Seasonal changes in hematology and serum biochemistry results, described by separate reference intervals for different seasons, have been reported in many animals. We developed a novel method to investigate seasonal variation in values and a reference tool (the reference curve) based on sine wave functions that, for suitable variables, represents data more appropriately than a fixed reference interval. We applied these techniques to values observed in blood samples from 126 adult wild platypuses (Ornithorhynchus anatinus; 58 females and 68 males). Samples were collected under isoflurane anesthesia from animals captured in the Inglis Catchment in northwest Tasmania. In general, packed cell volume (PCV), red cell count (RCC), and hemoglobin (Hb) values appeared to be lower than those in two studies that previously reported platypus hematology reference intervals. This likely resulted from reduced stress-related splenic contraction or isoflurane-associated splenic sequestration of red blood cells in our study. Reference curves were described for five variables (PCV, RCC, Hb, albumin, and magnesium). We found evidence that this seasonal variation may result from metabolic changes associated with seasonal variations in environmental temperature. These observations suggest that it is important for researchers reporting platypus hematology and serum biochemistry to look for seasonal changes in their data to ensure it is appropriately interpreted.
Seasonal changes in hematology and serum biochemistry occur in many wildlife species (Sealander 1964; DelGiudice et al. 1992; Collazos et al. 1998; Woods and Hellgren 2003). These changes have generally been described by separate statistics for spring, summer, autumn, and winter, each derived from data collected in the relevant season. This approach can lead to seasonal descriptive statistics being based on relatively small numbers of individuals, with an increased potential for misleading results. Additionally, it does not investigate variations within the seasons. An approach to describing seasonal variation in hematology and serum biochemistry parameters that can capture the magnitude of the changes more reliably, and the timing more precisely, would be an important step in understanding the causes of observed seasonal variation and in interpreting individual data.
For assessment of hematology and biochemistry parameters in adult platypuses (Ornithorhynchus anatinus) there are two sets of reference intervals available based on 90 and 131 conscious individuals (Booth and Connolly 2008; Geraghty et al. 2011). Data from another nine conscious platypuses and 31 individuals under ether anesthesia have also been published (Whittington and Grant 1984). Our aim was to add to this sparse data set and to investigate dynamic patterns of seasonal variation.
MATERIALS AND METHODS
Blood samples were collected from 126 wild platypuses (58 adult females—no spur or spur sheath present, 68 adult males—spur but no spur sheath present) captured in the Inglis Catchment (41°3′S, 145°38′E) in northwest Tasmania 29 August 2011–31 August 2013 with a fieldwork schedule that, in part, aimed for captures to be distributed evenly among seasons. During the 24-mo period, 34 platypuses were captured in spring, 27 in summer, 32 in autumn, and 33 in winter. Platypuses were examined in the field close to the site of their capture either in a tent erected on-site or in a farm shed (if available). After capture in a fyke net (Macgregor et al. 2010), each platypus was held in a cotton bag inside a hessian sack for a minimum of 60 min (mean=125 min, SD=50 min, maximum=324 min) to allow its fur to dry and its stomach to empty (Booth and Connolly 2008) before being anesthetized for examination and sampling. When the air temperature was below 10 C, a 32–34 C hot water bottle was placed between the cotton bag and hessian sack in a position where the platypus could chose to be close to it, or move away from it.
Anesthesia was induced with 5% isoflurane in oxygen at 2 L/min by face mask and was generally maintained with isoflurane at 1.5% in oxygen at 1.5 L/min. The mean duration of anesthesia was 25 min (SD=6 min, minimum=9 min, maximum=40 min). Body temperature was maintained by a thermostatically controlled heating pad and a bubble-wrap blanket, which were required because of the cooler environmental temperatures associated with anesthesia of platypuses in the field in Tasmania. Each platypus was checked for a microchip using a handheld Trovan® GR-251 Universal Reader (Trovan Ltd., Microchips Australia Pty Ltd., Keysborough, Victoria, Australia). If no microchip was present, a Trovan Unique® microchip (Trovan) was implanted aseptically into the subcutaneous tissues between the scapulae of the platypus, following the methods of Grant and Whittington (1991), to allow subsequent identification. Up to 2 mL of blood was collected using a 3-mL syringe and 23G needle from the bill sinus (Whittington and Grant 1984) during recovery from anesthesia after the face mask had been removed.
A clinical examination was performed on each platypus, and all were free from clinically significant disease. In addition, a range of morphometric examinations was performed. Body mass (kg) was measured with Rapala® digital scales (Rapala VMC Corporation, Vääksy, Asikkala, Finland). Tail volume index was assessed as a measure of body condition (range 1–5, where 5 is very low tail volume and 1 is very high tail volume; Grant and Carrick 1978). Bill width (mm) was measured using vernier calipers. Total body length (cm) was measured using a tape measure (tip of bill to tip of tail, measured over dorsum). Dorsoventral tail fat depth adjacent to the spinal muscles half way along the length of the tail (midtail fat depth) was measured by ultrasound.
In addition to tail volume index, body condition was assessed using fat depth relative (RFD) to total body length (TBL) (i.e., RFDTBL), where
and body condition index (BCI) relative to TBL,
as per Macgregor (2015).
The cloacal body temperature of each platypus was recorded every 5 min during anesthesia using a Welch Allyn SureTemp® Plus 690 Electronic Thermometer (Welch Allyn, Skaneateles Falls, New York, USA). Water temperature at capture sites was measured using a Testo probe thermometer (Testo Pty Ltd., Croydon, Victoria, Australia). The minimum temperature recorded at Wynyard (Australian Government, Bureau of Meteorology 2016) for each night of fieldwork was used as a measure of ambient temperature.
Sample handling and processing
Blood samples up to 0.5 mL in volume were transferred to an ethylenediaminetetraacetic acid blood tube. For larger blood samples, 0.5 mL was transferred to an ethylenediaminetetraacetic acid tube, and the remainder was transferred to a plain blood tube. Samples were held in a refrigerator and transported on ice blocks in a cooler. If enough blood was collected for some to be placed in a plain tube, this was centrifuged, and the serum was transferred to a separate plain tube. The maximum time between sampling and centrifugation was 18 h. Laboratory testing was performed at the Department of Primary Industries, Parks, Water, and Environment (DPIPWE, Tasmania Animal Health Laboratory, Mount Pleasant Laboratories, Prospect, Tasmania), a maximum of 60 h after sampling. Packed cell volumes were determined manually (Graeme Knowles, Veterinary Pathologist, DPIPWE), and other hematology parameters were determined using a Sysmex KX-21N automated hematology analyzer (Sysmex Corporation, Hyogo, Japan). All blood smear slides were reviewed for blood parasites and consistency with automated analyzer results. A Konelab 20XTi (Thermo Fisher Scientific, Waltham, Massachusetts, USA) analyzer was used to analyze sera for biochemical analytes.
The Animal Health Laboratory performed quality control runs for each serum biochemistry analyte on each day samples were tested. Cell count and hemoglobin (Hb) quality controls were run weekly, but these hematology quality control data were not stored and hence not available for analysis. Packed cell volume (PCV) quality controls were run quarterly. Target values and ranges for quality control data were used to assess analytical performance on an ongoing basis.
Hematology and biochemistry reference intervals
Of the 126 individuals captured, 116 were captured once, eight were captured twice, and two were captured three times. For the 10 individuals that were captured more than once, only hematology/biochemistry data from the first capture were used for analysis. Fixed reference intervals were produced for all parameters using standard methods (Friedrichs et al. 2012) using Reference Value Advisor 2.1 (Geffre et al. 2011) as an add-in to Microsoft Excel (Redmond, Washington, USA). We took a conservative approach to outlier identification and exclusion. Box plots (Statistica 8.0, StatSoft Inc., Tulsa, Oklahoma, USA) were used to identify extreme outliers: less than or equal to the first quartile minus three times the interquartile range or greater than the third quartile plus three times the interquartile range. Extreme outliers were removed before running standard descriptive statistics (Warren et al. 2015). The following parameters had extreme outliers in the development of fixed reference ranges (number of outliers in parentheses): red cell count (RCC) (1), mean cell hemoglobin concentration (2), lymphocytes (1), eosinophils (2), fibrinogen (1), creatinine kinase (5), aspartate aminotransferase (AST) (3), alanine aminotransferase (2), glutamate dehydrogenase (GLDH) (4), gamma-glutamyl transferase (3), and urea (1). Most extreme outliers occurred in isolation in individual platypuses apart from two individuals with extreme outlier results for AST, alanine aminotransferase, and GLDH and another with extreme outliers for AST and GLDH. All extreme outliers were removed before further analyses. An additional source of variation in the number of results for each parameter was the quantity of blood collected from each individual. We prioritized the use of blood for hematology. Depending on the quantity of blood collected, it was not always possible to perform all/any biochemistry tests. Fibrinogen and total solids were not assayed in one individual. White blood cell differential counts were not performed in two platypuses because of poor preservation of white blood cell morphology. Otherwise, it was not necessary to exclude any data because of sample quality.
Data were tested for normality using the Shapiro-Wilk test (Statistica 8.0). To test for differences between the sexes in the distributions of observed values, we used unpaired t-tests for normally distributed parameters and Mann-Whitney U-tests (two factors) for nonparametric data.
A novel method was used to investigate seasonal variation in observed values and to develop a reference tool that would more appropriately represent the data than a fixed reference interval. There were five possible outcomes of this approach for each hematology or biochemistry variable: a fixed reference range for all platypuses, separate fixed reference ranges for males and females, sinusoidally varying reference curves for all platypuses, separate reference curves for males and females, or a fixed reference range for one sex and a reference curve for the other sex. The main concept underlying this method was assessment of the correlation between the observed data and a series of parameters that varied sinusoidally over the year. To do this, the day of the year (DOY; 1 January = 1, 31 December = 365) was converted to a sine wave with an amplitude of one and a period of 1 yr. We created 365 sine functions, each shifting the minimum and maximum values forward by 1 d. The phase of seasonal change for each blood parameter was selected by comparing data for each blood parameter against these sine functions using a correlation matrix; the DOY sine wave parameter that produced the highest coefficient of determination (optimum DOY sine wave parameter) was selected for further analysis. The P value and standardized regression coefficient (beta) of this correlation were also determined. Where appropriate, data were converted to an adjusted value using the formula: adjusted value = observed value − (optimum DOY sine wave parameter on date of sample × beta).
A median for the adjusted values was then determined using Microsoft Excel, and a reference interval for adjusted values was determined using Reference Value Advisor 2.1 (Geffre et al. 2011) as an add-in to Excel. Where appropriate, a seasonally varying median and upper and lower reference curve limits were calculated for each DOY using the formula (showing median as an example): median for day i = median of adjusted values + (optimum DOY sine wave parameter value on day i × beta)
A stepwise process was taken for each variable, with the possibility of several iterations of the same or similar steps, until one of the possible outcomes was reached (see Supplementary Figure). Steps were undertaken 1) on observed data or data adjusted for possible sinusoidal seasonal variation and 2) on data for all platypuses or each sex separately. Steps were undertaken to remove outliers (box plots; Statistica 8.0), to assess the correlation of data with DOY parameters (correlation matrix; Statistica 8.0), and to assess whether the distributions of males and females were different (Mann-Whitney U-test; Statistica 8.0).
Correlation coefficient (Pearson's r), and the P value of correlations between seasonally varying hematology and biochemistry parameters and body condition indices were calculated. To investigate possible causes of seasonal variation in certain red blood cell hematology parameters, the correlation coefficient r and P value of correlations between PCV, DOY sine wave parameter, and individual/environmental temperature parameters were calculated. To investigate possible causes of seasonal variation in certain serum biochemistry parameters, the correlation coefficient r and P value of correlations between albumin, DOY sine wave parameter, and individual/environmental temperature parameters were calculated. The quality control data closely associated in time with each albumin test run allowed investigation of the possible influence of laboratory techniques on seasonal variation in certain serum biochemistry parameters by calculating the correlation coefficient r and P value of correlations between platypus albumin results and the associated laboratory quality control data.
Hematology and biochemistry reference intervals
The fixed hematology and biochemistry reference intervals and descriptive statistics for observed values are listed in Tables 1 and 2. Females had significantly higher values for gamma-glutamyl transferase, cholesterol, calcium, mean cell hemoglobin, and mean cell hemoglobin concentration (Table 3). Males had significantly higher values for creatinine kinase, creatinine, albumin, albumin:globulin ratio, and eosinophils (Table 3). There was seasonal variation in five parameters (PCV, RCC, Hb, albumin, and magnesium), and these were represented by reference curves for all adult platypuses (Fig. 1). The available laboratory quality control data for the seasonally varying parameters are presented in Table 4.
The results of observations of intrinsic and extrinsic factors relating to the platypuses and the dates of their captures are summarized in Table 5. The five seasonally varying parameters were significantly correlated, indicating similar responses to seasonal changes (Table 6). Of the 15 correlations tested between seasonally varying parameters and body condition indices, the negative relationship between albumin and RFDTBL was the only significant result (Table 6). There were multiple significant correlations between measures of ambient temperature, measures of platypus body temperature during examination, and optimum DOY sine wave parameter value (Tables 7, 8), indicating covariance between these variables. Also, PCV was highly significantly correlated with water temperature and optimum DOY sine wave parameter value, and albumin was highly significantly correlated with both measures of ambient temperature and optimum DOY sine wave parameter value. However, neither PCV nor albumin was significantly correlated with measures of platypus body temperature during examination. Albumin was not correlated with associated quality control data (low control: r=−0.06, P=0.995; high control: r=−0.33, P=0.735).
Our data supplement two previously reported sets of serum biochemistry and hematology reference intervals for platypuses. We also present a novel method of assessing seasonal variation in these variables, and we observe and represent sinusoidal variation over the year and introduce the reference curve as a new concept for representing dynamic serum biochemistry and hematology variables.
In general, although reported differently from this study and from each other, the previously reported ranges appear to have been derived from overall parameter distributions similar to those in this study (Booth and Connolly 2008; Geraghty et al. 2011). The exceptions to this are the red blood cell parameters PCV, RCC, and Hb, which all appear to have lower means/medians and narrower ranges than those reported by Booth and Connolly (2008) and Geraghty et al. (2011). For comparison, the minimum and maximum values for PCV observed by Booth and Connolly (2008) were 35 and 62, respectively, and the unadjusted 2.5 and 97.5 percentiles reported by Geraghty et al. (2011) were 42 and 64.8, respectively. A possible cause for this is that we took blood samples from platypuses under isoflurane anesthesia, whereas in the two previous studies, blood samples were taken from conscious animals. Isoflurane anesthesia can lead to decreased PCV, RCC, and Hb in ferrets (Mustela putorius furo), ponies (Equus ferus caballus), and rats (Rattus norvegicus) (Taylor 1991; Marini et al. 1997; Deckardt et al. 2007), and this may have occurred in the platypuses. In ferrets, this was due to splenic sequestration of red blood cells (Marini et al. 1997). Stress of handling, which may be higher during sampling in nonanesthetized platypuses, can lead to increases in these red blood cell parameters because of splenic contraction in other species (Raskin 2009; Harvey 2012). Peinado et al. (1993) observed lower PCV, RCC, and Hb results in the Spanish ibex (Capra pyrenaica hispanica) when restrained with tiletamine-zolazepam compared with physical restraint. Hawkey et al. (1983) observed lower PCV, RCC, and Hb in yaks (Bos grunniens) when sedated with xylazine compared with manual restraint, noting the effects of splenic contraction during handling and of splenic sequestration of red blood cells during xylazine sedation. Further study is needed to confirm that PCV, RCC, and Hb reference intervals are lower in isoflurane-anesthetized platypuses than in conscious individuals. Even if this is confirmed, it would be difficult to determine whether values for PCV, RCC, and Hb in platypuses sampled conscious or those sampled under isoflurane anesthesia are a better reflection of the values in the animals before capture. However, until such studies are performed and given the possibility that values may be different for platypuses sampled in these two situations, consideration should be given to restraint procedures during blood sampling when choosing appropriate reference intervals for interpretation of results. We used isoflurane anesthesia, which is advocated to reduce stress associated with physical restraint required for blood sample collection and other time-consuming or stressful procedures.
The reference curves (Fig. 1) allow a more meaningful assessment of results for the five parameters shown than would be possible with fixed reference ranges. The inclusion of these data illustrates that the shape and values of the reference curves provide an appropriate description of the observed data. Inclusion of both reference curves and fixed reference intervals for PCV demonstrates that, at certain times of the year, fixed reference intervals could lead to inappropriate interpretation of observed values.
Having observed seasonal variation in these five parameters, we can consider why their values might be lower at one time of year and higher at another. The laboratory quality control results (PCV, albumin, and magnesium) and the lack of a correlation between control data and test data (albumin) suggest that analytical variation was not a contributing factor. Low PCV, RCC, Hb, and albumin have been observed at times of negative energy/protein balance, such as reduced nutrition, pregnancy, and lactation, which might also be expected to lead to low body condition (Greig and Boynea 1956; EI-Nouty and AI-Haidary 1990; Artacho et al. 2007). However, the absence of significant positive correlations between the seasonal parameters and body condition (Table 6) suggests that the observed seasonal variation in hematology/biochemistry parameters did not primarily reflect seasonal changes in overall net energy/protein balance.
Climatic conditions and physiologic rhythms may also cause seasonal changes in these parameters (Sealander 1964; DelGiudice et al. 1992; Collazos et al. 1998; Woods and Hellgren 2003). Annual sinusoidal variations in a range of physiologic and pathologic processes have been related to similar variations in climate that result from the very precise sinusoidal variations in solar patterns, such as day length and maximum solar elevation angle (Fröhlich et al. 1997; Dowell et al. 2003; Ockene et al. 2004). Patterns of seasonal variation similar to those we observed for PCV, RCC, Hb, albumin, and magnesium have been seen in a range of local temperature-related variables, including 1) mean daily air temperatures and mean monthly rainfall for Wynyard (Australian Government, Bureau of Meteorology 2016), 2) water temperature recorded at three sites in the Inglis River Catchment over 3 yr (Bobbi et al. 2003), and 3) the first cloacal temperature taken after induction of anesthesia and, to a lesser extent, the last cloacal temperature before recovery of anesthesia (Macgregor 2015). These observations raise the possibility of a link between climate and the five seasonally varying hematology/biochemistry parameters. Such a link could arise 1) as an artifact caused by variation in body temperature of platypuses during fieldwork, 2) as a result of variation in food availability (ruled out earlier), or 3) because of variations in metabolic demands. Our results suggest that the variations of the five hematology/biochemistry parameters were not related to body temperature at or shortly before sampling (Table 6) and are unlikely to be an artifact of capture, holding, or handling. However, correlations in Tables 7 and 8 suggest that the seasonally varying parameters vary with ambient temperature in recent hours or days or over a longer period.
Previous findings suggest that variations in metabolic demands may cause the correlations we observed between PCV and ambient (particularly water) temperatures. Increases in PCV or hematocrit (a parameter equal to or very closely related to PCV) over periods of hours to weeks have been observed with exposure to reduced ambient temperatures in a variety of species (Sutherland et al. 1958; Keatinge et al. 1984; Yahav et al. 1997). Such increases can be a response to the need to deliver increased amounts of oxygen to tissues to support an increased metabolic rate for heat production (Keatinge et al. 1984; Yahav et al. 1997). Platypuses can maintain their body temperature in air or water temperatures down to 5 C (Grant and Dawson 1978a), and Bethge (2002) reported a near doubling of metabolic rate between January and July. Because of increased conductance of the fur when wet, water temperature is the greatest determinant of daily metabolic rate (Grant and Dawson 1978a, b).
Although increased PCV with cold exposure appears to be a consistent finding, varying changes in plasma proteins can occur (Sutherland et al. 1958; Yahav et al. 1997). Sutherland et al. (1958) found that some hematology/biochemistry parameters other than PCV changed quickly with cold exposure and then returned to values closer to normal. At least in the early stages of cold exposure, increased PCV levels may have been mediated by decreased blood volume resulting from decreased plasma albumin levels (Sutherland et al. 1958; Yahav et al. 1997). This mechanism would not be consistent with the similar variation over time in our study of red blood cell– and albumin-related parameters. However, in our study, platypuses were not in an experimental situation but were exposed to temperatures changing gradually over months in the environment in which they have evolved. It is possible that, in this situation, elevated PCV levels in response to cold might be maintained by factors other than albumin-mediated changes in blood volume, such as increased red blood cell production. The seasonal changes in magnesium likely reflect similar changes in albumin, magnesium being ∼25% bound to albumin in the blood (Kroll and Elin 1985; De Swiet 2002; Randell et al. 2008).
Red cell parameters can be indicators of physiologic state, but their use can be limited by the scarcity of reference intervals, natural variations in these parameters, and uncontrolled environmental conditions (Artacho et al. 2007). Nevertheless, assessment of biochemistry and hematology results is an important tool. Regardless of the cause(s) of the seasonal variations we observed in five parameters, awareness of the potential for seasonal variation is important for assessing individual health. Given the large geographic distribution of the platypus and the variation of climate and breeding season within this range, it is likely that the timing, magnitude, and possibly occurrence of these seasonal changes will vary among populations. Our observations suggest that platypus researchers reporting hematology and biochemistry parameters should look for seasonal changes in their data.
The reference curves we developed are two dimensional, being based on observed values and DOY. Reference intervals are one dimensional and based only on observed hematology/serum biochemistry values. However, reference curves and reference intervals are alike, in that a specified proportion (in this case 95%) of observed values lie between the upper and lower reference limits of each (Friedrichs et al. 2012). As such, we propose that the reference curve is a new concept that fits within the scope of existing theory on reference values. We postulate that reference curves, possibly based on functions other than the sine wave, may be more appropriate than fixed reference intervals for representing variation associated with a range of other continuous factors, such as age, stage of pregnancy/lactation, or altitude.
This study was approved by the Animal Ethics Committee of Murdoch University, Western Australia (permit RW 2422/11), DPIPWE, Tasmania (permits to take Wildlife for Scientific Purposes FA 11131 and FA 12165), and Inland Fisheries Service, Tasmania (exemption permit 2011-10). Thank you to Helen Robertson and other fieldwork volunteers for assistance, and to Sarah Munks for guidance. Funding was provided by the Winifred Violet Scott Estate, a Holsworth Wildlife Research Endowment, a Caring For Our Country Community Action Grant (project CAG 11-00128), the Central North Field Naturalists, the National Geographic Society (grant C217-12), the Cradle Coast Natural Resource Management, Tasmanian Alkaloids, the Australian Geographic Society, and the Forestry Practices Authority. We are grateful to the Forest Practices Authority, Tasmania, the University of Tasmania, and the DPIPWE for the loan of field equipment.
Supplementary material for this article is online at http://dx.doi.org/10.7589/2015-12-336.