to the density by a of and were not 1). communities are also by lake trout. The large-bodied snail Lymnaea is rare on the sediments of lakes with lake trout, the snail is (Hershey 1990, Merrick et al. This distribution the that adult Lymnaea are the major of lake trout, the smaller are of lake trout into a small lake resulted in a in density and size of adult Lymnaea (Merrick et al. lake trout indirectly interactions between the two snails because Lymnaea controls recruitment of where the two snails co-occur without lake trout, is by and Lymnaea is by (Hershey 1992). Large lake trout indirectly affect arctic communities by predation on from smaller fish in two they on smaller fish and reduce their and they the access of small fish to the of lakes, where the (O'Brien et al. 1979, McDonald et al. 1992). These effects are by the that in lakes that grayling but not lake trout, the large is absent. lakes have including and (O'Brien et al. the other large is also when grayling are but lake trout are is present, it is much smaller than it is in lakes with both lake trout and grayling Thus, the absence of the large and the size of in the presence of grayling show that grayling can structure the community to one that is However, this ability is when grayling occur with lake trout. for the effect of lake trout on from in Toolik Lake trout size due to fishing in the and the of lake trout from in to in (McDonald and Hershey 1989). this from in to than in the et al. This in is with small grayling and lake trout in the of Toolik we have no to this of grayling in the but a in the also the effect of lake trout on For example, Lake which is approximately of Toolik Lake in a different a large population of lake trout and a arctic In the summer of 1991, of the large lake trout were from the lake by the the density of from in to in the size of smaller in than in (Figure is most likely that small lake trout and arctic grayling were able to in the of the lake after the removal of the large lake trout, the of predation on the Lake trout indirectly affect the size distribution of A change in size distribution of occurred in Lake following experimental removal of adult lake trout. trout were in 1991, size in Large lake trout indirectly affect on large In to understanding the by which fish species controls invertebrate community it is of interest to determine the fish also controls the structure of the food A food web interactions among food web to relationships from which can be as a to the of through the The of can then be examined or of this the fish The of food web is to principles of food web The is to on history and experimental although the and important of species interactions are may that of linkage complexity that the incorporated in one species is widely throughout the food web through interactions with other have that the the number of for to a the the for the one because of local have to the that more complex webs are One given for of of species is that it will to complexity and presumably of ecosystems and the they Thus, it is important to the of the et al. have that complexity and are when a of has shown that the complexity of pelagic food webs does with on the structure of food webs of lakes the different decision of the hypothesis (Figure to the in a landscape In to the of et al. but in with of we have found that linkage complexity with number of fish species (Figure complexity is approximately links species in lakes without but each of fish species the number of links species by approximately a lakes the species with the two by and lakes with fishes present, and the by grayling lakes with or without The between landscape features and lake trophic structure in the arctic landscape the of landscape features to lake food the species diversity of arctic invertebrate communities is their trophic diversity is similar to that of their temperate (Hershey et al. These arctic invertebrate communities the of predator effects at different points in the food By how the landscape within which arctic lake communities are embedded the distribution of fish species, the geomorphic–trophic hypothesis serves as a unifying that and the history of the region into the geomorphic–trophic hypothesis the understanding of species interactions to the of arctic as a For example, outflow gradient with glacial and postglacial the presence of fish in lakes that have a outflow gradient. By the and interactions of fish with each other through a of invertebrate (Figure landscape to indirectly control the of throughout the food to in and between lakes, as depth, lack of to deeper lakes, and outflow as landscape that the and of different fish species, on the of the Thus, because landscape structure fish indirectly invertebrate communities, it in food webs in which are well than where more fish species are (Figure offish species on linkage complexity, which is approximately the number of links species (see for with the number offish species present, which is landscape control lakes with no lakes with grayling only; lakes with grayling and sculpin; lakes with lake trout or or four other fish Thus, by the distribution the landscape controls the of within a food In to for better understanding of the structure and of arctic lake food webs, the geomorphic–trophic hypothesis provides a for better of the of arctic lakes. By fish communities of most lakes on geomorphic this model would provide with a and to or to fishing on lakes in a region. of a that the fish populations are but the between and of linkage remains more on food webs of arctic lakes in to geomorphic barriers to fish distribution, and interactions is before can be For example, of and fishes in lakes and a throughout the the food web and to local of species and 1997). relationships will be by other factors as of fishes or of but relationships will be a of and Lake trout and arctic char population are of particular in of and in lakes the Arctic because lake trout has been widely as a fish in the of the to the of and which is to arctic char. arctic char to occur in higher lakes than do lake trout, and the two species rarely in distribution and In the arctic lakes we are of arctic char by lake trout to be a much than with Thus, would that lake trout in char lakes be in the and not in the Alaskan The geomorphic–trophic hypothesis of landscape control in the Alaskan Arctic be to other as well. However, the relationships between landscape control of fish species distributions and fish control of lake trophic structure to be studied for the geomorphic and in which they are For example, lake in and has shown that geomorphic setting is important in and offish species et al. because fishes are known to control food web structure and in the lakes et al. 1985, et al. and factors determine distribution of fishes et al. 1998), the geomorphic template also controls food webs in lakes. Thus, although arctic lakes have unique food webs and a unique geomorphic setting, the principles for understanding landscape control of trophic structure in arctic lakes also be to food webs in other lake We the for its support through to and with to and and to with to and We also the of and Finally, the and at the Toolik Lake to this in important without their the would not have been