Final Report: Genetic Diversity in California Native Fish Exposed to Pesticides

EPA Grant Number: R826603
Title: Genetic Diversity in California Native Fish Exposed to Pesticides
Investigators: Anderson, Susan L. , Kuivila, Katherine , May, Bernard , Wilson, Barry W.
Institution: University of California - Davis
EPA Project Officer: Packard, Benjamin H
Project Period: August 1, 1998 through January 31, 2003
Project Amount: $649,003
RFA: Ecological Indicators (1998) RFA Text |  Recipients Lists
Research Category: Ecological Indicators/Assessment/Restoration , Ecosystems


The overall goal of this project was to determine the effect pesticide exposure may have on genetic variation in a California native fish. This topic is significant because pesticide contamination is extensive in California; yet, almost nothing has been done to evaluate effects on native fish species.

Our first objective was to examine populations of Sacramento sucker (Catostomus occidentalis) exposed to landscape-scale pesticide releases to determine whether changes in genetic diversity are associated with indicators of pesticide exposure or natural genetic variation. We utilized two complementary molecular techniques: (1) Amplified Fragment Length Polymorphisms (AFLP); and (2) variation in microsatellite loci. The design of this study also included an exposure assessment, which was comprised of both analytical chemistry of pesticides as well as an extensive evaluation of California Department of Pesticide Regulation (CDPR) data. In addition, short-term biomarkers (acetylcholinesterase [AChE] enzyme activity and DNA strand breaks) were employed to elucidate pesticide exposure and potential genotoxic responses.

Our second objective was to compare the AFLP technique with the Randomly Amplified Polymorphic DNA (RAPD) technique to determine which produced the most informative and reproducible DNA fingerprints. The RAPD technique is significant because it has been used in U.S. Environmental Protection Agency (EPA) studies, and is a popular technique in scientific literature; however, the AFLP technique permits the examination of more of the genome per unit effort and is reputed to be more reproducible.

Our third objective was to evaluate potential linkages between any observed changes in genetic variation and fitness parameters in individuals and populations. With approval from our Project Officer, we expanded the scope of Objective 1 and deleted Objective 3 midway through the project. This change allowed us to explore more directly the implications of our findings to environmental management. Our study is unique because genetic variation associated with contaminant exposure has not been evaluated in fish populations on a large geographic scale.

Summary/Accomplishments (Outputs/Outcomes):

Objective 1

Population Genetics. Exposure to contaminants can affect survivorship, recruitment, reproductive success, mutation rates, and migration, and may play a significant role in the partitioning of genetic variation among exposed and unexposed populations. However, the application of molecular population genetic data to evaluate such influences has been uncommon and often flawed. We tested whether patterns of genetic variation among native fish populations (C. occidentalis) in the Central Valley of California were consistent with long-term pesticide exposure history, or primarily with expectations based on biogeography.

The application of pesticides in the Central Valley of California represents a compelling case study to examine the potential genetic effects of contaminant exposure at the landscape scale. Intensive farming exists throughout approximately 9 million acres of the large interconnected watersheds of the Sacramento and San Joaquin Valleys, which drain 40 percent of the land area of California. Application of organophosphates and carbamates to orchards in the dormant season is documented to have occurred for at least 3 decades. Studies by Dr. Kuivila and coworkers have documented widespread pesticide contamination in waterways and have shown that pesticides reach the rivers in amounts that are toxic to test invertebrates. C. occidentalis was selected for this work because the species could be collected in rivers flowing into both the Sacramento and San Joaquin Basins and because the species was present both above and below major dams.

Field sampling was designed to rigorously test for both geographic and contamination influences. One strength of this project was the extensive sampling design, which allows for a high level of field replication and a sophisticated exposure analysis. We collected more than 1,000 fin clips at eight reference and four impacted stations in six watersheds. The selection of sites permitted us to test for patterns in genetic structure at different geographical hierarchies; for example, we could compare among exposed and reference sites at the basin level, the river level, or the level of site within river.

Fine-scale structure of these interconnected populations was detected with both AFLP and microsatellite markers, and patterns of variation elucidated by the two marker systems were highly concordant. We found that biogeographic hypotheses described the dataset better than hypotheses relating to common historical pesticide exposure. Downstream populations had higher genetic diversity than upstream populations, regardless of exposure history, and genetic distances showed that populations from the same river system tended to cluster together. Relatedness among populations primarily reflected directions of gene flow, rather than convergence among contaminant-exposed populations. Watershed geography accounted for significant partitioning of genetic variation among populations; whereas, contaminant exposure history did not. Genetic patterns indicating contaminant-induced selection, increased mutation rates, or recent bottlenecks, were weak or absent. Our data indicate that contaminant exposure was not a significant factor in the genetic structure of C. occidentalis populations, and these findings highlight the importance of testing contaminant-induced genetic change hypotheses within a biogeographic context.

Strategic application of molecular markers for analysis of fine-scale structure, and for evaluating contaminant impacts on gene pools, is a significant aspect of this study's design. We know of no previous studies that have used AFLP and analysis of microsatellite loci in a complementary approach toward the goals above. Microsatellite loci are sensitive for detecting new mutations and for analysis of fine structure of populations. In contrast, AFLP analyses afford random sampling of a larger portion of the genome per unit effort and do not require prior sequence information.

In summary, this aspect of our work was significant because it provided an example of rigorous application of molecular techniques to assess genetic effects of contaminant exposure. The most unique facets of our research were: (1) the sampling design permitted field replication and the comparison of various aspects of biogeographic structure; (2) two complementary molecular techniques were used; and (3) CDPR data and related documentation were used to characterize landscape-scale pesticide exposure over multiple generations. A further significant aspect of our work was that the focus on native species was unique; hence, methods and markers that we developed already have been useful to other researchers working in conservation biology of fishes (see below).

Short-Term Biomarkers and Exposure Analysis. When pesticides are introduced into watersheds of the Central Valley of California during storm events, toxicants often reach concentrations acutely toxic to standard test invertebrates. However, effects of these dramatic pesticide pulse events on resident species in the field have not been examined. To determine whether pesticide inputs cause significant detrimental effects in C. occidentalis, we combined field-caging experiments with controlled laboratory exposures to field-collected water samples and examined sublethal physiologic responses (biomarker measurements). In addition, analytical chemistry was performed to characterize toxicant exposures.

Field-caging sites included the mainstream San Joaquin River, downstream of all tributaries (SJ, downstream of agriculture), Orestimba Creek just above the confluence with the San Joaquin (OD, downstream of agriculture), and Orestimba Creek upstream of all agriculture (OU, reference site). Suckers were placed in cages (eight fish per cage), multiple cages were deployed per site, and individual cages were retrieved at different time points to evaluate changes in biomarker responses following the rise and fall of pesticide concentrations. Fish were sacrificed, and multiple tissues were excised and archived for subsequent biomarker analysis. DNA strand breaks (Comet Assay) and AChE activity were analyzed. We selected AChE enzyme inhibition as a biomarker of high specificity to organophosphate and carbamate insecticides and DNA strand breakage as a biomarker responsive to a wide range of chemical stressors. Pesticide concentrations were measured concurrently using gas chromatography/mass spectrometry (GC/MS) analysis on samples collected at several time points throughout the storm. Finally, to complement DNA strand break data, we included the Ames mutagenicity assay as another indicator of genotoxicity.

Data from these experiments indicated that concentrations of dormant season pesticides during 2000 and 2001 were much lower than in previous years, and did not induce AChE enzyme inhibition in exposed fish in the field or the laboratory. During 2001, we expanded our list of pesticide analytes from 2000 experiments, and several pesticides other than organophosphorus compounds (OPs) were detected in our samples including carbaryl, hexazinone, simazine, molinate, trifluralin, 4-keto-molinate, atrazine, prometryn, metolachlor, thiobencarb, dacthal, pendimethalin, napropamide, and oxyfluorfen. Pesticide concentrations were lower than had been observed in previous years at OD. Yet, concentrations of some chemicals were higher in the San Joaquin River. At OD, combined concentrations of diazinon plus methidathion did not exceed 24 ng/L. The reference site at upstream Orestimba Creek did not have detectable concentrations of any of the pesticides. The San Joaquin River site had maximum concentrations of 137 ng/L diazinon and 179 ng/L methidathion, almost twice as high as in 2000, and the entire peak lasted 5 to 6 days.

In contrast to the analytical chemistry and the AChE data, DNA strand breaks significantly were elevated in fish exposed to San Joaquin River water (38.8 percent, 28.4 percent, and 53.6 percent DNA strand breakage in 2000 field, 2000 lab, and 2001 field exposures, respectively) compared to a nearby reference site (15.4 percent, 8.7 percent, and 12.6 percent in 2000 field, 2000 lab, and 2001 field exposures, respectively). Measurements in 2001 indicated that DNA strand breakage was indeed correlated with the timing of storm runoff, but the chemistry results indicated that toxic substances, other than detected organophosphate and carbamate insecticides, were responsible for the effects observed.

In collaboration with Dr. Sergei Kotolevstev of Moscow State University, whose effort on this project was donated by the Distinguished Research Fellow Program at BML, we also examined mutagenicity of site waters using the Ames assay. Our studies indicated that water collected from the San Joaquin River was significantly mutagenic, using two tester strains. These findings, coupled with DNA strand break data, provide strong evidence for the genotoxicity of San Joaquin River water during dormant season runoff events. Two additional manipulations of San Joaquin River water indicate that metals are not responsible for observed mutagenicity indicated by the Ames assay. Metals were removed from San Joaquin River whole water by ion exchange, and this treatment did not reduce mutagenicity in the TA 98 strain. Reconstitution of metals by elution off the ion exchange column into clean water did not restore any mutagenicity.

In summary, this aspect of our study was significant because we observed that toxic substance(s) other than the currently understood "contaminants of concern" in the San Joaquin River may be causing genotoxic effects in aquatic life. The effect was observed in 2 years with two different test approaches. Analytical chemistry coupled with AChE enzyme activity revealed that the toxicant was not organophosphate or carbamate pesticides. In addition, bioassay-directed fractionation of site water further indicated that an organic toxicant was responsible for the effects observed. Additional aspects of the significance of this research are presented below.

Objective 2: Comparison of AFLP and RAPD

Dr. Bernie May and former postdoctoral researcher Dr. Mark Bagley, now with the EPA Office of Research and Development (EPA/ORD) in Cincinnati, completed an assessment of the informativeness and reliability of RAPD and AFLP fingerprinting methodologies for population genetic analyses using fish with a well-established pedigree. Characteristics of gel bands within DNA fingerprints that appeared reproducible, and bands that appeared artifactual, were compared to determine what factors most strongly influenced reproducibility. This information subsequently could be used to select among bands (within DNA fingerprints) for population genetic analysis. Our focus strictly was on gel phenotypes; other studies have addressed the influences of polymerase chain reaction (PCR) conditions on reproducibility of DNA fingerprints.

Three full-sib families of rainbow trout (Oncorhyncus mykiss) were constructed by crossing an F1 progeny of an anadromous strain and a freshwater-resident strain back to the freshwater resident strain. A total of 30 DNA samples were prepared, consisting of 18 progeny samples and 2 sets of 6 parental samples. RAPD and AFLP protocols were similar to those used by the investigators that originally described the methodologies, but were modified to allow detection of DNA fingerprints with fluorometry. A total of eight 10-mer RAPD primers and eight AFLP primer combinations were evaluated. Highly conservative band-scoring criteria were utilized so that the repeatability (band amplified in both extractions from each of the six parental samples) of bands could be characterized and so that heritability (band identified in any of the progeny also observed in at least one adult) and transferability (band identified in one of the parents also is observed in at least one of its six progeny) of bands determined.

We found that the number of segregating bands is higher using RAPD than AFLP, but also that the reproducibility of RAPD bands is far lower. The reproducibility of AFLP bands was found to be a function of band intensity. Thus, we adopted a criterion that only bands that represent greater than 1 percent of total band intensity should be scored. Using this approach, only 1 percent of the AFLP bands were judged to be irreproducible. In contrast, using the same 1 percent band-intensity criterion, 16.1 percent of the RAPD bands were irreproducible.

These results suggest that any procedure implemented to cull unreliable RAPD bands is likely to remove a large portion, perhaps most, of the segregating bands from the analysis. Thus, the numbers of segregating RAPD bands available to be scored will be similar to or less than the number of segregating AFLP bands that are scored. At present, there is no reason to believe that any criterion we can develop will increase the reliability of RAPD methodology to a level that is comparable to AFLP analysis.

In summary, this portion of our research was significant because we demonstrated that AFLPs are superior to the RAPD methodology, and this finding has already influenced design of research programs within EPA/ORD (see below).


We have advanced the science of ecotoxicology as it relates to pesticide effects on fish populations and fish conservation biology in the following ways:

• We have shown that studies regarding the effects of toxicants on gene pool variation must carefully analyze the biogeographical context of population data. Rigorous study design must include adequate field replication at appropriate geographic scales. Many prior studies have not emphasized field replication, and we know of no study to assess effects of toxicants on a large geographic scale.

• We elucidated the complementary nature of microsatellite and AFLP markers for fish population studies and we are the first to use these techniques in combination for fisheries biology and toxicology. Although we did not detect significant pesticide effects, we recommend that this powerful combination of tools be utilized by other researchers where lower gene flow or stronger selection by toxicants may be in evidence.

• We demonstrated the efficacy of the AFLP technique over RAPDs to survey large numbers of anonymous DNA sequences for variability. This finding likely will be considered by numerous researchers in the design of future studies. As a direct result of our research, EPA/ORD has substituted AFLP for RAPD in its work to evaluate regional trends in patterns of genetic diversity of stream fish.

• We found that there is a need to characterize additional pesticides of concern in the San Joaquin River. It is possible that increasing use of pyrethroid insecticides is responsible for the genotoxic effects we observed. Improved analytical methods are needed to discern low doses of pyrethroids in riverine waters.

• Our research team utilized a state-of-the-art blend of analytical chemistry, hydrology, geographical information systems (GIS), biomarkers, Ames assay, bioassay-directed fractionation, and sophisticated molecular genetics to study the effects of pesticides on fish populations from the landscape to the molecular levels. We also utilized parallel field and laboratory exposures to increase the scope of inference of our findings. Each of these components was critical to designing and conducting an investigation that will impact local management, yet hopefully will prove to be a significant advancement in biomarker research. Related investigations in Canada have shown the value of the Ames assay in identifying genotoxic loads in waterways and numerous other published papers combine some of the biomarker approaches with chemistry. However, this investigation uses an unusually large and complementary range of techniques, and is the only investigation of its kind on pesticide inputs in native fish.

• Our integrated approach has permitted a more quantitative understanding of pesticide exposure and effects in a significant watershed. There is a great need for further studies of this nature in the arid West, where inputs are strongly seasonal, pulsed, and quite challenging to characterize. In addition, research of this nature can impact controversial water rights decisions, because it affects our understanding of how water quantity may be interrelated to water quality through the dilution and transport of toxicants.

We have contributed to the management of both pesticide inputs and native fish populations in California in the following ways:

• We observed that San Joaquin River water elicits genotoxic responses in native fish and in Ames mutagenicity assays. The effect is most pronounced following winter storms. We are seeking peer review of these data and have made initial reports to EPA Region 9, CALFED, State of the Estuary Conference, and other groups. After we benefit from further review, we envision that this information will be reported widely to the California State Water Quality Control Board, the California Department of Fish and Game, CALFED, and the San Francisco Estuary Institute. We will propose that the chemicals of concern be identified. Further information would then be used to inform the total maximum daily load (TMDL) process for agricultural chemicals in the San Joaquin River as well as the CALFED effort for restoring populations of target species in the San Francisco Bay Delta system. It is important to note that effects are not attributable to chemicals that are now considered the "contaminants of concern" in this watershed. In addition, because the effects were observed in multiple years and timepoints at the mouth of a large watershed, we surmise that effects may be widespread and more pronounced closer to the source of contamination.

• The microsatellite primers we developed during this project will be used for newly funded research on two endangered species. First, Dr. May and colleagues are using the markers to discriminate pure Modoc sucker populations from those that have introgressed with C. stomus occidentalis. The results of this study will be used by the U.S. Fish and Wildlife Service (FWS) to choose Modoc populations to be selected for reintroduction. Secondly, the microsatellite markers also are being utilized by Dr. May and Dr. Tranah at the Harvard School of Public Health to study Klamath suckers. Results of this study will be used by the U.S. Bureau of Reclamation and the FWS to help regulate water allocation of the Klamath River to lakes, farms, industry, and human consumption.

• We also provided microsatellite primers for EPA/ORD studies at Cincinnati, OH. The primers are being employed by Dr. Mark Bagley in an ongoing investigation to determine the potential long-term genetic effects of acid precipitation on white suckers in the United States and Canada.

• We verified that organophosphate and carbamate pesticide concentrations have been decreasing in the San Joaquin River and tributaries due to decreasing use and timing of application relative to rainfall. The combined use of analytical chemistry to assess exposure, in combination with AChE enzyme activity, to evaluate absorbed dose, provides a clear picture that biologically effective doses are declining. This type of information has not been available previously. These data will be reported to the Central Valley Regional Water Quality Control Board (CVRWQCB). The CVRWQCB works with other agencies and growers to manage the effects of agricultural practices on water quality. Knowledge of the physical, chemical, and biological bases of runoff and its effects helps in constructing recommendations to growers and wildlife managers to minimize pollution of streams and rivers.

• For this project, and for other ongoing research in the Wilson Laboratory, a standard for clinical and research laboratories to harmonize AChE enzyme assays was developed. Dr. Wilson's work was reported to Congress by the National Institute of Environmental Health Sciences (NIEHS), and this action has led to a rewrite of the guidelines for clinical laboratory licensing by the State of California. Accurate determination of AChE levels in species of concern will help establish meaningful no-observed-adverse-effects-levels (NOAEL) for pesticide exposures critical to risk assessment and regulatory recommendations and actions.

• We organized a preliminary study of possible pesticide effects on chinook salmon in the San Francisco Bay Delta using biomarker techniques and analytical chemistry. Salmon were caged in two tributaries and, using the same integrated approach, chemistry and biomarker responses were quantified. This project was funded by CALFED, and the concept was developed following extensive dialogue with Dr. Bruce Herbold at EPA Region 9 after regular briefings about the research for this project. Data will be used to determine whether toxicants are implicated in decreased survivorship of chinook salmon in the Old River. Depending on the outcome, this research would be used to design more definitive research that could be used in controversial water rights decisions. The development and funding of this study highlights the significant partnerships we have formed with agency colleagues, and the recognition we have received for our integrated approach.

Journal Articles on this Report : 3 Displayed | Download in RIS Format

Other project views: All 32 publications 5 publications in selected types All 4 journal articles
Type Citation Project Document Sources
Journal Article Bagley MJ, Anderson SL, May B. Choice of methodology for assessing genetic impacts of environmental stressors: Polymorphism and reproducibility of RAPD and AFLP fingerprints. Ecotoxicology 2001;10(4):239-244. R826603 (2000)
R826603 (2001)
R826603 (Final)
  • Abstract from PubMed
  • Abstract: SpringerLink Abstract
  • Other: SpringerLink PDF
  • Journal Article Belfiore NM, Anderson SL. Effects of contaminants on genetic patterns in aquatic organisms: a review. Mutation Research-Reviews in Mutation Research 2001;489(2-3):97-122. R826603 (1999)
    R826603 (2000)
    R826603 (2001)
    R826603 (Final)
  • Abstract from PubMed
  • Abstract: Science Direct Abstract
  • Other: Science Direct PDF
  • Journal Article Roach JL, Colayco R, To C, Batey D, May B. Microsatellite genotyping using the BaseStation® DNA fragment analyzer. MJ Research Application Note 2003;2(7). R826603 (Final)
  • Other: RSMAS PDF
  • Supplemental Keywords:

    watersheds, estuary, precipitation, mutagen, ecological effects, population, enzymes, genetic polymorphisms, toxics, ecosystem indicators, aquatic, integrated assessment, environmental chemistry, ecology, genetics, zoology, emap, monitoring, western, EPA Region 9, agriculture., RFA, Scientific Discipline, Toxics, Geographic Area, Water, Ecosystem Protection/Environmental Exposure & Risk, Hydrology, Ecology, Water & Watershed, exploratory research environmental biology, Genetics, Environmental Chemistry, Ecosystem/Assessment/Indicators, Ecosystem Protection, Chemistry, pesticides, State, Ecological Effects - Environmental Exposure & Risk, EPA Region, Watersheds, Ecological Indicators, ecological exposure, Region 9, anthropogenic stresses, pesticide exposure, risk assessment, genotype, agricultural watershed, biomonitoring, AFLP, aquatic ecosystems, pesticide runoff, DNA, water quality, fish , California (CA), agriculture ecosystems

    Relevant Websites: Exit Exit

    Progress and Final Reports:

    Original Abstract
  • 1999 Progress Report
  • 2000 Progress Report
  • 2001 Progress Report
  • 2002