SITE DESCRIPTION

The upper Green Lakes Valley is an east-facing glacial valley, headed on the Continental Divide in the Colorado Front Range (40°03′N, 105°35′W). Named for a series of shallow paternoster lakes, the Green Lakes Valley is the headwaters of North Boulder Creek and lies within the City of Boulder Watershed with no public access. The upper valley is approximately 225 ha in area, and the elevation ranges from 4084 m at the Continental Divide to 3515 m at the outlet of Green Lake 4 (GL4; Fig. 1A). The floor of Green Lakes Valley has the stepped form of a glaciated mountain valley. At the foot of the north-facing Kiowa Peak is the Green Lake 5 rock glacier (RG5), a lobate rock glacier approximately 8 ha in area and at an elevation of 4000 m, where extensive research on hydrochemistry has been conducted (Williams et al., 2006, 2007). Niwot Ridge, the northern boundary of the Green Lakes Valley, is the site of other experimental areas, including an Aerometrics wet-chemistry precipitation collector at the Saddle site (Fig. 1A), operated by NWT-LTER as part of the National Atmospheric Deposition Program (NADP).

Figure 1

(A) Study location at the Niwot Ridge Long-Term Ecological Research (NWT-LTER) site. Contour interval is 100 m. (B) Bathymetric map of Green Lake 4. Monitoring samples were taken from the point marked with a black star (40.05°N, 105.64°W).

The continental, high mountain climate of Green Lakes Valley has been recorded continuously at the D-1 meteorological station on Niwot Ridge (Fig. 1A) for over 50 years and for shorter periods on the valley floor. At an elevation of 3750 m, the D-1 station is the highest continuously operating weather station in North America. The climate is characterized by long, cool winters and a short growing season (Williams et al., 2002). Mean annual temperature at D-1 is −3.7 °C (Williams et al., 1996). Almost 80% of the approximately 1000 mm of recorded annual precipitation falls as snow (Caine, 1996). The bulk snowpack temperature remains below 0 °C until late spring, causing a lag in the hydrological cycle by concentrating the release of meltwater in a short, intense period of runoff (Caine, 1996).

GL4 is an oligotrophic alpine lake with low summer chlorophyll a concentrations and dilute surface waters (NWT-LTER database:  http://culter.colorado.edu). The average depth is 4.0 m, while the maximum depth is 13.1 m. GL4 has a surface area of 5.34 ha. The lake has become thermally stratified in mid-summer since limnological monitoring began in 2000. In general the lake mixes soon after ice-out and then becomes more stratified until the end of July when the differences in temperature between the epilimnion and hypolimnion decrease until fall mixing occurs in early September. The lake is generally ice free from the beginning of July until mid-late September, though there are large year-to-year variations. Historically, the lake was fishless, but due to accidental stocking in 1998, Yellowstone cutthroat trout now inhabit both GL4 and the upstream Green Lake 5.

FIELD SAMPLING AND MEASUREMENTS

Discharge from GL4 has been measured continuously since 1983 from about 1 May to 30 October (Williams and Caine, 2001). Hydraulic residence times were calculated by dividing the volume of the lake by the discharge.

The GL4 limnological monitoring program began in 2000; in-lake physical, chemical, and biological parameters were collected according to protocols in Gardner et al. (2008) during the ice-free period of the summer months of June, July, and August, 2000–2005. The first sampling date each summer occurred shortly after the ice had melted on the lake.

Using an inflatable raft to reach the deepest part of the lake, grab samples were collected at the surface and at depths of three and nine meters (Fig. 1B) with a Van Dorn sampler. The surface and 3 m sampling depths are representative of the epilimnion in terms of water quality. The surface samples may also reflect any effects of photoinhibition on the phytoplankton community. The 9 m depth is representative of the water quality and phytoplankton community in the hypolimnion. The hypolimnion accounts for a small portion of the lake volume, but is in closer contact with lake sediments. Subsamples were taken for algal biomass, phytoplankton community composition, and water quality analysis. Samples were stored in polyethylene bottles that were rinsed three times with sample water at the time of collection. Water samples were transported the same day as collection to the laboratory.

Temperature, dissolved oxygen, and solar irradiance measurements were measured from the raft at each depth using a YSI dissolved oxygen meter and a Li-Cor datalogger. Secchi depth readings were taken all years except 2004.

PHYSICAL AND CHEMICAL PARAMETERS

Water samples were analyzed for pH, acid neutralizing capacity (ANC), conductance, major ions, reactive silicate (Si), and dissolved organic material, following the protocols presented in Williams et al. (1996). ANC and pH were measured immediately on return to the laboratory using the Gran titration technique. Subsamples for major ions and nutrients were immediately filtered through pre-rinsed (300 mL), 47 mm Gelman A/E glass-fiber filters with 1.2 µm pore size. DOC samples were also filtered in the laboratory on the same day as collection using pre-combusted 47 mm Whatman glass-fiber filters (pore size 1.2 µm) and a hand-pump filtration system. Filtered samples were stored in the dark at 4 °C and analyzed within 1–4 weeks. Anions were measured by ion chromatography (Dionex DX 500) with chemical ion suppression and conductivity detection. Calcium, magnesium, sodium, and potassium were analyzed with a Varian AA6 atomic absorption spectrophotometer. Total phosphorus (TP) and total dissolved phosphorus (TDP) were measured on a Lachat QuikChem 8000, where dissolved organic phosphorus (DOP) is equal to TDP minus inorganic phosphorus (IP), and IP was measured as orthophosphate, PO₄³⁻. Analytical precision for inorganic solutes and nutrients was less than 2% and detection limits less than 1 µeq L⁻¹. Dissolved organic carbon (DOC) concentrations were determined by high-temperature catalytic oxidation using a Shimadzu Total Organic Carbon Analyzer. Three replicate analyses were done for each sample. Standard deviation was typically 0.06 mg C L⁻¹ with a range of 0.01–0.22 mg C L⁻¹.

Chlorophyll a samples were filtered within 12 hours through Whatman glass fiber filters (pore size 1.2 µm) with a hand-pump filtration system, and filters were frozen in aluminum foil until processing. The filters were extracted with hot ethanol, and chlorophyll a was quantified spectroscopically as described by Marker et al. (1980) and Nusch (1980).

Table 1 lists GL4 environmental characteristics and associated abbreviations used in later analyses.

Table 1

A subset of Green Lake 4 environmental variables, and corresponding abbreviations, used in analyses.

PHYTOPLANKTON ENUMERATION

Samples taken for phytoplankton community composition analysis were preserved with Lugol's solution (1%) and stored at room temperature. A 50-, 40-, 20-, 10- or 5-mL subsample was settled in Utermöhl or Hydro-Bios settling chambers overnight and identified at 1000× with an inverted microscope. A minimum of 400 cells were identified and counted in each sample.

Consistency in identification of algal taxa by the four phycologists involved in this study was maintained by consultation and use of photomicrographs. To minimize potential identification bias, taxa with very similar morphotypes, and hence greater difficulty to resolve, were considered as combined taxonomic groups in the statistical analysis. Table 2 includes a species list of algae and taxonomic groups found in Green Lake 4 samples, and acronyms used to represent taxa in subsequent analyses.

Table 2

Species list of algae found in Green Lake 4 samples. Corresponding symbols next to species indicate those species aggregated together into taxonomic groups for the ordination analyses. Abbreviations indicated represent taxa in statistical analyses.

Algal biomass was calculated based on lake-specific algal biovolumes, a technique that evaluates biomass and expresses abundance of each species (Wetzel, 2001). Mean dimensions of the cells (ca. five measurements per taxa) determined average cell volume (µm³) by corresponding geometric equations (Tikkanen, 1986). Several taxa were analyzed individually to illustrate the temporal patterns seen in the phytoplankton community, but all of the counts were utilized to calculate the total biomass by division or phyla.

PHYSICAL AND CHEMICAL CONDITIONS OF GREEN LAKE 4

During the six-year study, precipitation was at or below average throughout 2000–2002; which corresponded to the last three years of the five year regional drought. Precipitation was above average throughout 2003–2005 (Fig. 2A). In 2002, precipitation was the lowest at 860 mm (Fig. 2A) (Williams et al., 2006). Since 1951, the mean summer temperature has averaged 6.8 °C (Fig. 2B). The mean summer temperature was higher during the first three years and lower during the final years of the study, with 2002 the warmest summer at 9.3 °C (Fig. 2B). Thus, average or below average precipitation and above average temperatures typified the years of 2000 through 2002. Further, the extreme drought conditions which occurred in 2002 in the Green Lakes Valley were consistent with the extreme drought conditions prevalent throughout the State of Colorado that year.

Figure 2

Broad-scale climatic factors and historical average values as measured at NWT-LTER D-1 alpine station. Lines indicate the historical average from 1953 to 1996. (A) Total Water Year Precipitation (Oct–Sept); B) Mean summer (Jun–Jul–Aug) temperatures.

The environmental conditions of Green Lake 4 reflected the climatic changes over the study period, especially during the 2002 drought (Table 3). The annual discharge of 1,200,000 m³ during the 2002 drought was 20% lower than the next lowest discharge and only 65% of the historical average from 1983 to 2005. The average summer hydraulic residence time (HRT) in 2002, 21.5 days, was almost 40% longer than the historical average HRT. The date of ice-out in 2002, June 8, was about one month earlier than the historical average. Further, the length of the ice-free season of 132 days in 2002 was the longest over the 6-year study period.

Table 3

A summary of Green Lake 4 physical and chemical characteristics as measured during the summer monitoring period at the outlet to the lake, between 24 May and 31 August 2000–2005. Differences among years for each variable were tested with a one-way ANOVA followed by the Student-Newman Keuls post-hoc test: those differences significant at p < 0.05 are indicated with lower case letters, where statistically distinguishable means between years are indicated by varying letters, proceeding from highest mean value(s) (i.e., a) to lowest mean value(s) (e.g., b, c, d, e, or f). (nd =  no data; ^*number of days between ice out and ice cover; ^§sum of discharge gage measurements at the GL4 outlet, 24 May–31 August; ^¶computed as discharge divided by GL4 volume, 24 May–31 August).

Concomitant with these drought related hydrologic extremes, the mean summer water temperature was highest in 2002 at 10.9 °C. Moreover, although the lake is generally only weakly stratified, stratification was accentuated in 2002 with a maximum epilimnetic water temperature of 14 °C (at the surface) while hypolimnetic temperatures remained at about 7 °C, similar to other years (Fig. 3). The water temperature was lower during the post-drought period, 2003–2005. The pH of 6.7 and conductivity of 19.3 µS cm⁻¹ were significantly higher in 2002 compared to other years. In contrast, light penetration, dissolved oxygen, and percent saturation of dissolved oxygen remained relatively constant throughout the six-year period. DOC was lowest in 2000–2001, averaging 0.6 mg L⁻¹, and was highest in 2002, 2004, and 2005, averaging 1.0 mg L⁻¹.

Figure 3

Green Lake 4 temperature depth profile circa 20 July, peak growing season, 2000–2005.

In all years, the concentrations of most major cations and anions gradually increased following the loss of ice-cover. For example, there was a significant negative linear relationship between Ca²⁺ concentrations and discharge (p < 0.05), with an overall range for Ca²⁺ concentrations from 62 to 122 µeq L⁻¹. In contrast, DOC concentration gradually decreased during the summer. The changes in major ion concentrations and DOC during the summer period were greater than the variation in concentration with depth on a given sampling date (NWT-LTER database:  http://www.culter.colorado.edu). Therefore, to examine interannual variation, we compared the average concentrations of major cations, anions, and nutrients for the six-week summer period subsequent to ice-out in GL4 (Fig. 4). The average concentrations of major cations were high in the summers of 2002 and 2004 compared to the other years, and the concentrations of Ca²⁺ and K⁺ were at their maximum in 2002 (115 and 7 µeq L⁻¹, respectively). The anions SO₄²⁻ and Cl⁻ also exhibited highest concentrations in summer 2002 (90 and 4 µeq L⁻¹, respectively), whereas ANC concentrations generally decreased during the study. However, ANC concentrations were greater in 2002 than in 2001 or during the post-drought years. Furthermore, it should be noted that ANC concentrations peaked at the end of the summer 2000; whereas ANC concentrations peaked in mid-July in 2002.

Figure 4

All-depths averaged (Mean + 1SE) Green Lake 4 chemical characteristics, 2000–2005. Differences among years were tested with a one-way ANOVA followed by the Student-Newman Keuls post-hoc test: those differences significant at p < 0.05 are indicated with lower case letters, with “ns” indicating no significant difference among years. (Units for all parameters in µeq L⁻¹ except for Si and TDP  =  µmol L⁻¹). Acronyms for the environmental characteristics are included in Table 1.

The variations in summer nutrient concentrations followed different patterns than those for major ions. The nitrate concentrations generally decreased following ice-out (NWT-LTER database:  http://www.culter.colorado.edu). Concentrations of ammonium, phosphate, and total dissolved phosphorus (TDP) were low with no consistent pattern of variation following ice-out. The concentrations of Si, a required nutrient in the growth of diatoms, remained relatively constant throughout the summer in all years except for 2002, when a sustained decline occurred in July (Fig. 5). The silica concentrations measured during the 2002 ice-free sampling season (mid June to end of August) were the lowest measured during any ice-free sampling season over the 10 year period from 1995 to 2005. Also, the average silica concentration during the 2002 ice-free season was lower than the average over the entire 10 year period.

Figure 5

Silica concentrations measured at the outlet to Green Lake 4 during the 2002 sampling season. Also shown are the average silica concentrations for the 2002 ice-free sampling season and the average for the 10 year period from 1995 to 2005 ice-free sampling seasons.

In terms of interannual variation, average summer NO₃⁻ concentrations were highest in 2004 at 19.6 µeq L⁻¹ and lowest in 2000 at 7.1 µeq L⁻¹ (Fig. 4). Average concentrations of NH₄⁺ were highest in 2003 at 1.3 µeq L⁻¹, and there was no significant difference among the other years. Concentrations of PO₄³⁻ and TDP fluctuated in a similar manner, with higher values of PO₄³⁻ occurring in 2000, 2002, and 2005.

The PCA of the environmental variables indicated that the overall chemical composition of the lake water was unique for each year of the study (Fig. 6). The first two components explained 54% of the variance in the environmental data. ANOVA confirmed that differences between loadings of yearly averages were statistically significant for all yearly comparisons on Axis 1 and for all yearly comparisons with the exception of 2001 and 2003 on Axis 2 (p < 0.0001). The first axis was associated with the concentrations of most of the major cations and anions with a gradient from low concentrations in 2000 to high concentrations in 2002 and 2004. The variables loading most on the second PCA axis included temperature, silica, ANC, and pH as well as nitrate, which distinguished the environmental conditions in 2002 from 2004. The various forms of phosphorus were only weakly correlated with the first two PCA axes.

Figure 6

Biplot of the first two axes of a principal components analysis of the environmental variables which explained 54% of the variance. Solid symbols are the annual means. Arrows represent the correlations of the environmental variables with the two axes. Acronyms for the environmental characteristics are included in Table 1.

PHYTOPLANKTON ABUNDANCE AND VARIATION

Over the course of the study, chlorophyll a concentrations were consistently greater in the hypolimnion as compared to the epilimnion (Flanagan, 2007). This was especially true during the summers of 2002 and 2005. In general, chlorophyll a concentrations increased in mid-late July (epilimnion maximum concentrations ∼2–4 µg L⁻¹; hypolimnion maximum concentrations ∼4–10 µg L⁻¹) and decreased by the end of the summer sampling season. In general, the changes in phytoplankton species composition throughout the summers were much greater than the variation with depth, despite the greater hypolimnetic chlorophyll a concentrations (Flanagan, 2007).

The average algal biovolume increased following the extreme regional-scale drought in 2002 (Fig. 7). A one-way ANOVA followed by the Student-Newman Keuls post-hoc test indicates that the post-drought years of 2003–2005 had significantly greater algal biovolume at α  =  0.05 than the regional-scale drought years of 2000–2002. This increase in biovolume mainly reflects the increase in abundance of taxa in the chrysophyte, chlorophyte, and haptophyte divisions. Calculations of Shannon-Wiener Diversity Index (H′) for each year indicate that phytoplankton diversity was significantly higher in 2001 and 2002 (2.01 ± 0.08 and 2.04 ± 0.10, respectively) than in 2003, but was not significantly different from H′ in 2000, 2004, or 2005 (Flanagan, 2007).

Figure 7

Total algal biomass as biovolume-weighted abundance (Mean + 1SE) for each division in each year, Green Lake 4, 2000–2005. Differences among years were tested with a one-way ANOVA followed by the Student-Newman Keuls post-hoc test: those differences significant at p < 0.05 are indicated with lower case letters.

Detailed analyses of several genera highlight the observed patterns at higher taxonomic levels (Fig. 8). Chrysococcus sp. was the major species which accounted for the post-drought increase in biovolume of Chrysophyta (Fig. 8A). The biovolume of the chlorophyte Chlorococcum sp. also was greatest in 2003 and 2004 (Fig. 8B). Another chlorophyte, Chlamydomonas sp., thrived only during 2005 (Fig. 8C). Finally, the chlorophyte Ankyra sp. was abundant only during the summer of the extreme 2002 drought (Fig. 8D). Likewise, the diatom Synedra sp. was also a dominant algal species only throughout the summer of 2002 (Fig. 8E), with biovolumes exceeding those in any other year by more than a thousandfold. The abundance of Plagioselmis sp., a cryptomonad, remained relatively consistent throughout the study (Fig. 8F).

Figure 8

Average biovolume-weighted abundance (Mean + 1SE) of selected Green Lake 4 phytoplankton species, 2000–2005. Differences among years for each taxon were tested with a one-way ANOVA followed by the Student-Newman Keuls post-hoc test: those differences significant at p < 0.05 are indicated with lower case letters.

A principal components analysis of biovolume-weighted abundance data for all sampling depths and dates was performed to examine the relationships among algal species. In the PCA biplot of the GL4 phytoplankton data (Fig. 9), Axes 1 and 2 together accounted for 39% of the variance in community composition. Results of the ANOVA tests on the PCA scores indicate significant interannual differences in the phytoplankton community composition on the first two axes (p < 0.0001). Axis 1 clearly separated the regional scale drought years (2000–2002) from the post-drought years (2003–2005). The 2002 algal assemblage was associated with high abundance of Synedra sp. and Ankyra sp. In general, the drought years were associated with low abundance of Chrysococcus sp. and Chlorococcum sp. The second axis distinguished the community in 2002 from 2000–2001, the first years of the regional scale drought, and separates the communities in the post-drought years. The lake stratification was greatest in 2002, but even during this year, the species composition did not significantly differ among the sampled depths based on ANOVAs of the species scores for the first 2 axes.

Figure 9

Biplot of the first two axes of a principal components analysis (PCA) of Green Lake 4 phytoplankton community composition based on biovolume-weighted abundance, 2000–2005, explaining 39% of the variance. Solid symbols represent the annual means. The arrows represent the correlations of the six indicator species with the first two PCA axes. Taxon names indicated by acronyms are given in Table 2.

The relationship between environmental variables and phytoplankton species composition was investigated using RDA. The species scores, fit for the first two RDA axes, total fraction of the variance explained, and the total variance explained are shown in Table 4. The first two RDA axes were both significant and explained 16% of the variance in the species data. The forward selection process identified six significant variables (ANC, NO₃⁻, SO₄²⁻, Cl⁻, NH₄⁺, K⁺). Based on the variance partitioning these variables could individually explain 3.3, 5.8, 2.2, 4.0, 2.4, and 2.3%, respectively, of the total variance. The correlation triplot depicts the relationships among these variables and selected phytoplankton taxa (Fig. 10). The first RDA axis explained 10.3% of the variance (λ₁  =  0.103; p  =  0.001) and the second RDA axis explained 6.1% of the variance (λ₁  =  0.061; p  =  0.001). The first axis represents a gradient from higher acid neutralizing capacity and chloride concentrations in the positive direction and higher nitrate concentrations in the negative direction. The second RDA axis is most highly correlated with sulfate and potassium concentrations. The taxa most strongly associated with the first axis were Synedra sp. and Chrysococcus sp. Synedra sp. and Ankyra sp. appear to define the 2002 assemblage, while Chrysococcus sp. is negatively associated with Axis 1 and aligns with 2003. The 2002 assemblage is identified as having almost no overlap with those from the other years. Furthermore, during the summers, when the lake was the most strongly stratified (2000, 2002, and 2004; Fig. 3), the epilimnetic samples (surface and 3 m) had significantly greater scores (p < 0.05) on Axis 2 than the hypolimnetic samples (9 m). During summers when the lake was weakly stratified, there was not a consistent epilimnetic vs. hypolimnetic trend with respect to Axis 2.

Figure 10

Correlation triplot based on redundancy analysis (RDA) depicting the influence of selected environmental variables on Green Lake 4 phytoplankton community data, 2000–2005. Axis 1 and Axis 2 cumulatively account for 16% of the variation in phytoplankton community composition. Solid arrows represent the correlations of the indicator species and dashed arrows represent the correlations of the environmental variables with the two RDA axes. Solid symbols represent the means for a given year. Names indicated by acronyms are given in Tables 1 and 2.

Table 4

Redundancy analysis (RDA) output for all phytoplankton taxa. The species scores and fit for the first two RDA axes are presented along with the total fraction of variance explained by the RDA model and the total variance.

PHYTOPLANKTON COMMUNITY RESPONSE TO EXTREME DROUGHT

This study of Green Lake 4 began in the summer of 2000, the third year of a 5-year regional drought. The final year of the drought, 2002, was an extreme event, corresponding to the lowest annual precipitation in the 110-year-long climate record for the Colorado Rocky Mountains. The physical and chemical conditions in Green Lake 4 in 2002 were strongly influenced by the drought conditions in the watershed. In 2002, the discharge from the lake outlet was the lowest in 23 years of hydrologic monitoring. The highest values for pH, conductivity, and acid neutralizing capacity also occurred in 2002, along with the highest concentrations of major cations (i.e., Ca²⁺, Mg²⁺, Na⁺, K⁺) and anions (i.e., Cl⁻, SO₄²⁻). Associated with the 2002 drought was an increase in summer air temperature of almost 1 °C. The drought conditions and the warmer summer air temperatures were amplified in the lake, causing an earlier ice-out by one month and an increase in summer surface water temperatures to 14 °C, over 3 °C greater than observed in other years of the study. The post-drought years were typified by higher precipitation and discharge, cooler water temperatures, and decreased concentrations of major ions. This decrease in major ion concentrations can likely be attributed to changes in the hydrology of the system at the catchment scale. For example, Ca²⁺ and SO₄²⁻ concentrations entering surface waters from a rock glacier in the Green Lakes Valley were approximately four times greater during the regional-scale drought years examined in this study as compared to earlier years (Williams et al., 2006).

The changes in the summer phytoplankton community during the 2002 drought were prominent, with two otherwise uncommon algal species becoming dominant. These algal species were Synedra sp. (Bacillariophyta) and Ankyra sp (Chlorophyta). The most dramatic increase in abundance was for Synedra sp. This diatom was extremely rare or undetected in the first two years of the study and in the prior study of a sediment core by Waters (1999). Of diatoms found in the sediment core of Green Lake 4, Synedra sp. composed less than 1% of the total diatom abundance at any given depth. Moreover, data from sediment traps deployed in the lake confirmed that the dominant species in the water column are reflected in those species found in the sediment core. The growth of Synedra sp. in 2002 was sufficient to deplete the concentrations of dissolved silica as the summer progressed, which was not observed in any other year. Although it has been noted that diatoms generally exhibit the greatest tolerance to warm temperatures (Reynolds, 1984), the increased temperatures were not identified in the statistical analyses as a significant driver contributing to the dominance by Synedra sp. Another distinctive feature of 2002 was the high hypolimnetic chlorophyll a concentrations, a finding consistent with the deep chlorophyll maximum observed in oligotrophic alpine lakes of the Beartooth Mountains, Montana/Wyoming (Saros et al., 2005).

The post-drought phytoplankton community was likewise distinctive. Firstly, populations of Synedra sp. and Ankyra sp. decreased to low levels in the year immediately after the 2002 drought and their abundance remained quite low throughout the post-drought period. Furthermore, a sustained increase in abundance of Chrysophyta, especially Chrysococcus sp., accounted for significantly greater algal biovolume concentrations in the post-drought period (Fig. 8). Because of the general adaptation of Chrysophyta to low temperatures (Roijackers, 1986), the cooler conditions may have contributed to the dominance of the phytoplankton by Chrysococcus sp. in the post-drought period.

These observations complement those of other researchers who have found a response in phytoplankton community composition of alpine lakes to changes in climate-related parameters. McMaster and Schindler (2005) found that the length of the ice-free season, water temperature, pH, and conductivity were all associated with interannual variability in phytoplankton communities in alpine ponds of Banff National Park, Canada. In greenhouse warming experiments in fishless alpine ponds, Strecker et al. (2004) demonstrated that warming had a significant time-dependent effect on phytoplankton community composition, in part reflecting the emergence of resting stages from pond sediments.

In these studies, as well as in this study of Green Lake 4, it is inherently difficult to separate the drought-related effects of increased hydraulic residence time, changes in water quality, and increases in water temperature as the phytoplankton community composition changes through a summer. The effect of increased hydraulic residence time may be greatest during snowmelt at the beginning of summer, when an alpine lake essentially becomes a wide place in the stream and species require rapid growth rates to persist in the lake (McKnight et. al, 1990), whereas the effects of higher water temperatures may be greatest in late summer. Moreover, phytoplankton communities may be influenced by the interactive effects of temperature and hydraulic residence time on finer scale mixing regimes in the epilimnion (Hauer et al., 1997; Lotter and Bigler, 2000; Battarbee et al., 2002; Anneville et al., 2005). Finally, the higher degree of stratification in 2002 may have indirectly influenced the species distribution by influencing variables such as light penetration and zooplankton grazing rates.

INTERPRETATION OF STATISTICAL RELATIONSHIPS

Taken as a whole, the statistical analyses of the environmental and phytoplankton species data serve to emphasize the distinct effects of the extreme drought of 2002 on the physical, chemical, and biological characteristics in Green Lake 4. The environmental conditions in the lake in 2002 were unique among the six years studied with high concentrations of major cations and anions (except for silica), as well as high temperatures. However, there is little evidence in the water chemistry of the regional drought that was occurring. Nor was there any suggestion of a major change in water quality after the 2002 drought. In addition, the concentrations of nitrogen and phosphorus exhibited relatively little significant interannual variability with the exception of the high ammonium concentrations in 2003 and high nitrate concentrations in 2004. It has been shown that reduced hydraulic residence time can act to increase internal loading of nutrients (Rippey et al., 1997). However, the fact that an increase in nutrient concentrations in Green Lake 4 did not occur in 2002 in concert with the increased residence time is likely due to the relatively low residence times in Green Lake 4 (maximum mean of 21.4 days).

The phytoplankton community was unique during the year of extreme drought. While some taxa, such as Plagioselmis sp. showed little consistent trend over the six years of the study, two interesting patterns stood out. First, the abundance of two taxa, Synedra sp. and Ankyra sp., increased by several orders of magnitude in 2002 and decreased just as quickly by the following year. However, the regional drought years (2000–2002) and the post-drought years (2003–2005) were separated along the first axis of the species PCA. This distinction was exemplified by the strong relationships of Chrysococcus sp. on this axis. In addition, the total algal biovolume increased dramatically in the post-drought years. While this record is relatively short, it appears that a large shift occurred in the phytoplankton community composition and biomass in response to the drought.

The measured environmental variables explained a significant portion of the phytoplankton species variance. Similar to the unconstrained ordination, the first RDA axis distinguished the early (2000–2002) and late (2003–2005) years of the study. The result that during summers years when the lake was stratified, the second axis of the RDA separated the epilimnetic samples from the hypolimentic samples suggests that light penetration and/or temperature may play a role in determining community composition. The taxa associated with the drought, Synedra sp. and Ankyra sp., were highly correlated with ANC, SO₄²⁻, Cl⁻, and K⁺, whereas the taxa which increased in abundance post-drought, Chrysococcus sp. and Chlorococcum sp., were strongly correlated with NO₃⁻. These patterns were similar to the a posteriori analysis of the species PCA suggesting that the environmental variables are explaining real variability in the community data. Thus, the statistical analyses support the conclusion that the environmental conditions of the extreme drought could explain some of the change in the structure of the phytoplankton community. Specifically, results of the RDA suggest that water quality parameters are as important, if not more so, than physical conditions as drivers of phytoplankton community response. Further, one important aspect of the function of the phytoplankton community, specifically the uptake of dissolved silica from the lake water by diatoms, was also much more pronounced during the extreme drought than during the other years of the study. The environmental conditions of the summer of 2002 may have surpassed a threshold for the dominant species of the phytoplankton community of Green Lake 4, allowing a few of the previously rare species, such as Synedra sp., to become established.

Although the statistical analysis of the monitoring data presented here does not provide an explanation for the post-drought increase in algal biovolume, the general increase in nutrient concentrations and the correlation of the abundance of Chrysococcus sp. and Chlorococcum sp. with NO₃⁻ in the RDA suggest a possible connection with drought-related changes in hydrologic flowpaths in the alpine watershed. Drought has the potential to influence surface water hydrochemistry through groundwater/surface water interactions, and through changes in source waters and flow paths (Brække, 1981; Williams et al., 1996). Additionally, variations in the relative contributions of snowmelt and groundwater directly affect surface hydrochemistry. For example, Williams and Melack (1997) observed that the beginning of snowmelt runoff after antecedent dry periods commonly produced elevated water chemistry concentrations.

SYNERGISTIC INTERACTIONS

In alpine lakes, the response of the phytoplankton community to climate-driven conditions may be amplified by synergistic interactions with other drivers of environmental change. Atmospheric nitrogen deposition at Niwot Ridge is presently increasing at a rate of 0.3 kg N ha⁻¹ yr⁻¹ (Williams and Tonnessen, 2000). This sustained nitrogen deposition can affect the stoichiometry of the lake ecosystem by inducing or accentuating phytoplankton P limitation and thus resulting in high N:P and C:P ratios in phytoplankton biomass (Sterner and Elser, 2002). Such stoichiometric shifts may create ultra P-limited conditions, especially in otherwise low nutrient lakes such as Green Lake 4. Furthermore, this altered state could eliminate algal species that cannot survive in severely P-limiting conditions, while elevated phytoplankton C:P ratios could propagate beyond the level of primary producers to impair zooplankton because high dietary C:P ratios can result in direct zooplankton P-limitation (Sterner and Elser, 2002). The higher biovolume of the post-drought years could possibly reflect a response to a moderation of phosphorus limitation by the phytoplankton. If there was a change in groundwater/surface water interactions caused by the drought, such conditions could increase the flux of phosphorus to the lakes from mineral weathering in shallow groundwater. Thus, this study suggests that watershed-scale changes following drought may also have an indirect and sustained effect on the phytoplankton community, prolonging the response to extreme drought.