https://orcid.org/0000-0003-1547-5677
Figures
Abstract
Madagascar’s radiated tortoises (Astrochelys radiata) are critically endangered, threatened by illegal collection, and confiscated in alarming numbers in recent years. Robust population- and technique-specific hematology and biochemistry reference intervals are valuable yet heretofore missing tools for triage, rehabilitation, and reintroduction of confiscated radiated tortoises. We determined reference intervals in 120 previously confiscated, clinically healthy subadult radiated tortoises living under human care within their native habitat at the Tortoise Conservation Center (TCC). Specific analytes measured were manual packed cell volume, total solids, white blood cell (WBC) count and differentials, and biochemistry analytes using a point of care system. To evaluate the effects of different commonly used techniques on these analytes, we compared results between two venipuncture sites (subcarapacial sinus and brachial vein) and three different WBC quantification methods (Natt and Herrick, Leukopet, and slide estimate). Reference intervals were narrower for most analytes, and sodium and potassium were qualitatively higher in the TCC population compared to previously published values from radiated tortoises housed in North American institutions. Creatine kinase, aspartate aminotransferase, glucose and inorganic phosphorus were all significantly greater in brachial samples than in subcarapacial samples. There was poor agreement and evidence of constant and/or proportional bias between all WBC quantification methods. Differences based on time of sample collection were incidentally found in some analytes. These results highlight the need for considering technique, demographic, and environmental factors in creating and applying reference intervals, and contribute foundational knowledge for improving care of radiated tortoises throughout the confiscation-to-release pathway.
Citation: Brenn-White M, Raphael BL, Rakotoarisoa NAT, Deem SL (2022) Hematology and biochemistry of critically endangered radiated tortoises (Astrochelys radiata): Reference intervals in previously confiscated subadults and variability based on common techniques. PLoS ONE 17(3): e0264111. https://doi.org/10.1371/journal.pone.0264111
Editor: Jorge Ramón López-Olvera, Universitat Autònoma de Barcelona, SPAIN
Received: March 19, 2021; Accepted: February 3, 2022; Published: March 14, 2022
Copyright: © 2022 Brenn-White et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All data are included in a single file (a.radiate_bloodwork) available from Zenodo at http://doi.org/10.5281/zenodo.4614664 (DOI 10.5281/zenodo.4614664). Data and Read Me file including metadata have been uploaded.
Funding: Primary funding was provided by grant #2019.07 from the Saint Louis Zoo WildCare Institute Field Research for Conservation grant (Principal investigator: MBW, Co-investigators: SLD and NATR). Salary for MBW was provided by an anonymous donor to the Saint Louis Zoo. Support for additional project members was provided by the Saint Louis Zoo, Turtle Survival Alliance, and Wildlife Conservation Society Bronx Zoo. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: All authors have declared that no competing interests exist.
Introduction
Radiated tortoises (Astrochelys radiata) are critically endangered chelonians endemic to the spiny forests of southern Madagascar [1]. Habitat loss and poaching for trade and food are considered the primary threats to conservation of the species [1–5]. The scale of poaching is reflected in the massive confiscations of radiated tortoises that have taken place in recent years with over 10,000 tortoises recovered in a single 2018 event [6,7]. Over 24,000 confiscated radiated tortoises are currently in rehabilitation facilities in Madagascar, all managed by the Turtle Survival Alliance (TSA), the primary organization responsible for the care of rescued tortoises. Along with measures to minimize illegal collection, successful treatment, rehabilitation, and reintroduction of these and future confiscated radiated tortoises are critical to reversing the steep decline of this species in the wild.
Baseline biomedical information is essential to ensuring effective rehabilitation and safe reintroduction of wildlife, including chelonian species [8–10]. Hematology and biochemistry reference intervals based on clinically healthy tortoises are necessary for improving treatment of confiscated animals, selecting healthy candidates for release, and making evidence-based decisions about triage in confiscation events if and when resources allow [11–15]. Species and population-specific reference intervals also provide foundational information for subsequent research evaluating the efficacy of rehabilitation protocols and how these analytes may vary under different environmental conditions.
Robust hematology and biochemistry reference intervals for radiated tortoises have not previously been determined. Blood samples from more than 40 healthy individuals from a comparable reference population are recommended to establish de novo reference intervals with an absolute minimum of 20 individuals and 120 or more considered ideal [16–18]. To date, hematology and biochemistry research published for this species has been descriptive and based on sample sizes below 19 [19,20]. Additionally, prior published values and unpublished reference intervals have been based entirely on mixed age groups of individuals bred and/or managed within North American institutions [19–21]. As hematology and biochemistry values of reptile species change across demographic and environmental factors such as ambient temperature, humidity, rainfall patterns, food availability, and diet, there may be important differences between previously available hematology and biochemistry data and values in previously-confiscated tortoises being rehabilitated and conditioned for release in their native habitat [22–24].
Hematology and biochemistry values can vary not only by population, but also by sample collection and analytical methods. Choice of venipuncture site alone or in combination with lymph contamination has been shown to alter hematologic and biochemical analytes in a number of chelonian species [25–31]. Similarly, great variability in leukocyte counts using different quantification methods has been demonstrated in Galápagos tortoises (Chelonoidis spp.), eastern box turtles (Terrapene carolina carolina), and marine turtle species [32–36]. Veterinary and para-veterinary practitioners working with radiated tortoises may need to choose from multiple venipuncture sites and analytic methods based on patient size and temperament, practitioner training, cost and availability of supplies, and field conditions. Understanding how hematology and biochemistry analytes vary by technique is required to accurately apply population-specific reference intervals to interpretation of individual results.
To address these knowledge gaps in radiated tortoise health, our study had three objectives: 1) establish population-specific hematology and biochemistry reference intervals for previously confiscated, clinically healthy subadult radiated tortoises living under human care within their natural range; 2) evaluate differences in hematology and biochemistry values in blood from subcarapacial and brachial venipuncture sites within this population (between-animal comparison); and 3) evaluate agreement between three common leukocyte quantification methods within this population (within-animal comparison).
Study site & population
In February of 2020, we examined and sampled radiated tortoises housed at the TCC, a TSA rehabilitation and captive care facility located in the Ala Mahavelo spiny forest in southern Madagascar within the radiated tortoise’s current range [5]. For security reasons, GPS coordinates of the TCC cannot be provided.
We selected study subjects from a population of 1,000 juvenile and subadult radiated tortoises (carapacial length 15–26 cm) [37] undergoing pre-release health screening. All tortoises were illegally collected from their native habitat, confiscated by law enforcement, and placed under TSA management for rehabilitation and conditioning for release. Time at the TCC prior to sampling was between six months and two years for all individuals. Only subadults were included in this study as this is the age group targeted for this and subsequent releases. Sex of these tortoises is unknown as sexual dimorphism is not reliably apparent in this age group. The tortoises are housed year-round in 1000m2 open pens filled with native vegetation and natural substrate. Diets are supplemented with locally cultivated plants and native vegetation foraged by neighboring communities.
Study subjects, design, & sample collection
To allow time for same-day laboratory analyses, we screened the first 8 to 18 tortoises examined between 6:30 AM and 11:00 AM each day for inclusion in this study. Occasional afternoon screening and sampling of tortoises occurred between 4:00 PM and 5:30 PM when time permitted. Screening in the field consisted of a physical examination by a veterinarian. This included routine visual inspection and palpation, oral and cloacal examination, and coelomic palpation as size and temperament allowed. We collected oral-cloacal swabs for post-hoc screening for Mycoplasma spp., herpesviruses, and the intranuclear coccidian parasite of Testudines via polymerase chain reaction conducted by the Wildlife Conservation Society Molecular Lab as previously described [10,38–41]. Only tortoises that were apparently healthy on physical examination and negative for all pathogens via polymerase chain reaction were included in the study. We sampled 136 tortoises, with 120 samples of sufficient volume and gross quality for analysis. All handling and sample collection was performed according to Saint Louis Zoo’s Institutional Animal Care and Use Committee application 19–09 and Direction de la Gestion des ressources naturelles renouvelables et des ecosystemes research permit 026/20/MEDD/SG/DGEF/DGRNE.
We manually restrained tortoises [42] for blood collection from either the subcarapacial sinus (SC) or brachial vein using pre-heparinized 1 ml or 3 ml syringes with 0.64 mm (23 g) or 0.51 mm (25 g) needles of 25.4 mm (1”) or 38.1 mm (1.5”) in length as appropriate for patient size and temperament [43–45]. We collected no more than 3 ml of blood per tortoise, which was equivalent to less than 0.5% of body weight [45], and discarded samples with evidence of lymph contamination or clotting on gross examination. Using fresh blood directly from the syringe, we made two blood smears on glass slides and fixed them for more than one minute in methanol after air-drying. We removed the needle from the syringe and immediately transferred the remaining blood to lithium heparin blood collection tubes (Becton, Dickinson and Company, Franklin Lakes, New Jersey 07417, USA), which were filled > 2/3 of capacity and kept in coolers until analysis.
Hematology and biochemistry analyses
We performed all sample processing and analyses at an on-site field laboratory within 0.5 to 11 hours of sample collection with the exception of review of blood smears, which were stained the same day with a modified Wright-Giemsa stain (Dip Quick Stain, Jorgensen Laboratories, Loveland, Colorado 80538, USA) and evaluated within 14 days. We measured packed cell volume (PCV) using non-heparinized microhematocrit tubes centrifuged for six minutes at 1534 G (3,500 rpm) using hematocrit-specific adaptors (Crit Carrier, LW Scientific, Lawrenceville, GA 30045, USA) in a portable angled centrifuge (E8, LW Scientific, Lawrenceville, GA 30045, USA). We measured total solids (TS) in plasma from the microhematocrit tube using a manually calibrated, handheld clinical refractometer (J-351, Jorgensen Laboratories, Loveland, Colorado 80538, USA). We performed biochemistry analyses on whole blood using a VetScan® VS2 chemistry analyzer (Abaxis Inc., Union City, California 94584, USA) with VetScan Avian Reptilian Profile Plus rotors (Abaxis Inc., Union City, California 94584, USA). The biochemical panel included: albumin (Alb), bile acids (BA), aspartate aminotransferase (AST), total calcium (Ca), creatinine kinase (CK), glucose (GLU), potassium (K), sodium (Na), inorganic phosphorus (P), globulin (Glob), total protein (TP), and uric acid (UA).
We performed total white blood cell (WBC) quantification or estimates on each sample using three techniques. The Natt and Herrick’s method (NH) was performed using Natt-Herricks-TIC® 1:200 staining kits (Bioanalytic GmbH, D79224, Umkirch, Germany) per manufacturer instructions, charged into a Neubauer hemocytometer and calculated as: WBC x 109/L = number of WBC counted in 9 squares x 0.2/ 9. The Leukopet method (LP) was performed using Avian LeukopetTM Kits (Vetlab Supply, Palmetto Bay, FL) per manufacturer instruction, charged into a Neubauer hemocytometer and calculated as: WBC x 109/L = (number of eosinophilic staining cells [heterophils and eosinophils] counted in 18 squares x 1.1 x 16 x 0.1)/(% heterophils + % eosinophils determined from differential counts). For NH and LP, samples were incubated at room temperature with stain for five and ten minutes, respectively, and all were allowed to settle in the hemocytometer in a humidity chamber for three minutes prior to reading. Total WBC estimates (EST) were performed on the highest quality blood film from each tortoise by counting all leukocytes in ten fields of the monolayer using a 40x objective lens and calculated as: WBC x 109/L = (average number of leukocytes per field) x (objective power)2 x 0.001 [46]. We performed 100-cell leukocyte differentials on the same blood film using a 100x objective lens under oil immersion. We calculated method-specific absolute values for each WBC type by multiplying the percentage of a given cell type from the differential by the total leukocyte count from each WBC quantification method. All NH, EST, and differentials were performed by one observer. All LP were performed by a second observer after verifying that LP results performed by both observers were identical on a subset of samples.
Statistical analyses
All statistical analyses were performed in R v.4.0.3 [47] using the RStudio v. 1.3.1093 interface [48] unless specified otherwise. For all statistical tests, p < 0.05 was considered significant with the exception of the Anderson-Darling test for Gaussian distribution for reference interval calculation, where p < 0.3 was used per recommendations for samples of this size [49].
Pre-analysis data preparation.
Twenty samples were excluded from all analyses for gross lymph contamination (SC: n = 14/94 [14.8%]; brachial: n = 6/54 [11.1%]) and eight for clotting (SC: n = 0/94 [0.0%]; brachial: n = 8/54 [14.8%]). Slide quality issues (n = 9) and cell clumping on LP (n = 8) precluded additional samples from applicable WBC analyses. For values that fell outside of the dynamic range of the VetScan VS2 analyzer, we substituted the reported qualitative values with the quantitative value closest to the analyzer threshold to reduce the bias that would be introduced by removing these high and low values from the dataset [50]. Substitutions were made as follows: Ca– 4.01 for > 4.00 mmol/L; K– 8.6 for > 8.5 mmol/L; TP– 19 for < 20 g/L; and UA– 17 for < 18 μmol/L.
We evaluated the distribution of each hematologic and biochemical analyte by inspection of histograms, the Anderson-Darling test for Gaussian distribution [51], and a distribution-free test for symmetry [52]. For data that had a non-Gaussian distribution, we applied a Box-Cox transformation [53] and repeated evaluation of the distribution as above. We used Tukey’s interquartile range to identify statistical outliers on untransformed Gaussian data or on Box-Cox transformed data for non-Gaussian data. We removed only those outliers that could be attributed to sample, handling, or analytical error. After removing an outlier, we repeated the above procedures until no further outliers warranting removal were identified. Outliers removed from applicable analyses were: NH– 18.87 WBC x 109/L, LP– 28.86 WBC x 109/L, EST– 29.60 WBC x 109/L, and proportion of eosinophils– 0.15. Values altered specifically by hemolysis or extended time from sample collection to analysis were identified and handled as described below.
Effects of venipuncture site, sample artifacts, and time of sample collection (between-animal).
To identify potential confounds and presence of sample artifacts, we tested the associations between venipuncture site (SC, brachial), hemolysis (0, 1+, 2+, 3+), time of sample collection (AM, PM), and time from sample collection to biochemistry analysis using Kruskal-Wallis, Wilcoxon rank sum, and chi-square tests. As significant relationships existed between hemolysis and both venipuncture site and time to analysis as well as between time of sample collection and time to analysis, we used univariate generalized linear models (GLM) fit by the glm function [47] to simultaneously evaluate the effects of these four variables on our variables of interest. We fit separate models with each hematological analyte as the response and venipuncture site, hemolysis, and time of sample collection as predictors. We evaluated effects on biochemical analytes in the same manner with time to analysis included as an additional predictor. Model assumptions were checked using residual vs. fitted, normal Q-Q, scale-location, and residual vs. leverage plots. When model fit was poor, Box-Cox transformation was used. If this failed to provide sufficient fit, we performed stratified non-parametric tests with Hommel corrected p-values for each predictor-response pair. We conditionally averaged the best fit models defined as ΔAICc < 2. We determined the threshold at which hemolysis and/or time to analysis no longer exhibited a statistically significant effect on a given hematological or biochemical analyte by iteratively removing samples with the greatest hemolysis and/or time to analysis from the dataset and rerunning the above models or tests. We additionally evaluated differences in sample quality (hemolysis, gross clotting, and gross lymph contamination) based on venipuncture site using chi-square and Fisher exact tests.
Reference interval & descriptive statistics.
For a given analyte, we excluded samples determined to be affected by hemolysis or time to analysis as described above from reference interval calculation. For all analytes, samples collected in both the AM and PM were included in reference interval calculation as we considered variation based on time of sample collection to represent physiologic variation rather than pathology or handling artifact. Additionally, the number of samples collected in the PM (n = 13) was below the threshold recommended by American Society of Veterinary Clinical Pathology (ASVCP) guidelines for partitioning of reference intervals [16]. We used Reference Value Advisor v. 2.1 [54] to calculate reference intervals, 90% confidence intervals (CI) of the lower and upper reference interval limits, and the descriptive statistics mean, standard deviation, and median for each hematological and biochemical analyte. We calculated these values separately for each WBC quantification method and for each venipuncture site when venipuncture site was found to have an effect on an analyte. We calculated reference intervals and CIs according to the ASVCP guidelines [16] as follows: non-parametric methods for reference intervals and CIs when n = 120; robust method for reference intervals and bootstrap method for CIs when 40 ≤ n < 120 and distribution was symmetric; non-parametric method for reference intervals and bootstrap method for CIs when 40 ≤ n < 120 and distribution was not symmetric; and robust method for reference intervals and bootstrap method for CIs when 20 ≤ n < 40 and distribution was symmetric. When n < 120, we used Box-Cox transformed data for reference intervals and CI calculations of non-Gaussian distributed analytes when transformation improved distribution symmetry. Reference intervals were not calculated when n < 20 or when 20 ≤ n < 40 and analyte distribution was neither Gaussian nor symmetric.
Agreement between WBC quantification method (within-animal).
We evaluated agreement in total WBC count from the NH, LP, and EST methods using Passing-Bablok (PB) regression and Bland Altman plots. Passing-Bablok assumes a high positive correlation between the diagnostic methods being evaluated [55,56], which we tested using Kendall’s tau. We considered the methods in statistical agreement per PB when the 95% CI for the slope, the indicator of constant bias, contained one and the 95% CI for the y-intercept, the indicator of proportional bias, contained zero. We used Bland-Altman plots, estimates of bias (defined as the mean difference between methods), and limits of agreement (defined as the bias ± 1.96 times the standard deviation of the differences) to further evaluate agreement, systematic bias, and proportional bias [57]. As it has been proposed that LP results vary according to lymphocyte and heterophil proportions [24], we used Kendall’s tau to test for correlation between proportions of these cell types and the difference in total WBC count for each method pair.
Results
Reference intervals & descriptive statistics
Reference intervals calculated for hematology and biochemistry analytes of tortoises are reported in Tables 1 and 2 respectively. See S1 Table for biochemistry reference intervals and descriptive statistics presented in United States conventional units.
Reference intervals could not be calculated for three biochemistry analytes due to large proportions of the results falling outside of the dynamic range of the VetScan VS2 analyzer. All bile acid results (n = 120) were reported as < 35 μmol/L. Of 120 albumin results, 107 were reported as < 10 g/L with the remaining 13 results yielding a mean and standard deviation of 12 +/- 2 g/L, median of 11 g/L, and range of 10–17 g/L. As globulin concentration is calculated from total protein and albumin, only 13 values were reported for globulin with a mean and standard deviation of 24 +/- 3 g/L, median of 23 g/L, and range of 17–28 g/L.
Effects of venipuncture site (between-animal)
No significant effects on hematology analytes were found. All biochemistry analytes on which venipuncture site had a significant effect were greater in brachial samples than in subcarapacial samples. These included: AST (GLM: β = 19.9 [3.35–36.5] U/L, p = 0.018), CK (GLM: β = 893 [564–1222] U/L, p < 0.001), Glu (GLM: β = 0.630 [0.851–0.410] mmol/L, p < 0.001), and Phos (GLM: β = 0.107 [0.191–0.0222] mmol/L, p = 0.013). Conditionally averaged best-fit GLMs evaluating the effects of venipuncture site along with time of sample collection, hemolysis, and time from sample collection to biochemistry analysis are reported in Table 3. Effects of venipuncture site on uric acid, proportion of eosinophils, and presence of immature lymphocytes were evaluated with non-parametric tests due to inability to obtain appropriate GLM fit and no significant effects were identified. Gross clotting and hemolysis levels greater than 1+ were more frequent in brachial samples than SC samples (chi-square p < 0.001 and Fisher exact p < 0.001, respectively). There was no significant difference in gross lymph contamination (chi-square p = 0.517) between venipuncture sites.
Effects of time of sample collection (between-animal)
Time of sample collection did not have a significant effect on total WBC counts by any method, but the proportion of individual WBC types was significantly affected with PM samples being associated with increased heterophils (GLM: β = 0.0769 [0.0153–0.139], p = 0.016), decreased lymphocytes (GLM: β = -0.111 (-0.191–-0.0315), p = 0.006), and increased heterophil:lymphocyte ratios (GLM: β = 0.580 [0.161–1.000], p = 0.007) compared to AM samples. Of the biochemistry analytes, only Ca (GLM: β = -0.226 (-0.406–-0.0471) mmol/L, p = 0.013) and K (GLM: β = 1.06 [0.431–1.68] mmol/L, p < 0.001) were significantly affected.
Agreement between WBC quantification method (within-animal)
There was poor statistical agreement between all method pairs based on PB regression and Bland Altman plots (Fig 1, Table 4) with limits of agreement wider than 12.00 WBC x 109/L.
(a, c, e) Bland-Altman–solid line is the line of perfect agreement, dashed lines are mean difference between the methods (center), upper limit of agreement (upper), and lower limit of agreement (lower) with shading indicating associated 95% confidence intervals. (b, d) Passing-Bablok regression–dashed line is the line of perfect agreement, solid line is the regression line with shading indicating associated 95% confidence intervals. WBC = white blood cell.
Bland Altman plots revealed systematic bias between each pair with the EST method yielding higher values on average than the LP method (bias = 2.57 [1.49–3.66] WBC x 109/L), which in turn yielded higher values than the NH method (bias = 1.26 [0.21–2.30] WBC x 109/L). Passing-Bablok revealed constant bias between LP and EST values (intercept = 5.37 [3.08–7.40] WBC x 109/L) and proportional bias between EST and NH values (slope = 1.44 [1.03–1.83]), though the 95% CI approaches one. Results of the LP and NH methods failed the test for positive correlation prohibiting PB regression analysis. On Bland Altman plots, all pairs showed a trend toward larger differences as mean WBC counts increased. The difference between LP and NH values was positively correlated with lymphocyte proportion (τ = 0.332, p < 0.001) and negatively correlated with heterophil proportion (τ = -0.399, p < 0.001), as was the difference between LP and EST values (τ = 0.348, p < 0.001 and τ = -0.427, p < 0.001 respectively). The difference between NH and EST values was not significantly correlated with either lymphocyte proportion (τ = 0.0152, p = 0.814) or heterophil proportion (τ = -0.0313, p = 0.632).
Discussion
Robust population and technique-specific hematology and biochemistry reference intervals are valuable yet previously missing tools throughout the confiscation-to-release pathway that is important to reversing radiated tortoise population declines. Low cost, low technology analytes such as PCV and TS have proven useful as triage criteria with avian species in large scale oil spills [58–60]. They could be similarly informative in directing resources during tortoise confiscation events, though measuring even these two analytes may be challenging during large scale confiscations in low resource settings. After triage and when resources allow, hematology and biochemistry reference intervals can be used to help guide clinical care of confiscated tortoises just as they are used for a wide variety of taxa in rehabilitation settings [13,61–64]. These reference intervals also contribute to evaluating the health status of tortoises prior to release, as was the context of this study, with the goal of identifying and removing tortoises with evidence of disease from the release population [10,65]. Lastly, baseline values in the pre-release population provide a foundation for future research aimed at evaluating and improving management at all stages of care from confiscation until release [10,14,66–68].
This study provides statistically robust population- and technique-specific reference intervals for clinically healthy radiated tortoises in the priority age group for reintroduction living in a managed setting within native habitat. While reference intervals reported here are largely comparable to previously reported ranges for radiated tortoises in North American institutions [19–21], some important differences are apparent. Potassium and Na are notably higher in this population, which is most likely attributable to differences in water balance and diet between these settings. Drought-adapted chelonians tolerate wide variation in circulating K, Na, and Cl levels [69–71], all of which have been found to be higher in free-living desert tortoises (Gopherus agassizii) during seasons with relatively low rainfall and where diets are high in K [23,71]. This study was performed at the end of an extended dry season and thus under conditions of decreased water availability and altered forage, which likely also contributed to the relative elevations in K and Na in this population. In general, reference intervals for this population are substantially narrower than reference intervals calculated by ZIMS, which combine values from multiple age groups, venipunctures sites, biochemistry analyzers, and seasons. When based on adequate sample sizes, narrower intervals provide increased sensitivity in detecting abnormalities and highlight the value of determining and applying reference intervals on a population- and technique-specific basis.
We additionally quantified the effects of venipuncture site on hematology and biochemistry analytes in radiated tortoises and, to the authors’ knowledge, provide the first direct comparison between SC and brachial venipuncture sites in any species. While jugular blood is generally considered the gold standard sample in chelonians [72], SC and brachial venipuncture are the two most accessible techniques in unsedated radiated tortoises across the age groups in this study. Statistically significant differences were observed in AST, CK, Glu, and Phos, all of which were greater in brachial samples than in SC samples. These differences were present independent of level of hemolysis, time from sample collection to biochemistry analysis, and time of venipuncture. Of these, the magnitude of differences in CK, Glu, and Phos exceeded the thresholds adopted by the ASCVP for creation of separate reference intervals for clinical use [73,74]. Venipuncture site-based differences in CK (GLM: β = 893 [564–1222] U/L) and Glu (GLM: β = 0.630 [0.851–0.410] mmol/L) are particularly relevant for evaluating the physiologic effects of field capture and handling techniques and accurately identifying hypoglycemia in emergency or triage scenarios.
Relatively increased AST and CK are most likely attributable to restraint technique and relatively increased, although minor, iatrogenic tissue trauma during brachial venipuncture. This is supported by greater proportions of brachial samples exhibiting gross clotting (n = 8/54 [14.8%], Fisher exact p < 0.001) and hemolysis levels greater than 1+ (n = 15/40 [37.5%], chi-square p < 0.001) as compared to SC samples (n = 0/94 [0.0%] and n = 6/80 [7.5%], respectively). The etiology of the relatively decreased Glu and Phos in SC samples is less clear. This finding is consistent with presumed or observed lymph dilution during SC venipuncture in other chelonian species [25,26,28]. While gross lymph contamination was noted in a slightly greater proportion of SC (n = 14/94 [14.8%]) compared to brachial (n = 6/54 [11.1%]) samples in this study, this difference was far from statistically significant (chi-square p = 0.517). Further, lymph contamination is a form of hemodilution that most commonly reduces PCV, WBC count, and TP [25,26,28–30], none of which were observed here. Alternatively, an undetected difference between tortoises in the SC and brachial sample groups may be responsible, though the overall study population is homogenous in terms of age, size, apparent health status, husbandry, and environmental conditions.
As with venipuncture site, choice of leukocyte quantification method has been shown to have marked impacts on results in all chelonian species in which comparison of methods has been performed [31–36]. Leukocyte results in this study demonstrated poor clinical agreement between all methods with LOAs ranging from 12.64–21.58 WBC x 109/L and markedly different method-specific reference intervals. Width of the LOAs is due in part to random error inherent in manual leukocyte quantification methods even when performed by a single highly-trained observer. While imprecision has rarely been quantified, recent studies reported coefficients of variation of 8.2% using the LP method in eastern box turtles (Terrapene carolina carolina) [36] and 12% using the NH method in multiple avian species [75]. Despite the effects of random error, systematic and/or proportional biases are evident between all method pairs. On average, EST overestimated both LP (bias = 2.57 [1.49–3.66] x 109/L) and NH (bias = 3.51 [2.89–4.13] WBC x 109/L). This is consistent with comparisons between EST and NH in Galapagos tortoises (Chelonoidis spp.) [35], but contrary to studies comparing EST and LP in other chelonian species, all of which have found higher WBC counts using the same or similar phloxine-based stains compared to EST [33–36]. The magnitude of disagreement between all methods increased as mean WBC count increased on corresponding Bland Altman plots, though this proportional bias was only statistically supported by Passing-Bablok regression for EST vs. NH (slope = 1.44 [1.03–1.83]). When the proportional bias between EST and NH is adjusted for via Passing-Bablok regression, the constant bias between the methods decreases substantially (intercept = -0.160 [-3.02–3.12] WBC x 109/L). Given this and the relatively narrow LOA (12.64 [10.50–14.78] WBC x 109/L), EST and NH showed the greatest agreement of all method pairs. In contrast, LP and NH exhibited the poorest agreement with the widest LOA (21.58 [18.00–25.16] WBC x 109/L) and a lack of even basic positive correlation. Additionally, the difference in total WBC count by LP and both other methods was positively correlated with the proportion of lymphocytes and negatively correlated with the proportion of heterophils. This supports the supposition that LP overestimates total WBC count in high lymphocyte/low heterophil samples [24,35] and raises questions about its accuracy and utility in lymphocyte dominant species such as the radiated tortoise. Without an accepted gold standard or method-specific measures of precision, it is not possible to determine which method is the most accurate or precise. Nevertheless, our findings support the use of NH or EST over LP in radiated tortoises and emphasize the inadvisability of comparing results from different methods in individual patients or across populations.
The effect of time of sample collection on some hematological and biochemical analytes was an incidental finding in this study that warrants further investigation. Statistically significant differences in PM samples relative to AM samples included higher K, heterophil proportion, and heterophil:lymphocyte ratio and lower Ca and lymphocyte proportion. These differences may be due to circadian rhythm, temperature, or activity level, all of which have been show to affect certain hematology and biochemistry analytes in other reptile species [13,22,76–79]. For example, the changes in heterophil and lymphocyte proportions are consistent with shifts seen in some reptile species as glucocorticoids increase [80], which in turn has been shown to vary with temperature [81,82] and diurnally [83]. Similarly, K and Ca may be secondarily affected by the demonstrated effects of temperature and activity-level on blood gas and acid-base status [77,79,84]. As this study was not designed to evaluate the effects of time of day, reasons for observed differences are speculative. Future experiments that measure hematological and biochemical analytes in a time series with concurrent ambient and tortoise temperature, an index of activity level, and blood corticosterone would elucidate underlying etiologies and clinical implications of differences along these axes.
Certain limitations of this study should be considered when interpreting and applying our results. Reference intervals must be based on healthy individuals and while tortoises included in this study were apparently healthy based on physical examination and pathogen screening, subtler indicators of poor health such as changes in activity level and appetite may have been overlooked in this high volume rehabilitation setting. Additionally, biochemistry results were restricted by the dynamic range of the VetScan VS2 for certain analytes, imposing artificial limits on reference intervals and hampering the ability to detect effects of venipuncture site, sample artifacts, and time of day. By substituting out of range values with the value closest to the analyzer threshold rather than removing out of range values from analyses, these effects were reduced but not eliminated. Bile acids, UA, Alb, and Glob (which is calculated from Alb and TP) were most affected with results for these analytes below the reportable range for several or all tortoises. Other researchers have encountered similar constraints using the VetScan in alligator snapping turtles (Macrochelys temminckii) [85] and loggerhead turtles (Caretta caretta) [86,87]. As liver and renal disease are associated with elevated BA and UA, the VetScan and values reported here are likely useful for detecting clinical disease, but lack precision for ecophysiological research applications. In contrast, the current VetScan system cannot be recommended for measurement of Alb or Glob in this species for either clinical or research purposes with protein electrophoresis being the preferred method where resources allow [88,89]. For other biochemistry analytes, the VetScan remains a valuable tool for clinical care and research in chelonian populations where access to reference laboratories is limited, though species-specific validation is strongly recommended.
By determining population- and technique-specific hematology and biochemistry reference intervals and quantifying how these analytes vary between commonly used techniques, this study provides a valuable tool for improving care along the confiscation-to-release pathway critical to radiated tortoise conservation. These results also contribute foundational knowledge for future clinical research for this critically endangered species. We focused our primary efforts on the target population for release, subadults at the TCC, at a single time point. Further research is needed to characterize variation in hematology and biochemistry reference intervals across age groups and seasons within the rehabilitation population and to evaluate associations between these analytes, clinical outcomes, and post-release survival.
Supporting information
S1 Table. Subadult radiated tortoise biochemistry reference intervals and descriptive statistics in United States conventional units based on clinically healthy individuals living in a rehabilitation setting within their natural range.
https://doi.org/10.1371/journal.pone.0264111.s001
(DOCX)
Acknowledgments
We thank Jane Merkel, Dalia Ferguson, Christel Griffioen, Riana Rakotondrainy, Tony Ralivaniaina, Paul Calle, Vontsoa, Mena, and all TCC staff for their many contributions and invaluable support in the field. We are also grateful to Rick Hudson, Herilala Randriamahazo, and all Turtle Survival Alliance Madagascar staff for their organizational support and the ongoing care of they provide to these and thousands of other radiated tortoises. Thank you to Jamie Palmer, Ainoa Nieto-Claudín, and all Saint Louis Zoo Institute for Conservation Medicine staff for assistance with study logistics, design, and analysis. Lastly, we thank the Ecole Supérieure des Sciences Agronomiques (ESSA) for their institutional support.
References
- 1. Leuteritz T, Rioux Paquette S. Astrochelys radiata. The IUCN red list of threatened species 2008: e. T9014A12950491. 2008.
- 2. O’Brien S, Emahalala ER, Beard V, Rakotondrainy RM, Reid A, Raharisoa V, et al. Decline of the Madagascar radiated tortoise Geochelone radiata due to overexploitation. Oryx. 2003;37(3):338–43.
- 3. Leuteritz TE, Lamb T, Limberaza JC. Distribution, status, and conservation of radiated tortoises (Geochelone radiata) in Madagascar. Biol Conserv. 2005;124(4):451–61.
- 4. Harper GJ, Steininger MK, Tucker CJ, Juhn D, Hawkins F. Fifty years of deforestation and forest fragmentation in Madagascar. Environ Conserv. 2007;34(4):325–33.
- 5. Rafeliarisoa TH, Walker RCJ, Louis EE. Decline in the range and population density of Radiated Tortoises, Astrochelys radiata, in southern Madagascar. Castellano CM, Rhodin AGJ, Ogle M, Mittermeier RA, Randriamahazo H, Hudson R, et al., editors. Chelonian Res Monogr. 2013;6:83–9.
- 6. Ogle M, Hudson R, Castellano CM, Larsen L. Radiated tortoise SAFE program plan 2018. Association of Zoos & Aquariums SAFE Program; 2018.
- 7. Hudson R, Randriamahazo H. Radiated tortoise crisis worsens in Madagascar. Turtle Survival. 2018;19–23.
- 8. Karesh WB, Cook RA. Applications of veterinary medicine to in situ conservation efforts. Oryx. 1995;29(4):244–52.
- 9. Deem SL, Karesh WB, Weisman W. Putting theory into practice: wildlife health in conservation. Conserv Biol. 2001;15(5):1224–33.
- 10. Raphael BL, Macey SK, Platt SG, Seimon TA, Ossiboff RJ, Horne BD, et al. Health screening of burmese star tortoises (Geochelone platynota) prior to introduction to the wild. Chelonian Conserv Biol. 2019;18(2):153–62.
- 11. Woodford MH. Quarantine and health screening protocols for wildlife prior to translocation and release in to the wild. Office International des Epizooties; 2001.
- 12. Mazet JA, Newman SH, Gilardi KV, Tseng FS, Holcomb JB, Jessup DA, et al. Advances in oiled bird emergency medicine and management. J Avian Med Surg. 2002;16(2):146–9.
- 13. Innis CJ, Ravich JB, Tlusty MF, Hoge MS, Wunn DS, Boerner-Neville LB, et al. Hematologic and plasma biochemical findings in cold-stunned Kemp’s ridley turtles: 176 cases (2001–2005). J Am Vet Med Assoc. 2009;235(4):426–32. pmid:19681727
- 14. Sharp SM, Knoll JS, Moore MJ, Moore KM, Harry CT, Hoppe JM, et al. Hematological, biochemical, and morphological parameters as prognostic indicators for stranded common dolphins (Delphinus delphis) from Cape Cod, Massachusetts, USA. Mar Mammal Sci. 2014;30(3):864–87.
- 15.
Raphael BL. Rehabilitation medicine of confiscated turtles. In: Miller RE, Lamberski N, Calle P, editors. In: Miller RE, Lamberski N, Calle PP, editors. Fowler’s zoo and wild animal medicine current therapy. Vol. 9. Saint Louis: Elsevier; 2019. pp. 404–11.
- 16. Friedrichs KR, Harr KE, Freeman KP, Szladovits B, Walton RM, Barnhart KF, et al. ASVCP reference interval guidelines: determination of de novo reference intervals in veterinary species and other related topics. Vet Clin Pathol. 2012;41(4):441–53. pmid:23240820
- 17.
Weiser G, Allison RW. Perspectives in laboratory data interpretation and disease diagnosis. In: Thrall MA, Weiser G, Allison RW, Campbell TW, editors. Veterinary hematology and clinical chemistry. 2nd ed. New Jersey: Wiley; 2012. pp. 40–50.
- 18. Walton R. Establishing reference intervals: health as a relative concept. Seminars in Avian and Exotic Pet Medicine. 2001;10(2):66–71.
- 19. Marks SK, Citino SB. Hematology and serum chemistry of the radiated tortoise (Testudo radiata). J Zoo Wildl Med. 1990;342–4.
- 20. Zaias J, Norton T, Fickel A, Spratt J, Altman NH, Cray C. Biochemical and hematologic values for 18 clinically healthy radiated tortoises (Geochelone radiata) on St. Catherines Island, Georgia. Vet Clin Pathol. 2006;35(3):321–5. pmid:16967417
- 21.
ZIMS expected test results for Astrochelys radiata [Internet]. Species360 Zoological Information Management System; 2020. Available from: http://zims.Species360.org filters: All_ages_combined__All_sexes_combined_All_restraint_types_combined_Standard_International_Units.
- 22. Anderson NL, Wack RF, Hatcher R. Hematology and clinical chemistry reference ranges for clinically normal, captive new guinea snapping turtle (Elseya novaeguineae) and the effects of temperature, sex, and sample type. J Zoo Wildl Med. 1997;28(4):394–403. pmid:9523632
- 23. Christopher MM, Berry KH, Wallis IR, Nagy KA, Henen BT, Peterson CC. Reference intervals and physiologic alterations in hematologic and biochemical values of free-ranging desert tortoises in the Mojave Desert. J Wildl Dis. 1999;35(2):212–38. pmid:10231748
- 24. Sykes JM, Klaphake E. Reptile hematology. Veterinary Clin North Am Exot Anim Pract. 2008;11(3):481–500. pmid:18675730
- 25. Gottdenker NL, Jacobson ER. Effect of venipuncture sites on hematologic and clinical biochemical values in desert tortoises (Gopherus agassizii). Am J Vet Res. 1995;56(1):19–21. pmid:7695143
- 26. López-Olvera JR, Montané J, Marco I, Martínez-Silvestre A, Soler J, Lavín S. Effect of venipuncture site on hematologic and serum biochemical parameters in marginated tortoise (Testudo marginata). J Wildl Dis. 2003;39(4):830–6. pmid:14733278
- 27. Medeiros NC, Locatelli-Dittrich R, Schmidt EMS, Alvares AA, Patricio LL, Lange RR, et al. Effect of the blood collection site on the hematological parameters in red-eared slider, Trachemys scripta elegans. Pesqui Veterinária Bras. 2012;32(Suppl.):37–40.
- 28. Bonnet X, El Hassani MS, Lecq S, Michel CL, El Mouden EH, Michaud B, et al. Blood mixtures: impact of puncture site on blood parameters. J Comp Physiol B. 2016;186(6):787–800. pmid:27146147
- 29. Eshar D, Gancz AY, Avni-Magen N, Wagshal E, Pohlman LM, Mitchell MA. Selected plasma biochemistry analytes of healthy captive sulcata (African spurred) tortoises (Centrochelys sulcata). J Zoo Wildl Med. 2016;47(4):993–9. pmid:28080924
- 30. Perpiñán D, Armstrong DL, Dórea F. Effect of anticoagulant and venipuncture site on hematology and serum chemistries of the spiny softshell turtle (Apalone spinifera). J Herpetol Med Surg. 2010;20(2):74–8.
- 31. Stewart K, Mitchell MA, Norton T, Krecek RC. Measuring the level of agreement in hematologic and biochemical values between blood sampling sites in leatherback sea turtles (Dermochelys coriacea). J Zoo Wildl Med. 2012;43(4):719–25. pmid:23272336
- 32. Arnold J. White blood cell count discrepancies in Atlantic loggerhead sea turtles: Natt-Herrick vs. eosinophil unopette. In: Proc 14th Annu Conf Assoc Zoo Vet Tech. Cleaveland, OH; 1994. pp. 15–22.
- 33. Deem SL, Dierenfeld ES, Sounguet GP, Alleman AR, Cray C, Poppenga RH, et al. Blood values in free-ranging nesting leatherback sea turtles (Dermochelys coriacea) on the coast of the Republic of Gabon. J Zoo Wildl Med. 2006;37(4):464–71. pmid:17315430
- 34. Deem SL, Norton TM, Mitchell M, Segars AL, Alleman AR, Cray C, et al. Comparison of blood values in foraging, nesting, and stranded loggerhead turtles (Caretta caretta) along the coast of Georgia, USA. J Wildl Dis. 2009;45(1):41–56. pmid:19204334
- 35. Sheldon JD, Stacy NI, Blake S, Cabrera F, Deem SL. Comparison of total leukocyte quantification methods in free-living Galapagos tortoises (Chelonoidis spp.). J Zoo Wildl Med. 2016;47(1):196–205. pmid:27010280
- 36. Winter JM, Stacy NI, Adamovicz LA, Allender MC. Investigating the analytical variability and agreement of manual leukocyte quantification methods in eastern box turtles (Terrapene carolina carolina). Front Vet Sci. 2019;6:398. pmid:31781584
- 37. Leuteritz TE, Gantz DT. Sexual dimorphism in radiated tortoises (Astrochelys radiata). Chelonian Res Monogr. 2013;6:105–12.
- 38. VanDevanter DR, Warrener P, Bennett L, Schultz ER, Coulter S, Garber RL, et al. Detection and analysis of diverse herpesviral species by consensus primer PCR. J Clin Microbiol. 1996;34(7):1666–71. pmid:8784566
- 39. Ossiboff RJ, Newton AL, Seimon TA, Moore RP, McAloose D. Emydid herpesvirus 1 infection in northern map turtles (Graptemys geographica) and painted turtles (Chrysemys picta). J Vet Diagn Invest. 2015;27(3):392–5. pmid:25943128
- 40. Ossiboff RJ, Raphael BL, Ammazzalorso AD, Seimon TA, Niederriter H, Zarate B, et al. A Mycoplasma species of Emydidae turtles in the northeastern USA. J Wildl Dis. 2015;51(2):466–70. pmid:25574806
- 41. Seimon TA, Horne BD, Tomaszewicz A, Pruvot M, Som S, In S, et al. Disease screening in southern river terrapins (Batagur affinis edwardmolli) in Cambodia. J Zoo Wildl Med. 2017;48(4):1242–6. pmid:29297823
- 42.
Boyer TH, Boyer DM. Tortoises, Freshwater Turtles, and Terrapins. In: Divers SJ, Stahl SJ, editors. Mader’s reptile and amphibian medicine and surgery. 3rd ed. Saint Louis: Elsevier; 2019. pp. 168–79.
- 43. Jacobson ER. Blood collection techniques in reptiles: laboratory investigations. In: Fowler M., editor. Zoo and wild animal medicine, Current Therapy. WB Saunders; 1993. pp. 144–52.
- 44. Mans C. Venipuncture techniques in chelonian species. Lab Anim. 2008;37(7):303–4. pmid:18568008
- 45.
Divers SJ. Diagnostic techniques and sample collection. In: Divers SJ, Stahl SJ, editors. Mader’s reptile and amphibian medicine and surgery. 3rd ed. Saint Louis: Elsevier; 2019. pp. 405–21.
- 46. Weiss DJ. Uniform evaluation and semiquantitative reporting of hematologic data in veterinary laboratories. Vet Clin Pathol. 1984;13(2):27–31. pmid:15311388
- 47.
R Core Team. R: A language and environment for statistical computing [software]. Vienna, Austria: R Foundation for Statistical Computing; 2020. Available from: https://www.R-project.org/.
- 48.
RStudio Team. RStudio: Integrated Development for R [software]. Boston, MA: RStudio, PBC; 2020. Available from: http://www.rstudio.com/.
- 49. Le Boedec K. Sensitivity and specificity of normality tests and consequences on reference interval accuracy at small sample size: a computer-simulation study. Vet Clin Pathol. 2016;45(4):648–56. pmid:27556235
- 50. Nieto-Claudín A, Palmer JL, Esperón F, Deem SL. Haematology and plasma biochemistry reference intervals for the critically endangered western Santa Cruz Galapagos tortoise (Chelonoidis porteri). Conserv Physiol. 2021;9(1). pmid:33959287
- 51. Stephens MA. EDF statistics for goodness of fit and some comparisons. J Am Stat Assoc. 1974;69(347):730–7.
- 52. McWilliams TP. A distribution-free test for symmetry based on a runs statistic. J Am Stat Assoc. 1990;85(412):1130–3.
- 53. Box GE, Cox DR. An analysis of transformations. J R Stat Soc Ser B Methodol. 1964;26(2):211–43.
- 54. Geffré A, Concordet D, Braun J-P, Trumel C. Reference Value Advisor: a new freeware set of macroinstructions to calculate reference intervals with Microsoft Excel. Vet Clin Pathol. 2011;40(1):107–12. pmid:21366659
- 55. Passing H, Bablok W. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, Part I. J Clin Chem Clin Biochem. 1983;21:709–720. pmid:6655447
- 56. Bablok W, Passing H, Bender R, Schneider B. A general regression procedure for method transformation. Application of linear regression procedures for method comparison studies in clinical chemistry, Part III. J Clin Chem Clin Biochem. 1988;26:783–790. pmid:3235954
- 57. Bland JM, Altman D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;327(8476):307–10. pmid:2868172
- 58. Grogan A, Pulquério MJ, Cruz MJ, Oaten P, Thompson R, Grantham M. Factors affecting the welfare of rehabilitation of oiled Murres. In: Proc 11th Int Conf Effects of Oil on Wildlife. New Orleans, LA; 2012. pp. 24–8.
- 59. Duerr RS, Ziccardi MH, Massey JG. Mortality during treatment: factors affecting the survival of oiled, rehabilitated common murres (Uria aalge). J Wildl Dis. 2016;52(3):495–505. pmid:27187030
- 60.
Oiled Wildlife Care Network. Protocols for the care of oil-affected birds. 3rd ed. Davis (CA): Wildlife Health Center, School of Veterinary Medicine, University of California; 2014.
- 61. Heatley JJ, Demirjian SE, Wright JC, Welles EG, Hoerr F, Wohl J. Electrolytes of the critically ill raptor. In: Proc Annu Conf Assoc Avian Vet. 2005. p. 23–4.
- 62. Caliendo V, McKinney P, Robinson D, Baverstock W, Hyland K. Plasma biochemistry and hematology values in juvenile hawksbill turtles (Eretmochelys imbricata) undergoing rehabilitation. J Herpetol Med Surg. 2010;20(4):117–21.
- 63. Flint M, Morton JM, Limpus CJ, Patterson-Kane JC, Murray PJ, Mills PC. Development and application of biochemical and haematological reference intervals to identify unhealthy green sea turtles (Chelonia mydas). Vet J. 2010;185(3):299–304. pmid:19709912
- 64. Ratliff C, Gentry J, Schmalz S, Hartke K, Russell KE, Acierno M, et al. Saving lives: critical blood analytes for rehabilitation of coastal birds. In: Proc Int Oil Spill Conf. Savannah, GA; 2014. pp. 408–16.
- 65.
IUCN/SSC. Guidelines for reintroductions and other conservation translocations. Version 1.0. Gland, Switzerland: IUCN Species Survival Commission; 2013 p. viiii + 57 pp.
- 66. Alberts AC, Oliva ML, Worley MB, Telford Jr SR, Morris PJ, Janssen DL. The need for pre-release health screening in animal translocations: a case study of the Cuban iguana (Cyclura nubila). Anim Conserv. 1998;1(3):165–72.
- 67. Lander ME, Harvey JT, Gulland FM. Hematology and serum chemistry comparisons between free-ranging and rehabilitated harbor seal (Phoca vitulina richardsi) pups. J Wildl Dis. 2003;39(3):600–9. pmid:14567222
- 68. Fazio E, Liotta A, Medica P, Giacoppo E, Ferlazzo A. Effects of different health status on blood haematochemical values of loggerhead sea turtles (Caretta caretta). Comp Clin Pathol. 2012;21(1):105–9.
- 69. Minnich JE. Adaptive responses in the water and electrolyte budgets of native and captive desert tortoises, Gopherus agassizii, to chronic drought. In: Proc Desert Tortoise Council Symp. Las Vegas, NV;1977. pp. 102–29.
- 70. Nagy KA, Medica PA. Physiological ecology of desert tortoises in southern Nevada. Herpetologica. 1986;42(1):73–92.
- 71. Peterson CC. Anhomeostasis: seasonal water and solute relations in two populations of the desert tortoise (Gopherus agassizii) during chronic drought. Physiol Zool. 1996;69(6):1324–58.
- 72.
Heatley JJ, Russell KE. Hematology. In: Divers SJ, Stahl SJ, editors. Mader’s reptile and amphibian medicine and surgery. 3rd ed. Saint Louis: Elsevier; 2019. pp. 301–17.
- 73. Lahti A, Hyltoft Petersen P, Boyd JC, Fraser CG, Jørgensen N. Objective criteria for partitioning Gaussian-distributed reference values into subgroups. Clin Chem. 2002;48(2):338–52. pmid:11805016
- 74. Lahti A, Hyltoft Petersen P, Boyd JC, Rustad P, Laake P, Solberg HE. Partitioning of nongaussian-distributed biochemical reference data into subgroups. Clin Chem. 2004;50(5):891–900. pmid:15010425
- 75. Carisch L, Stirn M, Hatt JM, Federer K, Hofmann-Lehmann R, Riond B. White blood cell count in birds: evaluation of a commercially available method. BMC Vet Res. 2019 Mar 14;15(1):93. pmid:30866927
- 76. Dessauer HC. Blood chemistry of reptiles: physiological and evolutionary aspects. Biol Reptil. 1970;2:1–4.
- 77. Gatten RE. Effects of activity on blood oxygen saturation, lactate, and pH in the turtles Pseudemys scripta and Terrapene ornata. Physiol Zool. 1975;48(1):24–35.
- 78. Choudhury S, De TK, Maiti BR, Ghosh A. Circadian rhythm in blood sugar and adrenomedullary hormonal concentrations in an avian and a reptilian species. Gen Comp Endocrinol. 1982;46(1):110–2. pmid:7060927
- 79. Innis CJ, Tlusty M, Merigo C, Weber ES. Metabolic and respiratory status of cold-stunned Kemp’s ridley sea turtles (Lepidochelys kempii). J Comp Physiol B. 2007;177(6):623–30. pmid:17431640
- 80. Martínez Silvestre A. How to assess stress in reptiles. J Exot Pet Med. 2014;23(3):240–3.
- 81. Hudson SB, Lidgard AD, French SS. Glucocorticoids, energy metabolites, and immunity vary across allostatic states for plateau side-blotched lizards (Uta stansburiana uniformis) residing in a heterogeneous thermal environment. J Exp Zool Part Ecol Integr Physiol. 2020;333(10):732–43.
- 82. Racic A, Tylan C, Langkilde T. Effects of temperature on plasma corticosterone in a native lizard. Sci Rep. 2020;10(1):1–8. pmid:31913322
- 83. Woodley SK, Painter DL, Moore MC, Wikelski M, Michael Romero L. Effect of tidal cycle and food intake on the baseline plasma corticosterone rhythm in intertidally foraging marine iguanas. Gen Comp Endocrinol. 2003;132(2):216–22. pmid:12812768
- 84. Reeves RB. The interaction of body temperature and acid-base balance in ectothermic vertebrates. Annu Rev Physiol. 1977;39(1):559–86. pmid:15501
- 85. DiGeronimo PM, Beaufrère H, Grasperge BJ, Nevarez JG. Comparison of alligator snapping turtle (Macrochelys temminckii) plasma biochemical profiles from two clinical analyzers. J Zoo Wildl Med. 2018;49(4):925–30. pmid:30592915
- 86. Atkins A, Jacobson E, Hernandez J, Bolten AB, Lu X. Use of a portable point-of-care (Vetscan Vs2) biochemical analyzer for measuring plasma biochemical levels in free-living loggerhead sea turtles (Caretta caretta). J Zoo Wildl Med. 2010;41(4):585–93. pmid:21370637
- 87. Wolf KN, Harms CA, Beasley JF. Evaluation of five clinical chemistry analyzers for use in health assessment in sea turtles. J Am Vet Med Assoc. 2008;233(3):470–5. pmid:18673037
- 88. Müller K, Brunnberg L. Determination of plasma albumin concentration in healthy and diseased turtles: a comparison of protein electrophoresis and the bromcresol green dye‐binding method. Vet Clin Pathol. 2010;39(1):79–82. pmid:19702665
- 89.
Heatley JJ, Russell KE. Clinical Chemistry. In: Divers SJ, Stahl SJ, editors. Mader’s reptile and amphibian medicine and surgery. 3rd ed. Saint Louis: Elsevier; 2019. pp. 319–332.