Irrigation-induced contamination of water, sediment, and biota in the western United States-synthesis of data from the National Irrigation Water Quality Program
In October 1985 the U.S. Department of the Interior (DOI), through the National Irrigation Water Quality Program (NIWQP), began a series of field investigations at 26 areas in the Western United States to determine whether irrigation drainage has had harmful effects on fish, wildlife, and humans or has reduced beneficial uses of water. In 1992 NIWQP initiated the Data Synthesis Project to evaluate data collected during the field investigations. Geologic, climatologic, and hydrologic data were evaluated and water, sediment, and biota from the 26 areas were analyzed to identify commonalities and dominant factors that result in irrigation-induced contamination of water and biota. Data collected for the 26 area investigations have been compiled and merged into a common data base. The structure of the data base is designed to enable assessment of relations between contaminant concentrations in water, sediment, and biota. The data base is available to the scientific community through the World Wide Web at URL http://www.usbr.gov/niwqp . Analysis of the data base for the Data Synthesis included use of summary statistics, factor analysis, and logistic regression. A Geographic Information System was used to store and analyze spatially oriented digital data such as land use, geology and evaporation rates. In the U.S. Department of the Interior (DOI) study areas, samples of water, bottom sediment, and biota were collected for trace-element and pesticide analysis. Contaminants most commonly associated with irrigation drainage were identified by comparing concentrations in water with established criteria. For surface water, the criteria used were typically chronic criteria for the protection of freshwater aquatic life. Because ground water can discharge to the surface where wildlife can be exposed to it, the criteria used for ground water were both the maximum contaminant levels (MCL's) for drinking water and the chronic criteria for the protection of freshwater aquatic life. Data collected by the NIWQP studies indicated that, in surface water, filtered and unfiltered samples had nearly the same concentrations of arsenic, boron, molybdenum, and selenium for concentrations greater than about 10 micrograms per liter. Therefore, in this concentration range, filtered concentrations can be directly compared to biological-effect levels developed for unfiltered samples. In the range of 1 to 10 micrograms per liter there may be a tendency for unfiltered arsenic concentrations to be greater than filtered concentrations. For selenium, however, the data suggest differences from equality in that range result from analytical imprecision and not a general tendency for unfiltered concentrations to be greater than filtered concentrations. This relation may not be true in lentic, nutrient-rich waters because in such settings algae can bioaccumulate large amounts of selenium and other trace elements. Selenium was the trace element in surface water that most commonly exceeded chronic criteria for the protection of freshwater aquatic life; more than 40 percent of the selenium concentrations in surface-water samples exceeded the U.S. Environmental Protection Agency (USEPA) aquatic-life chronic criterion (5 micrograms per liter). In 12 of the 26 areas at least 25 percent of the surface water-samples had selenium concentrations that either equaled or exceeded the chronic criterion (5 micrograms per liter). More than 28 percent of boron concentrations and almost 17 percent of the molybdenum concentrations exceeded the aquatic life criteria established by the State of California (550 and 19 micrograms per liter, respectively). In ground water, more than 22 percent of the arsenic concentrations and more than 35 percent of the selenium concentrations exceeded the MCL (10 and 50 micrograms per liter, respectively). Few samples of uranium in surface water exceeded a criterion for the protection of aquatic life (300 micrograms per liter), but 44 percent of the uranium concentrations in ground water exceeded the MCL (30 micrograms per liter). Molybdenum, selenium and uranium were the trace elements most commonly found in bottom-sediment samples that exceeded the upper limit of the 95th percentile expected range in soils of the Western United States. Selenium is the only trace element for which ecological sediment guidelines are used in this report. Selenium concentrations commonly exceeded the ecological sediment guideline of two micrograms per gram. DDT and its degradation products DDD and DDE were the most common pesticide residues found in surface water at concentrations exceeding criteria. However, almost all the samples exceeding the criteria were from a single study area, the Owyhee-Vale Reclamation Project areas in Oregon and Idaho. The organochlorine pesticide chlordane was detected in 30 percent of the bottom-sediment samples, and undegraded DDT was detected in 21 percent. DDT or its degradation products were detected in all 21 study areas where bottom-sediment samples were analyzed for organochlorine pesticides. A principal-components analysis indicated that elevated selenium concentrations in surface water are not associated with elevated boron, molybdenum, or arsenic concentrations. The occurrence of selenium is associated with sulfate and uranium. The association of boron and molybdenum with chloride suggests that evaporative processes control their concentrations. Arsenic is not associated with any other measured trace element and is associated negatively with selenium. This report focuses on selenium because it was the trace element most frequently found at concentrations exceeding criteria for the protection of aquatic life. Selenium concentrations in water are dynamic, and, at a given site, the selenium concentration can change by an order of magnitude during a year and from one year to another. In some areas, selenium contamination may not occur during normal or wet periods. However, during a drought, reduced water deliveries may result in selenium contamination by evaporative concentration. Marine sedimentary rocks, especially those of Late Cretaceous age, are likely to be seleniferous. Irrigation of soils derived from them can contribute large amounts of selenium to drainwater; shallow wells in and near irrigated areas contained hundreds to thousands of micrograms per liter of selenium. The median selenium concentration in surface-water samples from NIWQP sites associated with Upper Cretaceous marine sedimentary rocks is 7 micrograms per liter (range less than 1 to 8,300 micrograms per liter) and from sites not associated with such rocks is 0.4 micrograms per liter (range less than 1 to 390 micrograms per liter). Irrigation-induced selenium contamination has been observed only in arid or semiarid areas. In those NIWQP study areas having local geologic sources of selenium, typically more than 25 percent of the surface-water samples exceed the chronic criterion for selenium if the evaporation rate is 3.0 times greater than the annual precipitation. In terminal water bodies, selenium accumulates and is not flushed out. In both terminal and flow-through lakes and ponds, the median selenium concentrations in surface water for samples collected from June through August are nearly the same (1.0 and 0.8 micrograms per liter, respectively). However, the 75th-percentile selenium concentration for terminal water bodies (24 µg/L) is significantly higher than for flow-through systems (4 µg/L). Selenium concentrations in biota were compared with concentrations that have been demonstrated to have adverse effects on similar species (the effect level) or to have adverse effects on another species if consumed (the dietary effect level). Twenty-five percent of the plant samples had selenium concentrations exceeding the dietary effect level (3 micrograms per gram dry weight) whereas more than 57 percent of the invertebrate samples and 61 percent of the fish samples exceeded the dietary effect level. Of the more than 2,000 bird eggs collected, 44 percent had selenium concentrations exceeding 6 micrograms per gram, a threshold value for reduced hatchability. In 14 areas, selenium concentrations in eggs from some populations of birds exceeded 6 micrograms per gram. Selenium-caused deformities of bird embryos were found in four of the NIWQP study areas; however, most study areas were not systematically surveyed for such deformities. Eggs were sampled from 34 species of birds belonging to 10 orders. Nearly all the eggs collected come from aquatic species of birds, with American coots, mallards, and American avocets being the three species most frequently collected. Of the 34 species, at least one set of eggs from 16 species had a geometric-mean selenium concentration of at least 12.5 micrograms per gram, a high-risk threshold. All three species of grebes yielded at least one set of high risk eggs, as did four of five species of shorebirds and five of eleven species of waterfowl. Egg-set data were examined to determine if some feeding guilds are more at risk to selenium poisoning than others. Analysis of data for waterbird eggs from the study areas where the 75th percentile selenium concentration in surface water exceeded 5 µg/L suggests that herbivorous birds may bioaccumulate less selenium than insect- and fish-eating birds. For birds from these study areas, selenium concentrations for 39 percent of the egg sets from herbivorous birds fell in the normal range (less than 3 µg/g) while only 7 and 0 percent, respectively, of egg sets from insect- and fish-eating birds fall in the normal range. Although herbivorous birds may bioaccumulate less selenium, it does not appear that any waterbird feeding guilds are particularly well buffered from exposure to selenium contamination. Predictive tools were developed to aid managers in identifying specific land areas at risk for irrigation-induced selenium contamination. The tools range from identifying broad geographic regions where selenium contamination is likely, to assessing the probability that selenium concentrations in a specific stream or lake exceed the chronic criterion for selenium. A geographic information system was used to prepare a map that identifies land areas in the Western United States that are susceptible to selenium contamination if irrigated. On the basis of the 75th percentile, selenium concentration in surface water, 12 of the 26 NIWQP study areas were classified as contaminated, two as seleniferous, and 12 as uncontaminated. The map correctly identified both seleniferous areas and 10 of 12 selenium-contaminated areas as susceptible; 10 of 12 uncontaminated areas were correctly identified as not being susceptible. About 160,000 square miles are identified as being susceptible; of that area, about 4,100 square miles have been identified by satellite imagery as actively being irrigated. Principal-components analysis and pattern-recognition techniques indicate that major-ion chemistry of water samples alone can be used to identify selenium- and nonselenium-producing areas in the Western United States. Water samples composed of simple salts of sulfate typically have concentrations of selenium that exceed 3 µg/g, whereas samples composed of simple salts of chloride or carbonate typically have low selenium concentrations. Weathering of soils that contain reduced-sulfur minerals, such as pyrite, mobilizes sulfur and selenium because selenium commonly substitutes for sulfur in these minerals. In areas where the bedrock is composed of Upper Cretaceous marine sedimentary rocks, logistic regression of data from the NIWQP sites indicates that if the dissolved-solids concentration equals 1,000 milligrams per liter, the probability is about 69 percent that the selenium concentration will exceed 5 micrograms per liter, the U.S. Environmental Protection Agency chronic criterion. In areas where the bedrock is not composed of such rocks, the probability is only about 10 percent. The avian-egg data within the biotic data base were used to make a quantitative toxicological risk assessment. Of the 23 study areas where avian eggs were sampled, 14 areas yielded at least one egg containing 6 µg/g selenium, the threshold for embryotoxicity. However, only 6 of the study areas yielded eggs containing enough selenium to expect selenium-induced teratogenesis of duck embryos. Predicted probabilities of discovering embryo teratogenesis matched field observations in 13 of the 14 study areas reporting results of embryo assessments. Bird eggs were collected from 161 individual sampling sites and at 79 of those sites selenium concentrations in one or more eggs exceeded 6 µg/g. At the 79 sites where biological effects are expected on the basis of selenium concentrations in the eggs, the median rate was 3.9 percent of the hens losing at least one egg to selenium-induced embryotoxicity. This corresponds to about 1.2 percent selenium-induced egg inviability among otherwise viable eggs. Across all NIWQP study areas, the overall rate of hens projected to lose at least one egg to selenium-induced embryotoxicity is estimated to be 1.9 percent, which corresponds to about 0.3 percent selenium-induced egg inviability among otherwise viable eggs. After accounting for increased mortality of selenium poisoned-hatchlings due to other factors such as weather and predators, it was estimated that that increases of 0.3 and 1.2 percent in inviable eggs would cause approximately a 1.4 and 5.4 percent depression in nesting success. Regional surveys of nesting success among ducks revealed that duck populations commonly exist near their demographic break-even point. The vulnerable demographic condition of North American duck populations during the mid-1960's to mid-1980's was primarily due to noncontaminant factors, such as poor-quality nesting habitat and dry climatic cycles. Under such conditions, rates of 1.4- to 5.4-percent depression in nesting success caused by exposure to selenium from irrigation projects can be crucial for avian populations already close to their demographic break-even point. Even the worst-case levels of contaminant effects, however, could be tolerated by populations of ducks existing just modestly above demographic break-even points. This suggests the biotic risk to ducks could be addressed by reducing irrigation-induced water pollution but more effectively by restoring high-quality (more predator-safe) nesting habitat. An analysis at the nesting-site level was made of the relation between selenium content of water and bird eggs. Eggs from 93 bird populations were collected from nesting sites where the water sample collected during April-July contained less than 5 µg/L selenium. The average selenium concentration in egg sets was embryotoxic in 19 of the 93 populations. Of the populations collected at sites where the selenium in the water was less than 1 microgram per liter, only four of 54 populations contained embryotoxic concentrations. Eggs from 65 populations of birds were collected from nesting sites where selenium concentrations in water samples collected during April-July equaled or exceeded 5 µg/L, and 55 of those 65 populations contained embryotoxic concentrations of selenium in the eggs. An analysis at the study-area level was made of the relation between selenium contamination of water and selenium contamination of the food chain and egg loss due to selenium poisoning. Most food organisms, particularly aquatic invertebrates and fish, contained potentially harmful amounts of selenium in study areas where selenium concentrations in more than 25 percent of the water samples exceed 5 µg/L. The analysis also indicates that some hens are predicted to lose eggs to selenium poisoning in all study areas where selenium concentrations in more than 25 percent of the water samples exceed 5 µg/L. These results suggest that areas where selenium contamination of the food chain and loss of eggs to selenium poisoning is occurring may be identified using the same methods developed to identify areas where selenium contamination of water is likely to occur.