| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Evolution, Ecology, and Organismal Biology, The Ohio State University, Columbus, Ohio 43210
Submitted 29 July 2002 ; accepted in final form 26 March 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT MATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
total evaporative water loss; carbon dioxide production
CO
2).
CO2, as estimated by the DLW technique, has been compared with standard laboratory measurements of expired CO2 in species of mammals, birds, lizards, and the desert tortoise (Gopherus agassizii). Mean errors in these studies are generally ±11% (19), but individual deviations can vary by ±30% (32). However, the method has not been validated on snakes, i.e., ectotherms that have larger surface to volume ratios than most other animals (9, 17, 18, 22, 32, 35). Despite this lack of assurance that the DLW technique can provide reliable estimates of metabolic rate for snakes, it has been used on them in the wild in a number of studies (2, 4, 23, 25, 30). In addition, although snake diversity is greatest in tropical regions, most DLW studies on snakes have been conducted on species living in xeric and temperate habitats. This study validates the DLW technique for use in a tropical snake species.
The assumptions of the DLW method have been previously detailed (14). Violation of any of these assumptions can lead to errors in estimates of CO2, but the one that is most likely to be violated for studies on tropical forest snakes is that water or CO2 does not enter the body with inspired air or through the skin. In tropical environments, water vapor content of air is high, and its influx through the skin or mucous membranes and consequent dilution of the hydrogen and oxygen isotopes can potentially generate a significant source of error in DLW calculations. Working on kangaroo rats, Nagy and Costa (21) reported that errors in DLW estimates increased from 3 to 44% as relative humidity (RH) increased from 4 to 20%. Whereas the DLW technique was within ±10% when validated on small desert lizards in low-humidity conditions (9, 18), the technique overestimated energy expenditure by a factor of 
4 in small tropical lizards tested under high-humidity conditions (Ref. 17; Nagy KA, personal communication). In contrast, Van Marken Lichtenbelt (35) found only a small increase in errors (-1.5–7.4%) in DLW estimates of CO2 measured in adult green iguanas [air temperature (Ta): 30–35°C] as RH increased from 55 to 65%. Green iguanas are large, thick-skinned animals with a lower surface-to-volume ratio than the smaller lizards studied by Nagy and Costa (21), and the test range of RH was also well below saturation (35).
As part of a study of the biology of the introduced brown treesnake (Boiga irregularis) on the island of Guam, we conducted a validation study to assess the accuracy of the DLW technique in this tropical snake. We asked whether the DLW method could accurately predict CO2 and water flux in brown treesnakes, an arboreal ectotherm with a high surface-to-volume ratio that lives in an environment where the water vapor density is high. We found that in high humidity, DLW estimates of CO2 were several times greater than those resulting from indirect calorimetry. Given that DLW estimates can overestimate metabolism of snakes in humid environments, we recommend caution in interpretation of results from studies on this group that do not have validated methods. In addition, we found that it is difficult to validate the DLW technique for estimating CO2 in a tropical forest reptile under low-humidity conditions because large evaporative water losses result in physiological dehydration before the reptile produces enough CO2 to result in measurable drops in the oxygen isotope due to metabolism. In this case, it is necessary to design the validation study to maintain either a stable TBW pool or constant CO2.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
In this paper, total water efflux (TWE) represents total water molecules leaving the body. TWE does not infer a net loss and is used to describe DLW estimates of the total number of water moles exchanged between the body and the environment. Under low-humidity conditions, TWE approaches equality with net TWE because no water is available in the environment, so exchange of water with the environment is unidirectional, a net loss. Net total evaporative water loss (TEWL) represents the net loss of water molecules from the body through evaporation and is used to describe indirect calorimetry measurements of net water vapor loss.
The research reported here complies with The Principles of Animal Care, publication no. 86-23 (revised 1985) of the National Institutes of Health and with current laws of the state of Ohio.
Biology of Brown Treesnakes
Colubrid snakes native to Papua New Guinea, the Solomon Islands, Indonesia, and coastal Australia, i.e., brown treesnakes (B. irregularis), were introduced to the island of Guam in the late 1940s (27). Here, they inhabit the moist tropical forests, are chiefly arboreal, and are nocturnally active (27). Ta and RH within the rainforest ranges from 23 to 32°C (mean 27°C) and 40 to 100% (mean 72%), respectively (1). Average rainfall ranges from 8 to 40 cm/mo (27).
Validation of DLW Method in Brown Treesnakes in Low and High Humidity
We used adult brown treesnakes, captured on Guam, to compare DLW estimates of water flux and CO2
with values determined by indirect calorimetry and hygrometry techniques (11, 12, 15) under high-humidity conditions (n = 8; mean body mass = 130 ± 7 g) and under low-humidity conditions (n = 6; mean body mass = 134 ± 8 g). Snakes were housed at The Ohio State University for 3–6 mo before experiments (Ta: 24–29°C; 12:12-h light-dark cycle) and fed one mouse every 2–3 wk. Water was provided ad libitum. Snakes were fasted for 1 wk before each study. Because brown treesnakes on Guam feed approximately once every 4 days (1), we fed each snake one 3- to 5-g skink (Carlia fusca), their most common prey species, every 4 days of the experiment (27).
Before each trial, we weighed each snake (±0.1 g) and obtained three 15-µl blood samples (flame sealed in glass capillary tubes) via the tail vein or by cardiocentesis for measurement of background levels of D and 18O. All isotope concentrations were measured by Henk Visser by using mass spectroscopy at The Centre for Isotope Research, Gronigen, The Netherlands. We administered 0.006 g of DLW/g snake intraperitoneally (2.5 ml of 99.9% D2O in 25.0 ml of 10.0% H218O). After injection, we placed each snake in a cage for 2 h to allow the DLW to equilibrate with the TBW pool and then drew blood to determine the initial concentration of isotopes.
After feeding the snakes, we placed each snake in a wire mesh cage that we lowered into a 1.25-liter metabolic chamber and then sealed the chamber with a metal lid. Below a wire platform, we placed mineral oil on the bottom of the chamber to collect feces and urine. Chamber Ta was maintained at 26°C, the approximate mean body temperature of free-ranging brown treesnakes on Guam (1), by circulating water through copper coils surrounding the chamber. Water temperature was controlled (±0.1°C) by a Neslab circulating water bath (Neslab, Portsmouth, NH).
Air from a cylinder flowed through columns of Drierite (W. A. Hammond Drierite, Xenia, Ohio), soda lime, and Drierite (dew point temperature: less than -37°C, RH of <0.1%) at a rate of 198.7 ml/min STPD for the high-humidity trials and 102 ml/min in the low-humidity trials. For the low-humidity trials, air directly entered the chamber, but for the high-humidity trial, the air stream was humidified to a dew point temperature of 24.9°C (RH = 93.7%) by a dew point generator (Li-cor, Lincoln, NE) before entrance to the metabolism chamber. Water vapor was trapped downstream of the metabolic chamber in a column of Drierite that we changed and weighed (±0.1 g) daily. We measured the parts/million (ppm) of CO2 in air exiting the chamber with an infrared CO2 analyzer (Li-cor) calibrated with a primary CO2 standard (1,879.8 ppm). Exiting air also passed through Ascarite, as an additional quantification of CO2 (Mallinckrodt and Baker, Phillipsburg, NJ), then through another tube of Drierite. Both columns were weighed at the beginning and end of each trial (±0.0001 g), and the differences in masses were added to give a gravimetric measure of CO2. The linear regression coefficient between the analyzer estimates and gravimetric measurements of the mass of CO2 was not significantly different from 1.0 (r2 = 99.9%, P < 0.0001). Voltages were averaged over 5-min intervals by a Campbell data logger using PC208W software (Campbell Scientific, Logan, UT). Data from the logger was downloaded into Excel (Microsoft, Redmond, WA) spreadsheets for conversion of voltage values to data parameters (APPENDIX A) and for further data analysis.
After 4 days, we removed each snake from the chamber, weighed and fed it, and took a blood sample (<0.25 ml) for determination of isotope concentrations; the entire process required <30 min. If snakes lost >2 g of body mass (1–2% body mass loss), we injected them intraperitoneally with enough 0.9% saline (±0.0005 g via Mettler balance) to balance mass loss. We scored excrement in the chamber as not present (0 g), scant (0.1–1.9 g), moderate (2.0–3.9 g), or large (4.0–5.9 g). Wet mass estimates of feces were based on masses of similar fecal volumes obtained from other captive brown treesnakes (1). Because of the lack of precision of the fecal mass estimates (±2 g) and the large volume of urine associated with feces, we assumed that fecal masses represented fecal water loss.
Each snake was returned to the metabolism chamber and remained there until day 8, when we took a final blood sample. We converted ppm CO2 for each 5-min period into measurements of CO2 by using standard equations (11, 12) and then summed these to calculate the total mass of CO2 produced per 4-day period and for the entire 8-day period. Data are presented as ml · g-1
· h-1 to allow comparison with data from other studies.
Data Analysis
Statistics. We used Minitab Release 13 (Minitab, State College, PA) to perform linear regression and basic statistics. Statistical significance was set at P < 0.05. For snakes that defecated during trials, DLW estimates for TWE- (water lost in feces) were compared with indirect calorimetry measurements of net TEWL (or estimates of net TEWL for the high-humidity trial). Mean error was calculated as (Estimated mean using DLW-control measurement mean)/control measurement mean. In addition to t-tests (paired when appropriate), we used linear regression (zero intercept) to compare DLW estimates of CO2 and TWE to indirect calorimetry measurements and to determine whether equations correcting for fractionation improved DLW estimates.
Preliminary Studies and Results Performed to Ensure Appropriate Use of the DLW Technique
Slope of D-to-slope of 18O ratio. The rate of decline of D and 18O in the TBW pool of an animal injected with DLW is exponential (32). A logarithmic plot of this decline provides a slope that is called the fractional turnover rate (k). The slope of the D washout curve (kd) is used to calculate water influx and the difference between kd and 18O (ko) washout curves is used to calculate CO2. When water turnover is high relative to CO2, the ratio of the slope of the washout curves (ko/kd) approaches one, and the precision of the technique rapidly decreases (26). A ko/kd of >1.1 is necessary to ensure that errors inherent in measurement of isotope concentrations will not significantly influence results of DLW estimates (32). Because previous independent indirect calorimetry studies (1) found that brown treesnakes had high net TEWL rates compared with CO2, we calculated ko/kd for all time periods (0–4, 4–8, and 0–8 days) and humidity trials. All ratios (1.15–1.3) were >1.1.
Determination of Percent TBW of Brown Treesnakes
To determine whether the percent of TBW (%TBW) varied with body mass, five brown treesnakes (78.35–159.00 g) were killed with <0.2 ml of pentobarbital sodium solution, weighed (OHaus scale model no. ts400s), cut open, dried at 65°C for 6 days, and reweighed. We found no relationship between body mass and %TBW in brown treesnakes (mean %TBW: 66.2 ± 1.05%; P > 0.24).
TBW has been predicted by using dilution space of D and 18O (32). We compared dilution space of both isotopes with measured values, and we evaluated whether TBW changed between humidity trials. We found no difference in %TBW determined by drying between the high- and low-humidity trials (P > 0.25). For the low-humidity trial, mean %TBW determined by either 18O (66.3%) or D (68.2%) did not differ significantly from 66.2% (P > 0.11). For the high-humidity trial, mean %TBW determined by either 18O (67.7%) or D (69.4%) was >66.2% (P < 0.047). The difference was due to one snake (no. 2,719; 70.48% TBW) that was tested only in the high-humidity trial. Deuterium overestimated the %TBW determined by O18 by 2.77% in the low-humidity trial and by 2.60% in the high-humidity trial. We used the value of 66.2% to calculate TBW in this study.
DLW Technique
Low-humidity trial CO2
and TWE and high-humidity trial CO2. We calculated TWE and CO2
by using equations that did not correct for a change in body mass, that corrected for a linear change in body mass, and that corrected for an exponential change in body mass (16, 21, 32). In addition, we used several equations to investigate the effects of correction for fractionation (APPENDIX A).
High-humidity trial TWE. A validation of DLW estimates of TWE under humid conditions was problematic by using standard indirect calorimetry techniques because changes in isotope concentrations reflected total exchange of water molecules between the animal and the environment, whereas indirect calorimetry measured the net difference in water produced by the snake over the amount of water added to the system. We assumed that, at any given Ta, the rate of evaporation of water through skin was a first-order reaction [i.e., a constant property of the skin and not influenced by the concentration of water molecules (i.e., humidity) at the skin surface (24, 36)]. Therefore, we predicted that the total number of water molecules lost through evaporation through the skin in a normally hydrated snake should remain the same between high- and low-humidity trials. Thus net TEWL measured during the first 24 h of the low-humidity trial (multiplied by 4 to equal the 4-day DLW estimate period) should be an approximation of TWE in the high-humidity trials. We compared DLW estimates of TWE corrected for fecal water loss to average net TEWL (0.1301 g H2O/g snake per 4 days) and maximum net TEWL (TEWLmax; 0.2007 g H2O/g snake per 4 days).
Comparison of TWE, TEWL, and CO2 Between Humidity Trials and Time Periods
We compared DLW estimates of TWE, net TEWL measured gravimetrically, and CO2 measured by the CO2 analyzer for differences between the humidity trials (CO2 only) and time periods (0–4 and 5–8 days).
|
RESULTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Use of equations to correct for fractionation increased the mean error, decreased r2 values, and increased the range of 95% confidence interval of regression coefficients of DLW estimates of TWE and CO2
(33). We found that use of a linear correction for body mass change minimized the mean error and maximized r2 values compared with other equations for both TWE and CO2
(Table 1). Therefore, we used Eq. 4 of Nagy and Costa (21) and Eq. 2 of Nagy (16) to calculate TWE and CO2, respectively.
|
Accuracy of DLW Estimates of TWE
Low humidity. For all time periods, the linear DLW model yielded a mean error of -3.7 to -1.0% and regression coefficients that were not significantly different than unity (Table 1). With the use of paired t-tests, we found no difference between gravimetric and DLW estimates of water efflux (Table 1) for any time period (P > 0.14). In dry air, DLW estimates of TWE- fecal water loss were equivalent to gravimetric estimates of net TEWL.
High humidity. During high-humidity trials, average net TEWL accounted for 63–95% and TEWLmax accounted for 88–132% of DLW estimates of TWE- fecal losses. For all time periods, the linear DLW model- fecal water loss vs. TEWLmax produced mean errors from -5.2–6.8% and regression coefficients that were not significantly different than unity (Table 1). With the use of paired t-tests, we found that linear model DLW estimates of TWE- fecal water loss (Table 1) were not significantly different from TEWLmax for any time period (P > 0.32).
Accuracy of DLW Estimates of CO2
Low humidity. When compared with open-system respirometry measures of CO2, DLW estimates produced mean errors of 78–168% with standard deviations of >63% for all time periods. Regression coefficients ranged from 1.7 to 2.7 (mean 2.2), and r2 values ranged from 67 to 94% (Table 1). Because the regression was forced through the origin, the regression coefficients indicated that the DLW technique overestimated the CO2 of brown treesnakes by 1.7–2.7 times.
High humidity. When compared with the CO2 analyzer measurements, the DLW technique produced mean errors of 308–415% for all time periods. Regression coefficients ranged from 4.1 to 5.2 (mean 4.6). Values for r2 were 98.2–99.6%, suggesting a tight correlation (Table 1). Overall, the DLW technique overestimated the CO2 of brown treesnakes by a factor of 4.6.
Comparison of TWE and CO2 Between Low- and High-Humidity Trials
TWE estimated by the DLW technique for high-humidity trials was greater than low-humidity trials (P < 0.0001) (Table 1). Mean CO2 measured with the CO2 analyzer for high-humidity trials was greater than low-humidity trials for 0–4 days (P < 0.0001) (Table 1). There was no difference between the humidity trials for 5–8 days (P > 0.10) (Table 1).
Comparison of TWE, TEWL, and CO2 Between Days 0–4 and 5–8
TWEs estimated by the DLW technique were not different between 0–4 and 5–8 days for either humidity trial (P > 0.138) (Table 1). For the low-humidity trial, net TEWL measured gravimetrically showed that snakes lost 7% more water during 5–8 days than during 0–4 days (P = 0.02) (Table 1). The rate of CO2 measured by the CO2 analyzer was not different between 0–4 and 5–8 days for the high-humidity trial (P = 0.88) (Table 1). For the low-humidity trial, CO2 was 13% higher during 5–8 days than during 0–4 days (P = 0.03) (Table 1).
|
DISCUSSION |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
To evaluate the validity of the errors in the DLW method reported in this study for brown treesnakes, we first established that our indirect calorimetry measurements were reasonable for this species. Mean CO2 measured via indirect calorimetry measurements during the first 4 days of this study was 0.0379 ml CO2 · g-1 · h-1, or 0.0519 ml O2 · g-1 · h-1 if we assume a respiratory quotient of 0.73 (11). Our value of oxygen consumption (O
2) is within the range reported for standard metabolic rate (SMR) in other colubrid snakes of similar body mass at 25°C [Thamnophis elegans vagrans, 107.5 g, 0.03 ml O2 · g-1 · h-1
(34); Elaphe guttata, 144 g, 0.054 ml O2 · g-1 · h-1
(31); Natrix rhombifera, 237.5 g, 0.084 ml O2 · g-1
· h-1 (13)]. This concordance suggests that our measurements of metabolic rate were reasonable.
However, because snakes were fed during the study, we also considered the effects of feeding on metabolism, called the specific dynamic action (SDA). Secor showed that increase in O2 as a result of SDA was positively correlated with meal size and ranged from 1.4 to 5 times SMR in pythons fed 0% (allowed to catch and constrict a prey item, but not swallow it) to 5% of their body mass (28). Increases in O2 due to SDA are 2.4 times higher in infrequently feeding species such as pythons when compared with frequently feeding species such as colubrids (29). Because brown treesnakes are frequent feeders, and because we fed meals 2–3% of body mass, we expected peak O2
associated with digestion to be increased by a maximum factor of 2 (5/2.4). CO2 values that we recorded every 5 min showed that peak CO2 associated with digestion occurred 24 h after feeding and returned to fasting levels after 2.6 days (1). From this information, we calculated mean SMR to be 0.037 ml O2 · g-1 · h-1 and peak O2 associated with digestion to be 0.067 ml O2 · g-1 · h-1 (1), the latter representing an increase by a factor of 1.8. These calculations indicate that the large discrepancies found in this study between DLW estimates and indirect calorimetry measurements of CO2 were not the result of failure of the indirect calorimetry technique to measure increases in metabolism associated with SDA.
Snakes living in dry environments have lower rates of TEWL than those living in moist environments (10). Hence, comparisons with other snakes need to be habitat specific. The range for indirect calorimetry measurements for mean TEWL in brown treesnakes for the low-humidity trial was 0.94–1.02 mg H2O · g-1 · h-1. These values are within the range reported for other colubrid snakes living in mesic habitats, having similar body mass, measured between 25 and 30°C, and tested in dry air [Pituophis melanoleucus catenifer, 46.6 g, 0.80 mg H2O · g-1 · h-1 (10); Elaphe climacophora, 70.0 g, 0.92 mg H2O · g-1 · h-1 (10); Lampropeltis doliata triangulum, 119.8 g, 1.07 mg H2O · g-1 · h-1 (10); Lampropeltis getulus, 81.7 g, 2.10 mg H2O · g-1 · h-1 (8)].
Accuracy of the DLW Technique in Measuring TWE
Under low-humidity conditions, the linear model for DLW accurately predicted net water efflux in brown treesnakes (regression coefficient = 1; mean error < 4%). This accuracy is similar to findings in birds, mammals, and other reptiles (19, 20, 32). Although TWE in brown treesnakes was mostly through evaporative losses, suggesting that fractionation effects should be significant, traditional methods to correct for fractionation decreased accuracy. Under high-humidity conditions, DLW estimates of TWE in brown tree snakes were equivalent to maximum net TEWL values measured in normally hydrated individuals tested under low-humidity conditions.
Accuracy of the DLW technique in measuring CO2
In the low-humidity trial, the DLW method overestimated O2 as measured by indirect calorimetry by 200%. The error was not a result of fractionation effects because fractionation can account for a maximum error of 40% and the resulting error is to underestimate isotope turnover. The error was the result of violation of the assumption that the TBW pool and CO2 remained constant (16) and illustrates the difficulties inherent in validating DLW estimates of CO2 in animals with low metabolic rates and high water losses. Because only a small proportion of the total decline of the oxygen isotope is due to CO2, it is problematic to obtain a measurable drop in the oxygen isotope attributable to metabolism before an animal becomes dehydrated. Our solution was to replace water losses with intraperitoneal saline administered when snakes were fed. Although this technique allowed us to minimize handling effects and accurately measure TWE, fluid losses were only replaced every 4 days, and the TBW pool changed by 12%. This change in TBW combined with variations in CO2 associated with handling and SDA contributed to the observed error of 200% (16).
Solutions to the dilemma of balancing high rates of water loss with low metabolic rate include increasing metabolic rate, decreasing water loss, or allowing more frequent access to water to maintain a stable TBW pool. Larger meals would have resulted in a higher metabolic rate and a larger water influx. Increasing Ta would have increased metabolic rate but would also have increased evaporative water losses. Applying a water-impermeable coating to the skin (such as mineral oil or petroleum jelly) would have decreased evaporative water losses. More frequent handling of the snake to inject or administer fluids would have resulted in more variation in CO2 but would have minimized changes in the TBW pool.
In the humid trial, body mass of snakes changed by <1.5%. The linear model for DLW overestimated CO2 by a mean factor of 4.6 (range 4.1–5.2). None of the equations traditionally used to calculate fractionation effects accounted for this difference. One hypothesis for this overestimate was continuous diffusion of 18O molecules from labeled CO2 molecules to water molecules because the concentration of 18O molecules was decreasing disproportionately with the rapid turnover of the TBW pool (Nagy, personal communication). A correction factor of 4–5 is likely inappropriate for DLW work conducted on other reptiles in high-humidity conditions because errors are likely to be dependent on the rate of evaporative water turnover compared with metabolic rate and, therefore, influenced by humidity, physiological status of the organism, wind speed, physical characteristics of the skin, and other factors. However, the error reported in this study does emphasize the need to validate the DLW technique when used in humid conditions.
In the past, researchers using DLW have found that field metabolic rates (FMR) of reptiles are higher when animals are tested in high-humidity environmental conditions than when they are tested in low-humidity conditions (3–7). In Christian et al.'s (5) study on marbled geckoes (Oedura marmorata), SMR from populations living in high-humidity vs. arid habitats were similar, whereas FMR measured in the field with DLW was uniformly higher for the geckoes living in the moist conditions. Differences in body temperature did not account for the difference in FMR. Most researchers have assumed that FMR is higher in wet conditions because animals increase their food consumption/activity levels because of greater food availability. Whereas this assumption may be warranted, our study indicates that FMR values may be inflated when large amounts of water vapor are absorbed through the skin or respiratory passages. With the use of DLW, Peterson et al. (23) found that garter snakes (Thamnophis sirtalis) living in a semiaquatic environment had a FMR that was 2.5 times greater than other colubrid species, such as northern racer (Coluber constrictor) (25) and coachwhip snake (Masticophis flagellum) (30), which live in mesic and arid environments, respectively. Peterson et al.'s study (23) on garter snakes ruled out dilutional amplification of FMR by showing that feeding rates calculated from water influx rates were equivalent to feeding rates calculated from metabolic and growth rates. Peterson et al. (23) attributed the high FMR to high feeding rates (continuous addition of energy expended on digestion, SDA) resulting from a limited foraging season.
Comparison of TWE, TEWL, and CO2 Between Humidity Trials and Time Periods
Snakes in high-humidity trials turned over approximately twice as many water molecules than snakes in low-humidity trials. This finding may seem counterintuitive at first, since snakes lost more body mass in low humidity trials. However, although the loss of body mass in the low-humidity trial was a good indication of net water turnover (i.e., true water loss) because there was no moisture available to replace evaporated water molecules, this was not true for the high-humidity trial. In the high-humidity trial, the DLW technique measured the total turnover of water molecules rather than the net turnover. In the high-humidity trial, snakes lost more water but were able to replace losses by absorption of water vapor through their mucous membranes and/or skin, so there was no concurrent body mass loss or net loss of water. Caution should be used when the DLW technique is used to estimate net water requirements for animals in humid environments because it is likely to result in overestimates.
The rate of CO2 production in brown treesnakes was greater during high-humidity trials than low-humidity trials, likely because snakes were less active in the latter. A reduction of activity would decrease water loss through the respiratory passages and, probably more importantly, allow snakes to position their coils to minimize exposed surface area and thereby reduce transcutaneous water loss. The lack of fecal production during the low-humidity trial (1 scant defecation in twelve 4-day trials) compared with the high-humidity trial (11 defecations in sixteen 4-day trials) is evidence that snakes in the low-humidity trials were using strategies to conserve water.
Suitability of the DLW Technique for Humid Environments
Although the DLW method has been widely used, its reliability is in doubt when used on snakes in humid environments. This study shows that without a validation study, errors of up to 500% are possible when the DLW technique is used on snakes living in moist, tropical environments. It is reasonable to assume that errors of this magnitude are likely in other small ectotherms living in humid environments.
|
APPENDIX |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
Voltages from the thermocouples were converted by the Campbell data logger to temperatures within 0.1°C (Campbell Scientific User manual).
The voltages from the Li-cor CO2 analyzer were converted into CO2 concentration ([CO2]; in ppm) in the air stream by the following equation (Li-cor User's Manual)
 |
Percent TBW
After a blood draw to determine isotope levels, snakes that lost >2 g of body mass after the first 4-day trial received injections of saline. Because it was rapidly absorbed, this saline diluted the concentration of isotope from the preinjection concentration. To account for this, we calculated the starting TBW pool for 4–8 days as %body water (original body mass at 4 days + mass NaCl + mass skink fed) entered into equations for DLW calculations. This equation modeled the expected concentration of isotope once the percentage of TBW had normalized back to 66.2% (i.e., after dehydration was corrected by absorption of water in the saline).
Calculations
D and 18O concentrations were reported in delta SMOW units. In general, three values for background, initial, day 4, and day 8 concentrations were reported for each snake. We determined the mean delta SMOW [difference in isotope concentration from standard mean ocean water (32)] for each time period and each snake. Mean delta SMOW units were converted to absolute ratio for deuterium by using Eq. 14.4 of Speakman (32) {Ratiosample = [(delta sample SMOW/1000) + 1] · 0.00015595} and Eq. 14.9 of Speakman (32) {Ratiosample = [(delta sample 18SMOW/1000) + 1] · 0.0020052}. Absolute ratios were converted to ppm by using Eq. 14.9 {[Ratiosample/(Ratiosample + 0.000373 + 1)] · 1000000} for 18O and Eq. 14.8 from Speakman {[Ratiosample/(Ratiosample + 1)]/0.000001} for D (32). The background concentration of isotopes was subtracted from initial, day 4, and day 8 values to give concentrations in ppm over background.
Nagy and Costas' (21) equations for constant body water (Eq. 3), linear (Eq. 4), and exponential (Eq. 5) changes in body water volume were used to calculate g H2O efflux · g-1 · h-1
 |
|
|
|
|
where W is mean body water, H22 is the final concentration of D, M is the body mass, LN is natural logarithm, and t is the length of the trial.
Nagy's (16) equations for constant body water (Eq. 1), linear (Eq. 2), and exponential (Eq. 3) changes in body water volume were used to calculate milliliters of CO2 efflux per gram per hour of D, H2
|
|
|
|
 |
where M1 is initial wet body mass, M2 is final wet body mass, W1 is initial weight body water, and W2 is final weight body water.
For TEWL, each of the values for the constant, linear, exponential models was used in Coward and Cole's equation [from Speakman (32)] for fractionation. This equation allows the investigator to enter fractionation effects ranging from 0 to 100%. The investigator can also vary the value of the fractionation factor (F1) between the lowest and highest values reported. In the Excel spread sheet, we calculated fractionation effects every 10% from 0 to 100% at the F1 of 0.941 (assuming 100% equilibrium fractionation), 0.917 (the lowest recorded value assuming 100% kinetic fractionation), and 0.93225 (assuming 50% equilibrium and 50% kinetic fractionation). The maximum TEWL and the minimum TWE from this matrix were used to examine the greatest correction possible due to fractionation
|
|
where x is fractionated proportion of water, N is moles of body water and F1 is of D/H2O vapor over water.
For CO2 production, of the three analyses, the linear equation had the smallest mean error when compared with indirect calorimetry data. To limit the number of permutations to a manageable level, the linear model was used as a basis to investigate the effects that correction for fractionation would have on relative error. The values for CO2 and TEWL from the linear models were entered into Lifson and McClintocks' (14) Eq. 35 (assumes 50% H2O loss is fractionated, 100% equilibrium losses, average body temperature of 25°C)
|
|
Speakman's (32) Eq. 7.17 (assumes 25% fractionated water losses, 3:1 equilibrium-to-kinetic ratio, body temperature of 37°C) shows
|
|
and Coward and Cole from Speakman (32) show
|
|
where F2 is fractionation factor of 18O H2O vapor over water, F3 is fractionation factor of 18O CO2 over water, and x is fractionated proportion of water.
For Coward and Coles' equation (32), we simulated the fractionated proportion of water to vary in 10% increments from 0 to 100%. There is only one value for F3 (1.039). The possible values for (F2 - F1)/2F3 can only vary from 0.02478 to 0.02526. Because the magnitude of change of this coefficient is small and its variable is subtracted from a much larger number, use of the minimum vs. maximum value has minimal effects on the estimated CO2. We chose to use a value of 0.0249 that is the most commonly used value that assumes a 3:1 contribution of equilibrium to kinetic fractionation. The maximum TWE and the minimum TWE from this matrix were used to examine the greatest correction possible due to fractionation.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
This research was supported by funds from the Biological Resources Division of the United States Geological Service and the Kansas City Herpetological Society.
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Present address for N. Anderson: Lindsay Wildlife Museum, 1931 First Ave., Walnut Creek, CA 94597.
|
REFERENCES |
|---|
|
TOPABSTRACTMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXACKNOWLEDGMENTSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
