Scientific Investigations Report 2007–5117
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5117
Back to Table of Contents
Growth and decomposition of dense blooms of Aphanizomenon flos-aquae in Upper Klamath Lake frequently cause extreme water-quality conditions that have led to critical fishery concerns for the region. Two species of suckers endemic to the lake, the Lost River sucker and the shortnose sucker, are listed as endangered. The poor water-quality conditions associated with the long and productive algal blooms are believed to be the primary threat to these adult endangered suckers. The Bureau of Reclamation has asked the USGS to examine the water-quality data collected by the Klamath Tribes for relations with lake level. This work, which evaluates a 17-year dataset (1990–2006), serves as an update to the original analysis of a 5-year dataset (1990–94) performed by Wood and others (1996).
Both univariate hypothesis testing and multivariable analyses evaluated using an information-theoretic approach revealed the same results—no one overarching factor emerged from the data. No one single factor could be relegated from consideration either. The lack of statistically significant, strong correlations between water-quality conditions, lake level, and climatic factors does not necessarily show that these factors do not influence water-quality conditions; it is more likely that they all work in conjunction with each other to affect water quality. A few different conclusions could be drawn with the larger dataset than with the smaller dataset examined by Wood and others (1996), but for the most part, the outcome was the same. Using an observational dataset that may not capture all variation in water-quality conditions (samples were collected on a two-week interval) and that has a limited range of conditions for evaluation (confined to the operation of lake) may have confounded the exploration of explanatory factors. In the end, all years experienced some variation in poor water-quality conditions, either in the timing of the occurrence of the poor conditions or in their duration. The dataset of 17 years simply provided 17 different patterns of lake level, cumulative degree-days, timing of the bloom onset, and poor water-quality conditions, with no overriding causal factor emerging from the variations.
Water-quality conditions measured during 1990–2006 were evaluated for their potential to be harmful to the endangered sucker species. Based on high-stress thresholds established to calculate stress indices for Upper Klamath Lake suckers (Reiser and others, 2000), the values used in this study to delimit poor water-quality conditions were: water-temperature values greater than 28°C, dissolved-oxygen concentrations less than 4 mg/L, and pH values greater than 9.7. Only a few water temperatures recorded during May–October 1990–2006 were greater than 28°C. These were all measured at the surface, in the afternoon, and most were in the latter part of July, when air temperatures are expected to be elevated as well. Dissolved-oxygen concentrations of less than 4 mg/L were generally recorded in mid- to late-summer—most frequently in August, then July, and finally in September. Unlike the patterns in the dissolved-oxygen data, high pH values were more frequent and occurred earlier in the season and parallel with growth in the algal bloom, typically in June and July. Spatial patterns show that when high pH values were measured, they were often observed at sites all around the lake.
Total ammonia concentrations were screened against USEPA criteria, which apply only to samples with pH values of less than 9. For the half of the dataset that these criteria applied to, only a few samples (less than 1 percent for the acute criterion and less than 10 percent for the chronic criterion) exceeded the criteria; however, half of these exceedances occurred in the bay areas. Un-ionized ammonia concentrations were calculated for the entire dataset and evaluated against LC50 values of 530 µg/L for Lost River suckers (Saiki and others, 1999). About 3 percent of the un-ionized ammonia concentrations exceeded 530 µg/L, and there was at least one exceedance at each of the 10 sites. Although the USEPA criteria for total ammonia represent “acceptable no-effect levels” of total ammonia, the LC50 values for un-ionized ammonia concentrations represent “unacceptable severe-effect levels” for the health of the suckers.
To help evaluate relations in water-quality conditions between the years, three factors were used to rank the years in relation to each other: lake level, cumulative degree-days from April 1 to May 15, and timing of the onset of the AFA bloom. The end-of-month lake levels for May–August were compared to each other and to the post-dam historical (1922–2006) record. The years 1992, 1994, and 1991 were recognized as notably low lake-level years. The degree-days ranking serves as a measure of how warm each spring was, recognizing that climatic factors such as air temperature and cloud cover are expected to have an effect on when the bloom begins to grow and the rate at which it develops. The onset of the bloom for each year was defined by when the chlorophyll‑a concentration exceeded 20 µg/L. For the majority of the years, this occurred during the last week of May and the first week of June, with the earliest bloom starting in mid-May in 1992 and the latest bloom starting in mid-June in 2006. Comparison of the years by these three ranking variables reveals that these variables are interlinked, but no single one emerged as an overall controlling factor.
The 10 hypotheses relating water-quality variables, lake level, and climatic factors from Wood and others (1996) evaluation of the 1990–1994 dataset were tested in this analysis for the larger 1990–2006 dataset. These hypotheses proposed relations between lake level and chlorophyll‑a, pH, dissolved oxygen, total phosphorus, and water temperature. As in the previous study, there was no evidence in the larger dataset for any of these relations based on a seasonal (May–October) distribution. When analyzing just the June data, the previous study did find evidence for three hypotheses relating lake level to the onset of the bloom, June chlorophyll‑a concentrations, and the frequency of high pH values in June. These hypotheses were not supported, however, by the 1990–2006 dataset, but the two hypotheses related to cumulative degree-days from the previous study were: chlorophyll‑a concentrations were lower and the onset of the algal bloom was delayed when spring air temperatures were cooler. Other relations between water-quality variables and cumulative degree-days were not determined to be significant.
Multivariable analyses of the data revealed similar results. Multiple regressions were performed between lakewide water-quality measures—the percentage of low dissolved-oxygen concentrations in July and August and the percentage of high pH values in July and August—and six physical and biological variables—peak chlorophyll‑a concentrations, degree-days, water temperature, median October–May discharge in the Williamson River, median monthly wind speed, and median monthly lake level in Upper Klamath Lake. Model sets were developed for each combination of these chosen factors, both with and without lake level as an explanatory variable, and evaluated using an information-theoretic approach. In every instance, the addition of lake level to the regression resulted in an increase in the Akaike Information Criterion value of somewhere between 2 and 5, indicating that the model without lake level as a variable offered a better fit of the data.
The multiple regression models also were compared to each other as a means of explaining the variation observed between years in water-quality conditions in the lake. The generation of these multiple models with varying numbers of explanatory variables yielded no single overarching equation. For each water-quality measure tested, the models with the lowest AICc statistics, and therefore the best fit, were the models that considered each physical variable individually. The variables with the best fit for dissolved oxygen were water temperature and wind speed, whereas the variable with the best fit for pH was water temperature. The Akaike weights for the remaining variables were fairly evenly distributed, indicating that there was no clear hierarchy of importance among those variables.
Although water temperature and wind speed appear to be important explanatory variables for the variance observed in different water-quality measures, no overarching variable or combination of variables was revealed. As with the conclusion from the univariate analyses, it is suspected that these variables work in combination to affect water quality in Upper Klamath Lake. This conclusion makes sense when considering that no clear pattern has emerged from multiple approaches to examining interrelations among water-quality conditions, lake level, and climatic factors. Although the water-quality conditions and climatic factors around the lake can be defined, the dynamic nature of these variables and their interactions from year to year, within a season, and between sites around the lake confounds our ability to explain or predict water-quality conditions in Upper Klamath Lake. At present, no single causal factor can be clearly identified.
Back to Table of Contents