Numerous lake classifications have been developed during the past century with diverse origins and purposes for their use. Busch and Sly (1992) identify many approaches in their review of lake habitat classifications. Uses for classification include basic research, application in resource management and satisfying legal mandate. Lake classifications have been developed based on physical (Canfield et al., 1984; George and Maitland, 1984; Lewis, 1983; Larkin and Northcote, 1958; Pennak, 1958; Hutchinson, 1957; Rawson, 1955), chemical (Vollenwieder, 1968; Stockner and Benson, 1967), and biological characteristics (Tonn and Magnusson, 1982; Jensen and Van Der Maarel, 1980; Brundin, 1958; Round, 1958; Nygaard, 1949; Thieneman, 1909) and combinations of these characteristics (Carlson, 1977; Schindler, 1971; Ryder, 1974).

Water quality and the biological structure of high mountain lakes are influenced by elevation, lake morphology, and watershed characteristics. Mosello et al. (1990) reported that water chemistry in alpine lakes of NW Italy was determined by lithology, vegetation, and amount of atmospheric deposition. Pechlaner (1971) determined that primary production and phytoplankton biomass in high mountain oligotrophic lakes were affected by lake hydrology, temperature, winter snow cover and water chemistry. Lomnicky et al. (1989) provide a conceptual framework arguing that lakes interact with climate and watershed features, and that lake morphometry is important to the physical, chemical and biological characteristics of lakes at the spatial scales found in NOCA. Recently, Larson et al. (1994, in press) reported that limnological characteristics of high mountain lakes result from complex interactions of climatic conditions, watershed characteristics including geographical location, aspect, surface area, elevation, geology, hydrology, soil and vegetation, and lake morphometry.

NOCA lakes are diverse in many physical attributes, although on a global scale they are quite similar (Goldman and Home, 1983). Primarily glacially formed high mountain lakes, most would be classified oligotrophic (Likens, 1975). Lakes within NOCA differ in geologic and climatic setting, age, origin, elevation, aspect, extent of glacial influence, vegetational setting, morphometry including size and depth, and many other characteristics.

The goal of the work reported in this section was to better understand the similarities and dissimilarities among mountain lake systems in NOCA relative to physiographic characteristics (climate, topography, geology, vegetation) of the terrestrial environments.

The objectives were:

1. Develop a hierarchical classification incorporating physiographic (geological, vegetative and climatological) attributes of the terrestrial environment.

2. Determine physical and chemical properties of lakes and relate these to physiographic characteristics of the terrestrial environment through classification.

3. Use classification to identify the diversity of lake types (classes) parkwide.

Previous classification schemes reflect linkages between lake and watershed characteristics and processes, but have tended to deal either with geographical regions having a greater physical and chemical diversity among lakes or a more uniform environment than is present at NOCA. A different approach was necessary for the specialized conditions found at NOCA. A conceptual view of lake development and its relationship to the expression of lake and watershed characteristics was developed to provide context for this classification approach.

Lakes, like streams, are products of their surrounding watersheds. Likens and Borman (1974) recognized a linkage between land and water though they did not classify lakes. Others have taken a stronger view that water resources are embedded in land ecosystems (Swanson et al., 1988; Frissell et al., 1986; Lotspeich and Platts, 1982; Warren, 1979). Land ecosystems or watersheds are the environment of lakes (Aber and Mellilo, 1991; Warren, 1979). The influence of watershed characteristics (slope, geology, aspect, vegetation) determine rates of nutrient loading to streams and lakes (Mosello, 1990; Hem, 1989; Teti, 1984; Schindler, 1971; Vollenweider, 1968; OhIe, 1965; Gorham, 1961).

A conceptualized view of the complex interrelationships of watershed and lake components (Figure 1) highlights the importance of geology and climate in determining many performances of watersheds and lakes. The expression of physical, chemical and biological attributes, anything measurable, may be described as a performance of the lake or surrounding watershed. Examples include watershed area, vegetation, lake temperature, surface area, depth, and densities and types of organisms present. Geologic and climatic processes exert powerful influences on substrate and physical and chemical performances of lakes both directly, and indirectly through their impact on watershed structure (relief, aspect, soil maturation and vegetation development) and lake morphometry (area, depth).

Lakes can be thought of as dynamic systems that are continually changing or developing relative to changes in their watershed environments. For example, a lake recently exposed by glacial recession would likely have quite different chemical and physical characteristics than an otherwise similar lake formed by the same processes and exposed thousands of years ago. Lake performances, then, are influenced by both the stage of development of the lake and the stage of development of its watershed environment.

Identifying and understanding the developmental processes of this glacially influenced mountainous terrain aids in recognizing those processes which ultimately constrain and define classes of lakes. Temporal and spatial scales appropriate to NOCA for developmental processes and evolutionary events are given in Table 1.

Lake and Watershed Development in NOCA

Climate and geomorphic processes such as tectonics and glaciation mold a geologic template that determines lake evolution and development. Geologic and climatic processes influence physical and chemical performances of lakes through impact on watershed structure (Aber and Mellilo, 1991), lake morphometry (Rawson, 1955), rate of soil maturation (Buol, 1973), and vegetation (Mosello, 1990). Developmental processes that are constrained or enhanced by climate include drainage network development, organic material accumulation, and sedimentation (Table 1). These processes affect the rate and path of water movement through watersheds which affects nutrient concentrations in lakes.

Watersheds, lakes, and the biological communities in lakes can be thought of as codeveloping systems. Development can be considered over geologic, annual, and seasonal time frames. Watershed development over geologic time in the North Cascades has involved the working and reworking of the terrain by glaciers, downcutting by stream channels, development of soils from bedrock and till, and establishment and development of vegetation communities. This region is still in the process of change. Climatic change occurring only recently has led to the recession of glaciers and the formation of new lakes. Whether watershed changes are viewed over the short-term or long-term, lakes and streams will, to some extent, reflect these changes. We can conceptualize the development of lakes through time and the characteristics they exhibit at any particular time as being determined at least in part by the development and characteristics of their watershed environments.

NOCA lakes are predominantly of glacial origin, and their development has been generally initialized with glacial ablation. NOCA lake and watershed development can be understood as progressing through a scenario similar to our present day alpine, subalpine and forest states given enough time and favorable climatic (temperature, aspect) factors. Under present regional climatic conditions, alpine watersheds and lakes exist at the highest elevations while forest systems are at the lowest elevations. Subalpine systems occur between the two.

The alpine landscape is severe, a cold, rocky, barren, dessicated region filled with ice and snow for much of the year. Watershed relief tends to be steep except for lakes perched on ridgetops. There is minimal water retention in the basins except in frozen form since soil formation processes are just beginning, and steep relief hastens water exodus from the watershed. Ice-out in this environment comes late in the summer season if at all, minimizing vegetation development. A lake's local environment is not moderated by soils or surrounding vegetation. These systems are presently restricted to the higher elevations in the headwaters of drainages and consequently tend to have smaller watersheds.

Under favorable climatic conditions, development progresses from the alpine to the subalpine stage. Increasing soil maturity leads to enhanced hydrologic retention. This increases the potential for mineral weathering and nutrient liberation to lakes from bedrock. In a subalpine climatic regime, lake iceout is generally earlier in the year than in alpine systems but later than in forest lakes. The lakes are usually warmer than alpine lakes due to an increased period for direct insolation and generally warmer ambient air temperatures at lower elevations. Though glacial inputs are decreasing, other infilling processes (colluvial transport, avalanche chutes, stream transportation) add inorganic and organic materials to the lakes which decreases lake depth. The decreased time period of winter cover provides increased opportunity for vegetative growth. Successionally, this eventually leads to a Parkland mosaic (Franklin and Dyrness, 1973), a mixture of meadows and stringers of trees which provides added organic input to lakes in the form of leaf litter and woody debris. Subalpine lakes are found near the treeline usually tucked near valley headwalls or on west-facing ridges.

Under more temperate climatic conditions with a long enough snow-free period for tree establishment, the potential exists for watersheds to develop to a forest state. Warmer temperatures lead to a reduced period of winter snowpack, increased amount of vegetation, greater soil nutrient concentrations and accumulations, and an organic horizon in the soils. Increased soil/water interactions create elevated nutrient concentrations delivered to lakes via inlets and groundwater. Infilling processes carrying suspended sediments and organic inputs reduce the mean depths of lakes. Fed by the more abundant groundwater, a greater number of permanent inlet and outlet streams exist. Weathering processes such as frost riving and colluvial raveling tend to decrease watershed slope relief, further increasing the potential for soil and water interactions. Forest lakes are found in the lower elevations of the park. They are usually in valley bottoms away from ridge slopes since colluvial infilling processes tend to push lake shorelines towards the middle of valleys.

Factors Affecting Development

The rate of watershed and lake development can be affected by climatic regime and disturbance factors. Development may be arrested or reversed at any stage or state due to prevailing or changing climatic conditions. Elevation, latitude, basin aspect and the proximity to glaciers influence localized climate (Table 1). These varying conditions may affect the rate or potential for further development. For example, two lakes at the same elevation on a mountain may have differing aspects resulting in very different localized climates. West-facing lake basins, with warmer aftemoon sun and increased precipitation relative to east-facing basins, may have progressed to a forest stage while the east-facing basin still has alpine characteristics. As a result, elevation is just one factor determining local climate in a mountainous environment.

Disturbance regimes such as fires and annual snow events can maintain a particular developmental state by resetting successional trajectories following fire or other disturbance. In the subalpine zone, snow period minimizes tree seedling survival prohibiting forest states from developing. Additionally, plant community succession may be reset following fire or other disturbance.

Table 1. Processes or events controlling systems on different spatiotemporal scales in North Cascades National Park Service Complex.

Classification Rationale

Based on this view of development for glacially influenced high mountain lakes and their watersheds, a classification system was devised. We approached classification as a means of ordering systems having common properties and developmental processes. Variables used for classification should be highly determining of system performances (Warren, 1979). The variables used to define and classify systems at each hierarchical level (Table 1) must be appropriately scaled to the space-time frame of the systems composing that level (Frissell et al., 1986).

NOCA is located in north central Washington on the Canadian border (Figure 2) and contains 162 alpine, subalpine and forest lakes identified by Park Service biologists as important to fisheries management. The rugged topography of the park spans the crest of the North Cascade mountains separating the park into two geographical regions, westslope and eastslope. This is a tectonically active region (Press and Siever, 1982), and many of the landforms are still being altered by processes such as faulting, uplift and subsidence (McKee, 1972). Accretion of an island microcontinent (Alt, 1984), metamorphism and intrusions of magma, have created a complex park lithology. Hard crystalline bedrock, predominately gneiss and granite with some sedimentary and volcanic lithology (Misch, 1977), underlies the region. Local soil surveys have not been completed, but regional soils are primarily young cryandempts that form during cool summers (Jackson and Kimerling, 1993).

Over the last 50,000 years, climatic processes such as glaciation have reworked the geologic template. During the Fraser period of glaciation, 25,000 to 10,000 years ago (Crandell, 1965), the major river valleys (Chilliwack, Skagit, Stehekin) were broadened and deepened. Some larger lake basins were created. Neoglaciation reworked most of the present day landforms and existing lake basins at the higher elevations about 2500 years ago (Crandell, 1965). Flowing from the headwalls, glaciers carved depressions and created moraines and other depositional landforms of alluvium and outwash, some which have become morphogenetic features of lake basins. Today, remnant glaciers are reminders of a much more severe past climatic regime. These glaciers still influence processes within some lakes by providing turbid meltwater inputs to lakes.

Differential precipitation across the hydrologic divide separates the park into two major climatic units (Jackson and Kimerling, 1993). Westslope climate is maritime with warm, moist air coming off the Pacific Ocean and Puget Sound, creating wet relatively warm winters and cool summers. East of the crest, a more semi-arid continental climate exists with colder winters and hot dry summers. Precipitation, comprising the major weathering component of the present climatic regime, ranges from 125 cm to 305 cm per year on the eastslope and 152 cm to over 350 cm per year on the westslope (Lomnicky et al., 1989). Ice and snow blanket most lakes approximately nine months each year. A majority of the lakes are open by mid July (Lomnicky et al., 1989) and freeze over by late October to early November on the eastslope and two weeks later on the westslope (Tony Reece, pers. comm.).

Westslope vegetation is typified by lowland and montane forests. Lowland forests are generally below the permanent winter snowpack while montane forests exist in areas where snow remains the entire winter. The subalpine zone is composed of lush herbs and woody shrubs. The alpine zone is recognized by krummholtz, a growth form, rather than any particular species. Rock and ice dominate the alpine zone. Less understory exists in eastslope forests which are drier than westslope counterparts.

Physical

U.S. Geological Survey topographic maps (7.5') were used to determine lake watershed area, surface area, elevation, and permanent inlet streams. A BASIC program, GAP4 (GTCO 1982), was used to manually digitize the flat map watershed area and surface area of each lake. Lake elevation was recorded from the park service database.

Bathymetric maps of each surveyed lake were developed from depth measurements taken along transects using a hand-held sonar depth finder. Depth isopleths were developed and volumes, V= h/3(A₁ + A₂ + (A₁A₂)), determined using a digitizer. With this data, mean depth was determined for each lake (MZ= V/A_o where A_o = surface area).

Temperature was measured over the deepest spot in the lakes at one meter intervals from the surface to the bottom. An Omega HH70 series handheld thermometer with a single input Type K chromel-alumel thermocouple was used.

Date of ice-out, which was defined as lake surface virtually ice free was determined for most lakes from weekly aerial surveys over the park. For a few lakes, we relied on information from backcountry rangers. Precipitation values for lake watersheds were determined from a map developed by Agee and Pickford (1985) from state climatological data with revisions using local survey information.

Lake origin was determined by visual interpretation of air photographs of all park lakes using a stereoscope adapting descriptions from Hutchinson (1957). A total of 152 lakes were identified by basin origin. The origin of 10 lakes was unknown.

Chemical

A total of 142 water samples were collected (1989-1992) from 58 lakes. Water samples were collected 1 meter below the surface over the deepest spot in each lake from inflatable boats using a 1.5 liter Van Dom style bottle. Filtered (0.7µm prewashed Watman GE/C filters) nutrient and trace element samples were placed in 1 liter acid washed polyethylene bottles. Unfiltered samples (250 ml acid washed polypropylene bottles) were collected for determination of conductivity, alkalinity and total phosphorus. Water samples were transported out of the field, frozen and then shipped to the laboratory. Chemical analyses (Table 2) for total Kjeldahl-nitrogen (TKN), nitrate-nitrogen (NO3), total phosphorus (TP) orthophosphorus (OP), alkalinity (ALK), pH, conductivity (COND), sodium (NA), potassium (K), calcium (CA), and magnesium (MG) were performed at the Cooperative Chemical Analytical Laboratory, Oregon State University.

Nutrient and water quality values for each lake were determined by averaging all samples collected for each lake from 1989 to 1992. Average pH was calculated from hydrogen ion activity.

Table 2. Analytical procedures used by the Cooperative Chemistry Analytical Laboratory, Oregon State University. (Cameron Jones, personal communication)

Vegetation

Lake watershed class was identified by the predominate watershed vegetation, emphasizing nearlake vegetation. In the absence of sufficient data on climate, soil maturity, and hydrologic regimen, vegetation was used as a proxy for these factors at the watershed level. Vegetation was considered a good integrator of these components due to the long time frames for development and persistence of the vegetative communities (Aber and Melillo, 1991; Franklin et al., 1988).

A vegetation map, developed from LANDSAT multispectral scanner images and other data (slope, elevation, aspect) on a Geographic Information System (Agee and Pickford, 1985), was used to identify NOCA lake watershed cover types. Watersheds were identified as predominately alpine, subalpine, or forest. The forest zone was further subdivided into high and low elevation based on location above (maintains winter snow cover till spring) or below (intermittent snow free ground during the winter) permanent winter snowpack.

Predominant plant community types and associations were usually assessed in the field to verify that each lake had been placed into the correct vegetation zone. Vegetation cover of unsurveyed lakes was crosschecked with aerial photographs. Elevation ranges for each vegetation zone were determined from Agee and Pickford (1985) by grouping vegetation community types for a particular zone and recording the maximum and minimum mean ranges for the communities in each zone. The lakes were identified as west (westslope) or east (eastslope) of the hydrologic crest. Westslope lakes drain into Puget Sound and the Fraser River while eastslope lakes drain into the Columbia River.

Statistics

Kruskal Wallace/Mann Whitney-U tests (SYSTAT ver. 5.01) was used to test for statistically significant differences between various lake classes. Statistical significance was set at = 0.05. Level of significance for multiple comparisons with the same data was determined by dividing the chosen significance level ( = 0.05) by the number of comparisons to arrive at a conservative P value. Regression and covariance analyses (SYSTAT ver. 5.01) were used to determine the relationships of epilimnion temperature and ice-out date to vegetation class and elevation.

The diversity of lakes by origin within a vegetation zone Were determined using the AID1 program (Overton et al., 1987). Analysis of lake chemistry data was performed using ClusB4 (Overton et al., 1987), a non hierarchical clustering algorithm. Discriminate and correlation analyses (SPSS statistical package) were used on these cluster groups for graphical display and axis interpretations. Fishers Exact Test ( = 0.05) was used to compare lakes differing in maximum depth and near-surface and near-bottom temperatures.

Lake Classification

Three hierarchical levels were developed to classify NOCA lakes. These levels, listed in descending order within the classification hierarchy, were: (1) position relative to the hydrologic crest of the North Cascade Mountain Range, (2) vegetation zone, and (3) basin origin.

Hydrologic Crest Position

All lakes were classified either as westslope or eastslope (Figure 2) based on locations relative to the hydrologic crest of the North Cascade Mountains. Major patterns of air masses, which seasonally influenced air temperature, amount and form of precipitation, aspect of major watersheds, and geology, differed with crest position. Watersheds on the westslope of the crest were associated with the Skagit, Chilliwack, and Nooksack rivers. Eastslope watersheds were associated with the Stehekin river. Annual temperature extremes were greater on the eastslope than on the westslope because skies on the eastslope tended to be clearer than skies on the westslope. This resulted in hotter summers and colder winters eastslope compared to westslope (Jackson and Kimerling, 1993). The southern aspect of the Stehekin watershed (Figure 2) resulted in high solar insolation in the headwater areas where most eastslope lakes were located. The Nooksack and Chilliwack watersheds on the westslope were generally oriented in a north-facing position, whereas the large and complex Skagit watershed encompassed a variety of aspects. Westslope watersheds were geologically more complex than eastslope watersheds. Gneiss was the predominate bedrock of eastslope watersheds, with only a few lakes found on a granitic pluton. Westslope lithology was dominated by gneiss, granite, sedimentary deposits, and volcanic deposits.

Vegetation Zone

The second level of the classification was based on vegetation. The alpine zone was dominated by exposed rock and ice (Table 3). Dominant vegetation was clumped, low in stature, and included sedges (Carex sp.), stonecrop (Sedum sp.), and partridgefoot (Leutkia sp.). The alpine zone occurred primarily in westslope watersheds where the climate was conducive to the persistence of glaciers and alpine conditions. The semi-arid climate of eastslope watersheds generally was too dry and warm for glacial and alpine conditions to persist.

The subalpine zone was intermediate to the alpine and forest zones (Table 3). Dominant vegetation included herbs, woody shrubs, and some trees aggregated in open forests. Herbs included partridgefoot (Leutkia sp.), a variety of ferns and mosses, sedges (Carex sp.), and hellebores (Veratrumsp.). Common woody shrubs were heather (Phyllodoce sp.) and huckleberry ( Vaccinium sp.). Westslope trees included Mountain hemlock (Tsuga mertensiana), Pacific silver fir (Abies amablis), and Subalpine fir (Abies lasiocarpa). Eastslope trees included Mountain hemlock, an occasional Alaska yellow cedar (Chamaecyparis nootkatensis), Subalpine fir and, at the higher elevations in the zone, Whitebark pine (Pinus albicaulis) and Subalpine larch (Larix lyalli) (Table 3).

The forest zone on both the east and westslope was subdivided according to elevation. High-forest zones experienced permanent winter snowpack, whereas low- forest zones did not. High-forest zones were dominated by Silver fir on the westslope and by Mountain hemlock and Subalpine fir on the eastslope. The low-elevation forests included Ponderosa pine (Pinus ponderosa) and Douglas fir (Pseudotsuga menziesii) on the eastslope, and Douglas fir and Western hemlock (Tsuga heterophylla) on the westslope. Understory vegetation was sparser on the eastslope than on the westslope.

Climatic differences between the eastslope and westslope influenced the occurrence of the vegetation zones relative to elevation. All three vegetation zones on the eastslope occurred at higher average elevations than their counterparts on the westslope (Table 3). The alpine zone was restricted on the eastslope, but occurred in a significant portion of westslope watersheds. Subalpine and high-forest zones were predominant on both the eastslope and westslope. Low-forest zones in NOCA were most common westslope and rare eastslope (Table 3).

Table 3. Vegetation zones for lake basins with corresponding cover types (Agee & Pickford 1985) and cover type mean elevations for open and closed forest, lake elevations (min & max), habitat notes and NOCA lakes. Sampled lakes in bold type. Lake acronyms given in Appendix I.

Vegetation Zones	Major Cover Types found in lake basins	Mean Elevation Vegetation (m) (Lakes)	Habitat Notes	Lakes by originab
EASTSLOPE
Forest Low	Ponderosa pine, Douglas fir	880 (662)	Very dry, low elevation. Douglas fir to mid elevation. Disturbance oriented (fire, avalanche) communities.	B-COON
Forest High	Subalpine fir, Mountain hemlock, subalpine herb, heather	1127 1789 (1504 1717)	Subalpine fir found in more xeric areas than Mountain Hemlock. Forests communities tend to be more xeric tolerant than westside counterparts.	C-BATT MCAL T-DAGG, KETT, KETU, RAIN, WADD
Subalpine	Subalpine fir, Whitebark pine/larch, Subalpine herb, Rock, Snow	1558 2231 (1270 2072)	Mesic aspects east. Larch found at higher colder areas than pine. Summer warmth has allowed subalpine veg. types to extend to the ridgetops in most basins.	C-DOUB, MA3, MLy1, MM7, MR1, MR9, MR11, MR13-1, MR15-1, MR15-2, TRIL, T-GNVW, TRAP, I-JUAN, ML6, MM11, MR2, MR3, MR8, MR12, D-MM1, MR13-2, TRIU, B-MR16
Alpine	rock, snow	(1860 2127)	High elevation xeric environment. Limited summer season, and permanent winter snowpack.	I-MA2, MM3, D-MM8
WESTSLOPE
Forest Low	Hardwood forests, High shrub, Western hemlock, Douglas fir	488 880 (412 1031)	Warm low elevation, dry to moist areas generally below permanent winter snowpack. Marked by disturbance oriented communities and Western hemlock. High shrub extends up avalanche slopes. Disturbance oriented species (Douglas fir) may be seral to Silver fir.	C-PYRA, T-HOZO*, MP5, PANL, PANU, RIDL*, THUN, WILL* *Sedimentary geology
Forest High	Pacific silver fir, Douglas fir	728 1316 (1171 1687)	>100 inches/yr rainfall. Winter snowpack does not melt off periodically during winter	C-BOUC, JEAN, LS1, LS2, MC8, NONA, THRL, I-NERTa, D-SWEEb, S-EP3, K-PMR(1-6)
Subalpine	Rock, snow, subalpine herb, subalpine fir, Mountain hemlock, silver fir, heather	1220 1789 (1110 1967)	Basins tend to be meadow filled with clumps of trees if any. Often snow persists in basin throughout the summer.	C-BEAR, COPP, DD5, EGG, EP2, EP5-1, EP6, EP9-2, EP12, FP9, FP10, LS3, LS6, LS7b, M11b, M14, M17, MC3, MC7, MC16-2, MC29b, MC34, MONO, MP2, MS3, MSH4, TAP1, THRM, T-HIDD, M4b, M8, MC14-1, MC14-2, PRICc, REVL, SKYM, WILD, I-EP9-1, EP13, FP6, M24-1, M24-2, MC1, MC2, MC15, ML3, MP1(1-3), PM2, TAP2, TAP3, TAP4, D-DD8, M21, MC28c, ML1, PM12, REDOb, REVU, VULCb, B-M1, MC10, TTAR, F-M7, U-M15, MSH2
Alpine	Rock,snow, subalpine herb, subalpine fir	(1113 2083)	Severe, lots of snow, rock and ice. Vegetation, if any, is of krummholtz form.	C-EP4, EP10, LS5, M10, M13, M18, M22, MC13, MC16-1, MC23, MP8, MP9, MS2, MSH1c, OUZLc, T-EP11-1, KLAWc, MORAc, SILV, SKYU, I-EILE, EP14, FP5, LS4, MC22, WILE, D-FP1, M9, M16, MA1b, ML5, U-EP11-2, FP8, M25, MC30, RD4

Precipitation varied among vegetation zones. Average annual precipitation decreased from the alpine to the low-forest zone both east and west of the hydrologic crest (Table 4). Differences in precipitation were not evident between alpine and subalpine zones on the westslope. Average amounts of precipitation for each vegetation zone were lower for eastslope zones than for westslope zones (Table 4). The amount of precipitation in high-forest westslope watersheds was approximately equal to precipitation in alpine and subalpine watersheds on the eastslope.

Table 4. Elevations (EL), precipitation (PREC), and date of ice-out for NOCA lakes by position east or west of the Cascade crest and vegetation zone.

Origin

Eight morphogenetic lake classes were identified based on basin origin -- cirque, trough, ice scour, moraine, bench, fault, slump, and kettle lakes (Hutchinson, 1957). Cirques are amphitheatre-shaped basins generally occurring at the heads of glaciated valleys, and thought to be formed by frost riving at the firn line (permanent snow line). Troughs lakes were formed in glacially scoured U-shaped valleys, and tended to be long, narrow and wedge-shaped with the deepest spots in the lakes near their outlets. Ice scour lakes, found in irregular depressions, were formed by glacial scour along fault or joint fracture lines. The lakes occur on ridgetops. Moraine lakes formed behind the unconsolidated till of terminal or lateral moraines which were deposited by glaciers. This damming feature may act alone to form the lake or increase the depth of an existing cirque or trough lake. Bench lakes were the result of glacial scouring working at two different time periods to create a step or bench along the hillslope parallel to the valley floor. Oblong in shape, they have no sheltering basin. Fault lakes were formed by bedrock dams created by differential displacement of bedrock along tectonic faults. Slump lakes occur in the depression left by the rotational slip of deep seated soil. Kettle lakes were the result of irregularly deposited outwash (ground/terminal moraine) or remnant pieces of ice left in the outwash of retreating glaciers, and tended to have irregular shapes. When the ice melted, a lake was left in the resulting depression.

Applications of the Lake Classification

The lake classification was used to assess the following characteristics of NOCA lakes: distribution of the lake classes by basin origin, time of lake ice-out, epilimnetic water temperature, and water chemistry. Also considered were influences of lake depth on water temperature and influences of geology, glacial outwash, lake depth, and a measure of the integrated influence of watershed characteristics on water chemistry.

Basin Origin

Each morphogenetic class exhibited differences in landscape position, relative persistence, relative depth, presence of a permanent inlet stream, relief of the watershed basin, and several other morphogenic features (Table 5). Lakes situated in erosional landforms such as cirques and troughs tended to be deeper and more persistent than lakes situated in depositional landforms, such as moraines and slumps. Though there is a wide range of depths associated with troughs and cirques, their maximum depths were greater than other lake types. Slump lakes and kettle lakes, which were small, shallow depressions, may be shorter lived than the other lake classes. The presence of permanent inlet streams was related to lake position within a watershed. Low-elevation lakes in valleys, such as trough lakes, had the greatest occurrence of permanent inlet streams. High-elevation cirques and ice-scour lakes tended to lack inlet streams. Slump and kettle lakes also tended to lack permanent inlet streams.

Table 5. Morphogenetic lake classes listed with general descriptors associated with each class. Temporal frame of reference and depth for North Cascades lakes (Maximum age to approx. 7000 yrs.)

Lake class	Location	Relative Persistence	Relative deptha (m)	Permanent inlet stream	Basin relief	Morphogenetic feature
Cirque	headwall	long	shallow to deep (2-46.3)	generally no if headwall, yes if lower end of patemoster lakes	high	scoured depression at headwall with rock lip or moraine damming feature
Trough	valley	long	shallow to deep (3-137)	generally yes	low	glacially deepened valley depression with bedrock dam with occasional moraine influence
Ice Scour	ridgetop	long	shallow (1.2-8.2)	no	low	irregular depression created by glacial scour along fault or joint fracture
Moraine	valley, cirque	short to med	shallow to medium (4.3-27.4)	yes	variable	valley or cirque lake formed principally by terminal or lateral moraine dam
Bench	along valley side wall, a false ridge parallel to valley floor	medium to long	shallow to medium (3.6-5.8)	site specific	med to high	differencial valley glacier erosion along valley slope producing bench. Interaction with ancillary glacier
Fault	variable	variable	shallow to deep (11)	site specific	variable	tectonic displacement creating a bedrock dam with associated basin
Slump	on slope with deep soils, valley sidewall	short	shallow (no data)	no	medium	soil mass movement, rotational slip of deep seated soil
Kettle	ground moraine outwash	short	shallow (1-4)	no	low	stranded ice retreating glacier leaves depression in ground moraine after melting

^a Maximum depth relative to other classes of lakes by origin. Depths are for sampled lakes.

Morphogenetic classes were unequally distributed across and within vegetation zones (Table 6). Trough, cirque, and ice-scour lakes occurred in all zones. The forest zone (high and low zones combined) contained the greatest diversity (HE) of morphogenetic classes (HE=1.53), although this zone contained only 20% of all lakes. The alpine zone had the lowest diversity of morphogenetic classes (HE=1.30) and contained lakes situated primarily in erosional landforms. Subalpine systems contained an intermediate diversity of morphogenetic classes (HE=1.41). The majority (58%) of the lakes in NOCA occur in this vegetation zone (Table 6). The occurrence of particular morphogenetic classes differed among vegetation zones (Table 6). The proportion of trough lakes increased from alpine to forest zones (15% in alpine, 13% in subalpine, and 33% in forest). Ice-scour and dammed lakes were less prevalent in forest than in subalpine and alpine forest zones. Cirque lakes were a major component of all zones, and as previously mentioned, slump and kettle lakes were present only in the forest zones.

Table 6. Distribution of morphogenetic classes of NOCA lakes by vegetation zone. Number in parenthesis.

Time of Ice-Out

Time of lake ice-out, measured in Julian days, increased from low-forest lakes to alpine lakes based on an analysis of pooled data for eastslope and westslope lakes (Table 4). Eastslope lakes typically iced-out somewhat earlier than westslope lakes in a given vegetation zone. Date of ice-out also was influenced by basin aspect. For example, west-facing eastslope subalpine lakes tended to ice-out earlier than lakes with an easterly aspect (Figure 3). The pattern was not as clear in westslope subalpine lakes (Figure 3). EP6 and Bear are deep lakes. SKYMO (SKYM) receives little precipitation, and like Talus Tam (TTAR), is found in a rocky open basin which may have caused them to ice-out earlier than lakes in more sheltered locations.

Figure 3. Relationship between estimate date of ice-out (Julian days), elevation (m), and basin aspect of eastslope and westslope subalpine lakes in 1989.

Epilimnetic Water Temperature

Epilimnetic water temperature was warmest in low-forest lakes and coolest in alpine lakes based on measurements during the ice-free season in 1989 (Figure 4). The water temperature of subalpine and high-forest lakes was very similar, and intermediate to the temperature of alpine and low-forest lakes. Furthermore, water temperature of low-forest lakes was warmest in July, whereas the water temperature of high-forest, subalpine, and alpine lakes was warmest in September. Similar patterns were observed in 1990, even though a hot, dry August elevated water temperatures of subalpine, high-forest and low-forest lakes by approximately 4°C relative to August 1989 (data not shown). A comparison of regressions of water temperature against elevation for forest (high and low combined) and alpine zones revealed significant differences between these vegetation classes (Figure 5, ANCOVA P=0.002). This comparison suggests that local climate differed between these two zones. Water temperature varied inversely with elevation within forest and alpine vegetation zones based on an analysis of 1989 data (forest, r²=0.45, p=0.002; alpine r=0.51, p=0.046; Figure 5).

Figure 4. Relationship between epilimnion temperature (°C) and month for westslope vegetation zones. Each symbol represents the average temperature of all lakes sampled in a particular vegetation zone each month.

Figure 5. Relationship between epilimnion temperature (°C) and elevation (m) for forest and alpine zones.

Water Chemistry

Classification of lakes using crest position and vegetation zones did not clearly order the lakes relative to chemical characteristics. With the exception of westslope low-forest lakes, there were few significant differences in concentrations of nutrients (total Kjeldahl-nitrogen, total phosphorus, orthophosphorus, nitrate-nitrogen, and ammonia- nitrogen), pH, alkalinity, conductivity, and concentration of cations (sodium, potassium, calcium, and magnesium) between eastslope and westslope lakes and lakes in different between vegetation zones (Table 7). Westslope low-forest lakes differed significantly (Kruskall-Wallis, P<0.008) from all other westslope vegetation zones in pH, alkalinity, conductivity, cations (Na, K, Mg, Ca) and total Kjeldahl-nitrogen (except high forest). Although there were few significant differences among westlsope vegetation zones, pH, alkalinity, conductivity, concentrations of total Kjeldahl-nitrogen, potassium, calcium, and magnesium tended to decrease from low-forest to alpine westslope lakes, and concentration of nitrate-nitrogen tended to increase (ALP > WLF; Kruskall-Wallis, P<0.008). Total phosphorus was significantly higher in concentration in alpine than low forest lakes (Kruskall-Wallis, P<0.008). Westslope low-forest lakes also differed significantly (Kruskall-Wallis, P<0.008) from eastslope subalpine lakes in pH, conductivity, alkalinity, and cations (except Mg). Other significant differences between eastslope and westslope lakes were few (TP, ESA > WSA; Na, EHF > WALP, WLF> EH F) .

Table 7. Means and standard deviations (in parenthesis) for water chemistry variables 1989-1992. LF=Low Forest; HF=High Forest; SA=Subalpine; ALP=Alpine.

The lakes exhibited little seasonal and annual variation in their chemical characteristics based on a cluster analysis of all samples collected from 1989 to 1992. A five-cluster structure was selected. Cluster 5 was dominated by alpine and subalpine lakes receiving glacial outwash (Table 8). Subalpine lakes dominated cluster 1, and subalpine and high-forest lakes dominated cluster 4. Clusters 2 and 3 contained only low-forest westslope lakes. A discriminate analysis was performed on the clusters to graphically determine their relative similarity in ordination space (Figure 6). Axis one was correlated with alkalinity (r²=0.903), pH (r²=0.760), conductivity (r²=0.894), total Kjeldahl nitrogen (r²=0.437), Ca (r²=0.948), Na (r²=0.903), K (r²=0.790), and Mg (r²=0.363). Axis two was correlated weakly with potassium (r²=0.482). Axis three was negatively correlated with total phosphorus (r²= -0.632) and ortho-phosphorus (r²= -0.882). Thus, westslope low-forest groups (2 and 3) were distinguished from other groups (1,4, and 5) along axis one by tending to have higher alkalinity, pH, conductivity, TKN, and cations. Group 5, alpine and subalpine glacially influenced lakes, was distinguished by having higher levels of phosphorus than other groups (Table 9). Groups 1 and 4 contained the majority of the lakes, although the lakes in group 4 tended to be lower in elevation than the lakes in group 1.

Table 8. Summary of lake groups developed using cluster analysis, based on water chemistry variables, 1989-1992. Each lake sample was considered an individual point. Two lakes with multiple samples had a single sample in a second group. Lakes occurring in more than one group were placed in the group containing the majority of samples for that lake. Lake acronyms given in Appendix I.

Groups 2 and 3, both containing westslope low-forest lakes, were differentiated by geology. Group 2 contained lakes found in gneiss, a metamorphosed crystalline bedrock. Lakes in group 3 (HOZO, RIDL, and WILL) had the highest alkalinity, conductivity and pH, and highest concentrations of TKN, Mg, and Ca. These lakes occurred in a metamorphosed interbedded sedimentary and volcanic bedrock termed green stone.

Lakes of various depths occurred within each vegetation zone, and lake depth influenced water chemistry. For example, concentrations of total Kjeldahl-nitrogen and total phosphorus tended to decrease with increasing lake depth (Figure 7). A wide variety of watershed characteristics such as glacial influence (note KLAW, PRIC and OUZL) also probably influenced ambient chemical concentrations in the lakes, further adding to the variability observed within each vegetation zone. Schindler's ratio (A₀+ A0_d/V, where A₀= watershed area, A^{d= lake surface area, V= lake volume) indicates the potential flushing ratio for each lake (Schindler 1971). Total concentration of Kjeldahl- nitrogen appeared to increase and then decrease with an increase in Schindler's ratio for each vegetation class (Figure 8). Hand-drawn curves suggested that for a given ratio, total Kjeldahl-nitrogen decreased from low forest to alpine zones.}

NOCA lakes exhibited a relatively narrow range of water qualities and nutrient concentrations. The most productive lakes were marginally mesotrophic, whereas the majority of lakes were either ultraoligotrophic or oligotrophic based on concentrations of nitrogen and phosphorus (Likens, 1975). Hydrologic crest position, vegetation, and lake morphometry influenced the physical and chemical characteristics of NOCA lakes. Individual vegetation zones were located at higher elevations east of the hydrologic crest than west owing to the sharp climate gradient. In fact, the climatic differences essentially excluded the alpine zone on the eastside of the hydrologic crest. Vegetation state (alpine to low forest) was associated with increased epilimnetic water temperature and earlier dates of ice-out. For a given vegetation zone, lakes east of the hydrologic crest iced-out earlier than lakes west of the crest. Furthermore, lakes in west-facing watersheds iced-out earlier than lakes in east-facing watersheds in some vegetation zones. Lakes with late ice-out dates (i.e. alpine lakes) were generally colder throughout the ice-free season than lakes that iced-out earlier in the season.

Vegetation zones did not clearly order many water chemistry parameters, although there was a tendency for pH, alkalinity, and concentrations of TKN and certain cations to decrease from the low forest to the alpine zone on the westslope. The low forest zone had the latest vegetation successional state, greatest potential for water and soil interactions, and had been deglaciated for the longest period of time. However, concentrations of total phosphorus were high in alpine lakes receiving glacial outwash, comparable to those of low-forest lakes. Phosphorus concentrations in subalpine lakes were low relative to those in alpine lakes because glacial influences were much reduced. Sedimentary rock elevated the water quality (pH, alkalinity, conductivity) of several low-forest lakes relative to those lakes lacking this rock type in their watersheds.

Lake origin was useful in identifying potential differences in longterm persistence, basin location, and the diversity of lakes parkwide. Some morphogenetic lake types were found in all vegetation zones (e.g., ice scour, cirque, trough, and dammed lakes) whereas other were restricted to specific Vegetation zones. Lake basins formed by erosional processes may be more persistent than lakes created by depositional landforms (Table 5). Amount of infilling is a function of both morphogenetic origin and lake age. Older low-forest trough lakes no longer cover entire valley floors as young alpine trough lakes often do, and they tend to be shallower than alpine trough lakes. Relatively shorter-lived slump lakes originate from a rotational slip in deep seated soils, and only occur in the forest zone in NOCA, the zone which has the greatest soil maturity and soil depth. Kettle lakes occur in one park drainage though they are generally restricted to ground moraines deposited across the continental plains.

Limnologists have developed several types of lake classification systems during the last 60 years. Much of the research has addressed physical attributes, morphometrics, nutrients or combinations of these variables on very diverse groups of lakes. Larkin and Northcote (1958) sampled a group of British Columbia lakes with a wide range of mean depths and determined that conductivity decreased with increasing mean depth. The range of mean depths among NOCA lakes was narrow in relation to the lakes studied by Larkin and Northcote, and the relationship between mean depth and conductivity was weak and not significant (r²= -0.123). George and Maitland (1984) used lake basin morphometric characteristics as the basis of classifying lochs in Scotland which, unlike NOCA, differ little in climatic extremes. Hutchinson (1957) described 76 types of lakes by origin. Relatively few of these types occur in NOCA. Mean depth was found to be a good indicator of fish production in Canadian lakes much larger than those existing at NOCA (Rawson 1955). Pennak (1958) used elevation to classify lakes in the Colorado Rockies that were similar with respect to basin geology, origin, and aspect. Lewis (1983) used temperature and mixing to classify lakes in different climates worldwide. Based on this classification, NOCA lakes would be grouped into a single class (dimictic), although some of the lakes may be amictic in some years.

Some lake classification systems have focused primarily on rates of nutrient loading. Vollenwieder (1968) determined that total phosphorus loading rate was an indicator of the trophic status of lakes. Almost without exception, NOCA lakes are oligotrophic or ultraoligotrophic. The range of total phosphorus concentrations in NOCA lakes was small in comparison to the range used by Vollenweider. Schindler (1971) determined that concentrations of nitrogen and phosphorus in lakes in the Experimental Lakes Area (ELA) were associated with the ratio of [lake and watershed area /lake volume]. Unlike NOCA, the ELA area is of relatively uniform topography, vegetation, and geology. Schindler found linear relationships between concentration of nutrients and a relatively limited range of watershed-to-volume ratios (0 to 7). At NOCA the range of ratios was much larger than for the ELA lakes and the relationships appeared to be curvilinear. These results suggest that after watershed-to-volume ratios exceed about seven, concentrations of Kjeldahl N decrease, perhaps due to increased flushing rate.

Biological approaches to lake classification have existed for most of the century. Thieneman (1909) grouped a diverse range of lakes by the presence or absence of specific chironomid taxa. NOCA lakes include a very narrow range of these lake types. Tonn and Magnusson (1982) classified lakes based on the number of fish taxa present. Fish are not indigenous to NOCA high mountain lakes. Round (1958) and Nygaard (1949) used phytoplankton quotients to classify lakes. However, the taxa used are not important in NOCA phytoplankton communities.

A number of lake classifications have combined chemical and physical approaches. The Morphoedaphic Index (MEl) developed by Ryder (1974) combines mean depth and total dissolved solids (TDS) to classify lakes for potential fish production. The index was originally intended for lakes greater than 260 ha in surface area and less than 600 m in elevation with low flushing rates and inorganic turbidity and within an area with relatively homogenous environmental conditions. Climatic and limnological characteristics of NOCA lakes violate these restrictions. The Trophic State Index (TSI) developed by Carlson (1977) was developed using a set of lakes with a very wide range of Secchi disk transparencies and concentrations of chlorophyll and total phosphorus. Chlorophyll was used as an index of algal biomass. The TSI is not appropriate for glacial systems or lakes of extreme clarity. Algal biomass is very low in NOCA lakes (see Phytoplankton section) and the water is extremely clear (Secchi disk visibility extends to the bottom in many lakes). Thus the TSI is of little value for classifying NOCA lakes.