Regional inhibition of cholinesterase in free-ranging western pond turtles (Emys marmorata) occupying California mountain streams


  • Erik Meyer,

    Corresponding author
    1. Division of Resources Management and Science, Sequoia and Kings Canyon National Parks,Three Rivers, California, USA
    2. Department of Biology, California State University, Fresno, California, USA
    • Division of Resources Management and Science, Sequoia and Kings Canyon National Parks,Three Rivers, California, USA
    Search for more papers by this author
  • Donald Sparling,

    1. Cooperative Wildlife Research Laboratory, Southern Illinois University, Carbondale, Illinois, USA
    Search for more papers by this author
  • Steve Blumenshine

    1. Department of Biology, California State University, Fresno, California, USA
    Search for more papers by this author


The present study investigated the potential effects of cholinesterase (ChE)-inhibiting pesticides on western pond turtles (Emys marmorata) occupying streams in two regions of California, USA. The southern region was suspected of having increased exposure to atmospheric deposition of contaminants originating from Central Valley agriculture. The northern region represented reference ChE activities because this area was located outside of the prominent wind patterns that deposit pesticides into the southern region. Total ChE activity was measured in plasma from a total of 81 turtles from both regions. Cholinesterase activity of turtles was significantly depressed by 31% (p = 0.005) in the southern region after accounting for additional sources of variation in ChE activity. Male turtles had significantly increased ChE activity compared with females (p = 0.054). Cloaca temperature, length, mass, handling time, body condition, and lymph presence were not significant predictors of turtle ChE activity. In the southern region, 6.3% of the turtles were below the diagnostic threshold of two standard deviations less than the reference site mean ChE activity. Another diagnostic threshold determined that 75% of the turtles from the southern region had ChE activities depressed by 20% of the reference mean. The decrease in ChE activity in the southern region suggests sublethal effects of pesticide exposure, potentially altering neurotransmission, which can result in various deleterious behaviors. Environ. Toxicol. Chem. 2013;32:692–698. © 2012 SETAC


In 2010, two million kilograms of organophosphate and carbamate insecticides were sprayed for pest control on California, USA, agricultural lands 1. The primary mechanism of toxicity of organophosphates and carbamates is the deactivation of cholinesterase (ChE), an essential enzyme of invertebrate and vertebrate nervous systems. Cholinesterase, a B-esterase enzyme, is responsible for the hydrolysis of the neurotransmitter acetylcholine into choline and acetic acid 2. In turn, the neurotransmitter is removed from the postsynaptic nerve ending. By binding to ChE, antagonistic organophosphates and carbamates can lead to a buildup of acetylcholine, resulting in an overstimulation of neurons. Widespread use of ChE inhibitors for insect control have indiscriminately impacted nontarget organisms, including events of acute lethal poisoning and evidence of various sublethal deleterious neurological effects to both terrestrial and aquatic organisms 3–5.

In California, extensive research has identified ChE-inhibiting contaminants in numerous media in mountainous natural areas bordering intensively cultivated and chemically treated farmland 6–10. One region receiving great attention in terms of this topic is the Sierra Nevada Mountains, an extensive north–south-oriented mountain range spanning inland California adjacent and downwind of fertile Central Valley agricultural lands. Postapplication pesticides from the Central Valley enter the atmosphere and deposit in nontarget areas through air, rain, snow, fog, and dry particle deposition 6, 9, 11, 12. This phenomenon is accentuated in the southern Sierra Nevada Mountains, where local geography forms the Fresno Eddy, a circulating wind pattern that brings polluted air originating in the Central Valley down valley and upslope into the nearby mountains 13. In Sequoia National Park, which is located in this region, pesticide concentrations appear to have a negative relationship with elevation and distance from the source 9, 10, 14. Air particle models have predicted deposition routes of contaminants from the Central Valley into the park, and pesticide detections are correlated with the timing of agricultural application in adjacent agricultural lands 8, 9.

Once in the environment, ChE inhibitors can be converted to more toxic and stable metabolites. For instance, chlorpyrifos oxons have been measured at higher concentrations in plant foliage, air, and dry deposition samples in the Sierra Nevada foothills than their parent forms 6, 9. This is ecologically relevant because organophosphates metabolites are 10 to 100 times more toxic to Sierra Nevada amphibians than their parental forms 5. Furthermore, many currently used pesticides like organophosphates and carbamates are water-soluble and can quickly break down under natural conditions 15, 16. Therefore, they may not persist in the environment or may accumulate in food webs like longer lived, historically used organochlorine pesticides and industrial byproducts 2. This can create misleading results in organophosphate and carbamate field studies that focus on concentrations in the environment because the effect of ChE inhibitors may outlast the actual detection of pesticide residues or exposure event 17, 18.

An alternative method of measuring pesticide residues in ecotoxicology research is the biomarker approach. This technique measures a biological response to environmental contamination at a molecular, cellular, or whole-organism level 2, 17, 19. Total ChE (TChE) is a highly reliable biomarker for detecting the presence of pesticides, and, if inhibition is detected, a direct relationship to loss of neurological function can be projected on study organisms 17. Total ChE activity in plasma primarily consists of butyrylcholinesterase, although in some species acetylcholinesterase can constitute up to one-fourth of TChE in plasma 20, 21. All ChE enzymes can be depressed by ChE-inhibiting pesticides, but TChE and butyrylcholinesterase are typical measurements for plasma analysis, while TChE and acetylcholinesterase are commonly measured in the brain or muscle tissues 4, 19, 22. Although brain and muscle tissues are effective for detecting pesticide-induced inhibition, their use in toxicology requires sacrificing the study animal. Utilizing nonlethal methods such as plasma ChE analysis is beneficial, although there is some evidence that rapid recovery of plasma ChE activity may make it a less useful tool for field toxicology compared with brain or muscle tissue 23. Organophosphates can permanently inhibit ChE, requiring synthesis of new enzyme, whereas ChE recovers rapidly after carbamate exposure 2. Therefore, plasma ChE enzymes are still considered an excellent tool for assessing asymptomatic pesticide exposure; when used as a health biomarker, they can be effective in evaluating low levels of permanent or short-term ChE inhibition from pesticide exposure 19. Care should be taken when using this approach, however, as ChE activity varies naturally within a species. Sources of variation include species-specific attributes such as body size, weight, and condition or habitat features such as water temperature, salinity, conductivity, and dissolved oxygen 17, 24.

The purpose of the present study was to investigate the ecological risk of atmospherically deposited ChE inhibitors to western pond turtles (Emys marmorata) inhabiting mountain streams in natural areas of the southern Sierra Nevada foothills. Emys marmorata, a state species of special concern, is California's only widespread native turtle. Turtles have long been considered excellent sentinel species in toxicology research due to their long lifespan, generalist feeding behavior, widespread distribution, occupation of diverse habitats, and nonmigratory habits 25. The present study evaluated TChE activity in turtles from several streams within two regions of California: streams in the southern Sierra Nevada foothills, where exposure to atmospheric deposition of ChE inhibitors is increased compared with those in Northern California, where turtles inhabit streams outside of the predominant atmospheric deposition pathways of contaminants originating in California's Central Valley. To further determine whether ChE activity is influenced by ChE-inhibitor pesticides, the present study took into account biological and environmental explanatory variables that could contribute to variation of ChE activity within a species.


Animal care and use

Animal handling and care approval was obtained through the Institutional Animal Care and Use Committee (File 145), California State University, Fresno. Research was also completed under approval of California Department of Fish and Game scientific collecting permit (SCP 11633) and National Park Service permits (WHIS-2011-SCI-0007 and SEKI-2011-SCI-0022).

Study sites and watershed characteristics

Seven sites were selected across a latitudinal gradient of 700 km representing two regions in California (Fig. 1, Table 1). The southern region consisted of four lotic sites in the Sierra Nevada foothills adjacent to and downwind of intensively cultivated areas. The northern region consisted of two lotic sites in the Klamath Mountains and one in the California Coast Range. Sites in the northern region were outside of the predominant wind patterns that transport airborne pesticides into the southern region. The South Fork Trinity River and Mad River are oriented away from Central Valley agriculture. Clear Creek is located just north of the Central Valley. This site could be prone to pesticide drift from agriculture to the south, although to a lesser extent than the southern sites because it is not within the Fresno Eddy weather pattern. Two additional ponds located in the southern region near the Tyler Creek site were opportunistically trapped to capture turtles. The results were not analyzed with data obtained from the lotic sites, although summary statistics will be provided for these ponds.

Figure 1.

Location of the sampling sites in California. Northern and southern sites are identified by black and white stars, respectively. Each map includes the total aerial and ground agricultural applications by township of all cholinesterase (ChE)-inhibiting pesticides (organophosphates and carbamates). Each region includes radiating multipoint buffers of 25, 75, and 150 km.

Table 1. Watershed characteristics for each site using ArcGIS
SiteOwnershipaElevation (m)bArea (ha)cPrecip. (cm)dTemp. (°C)d
  • a

    Land ownership where turtles were captured. Entire contributing watershed area may not be under the same ownership.

  • b

    Minimum (min) elevation represents the furthest downstream area of turtle captures and locations of GIS pour points. Maximum (max) elevation is the highest point within the delineated watershed. All turtle captures were within the first 25 to 50 m elevation of the minimum elevation.

  • c

    Surface area calculated using area and volume statistics in 3D analyst. Included in surface area analysis was the delineated watershed area as defined by pour points at each site.

  • d

    Precipitation and temperature contours within each delineated watershed area were averaged using 30-year Prism Climate Group point-derived models.

    NF = national forest; NP = national park; NRA = national recreation area.

Southern region        
 Jose CreekSierra NF8961,9115,68983.80.73012.3
 North Fork Kaweah RiverSequoia NP5232,93025,493102.1−
 Sycamore CreekSequoia NP4821,68558276.20.732.213.9
 Tyler CreekPrivate7882,3514,43078.7−1.728.511.7
Northern region        
 Mad RiverPrivate1321,83584,233214.4−0.927.410.4
 South Fork Trinity RiverSix Rivers NF1622,387260,541182.6−1.229.310.1
 Clear CreekWhiskeytown NRA3711,89344,968161.3−1.429.911.2

Western pond turtles occupy lotic habitat throughout the state 26. Spatial analysis was used to derive site-specific attributes to represent variation among streams. Differences between sites may contribute to variability in the physiological status of a species; thus, it is important to take this into consideration when utilizing biomarkers to evaluate environmental effects of xenobiotic exposure (Table 1) 17. Watersheds were delineated using hydrology tools in ArcMap (ESRI v 9.3). Watersheds were defined using pour points located at the furthest downstream location of a capture site. Pour points are used in ArcGIS to define contributing watershed area above a defined location within a stream network. Contributing watershed surface area and elevation were calculated in 3D analyst using 10-m digital elevation models. All sites are located in natural areas, with all or much of the contributing watershed area in National Park Service or National Forest Service ownership. No site had more than 4% of developed land cover within the contributing watershed area (see the U.S. Geological Survey land use dataset at Predominant land cover for all sites is deciduous forest, evergreen forest, mixed deciduous–evergreen forest, shrub or scrub, and grassland or herbaceous. Thirty-year (1971–2000) point precipitation and temperature data were extracted for each delineated watershed using models developed by the PRISM Climate Group (Oregon State University, Corvallis, OR, USA).

To illustrate the regional differences in pesticide use near each designated study region, the most recent pesticide use reports were used to sum ground and aerial agricultural applications of ChE-inhibiting pesticides at three distances using a multipoint buffer in ArcGIS (Fig. 1, Table 2) 1. Pesticide use reported in each township was included if its centroid was located within each radial buffer. Buffer distances were chosen to best represent 1-d back-trajectory atmospheric transport models completed for several western U.S. National Parks 8. This time frame represents local atmospheric transport, which describes proximal pathways of pesticides into nontarget areas.

Table 2. Aerial and ground applications of organophosphate (OP) and carbamate (CB) pesticides (kg) within radiating distances (km) of sites in each regiona
  • a

    Multipoint buffers were utilized to capture total agricultural pesticide use near each site.

  • b

    Distance (km) of radial multipoint buffer from sites in each region.

  • c

    Total organophosphates applied, kg of active ingredient.

  • d

    Total carbamates applied, kg of active ingredient.

  • e

    Total organophosphates and carbamates combined, kg of active ingredient.

Southern region
Northern region

Field sampling of turtles

Turtles were captured from August to October of 2011 primarily by hand (snorkeling), and occasionally in sardine-baited commercial nylon net traps. Searches were conducted largely in low-gradient and backwater pools, which constitute the turtles' primary lotic habitat 26. Captures made via snorkeling included extensive searches in root wads, in vegetation masses, under boulders, and in undercut banks. Thus, the inconspicuous habitat of turtles captured underwater suggests sampling was not biased toward individuals in poor physiological health. All sites were visited once, except North Fork Kaweah River, which was visited three times.

We attempted to maintain turtles at a consistent internal body temperature to ensure that physiological condition was not influenced by an immediate change in temperature. Once captured, turtles were placed in a container of water from the capture location. Cloaca temperature (°C), straight-line maximum carapace length (mm), and mass (g) were recorded from each turtle. Gender (male, female, juvenile) was determined by the presence of secondary sex characteristics 26. Handling time (in minutes) was recorded to account for stress-induced physiological changes 27. Fulton's condition factor (K) was calculated to quantify individual health for each turtle 28. Maximum carapace length and body mass were used to calculate body condition, expressed as K (kg/m−3) = 1,000 × [mass (g)/length (cm3)].

Blood was collected from the subcarapacial venous sinus using a 5-ml heparinized syringe to prevent blood coagulation. If lymph contamination was prominent, a second draw was attempted. Samples suspected to have partial lymph contamination were noted (yes or no) and coded as a nominal variable for statistical analysis to investigate whether lymph influenced ChE results. Syringes were stored on ice and processed immediately upon return from the field. Water temperature data loggers were placed in a well-mixed riffle at the furthest downstream location of each site. Data loggers were set to record temperature at 15-min intervals.

Sample preparation and analysis

Blood was placed in a vacutainer or cryovial and centrifuged at 1,170g force for 15 min. Plasma was separated using a sterile pipette and placed in 1.8-ml cryovials. Plasma samples were shipped overnight on dry ice to Southern Illinois University (Carbondale, IL, USA). Here samples were stored in a –80°C freezer until further analysis. Total ChE activity was determined by the colorimetric method of Ellmann et al. 29. Volumes of plasma and reagents in this method were adjusted to accommodate a 96-well plate reader (BioTex Synergy). Analytical temperature was held at 28°C. Total ChE activity was determined from three to four replicates for each sample with interreplicate coefficients of variation <8%. Cholinesterase values were determined from the rate of color change and expressed as µmol min−1 g−1 tissue.

Statistical analysis

All statistical procedures were completed in R 2.15.0 statistical software 30. Utilizing the MASS package in R, a linear discriminant analysis (DA) with a jackknifed prediction was used 31. We used DA to determine whether body size (carapace length), body condition (K), and cloaca temperature are useful variables in discriminating between northern and southern populations. The sequence of assumptions associated with DA as outlined in Zuur et al. 32 was verified prior to analysis.

Water temperatures have been identified as a natural influence on ChE activity in ectotherms. In lieu of collecting water temperatures at each capture location, we determined that a more accurate measurement would be to collect individual body temperatures from the cloaca. A Spearman rank correlation test (ρ) was used (cor.test) to determine whether cloaca temperatures were reflective of stream temperatures, therefore possibly acting as a surrogate variable to water temperatures. Using the cloaca temperatures allowed us to obtain a measurement for each turtle, whereas water temperatures were collected from a data logger placed at one location in the stream. The water temperatures logged during each turtle capture period were averaged and paired with the mean cloaca temperatures from turtles of the corresponding site visit. The Sycamore Creek cloaca temperatures were excluded from analysis as outliers because two of the four captured turtles were in warm residual pools not reflective of cooler water temperatures recorded from the data logger.

Variation in plasma ChE activity was tested with a linear mixed-effects model (LME) due to nested, hierarchal data, an unbalanced design, and heterogeneous error structures. A quantile–quantile plot verified normality of the response variable 32. Model comparisons were not made because we were only interested in the effects of all response variables on ChE activity. Fixed effects included mass, cloaca temperature, handling time, K (continuous variables), sex, lymph presence, and region (nominal variables) for each turtle. Individual sites were designated as a random effect. Each site was given a different variance structure to allow for heterogeneity. Statistical analyses were performed using the nlme package 33. Following steps in Zuur et al. 34, the data were refit with an ordinary linear regression using the gls function and compared with the LME model using a likelihood ratio (LR) test. A corrected p value was obtained using the command (0.5 * [1 - pchisq{LR, 1}]), where pchisq represents a chi-squared distribution of the LR. The LME model with random effect showed a significant improvement over the gls model (LR = 14.6, p = 0.0001). Model validation plots and a Shapiro–Wilk test of the model residuals confirmed error normality and homogeneity in the preferred model 34.


A total of 81 turtles had successful blood draws and subsequent valid ChE results (Table 3). The DA was able to accurately predict 77% of the turtles captured in the northern sites and 88% of the turtles in the southern sites based on body length, condition, and temperature.

Table 3. Sex ratios, maximum carapace length (CL), mass, Fulton's condition factor (K), and cloaca temperatures from turtles captured at seven sites in two regions of California
SiteNo.M:F:JCL (mm)Mass (g)K (kg/m3)Temp (°C)
Southern region4221:13:8124 (± 18.4)291 (± 119.8)0.144 (± 0.01)22 (± 2.7)
 Jose Creek123:5:4116 (± 9.8)223 (± 49.0)0.142 (± 0.01)23 (± 1.2)
 North Fork Kaweah River2415:6:3129 (± 17.5)320 (± 113.1)0.143 (± 0.01)21 (± 2.8)
 Sycamore Creek42:1:1106 (± 19.1)200 (± 115.1)0.156 (± 0.01)26 (± 3.0)
 Tyler Creek21:1:0154 (± 13.4)535 (± 36.0)0.150 (± 0.03)20 (± 1.6)
Northern region3915:19:5151 (± 21.8)538 (± 194.1)0.149 (± 0.01)24 (± 9.8)
 Mad River115:5:1146 (± 22.8)480 (± 205.7)0.147 (± 0.01)25 (± 1.2)
 South Fork Trinity River206:12:2159 (± 17.6)609 (± 171.3)0.147 (± 0.01)25 (± 1.4)
 Clear Creek84:2:2138 (± 23.8)439 (± 183.1)0.156 (± 0.01)23 (± 1.2)

Cloaca temperatures were positively correlated to water temperatures (ρ = 0.87, p < 0.001). These results indicate that cloaca temperatures are reflective of local water temperatures. Thus, further analysis could advance the use of cloaca temperature as a variable in the LME.

The LME model residuals were normally distributed (W = 0.97, p > 0.05). A strong regional effect on total ChE activity was found (p = 0.005; Fig. 2, Table 4). In the southern region, ChE was depressed by 0.043 µmol min−1 g−1 (95% confidence intervals −0.066, −0.020) compared with the northern region. Male turtles had 0.013 µmol min−1 g−1 higher ChE activity compared with females (95% confidence intervals 0.000, 0.026, p = 0.054). No significant relationship (p > 0.05) was found between ChE activities and handling time, body mass, cloaca temperature, lymph contamination, or body condition.

Figure 2.

Plasma cholinesterase activity (µmol min−1 g−1) of western pond turtles (Emys marmorata) from seven sites in two regions of California. The northern reference sites include Mad River (MAD), South Fork Trinity River (STR), and Clear Creek (CLE). The southern sites include the North Fork Kaweah River (NKA), Sycamore Creek (SYC), Jose Creek (JOS), and Tyler Creek (TYL).

Table 4. Results from a linear mixed-effects model examining the effects of region, handling time, sex, mass, temperature, lymph presence, and body condition on plasma cholinesterase (ChE) levels
 EstimateStandard errordft valuep valueSite variance (residual variance)
      1.201 × 10−3 (0.020)
Region (South)−0.0430.0095−4.8250.005 
Handling time0.0000.000670.9560.343 
Sex (male)0.0130.007671.9610.054 
Sex (juvenile)−0.0070.00967−0.7660.446 
Cloaca temperature0.0000.001670.0330.974 
Lymph (yes)−0.0050.00667−0.8570.394 
Condition factor (K)−0.1070.26567−0.4050.687 

In addition to the seven lotic sites included in the statistical analysis, turtles were also captured from two stock ponds near Tyler Creek in the southern region. These results were not incorporated into the LME model because the lentic habitat at the Tyler Creek ponds was dissimilar to the lotic habitat of all other sites. At one stock pond sampled in late July 2011, four turtles (three females, one juvenile) were captured and had the following mean (±1 standard deviation) characteristics: carapace length (125.0 ± 41.4 mm), mass (313.8 ± 216.0 g), K (0.138 ± 0.04 kg/m3), cloaca temperature (25.2 ± 0.9°C), and ChE activity (0.104 ± 0.06 µmol min−1 g−1). At an additional stock pond sampled in early October 2011, two adult male turtles were captured resulting in the following mean (±1 standard deviation) carapace length (158.5 ± 19.1.4 mm), mass (604.3 ± 201.7 g), K (0.149 ± 0.0 kg/m3, cloaca temperature (19.1 ± 0.1°C), and ChE activity (0.075 ± 0.01 µmol min−1 g−1). Mean ChE activity of turtles at the Tyler Creek ponds was comparable to ChE activities of turtles captured at the southern lotic sites (Fig. 2).


Turtle TChE activity was significantly depressed by 31% in the southern study region compared with the northern region after accounting for known and potential sources of variation in ChE activity. In a similar study, TChE activity was used as a biomarker to infer pesticide exposure from nearby agricultural land use. This study found that an Argentina toad species (Bufo paracnemis) had significantly depressed TChE activity (37 and 84%) at two sites located near agricultural lands compared with a toad population located in a pristine reference site 35. A different study found another Argentina toad species (Chaunus schneideri) had plasma butyrylcholinesterase that was significantly depressed at agricultural sites compared with a reference site 36, and Sanchez-Hernandez 37 found that plasma butyrylcholinesterase was significantly reduced in lizards (Gallotia galloti palmae) occupying agricultural sites compared with reference sites in the Canary Islands.

Prior studies have identified additional methods for assessing the biological significance of ChE inhibition in relation to a reference condition. One is to quantify the proportion of individuals in the contaminated sites that have ChE activity two standard deviations below the reference site mean ChE activity 35. Another method is to identify the percentage of individuals that have ChE activities depressed by 20% of the reference mean 38. Of the turtles sampled in the southern region of the present study, 6.3% of the turtles had ChE activity below the two standard deviation threshold and 75% of the turtles had ChE activity below the 20% diagnostic threshold. In the Canary Islands, 23% of the lizards at one agricultural site were below the two standard deviation threshold 37. At another agricultural site in the same study, 7% of the lizards were below the two standard deviation threshold. Maul and Ferris 38 found that at sites treated with ChE inhibitors, 3% of northern cardinals (Cardinalis cardinalis) were below the two standard deviation threshold and 34% were below the 20% diagnostic threshold.

The effect that ChE depression has on E. marmorata at the individual or population level is unknown. Laboratory studies examining the physiology of this species would further our ability to predict the effect of ChE depression on E. marmorata in the wild. Sandahl et al. 4 found that chlorpyrifos-induced ChE inhibition resulted in decreased spontaneous swimming rate in Coho salmon (Oncorhynchus kisutch) when brain and muscle acetylcholinesterase was depressed by 23 and 12%, respectively. More pronounced deleterious effects to swimming and feeding behavior arose as chlorpyrifos concentrations increased and acetylcholinesterase activity decreased. Likewise, with significant depression of ChE activity present in the southern region, E. marmorata could exhibit loss of neurotransmission and neuromuscular function.

The LME model found significance in only one additional variable besides region. Male turtles had slightly significantly increased ChE activity compared with females. This is consistent with results of other studies with reptiles and should be considered as a natural source of variation in ChE activity 20. Also, environmental factors such as water temperature have long been identified as influencing ChE activity in aquatic organisms 39, but no other potentially influential variable was a significant predictor of turtle ChE activity in the LME.

In the DA analysis, region was a discriminating factor for size, body condition, and cloaca temperature. This analysis could suggest that depressed ChE in the southern region was responsible for differences in turtle body size and condition. However, the LME model found that mass, body condition, and cloaca temperatures were poor predictors of turtle ChE activity. An alternative explanation could be that there are additional between-region differences in turtle characteristics other than TChE activity. Geographic variation in body size has been previously described in this species, including larger turtles inhabiting the northern latitudes of its range 40.Therefore, there may be a natural region effect on ChE activity in lieu of pesticide exposure. Reference ChE activities are typically comparable among individuals of the same species. Cholinesterase activities also differ little among wild bird species related below the family level but are statistically different among bird families 41. Although still recognized as one species, recent genetic studies suggest a diverse phylogeny of E. marmorata in California. Turtles in both regions are from the same nuclear group, but mitochondrial grouping has identified turtles from sites in the north region as part of a northern clade and those in the southern region as part of a Central (San Joaquin) Valley clade 42. It is unknown whether these mitochondrial groupings would be the source of a large difference in normal ChE activities. A study that compared ChE activities of mountain plovers (Charadrius montanus) from two sites in California found that ChE activities were significantly depressed (32%) at clean reference sites in the Carrizo Plain compared with Central Valley polluted sites 43. These results suggest additional regional factors that may influence ChE because they contradict what would be expected based on site selection. The authors of the mountain plover study suggest that very small sample sizes and unknown physiological conditions could have impacted the results. Even so, the use of ChE is considered highly reliable for field toxicology. Multiple studies have used this biomarker for environmental assessment of organophosphate and carbamate contamination, and dose–response relationships have proved that ChE is a primary mechanism of toxicity for exposure 2.

The results of the present study are valuable given that this is the first to measure ChE activity in E. marmorata. Documented ChE values are beneficial for future biomonitoring of this widespread turtle species. Turtles at the northern sites should provide a reference condition for E. marmorata. The values at the southern site likely reflect chronic exposure to low levels of ChE inhibitors in seemingly pristine habitat. Prior studies in the southern Sierra Nevada have found ChE inhibitors in the air at concentrations ranging from ng m−3 to pg m−3 9, 14, in surface waters, rain, and snow at ng l−1 8, 9, 12, in lichen and vegetation at ng g−1 8, and in fish and larval amphibians at ng g−1 to pg g−1 7, 8, 44. These concentrations are below laboratory-determined lethal toxicities of some of the most sensitive species from this region 45. However, the true biological effect may manifest as sublethal and imperceptible behavioral alterations. Emys marmorata is also prevalent in some cultivated areas, including in sewage ponds, sloughs, rivers, and wetlands adjacent and downstream of direct agricultural pesticide applications 26. Future studies at monitoring sites with increased levels of pollution or potential events of acute lethal poisoning will be able to use the results of the present study for reference ranges.

The present study provides evidence that ChE depression in the southern region may be due to sublethal pesticide exposure. Depressed ChE could potentially contribute to sublethal effects in E. marmorata, including impaired behavior or altered physiological and immunological conditions. As a state species of special concern, sublethal exposure to pesticides should be considered as a potential conservation threat due to the widespread use of ChE-inhibiting compounds in California. Moreover, E. marmorata could be a valid biosentinel species to assess the widespread declines of foothill yellow-legged frogs (Rana boylii) in the southern Sierra Nevada Mountains. Disappearances of several ranid frog species have been linked to upwind pesticide use, and subsequent laboratory tests found that organophosphate pesticides can lead to mortality at very low concentrations in R. boylii 5, 45. During field sampling of turtles for the present study, R. boylii was never observed at the southern sites, but it was frequently encountered at all of the northern sites. Museum records and prior reports confirm its historical presence at some of the southern sites and throughout the southern Sierra Nevada region (Museum of Vertebrate Zoology, University of California, Berkeley, CA, USA; Because E. marmorata and R. boylii occupy similar range and habitat, reduced ChE activity in turtles in the southern region further an already compelling case that atmospheric deposition of pesticides contributed to region-wide extirpation of a once widespread amphibian. Similarly, because the function of ChE is common to all vertebrates and invertebrates, there may be additional taxa in the southern region facing harmful negative effects from sublethal concentrations of one or more ChE-inhibiting compounds.


We are appreciative of C. Sydoriak, D. Boiano, A. Esperanza, H. Werner, and E. Frenzel of Sequoia and Kings Canyon National Parks (SEKI) for supporting the project in many ways. We also thank J. Bettaso of the U.S. Fish and Wildlife Service, and J. and G. Versteeg and M. Provencio for extensive field and logistical support. We are also grateful to P. Crosbie of California State University-Fresno and the many volunteers who assisted in fieldwork. We are also thankful to Green Diamond Resource Company for landowner permission, L. Wilhoit of the California Department of Pesticide Regulation for providing pesticide use data, and D. Bradford for providing spatial data. The research described herein was funded by the U.S. National Park Service.