USGS - science for a changing world

Northern Prairie Wildlife Research Center

  Home About NPWRC Our Science Staff Employment Contacts Common Questions About the Site

Environmental Characteristics Associated with the Occurrence of Avian Botulism in Wetlands of a Northern California Refuge

Methods


All selected wetlands were drained in April–May 1986, and a 1.6-ha enclosure was built in each as previously reported (Rocke and Brand 1994). We placed 40–50 captive-reared mallards (males and females) in enclosures to act as sentinels for the occurrence of botulism in each wetland and to determine relative mortality rates from botulism (Rocke and Brand 1994). Sentinels were released in June or July (Sep for seasonally flooded wetlands) and maintained through October from 1987 to 1989. Free-flying birds had access to the uncovered enclosures at all times. All animal handling and husbandry protocols were approved by the NWHC Animal Care and Use Committee.

To recover any sick or dead sentinel birds, all enclosures were searched on foot or by canoe and with assistance of retrieving dogs, 3–4 times each week. If mortality was detected, wetland enclosures were searched daily. Healthy sentinel birds were counted during each search and trapped periodically to conduct an accurate inventory. Sick, dead, and missing sentinel birds were replaced with healthy birds to maintain a consistent number of bird-exposure-days.

All sick birds found within enclosures were removed; a blood sample was withdrawn from the jugular vein and tested for Type C botulinum toxin via the mouse neutralization test (Quortrup and Sudheimer 1943). Sick sentinel birds were then euthanized by cervical dislocation and necropsied. All intact fresh carcasses found within enclosures were also necropsied. Blood collected from the heart of each carcass was tested for botulinum toxin. Because botulinum toxin can form postmortem in carcasses, an enclosure was not confirmed to have had botulism mortality in birds if botulinum toxin was detected in only a single carcass. For enclosures where botulism outbreaks occurred, daily botulism mortality rates for specific time intervals (10–14 days) were calculated by dividing the number of sick and dead birds by the number of bird-exposure-days within the interval when mortality occurred (Samuel and Fuller 1994).

We collected environmental data from the water column and 2 strata in the sediments (S1, S2) of each wetland enclosure at 10–14-day sampling intervals. Water column samples were collected at 1 location along each of 5 transects equally spaced within each enclosure; sample locations were randomized for each sampling interval. At these collection sites, we pumped water from approximately 7.5 cm above the sediment-water interface via a hand pump operated from a small boat. Pumped water was allowed to run freely for a few minutes and then collected by running the flowing sample down the side of a collection jar to avoid aeration. Oxidation-reduction potential (redox potential) was measured immediately with a specific ion electrode (Orion Research, Boston, Massachusetts, USA). Standardized redox potential (redox potential corrected to pH 7.0 and a temperature of 25°C) was calculated according to the formula reported by Smoot and Pierson (1979). Specific conductance (Markson Science, Phoenix, Arizona, USA), pH (Model 51; Yellowsprings Instrument, Marion, Massachusetts, USA), and turbidity (Hach, Loveland, Colorado, USA) were measured via standard meters. After water column samples had been collected and processed from a small boat, we collected a core sample from the upper 2.5 cm of sediment via a 30.5- 2.54-cm stainless steel core sampler. These core samples were submitted to a local laboratory (Monarch Laboratories, Chico, California, USA) for determination of percent organic matter (POM) via standard techniques (American Public Health Association 1985).

Samples of interstitial water from the sediments (S1, S2) were collected at 5 permanent sampling stations arranged in an X-configuration to minimize disturbance of sediments prior to sampling and to ensure a representative sample. We used an interstitial water collection device described by Euliss and Barnes (1992) to collect samples from 2 sediment strata 2.5 cm below the sediment–water interface (S1) and 11.25 cm below the sediment–water interface (S2). Nitrogen gas was used to evacuate the samples to prevent contamination with atmospheric oxygen (Euliss and Barnes 1992). Water from the collecting device was slowly added down the sides of collection jars to avoid aeration, and redox potential, pH, and conductivity were measured with standard meters as described above.

We collected invertebrates from the water column and benthic level at each of the sites previously sampled for water column data via a modified 10-cm-diameter single-tube device described by Euliss et al. (1992) that was placed 5 cm into the substrate. Benthic samples were sectioned by hand as described by Mackay and Euliss (1994). Prior to storage in 80% ethanol, we used a modified self-cleaning screen to concentrate samples into residues (Euliss and Swanson 1989). Invertebrate samples were sorted by hand into taxonomic groups, enumerated, dried to a constant mass at 60°C, and weighed to the nearest milligram on an analytical balance. We used keys by Usinger (1956), Ward and Whipple (1959), Grodhaus (1967), Pennak (1978), and Merritt and Cummins (1984) to identify specimens.

Data Analysis

Means for each environmental variable measured within each enclosure were calculated for each 10–14-day sampling interval. We estimated missing values for several means (n = 10) via regression analysis with highly correlated subsets of variables (BMDP procedure AM; Dixon et al. 1988). Means of the environmental variables (environmental parameters) were inspected for outlying values and nonnormal data distributions. To stabilize the variance, samples of invertebrates (counts and biomass) were log10-transformed [e.g., log10(biomass + 1.0)] following the recommendations of Elliott (1977).

We used principal component analysis (SAS Institute 1989) to determine the variance structure among the 22 environmental parameters so as to aggregate highly correlated environmental measurements and to parsimoniously reduce the number of parameters for subsequent analyses (Morrison 1976). Principal components with eigenvalues >1.0 were selected from the 22 variables. We used varimax rotation to produce factors that maximized the contribution of each environmental parameter to a single factor (Cooley and Lohnes 1971), which provided enhanced biological interpretation of each rotated factor. The rotated environmental factors provided the basis for further analysis of the relation between environmental parameters and the occurrence of botulism in sentinel birds.

Using repeated measures analysis of variance (ANOVA; Kirk 1982), we tested for environmental differences between outbreak and nonoutbreak wetlands during 20 sampling intervals (6 from 1987, 6 from 1988, 8 from 1989). Scores from environmental factors (response variable) were analyzed by separate ANOVA. Changes among sampling intervals (the repeated measure) and sampling interval interactions with outbreak (or nonoutbreak) classification were also analyzed by ANOVA. We conducted all analyses using SAS PROC GLM with Type III sums of squares (SAS Institute 1989). We conducted F-tests for differences between outbreak and nonoutbreak wetlands with the mean square (MS) and degrees of freedom (df) for the wetland enclosures within each outbreak classification. The F-tests for differences among sampling intervals and for sampling interval and outbreak classification interactions were conducted with the MS and df for the interaction of sampling intervals and wetland enclosures within each outbreak classification.

We evaluated associations between environmental factors and the occurrence of botulism outbreaks in sentinel mallards during the 10–14-day intervals surrounding (5–7 days before and after) the collection of environmental samples from each outbreak wetland. A binary classification was used for each sampling interval with >1 botulism mortality or ≤1 mortality. In addition to environmental factors, we also included covariates for wetland enclosure, year, and autocorrelation (Bonney 1987) in the occurrence of botulism outbreaks between sequential sampling intervals. Analyses were conducted with the BMDPLR (Dixon et al. 1988) backwards stepwise logistic regression models. We used the procedures of Hosmer and Lemeshow (1989) in developing our results and assessing the appropriateness of our logistic regression models.

Finally, we determined the association between environmental factors and daily rates of botulism mortality in sentinel birds in wetland enclosures during outbreaks. We conducted a Pearson product-moment correlation analysis (SAS Institute 1989) between environmental factors and daily botulism mortality rates of sentinel birds during the 10–14-day intervals surrounding collection of environmental samples from each wetland enclosure.

Little is known about the relation between botulism outbreaks and environmental conditions. Because we considered our analyses exploratory in nature and because sample size was generally small, which resulted in lower statistical power, our results were considered significant at P ≤ 0.10.


Previous Section -- Study Area
Return to Contents
Next Section -- Results

Accessibility FOIA Privacy Policies and Notices

Take Pride in America logo USA.gov logo U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.npwrc.usgs.gov/resource/birds/botulsm/method.htm
Page Contact Information: Webmaster
Page Last Modified: Friday, 01-Feb-2013 18:14:35 EST
Reston, VA [vaww54]