Hematologic and serum biochemical reference intervals for free-ranging common bottlenose dolphins (Tursiops truncatus) and variation in the distributions of clinicopathologic values related to geographic sampling site

Lori H. Schwacke NOAA Ocean Service, Hollings Marine Laboratory, 331 Ft Johnson Rd, Charleston, SC 29412.

Search for other papers by Lori H. Schwacke in
Current site
Google Scholar
PubMed
Close
 PhD
,
Ailsa J. Hall Sea Mammal Research Unit, Gatty Marine Laboratory, University of St Andrews, Fife, Scotland, KY16 8LB.

Search for other papers by Ailsa J. Hall in
Current site
Google Scholar
PubMed
Close
 PhD
,
Forrest I. Townsend Bayside Hospital for Animals, 251 NE Racetrack Rd, Fort Walton Beach, FL 32547.

Search for other papers by Forrest I. Townsend in
Current site
Google Scholar
PubMed
Close
 DVM
,
Randall S. Wells Chicago Zoological Society, c/o Mote Marine Laboratory, 1600 Ken Thompson Pkwy, Sarasota, FL 34236.

Search for other papers by Randall S. Wells in
Current site
Google Scholar
PubMed
Close
 PhD
,
Larry J. Hansen NOAA Fisheries, Southeast Fisheries Science Center, 101 Pivers Island Rd, Beaufort, NC 28516.

Search for other papers by Larry J. Hansen in
Current site
Google Scholar
PubMed
Close
 MS
,
Aleta A. Hohn NOAA Fisheries, Southeast Fisheries Science Center, 101 Pivers Island Rd, Beaufort, NC 28516.

Search for other papers by Aleta A. Hohn in
Current site
Google Scholar
PubMed
Close
 PhD
,
Gregory D. Bossart Harbor Branch Oceanographic Institute at Florida Atlantic University, Fort Pierce, FL 24946.

Search for other papers by Gregory D. Bossart in
Current site
Google Scholar
PubMed
Close
 VMD, PhD
,
Patricia A. Fair NOAA Ocean Service Center for Coastal Environmental Health and Biomolecular Research, 219 Ft Johnson Rd, Charleston, SC 29412.

Search for other papers by Patricia A. Fair in
Current site
Google Scholar
PubMed
Close
 PhD
, and
Teresa K. Rowles NOAA Fisheries, Office of Protected Resources, 1315 E West Hwy, Silver Spring, MD 20910.

Search for other papers by Teresa K. Rowles in
Current site
Google Scholar
PubMed
Close
 DVM, PhD

Click on author name to view affiliation information

Abstract

Objective—To develop robust reference intervals for hematologic and serum biochemical variables by use of data derived from free-ranging bottlenose dolphins (Tursiops truncatus) and examine potential variation in distributions of clinicopathologic values related to sampling sites' geographic locations.

Animals—255 free-ranging bottlenose dolphins.

Procedures—Data from samples collected during multiple bottlenose dolphin capture-release projects conducted at 4 southeastern US coastal locations in 2000 through 2006 were combined to determine reference intervals for 52 clinicopathologic variables. A nonparametric bootstrap approach was applied to estimate 95th percentiles and associated 90% confidence intervals; the need for partitioning by length and sex classes was determined by testing for differences in estimated thresholds with a bootstrap method. When appropriate, quantile regression was used to determine continuous functions for 95th percentiles dependent on length. The proportion of out-of-range samples for all clinicopathologic measurements was examined for each geographic site, and multivariate ANOVA was applied to further explore variation in leukocyte subgroups.

Results—A need for partitioning by length and sex classes was indicated for many clinicopathologic variables. For each geographic site, few significant deviations from expected number of out-of-range samples were detected. Although mean leukocyte counts did not vary among sites, differences in the mean counts for leukocyte subgroups were identified.

Conclusions and Clinical Relevance—Although differences in the centrality of distributions for some variables were detected, the 95th percentiles estimated from the pooled data were robust and applicable across geographic sites. The derived reference intervals provide critical information for conducting bottlenose dolphin population health studies.

Abstract

Objective—To develop robust reference intervals for hematologic and serum biochemical variables by use of data derived from free-ranging bottlenose dolphins (Tursiops truncatus) and examine potential variation in distributions of clinicopathologic values related to sampling sites' geographic locations.

Animals—255 free-ranging bottlenose dolphins.

Procedures—Data from samples collected during multiple bottlenose dolphin capture-release projects conducted at 4 southeastern US coastal locations in 2000 through 2006 were combined to determine reference intervals for 52 clinicopathologic variables. A nonparametric bootstrap approach was applied to estimate 95th percentiles and associated 90% confidence intervals; the need for partitioning by length and sex classes was determined by testing for differences in estimated thresholds with a bootstrap method. When appropriate, quantile regression was used to determine continuous functions for 95th percentiles dependent on length. The proportion of out-of-range samples for all clinicopathologic measurements was examined for each geographic site, and multivariate ANOVA was applied to further explore variation in leukocyte subgroups.

Results—A need for partitioning by length and sex classes was indicated for many clinicopathologic variables. For each geographic site, few significant deviations from expected number of out-of-range samples were detected. Although mean leukocyte counts did not vary among sites, differences in the mean counts for leukocyte subgroups were identified.

Conclusions and Clinical Relevance—Although differences in the centrality of distributions for some variables were detected, the 95th percentiles estimated from the pooled data were robust and applicable across geographic sites. The derived reference intervals provide critical information for conducting bottlenose dolphin population health studies.

Reference intervals for hematologic and serum biochemical variables are important in assessments of animal health. Clinicopathologic test results for individual animals are compared with the distribution of values from a reference population to diagnose disease, to make treatment or management decisions, or as part of a broader physiologic assessment. In situations in which a marine mammal becomes stranded, an initial health assessment, including evaluation of clinicopathologic variables, is a preliminary and critical step in determining a course of treatment and assessing the suitability of the animal for rehabilitation.1 Clinicopathologic analyses are also basic and crucial components of population health assessment investigations. In recent decades, several such studies2–6 involving marine mammals, and specifically bottlenose dolphins, have been performed. These studies are becoming more common as scientists have realized the importance of investigating population health of marine mammals for effective conservation and management6 and have focused research on understanding the implications of stressors such as infectious disease and marine toxins and pollutants on the health of marine mammal populations.

Reference intervals are generally defined as the central 95% of a distribution of values obtained from samples collected from individuals that are selected randomly from the population. A clinicopathologic test result for a stranded or free-ranging animal can then be compared with the reference interval to provide perspective for the measured value; by comparison of the measured value with expected values from the population, information is gained that can aid in selecting appropriate diagnostic or management procedures for the animal. This process has long been established as a standard clinical practice in both human and veterinary medicine. To establish reliable reference intervals that are both accurate and precise, the collection of representative samples in sufficient numbers and the selection of an appropriate statistical approach are issues that must be considered. Such issues have been the focus of much research,7–10 and recommendations have been proposed by the IFCC for the establishment of laboratory reference intervals for use in human medicine.11 Unfortunately, because of logistic considerations, the proposed IFCC recommendations for human medicine have not been widely applied for the establishment of clinicopathologic reference intervals for wildlife species.

The selection of an appropriate reference population is important and is strongly dependent on the purpose for which the reference intervals are to be defined. When the goal is to define baseline reference intervals for use in distinguishing novel disease or unanticipated health conditions in wild populations, then individuals should be selected from the general population of live, free-ranging animals. Those animals are not likely to be disease free or completely lacking clinical issues. However, if the animals are randomly selected, then the samples should be representative of the general population and will provide the most appropriate baseline data.

Aspects of the distributions of many clinicopathologic values have been described for some cetacean species,12–17 and in particular for both captive18 and wild2,3,19 populations of common bottlenose dolphins (Tursiops truncatus). Those studies were primarily focused on evaluation of factors that affect values of clinicopathologic variables, such as age, sex, or season; thus, differences in the means of distributions among cohorts were examined. Although a difference between 2 or more means may be of interest for investigating shifts in the overall distribution of values, it does not necessarily have implications for the definition of reference intervals, which are determined by the tails of the distribution. Two distributions may have similar means but very different 95th percentiles, or similar percentiles but different means. Therefore, a difference in means between groups detected via standard statistical methods, such as ANOVA, should not necessarily determine whether the groups should be partitioned for the calculation of reference intervals. This point has been explored extensively for clinicopathologic reference intervals in human populations, specifically for variables that do not follow standard Gaussian distributions, and alternative statistical approaches to the classic ANOVA for comparing intervals among cohorts and determining the need for partitioning have been proposed.20,21

On a similar note, although a mild departure from a Gaussian distribution may not significantly affect the calculation of means or affect statistical tests of differences in means (eg, the F test), it can significantly influence estimation of percentiles for the tails of the distribution. For this reason, nonparametric methods for the estimation of clinicopathologic reference values have been recommended.11 Nonparametric methods allow for the estimation of percentiles without a distribution assumption; thus, they are less vulnerable to the influence of outliers. A nonparametric approach is particularly appropriate for highly skewed variable distributions, such as those that are often associated with clinicopathologic values.

Two nonparametric approaches have been commonly used: a simple, rank-based calculation on the sorted sample and a bootstrap (ie, resampling) procedure that also allows for the calculation of uncertainty bounds. Aside from the benefit of allowing for estimation of confidence intervals, the bootstrap approach has been recommended to provide the greatest precision and efficiency, thereby requiring smaller sample sizes.22 Another consideration is the inclusion of appropriate samples in sufficient numbers. Particularly for nonparametric estimation, a large sample size is essential to adequately represent the tails of the distribution and obtain accurate estimates of outer percentiles. Generally, a minimum sample size of 40 animals is recommended.23 Preliminary clinicopathologic reference ranges reported for bottlenose dolphins were based on a relatively small sample size.13,24 More recent studies18,19 have established reference ranges that were based on larger sample sizes, but those studies focused on effects of biological or environmental factors and included > 1 sample from the same individuals. For testing relationships between variables and a specific factor of interest, an experimental design that includes repeated sample collections from individuals is highly appropriate—it controls for individual variation and increases power to detect relationships. However, repeated sample collections from the same individuals are not appropriate for the definition of reference intervals and may result in an underestimation of interindividual variation. Unique samples from many individuals provide a better representation of the overall population. Unfortunately, because of the cost and effort associated with the collection of samples from wild animal populations, most studies have acquired only a limited number of unique samples from individuals. In addition, those studies were designed to target a particular study area, so the generalization of results for animals that become stranded or that are located outside of the study area is uncertain. In an attempt to address these issues, we proposed to undertake a pooled analysis approach involving values obtained from multiple, independent studies conducted across a range of geographic sites.

The purpose of the study reported here was to develop robust reference intervals for hematologic and serum biochemical variables by use of data derived from free-ranging bottlenose dolphins and to examine potential variation in distributions of clinicopathologic values related to geographic location of sampling sites. In 2000 through 2006, several capture-release projects involving wild bottlenose dolphins were conducted at various geographic sites along the southeastern US coast as part of baseline health or stock assessment studies. Although the studies were conducted by independent researchers, protocols for collection of samples and selection of a diagnostic laboratory were coordinated among the projects in an effort to ensure consistent sample quality and interstudy comparability.a For purposes of the study, previously published and unpublished data from capture-release studies performed at 4 sites along the southeastern US coast were combined: Sarasota Bay, Fla6,19; Indian River Lagoon, Fla3; Charleston, SC2; and Beaufort, NC. Selection of dolphins for sample collection was not necessarily random in that specific age classes were often targeted. However, selections were not made on the basis of an observed disease state; therefore, the samples from the studies should be reasonably representative of ostensibly normal individuals from seemingly stable stocks. The studies also provided a large number of unique samples from individual wild dolphins in the southeastern United States. For the combined data set, the need for sex and age-class stratification was evaluated via assessment of differences in estimated reference thresholds rather than distribution means. The derived reference intervals from the pooled data were applied to evaluate the proportion of out-of-range samples for each of the 4 populations and to examine the validity of generalizing ranges between geographically distinct populations. The intent was to define reference ranges that would provide critical baseline data for the evaluation of unexpected disease states among stranded dolphins. Furthermore, the defined reference ranges would facilitate epidemiologic investigations that are aimed at evaluating the prevalence of specific disease conditions in wild dolphin populations.

Materials and Methods

Dolphins and blood sample collection—Methods used for capture of dolphins and blood sample collection in the studies performed in Sarasota Bay,6,19 the IRL,3 and Charleston2 have been previously described. Collection of samples near Beaufort followed similar protocols. Protocols were comparable among all studies. Briefly, dolphins were captured by encircling them with a net and restraining them by hand. Once restrained, every effort was made to perform an ultrasonographic examination to determine pregnancy status for adult females. Late-term pregnant females were immediately released; for early-term pregnant females, blood sample collection was performed, and they were then released. Blood samples were collected from the ventral fluke vasculature by use of a 19-gauge butterfly catheter.b The time at which the dolphin was first encircled and the time at which blood samples were collected were recorded, and the elapsed time (capture to sample collection) was computed. For serum biochemical analyses, blood samples were collected in a 10-mL serum separator tube, slowly inverted several times, and then immediately refrigerated in a cooler on board the research vessel for at least 20 minutes before being centrifuged. For Beaufort, Charleston, and IRL studies, samples were centrifuged in an onboard centrifuge within approximately 40 minutes of collection. Samples collected from the Sarasota Bay study were taken back to an onshore laboratory for processing; thus, it was possible that some of those samples were held in refrigeration for several hours prior to centrifugation.

Samples for hematologic analysis were collected in an evacuated tubec containing EDTA as an anticoagulant and immediately stored in a cooler or refrigerator on board the research vessel. Upon returning to the dock each day, samples were shipped overnight to the diagnostic laboratory. In several of the studies, blood was collected twice from the same dolphin during the health assessment. The first blood collection occurred prior to physical examination and any other sample collection or diagnostic procedures; a second blood sample was obtained just before release of the dolphin to investigate potential alterations in clinicopathologic variables attributable to the stress of brief capture and restraint. For our analysis, only results from the initial blood collection were included. Furthermore, if duplicate samples were obtained from a given individual on multiple days or during > 1 season, only the first sample was included. All analyses were carried out by a single laboratory.d Potential interlaboratory bias is a concern in comparing data obtained at different sites because significant variation for certain variables among laboratories has been reported.6,19 For this reason, we chose to include only samples processed by a single laboratory.

At the selected laboratory, the equipment used for analysis of samples was consistent during all of the sample collection periods. Hematologic analyses were performed by use of an automated analyzer,e and serum biochemical analyses were performed by use of an automated biochemical analyzer.f Leukocyte counts and leukocyte subgroup counts were determined manually via microscopic examination of blood smears stained with modified Wright stain.g

The IRL study targeted 2 separate areas of the lagoon: a northern area that included the Mosquito Lagoon and portions of the Indian River and Banana River above 28°15'0″N latitude and a southern area that included the St Lucie Inlet and portions of the Indian River south of 27°25'0″N latitude.25 Only samples from the northern study area were included in this analysis. Samples from the southern area were excluded because health assessment efforts in the southern portion of the lagoon had been initiated in part because of skin disease among dolphins that had been observed during prior photo-identification studies; those health assessment efforts confirmed an unusually high prevalence of lobomycosis in dolphins in the southern IRL.25 Exclusion of those data was done to avoid potential sample bias that could be introduced by inclusion of data from animals that were specifically targeted because of a suspected high prevalence of a specific disease that could potentially alter clinicopathologic values.

Samples were stratified into 3 groups according to lengths of the dolphins; length groups were selected to approximately represent calves (< 200 cm in length), juveniles or subadults (≥ 200 cm and < 240 cm in length), and adults (≥ 240 cm in length).26 Length (as a measure of body size) was used rather than age to stratify samples because length can be readily measured on stranded or live-captured dolphins, whereas determination of age most often requires the extraction of a tooth to count dentinal growth layers, the results of which are not immediately known. Furthermore, length is measurable on a fine scale and may correlate better with clinicopathologic measurements, particularly during the very early years or at an age near the time of sexual maturity, when rapid growth spurts occur.27

Among the data, outlier values were examined by use of previously described methods.23 Briefly, the difference between the highest (or lowest) value and the second highest (or lowest) value was calculated. The value was deemed an outlier if the difference was greater than a third of the range of all values.

Serum electrophoresis was not conducted on samples analyzed prior to 2003. Furthermore, laboratory protocols for the interpretation of results of electrophoresis in 2003 differed from the protocols used in subsequent years; thus, the threshold values used in 2003 may have differed. Because of this potential inconsistency, only electrophoresis samples that were analyzed in 2004 or later were included.

Calculation of reference intervals—As a preliminary step, the need for partitioning each clinicopathologic variable by length class was assessed by use of a bootstrap approach, which uses repeated random resampling of the original observations with replacement. Briefly, a sample was selected, with replacement, from each of the original data sets for calves, juveniles and subadults, and adults (males and females combined), and upper (97.5th) and lower (2.5th) percentiles were computed for each sample. The difference between each pair of samples (calves, juveniles and subadults, and adults) was computed and stored electronically. The process was repeated 1,000 times, providing 1,000 differences for each variable (upper and lower percentiles) between each pair of groups (calves, juveniles and subadults, and adults). This provided the sampling distribution of differences from which a 95% confidence interval was computed. The difference between 2 groups was considered significant if the 95% confidence interval around the difference did not include zero. Variables that differed significantly among those groups were graphed as a function of length and examined for evidence that a dependency between each of those values and length still existed even after partitioning into calf, juvenile-subadult, and adult groups. In such instances, regression-based reference thresholds have been proposed,28 so linear quantile regression29 was used to compute the 95th conditional quantile function for each variable as a function of length.

Results of prior studies3,18,19 of dolphins, as well as other species, have suggested sex-based differences in clinicopathologic values for adult animals. Therefore, the same method applied to determine the need for length class partitioning was also applied to determine whether some variables required partitioning between sexes for adult (> 240 cm in length) dolphins.

All bootstrap analyses were coded by use of a specialized programming language.h Quantile regression was conducted by use of a specialized statistical package.i Many of the clinicopathologic variables were not independent (eg, hemoglobin, Hct, MCV, MCH, MCHC); such variables typically covary, and individuals that have a value that is out of range for 1 variable are likely to have values that are out of range for other related variables. Although separate statistical tests were conducted for each variable, the need for partitioning certain variables as groups or panels of related variables rather than as independent variables was examined.

For reference threshold point estimates, the original data were resampled 1,000 times, and percentiles were computed. Reference thresholds were computed as the median of the computed percentiles from the 1,000 resamples. This follows recommendations of an expert panel organized by the IFCC.11,30

Generalization of ranges across populations—The fact that values of clinicopathologic variables could vary among dolphin cohorts from different geographic areas as a result of genetic variability or acclimation to unique environmental conditions was a concern. Such variation among populations has been described for values of clinicopathologic variables in humans.31 Furthermore, variation in hematologic values between coastal and offshore morphotypes of bottlenose dolphins has also been described32 and is likely related to diving adaptations in offshore species that occupy deeper water habitats. Although all of the samples used in this analysis were from estuarine or coastal morphotype dolphins, we were interested in verifying that the reference values generated from the pooled samples were appropriate when applied to each of the geographic cohorts.

Several methods have been proposed to examine potential analytic bias in the use of common reference intervals derived from pooled data collected from multiple cohorts.20,21,33 Most involve specific criteria to examine the percentages of each subgroup outside of the reference intervals of the combined distribution. For purposes of this study, a similar approach was adopted—the computed reference intervals were applied to determine the number of out-of-range samples for each of the populations. If measurements were truly homogeneous across sites, then among samples from each geographic site, one would expect 2.5% to be greater than and 2.5% to be less than the computed reference intervals. Subsequently, the probability of obtaining the observed number of out-of-range samples greater and less than the interval for each of the sampled populations was examined, given that the true probability was 0.025, by applying a binomial distribution. Because both the upper and lower thresholds were tested for each variable, a Bonferroni adjusted critical value (A = 0.05/2 = 0.025) was used for the binomial test. It should be noted that because of the limited number of samples from each population (maximum number of samples, 103), it would be impossible to conclude that the number of observed out-of-range samples was less than the expected number. It would require at least 146 samples to obtain statistical significance (ie, the probability of obtaining zero out-of-range samples from 146 samples when the true probability is 0.025 is 0.0248). All analyses to examine proportions of out-of-range samples were conducted by use of a statistical packagej for computing binomial probabilities. Serum electrophoresis values were excluded from this portion of the analysis because of the limited number of samples.

A MANOVA was also conducted to examine variation in the centrality (mean) of distributions for absolute counts of leukocyte subgroups among geographic sites. Length class was included as a covariate. Before the MANOVA was conducted, homogeneity of variances was examined by use of a Levene test for each dependent variable; when a deviation from this assumption was indicated (P < 0.05), then data were logarithmically transformed. The MANOVA was followed with univariate F tests for each of the dependent variables. Pairwise comparisons were made by use of a Tukey honestly significant difference test for unequal sample sizes. The MANOVAs and associated analyses were conducted by use of computer software.k For all analyses, a value of P < 0.05 was considered significant.

Results

Dolphins—Initially, single samples from each of 263 dolphins were available for analysis. Ultrasonographic or physical examination revealed that 7 females were pregnant; data from those dolphins were excluded from the analyses. One additional sample was excluded because of a lack of information regarding length and age of the dolphin. Thus, 255 samples (obtained from 255 dolphins) were available for use in the study. A C2 test was conducted to examine differences in the number of samples from adult versus nonadult (calf, juvenile, and subadult) dolphins among geographic sites. An additional C2 test was conducted with data from the nonadults only to examine the distribution of samples for calf versus juvenile classes (Table 1). The median elapsed time between capture and blood sample collection for all samples was 21 minutes (interquartile [25th to 75th] range, 15 to 29 minutes).

Table 1—

Number of samples obtained from 255 free-ranging bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006.

SiteSample collection periodLength classTotal No. of dolphins/site
No. of calves (%)No. of juveniles and subadults (%)Total No. of nonadult dolphins* (%)No. of adults (%) 
Sarasota Bay, Fla2000–200618 (17)37 (44)55 (53)48 (47)103
Northern IRL, Fla2003–20065 (8)36 (63)41 (66)21 (34)62
Charleston, SC2003–200519 (27)23 (44)42 (59)29 (41)71
Beaufort, NC20066 (32)7 (54)13 (68)6 (32)19
Total No. of dolphins/group48103151104  

Data are stratified by length class (ie, dolphins were assigned to 1 of 3 groups according to their length); length groups were selected to approximately represent calves (< 200 cm in length), juveniles and subadults (r 200 cm and < 240 cm in length), and adults (r 240 cm in length). Distribution of samples from adult versus nonadult length classes did not differ (P = 0.34) among sites. For the nonadults, the distribution of samples from calves versus juveniles and subadults did differ significantly (P < 0.01) between sites because of the low number of IRL calves that were evaluated. However, when the χ2 test was repeated excluding IRL samples, sample distributions were no longer significantly (P = 0.39) different.

Number of calves, juveniles, and subadults combined.

Single outlying values for serum concentrations of magnesium (2.7 mEq/L) and sodium (180 mEq/L) and activity of CK (1,255 U/L) were identified; 2 outlying values were determined for serum ALT activity (185 and 120 U/L). In addition to the aforementioned outlying value in each of those 5 samples, the value of at least 1 related variable was the minimum or maximum value for that variable among all samples. For example, the sample with the outlying value for serum sodium concentration also had the maximum serum chloride concentration, and the sample with the higher outlying value for serum ALT activity also had the maximum serum LDH and AST activities. Therefore, all 5 samples were removed entirely from the data set.

Reference intervals were computed for 52 clinicopathologic variables (Tables 2–4). Not all of the 52 variables were assessed in every sample; thus, the total number of samples evaluated for each variable varied. The number of samples available for assessment of a few variables was limited following stratification for length and sex, and the nonparametric estimation of percentiles was inappropriate. When the available number of samples was < 40, minimum and maximum values were determined rather than 95th percentiles; in those instances, 90% confidence intervals could not be calculated.

Table 2—

Computed reference thresholds (2.5th and 97.5th percentiles) and associated 90% confidence intervals (CIs) of clinicopathologic variables computed for free-ranging calf and juvenile or subadult bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006. When available sample size is < 40, only minimum and maximum values are given; thus, CIs could not be determined.

VariableCalfJuvenile
No. of dolphinsLower threshold value90% CI for lower threshold valueUpper threshold value90% CI for lower threshold valueNo. of dolphinsLower threshold value90% CI for lower threshold valueUpper threshold value90% CI for lower threshold value
Hematologic analyses
   Erythrocytes (× 106 cells/μL)462.92.8–3.243.8–4.2993.23–3.243.9–4.8
   Hemoglobin (g/dL)4612.311–12.715.515–15.59912.911.5–13.115.815.6–18.2
   Hct (%)463433–374645–46993735–374747–50
   MCV (fL)469897–103133130–1349910196–106127124–129
   MCH (pg)463331–374444–46993635–374443–44
   MCHC (g/dL)463230–323736–38993230–333736–38
   RDW (%)4611.211.1–11.515.514.8–15.510111.311.1–11.515.914.7–16.6
   Leukocytes (× 103 cells/μL)467.96.4–8.516.314.5–16.8996.45.8–7.318.115.2–20.5
   Neutrophils (× 103 cells/μL)451.61.5–27.96.3–9.3992.21.8–2.67.56.7–8.9
   Lymphocytes (× 103 cells/μL)451.20.8–1.65.75.1–7.1991.10.8–1.14.53.9–5.6
   Monocytes (× 103 cells/μL)450.0NA1.30.8–1.4990.0NA0.80.6–0.9
   Eosinophils (× 103 cells/μL)451.71.3–1.88.16.1–8.8991.31.3–1.58.46.5–10.6
   Basophils (× 103 cells/μL)450.0NA0.30.2–0.3990.0NA0.60.4–0.6
   Platelets (× 103 platelets/μL)46140108–164314260–3349611073–119277238–331
Serum biochemical analyses
   Glucose (mg/dL)478078–86153142–1731016647–75141130–177
   Cholesterol (mg/dL)3788ND268ND71102100–118217198–219
   Triglyceride (mg/dL)3457ND177ND804742–53148122–178
   Sodium (mEq/L)47150150–152162158–162101151149–151158158–164
   Potassium (mEq/L)473.33.3–3.54.94.5–4.91013.13.1–3.44.54.2–5.3
   Chloride (mEq/L)47107104–109119116–122101107106–108120119–121
   Calcium (mg/dL)478.98.6–9.110.410.0–10.71018.78.5–8.810.210.2–10.5
   Phosphate (mg/dL)473.52.8–4.46.76.3–7.11013.63–3.86.76.3–7.1
   Magnesium (mEq/L)471.21.1–1.31.81.6–1.91011.21.1–1.31.81.7–1.9
   ALP (U/L)4711975–1891012886–11151039641–131702591–873
   ALT (U/L)3713ND75ND73128–177066–75
   AST (U/L)47135130–145300287–316103165153–183371333–450
   SDH (U/L)341ND36ND7011–24632–56
   LDH (U/L)37323ND612ND73377353–403611583–627
   GGT (U/L)471514–183228–32103127–153433–36
   Total bilirubin (mg/dL)440.0NA0.20.1–0.21030.0NA0.20.2–0.2
   Direct bilirubin (mg/dL)440.0NA0.10.1–0.11030.0NA0.10.1–0.1
   Indirect bilirubin (mg/dL)440.0NA0.10.1–0.11030.0NA0.20.1–0.2
   BUN (mg/dL)474040–458574–871014744–498078–87
   Creatinine (mg/dL)470.60.5–0.71.41.3–1.51010.70.6–0.81.51.4–1.6
   BUN-to-creatinine ratio4131.428.6–40.911387.1–1149033.831.3–41.310093.8–110
   Total protein (g/dL)475.95.7–6.48.67.8–9.21016.45.3–6.58.38.2–9.2
   Albumin (g/dL)4743.9–4.14.94.9–5.11013.83.5–4.054.9–5.2
   Globulin (g/dL)471.11.0–1.54.13.3–5.11011.91.6–2.13.83.7–4.9
   Albumin-to-globulin ratio470.950.80–1.44.23.2–4.91011.10.76–1.22.42.2–3.2
   Iron (μg/dL)465756–72282246–2921016549–67215186–240
   TIBC (μg/dL)47243243–245513461–579101183171–191419381–498
   Saturation (%)471711–238064–941012315–276256–77
   Uric acid (mg/dL)370.10ND2.9ND710.140.10–0.202.01.6–2.3
   CK (U/L)47138134–148494382–49910110482–119358283–371
Serum electrophoresis
   Albumin (g/dL)242.93ND4.36ND463.13.1–3.34.34.1–4.4
   α1-Globulin (g/dL)240.17ND0.84ND460.20.2–0.21.20.7–1.2
   α2-Globulin (g/dL)240.81ND1.41ND460.60.3–0.81.31.2–1.4
   Total α-globulin (g/dL)241.14ND1.81ND461.11.1–1.21.81.7–1.9
   Total β-globulin (g/dL)240.29ND0.79ND460.40.4–0.40.70.6–0.7
   γ-Globulin (g/dL)240.33ND3.07ND460.80.8–1.032.8–3.7
   Total globulin (g/dL)241.88ND5.52ND462.72.3–3.14.94.6–5.8
   Total protein (g/dL)245.91ND9.19ND466.56.4–6.78.38.1–9.2

NA = Not applicable. ND = Not determined.

Table 3—

Computed reference thresholds (2.5th and 97.5th percentiles) and associated 90% CIs of clinicopathologic variables computed for free-ranging adult bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006. Only variables for which partitioning by sex was not required are shown.

VariableNo. of dolphinsLower threshold value90% CI for lower threshold valueUpper threshold value90% CI for upper threshold value      
Hematologic analyses      
   Leukocytes (× 103 cells/μL)1017.14.8–7.717.515.6–18.7      
   Neutrophils (× 103 cells/μL)1002.52.3–2.89.96.9–10.7      
   Lymphocytes (× 103 cells/μL)1000.60.3–0.74.23.4–4.5      
   Monocytes (× 103 cells/μL)1000.0NA10.8–1      
   Eosinophils (× 103 cells/μL)1001.81.4–2.28.17.1–8.2      
   Basophils (× 103 cells/μL)1000.0NA0.40.3–0.6      
   Platelets (× 103 platelets/μL)9810486–109253240–277      
Serum biochemical analyses      
   Glucose (mg/dL)1016020–71121112–124      
   Cholesterol (mg/dL)718888–114236224–237      
   Triglyceride (mg/dL)763535–41134122–162      
   Sodium (mEq/L)101152151–153160159–161      
   Potassium (mEq/L)1013.23–3.34.54.4–4.7      
   Chloride (mEq/L)101107106–109124119–129      
   Calcium (mg/dL)1018.58.5–8.710.110.0–10.6      
   Phosphate (mg/dL)1013.32.5–3.66.86.3–7.0      
   Magnesium (mEq/L)1011.21.2–1.31.71.7–1.8      
   ALP (U/L)1015551–59342274–454      
   ALT (U/L)711514–196659–82      
   AST (U/L)101160118–169586369–733      
   SDH (U/L)6641–54531–62      
   LDH (U/L)71329325–351530512–596      
   GGT (U/L)1011511–163330–38      
   Total bilirubin (mg/dL)1010.0NA0.20.2–0.3      
   Direct bilirubin (mg/dL)1010.0NA0.10.1–0.1      
   Indirect bilirubin (mg/dL)1010.0NA0.20.1–0.3      
   Total protein (g/dL)1016.76.5–6.88.88.6–8.9      
   Albumin (g/dL)1013.83.8–3.95.14.9–5.4      
   Globulin (g/dL)1012.12.1–2.24.54.4–5      
   Albumin-to-globulin ratio1010.90.8–1.02.22.1–2.6      
   Uric acid (mg/dL)710.10.1–0.11.40.9–2.3      
   CK (U/L)1019165–94213190–325      
Serum electrophoresis      
   Albumin (g/dL)423.12.9–3.14.24–4.4      
   α1-Globulin (g/dL)420.10.1–0.21.10.5–1.1      
   α2-Globulin (g/dL)420.60.6–0.81.41.3–1.4      
   Total α-globulin (g/dL)421.21.2–1.21.81.7–1.8      
   Total β-globulin (g/dL)420.40.3–0.40.60.6–0.7      
   γ-Globulin (g/dL)421.01.0–1.43.02.5–3.6      
   Total globulin (g/dL)422.92.6–3.35.34.6–5.3      
   Total protein (g/dL)426.86.7–78.78.1–8.8      

See Table 2 for key.

Table 4—

Computed reference thresholds (2.5th and 97.5th percentiles) and associated 90% CIs of clinicopathologic variables computed for free-ranging adult male and female bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006. Only variables for which partitioning by sex was required are shown.

VariableAdult maleAdult female
No. of dolphinsLower threshold value90% CI for lower threshold valueUpper threshold value90% CI for upper threshold valueNo. of dolphinsLower threshold value90% CI for lower threshold valueUpper threshold value90% CI for upper threshold value
Hematologic analyses
   Erythrocytes (× 106 cells/μL)593.33.2–3.34.24.1–4.3443.13.0–3.23.93.8–4
   Hemoglobin (g/dL)5912.612.1–13.315.715.4–15.94412.011.5–12.615.515.1–16.0
   Hct (%)593737–374746–47443533–374645–47
   MCV (fL)59103102–105124121–12944106103–110131126–134
   MCH (pg)593535–364241–43443634–384343–45
   MCHC (g/dL)593232–333736–37443230–333736–37
   RDW (%)5911.511.3–11.716.315.2–16.64411.110.7–11.515.213.5–18.7
Serum biochemical analyses
   BUN (mg/dL)594240–467672–79444341–477572–80
   Creatinine (mg/dL)591.01.0–1.11.91.8–2.0440.90.8–1.01.61.4–1.7
   BUN-to-creatinine ratio5325.722.2–29.362.259–644127.924.1–36.282.469.0–88.9
   Iron (Mg/dL)594232–56144130–171446155–72156145–165
   TIBC (Mg/dL)59166147–178300290–30844192181–222388358–413
   Saturation (%)591915–247357–93442116–255954–64

The need for stratification on the basis of length class was indicated for approximately half (21/52) of the clinicopathologic variables; for those variables, 1 or more of the threshold bounds (ie, lower, upper, or both) differed significantly among 2 or more length classes. Variables with different thresholds among length classes included kidney- and liver-associated analytes (serum BUN and creatinine concentrations and ALP, AST, LDH, and GGT activities), lymphocyte count, and serum concentrations of glucose, cholesterol, triglycerides, total globulin, and iron. Because of the number of variables that differed significantly among length classes and to simplify application, all of the variables were partitioned by length class.

For 4 variables (serum ALP and CK activities, iron concentration, and TIBC), there was a dependency related to length that was not completely accounted for by the partitioning of calves and juvenile-subadult length classes. Quantile regression was conducted for these 4 variables to determine 95th conditional quantile functions for the dolphins < 240 cm (calf and juvenile-sub-adult length classes; Figure 1). Regression coefficients for the 2.5th and 97.5th percentiles were significant for serum CK activity and TIBC, indicating a relationship between length and both lower and upper reference thresholds for those variables (Table 5). For ALP activity and iron concentration, only the regression of the 97.5th percentile was significant, indicating a relationship between length and the upper threshold only.

Figure 1—
Figure 1—

Quantile regression functions for serum ALP (A [n = 251]) and CK (B [249]) activities, iron concentration (C [250]), and TIBC (D [251]) with respect to length in free-ranging bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006. Outer boundaries of the hashed area represent the 2.5th and 97.5th percentiles; center line represents standard regression of length versus the variable of interest.

Citation: American Journal of Veterinary Research 70, 8; 10.2460/ajvr.70.8.973

Table 5—

Results of regression analysis of serum ALP activity (n = 251), CK activity (249), iron concentration (250), and TIBC (249) as a function of body length in free-ranging bottlenose dolphins that were evaluated during capture-release projects conducted at 4 southeastern US coast locations in 2000 through 2006.

VariableLower threshold valueUpper threshold value
InterceptP valueβP valueInterceptP valueβP value
ALP activity407.80.051−1.5200.1252775.3< 0.001−9.667< 0.001
CK activity355.2< 0.001−1.120< 0.0011353.30.001−4.6810.003
Iron concentration32.3910.5110.1300.574801.680.001−2.7930.001
TIBC559.67< 0.001−1.667< 0.0011004.30.025−2.714< 0.001

Lower and upper threshold values were determined on the basis of the 2.5th and 97.5th percentiles. Significance of regression coefficients was set at a value of P < 0.05 (ie, α = 0.05).

Because of the limited number of samples from which serum electrophoresis variables were available (n = 42; adult values combined), sex class differences were not examined. Of the remaining 44 clinicopathologic variables, 5 differed significantly between sex classes for adult dolphins. Specifically, erythrocyte indices varied such that males had a higher upper threshold for erythrocyte count and higher upper and lower thresholds for Hct, compared with females. Females had higher upper and lower thresholds for TIBC and a higher lower threshold for serum iron concentration than males. Males had a higher upper threshold for serum creatinine concentration. Adult values were further partitioned by sex for those clinicopathologic panels that had significantly different upper or lower thresholds (Table 4).

Hemolysis indices were examined to verify that hemolysis of samples did not significantly interfere with results of serum biochemical analyses. None of the samples were classified by the laboratory as markedly hemolyzed, and only 2 of the 249 samples were classified as moderately hemolyzed (index, 100 to 400). Fifty (20%) of the samples were classified as slightly hemolyzed (index, 20 to 100), and the remaining 197 (79%) samples had hemolysis indices of < 20. All samples with hemolysis indices ≥ 20 were removed, and the reference interval calculations for serum iron and uric acid concentrations and CK, AST, LDH, ALP, and GGT activities were repeated because laboratory personnel had determined that those variables are potentially altered if the extent of hemolysis is severe. The recomputed reference thresholds were highly similar, and all were within the 90% confidence limits for the original estimates.

Geographic variation—Analyses of the number of out-of-range samples for each variable at each geographic site revealed few deviations from the expected proportions. The only significant deviation was for MCV; among IRL samples, a higher than expected number of values for MCV (5/61 = 0.08; P = 0.01) were less than the lower reference limit.

Results of MANOVA of leukocyte subgroup counts—Although the proportion of out-of-range samples did not differ significantly among geographic sites, variation in the centrality of the distributions for leukocyte subgroup counts was evident. For the MANOVA, logarithmic transformation was required for all dependent variables (neutrophil, lymphocyte, monocyte, eosinophil, and basophil counts) to meet assumptions of homogeneity of variance. Results of the MANOVA indicated a significant (P < 0.001) overall effect of geographic site on leukocyte subgroup counts. Individual univariate F tests revealed that counts for all leukocyte subgroups (with the exception of monocytes) varied significantly among geographic sites (P = 0.003, P = 0.009, P < 0.001, P = 0.016, and P = 0.092 for neutrophils, lymphocytes, eosinophils, basophils, and monocytes, respectively). Among length classes, significant (P < 0.001) differences were identified for only lymphocytes and eosinophils. Results of pairwise multivariate comparisons among geographic sites indicated that leukocyte subgroup counts did not differ significantly between the 2 Carolina sites (Charleston and Beaufort; P = 0.306) nor between the 2 Florida sites (Sarasota and IRL; P = 0.108). However, all but one of the pairwise comparisons between Florida sites and Carolina sites yielded significant differences in leukocyte subgroup counts: Charleston versus IRL (P < 0.001), Charleston versus Sarasota Bay (P < 0.001), and Beaufort versus IRL (P = 0.003). These differences were significant even with a Bonferroni adjustment to the critical threshold value for multiple comparisons (adjusted threshold, 0.008). Comparison of leukocyte counts at the Sarasota and Beaufort sites did not yield any significant difference (P = 0.054). Pairwise comparisons were also made for each leukocyte subgroup (Figure 2). Higher neutrophil counts and lower lymphocyte and eosinophil counts were detected at the Florida sites (northern IRL and Sarasota Bay), compared with findings at the Carolina sites. However, differences in neutrophil counts were significant only between Charleston and the 2 Florida sites, and differences in lymphocyte and eosinophil counts were significant only between Charleston and the IRL site. In part, this was likely a result of the limited sample size for Beaufort.

Figure 2—
Figure 2—

Least squares means ± 95% confidence intervals for leukocyte subgroup counts in free-ranging bottlenose dolphins that were evaluated during capture-release projects2,3,6,19 conducted at 4 southeastern US coast locations in 2000 through 2006. Data are reported for each site (Beaufort, n = 15; Charleston, 68; northern IRL, 60; and Sarasota Bay, 103).a–f Within a leukocyte subgroup, different letters indicate heterogeneous groups as determined by use of a Tukey honestly significant difference test for unequal sample sizes; a single homogeneous group was identified for monocyte and for basophil values.

Citation: American Journal of Veterinary Research 70, 8; 10.2460/ajvr.70.8.973

Discussion

In the present study, reference ranges for 52 clinicopathologic variables in free-ranging common bottlenose dolphins were determined on the basis of data obtained from 250 samples (after removal of 5 outliers) collected from 250 wild dolphins that were captured and released at 4 different sites along the southeastern US coast. Our analysis revealed a strong need for partitioning of samples by length class for the calculation of most reference intervals. The variables for which differences among length classes were detected in the present study are those for which length-associated variation in bottlenose dolphins2,3,6,18 and other mammalian species34–36 has been previously suggested. We also determined a need to partition some groups of variables (eg, kidney-associated analytes, erythrocyte indices, and iron measures) by sex class for adult animals. These variable groups are generally stratified by sex for calculation of reference ranges in humans and have been reported to differ between sexes in other mammalian wildlife populations.37,38 Although the analysis approach used in the present study combined data from multiple sites, which dramatically increases sample size for a more robust analysis, the need for partitioning by sex and length classes suggests that further refinements could be made in the future as the number of samples collected and number of sample collection sites investigated expand. This emphasizes a need for a centralized repository for health information for not only bottlenose dolphins, but also for other marine wildlife species as well.

The investigations2,3,6,19 that contributed data to the present study all involved the same diagnostic laboratory. Caution should be used in applying the computed reference intervals for values reported by other laboratories because interlaboratory bias can impact some variables. In prior studies,6,19 the concordance of dolphin clinicopathologic values among specific laboratories was evaluated, and results indicated that assessments of only approximately half of the examined variables were comparable. Specifically, manual total leukocyte and differential counts were generally comparable among laboratories, but assessments of some variables, such as hemoglobin concentration or other erythrocyte indices that are determined by use of an automated analyzer, were affected by laboratory bias. Such bias is likely to be greater when different instruments are used by the laboratories. Serum biochemical variables were even more vulnerable to potential interlaboratory bias; Hall et al19 reported that only glucose, calcium, phosphate, and albumin concentrations were comparable among 3 independent laboratories. Although interlaboratory variation does not preclude comparison of data that are analyzed by different laboratories, it does at least indicate the need to consider the potential bias inherent in such comparisons for some variables.

For purposes of pooling of data, sample collection and processing protocols were coordinated among the investigations2,3,6,19 that were used as the basis of the present study. However, because of logistic difficulties in collection of samples from wild dolphins, some variation in sample collection and handling still occurred. Specifically, although blood samples were centrifuged within 40 minutes of collection, samples from Sarasota Bay dolphins were held in refrigeration until delivered to the onshore laboratory, which in some instances may have been a period of several hours. One potential effect of this delay in sample centrifugation is excessive hemolysis. We investigated the degree of hemolysis in samples and found that most samples had little or no hemolysis. Moreover, there was no apparent effect of the slightly hemolyzed samples on the estimated reference thresholds; reference values computed after removal of all samples with a hemolysis index > 20 did not differ from the original estimates that were computed on the basis of the entire data set. Another potential issue associated with delay in serum separation is the development of artificially low glucose concentration in the sample as a result of continued glycolytic activity of blood cells.13,39 Five of 103 serum samples from Sarasota Bay dolphins had glucose concentrations less than the computed lower reference threshold, and no samples had values greater than the upper reference threshold. Although the number of out-of-range samples did not significantly differ from the expected number (values were less than the lower threshold in 4/99 [4%] samples and were greater than the upper threshold in 0/99 [0%] samples), the estimated P values (P = 0.072 and 0.074, respectively) were close to a critical value of 0.05, suggesting a potential bias in the assessments of serum glucose concentration in samples from Sarasota Bay that was related to delay in serum processing.

In contrast, 2 of 18 samples from Beaufort had high glucose concentrations, and no samples had values less than the lower reference limit for this variable. Because serum glucose concentrations increase during capture or restraint and have been positively correlated with the duration of animal holding prior to blood sample collection,39 we examined the elapsed time between capture and blood sample collection for the individual dolphins from Beaufort that had out-of-range serum glucose concentrations. The interval between capture and sample collection for the 2 individuals was 59 and 36 minutes. These values were high, compared with the intervals for other dolphins (median interval for all sites combined, 21 minutes; interquartile [25th to 75th percentile] range, 15 to 29 minutes). Furthermore, elapsed time for completion of blood sample collection at the Beaufort site was in general higher than findings at the other sample collection sites. Sample collection in Beaufort was for the most part carried out in deeper waters and was therefore logistically more difficult, resulting in a longer interval between initial capture and blood sample collection. Results of at least 1 other study39 have indicated that circulating glucose concentration increases with increasing duration of restraint for individual dolphins. Taken collectively, these results reinforce findings of previous studies12,13,39,40 that have examined the potential influence of sample collection and processing methods on some serum biochemical variables and emphasize that even though efforts are made to collect samples as soon as possible after capture and to process samples within a timely manner as logistics allow, sampling variables must be considered when interpreting sensitive serum biochemical variables such as glucose concentration. The implication for the present study is that the computed reference thresholds and confidence intervals are likely wider than desired and wider than what would be obtained if samples could be collected and processed under ideal conditions. As a follow-up analysis, separate ANOVAs were conducted to determine whether lag time between capture and blood sample collection affected other clinicopathologic variables (data not shown). Serum glucose concentration was the only variable that was significantly influenced by blood collection lag time (P < 0.001 for glucose concentration and P ≥ 0.05 for all other variables).

Aside from the analysis and removal of outliers in the present study, no effort was made to exclude nonhealthy individuals, leaving open the possibility that derived reference thresholds do not represent a strictly disease-free population. The intent of this analysis was to establish reference ranges for wild dolphin populations to assist in the evaluation and management of stranded animals and to aid in the assessment of disease conditions in dolphin cohorts that are presumably exposed to high levels of chemical, biological, or physical stressors. Therefore, the aim was to determine reference thresholds that reflect the population of live, free-ranging dolphins, which, like any other wild population of animals, is not likely to be completely disease free. For this purpose, inclusion of nonhealthy individuals is acceptable, and in fact, exclusion of such individuals would introduce bias into the results. This question has been evaluated and debated for human populations and has led to a similar conclusion that the inclusion of nonhealthy subjects is sometimes both acceptable and necessary.8,10

Some distributional differences were evident among geographic sites in the present study. Specifically, differences in the centrality of distributions for immunerelated variables (eg, leukocyte subgroup counts) were detected. Total leukocyte count did not differ among dolphins at the 4 geographic sites, but mean counts of leukocyte subgroups did differ among sites, particularly between the Florida and Carolina sites. The difference in leukocyte subgroup counts could relate to environmental factors; dolphins in different areas may be exposed to different pathogens or have differing susceptibilities to pathogens. Neutrophils provide the major defense against pyogenic bacteria, and counts may fluctuate in response to bacterial infection.13,41 The higher neutrophil counts in dolphins in the warmer waters of the Florida sites (Sarasota and IRL) may reflect an increased exposure to certain bacterial pathogens. Conversely, eosinophils, which were highest in dolphins in the more northern Carolina sites, are generally associated with a parasiticidal role,13,42 although eosinophils have also been suggested to mediate bactericidal effects43,44 and their function in bottlenose dolphin immune response is not well understood.13 As well as potentially different pathogen exposures, exposures to chemical contaminants also differ among these geographic sites.45–47 Persistent organochlorine pollutants have an immuno-modulatory effect in many mammalian species,48 and a recent study49 revealed that such persistent organochlorine pollutants are inversely correlated with neutrophil counts in humans. The results of the MANOVA in the present study indicated a shift in the centrality of the distributions of leukocyte subpopulation counts but did not necessarily indicate a significant increase in the number of dolphins with leukocyte subgroup counts outside of the operating (reference) limits at any geographic site. There were not unexpectedly high numbers of measured values outside the computed reference intervals for any leukocyte subgroup count for any of the sites. We conclude that the differences detected among the sites did not preclude the use of a generalized set of reference intervals for multiple regions, but indicated that caution should be taken when applying the reference thresholds for certain clinicopathologic variables in dolphins from areas in which free-ranging individuals have not yet been evaluated and geographic variation is not well understood. The 90% confidence intervals are an indication of the uncertainty in a given threshold and should be considered when comparisons are made. When a high degree of variability or uncertainty exists in the computed reference threshold, it is manifested as a wide confidence interval around the estimated threshold point values.

Reference intervals are a necessary and critical diagnostic tool used in the care of stranded animals and for population health investigations. Although baseline reference intervals for each geographic site could be more sensitive for detecting out-of-range values (ie, cases), acquisition of sufficient numbers of samples for determination of precise intervals in each site is not practical. In a few instances, long-term monitoring has established a solid understanding of site-specific reference intervals,19 but such intensive monitoring across a broad range of sites such as that investigated in the present study would be unrealistic. Perhaps the greatest need for an understanding of the distributions of clinicopathologic values in dolphins comes from investigations of unusual deaths or morbidity events, defined as a significant increase in the magnitude or change in the nature, pattern, or demographic structure of deaths and illnesses, compared with prior records. Among bottlenose dolphins, unusual deaths or morbidity events have often occurred in areas such as the northern Gulf of Mexico,50,51 in which there was little or no monitoring or population health investigation prior to the event. Health assessments of surviving dolphins in investigations following episodes of unusual deaths or illnesses have been conducted in the Florida panhandle by 2 of the authors (TKR and LHS) and along the Texas coast,5 but the interpretation of findings is difficult without an understanding of baseline measurements from a reference population. Generalized ranges that are based on a large sample size and are analyzed to determine variables that could potentially vary among sample collection sites, such as those established in the study of this report, provide a necessary baseline for comparisons.

ABBREVIATIONS

ALT

Alanine aminotransferase

ALP

Alkaline phosphatase

AST

Aspartate aminotransferase

CK

Creatine kinase

GGT

γ-Glutamyltransferase

IFCC

International Federation of Clinical Chemistry and Laboratory Medicine

IRL

Indian River Lagoon

LDH

Lactate dehydrogenase

MANOVA

Multivariate ANOVA

MCV

Mean corpuscular volume

MCH

Mean corpuscular hemoglobin

MCHC

Mean corpuscular hemoglobin concentration

RDW

Red cell distribution width

SDH

Sorbitol dehydrogenase

TIBC

Total iron-binding capacity

a.

Schwacke LH, Hall AJ, Wells RS, et al. Health and risk assessment for bottlenose dolphin (Tursiops truncatus) populations along the southeast United States coast: current status and future plans (Paper SC/56/E20). 56th Annu Meet Int Whaling Commission, Sorrento, Italy, July 2004.

b.

Becton-Dickinson, Franklin Lakes, NJ.

c.

BD 6452, Becton-Dickinson, Franklin Lakes, NJ.

d.

Animal Health Diagnostic Laboratory, Cornell University, Ithaca, NY.

e.

ADVIA 120 hematology system, Bayer Diagnostics, Tarrytown, NY.

f.

Hitachi 917 chemistry analyzer, Roche Diagnostics Corp, Indianapolis, Ind.

g.

HEMA-TEK Stain Pack, Bayer HealthCare, Tarrytown, NY.

h.

R statistical computation and graphics system, version 2.4.1, R Foundation for Statistical Computing, Vienna, Austria. Available at: www.r-project.org/. Accessed Mar 2, 2007.

i.

Quantile regression, R statistical computation and graphics system, version 4.06, R Foundation for Statistical Computing, Vienna, Austria. Available at: www.r-project.org/. Accessed Mar 2, 2008.

j.

Dbinom function, R statistical computation and graphics system, version 2.4.1, R Foundation for Statistical Computing, Vienna, Austria. Available at: www.r-project.org/. Accessed Mar 2, 2007.

k.

Statistica, version 7.1, StatSoft Inc, Tulsa, Okla.

References

  • 1.

    Moore M, Early G, Touhey K, et al. Rehabilitation and release of marine mammals in the United States: risks and benefits. Mar Mamm Sci 2007;23:731750.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2.

    Fair P, Hulsey T, Varela R, et al. Hematology, serum chemistry, and cytology findings from apparently healthy Atlantic bottlenose dolphins (Tursiops truncatus) inhabiting the estuarine waters of Charleston, South Carolina. Aquat Mamm 2006;32:182195.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3.

    Goldstein JD, Reese E, Reif JS, et al. Hematologic, biochemical, and cytologic findings from apparently healthy Atlantic bottlenose dolphins (Tursiops truncatus) inhabiting the Indian River Lagoon, Florida, USA. J Wildl Dis 2006;42:447454.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4.

    Hansen L, Wells R. Bottlenose dolphin health assessment: field report on sampling near Beaufort, North Carolina, during July, 1995. NOAA technical memorandum NMFS-SEFSC-382 1996. Miami: NOAA/NMFS Southeast Fishery Science Center, 1996.

    • Search Google Scholar
    • Export Citation
  • 5.

    Sweeney J. Veterinary assessment report, Tursiops truncatus, Matagorda Bay, Texas, July, 1992. NOAA contract report MIA-92/93–41 1992. Miami: NOAA/NMFS Southeast Fishery Science Center, 1992.

    • Search Google Scholar
    • Export Citation
  • 6.

    Wells RS, Rhinehart HL, Hansen LJ, et al. Bottlenose dolphins as marine ecosystem sentinels: developing a health monitoring system. EcoHealth 2004;1:246254.

    • Search Google Scholar
    • Export Citation
  • 7.

    Klee GG. Clinical interpretation of reference intervals and reference limits. A plea for assay harmonization. Clin Chem Lab Med 2004;42:752757.

    • Search Google Scholar
    • Export Citation
  • 8.

    Ritchie RF, Palomaki G. Selecting clinically relevant populations for reference intervals. Clin Chem Lab Med 2004;42:702709.

  • 9.

    Horn PS, Feng L, Li YM, et al. Effect of outliers and non-healthy individuals on reference interval estimation. Clin Chem 2001;47:21372145.

  • 10.

    Rustad P, Felding P, Lahti A. Proposal for guidelines to establish common biological reference intervals in large geographical areas for biochemical quantities measured frequently in serum and plasma. Clin Chem Lab Med 2004;42:783791.

    • Search Google Scholar
    • Export Citation
  • 11.

    Solberg HE. The IFCC recommendation on estimation of reference intervals. The RefVal program. Clin Chem Lab Med 2004;42:710714.

  • 12.

    Tryland M, Brun E. Serum chemistry of the minke whale from the northeastern Atlantic. J Wildl Dis 2001;37:332341.

  • 13.

    Bossart G, Reidarson T, Dierauf L, et al. Clinical pathology. In: Dierauf L, Gulland F, eds. CRC handbook of marine mammal medicine. Boca Raton, Fla: CRC Press Inc, 2001;383486.

    • Search Google Scholar
    • Export Citation
  • 14.

    Heidel JR, Philo LM, Albert TF, et al. Serum chemistry of bowhead whales (Balaena mysticetus). J Wildl Dis 1996;32:7579.

  • 15.

    Tryland M, Thoresen SI, Kovacs KM, et al. Serum chemistry of free-ranging white whales (Delphinapterus leucas) in Svalbard. Vet Clin Pathol 2006;35:199203.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16.

    Koopman HN, Westgate AJ, Read AJ. Hematology values of wild harbor porpoises (Phocoena phocaena) from the Bay of Fundy, Canada. Mar Mamm Sci 1999;15:5264.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 17.

    Koopman HN, Westgate AJ, Read AJ, et al. Blood-chemistry of wild harbor porpoises Phocoena-Phocoena (L). Mar Mamm Sci 1995;11:123135.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 18.

    Venn-Watson S, Jensen ED, Ridgway SH. Effects of age and sex on clinicopathologic reference ranges in a healthy managed Atlantic bottlenose dolphin population. J Am Vet Med Assoc 2007;231:596601.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 19.

    Hall AJ, Wells RS, Sweeney JC, et al. Annual, seasonal and individual variation in hematology and clinical blood chemistry profiles in bottlenose dolphins (Tursiops truncatus) from Sarasota Bay, Florida. Comp Biochem Physiol A Mol Integr Physiol 2007;148:266277.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 20.

    Lahti A, Petersen PH, Boyd JC, et al. Objective criteria for partitioning Gaussian-distributed reference values into subgroups. Clin Chem 2002;48:338352.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 21.

    Lahti A, Petersen PH, Boyd JC, et al. Partitioning of nongaussian-distributed biochemical reference data into subgroups. Clin Chem 2004;50:891900.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 22.

    Linnet K. Nonparametric estimation of reference intervals by simple and bootstrap-based procedures. Clin Chem 2000;46:867869.

  • 23.

    Lassen ED. Perspectives in data interpretation. In: Thrall MA, ed. Veterinary hematology and clinical chemistry. Philadelphia: Lippincott Williams & Wilkins, 2004;4554.

    • Search Google Scholar
    • Export Citation
  • 24.

    Asper ED, Cornell LH, Duffield DA, et al. Hematology and serum chemistry values in bottlenose dolphins. In: Leatherwood S, Reeves RR, eds. The bottlenose dolphin. San Diego: Academic Press Inc, 1990;235244.

    • Search Google Scholar
    • Export Citation
  • 25.

    Reif JS, Mazzoil MS, McCulloch SD, et al. Lobomycosis in Atlantic bottlenose dolphins from the Indian River Lagoon, Florida. J Am Vet Med Assoc 2006;228:104108.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 26.

    McFee WE, Hopkins-Murphy SR. Bottlenose dolphin (Tursiops truncatus) strandings in South Carolina, 1992–1996. Fish Bull 2002;100:258265.

    • Search Google Scholar
    • Export Citation
  • 27.

    Mattson MC, Mullin KD, Ingram GW, et al. Age structure and growth of the bottlenose dolphin (Tursiops truncatus) from strandings in the Mississippi sound region of the north-central Gulf of Mexico from 1986 to 2003. Mar Mamm Sci 2006;22:654666.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 28.

    Virtanen A, Kairisto V, Irjala K, et al. Regression-based reference limits and their reliability: example on hemoglobin during the first year of life. Clin Chem 1998;44:327335.

    • Search Google Scholar
    • Export Citation
  • 29.

    Koenker R, d'Orey V. Computing regression quantiles. Appl Stat 1987;36:383393.

  • 30.

    Solberg HE. The theory of reference values part 5. Statistical treatment of collected reference values. Determination of reference limits. J Clin Chem Clin Biochem 1983;21:749760.

    • Search Google Scholar
    • Export Citation
  • 31.

    Rustad P, Felding P, Lahti A, et al. Descriptive analytical data and consequences for calculation of common reference intervals in the Nordic Reference Interval Project 2000. Scand J Clin Lab Invest 2004;64:343369.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 32.

    Duffield D, Ridgway S, Cornell L. Hematology distinguishes coastal and offshore forms of dolphins (Tursiops). Can J Zool 1983;61:930933.

  • 33.

    Harris EK, Boyd JC. On dividing reference data into subgroups to produce separate reference ranges. Clin Chem 1990;36:265270.

  • 34.

    Harewood WJ, Gillin A, Hennessy A, et al. Biochemistry and haematology values for the baboon (Papio hamadryas): the effects of sex, growth, development and age. J Med Primatol 1999;28:1931.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 35.

    Taylor MRH, Holland CV, Spenser R, et al. Haematological reference ranges for schoolchildren. Clin Lab Haematol 1997;19:115.

  • 36.

    Quinto L, Aponte JJ, Sacarlal J, et al. Haematological and biochemical indices in young African children: in search of reference intervals. Trop Med Int Health 2006;11:17411748.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 37.

    Swaminathan R, Ho CS, Chu LM, et al. Relation between plasma creatinine and body size. Clin Chem 1986;32:371373.

  • 38.

    Ekanayake DK, Horadagoda NU, Sanjeevani GK, et al. Hematology of a natural population of toque macaques (Macaca sinica) at Polonnaruwa, Sri Lanka. Am J Primatol 2003;61:1328.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 39.

    Varela RA, Schwacke L, Fair PA, et al. Effects of duration of capture and sample handling on critical care blood analytes in free-ranging bottlenose dolphins. J Am Vet Med Assoc 2006;229:19551961.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 40.

    Tryland M. ‘Normal' serum chemistry values in wild animals. Vet Rec 2006;158:211212.

  • 41.

    Roitt I. Essential immunology. 8th ed. Oxford: Blackwell Scientific Publications, 1994;247255.

  • 42.

    Smith JA. Molecular and cellular properties of eosinophils (a review). Ric Clin Lab 1981;11:181193.

  • 43.

    Carreras E, Boix E, Rosenberg HF, et al. Both aromatic and cationic residues contribute to the membrane-lytic and bactericidal activity of eosinophil cationic protein. Biochemistry 2003;42:66366644.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 44.

    Jong EC, Henderseon WR, Klebanoff SJ. Bactericidal activity of eosinophil peroxidase. J Immunol 1980;124:13781382.

  • 45.

    Hansen LJ, Schwacke LH, Mitchum GB, et al. Geographic variation in polychorinated biphenyl and organochlorine pesticide concentrations in the blubber of bottlenose dolphins from the US Atlantic coast. Sci Total Environ 2004;319:147172.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 46.

    Houde M, Pacepavicius G, Wells RS, et al. Polychlorinated biphenyls and hydroxylated polychlorinated biphenyls in plasma of bottlenose dolphins (Tursiops truncatus) from the Western Atlantic and the Gulf of Mexico. Environ Sci Technol 2006;40:58605866.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 47.

    Houde M, Wells RS, Fair PA, et al. Polyfluoroalkyl compounds in free-ranging bottlenose dolphins (Tursiops truncatus) from the Gulf of Mexico and the Atlantic Ocean. Environ Sci Technol 2005;39:65916598.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 48.

    Ross PS. The role of immunotoxic environmental contaminants in facilitating the emergence of infectious diseases in marine mammals. Hum Ecol Risk Assess 2002;8:277292.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 49.

    Lee DH, Jacobs D, Kocher T. Associations of serum concentrations of persistent organic pollutants with prevalence of periodontal disease and subpopulations of white blood cells. Environ Health Perspect 2008;116:15581562.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 50.

    Gaydos J. Bottlenose dolphins and brevetoxins: a coordinated research and response plan. NOAA report NMFS-OPR-32 2007. Silver Spring, Md: NOAA/NMFS Office of Protected Resources, 2007.

    • Search Google Scholar
    • Export Citation
  • 51.

    Gulland FMD, Hall AJ. Is marine mammal health deteriorating? Trends in the global reporting of marine mammal disease. EcoHealth 2007;4:135150.

    • Crossref
    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 99 0 0
Full Text Views 856 591 58
PDF Downloads 271 150 15
Advertisement