Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Estimating metabolic rate in tegu lizards (Tupinambis merianae) : can one calculate oxygen consumption… Piercy, Joanna Carolyne 2005

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2005-0597.pdf [ 7.78MB ]
Metadata
JSON: 831-1.0092147.json
JSON-LD: 831-1.0092147-ld.json
RDF/XML (Pretty): 831-1.0092147-rdf.xml
RDF/JSON: 831-1.0092147-rdf.json
Turtle: 831-1.0092147-turtle.txt
N-Triples: 831-1.0092147-rdf-ntriples.txt
Original Record: 831-1.0092147-source.json
Full Text
831-1.0092147-fulltext.txt
Citation
831-1.0092147.ris

Full Text

E S T I M A T I N G M E T A B O L I C R A T E I N T E G U LIZARDS (TUPINAMBIS MERIANAE): C A N O N E C A L C U L A T E O X Y G E N C O N S U M P T I O N F R O M H E A R T RATE? by Joanna Carolyne Piercy B.Sc, M c G i l l University, 1996 A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in The Faculty of Graduate Studies (Zoology) University of British Columbia September, 2005 ©Joanna Carolyne Piercy, 2005 A B S T R A C T Estimating Metabolic Rate in Tegu Lizards (Tupinambis merianae): Can One Calculate Oxygen Consumption from Heart Rate? by Joanna Piercy There are a number of methods commonly used to estimate metabolic rate in animals, one of which is the Heart Rate Method. Heart rate (fn) can theoretically be used as a direct proxy for O z consumption ( V 0 ) according to Fick's equation ( V 0 = fH x 0 2 pulse) as long as the 0 2 pulse varies in a predictable manner. The benefit of this method is that heart rate is easily monitored both in the lab and in the field, is sensitive to short-term changes in activity, and can be monitored continuously on the order of weeks or months. Studies have confirmed that in some birds, mammals, and at least one species of reptile, the accuracy of this method is at least as great as that of the doubly-labelled water method. My goal was to determine if heart rate can be used as a reliable estimate of metabolic rate in tegu lizards. To this end, fu and V D were measured simultaneously in fasting lizards at 17°C, 27°C and 37°C, and in digesting lizards at 37°C. Regression analysis showed that, at any given temperature or digestive state, the relationship between fu and V 0 was highly variable between individuals (as indicated by different slopes and intercepts) as well as within individuals (as indicated by low r2 values). Regression analysis on data pooled within each treatment failed to account for this variability. However, regression analysis of the dataset in its entirety accounted for 7 4 0 / 0 of the variability. The equation which best described the relationship in unstressed animals was lnsV 0 = —3.441 + 0 . 6 7 9 In/ R • Predictions of total metabolic rate calculated from this general calibration equation were typically more accurate than predictions based on the Time Energy Budget method. 11 T A B L E OF C O N T E N T S Topic Page Abstract ii Table of Contents iii List of Tables v List of Figures viii List of Abbreviations x Glossary xi Acknowledgments xii C H A P T E R 1: Introduction 1 Methods available to estimate field metabolic rate: 1 Validation of the Heart Rate Method in different species 4 Cautionary notes on the Heart Rate Method 4 Tupinambis merianae 10 A note on taxonomy. 10 Natural history 10 Current questions involving metabolism in tegu lizards 11 The potential of the Heart Rate Method in tegu lizards 12 Flypotheses 14 C H A P T E R 2: Changes in the rate of oxygen consumption and associated circulatory and respiratory variables with changes in temperatures and activity in tegu lizards 15 Introduction , 15 Methods F7 24 hour confined calibration experiments 18 Experimental setup 18 Training and Controls 20 Surgical Procedure 21 Experimental Protocol 21 3 hour roarning calibration experiments 24 Experimental setup 25 Surgical Procedure 25 Experimental Protocol 25 Data analysis 26 Results 30 Effects of mass 31 Controls 31 iii T A B L E O F C O N T E N T S Topic Page Effects of temperature and state on gas exchange. 31 24 hour confined experiments 31 3 hour roarriing experiments 35 Effects of digestion on gas exchange 40 Discussion 41 Does technique make a difference? 44 Effects of temperature 47 Effects of state 48 Relative contributions of heart rate and O 2 pulse to O 2 consumption 49 Relative contributions of breathing frequency, tidal volume and O 2 extraction to O 2 consumption 51 Does RER change as a function of changes in temperature or changes in activity? 52 Summary 53 CHAPTER 3: The relationship between heart rate and the rate of oxygen consumption in tegu lizards at different temperatures and digestive states 56 Introduction 56 Methods 59 Data analysis 59 Results 61 Heart rate vs. O 2 consumption 61 24 hour confined experiments 61 The Effect of Sample duration 68 3 hour roaming experiments 70 O 2 pulse 77 Breadiing frequency vs. O 2 consumption 79 24 hour confined experiments 79 3 hour roaming experiments 85 Discussion 92 The correlation between heart rate and metabolic rate 92 Accounting for the variability in the heart rate and oxygen consumption data...92 The correlation between breathing frequency and metabolic rate 102 Summary and future directions for research 103 C H A P T E R 4: Conclusion 105 References 107 Appendix 113 iv LIST OF T A B L E S Number Caption Page Table 1: Q 1 0 values of directly measured variables for 24 hour confined data. Values in bold correspond to significant changes in the variable for the given activity and temperature transition 35 Table 2: Q 1 0 values of directly measured variables for 3 hour roaming data. Values in bold correspond to significant changes in the variable for the given activity and temperature transition 40 Table 3: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each temperature and activity, as well as for pooled data at each temperature and activity. Values in bold indicate that the relationship was significantly different from the mean mass-specific 0 2 consumption. When N=0 or 1, regression analysis could not be performed. Similarly, when N=2, statistical significance could not be tested 64 Table 4: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each temperature, as well as for pooled data at each temperature. Values in bold indicate that the relationship was significant 67 Table 5: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each activity as well as for pooled data at each activity. Values in bold indicate that the relationship was significant 67 Table 6: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for pooled untransformed and pooled transformed data for 24 hour experiments. Values in bold indicate that the relationship was significant 68 Table 7: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each temperature and activity, as well as for pooled data at each temperature and activity. Values in bold indicate that the relationship was significantly different from the mean mass-specific 0 2 consumption. When N=0 or 1, regression analysis could not be performed. Similarly, when N=2, statistical significance could not be tested 73 Table 8: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each temperature, as well as for pooled data at each temperature. Values in bold indicate that the relationship was significant 76 v LIST O F T A B L E S ( C O N T I N U E D ) Number Caption Page Table 9: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each activity as well as for pooled data at each activity. Values in bold indicate that the relationship was significant 76 Table 10: Heart rate vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for pooled untransformed and pooled transformed data for 3 hour experiments. Values in bold indicate that the relationship was significant 77 Table 11: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each temperature and activity, as well as for pooled data at each temperature and activity. Values in bold indicate that the relationship was significantly different from the mean mass-specific 0 2 consumption. When N=0 or 1, regression analysis could not be performed. Similarly, when N=2, statistical significance could not be tested 81 Table 12: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each temperature, as well as for pooled data at each temperature. Values in bold indicate that the relationship was significant 84 Table 13: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for each lizard at each activity as well as for pooled data at each activity. Values in bold indicate that the relationship was significant 84 Table 14: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 24 hour confined data. Simple linear regressions were calculated for pooled untransformed and pooled transformed data for 24 hour experiments. Values in bold indicate that the relationship was significant 85 Table 15: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each temperature and activity, as well as for pooled data at each temperature and activity. Values in bold indicate that the relationship was significantly different from the mean mass-specific 0 2 consumption. When N=0 or 1, regression analysis could not be performed. Similarly, when N=2, statistical significance could not be tested 87 Table 16: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each temperature, as well as for pooled data at each temperature. Values in bold indicate that the relationship was significant 90 v i LIST OF T A B L E S ( C O N T I N U E D ) Number Caption Page Table 17: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for each lizard at each activity as well as for pooled data at each activity. Values in bold indicate that the relationship was significant 90 Table 18: Breathing frequency vs. mass-specific 0 2 consumption regression variables and estimation of fit (r2) for 3 hour roaming data. Simple linear regressions were calculated for pooled untransformed and pooled transformed data for 3 hour experiments. Values in bold indicate that the relationship was significant 91 Table 19: A comparison of the prediction accuracies of metabolic rate of two different methods in two different individuals. For any given comparison, the prediction value closest to the actual value of metabolic rate is indicated in bold. Note that the two 37°C/digesting treatments correspond to those in Figure 15. See text for details 106 Table 20: Means of raw data and derived variables for 24 hour confined experiments 113 Table 21: Means of raw data and derived variables for 3 hour roaming experiments 114 vii LIST O F F I G U R E S Number Caption Page Figure 1: Change in 0 2 pulse with oxygen consumption when the regression of heart rate on oxygen consumption is linear. A) When 0 2 pulse is constant, the regression of heart rate on oxygen consumption passes through the origin. B) When 0 2 pulse is curvilinear, the regression has a positive intercept. Adapted from Butler (1993) 6 Figure 2: Experimental set-up. Three heart rate electrodes are represented by a single yellow disk. The mask is represented by a dashed line. Exhalation dynamics are indicated in red; Inhalation dynamics are indicated in blue. (dP) differential pressure transducer; (dc) drying column; (fH) heart rate; (vE) ventilation; (v02) oxygen consumption; (vC02) carbon dioxide production. (Adapted from Wang & Warburton, 1995) 19 Figure 3: Typical postural differences between: A) "Sleep", B) "Quiet", C) "Alert", and D) "Moving" as viewed from above 23 Figure 4: Typical traces of changes in FIQ (%02), F,CO (%C02), EKG and respiratory air flow (i.e. breathing trace) for eacfi activity level! Each trace is taken from the same animal during the 27°C/fasting trial (24 hour confined experiments), and represents 2 minutes of recording 30 Figure 5: Mean values of each variable in Fick's equation for 24 hour confined data. Values for activities that were statistically different from those measured in the quiet state are indicated by a * symbol, and values for temperatures that were statistically different from those measured at 27°C are indicated by a • symbol. Black symbols indicate differences that are independent of temperature (*) or that are independent of activity (•), while coloured symbols indicate incidents that are dependent on both factors (i.e. statistical interactions between activity and temperature) 32 Figure 6: Mean values of each variable in Fick's equation for 3 hour roaming data. Values for activities that were statistically different from those measured in the quiet state are indicated by a * symbol, and values for temperatures that were statistically different from those measured at 27°C are indicated by a • symbol. Black symbols indicate differences that are independent of temperature (*) or that are independent of activity (•), while coloured symbols indicate incidents that are dependent on both factors (i.e. statistical interactions between activity and temperature). Note that only two of the six lizards slept during these experiments, and only during the 17°C trial. For this reason, sleep values were excluded from statistical analysis 36 V l l l LIST OF F I G U R E S ( C O N T I N U E D ) Number Caption Page Figure 7: Mean values of A) mass-specific 0 2 consumption (standardized by the scaling factor M 0 7 9), B) mass-specific C 0 2 production, C) breathing frequency and D) tidal volume (standardized by the scaling factor M°75) in the present study, and in four other studies. Data from other studies were plotted corresponding to the most similar temperature and activity level (i.e. X-axis categories) used in the present study to facilitate comparison. Note that data from other studies were converted to match the units used in the present study. Data from Klein et al. (2003) are approximated from visual inspection of Figures 2 and 3 in same 43 Figure 8: Pooled heart rate and mass-specific 0 2 consumption data from 24 hour confined experiments, colour coded by A) treatment, and B) activity 66 Figure 9: The effect of sampling duration on regressions of mass-specific 0 2 consumption on heart rate. Data were collected from the same lizard while fasting at 27°C, during the 3 hour roaming experiments. When activity did not remain constant for the full duration of the sample, the data point was not included in the regression; hence, N is lower for longer sample durations. Confidence intervals for each regression are indicated by the dashed lines 69 Figure 10: Pooled heart rate and mass-specific 0 2 consumption data from 3 hour roaming experiments, colour coded by A) treatment, and B) activity 75 Figure 11: Mean mass-specific 0 2 pulse for 24 hour confined experiments, as a function of A) temperature and activity, and B) digestive state and activity 78 Figure 12: Mean mass-specific 0 2 pulse for 3 hour roaming experiments, as a function of A) temperature and activity, and B) digestive state and activity. Axes are standardized to those of Figure 11 in order to facilitate comparison 78 Figure 13: Pooled breathing frequency and mass-specific 0 2 consumption data from 24 hour confined experiments, colour coded by A) treatment, and B) activity 83 Figure 14: Pooled breathing frequency and mass-specific 0 2 consumption data from 3 hour roaming experiments, colour coded by A) treatment, and B) activity 89 Figure 15: Regressions of two different lizards under the same conditions (37°C/digesting, 3 hour roaming experiments). The data for A was collected from "LB" and for B was collected from "Pi." Confidence intervals for each regression are indicated by a dashed line 96 ix LIST O F A B B R E V I A T I O N S ACR Air Convection Requirement Cao Oxygen content in systemic arterial blood C Oxygen content in systemic venous blood ECG Electrocardiogram fR Breathing frequency FE o Fractional content of oxygen in exhaled air (air leaving the mask) fH Heart rate F; Fractional content of oxygen in inspired air (air entering the mask) Q_ Cardiac output QJO Temperature quotient r2 Coefficient of determination (in regression analysis) RER Respiratory Exchange Ratio RMR Resting Metabolic Rate SDA Specific Dynamic Action sVco Mass-specific rate of carbon dioxide production sVQ Mass-specific rate of oxygen consumption V c o Rate of carbon dioxide production VE Minute ventilation V 0 Rate of oxygen consumption V s Stroke Volume V T Tidal Volume V The rate of oxygen delivery to the lung x G L O S S A R Y Air Convection Requirement. The amount of air that must be ventilated for each unit of oxygen consumed. The units for air convection are ml air/ml 0 2. Coefficient of determination. A statistical measure of the amount of variation in the data that is explained by a regression model. O z pulse. The amount of oxygen delivered to the tissues per heart beat. Q 2 pulse = Vs(Ca^ -CvJ Respiratory Exchange Ratio. The ratio of the rates of gas (CO z and 0 2) transfer across the lung. Respiratory Quotient. The ratio of the rate of carbon dioxide production to the rate of oxygen consumption by cellular metabolism. Resting Metabolic Rate. Minimum metabolic rate required to maintain basic physiological functions at a given temperature. Specific Dynamic Action. The increase in metabolic rate associated with digestion. X I A C K N O W L E D G M E N T S Many thanks to all members of the Milsom lab, both past and present, whom I have had the fortune to know. There are a few people in particular who I would like to thank for specific aspects of their support. First, thanks to Colin Sanders for his assistance with the experimental setup in the early phases of the project. I would also like to credit Kim Borg for her moral support and her assistance with animal care; her flesh sacrifice will not be forgotten. Catalina Reyes also gave up several pints of blood to the cause, and deserves special mention for all her advice and encouragement. Finally, many thanks go to Bill Milsom for guidance which, despite his heavy workload, was uncompromising in quality. This project was funded by NSERC and UBC UGF. X l l C H A P T E R 1: I n t r o d u c t i o n There are a number of methods commonly used to estimate metabolic rate in animals, however every method has its limitations. The ideal method to measure metabolism in wild animals would allow metabolic rate to be measured remotely, on an instantaneous basis, and throughout the duration of long term studies. With the advent of data loggers and associated wireless technology, it is now possible to measure some physiological variables (e.g. heart rate, body temperature) in this manner; however, there is no direct way to measure metabolic rate in a field setting. Generally oxygen consumption is used as a measure of metabolism, with the assumption that normally all metabolism is aerobic. However oxygen consumption is also difficult to measure directly without removing the animal from its natural setting. Consequently indirect methods are required. In this introduction, I will outline the major methods at our disposal to estimate metabolic rate in field studies, and elaborate on the theoretical and practical issues surrounding the Heart Rate Method. I will then introduce the tegu lizard (Tupinamhis merianae) and discuss the potential applications for this method of metabolic estimation in these animals. Finally, I will outline the hypothesis that my thesis will address. Methods available to estimate field metabolic rate: Three methods are already commonly used to estimate metabolic rate: i) the Time Energy Budget method, 2) the Doubly Labelled Water method, and 3) the Heart Rate Method. Each method has its advantages and disadvantages. 1 The Time Energy Budget (TEB) method involves monitoring the animals' behavioural patterns in the field. Metabolic rate is then estimated by multiplying the duration of each activity by its average energetic cost, as already determined in the laboratory by respirometry. This method has the advantage of being easy to use in the field and doesn't involve handling stress or human interference (Hawkins et al., 2000). However, it is extremely time consuming and computationally laborious. Also, an energy budget must already be available for the various age classes, behaviours (some of which are difficult to elicit in a laboratory setting), and when considering ectotherms, body temperatures of the species in question. This is not yet the case for many species, including tegu lizards. The Doubly Labelled Water (DLW) method involves using the half-life of isotopes in the body to estimate the rate of carbon dioxide production. More specifically, an animal is caught, a blood sample is taken, and oxygen and hydrogen isotopes are injected into the animal. The isotopes are allowed to equilibrate with the body fluids before a second blood sample is taken. The animal is then released, and later is recaptured to take a third blood sample. The levels of oxygen and hydrogen isotopes in each blood sample are assessed, and because the period between sampling times is known, the rate of isotope turnover can be calculated. This turnover rate is used to calculate the rate of carbon dioxide production, which in turn is used to calculate the rate of energy expenditure. This method has the advantage of being minimally invasive, and the isotopes are non-toxic (Hawkins et al., 2000). However, there are a number of disadvantages: First, it is expensive to carry out (particularly in larger animals). Second, it requires disturbing the animal to take blood samples, so metabolic rate may be influenced by handling stress (this may preclude accurate readings in dormant animals). Third, the duration of DLW studies depends on the half-life of the isotope, so studies are limited to a few hours or a few days depending on the animal. Fourth, this method is limited to estimating 2 the average metabolic rate in an animal over the duration of the experiment, so the metabolic cost of each individual activity cannot be determined (Boyd et al., 1999; McPhee et al., 2003). The Heart Rate Method involves using a known equation relating heart rate to oxygen consumption to predict metabolic rate from heart rate telemetry or logger data. This method has proven in many cases to be at least as accurate as the DLW and TEB methods (Bevan et al., 1994; Bevan et al., 1995a; Bevan et al., 1995b; Butler, 1993; Hawkins et al., 2000; Nolet et al., 1992), and surpasses the DLW method in terms of sensitivity (i.e. to individual activities), and applicability to long term studies (McPhee et al., 2003). However, it is the most invasive of these three methods, and requires more elaborate instrumentation during the experiments that must be performed in advance to derive the relationship and confirm its accuracy in the species of interest. To elaborate, the first step in developing the Heart Rate Method is to perform calibration experiments, which involve measuring heart rate and oxygen consumption simultaneously in one population and performing a regression analysis in order to derive the calibration equation. The next step is to perform validation experiments, which involve recording the same data in a different population and comparing known oxygen consumption values with oxygen consumption values predicted from the calibration equation. These validation experiments are imperative in order to verify the robustness of the calibration equation beyond the original population. If successful, metabolic rate can be estimated in any population, for any given type of activity, and over the long term from little more than heart rate. 3 Validation of the Heart Rate Method in different species As noted above, the accuracy of the Heart Rate Method has been verified in a number of species, including birds (Bevan et al., 1994; Butler et al., 2000; Green et al., 2001; Hawkins et al., 2000; Nolet et al., 1992) and mammals (Boyd et al., 1999; Butler et al., 1992; Butler, 1993; McPhee et al., 2003). Unfortunately, use of this method has been problematic in many fish (Lefrancois et al., 1998; Thorarensen et al., 1996) partly because the relationship between heart rate and metabolic rate is highly dependent on physiological and environmental conditions (see discussion on pages 6-8). This is probably also true in other ectotherms, such as reptiles. It may be for this reason that, to date, only one published study has focussed on reptiles; Butler et al. (2002) explored the reliability of this method in Galapagos marine iguanas and found that if temperature was controlled, the relationship was well defined with little variability (regressions at two different temperatures had r2 values of 0.86 and 0.91). Currently this trend is also being investigated in goannas (Clark et al., 2003) and other varanid lizards (Frappell, personal communication). Thus there is evidence that, with a few modifications, the Heart Rate Method can be used successfully in reptiles. Cautionary notes on the Heart Rate Method Theory It is important to review the theory behind the Heart Rate Method in order to understand its inherent assumptions. The theoretical basis of the Heart Rate Method lies with Fick's equation: V D J = / H X V S ( C „ 0 I - C „ J ( I ) 4 which states that the rate of oxygen consumption = heart rate x stroke volume x 0 2 content difference between systemic arterial and venous blood. In order for the relationship between heart rate and oxygen consumption to be linear, two scenarios are possible: i) The term V s (c ~C„D j (otherwise known as the 0 2 pulse) is constant (Butler, 1993; Butler et al., 2002; Thorarensen et al., 1996), such that both stroke volume and oxygen extraction efficiency are either constant, or vary inversely and in proportion to each other (McPhee et al., 2003). This scenario results in a linear relationship between heart rate and oxygen consumption in which the intercept passes through the origin (figure iA). 2) Alternatively, 0 2 pulse could vary with oxygen consumption in a predictable (i.e. curvilinear) manner (Butler, 1993; Butler et al., 2002). According to Butler (1993), this scenario results in a linear relationship with a positive intercept (figure iB) and is the more common of the two scenarios. Note that a positive intercept here is equivalent to a negative intercept when the axes are reversed, and recent calibration studies, where heart rate is plotted as the independent variable, have found that linear regression equations with negative intercepts best describe the relationship between heart rate and oxygen consumption in black-browed albatrosses (Bevan et al., 1994), macaroni penguins (Green et al., 2001) and Galapagos marine iguanas (Butler et al., 2002). The implication is that the relationship between heart rate and oxygen consumption can be linear in vertebrates under a variety of conditions, including steady state or varying 0 2 pulse. It is also possible that heart rate and oxygen consumption are related in a nonlinear fashion. In this case, 0 2 pulse certainly varies. Therefore, one cannot necessarily distinguish between a linear and a nonlinear relationship based on varying 0 2 pulse alone. In order to determine the best model to fit the data, many different regression models must be attempted. 5 Oxygen consumption (ml/min/kg) Oxygen consumption (ml/min/kg) — A — Heart Rate — -o— 0 2 pulse Figure 1: Change in 0 2 pulse with oxygen consumption when the regression of heart rate on oxygen consumption is linear. A) When 0 2 pulse is constant, the regression of heart rate on oxygen consumption passes through the origin. B) When 0 2 pulse is curvilinear, the regression has a positive intercept. Adapted from Butler (1993) Confounding factors The accuracy of predictions of metabolic rate will depend on a number of confounding factors that influence the relationship between heart rate and oxygen consumption in the species of interest. In all animals, the relationship may depend on physiological conditions (including age, sex, breeding state, digestive state, physical fitness, stress or activity level), and/or environmental conditions (including season and temperature) (Butler et al., 2002; Butler et al., 2000; Clark et al., 2003; Green et al., 2001; Hawkins et al., 2000; McPhee et al., 2003; Thorarensen et al., 1996). With this in mind, the Heart Rate Method must be verified for accuracy under a variety of different conditions before it can be usefully applied to field studies (Bevan et al., 1994; Green et al., 2001); In some species, the relationship is stable despite varying conditions (e.g. tufted ducks, Bevan & Butler, 1992), while in other species the relationship is highly dependent on conditions (e.g. marine iguanas, Butler et al., 2002; macaroni penguins, Green et al., 2001). 6 Considerations specific to reptiles Reptile physiology can complicate matters when calculating the relationship between heart rate and oxygen consumption. For example, the most obvious potentially confounding factor in these animals is arguably temperature, since some species undergo drastic changes in body temperature within hours under normal environmental conditions. If heart rate and the rate of oxygen consumption are not equally temperature-sensitive, then they will not change in proportion to one another with increasing temperature. However this does not need to be a problem. Butler et al. (2002) controlled for temperature in calibration and validation studies with marine iguanas and observed a remarkably tight regression at each of two temperatures (representing the daily minimum and maximum body temperatures). At the higher temperature, the regression was right-shifted (i.e. the regressions had similar slopes but different intercepts). With these two regressions, the authors were able to develop a common calibration equation incorporating the effect of temperature as a Q ,^ effect. The significance of this study is the indication that differences in temperature sensitivity of heart rate and oxygen consumption in reptiles can be accommodated by modifications to the Heart Rate Method. Not only can the relationship between heart rate and oxygen consumption be affected by temperature, but it can also be affected by digestion. Digestion can be extremely energetically costly in reptiles (Andersen & Wang, 2003; Andrade et al., 1997; Bennett & Hicks, 2001). In many species the post-prandial increase in oxygen consumption (i.e. specific dynamic action, or SDA) is greater than the increase in oxygen consumption during exercise (Bennett & Hicks, 2001). Although the effect of digestion has not been considered to date in any published study on the Heart Rate Method in reptiles, it was briefly investigated in a single Stellar sea lion (McPhee et al., 2003). In this study, heart rate clearly co-varied with oxygen consumption during fasting, but the relationship broke down during digestion. On 7 the other hand, preliminary studies performed on tegu lizards suggested that the relationship remains intact in this species during digestion, but is right-shifted (same slope, different intercept) (Piercy, unpublished data). If this proves to be true, the Heart Rate Method could still be used in digesting lizards as long as digestive state was known and taken into account. Finally, the uniquely reptilian trait of intra-cardiac shunting must be considered. When shunting is significant, oxygenated and deoxygenated blood mix since blood is either recirculated to the lungs (L-R shunt) or to the systemic circulation (R-L shunt) (Burggren, 1987; Hicks, 2002; Hicks & Wang, 1996; Wang & Hicks, 1996; Wang et al., 1997). The result is a mismatch between ventilation and perfusion. Not all reptiles have the same degree of shunting; the magnitude and direction of shunting depends on physiological controls such as changing peripheral resistance, as well as anatomical features such as the ventricular muscular ridge. Reptiles with a well-developed ventricular muscular ridge have good pressure separation between ventricular compartments and consequently are capable of a low degree of shunting (Burggren, 1987; Wang et al., 2003; Wang et al., 1997). Since ventilation and perfusion are relatively well matched in this situation, heart rate and oxygen consumption should also be closely correlated. Varanid lizards are the classic example of reptiles with a relatively well-developed ventricular muscular ridge, and it is often suggested that this may be key to their relatively active lifestyle and high metabolic rate. Recently pythons have also been shown to have good pressure separation, possibly supporting their relatively high metabolic rates during digestion or shivering thermogenesis during egg incubation (Wang et al., 2003). Tegu lizards are also relatively active reptiles and have been shown to have a well-developed muscular ridge (Wang et al., 2001), so one could hypothesize that shunting is limited in these reptiles. Indeed one study has indicated that shunting is slight in tegu lizards (Johansen et al., 1987). Thus, it would appear to be reasonable to use heart 8 rate as a proxy for oxygen consumption in these lizards since ventilation-perfusion matching may be almost complete. However, results must be analyzed with this assumption in mind. Limitations of the calibration equation Once calibration equations have been validated, there are still two major limitations of the Heart Rate Method to keep in mind. First, because regression equations between individuals (and sometimes within individuals) are often different (Hawkins et al., 2000; Nolet et al., 1992), a calibration equation represents the average relationship in a variable population. Accordingly, the Heart Rate Method is not reliable if used to predict oxygen consumption in an individual (unless a calibration equation is derived and validated for that specific individual) (Bevan et al., 1992; Green et al., 2001). Consequently prediction accuracy is severely reduced when calibrating, validating, or even employing the Heart Rate Method with smaller sample sizes. Second, it is important to remember the statistical limitations of regression analysis: a regression cannot be used to predict variables that lie outside the limits of the regression line itself. In other words, a regression cannot be extrapolated (Green et al., 2001). To put this in perspective, if the calibration and validation experiments are performed only with resting animals, the resulting regression will be limited in its predictive capacity; it can be used to predict resting metabolic rate, but not metabolic rates approaching maximum. For this reason it is imperative to perform calibration and validation experiments that include the most diverse range of oxygen consumption values possible, so that, once validated, the practical application of the method is not limited. 9 Tupinamhis merianae A note on taxonomy Tupinamhis merianae, otherwise known as the Argentine Black and White tegu, has had a complicated taxonomic history (Fitzgerald et al., 1999). Up until 1995, T. merianae was known as T. teguixin, whereas the species currently known as T. teguixin (the Columbian tegu) was known as T. nigropunctatus. The implication of this is that older literature referring to T. teguixin is actually referring to T. merianae. Natural history T. merianae is the largest of the tegu species (Fitzgerald et al., 1999; Milstead, 1961), is native to South America and is commonly found in Brazil, Paraguay, Uruguay, and Argentina. These lizards inhabit a diverse range of habitats (Milstead, 1961; van Sluys & Rocha, 1999), including forests and savannah in both temperate and tropical regions, but excluding deserts; they usually stay close to water. T. merianae is one of three species in the Tupinamhis genus which undergo dormancy. T. merianae becomes inactive during the dry season when food is scarce (Abe, 1995; de Souza et al., 2004; van Sluys & Rocha, 1999), usually from May to September. During this time, the lizards remain underground in their burrows, at approximately i7°C (based on measurements taken from artificial burrows at Jacarezario, U N E S P , Rio Claro, in Sao Paulo state, southeastern Brazil) and survive on endogenous fuels. Shortly after emerging, lizards begin eating again to replenish endogenous stores. Mating begins and lasts throughout September, and a month later females wi l l lay up to 65 eggs. The eggs incubate for 60-80 days and hatch in December and January. The hatchlings are only 22 cm long, but wi l l double in size 10 over their first year as they prepare for dormancy. Lizards become mature within 3 years, and can grow up to 4 feet long and 4 to 8 kg in weight. Mature males tend to develop jowls and have slight prominences on either side of the cloacal orifice. T. merianae is an opportunistic eater (Milstead, 1961); lizards will eat vegetables, fruit, insects, rodents, and other lizards. They are sit-and-wait predators, so although they are capable of short bursts of intense activity, usually it is only as a fight-or-flight measure. Otherwise, they move at moderate speeds. They are prey to large cats, hawks, dogs and humans (Palacios et al., 1997). Current questions involving metabolism in tegu lizards Metabolic depression Although much is known about the molecular mechanisms that are involved in metabolic depression (for reviews, see Boutilier, 2001; Storey & Storey, 1990; Storey & Storey, 2004), assigning cause and effect is not a simple matter. Most research on metabolic depression has focussed on animals that enter into metabolic depression as temperature falls (e.g. hibernation, hypothermia), but the role of temperature in metabolic depression is not necessarily obvious. One may ask: is falling body temperature the cause or the result of metabolic depression? Assigning cause and effect may be simpler in tegu lizards. These lizards undergo seasonal periods of dormancy; however, they do so without experiencing much change in ambient (and therefore body) temperature. In fact, lizards can enter dormancy even if they are housed in a lab at constant temperature (e.g. Abe, 1983; Andrade & Abe, 1999a). With this in mind, tegu lizards may provide a unique opportunity to study the changes in cardio-respiratory function associated with 11 metabolic depression in an animal in which body temperature can be kept constant, thereby eliminating one complicating factor. Before we can study metabolic depression in tegu lizards, we must be able to differentiate between sleep and the dormant state using a physiological indicator. The obvious candidate is oxygen consumption. However, to verify this, one must be able to monitor oxygen consumption in lizards without disrupting them, and preferably under field conditions where lizards are more likely to go dormant. For this reason, the Heart Rate Method would be a useful tool in such studies. Energy budget Currently little is known about the energetic costs facing a tegu lizard in the wild. Studies have been performed on dormant, resting and exercising animals in the laboratory (Abe, 1983; Andrade & Abe, 1999b; de Souza et al., 2004), but there is much more to learn: for example, the metabolic costs for varying types of activities performed by wild lizards (e.g. running, digging, fighting, eating, reproducing), and the costs of these activities at different temperatures. Indeed, because physical fitness and stress are known to affect a host of physiological variables, there is no reason to assume that metabolic rates measured in a laboratory setting will be equivalent to those in wild lizards. A detailed budget is an essential tool for any physiologist or ecologist, and is vital for conservation policy. The potential of the Heart Rate Method in tegu lizards Clearly there is a gap in our understanding of metabolic costs and their nutritional, behavioural and ecological implications in tegu lizards. To fill this gap, we need to determine metabolic rates over shorter periods of natural activities in wild animals. Traditional methods of estimating metabolic rate (i.e. DLW and the Time-Energy 12 Budget) do not provide fine time resolution, long-term potential, and unobtrusive data collection. It is for these reasons that I decided to attempt to calibrate the Heart Rate Method in this species. In 2003, preliminary studies were carried out to determine if the Heart Rate Method would reliably estimate metabolic rate in tegu lizards (Piercy, unpublished data). Temperature, activity level and digestive state were considered as potential confounding factors. Six lizards were used in calibration experiments at I5°C (fasting), 22-5°C (fasting) and 3o°C (both fasting and after feeding). The lizards were implanted with subcutaneous heart rate electrodes and placed in a closed-circuit respirometer. A camera was mounted to observe the lizards, and activity was categorized based on postural indicators. Data for each treatment were pooled for regression analysis. Equations for different temperatures had similar slopes but different intercepts; similarly equations before and after feeding at 3o°C had similar slopes but different intercepts. Regressions accounted for 61-66% of the variation (n=6), except at 30°C after feeding when only 43% of the variation was explained by the regression (n=4). These experiments suggested that oxygen consumption co-varies with heart rate in a relatively predictable manner at any given temperature. Nevertheless, the methodology needed to be refined in order to eliminate error associated with the closed-circuit respirometry. Thus I employed a more sophisticated open-circuit design and more advanced analysis to confirm the trends observed in these preliminary studies. 13 Hypotheses In this study, various respiratory and circulatory variables were measured or calculated in lizards at four different levels of activity, at three different temperatures, and using two different methodologies. The objective of Chapter 2 is to describe the mean values of these variables under each condition, in the process developing a more detailed energy budget than had been previously collected. I also determined whether lizards in different states (particularly sleep vs. awake but quiet) could be differentiated based on these physiological variables. Chapter 3 addresses the primary goal of this thesis, which was to test the potential of the Heart Rate Method to estimate metabolic rate in tegu lizards. To do this, the relationship between oxygen consumption and heart rate in one group of lizards was calculated in two sets of "calibration" experiments. I hypothesized that heart rate is indeed an accurate indicator of metabolic rate in tegu lizards for all behaviours (i.e. activity states), provided that temperature and digestive state are taken into account. 14 C H A P T E R 2: C h a n g e s i n the r a t e o f o x y g e n c o n s u m p t i o n a n d a s s o c i a t e d c i r c u l a t o r y a n d r e s p i r a t o r y v a r i a b l e s w i t h c h a n g e s i n t e m p e r a t u r e s a n d a c t i v i t y i n t e g u l i z a r d s I N T R O D U C T I O N In reptiles, it is often difficult to determine the difference between animals that are dormant, asleep, or awake but quiet (eyes closed) (Lillywhite et al., 1999). This problem is of particular relevance in facultative hibernators at certain times of the year since the difference between dormancy and sleep is not obvious without disturbing the animal. Even alert but motionless animals occasionally may be mistaken for resting animals. When these states are misidentified, measurements such as resting metabolic rate (RMR) may be under- or over-estimated. Although these different states may be difficult to distinguish on the surface, one would expect metabolism to be higher in animals when awake than when sleeping, and to be suppressed below sleeping levels during dormancy. If this is true, one could use metabolic rate to distinguish between different activity states. Tegu lizards, which are facultative hibernators, have been the focus of several metabolic studies, but to date none have explored in detail the use of metabolic rate to define activity state. In order to do so, metabolic rate must be determined for each activity state over a variety of temperatures representing the normal range that would be experienced in the wild. On first inspection, it may seem as though there is already an adequate database to address this. Many publications have focused on the effects of temperature on 15 metabolic rate in tegu lizards (e.g. Abe, 1995; Andrade & Abe, 1999a; Andrade & Abe, 1999b; de Souza et al., 2004; Toledo et al., 2005, Submitted) and other reptiles, while others have quantified resting metabolic rate and maximal metabolic rate (e.g. Klein et al., 2003). However, few have investigated the interactive effects of temperature and activity level on metabolic rate in a reptile, and those that have done so are usually limited to two temperatures and two activity levels (usually resting and maximal). Additionally, protocols are rarely (if ever) consistent between studies, and yet none have explored the influence of methodology on the results. These are some of the issues that will be considered in this chapter. These considerations also can be applied to the circulatory and respiratory variables that are closely associated with metabolic rate as per Fick's equation: V o a = / H X 02pulse = fR x V T (FIoi - F £ O J ) (2) where V 0 is the rate of oxygen consumption, fu is heart rate, O z pulse is the amount of oxygen taken up by the tissues per heart beat, f% is the breathing frequency, V j is the tidal volume, and \FJ — F £ q j is the amount of oxygen extracted from the lung per breath. It is to be expected that changes in oxygen consumption must be accommodated by adjustments in some of these variables. However it is not obvious which variables play the greatest role in matching oxygen supply to demand in tegu lizards. It is also not obvious whether temperature-induced increases in oxygen consumption are associated with the same qualitative adjustments as activity-induced increases in oxygen consumption. If these questions can be resolved, it might be possible to define activity state based on one or more of these physiological variables along with (or instead of) metabolic rate. 16 This chapter addresses some of these gaps in the literature by focusing on three interrelated objectives. The first objective was to quantify the effects of temperature and activity on metabolic rate and associated circulatory and respiratory variables in adult tegu lizards in greater resolution than has been reported previously. The second objective was to identify the relative contributions of each of the associated circulatory and respiratory variables to oxygen consumption under different circumstances. Finally, the third objective was to explore the possibility that methodology significantly influences results. M E T H O D S Two experimental protocols were used; the first set of experiments, lasting 24 hours each, began after two weeks of trial runs, and ran from May 18, 2004 to August 25, 2004. The second set of experiments, lasting 3 hours each, was carried out from November 18, 2004 until December 2, 2004. All experiments were performed using three male and three female tegu lizards, each approximately 3-4 years of age with weights ranging from 2.2 to 3.9 kg (mean = 3.08 ±0.24 kg). All lizards were bred in Brazil and were brought back to UBC within a few weeks of hatching. As such all lizards have been in captivity all their lives. Note that two lizards (one male and one female) were each missing the front right limb. Treatments were applied using a repeated-measures design. 17 Z 4 hour confined calibration experiments Experimental setup These experiments involved open-circuit respirometry to collect respiration data. As such, custom-fitted masks for each lizard were made as described by Glass et al. (1978), with a few modifications as per Wang and Warburton (1995). Lizards were anaesthetized and casts of their heads were made using Jeltrate® (Dentsply International) dental impression material. Plaster of Paris was then poured into these casts and allowed to solidify. The masks were fitted by warming and stretching mouthguard plastic over these plaster molds. The mouthguard plastic was then cut to cover the snout and the top of the head, such that the mask would not interfere with the eyes or the mouth. Nostril holes were drilled, and rubber tubing was glued in place to direct air to and from the nostrils through a pneumotachograph mounted on top of the mask (see Figure 2). The pneumotachograph was hand-made from the barrel of a 3cc syringe packed with polyethylene tubing. Three 16G1V2 needles were used as outflow tubes from the pneumotachograph: one leading through a drying column containing Drierite (Hammond, U S A ) to the oxygen and carbon dioxide analyzers (Models OM-11 and LB-2, Beckman Instruments, U S A ) and two leading to the differential pressure transducer (Model DP103-18, Validyne Engineering Corp, U S A ) which was in turn connected to an amplifier (Model 7P122E, Grass Instruments, U S A ) for tidal volume analysis. The other end of the pneumotachograph was left open to ambient air. Masks were fixed to the lizards using Impregum F™ (3M, U S A ) , an inert, rapidly polymerizing dental impression material that is removable without damaging the underlying scales. Because lizards were often capable of removing the masks, Nexcare™ Athletic Wrap (3M, U S A ) was wrapped around the snout in front of the 18 eye to further secure the mask. Each lizard was unable to open its mouth with this bandage in place; thus it gave the additional benefit of ensuring that all exhaled air was directed out through the nostrils instead of the mouth. With training, lizards learned to tolerate the mask for over 2 4 hours. Figure 2: Expe r imen ta l se t -up . T h r e e heart rate e lec t rodes are rep resen ted by a s ing le ye l l ow d isk . T h e m a s k is rep resen ted by a d a s h e d l ine. Exha la t i on d y n a m i c s a re ind icated in red ; Inhalation d y n a m i c s are ind ica ted in b lue . (dP) dif ferent ial p ressu re t ransducer ; (dc) dry ing c o l u m n ; (fH) heart rate; ( V E ) vent i lat ion; { V Q 2 ) o x y g e n c o n s u m p t i o n ; ( V C C , 2 ) c a r b o n d iox ide product ion. (Adap ted f rom W a n g & Warbu r ton , 1995) The environmental chamber was temperature controlled, and contained a Plexiglas box ( 4 6 c m wide by 4 5 . 5 c m long by 3 2 c m high), large enough to allow movement in even the largest lizard. The lid of the box was designed with a long slit cut down the middle to allow heart rate leads and tubing from the pneumotachograph to exit 19 the box without becoming entangled as the animal moved around. This slit also allowed the box to be open to ambient air, while preventing lizards from escaping. The photoperiod in the chamber was set at I2h:i2h (on at 8:35am, off at 8:35pm), and two infrared cameras were mounted in order to record images of lizard activity and posture to a VCR during the experiment. Training and Controls To begin training, a fasting (minimum 7 days) lizard was placed in the Plexiglas box inside the environmental chamber for approximately 24 hours. The lizard was then fitted with a mask and placed back in the Plexiglas box inside the environmental chamber for another 24 hours; if the lizard shook off the mask then the mask would be replaced and the training would resume. Once the lizard was willing to wear the mask for a full 24 hours, the lizard was considered ready to start trials. The minimum training time required was 2 days. Once training was complete, the lizard was placed back in the environmental chamber at 27°C to acclimate for approximately 24 hours. After this time, the analyzers and the pneumotachograph were calibrated (see details under Experimental Protocol), and the lizard was taken out, fitted with the mask and returned to the chamber. Oxygen consumption, tidal volume and breathing frequency were recorded for 24 hours to serve as a control, to be compared with post-surgery values. Since heart rate could not be recorded prior to surgical implantation of the electrodes, heart rate was assumed to be affected by surgery in the same manner as oxygen consumption. 20 Surgical Procedure Lizards were already fasted prior to surgery. Each lizard was anaesthetized with halothane, and three electrodes (gold plated, Grass Instruments, U S A ) were implanted subcutaneously in the thorax/abdomen, triangulating the heart in order to monitor heart rate. The lizards were allowed to recover and acclimate for one day in the environmental chamber at the first temperature treatment. Experimental Protocol Experiments were carried out from May until August, 2004, and involved four treatments: i7°C/fasting, 27°C/fasting, 37°C/fasting, and 37°C/digesting. Each trial lasted 24 hours, and whenever applicable, was followed by a 24 hour acclimation period to the next treatment temperature (i.e. no acclimation was necessary when the 37°C/fasting treatment was followed by the 37°C/digesting treatment). For the sake of efficiency, the final trial was always the digesting trial. Thus, the protocol was as follows: Day 1: control trial at 27°C Day 2: surgery, recovery and acclimation to first treatment temperature Day 3: 1st trial Day 4: acclimation to next treatment temperature Day 5: 2nd trial Day 6: acclimation to next treatment temperature (usually 37°C) Day 7: 3rd trial (usually 37°C/fasting) Day 8: 4th trial (37°C/digesting) Day 9: surgery to remove heart rate electrodes 21 This protocol was a guideline only, and was amended as necessary as technical issues arose. All equipment was carefully calibrated before each trial. The pneumotachograph on the mask was connected to the gas analyzers and the differential pressure transducer. The pneumotachograph itself was calibrated by injecting and removing known volumes of air through the nostril holes of the mask. The pressure difference across the pneumotachograph, and hence flow rate, was detected by the pressure transducer and was recorded on the computer as voltage, and a linear regression of volume on integrated voltage (r2>o.o8 in all cases) was used to convert differential pressure traces to air flow rate (which in turn was integrated to calculate tidal volume) throughout the course of the experiment. This regression was also used to calculate the pump flow rate from recordings of the voltage change that occurred when the pump was connected and disconnected downstream of the gas analyzers (nostril holes were blocked for this, so all airflow was through the pneumotachograph grid, see Figure 2). The pump flow rate varied over time, so it was recalculated before and after each experiment. In all cases, the flow rate was sufficiently high to assume that all exhaled air was drawn into the tubing leading to the gas analyzer (ranging from 3ooml/min to 6ooml/min). Finally, the gas analyzers were calibrated using i) iooo/0 N 2 and 2) a mixture of 21% 0 2, 5% C 0 2 and 7 4 0 / 0 N 2 ; these gases were fed through the nostril holes of the mask which in turn was connected to the drying column and the gas analyzers. Once calibrations were satisfactory, the lizard's weight was recorded, and the mask was fixed to the lizard. Impregum F™ was used to create an airtight seal around the nostrils, and the lizard was placed in the Plexiglas box in the environmental chamber. The outflow tubes from the pneumotachograph were connected to the oxygen and carbon dioxide analyzers (via the drying column) and to the differential 22 pressure transducer, and the heart rate leads were connected to the amplifier (Model 7P511K, Grass Instruments, USA) (Figure 2). The lid was sealed and the chamber was shut. As mentioned earlier, the pump was connected downstream of the gas analyzers. There was a lag time of approximately 3 seconds from the time that the lizard exhaled to the time that changes in 0 2 and C O z were detected at the gas analyzers. The trial began once the data acquisition device began recording, and the lizard was not disturbed for 24 hours unless it was necessary to adjust equipment. The lizard's posture and activities were recorded with a VCR, and were associated with the corresponding physiological data during analysis. Activity level was scored based on the following guidelines: Sleep: curled up and/or legs stretched back (e.g. Figure 3A) Quiet: eyes occasionally open (if visible), body lying flat on floor (e.g. Figure 3B) Alert: eyes open (if visible), or body held up off floor, or resting between periods of activity (e.g. Figure 3C) Moving: walking, digging, etc. (e.g. Figure 3D) Note: data collected when the animal was obviously exercising to exhaustion were excluded from analysis to avoid inclusion of anaerobic activity 23 After each trial, the mask was removed to prevent damage, and the lizard was offered water and kept in the environmental chamber while acclimatizing to the next treatment temperature. Prior to the 37°C/digesting trial no acclimation was necessary since the third fasting trial was usually conducted at 37°C, but because lizards would often refuse to eat immediately after completing a trial, they were usually left in the environmental chamber overnight to rest. The next day the lizard would be offered a standard meal of ground beef, vegetables, fruit and vitamin and mineral supplements and was allowed to feed to satiation; lizards never ate more than 4% of the body weight despite being fasted for several weeks. Once the lizard showed no further interest in food, and once the equipment was calibrated, the lizard was once again fitted with the mask. Data collection typically began approximately 1 hour from the start of feeding. 3 hour roaming calibration experiments In these experiments, lizards were allowed to roam around in an open space (approximately 4 feet wide by 10 feet long) in order to record more natural levels of activity, with less stress. Tegu lizards have been encouraged to exercise (albeit reluctantly) in previous studies (Klein et al., 2003), but because they are sit-and-wait predators, they tend to run only in short sprints. With this in mind, it was more ecologically relevant to collect data on animals moving at a moderate pace. Thus the experiments were repeated in a closed, temperature controlled room where the animals could wander at their own pace. Note that all data are assumed to be aerobic under these conditions. These experiments were performed between mid-November to early December 2004. 24 Experimental setup Masks were made as before with the exception that outflow tubes were longer in order to accommodate the larger roaming area. Equipment was also set up as before (see Figure 2); however, instead of carrying out the trials in a Plexiglas box inside a small environmental chamber, trials were conducted in a temperature-controlled room with a large area of the floor fenced off. Another section of the floor was fenced off as an acclimation section. With this setup, trials could be conducted on one lizard as all other lizards were being acclimated simultaneously. The fasting lizards were briefly trained as before, but because the trials in this series of experiments only lasted a few hours, it was not necessary to train them to tolerate the mask for a full 24 hours. Surgical Procedure Heart rate electrodes were implanted as before, and the animals were allowed to recover for a minimum of 24 hours in the temperature-controlled room. Experimental Protocol The lizards were exposed to the same treatments as in the 24 hour confined experiments: i7°C/fasting, 27°C/fasting, 37°C/fasting, and 37°C/digesting. The lag time for gas flow between the nostrils and the gas analyzers was longer in these experiments, at approximately 4-6 seconds, due to the longer lengths of tubing required to accommodate the free-ranging area. Since all lizards could be instrumented and acclimated at the same time, experiments were carried out one after another without break. Additionally, since circadian rhythms were taken into 25 account in the first series of experiments, it was not necessary to conduct these trials for the full 24 hours. Thus three trials were completed per day, (i.e. three different animals at the same temperature, each trial lasting approximately 3 hours). For the 37°C/digesting trial, the lizards were fed approximately 36 to 40 hours prior to recording in order to detect the effect of digestion within the duration of the experiment as tegu lizards reach their peak specific dynamic action during this time period (Klein et al., 2005, In review) Data analysis Four channels of data were collected (see Figures 2 and 4): 0 2 (in percent), C 0 2 (in percent) electrocardiogram (EKG) and differential pressure. Thus it was possible to calculate the rate of oxygen consumption (V 0 ) (Standard temperature and pressure, dry, or STPD), the rate of carbon dioxide production ( V c o ) (STPD), heart rate ( / H ) > breathing frequency (/R) and tidal volume (VT). It should be noted that the breathing trace during movement was often noisy, so breathing frequencies and tidal volumes were difficult to asses and values may not be as accurate as "sleep", "quiet" and "alert" values. Nevertheless, the qualitative changes in these variables during movement should be real. Oxygen consumption was obtained by integrating the area under the curve and was corrected for a H 2 0 absorbent/no C 0 2 absorbent setup according to Withers (1977) equation 3b: v o, = 5 1 (3) i - F , 26 Where V „ is the rate of airflow out of the mask at STPD, F, is the fractional pump 7 lQ content of 0 2 entering the mask, and F E q is the fractional content of 0 2 leaving the mask. The measured variables were used to derive several related variables, including ventilation (VE, the product of fR and V T) , 0 2 pulse (VQ divided by fH , as per Fick's equation), the rate of oxygen delivery to the lung (V , 21% of VE), percentage oxygen extraction from the lung (%02 Extraction, or the ratio of V 0 to V t o t o ), Air Convection Requirement (ACR, the ratio of VE to V 0 ), and Respiratory Exchange Ratio (RER, the ratio of V c o to V 0 ). Finally, Q ,^ values were calculated for the rates of oxygen consumption and carbon dioxide production, heart rate and breathing frequency: 0,= V * T . J T-T, (4) where T, and T2 were the lower and higher temperatures respectively, and xT and xTx were the rates at these temperatures. Data points for each physiological variable were calculated from the averages of 1 to 2 minute samples during the 24 hour confined experiments (this sample duration was chosen based on a similar protocol followed by Butler et al. (2002), in which samples were taken from 30-60 second blocks of data). The first data sample was taken at least one hour into the experiment, and subsequent samples were generally taken 20 minutes apart. However, the sampling protocol was adjusted for the 3 hour 27 roaming experiments; at 27°C/fasting, each data point was reanalyzed ten different ways; first taking average values from i minute samples, 2 minute samples, 3 minute samples and so on, to 10 minute samples. This was carried out in order to determine if the sample duration affected the quality of the regression analysis in Chapter 3. The rest of the 3 hour roaming experiments (at i7°C/fasting, 37°C/fasting and 37°C/digesting) were analyzed using 5 minute samples of data approximately every 10 minutes. The first data sample for these experiments was always taken at least 30 minutes after commencing recording. Lizards were highly variable in mass, ranging from 2.2 kg to 3.9 kg, and such variation undoubtedly influenced oxygen consumption. Similar studies have attempted to take such variation into account by converting oxygen consumption to mass-specific terms (e.g. Butler et al., 2002). However, there has been much controversy in recent years regarding the use of mass-specific measures of physiological variables. Many authorities believe that expressing these variables in terms of mass does not remove the effect of mass (McNab, 1999; McPhee et al., 2003; Packard & Boardman, 1999) since mass rarely scales with these variables in a 1:1 ratio. In order to characterize the relationship between mass and resting oxygen consumption, both linear and nonlinear regression analyses were performed on these data, and the best regression model dictated the standardization of VQ (as per Green et al. (2001)). Mean values were calculated for each treatment and activity for the 24 hour confined and the 3 hour roaming experiments. For each protocol, the means for each variable in Fick's equation as presented in equation 2 were then compared via a z~ way repeated measures A N O V A with treatment and activity as the two effects, followed by post-hoc Holm-Sidak tests to identify specific differences. The effect of temperature was tested using averages from i7°C/fasting, 27°C/fasting, and 28 37°C/fasting experiments, while the effect of digestion was tested on averages from 37°C/fasting and 37°C/digesting experiments. Differences between the two protocols were assessed via a paired t-test. Sleep data were excluded from statistical analysis of 3 hour roaming data due to low sample size. All statistics were performed using Microsoft Excel 2003, SigmaStat 3.1 and JMPIN 4.0.4. Results were considered significant if P<o.o5. 29 RESULTS Figure 4 shows traces collected at 27°C for a single tegu lizard at each of the four activity levels as previously defined (see Figure 3). There was some individual variation in the regularity of the breathing trace, particularly for sleep (some lizards displayed episodic patterns while sleeping); otherwise these traces are representative. Sleep Quiet 0 20 40 60 80 100 120 0 20 40 60 80 100 120 T i m e ( s e c o n d s ) T i m e ( s e c o n d s ) Figure 4 : T y p i c a l t races of c h a n g e s in FJQ^ (%02) , FLCC^ ( % C 0 2 ) , E K G a n d respi ratory a i r f l o w (i.e. breath ing t race) for e a c h activi ty leve l . E a c h t race is taken f rom the s a m e an ima l dur ing the 27°C/ fas t ing trial (24 hour con f ined exper imen ts ) , a n d rep resen ts 2 m inu tes of record ing . 30 Effects of mass In order to correct for mass, the relationship between mass and resting oxygen consumption was explored for each treatment using linear and nonlinear regressions; oxygen consumption generally increased with mass, and linear regressions tended to fit the data as well (r2 ranging from 0.04 to 0.62 for 24 hour data) as any of the nonlinear regressions, including logarithmic regressions (r2 ranging from 0.05 to 0.60 for 24 hour data). Therefore it was assumed that the relationship was linear, and oxygen consumption was simply divided by mass. Controls Paired t-tests on mean sVQ and sV c o showed no significant difference between tegus pre- and post-surgery at any activity level. Similarly, Paired t-tests on transformed data showed that breathing frequency and tidal volume were also unaffected. Thus, it was assumed that heart rate was similarly unaltered 24 hours post-surgery. Effects of temperature and state on gas exchange 24 hour confined experiments Mean values and standard errors of all measured and derived variables are compiled in Table 20 in the Appendix. Fick's equation variables are plotted in Figure 5, illustrating how mean values change with temperature and activity during the 24 hour confined experiments. 31 Sleep Quiet Alert Moving Sleep Quiet Alert Moving Sleep Quiet Alert Moving Sleep Quiet Alert Moving Activity Activity Activity Activity sV02 = fR x VT x % 0 2 extracted Figure 5: Mean values of each variable in Fick's equation for 24 hour confined data. Values for activities that were statistically different from those measured in the quiet state are indicated by a * symbol, and values for temperatures that were statistically different from those measured at 2 7 ° C are indicated by a • symbol. Black symbols indicate differences that are independent of temperature (*) or that are independent of activity (•>), while coloured symbols indicate incidents that are dependent on both factors (i.e. statistical interactions between activity and temperature). 32 The rate of oxygen consumption in sleeping animals was not significantly different with respect to quiet animals. However, oxygen consumption increased significantly when the animals were alert or moving. Temperature had no significant affect on mean oxygen consumption values. Heart rate followed a more complex pattern. As with oxygen consumption, "sleep" heart rates and "quiet" heart rates were similar, whereas "alert" and "moving" heart rates were higher. However, the magnitude of the increase for each of these states was dependant on temperature. At i7°C, the increase in heart rate was not significantly different between the alert and moving states, whereas at 27°C and 37°C, heart rate was greater still during movement than when the animals were alert. Accordingly, temperature had a significant effect on heart rate, where values were generally lower at i7°C. This trend was less pronounced at lower activity levels (i.e. "sleep" and "quiet"), when heart rates at i7°C were significantly lower than those at 37°C, but neither of these were significantly different from 27°C. During the "alert" and "moving" states however, the temperature effect was more pronounced, with mean values at i7°C significantly lower than both 27°C and 37°C. Mean values of 0 2 pulse (as will be further discussed in Chapter 3) were not significantly different in sleeping, quiet and alert animals. However, 0 2 pulse was significantly lower in moving animals at all temperatures. There was a temperature effect, such that 0 2 pulse was significantly higher at i7°C than at 27°C or 37°C. However, there was no detectable difference in mean values between 27°C and 37°C. Breathing frequency was significantly different at each activity level, with values increasing from sleep to quiet to alert to moving. On the other hand, there was no difference in breathing frequency between the three temperature treatments. 33 Unlike the other variables, mean tidal volume did not change significantly with increasing activity or temperature. Nevertheless, it is apparent from the error bars that tidal volume was quite variable over short periods, with error bars encompassing the range of 14 ml to 23 ml (representing an increase of 6 4 % ) . Oxygen extraction generally decreased with increasing activity. The only exception was that mean "alert" values were not significantly different from mean "quiet" values or from mean "moving" values, whereas quiet and moving values were significantly different from each other. Regardless, oxygen extraction was highest during sleep, and lowest during movement. There was no significant change with increasing temperature. Table 1 quantifies the effect of temperature on the rate of oxygen consumption, the rate of carbon dioxide production, heart rate, and breathing frequency for 24 hour confined experiments. values generally ranged between 1 and 2, since temperature had little effect on most variables. As is apparent from Figure 5, the largest increase with temperature was seen in heart rate data, where Q m values ranged from 1.6 to 2.3 for the increase from i7°C to 27°C, and from 1.1 to 1.7 for the transition from 27°C to 37°C. Overall, the for heart rate over the full temperature range (i7°C to 37°C) was calculated to be 1.7 during sleep, to 1.4 during movement. 34 Table 1: Q 1 0 v a l u e s of direct ly m e a s u r e d va r i ab les for 24 hour con f i ned da ta . V a l u e s in bo ld c o r r e s p o n d to s igni f icant c h a n g e s in the var iab le for the g iven act iv i ty a n d tempera ture t ransi t ion. Variable Activity Q10 17°C -27 °C 27°C - 37°C 17°C -37 °C Mass-specific Sleep 0 2 consumption Quiet Alert Moving 1.53 ± 0 . 1 6 1.31 ± 0 . 1 3 1.41 ± 0 . 3 3 1.77 ± 0 . 3 0 1.27 ± 0 . 1 3 1.20 ± 0 . 1 6 0.97 ± 0.14 0.91 ± 0.33 1.36 ± 0 . 0 6 1.21 ± 0 . 0 6 1.10 ± 0 . 1 1 1.11 ± 0 . 1 3 Mass-specific Sleep C O z production Quiet Alert Moving 1.47 ± 0 . 1 1 1.34 ± 0.15 1.25 ± 0.23 1.49 +0.23 1.31 ± 0 . 1 1 1.18 ± 0 . 1 4 0.94 ± 0.09 0.89 ± 0.22 1.38 ± 0 . 0 7 1.22 ± 0 . 0 4 1.06 ± 0 . 1 1 1.09 ± 0 . 0 7 Heart Rate Sleep Quiet Alert Moving 1.71 ± 0 . 1 7 1.58 ± 0 . 0 9 1.91 ± 0.36 2.26 ± 0.47 1.68 ± 0 . 1 9 1.51 ± 0 . 2 5 1.13 ± 0 . 1 3 1.06 ± 0.32 1.68 ± 0.15 1.50 ± 0.10 1.39 ± 0.09 1.41 ± 0.09 Breathing Sleep Frequency Quiet Alert Moving 1.14 ± 0 . 2 1 1.02 ± 0 . 2 3 0.86 ± 0.19 1.07 ± 0 . 1 1 1.35 ± 0 . 1 8 1.40 ± 0 . 4 3 1.35 ± 0 . 3 0 0.87 ± 0.10 1.21 ± 0 . 1 9 1.09 ± 0 . 1 6 0.92 ± 0.07 0.91 ± 0.04 3 hour roaming experiments All measured and derived variables for the 3 hour roaming experiments are presented in Table 21 in the Appendix. Graphic representations of the means of Fick's equation variables are presented in Figure 6 in the same manner as in Figure 5. Recall that statistical analysis of this dataset was limited to data from the quiet, alert and moving categories, because no animals slept during the zj°G and 37°C treatments. 35 Sleep Quiet Alert Moving Sleep Quiet Alert Moving Sleep Quiet Alert Moving Sleep Quiet Alert Moving Activity Activity Activity Activity sVn = fR x VT x % 0 2 extracted Figure 6: Mean values of each variable in Fick's equation for 3 hour roaming data. Values for activities that were statistically different from those measured in the quiet state are indicated by a * symbol, and values for temperatures that were statistically different from those measured at 27°C are indicated by a • symbol. Black symbols indicate differences that are independent of temperature (* ) or that are independent of activity (•) , while coloured symbols indicate incidents that are dependent on both factors (i.e. statistical interactions between activity and temperature). Note that only two of the six lizards slept during these experiments, and only during the 17°C trial. For this reason, sleep values were excluded from statistical analysis. 36 As in the 24 hour confined experiments, the rate of oxygen consumption during rest was significantly lower than when the animals were alert or moving. However, contrary to the 24 hour confined experiments, there was a strong temperature effect with greater oxygen consumption with increasing temperature. It is notable that with respect to the 24 hour confined dataset, oxygen consumption was lower during the 3 hour roaming experiments for each temperature treatment except at 37°C. Heart rate increased with activity and temperature, as it did in the 24 hour confined experiments. However these trends were more defined in the 3 hour roaming data. Heart rate was significantly higher when the animals were alert than when they were quiet at i7°C and 37°C, and when the animals were moving at all temperatures. As with the 24 hour experiments, the degree of the increase in heart rate at higher activity levels was temperature-dependent. Specifically, there was no significant change between alert and moving at i7°C. On the other hand, at 27°C and 37°C heart rate jumped significantly from alert values to moving values. In addition, for any given activity, heart rate increased with temperature. Generally heart rates were in the same range as their 24 hour confined counterparts. The changes in 0 2 pulse were similar to those seen in the 24 hour roaming experiments, but with a few differences. The changes in mean values with increasing activity were qualitatively different depending on the temperature. At i7°C, the mean O z pulse during the "alert" state was significantly higher than mean 0 2 pulse during the "quiet" state. However, with movement mean 0 2 pulse decreased back to levels similar to quiet values. Conversely, at 27°C and 37°C, mean 0 2 pulse followed the same trend as in the 24 hour confined dataset. In other words, mean 0 2 pulse was significantly lower during movement. The temperature effect was also slightly different from the 24 hour confined experiments. Only during the "alert" state and the "moving" state were mean values at i7°C significantly higher 37 than those at 2J°C and 3 7 ° C . Finally, 0 2 pulse was consistently lower during the 3 hour roaming experiments, with values in the range of half those of the 24 hour confined experiments. The trends in breathing frequency are also not as straightforward as in the 24 hour confined data, as the magnitude of change with activity was temperature-dependent. Breathing frequency tended to increase with activity, with mean "moving" values significantly greater than resting values for all temperatures. However, only at 3 7 ° C was the "alert" breathing frequency also significantly greater than quiet values. Unlike the 24 hour confined breathing frequencies, increases in temperature were also associated with increases in breathing frequency. Mean values at i 7 ° C were consistently lower at each activity level than mean values at 27°C and 3 7 ° C . However, the difference between the two higher temperatures was not so great, with the only difference existing between "alert" mean values. With respect to the 24 hour confined values, breathing frequencies in this dataset were, in general, lower at i 7 ° C , similar at 27°C, and higher at 3 7 ° C . Tidal volume did not prove to be significantly different at any activity level or temperature, as was the case in the 24 hour confined experiments. In addition, tidal volume was similarly variable over short term measurements, with error bars encompassing a range of 12 ml to 20 ml (representing almost a 6 7 0 / 0 increase over the minimum value). Finally, although it would appear in Figure 6 that oxygen extraction decreased with increasing activity (as in the 24 hour confined experiments), the scope was much reduced and statistical analyses revealed that this trend was not significant. There was also no statistical difference in oxygen extraction with different temperature treatments. Note that oxygen extraction was much lower in these experiments 38 (around half of the equivalent values in the 24 hour confined experiments) for all activity levels except moving. Table 2 shows the Q ^ , values for the rate of oxygen consumption, the rate of carbon dioxide production, heart rate, and breathing frequency for 3 hour roaming data. Once again, the greatest temperature effect was seen in heart rate. The greatest values were found in the transition from i7°C to 27°C, with values ranging 2.0 to 3.4. The transition from 27°C to 37°C produced slightly lower values of 1.5 to 2.5. Overall, the Q u , values for the full temperature range were calculated to be around 2.2. These values are generally higher than those found for the 24 hour confined data (see Table 1). In contrast to the 24 hour confined data, there was a significant temperature effect on the rate of oxygen consumption and carbon dioxide production. values for the full temperature range fell between 1.6 and 2.1 for oxygen consumption, and between 1.5 and 1.9 for carbon dioxide production. Similarly, values for breathing frequency were also higher than the 24 hour counterparts, ranging from 1.4 to 2.6 for the i7°C to 37°C transition. 39 Table 2: Q 1 0 v a l u e s of direct ly m e a s u r e d va r i ab les for 3 hour roam ing da ta . V a l u e s in bold c o r r e s p o n d to s igni f icant c h a n g e s in the var iab le for the g i ven activi ty a n d tempera tu re t ransi t ion. Variable Activity Q10 17°C - 2 7 ° C 27°C - 37°C 17°C - 3 7 ° C Mass-specific Sleep 0 2 consumption Quiet Alert Moving N/A 2.27 ± 0.75 1.25 + 0.10 1.92 ± 0 . 2 3 N/A 2.43 ± 0.79 2.13 ± 0.28 1.44 ± 0.21 N/A 2.05 ± 0.20 1.60 ± 0 . 0 6 1.60 ± 0 . 0 6 Mass-specific Sleep C 0 2 production Quiet Alert Moving N/A 2.25 ± 0.39 1.65 ± 0.30 1.76 ± 0.22 N/A 1.67 ± 0.16 1.60 ± 0.28 1.28 ± 0 . 1 8 N/A 1.88 ± 0 . 1 5 1.52 ± 0 . 0 5 1.45 ± 0 . 0 5 Heart Rate Sleep Quiet Alert Moving N/A 2.03 ± 0.33 2.48 ± 0.62 3.39 ± 0.30 N/A 2.50 ± 0.54 2.16 ± 0.38 1.46 ± 0.08 N/A 2.16 ± 0 . 1 1 2.15 ± 0.17 2.20 ± 0.07 Breathing Sleep Frequency Quiet Alert Moving N/A 2.89 ± 0.46 2.22 ± 0.59 1.91 ± 0 . 1 6 N/A 2.39 ± 0.59 2.01 ± 0.26 1.08 ± 0 . 0 8 N/A 2.55 ± 0.14 1.97 ± 0.17 1.42 ± 0.05 Effects of digestion on gas exchange Mean values of each variable were also measured or calculated from digesting animals at 37°C. The means from these 37°C/digesting trials are presented alongside the means collected from the 37°C/fasting trials in Tables 20 and 21 in the Appendix. Because feeding was voluntary, each lizard ate no more than 4 0 / 0 of their body weight. Consequently, statistical analysis on 37°C/digesting and 37°C/fasting data failed to detect an effect of digestion on any of the variables measured in these experiments. 40 DISCUSSION Toledo et al (2005, Submitted) recently found that, in tegu lizards ranging from io.4g to 3.75kg, metabolic rate scaled with mass according to M°'7 9. If all mass-specific values in the present study are corrected using this scaling factor, the corrected rates of oxygen consumption are in good agreement with the corrected values reported from other studies (Andrade & Abe, 1999b; Skovgaard & Wang, 2004; Toledo et al., 2005, Submitted) (see Figure 7A). The one exception is between the rates of oxygen consumption reported by Klein et al. (2003) and those of all other studies. In the present study at 37°C mean values of mass-specific 0 2 consumption during sleep, rest, and movement were only a fraction of those reported at 35°C by Klein et al. (2003). It is unclear why this disparity is so extreme, particularly since breathing frequencies and tidal volumes were relatively similar in both studies. It is possible that this may be in part due to effects of season. Several studies have shown that metabolism in the spring and summer is significantly greater (up to three times) than in the autumn or winter (Andrade & Abe, 1999b; de Souza et al., 2004; Toledo et al., 2005, Submitted). Increased metabolism during spring and summer may be associated with changes in breeding state and restructuring of the gut lining after arousal from dormancy. If the study of Klein et al. took place in the spring or summer, this could account for much of the difference seen when comparing metabolic rates with our animals, which had been in captivity under constant conditions so long as to lose most, if not all seasonal rhythms. Measurements of mass-specific C O z production in our lizards were also in close agreement with values reported in other studies (Andrade & Abe, 1999b; Skovgaard & Wang, 2004) (see Figure 7B). 41 The mean values of breathing frequency presented here were generally similar to those previously collected from tegu lizards at similar temperatures and activity states (Andrade & Abe, 1999b; Klein et al., 2003; Skovgaard & Wang, 2004) (see Figure 7C). The exception is the daytime resting breathing frequencies at 35°C reported by Klein et al. (2003), which are somewhat higher (around two times greater) than our breathing frequencies in quiet animals at 37°C. This discrepancy may be associated with the higher rates of oxygen consumption in Klein's animals, although their nighttime and exercising breathing frequencies were similar to ours despite similar differences in oxygen consumption. Tidal volumes in our lizards were generally higher than have been reported in other studies (Andrade & Abe, 1999b; Klein et al., 2003; Skovgaard & Wang, 2004). Because our lizards were two to four times larger and two to three years older than lizards in these previous studies, their lung capacities were undoubtedly greater. To determine if the discrepancy in tidal volumes was simply due to differences in mass (M), tidal volumes were standardized by M°'75, which has been determined to be the scaling factor for tidal volumes in agamid lizards (Frappell & Daniels, 1991). (The following comparisons are also qualitatively similar to those when tidal volume is standardized by M 1 , the mammalian scaling factor as cited by Frappell & Daniels (1991)) Once standardized, tidal volumes in our lizards were similar to those reported in other studies under similar conditions (see Figure 7D), with the exception of exercising values in Klein's study. Their tidal volumes were higher than those measured in our moving lizards. Again, this may be associated with the greater rates of oxygen consumption in the study of Klein et al. 42 10.0 8.0 6.0 4.0 2.0, 1.0 • 0.5 • 0.0 • o 2 <£0° ci 17°C I | i . » C P ' 27°C 37°C 0.5 0.4 o o 0.3 0.2 0.1 0.0 B 17°C A 27°C _ 60 c I "5 50 .c ro <D s.40 >> O <= 30 3 CT a> £ 20 c £ 10 ro 3* 17°C 5 A o 27°C D 20 ro 15 | 10 o > i= 5 I i 37°C 17°C 27°C 37°C 0 Present study (24 hour confined data) A Present study (3 hour roaming data) V Andrade and Abe (1999) (17°C, dormant) B Toledo et al. (2005, submitted (17°C, resting) A Skovgaard and Wang (2004) (25°C, fasting) © Toledo et al. (2005, submitted) (25°C, resting) O Klein et al. (2003) (35°C, fasting, sham operated) Figure 7: Mean values of A) mass-specific 0 2 consumption (standardized by the scaling factor M° 7 9 ) , B ) mass-specific C 0 2 production, C) breathing frequency and D) tidal volume (standardized by the scaling factor M 0 7 5 ) in the present study, and in four other studies. Data from other studies were plotted corresponding to the most similar temperature and activity level (i.e. X -axis categories) used in the present study to facilitate comparison. Note that data from other studies were converted to match the units used in the present study. Data from Klein et al. (2003) are approximated from visual inspection of Figures 2 and 3 in same. 43 Does technique make a difference? Different protocols produced different results. Although effects of activity and temperature on heart rate and tidal volume appeared to be unaffected by protocol, breathing frequency was temperature-independent in the 24 hour confined experiments but not in the 3 hour roaming experiments. Also, the mean rates of oxygen consumption during the 24 hour confined experiments were approximately 2 times greater than those measured in the 3 hour roaming experiments for all temperatures except at 37°C. One might be tempted to ascribe these discrepancies to differences in experiment duration and the inclusion or exclusion of circadian rhythms. However, comparisons made between the "quiet", "alert" and "moving" activity states involved data typically collected at the same time of day for both protocols. Differences in sample duration were also unlikely to play a role. As is discussed in Chapter 3, by increasing the sample duration to 5 minutes in the 3 hour roaming experiments, variability was reduced, but the relationship between heart rate and oxygen consumption was otherwise unaffected. Assuming this was also true for all other variables, the means for each variable at each activity level should also be unaffected. There are several other possible explanations for the differences seen in the data collected using the two protocols. One possible explanation may lie in the difference in acclimation times in the two different experimental series. In the 24 hour confined protocol, animals were generally acclimated for one day prior to recording. This short period was sufficient to change the temperature in the smaller chamber and has routinely been used in other studies (e.g. Andrade et al., 2004a; Klein et al., 2003; Skovgaard & Wang, 2004; Wood et al., 1987). The 3 hour roaming 44 protocol typically involved acclimation for 4 to 7 days. Because the chamber was much larger, more time was needed to adjust the temperature of the room. If 24 hours was not enough time for lizards to reach a new steady state, this may account for the lack of temperature effect on oxygen consumption in the 24 hour confined experiments. On the other hand, Bennett and Dawson (1976) have noted that approximately 2 weeks are required for acclimation of oxygen consumption to a new temperature to occur. Since acclimation should partially compensate for temperature effects, the 24 hour measurements should be more temperature-dependent than the 3 hour measurements. An alternative explanation involves seasonal differences in temperature sensitivity. It is possible that metabolic rate was temperature insensitive (i.e. of 1) during the 24 hour confined experiments because these experiments were carried out during the South American winter (which in Brazil lasts from May to September). During this time, wild lizards often remain inactive regardless of daily temperature fluctuations. Indeed, reports of low values during this season are common (Abe, 1983; Abe, 1995; Andrade & Abe, 1999b; de Souza et al., 2004; Milsom et al., 2005, Submitted). On the other hand, metabolism may have been temperature-sensitive during the 3 hour confined experiments because they coincided with the South American spring/summer (November). During this time, wild lizards usually conserve energy when cool, and are active when warm. However, this premise assumes that our lizards were still under the influence of circannual rhythms after 3-4 years in indoor captivity with no seasonal cues. In fact, our lizards have not shown any indication of reduced or enhanced activity during different times of the year, and have never gone into dormancy as many wild tegu lizards do (Abe, 1983; Andrade et al., 2004b). 45 Finally, recall that the 3 hour roaming protocol was implemented to alleviate some of the signs of stress that had been observed in the 24 hour confined experiments. It is possible that lizards approached their maximum rates of oxygen consumption during the 24 hour experiments due to stress and there was little scope for further increase. It may be relevant that our lizards were likely not as physically fit as their wilder counterparts since they had been living in captivity for several years. Consequently, their maximum rates of oxygen consumption were undoubtedly already relatively low, allowing the lizards less scope for increase. The dataset collected from the 24 hour confined experiments was more comprehensive than that collected from the 3 hour roaming experiments; data were available for each of the four levels of activity at every temperature in the former, but not the latter. Nonetheless, the remainder of this discussion will be based on the assumption that differences between the two datasets are due to different levels of stress, and that the 3 hour roaming dataset is more representative of the natural state. If this is true, then it would appear that stress elevates oxygen consumption to the same level regardless of temperature. Given that breathing frequency is associated with metabolic rate, it would not be surprising that stress would also eliminate the temperature dependence of breathing frequency. In fact, this does appear to be true. Following the same logic, one would expect that stress should eliminate the temperature-dependence of heart rate as well, but this was not the case. These observations will be discussed in more detail in later sections (see pages 50 and 51). 46 Effects of temperature Because tegu lizards are ectotherms, I had anticipated that all physiological rate processes, including the rate of oxygen consumption (and carbon dioxide production), heart rate and breathing frequency, would increase with ambient temperature. In general this was true in relatively unstressed animals (i.e. during the 3 hour roaming experiments). Q m values indicate the proportional increase in the rate of a reaction for an increase in temperature of io°C. Biochemical reactions generally have a Q ^ , between 2 and 3 in the absence of active up- or down-regulation. Reptiles in general have 0^ values ranging from 1.5 to 3.1 for metabolic processes (Bennett & Dawson, 1976). Our data show that there was no consistent difference between Q ^ , values calculated over the i7°C-27°C range and the 27°C-37°C range, so the focus of this discussion will lie on the full temperature range of i7°C-37°C. Oxygen consumption, heart rate and breathing frequency, which are normally all closely associated according to Fick's equation, had similar Q ^ , values close to 2.0 for the 3 hour roaming experiments. In other words, temperature sensitivity was generally low, but within the normal range reported for other reptiles. The acclimation time involved in these experiments may be responsible for these values. Because it takes approximately two weeks for oxygen consumption to stabilize at a new temperature (Bennett & Dawson, 1976), partial compensation may have occurred after the week or so of acclimation in these experiments. One would expect that, if all other things were equal, these values would be greater within the first few days of changing the temperature, and would decline over time. Normally it would be difficult to distinguish whether changes in heart rate and breathing frequency were primarily associated with changes in temperature, changes in oxygen consumption, or both. However, we can use the differences between 4 7 these two datasets to make statements to this effect. In the 24 hour confined experiments, heart rate generally changed with temperature despite the fact that oxygen consumption did not. Thus heart rate would seem to be more closely associated with changes in temperature than changes in oxygen consumption. On the other hand, in the same experiments, changes in temperature were not associated with changes in breathing frequency, suggesting that the latter was more closely associated with oxygen consumption. These conclusions are also supported by the 3 hour roaming data; heart rates changed with temperature in the same manner as in the 24 hour confined experiments, whereas oxygen consumption and breathing frequency now both changed in parallel with temperature. This evidence is circumstantial at best, but seems worthy of further investigation. Effects of state Unfortunately, only the "quiet", "alert" and "moving" states could be analyzed for the 3 hour roaming experiments, since no sleep data was collected for the two higher temperatures and since only a few lizards slept at the lowest temperature. However, the trends seen during the 24 hour confined experiments are similar to those seen in the 3 hour roaming experiments, so it could be argued that stress did not override the effects of state. If so, it may be possible to draw conclusions about sleep based on the 24 hour confined dataset. However, any such discussion must be made with caution since the data may not be representative of the natural state. The 24 hour confined data suggest that there was no statistical difference between the rate of oxygen consumption in sleeping animals and quiet animals at any temperature. If this is true, one need not distinguish between sleeping animals and quiet animals when estimating RMR and other related variables (e.g. metabolic 48 scope). However, this also implies that metabolic rate cannot be used to distinguish a sleeping lizard from a quiet lizard, so some other defining characteristic must be identified. It is notable that the "alert" state is associated with a higher rate of oxygen consumption than the "quiet" state despite the fact that the lizards are motionless for both activity levels. The upshot is that there is a danger of overestimating RMR in these lizards if measures are not taken to ensure that individuals are truly resting and not alert. This may seem like an obvious statement, but it is probably naive to assume that there is no inconsistency in the literature in this regard. Thus far, the effects of state have only been discussed in the context of oxygen consumption. Details on how other variables (i.e. heart rate, breathing frequency, etc.) contribute to this increase follow in the next two sections. Relative contributions of heart rate and O z pulse to 02 consumption It has already been said that increasing temperature or activity will induce an increase in oxygen consumption. Consequently, either heart rate or 0 2 pulse, or both, must increase to match oxygen delivery to demand. It is clear from the 3 hour roaming data that only heart rate, not O z pulse, increases with temperature-induced increases in oxygen consumption. The same can be said of activity-induced increases in oxygen consumption. In view of this, it is clear that heart rate plays the greatest role in accommodating changes in oxygen consumption brought on by temperature or activity. This result is an affirmation of the potential use of heart rate as a proxy for oxygen consumption in studies of field metabolic rate (see next 49 chapter). The analysis presented in Chapter 3 will expand on this idea and explore the relationship between the two variables in the current dataset in more depth. Although heart rate is the primary factor involved in mediating changes in oxygen consumption, 0 2 pulse is not constant for all activities and temperatures. Less oxygen is removed by the tissues per beat when the lizards are moving, and at higher temperatures. This effect seems somewhat counterproductive in the face of increasing oxygen demand, so one might wonder how it could occur. If heart rate was so rapid that there was inadequate time left to fill the heart with each beat, then stroke volume could conceivably decrease, thereby causing 0 2 pulse to drop. Alternatively, if heart rate was so rapid that diffusion time at the tissues was reduced, less oxygen could be extracted by the tissues. Why heart rate would increase so much as to impair oxygen delivery is unclear, but there are other factors influencing the circulatory system that could dominate over oxygen delivery under these circumstances. For example, glucose mobilization during exercise, rather than oxygen consumption could drive cardiac performance. It is also possible that the drop in 0 2 pulse is not related to high heart rate, but is actually due to the development of a net R-L intracardiac shunt. In this scenario, deoxygenated blood would be recirculated systemically, arterial oxygen levels would drop, and the diffusion gradient into the tissues would decrease. Unfortunately, based on the available data, it cannot be said for certain why 0 2 pulse decreases with activity and temperature. It is interesting to note that these lizards accommodated stress-related increases in oxygen consumption (i.e. the difference between 3 hour roaming and 24 hour confined values of oxygen consumption) via a different mechanism. Rather than increase heart rate, these lizards appeared to increase the O z pulse by a factor of 50 approximately two when they were under stress. This could be mediated either by increasing stroke volume, or by reducing intracardiac shunting such that arterial oxygen levels were increased and oxygen diffusion was enhanced. Relative contributions of breathing frequency, tidal volume and O z extraction to 0 2 consumption Increases in oxygen consumption must also be accompanied by respiratory adjustments. Either breathing frequency or tidal volume or oxygen extraction at the lung (or a combination of these factors) must be enhanced to match oxygen supply to demand. With increases in temperature, in general there was a significant increase in breathing frequency but not in tidal volume or oxygen extraction. The same effect was seen with increases in activity, such that the lizards breathed faster but not deeper, and oxygen extraction at the lung was relatively unchanged (and actually decreased during the 24 hour confined experiments). These results suggest that breathing frequency is the primary contributor to both temperature- and activity-related increases in oxygen consumption (the latter of which can be seen in the representative traces shown in Figure 4). In contrast, stress-induced increases in metabolism did not cause lizards to breathe faster or deeper. The implication is that they simply extracted more oxygen from the lung. This could occur if stroke volume increased (recall that stress also induced an increase in 0 2 pulse), thereby increasing cardiac output. The increased blood flow to the lung would facilitate oxygen extraction. Alternatively, an increase in a net R-L intra-cardiac shunt would cause deoxygenated blood to be recirculated systemically, arterial oxygen levels to decrease, and combined with increased 51 oxygen consumption, would lead to decreased venous oxygen levels. This would produce a greater diffusion gradient in the lung. Note that oxygen extraction at the lung can be increased only if there is sufficient time for diffusion to occur. Accordingly, when breathing frequency was very high during movement, the difference between oxygen extraction for the two protocols is negligible. On the other hand, during the 24 hour confined experiments, oxygen extraction was greatest when the animals were breathing infrequently during sleep. Presumably the apneas allowed greater time for diffusion to occur across the lung. As activity increased, breathing frequency became more important to supply sufficient oxygen to the lung to meet the increasing metabolic demands, so there was less time for oxygen to diffuse across the lung. Regarding the 3 hour roaming experiments, although there was no significant activity-related decrease in oxygen extraction at the lung, it should be noted once again that no sleep data were included in the statistical analysis. It is possible that a significant decline in oxygen extraction may have been detected had more "sleep" data been recorded when apneas were common. Does RER change as a function of changes in temperature or changes in activity? In general, RER values tended to increase with activity, particularly in the 24 hour confined experiments. On the other hand, there was no consistent change in RER values with increasing temperature in either dataset (See Tables 20 and 21 in the Appendix). 52 When the lizards were sleeping, RER values were close to 0.7. As they awoke, RER values appeared to increase, however the magnitude of this increase was variable particularly during the 3 hour roaming experiments ("Quiet" RER averaged around 0.8, but ranged from 0.7 to 0.9 depending on the treatment). If this trend is real (i.e. if more sleep data had been collected in the 3 hour roaming experiments to carry out statistical analysis), and if we can assume that lizards were at steady state when they were asleep or quiet, then this would suggest that the lizards switch from metabolizing primarily fat while sleeping to using a combination of metabolic substrates when awake and quiet. When animals were not in steady state (i.e. alert or moving), RER cannot be used to gain insight into metabolic fuel use. However, RER values will reflect the relative balance of respiratory and metabolic processes. When the lizards were alert and moving, RER values tended to be higher (although again there was great variation in the magnitude of the increase between treatments and protocols). This would suggest that the animals were hyperventilating, eliminating a greater proportion of C O z relative to oxygen consumed. In other words, there was a respiratory/metabolic imbalance at higher activity levels. This is not surprising, given that air convection requirement, which is a measure of the relative amount of air that is ventilated for a given unit of oxygen consumed, rose dramatically with increasing activity (see Tables 20 and 21 in the Appendix). Summary This study quantifies the effects of temperature and activity on oxygen consumption and associated variables in greater detail than has been explored previously. While it may seem obvious that oxygen consumption, heart rate and 53 breathing frequency would increase with temperature and activity, there appears to be special circumstances when temperature has no effect on oxygen consumption and breathing frequency. The conditions that elicit this response (or lack thereof) are likely caused by stress associated with confinement. The relative roles of heart rate and 0 2 pulse were examined in further detail to determine if increases in oxygen consumption were accompanied by increases in both heart rate and O z pulse, or heart rate only. Similar comparisons were made between oxygen consumption and breathing frequency, tidal volume, and oxygen extraction from the lung. Temperature-induced and activity-induced increases in the rate of oxygen consumption were accommodated by the same pattern of changes; increased oxygen demand was mediated entirely by increasing heart rate (not O z pulse), and breathing frequency (not tidal volume nor oxygen extraction from the lung). However, a different pattern was associated with stress-induced increases in oxygen consumption. 0 2 pulse and oxygen extraction from the lung was enhanced under these conditions, whereas heart rate and breathing pattern were unchanged. Clearly methodology can significantly influence the qualitative nature of the recorded data. The implications are obvious; conclusions may be inaccurate if conditions are not standardized. For example, as will be detailed in Chapter 3, Heart Rate Method calibrations produce different results depending on the experimental protocol employed. It would appear that the Heart Rate Method may be useful to estimate metabolic rate in tegu lizards at different temperatures and activities, but the relationship between heart rate and oxygen consumption will certainly be different in confined and therefore stressed animals. Similarly, estimates of resting metabolic rate, maximal metabolic rate, or metabolic scope will be different depending on which protocol is used. It may be intuitive that methodology would 54 influence results, but the disparity is perhaps more extreme than is commonly assumed. Finally, although increases in activity were associated with increases in oxygen consumption, heart rate and breathing frequency, not all of these variables can be used to distinguish the difference between sleeping animals and quiet animals. Breathing frequency alone was significantly different when animals were asleep, due to the emergence of apneas. Breathing frequency also proved to increase with each increase in activity level. If this trend is real (i.e. if stress does not influence this result) then it is possible that breathing frequency could be used to distinguish between these activity states in tegu lizards. 55 C H A P T E R 3 : The relationship between heart rate and the rate of oxygen consumption in tegu lizards at different temperatures and digestive states I N T R O D U C T I O N There are plenty of methods available to measure metabolic rate in confined animals, but relatively few of these can be used to estimate metabolic rate in undisturbed animals in the field. Two commonly used methods are the Time Energy Budget (TEB) Method and the Doubly-Labelled Water (DLW) Method. Although the reliability of these methods is well-established, each has its limitations. For example, the TEB method is time consuming and computationally laborious. The D L W method produces one estimate per experiment, and the duration of the experiment is limited by the half-life of the isotope. Thus the DLW method only estimates mean metabolic rate, often over a period of days. In addition, DLW requires disturbing the animal to take blood samples, so metabolic rate may be influenced by handling stress. This may preclude accurate readings in dormant animals. For these reasons, in recent years there has been much emphasis on testing the use of heart rate as a proxy for metabolic rate. The Heart Rate Method is a relatively new method which offers the potential to estimate metabolic rate over short-term activities because heart rate can change on an instantaneous basis. Coupled with the fact that improvements in data logger technology now allow heart rate to be monitored over weeks to months, this suggests that one could improve both the resolution and scope of metabolic studies of animals in the field if the method is sufficiently accurate in the species of interest. 56 Furthermore, because data loggers can record heart rate in animals without eliciting handling stress, it could provide accurate estimates of metabolic costs of activities under natural conditions, including sleep and dormancy. Using the DLW and TEB methods as a benchmark, the Heart Rate Method has now been found to be at least as accurate in several species (Bevan et al., 1994; Bevan et al., 1995b; Hawkins et al., 2000; Nolet et al., 1992). The Heart Rate Method has been tested in a wide variety of animals, including birds (Bevan et al., 1994; Bevan et al., 1995b; Froget et al., 2001; Froget et al., 2002; Green et al., 2001; Hawkins et al., 2000; Nolet et al., 1992), mammals (Boyd et al., 1999; Butler et al., 1992; McPhee et al., 2003; Nilssen et al., 1984), and, to a lesser extent, in fish (Thorarensen et al., 1996). In the latter group, the Heart Rate Method has had limited success partly because, as ectotherms, their physiological functions are dependent on body temperature. If heart rate and the rate of oxygen consumption are not equally temperature-sensitive, then they will not change in proportion to one another with increasing temperature. With this in mind, one published study to date has attempted to take temperature into account while testing the use of heart rate as a proxy for metabolic rate in a reptile (Butler et al., 2002). This study found that at each of two temperatures tested, heart rate was very closely associated with oxygen consumption in Galapagos marine iguanas, but the relationship was indeed different at each of the two temperatures. The implication was that the Heart Rate Method could be used in these reptiles but only if corrected for temperature. With this in mind, one might ask if the Heart Rate Method could similarly be used to estimate oxygen consumption in tegu lizards, if potentially confounding factors were taken into account. These lizards have been the focus of multiple metabolic studies, including studies in metabolic depression during dormancy (Abe, 1983; Abe, 1995; Andrade & Abe, 1999b; de Souza et al., 2004), metabolic acclimation at constant 57 temperature and photo-period (Milsom et al., 2005, Submitted), and metabolic responses to hypercapnia (Piercy, unpublished data). Given the effect that confinement had on the physiological variables presented in Chapter 2, the Heart Rate Method would be a valuable tool since it would allow these studies to be performed on free-ranging animals under more natural conditions. The basis of the Heart Rate Method lies in Fick's equation, whereby the rate of oxygen consumption is described in terms of heart rate multiplied by 0 2 pulse (Equation 1, page 4). If changes in the rate of oxygen consumption are mediated primarily by heart rate (i.e. if 0 2 pulse is constant or curvilinear with respect to changing oxygen consumption), heart rate and oxygen consumption should covary in a predictable, linear manner. Indeed, heart rate does play the greater role in accommodating non-stress-related changes in mean oxygen consumption, as was mentioned in the previous chapter. However, it remains to be seen whether this is true not just of mean values at specified temperatures and activity levels, but in general. Regression analyses of heart rate on oxygen consumption must be performed to examine the relationship in greater detail and to appraise the predictive power of the Heart Rate Method on a sample-by-sample basis. The goal of this chapter was to test the potential of the Heart Rate Method in tegu lizards, and to determine whether or not temperature, activity, and/or digestion influence the nature of the relationship. I hypothesized that heart rate would be tightly associated with the rate of oxygen consumption over all activities, provided that temperature and digestive state are taken into account, and provided that calibration conditions allow for natural activity. 58 M E T H O D S The experimental setup and protocol was described at length in the Methods section of Chapter 2. Data analysis For each of the 24 hour confined and 3 hour roaming experiments, oxygen consumption and heart rate data were pooled in a variety of combinations, including: 1) per tegu, activity and treatment 2) per activity and treatment 3) per tegu and treatment 4) per treatment 5) per tegu and activity 6) per activity 7) all pooled The relationship between oxygen consumption and heart rate data was analyzed for each of these groupings. Some relationships were best described by linear regressions, while others appeared to be better described by various nonlinear regressions. However, in most cases, a nonlinear regression did little to improve the fit to the data. Consequently, a simple linear model was applied to all regressions in order to facilitate comparison (except where noted). For any given treatment or activity, slopes and intercepts for different tegus were compared using an A N C O V A (Zar, 1999, pp 369-375). Statistics were performed using Microsoft Excel 2003, SigmaStat 3.1 and JMPIN 4.0.4. Results were considered significant if P<o.o5. 59 Fick's equation may also be described in terms of breathing frequency: V 0 , = / R x V T ( F I o i - F E E J (5) Therefore, an analogous regression analysis was performed on oxygen consumption and breathing frequency data to see if this provided a more accurate prediction of metabolic rate. 60 RESULTS Heart rate vs. Oj, consumption 24 hour confined experiments Regression analyses of mass-specific 0 2 consumption on heart rate were initially performed on each lizard at each temperature and activity level. Table 3 shows the primary results of this analysis for the 24 hour confined experiments. Many groupings had low sample sizes, and in such cases regressions had wide confidence intervals and were highly unreliable; the majority of the relationships were not significant (63 of 89 regressions had P>o.o5). In fact some groupings had sample sizes of o or 1, in which case no regression could be performed. Of the regressions that were significant, slopes and intercepts were highly variable between lizards at the same temperature and activity level, and the coefficient of determination (r2) ranged from very low (e.g. 0.15) to high (e.g. 0.93), probably due to the great heterogeneity of sample sizes. In other words, the significant regressions explained from 150/0 to 930/0 of the variability in the data. Pooling data within temperature and activity categories did not help; again, many relationships were still not significant (9 of 16 regressions had P>o.o5), and of those that were significant, r2 values were always low (ranging from 0.04 to 0.52). Consequently it was decided that this level of grouping would not produce relationships of sufficient predictive power to be useful with this sample size. Figure 8 shows all data for mass-specific 0 2 consumption plotted against heart rate for the 24 hour confined experiments. It is immediately evident that there was a great deal of variation in the relationship, even when treatment or activity type is 61 taken into account. Table 4 shows the primary regression variables for two different levels of regression analyses on this data: a) each lizard at each treatment, and b) pooled data for each treatment (corresponding to Figure 8A). Beginning with the former, almost all regressions proved to be significant (only 5 regressions had P>o.o5, most of which were from data collected during the 37°C/digestion experiments), but r2 values were highly variable. Of the regressions that were significant, the highest r2 values were found in individuals during the i7°C/fasting experiments (r2 ranging from 0.29-0.83), and the lowest r2 values were found in individuals during the 37°C/digesting experiments (r2 ranging from 0.23-0.26). Slopes and intercepts were also highly variable; tests showed that slopes were statistically different between lizards within each treatment, as were intercepts. For this reason, it was considered inappropriate to calculate group regressions from each lizard's mean mass-specific 0 2 consumption and mean heart rate as has been done in other studies (e.g. Bevan et al., 1995b; Boyd et al., 1999; Butler et al., 1992; Butler et al., 2002; Froget et al., 2001; McPhee et al., 2003). Instead, pooled regressions were calculated for each treatment (as in Bevan et al., 1994; Bevan et al., 1995a; Hawkins et al., 2000; Nolet et al., 1992); all of these regressions were significant, but r2 values were relatively low (ranging from 0.02 at 37°C/digesting to 0.44 at i7°C/fasting). Thus even in the best case, only 44% of the variation was explained by the regression model. Again, it is obvious from Figure 8A that there is a great deal of variation even within each treatment. Similar results were found when regression analyses were performed using activity as a grouping factor (Table 5, based on data grouped as per Figure 8B). Most regressions for each tegu at each activity level proved to be significant (only 3 regressions had P>o.o5), but again r2 values were variable (r2 ranged from 0.17 to 0.63 among the significant regressions), and there was no clear trend in the r2 values with respect to activity. Again, slopes and intercepts were highly variable; slopes were 62 significantly different between individuals within each activity category except during movement, while intercepts were significantly different in all activities. Again, calculating group regressions would be inappropriate, so pooled regressions were calculated. Pooled regressions were significant in all cases, but as with the pooled treatment regressions, r2 values were consistently low (ranging from 0.09 for alert animals to 0.32 for sleeping animals). Thus once again even in the best case scenario, only 32% of the variation was explained by the regression model. This is also evident from the scatter within each activity category in Figure 8B. It must be noted that all regressions were repeated on transformed data (In / H VS. sVD , and ln/n vs. lnsV0 ), with no consistent improvement in r2 values. For this reason, only the simple linear regression results are given. The final series of regression analyses were performed on all data pooled together regardless of treatment or activity. Linear regressions were applied to untransformed and transformed data, and the primary regression variables for each are shown in Table 6. All regressions were significant. The regression based on untransformed data was the poorest of all three regressions, with only 28% of the variability in oxygen consumption being explained by changes in heart rate. However, transforming the pooled data did improve the fit. When heart rate was natural log-transformed, the regression explained 32% of the variability in the data, whereas when both heart rate and oxygen consumption were natural log-transformed, 41% of the variability in oxygen consumption was explained by heart rate. Although this was a significant improvement in the fit, this still left 590/0 of the variability in oxygen consumption unaccounted for. 63 Table 3: Hear t rate vs . m a s s - s p e c i f i c 0 2 consumpt ion reg ress ion var iab les and est imat ion of fit (r 2) for 24 hour con f ined da ta . S i m p l e l inear r e g r e s s i o n s were ca lcu la ted for e a c h l izard at e a c h tempera tu re and activity, a s wel l a s for poo led data at e a c h temperature and activity. V a l u e s in bold indicate that the re lat ionship w a s s igni f icant ly different f rom the m e a n m a s s - s p e c i f i c 0 2 c o n s u m p t i o n . W h e n N=0 or 1, regress ion a n a l y s i s cou ld not be per fo rmed. Simi lar ly , w h e n N=2, stat ist ical s ign i f i cance cou ld not be tes ted . 24 hour confined experiments 17°C, Fasting 27°C, Fasti ng Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 2.98 55 0.018 0.148 0.754 <0.0001 28 0.012 0.349 0.009 0.626 LB 2.63 35 0.042 -0.073 0.602 <0.0001 23 0.076 -0.590 0.305 0.006 Lefty 3.26 35 0.026 -0.126 0.614 <0.0001 28 -0.005 0.430 0.020 0.470 Mu 3.85 41 0.027 0.071 0.489 <0.0001 26 -0.002 0.376 0.004 0.757 Pi 2.09 33 0.033 -0.013 0.178 0.014 25 0.004 0.462 0.002 0.817 Rho 3.18 42 0.038 0.001 0.082 0.066 11 0.044 -0.173 0.195 0.174 Quiet Gimpy 2.98 1 3 0.045 -0.098 0.925 0.177 LB 2.63 11 0.026 0.027 0.525 0.012 12 -0.052 1.069 0.042 0.522 Lefty 3.26 14 0.025 -0.164 0.834 <0.0001 32 0.009 0.076 0.425 <0.0001 Mu 3.85 10 0.021 0.088 0.755 0.001 22 0.014 0.215 0.363 0.003 Pi 2.09 19 0.069 -0.328 0.783 <0.0001 10 0.018 0.224 0.330 0.083 Rho 3.18 6 0.085 -0.462 0.378 0.194 6 0.003 0.367 0.006 0.888 Alert Gimpy 2.98 2 -0.051 0.889 1.000 N/A 4 0.035 0.029 0.696 0.166 LB 2.63 7 0.044 -0.735 0.823 0.005 5 0.043 -0.605 0.974 0.002 Lefty 3.26 7 0.036 -0.529 0.318 0.187 2 0.006 0.245 1.000 N/A Mu 3.85 7 0.024 -0.103 0.602 0.040 3 0.015 -0.108 1.000 0.004 Pi 2.09 1 5 0.031 -0.444 0.759 0.054 Rho 3.18 5 -0.005 0.451 0.016 0.838 2 0.000 0.686 1.000 N/A Moving Gimpy 2.98 1 9 0.007 0.595 0.224 0.198 LB 2.63 10 0.023 -0.238 0.248 0.143 1 Lefty 3.26 2 0.103 -3.825 1.000 N/A 1 Mu 3.85 2 0.022 -0.451 1.000 N/A 3 -0.013 0.892 0.702 0.368 Pi 2.09 2 0.104 -1.396 1.000 N/A 2 -0.150 9.663 1.000 N/A Rho 3.18 6 0.019 0.109 0.492 0.120 3 -0.025 2.646 0.567 0.457 Sleep Pooled 241 0.013 0.161 0.247 <0.0001 141 -0.010 0.551 0.047 0.010 Quiet Pooled 61 0.026 0.047 0.315 <0.0001 85 0.005 0.332 0.032 0.101 Alert Pooled 29 0.021 0.090 0.475 <0.0001 21 0.024 -0.033 0.520 0.000 Moving Pooled 23 0.003 0.467 0.008 0.689 19 0.013 0.205 0.243 0.032 64 Table 3: C o n t i n u e d . 24 hour confined experiments 37°C, Fasti ng 37°C, Digesting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 2.98 23 0.096 -1.052 0.222 0.023 37 0.015 0.121 0.035 0.268 LB 2.63 27 -0.011 0.723 0.021 0.469 15 0.006 0.332 0.013 0.680 Lefty 3.26 27 0.009 0.186 0.149 0.047 35 0.038 -0.659 0.200 0.007 Mu 3.85 33 -0.009 0.858 0.094 0.083 28 -0.006 0.965 0.029 0.388 Pi 2.09 - 26 0.020 0.206 0.077 0.169 18 0.078 -0.826 0.515 0.001 Rho 3.18 36 0.029 -0.061 0.051 0.184 26 0.010 0.254 0.061 0.223 Quiet Gimpy 2.98 29 -0.014 0.723 0.026 0.400 11 0.047 -0.478 0.304 0.078 LB 2.63 24 0.003 0.435 0.011 0.630 38 0.001 0.589 0.002 0.797 Lefty 3.26 23 0.006 0.242 0.158 0.061 15 0.002 0.555 0.011 0.709 Mu 3.85 6 0.012 0.037 0.489 0.122 16 -0.001 0.577 0.005 0.802 Pi 2.09 21 0.011 0.427 0.200 0.042 27 0.022 0.273 0.247 0.008 Rho 3.18 17 0.013 0.346 0.130 0.156 25 -0.009 1.004 0.138 0.068 Alert Gimpy 2.98 5 0.026 0.175 0.202 0.448 4 -0.105 4.653 0.928 0.037 LB 2.63 7 0.005 0.479 0.189 0.330 4 0.045 -1.686 0.896 0.053 Lefty 3.26 4 0.037 -1.270 0.894 0.055 5 0.023 -1.171 0.693 0.080 Mu 3.85 11 0.005 0.291 0.206 0.160 9 -0.006 1.200 0.088 0.440 Pi 2.09 12 0.012 0.470 0.838 <0.0001 7 0.016 0.262 0.385 0.137 Rho 3.18 4 0.010 0.394 0.314 0.440 4 -0.002 0.747 0.007 0.919 Moving Gimpy 2.98 4 -0.020 1.352 0.068 0.739 5 -0.001 0.791 0.001 0.970 LB 2.63 1 2 0.018 -0.621 1.000 N/A Lefty 3.26 0 2 0.012 -0.405 1.000 N/A Mu 3.85 6 -0.002 0.796 0.043 0.692 6 0.005 0.131 0.638 0.057 Pi 2.09 5 0.023 -0.496 0.929 0.008 10 0.007 0.482 0.329 0.083 Rho 3.18 0 5 0.014 -0.143 0.164 0.499 Sleep Pooled 172 0.002 0.466 0.011 0.166 159 0.002 0.573 0.003 0.467 Quiet Pooled 120 0.001 0.531 0.002 0.651 132 -0.004 0.796 0.037 0.027 Alert Pooled 43 0.003 0.594 0.081 0.064 33 -0.006 1.236 0.081 0.109 Moving Pooled 16 0.003 0.482 0.067 0.332 30 0.001 0.683 0.011 0.588 65 o CL E OO o o CD 21 — E 0 C L OO I 0) 00 ca 2.5 2.0 1.5 1.0 • 0.5 0.0 0 20 40 60 80 Heart Rate (beats/min) 100 • 17°C, fasting • 27°C, fasting • 37°C, fasting • 37°C, digesting B 2.5 n 20 40 60 80 Heart Rate (beats/min) Figure 8: P o o l e d heart rate a n d m a s s - s p e c i f i c 0 2 consump t i on da ta f rom 2 4 hour con f ined expe r imen t s , co lou r c o d e d by A ) t reatment, a n d B) activity. 66 T a b l e 4: Hear t rate v s . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r iab les a n d es t imat ion of fit (r 2) for 24 hour con f ined da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h tempera ture , a s wel l a s for poo led da ta at e a c h tempera tu re . V a l u e s in bold ind icate that the re lat ionship w a s signi f icant. 24 hour confined experiments 17°C, Fasting 27°C, Fasting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 2.98 59 0.018 0.148 0.735 <0.0001 44 0.012 0.389 0.425 <0.0001 LB 2.63 63 0.017 0.121 0.613 <0.0001 41 0.025 0.032 0.743 <0.0001 Lefty 3.26 58 0.021 -0.068 0.679 <0.0001 63 0.007 0.148 0.300 <0.0001 Mu 3.85 60 0.012 0.178 0.544 <0.0001 54 0.005 0.308 0.167 0.002 Pi 2.09 55 0.066 •0.258 0.833 <0.0001 42 0.020 0.225 0.697 <0.0001 Rho 3.18 59 0.018 0.157 0.291 <0.0001 22 0.019 0.193 0.802 <0.0001 Pooled 354 0.017 0.131 0.442 <0.0001 266 0.014 0.185 0.400 <0.0001 37°C, Fasting 37°C, Digesting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 p Gimpy 2.98 61 0.008 0.346 0.056 0.066 57 0.002 0.633 0.001 0.783 LB 2.63 59 0.007 0.332 0.124 0.006 59 0.009 0.294 0.268 <0.0001 Lefty 3.26 54 0.010 0.120 0.294 <0.0001 57 0.001 0.592 0.010 0.455 Mu 3.85 56 0.004 0.363 0.263 <0.0001 59 0.000 0.628 0.002 0.744 Pi 2.09 64 0.011 0.391 0.515 <0.0001 62 0.007 0.612 0.236 <0.0001 Rho 3.18 57 0.016 0.220 0.129 0.006 60 0.003 0.558 0.027 0.210 Pooled 351 0.004 0.445 0.087 <0.0001 354 0.002 0.609 0.017 0.013 T a b l e 5: Hear t rate v s . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r iab les a n d es t imat ion of fit (r 2) for 24 hour con f i ned da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h activi ty a s wel l a s for poo led da ta at e a c h activity. V a l u e s in bo ld ind icate that the re lat ionship w a s signi f icant. 24 hour confined experiments Sleep Quiet Tegu Mass (kq) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 2.98 143 0.012 0.234 0.270 <0.0001 44 0.028 0.064 0.335 <0.0001 LB 2.63 100 0.012 0.186 0.399 <0.0001 85 0.010 0.246 0.290 <0.0001 Lefty 3.26 125 0.016 0.021 0.586 <0.0001 84 0.010 0.076 0.533 <0.0001 Mu 3.85 128 0.010 0.193 0.604 <0.0001 54 0.005 0.346 0.243 0.000 Pi 2.09 102 0.037 -0.014 0.585 <0.0001 77 0.024 0.191 0.383 <0.0001 Rho 3.18 115 0.012 0.234 0.390 <0.0001 54 0.007 0.420 0.174 0.002 Pooled 713 0.010 0.236 0.323 <0.0001 398 0.006 0.381 0.097 <0.0001 Alert Moving Tegu Mass (kg) N Slope Intercept r2 p N Slope Intercept r2 P Gimpy 2.98 15 0.032 0.153 0.372 0.016 19 0.012 0.168 0.475 0.001 LB 2.63 23 0.019 0.033 0.410 0.001 14 0.009 0.255 0.444 0.009 Lefty 3.26 18 0.000 0.763 0.000 0.984 5 0.005 0.147 0.736 0.063 Mu 3.85 30 0.004 0.415 0.146 0.037 17 0.005 0.177 0.629 0.000 Pi 2.09 25 0.015 0.416 0.536 <0.0001 19 0.005 0.703 0.058 0.320 Rho 3.18 15 0.009 0.353 0.297 0.036 14 0.009 0.357 0.331 0.031 Pooled 126 0.005 0.556 0.089 0.001 88 0.005 0.480 0.108 0.002 67 T a b l e 6: Hear t rate v s . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 24 hour con f ined da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for poo led un t rans fo rmed a n d poo led t rans fo rmed da ta for 24 hour expe r imen ts . V a l u e s in bo ld indicate that the re lat ionship w a s s igni f icant. 24 hour confined experiments N x y Slope Intercept r2 P Overall Pooled 1325 In fH sV02 In fH In sV0i 0.009 0.299 0.286 <0.0001 0.240 -0.208 0.328 <0.0001 0.503 -2.316 0.413 <0.0001 The Effect of Sample duration After analyzing the 24 hour confined experiments, the question arose as to whether the observed variability might be due to the short sample duration (approximately 1-2 minutes). Thus for a given starting point, samples lasting one minute, two minutes, and so on up to 10 minutes were taken, and regressions were repeated using each of these sample durations. These analyses were performed on each lizard at 27°C/fasting for the 3 hour roaming experiments. Figure 9 shows the results for one representative individual. Regressions were significant in all cases, while A N C O V A s showed that sample duration had no effect on slope or intercept for any given lizard. However, the r2 value was affected by sample duration; generally, sampling the data over a greater duration produced a regression with a better fit (i.e. higher r2 value). However, when activity did not remain constant for the full duration of the sample, the data point could not be included in the regression; hence, N is lower for longer sample durations. Thus in many cases there was a point where the benefit of longer sample duration was outweighed by diminishing sample size. In general this point occurred around an intermediate sampling duration, approximately 5 minutes long. Note in Figure 9 68 1 minute samp les 2 minute samp les 3 minute samp les 4 minute s a m p l e s 5 minute samp les 6 minute samp les Heart Rate (beats/min) 7 minute s a m p l e s 8 minute samp les 9 minute samp les SV^ = 0.008 fH +0.248 i* = 0.297 10 20 30 40 50 60 70 Heart Rate (beats/min) 1 4 1 2 1 0 0 8 0 6 0.4 0 2 , =0.010(H»0.215 r ! » 0.413 (N = 10) 10 20 30 40 50 60 1 4 12 1 0 0.8 0.6 0 4 0 2 SKJ, =0.011 fH* 0.211 ^ = 0.411 (N = 8) 10 20 30 40 50 60 70 10 minute samp les 1.4 1.2 _ 1.0 .1 0.8 | o , 0.4 0.2 0.0 sV 0 , = 0 .012CH* 0.175 r 2 • 0.536 (N = 6) • Quiet o Alert o Moving 10 20 30 40 50 60 70 Heart Rate (beats/min) F i g u r e 9 : T h e effect of s a m p l i n g durat ion o n r e g r e s s i o n s of m a s s - s p e c i f i c 0 2 c o n s u m p t i o n o n hear t rate. D a t a we re co l l ec ted f rom the s a m e l izard whi le fast ing at 27°C, dur ing the 3 hour roaming expe r imen ts . W h e n act iv i ty d id not rema in cons tan t for the full durat ion of the s a m p l e , the da ta point w a s not inc luded in the reg ress i on ; h e n c e , N is lower for longer s a m p l e dura t ions . C o n f i d e n c e intervals for e a c h r eg ress i on are ind icated by the d a s h e d l ines. 69 that from i minute to 6 minutes, the r2 value increases and the confidence intervals decrease, but beyond 7 minutes the r2 rapidly drops and confidence intervals widen. With these results in mind, the 3 hour roaming experiments were analyzed using 5 minute sample durations. 3 hour roaming experiments The results of the 3 hour roaming experiments are presented in Tables 7, 8, 9 and 10 in a manner analogous to the 24 hour experiments (Tables 3, 4, 5 and 6). Table 7 shows the regression variables for each lizard at each temperature and activity. Low sample sizes were common; the greatest sample size was 13, but was often zero for some activity/treatment categories. Consequently, most of the relationships were not significant (50 of 67 regressions had P>o.o5). Regressions were not improved by pooling data within temperature and activity category. As with the 24 hour confined data in Table 3, many relationships were not significant (5 of 13 regressions had P>o.o5), and of those that were significant, r2 values were always low (ranging from 0.17 to 0.46). For this reason these regressions had no potential as a predictive tool, and once more this line of analysis was abandoned in favour of pooling the data. Figure 10 shows all mass-specific 02 consumption data grouped by treatment and activity, and plotted against heart rate for the 3 hour roaming experiments. It is immediately evident that the variation in the relationship is much less than was found in the 24 hour confined experiments. Table 8 shows the primary regression variables for two different levels of regression analyses on this data: a) each lizard at each treatment, b) pooled data for each treatment (corresponding to Figure 10A). 70 Analysis of each lizard at each treatment revealed that almost all regressions proved to be significant (only 2 of 24 regressions had P>o.05), and r2 values were generally improved with respect to 24 hour confined experiments (see Table 4). However, slopes and intercepts were still variable. All slopes were statistically equal during the 3 hour roaming, i7°C/fasting experiments, but slopes were significantly different in all other treatments. Intercepts were significantly different in all treatments. Again, calculating group regressions would be improper, so pooled regressions were calculated. These regressions were significant in all cases, but although r2 values were consistently better than their counterparts in the 24 hour confined experiments, there was still plenty of variation not accounted for by the regression (the r2 values ranged from 0.30 at 37°C/fasting, to 0.63 at i7°C/fasting). Thus this family of regressions accounts for 30-639/0 of the variation seen in Figure 10A. When regression analyses were performed using activity as a grouping factor (Table 9, based on data grouped as per Figure 10B), individual regressions were significant at each activity level (18 of 18 regressions had P<o.o5) except during sleep (2 of 2 regressions had P>o.o5), when sample sizes were extremely low (3 hours was generally not sufficient time for the lizards to fall asleep). For this reason sleep was excluded from the following analyses. Compared to 24 hour confined experiments (Table 5), r2 values were higher, but still slopes and intercepts were significantly different between individuals within each activity category so the data were pooled. Pooled regressions were significant in all cases except for sleep (probably due to low sample size of 15), and of the regressions that were significant, r2 values were higher than the 24 hour confined experiment values. In fact, ranging from 0.47 to 0.66, these r2 were even higher than those achieved from the regressions of the 3 hour roaming data pooled by treatment (see Table 8). Thus the best family of simple 71 linear regressions on untransformed data accounts for 4 7 - 6 6 0 / 0 of the variation seen in Figure 10 B. Once again, transformations were performed on heart rate and oxygen consumption data ( ln/ H vs. sVQ , and l n / H vs. lnsV0 ), but these transformations did not improve the regression analyses. For this reason, only the results for simple linear regressions on untransformed data are shown here. Finally, all data were pooled together to perform one last regression analysis. As with the 24 hour confined data, linear regressions were carried out on untransformed and transformed data, and the primary regression variables for each are shown in Table 10. All regressions were significant. In all cases, regressions for 3 hour roaming data explained a greater proportion of the variability than did the 24 hour confined regressions (see Table 6 ) , and as before, transforming the data produced better regressions. For the untransformed data, 6 4 0 / 0 of the variability in oxygen consumption was explained by heart rate. When only heart rate was natural log-transformed, 6 8 0 / 0 of the variability was accounted for. When both heart rate and oxygen consumption were natural log-transformed, 7 4 0 / 0 of the variability in oxygen consumption could be accounted for by changes in heart rate. Therefore when temperature, digestive state and activity were ignored, and when the data were transformed, only 269/0 of the variability in oxygen consumption was left unexplained. 72 Table 7: Hear t rate v s . m a s s - s p e c i f i c 0 2 consump t i on regress ion var iab les and est imat ion of fit (r 2) for 3 hour roaming da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h tempera ture and activity, a s wel l a s for poo led data at e a c h temperature and activity. V a l u e s in bold indicate that the re lat ionship w a s signi f icant ly different f rom the m e a n m a s s - s p e c i f i c 0 2 c o n s u m p t i o n . W h e n N=0 or 1, regress ion ana l ys i s could not be per fo rmed. Simi lar ly , w h e n N=2, stat ist ical s ign i f i cance cou ld not be tes ted . 3 hour roaming experiments 17°C, Fasting 27°C, Fasting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 3.28 3 0.021 -0.041 0.181 0.721 0 LB 2.58 0 0 Lefty 3.34 2 1.200 -8.040 1.000 N/A 0 Mu 3.88 0 0 Pi 2.22 0 0 Rho 3.22 10 0.005 0.025 0.036 0.601 0 Quiet Gimpy 3.28 9 0.080 -0.340 0.799 0.001 0 LB 2.58 3 0.007 0.047 0.396 0.566 2 -0.022 0.951 1.000 N/A Lefty 3.34 7 0.031 -0.133 0.736 0.014 6 0.004 0.048 0.067 0.619 Mu 3.88 10 0.028 -0.075 0.634 0.006 0 Pi 2.22 8 0.045 -0.297 0.894 0.000 9 0.004 0.292 0.112 0.381 Rho 3.22 3 -0.037 0.322 0.989 0.066 8 0.032 -0.151 0.770 0.004 Alert Gimpy 3.28 2 0.575 -4.376 1.000 N/A 0 LB 2.58 6 0.000 0.261 0.000 0.990 5 0.008 0.030 0.578 0.136 Lefty 3.34 6 0.039 -0.181 0.850 0.009 2 -0.001 0.247 1.000 N/A Mu 3.88 4 0.031 -0.092 0.969 0.015 2 0.002 0.135 1.000 N/A Pi 2.22 3 0.059 -0.571 0.820 0.279 4 0.002 0.405 0.022 0.861 Rho 3.22 2 0.072 -0.579 1.000 N/A 1 Moving Gimpy 3.28 4 0.019 -0.008 0.270 0.481 13 0.004 0.236 0.306 0.050 LB 2.58 6 -0.017 0.702 0.019 0.793 8 0.044 -2.177 0.605 0.023 Lefty 3.34 2 0.004 0.147 1.000 N/A 8 -0.013 1.050 0.683 0.012 Mu 3.88 3 0.018 -0.054 0.426 0.548 7 0.024 -0.916 0.667 0.025 Pi 2.22 6 0.042 -0.662 0.571 0.083 1 Rho 3.22 2 -0.019 0.369 1.000 N/A 2 0.053 -2.227 1.000 N/A Sleep Pooled 15 0.005 0.037 0.022 0.600 0 Quiet Pooled 40 0.020 -0.028 0.466 <0.0001 25 0.003 0.206 0.020 0.549 Alert Pooled 23 0.010 0.095 0.328 0.004 14 -0.001 0.348 0.013 0.669 Moving Pooled 23 0.013 0.015 0.446 0.001 39 0.008 -0.036 0.404 <0.0001 73 Table 7: C o n t i n u e d . 3 hour roaming experiments 37°C, Fasting 37°C, Digesting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 3.28 0 0 LB 2.58 0 0 Lefty 3.34 0 0 Mu 3.88 0 0 Pi 2.22 0 0 Rho 3.22 0 0 Quiet Gimpy 3.28 5 0.001 0.512 0.021 0.818 3 0.022 -0.712 0.441 0.538 LB 2.58 10 0.014 -0.095 0.257 0.135 9 0.009 -0.002 0.270 0.152 Lefty 3.34 8 0.005 0.323 0.433 0.076 11 -0.002 0.509 0.023 0.659 Mu 3.88 0 2 0.088 -4.637 1.000 N/A Pi 2.22 5 0.008 0.051 0.933 0.008 8 -0.002 0.526 0.068 0.533 Rho 3.22 9 0.013 -0.002 0.538 0.024 8 0.046 -0.786 0.202 0.264 Alert Gimpy 3.28 8 0.003 0.497 0.049 0.598 6 0.007 0.158 0.725 0.031 LB 2.58 4 0.011 -0.010 0.521 0.278 4 0.010 -0.113 0.859 0.073 Lefty 3.34 7 -0.003 0.750 0.534 0.062 4 0.012 -0.162 0.495 0.297 Mu 3.88 3 0.013 -0.367 0.838 0.264 5 -0.013 1.145 0.711 0.073 Pi 2.22 5 0.008 0.170 0.082 0.641 4 -0.024 2.514 0.726 0.148 Rho 3.22 6 0.019 -0.301 0.728 0.031 7 0.020 -0.128 0.494 0.078 Moving Gimpy 3.28 5 -0.005 1.057 0.562 0.145 7 -0.003 0.948 0.134 0.419 LB 2.58 3 -0.073 7.351 0.947 0.147 5 0.012 -0.284 0.964 0.003 Lefty 3.34 2 0.127 -11.315 1.000 N/A 0 Mu 3.88 13 0.006 -0.006 0.329 0.040 10 0.009 -0.332 0.333 0.081 Pi 2.22 6 0.002 0.604 0.062 0.633 6 -0.035 4.256 0.191 0.387 Rho 3.22 1 2 -0.051 3.708 1.000 N/A Sleep Pooled 0 0 Quiet Pooled 37 0.005 0.258 0.435 <0.0001 41 0.001 0.355 0.032 0.265 Alert Pooled 33 0.004 0.368 0.101 0.072 30 0.004 0.282 0.177 0.021 Moving Pooled 30 0.007 -0.005 0.199 0.013 30 0.006 0.064 0.211 0.011 74 A Heart Rate (beats/min) B c o oo c O O) o ^ O £ o £: ™ E d) Q. OO oo 00 CO 1.0 0.8 0.6 0.4 0.2 0.0 g, 00 =6 ,,o o o 20 40 60 80 Heart Rate (beats/min) 100 F i g u r e 10: Pooled heart rate and mass-specific 0 2 consumption data from 3 hour roaming experiments, colour coded by A) treatment, and B ) activity. 75 T a b l e 8 : Hear t rate v s . m a s s - s p e c i f i c 0 2 consump t i on reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 3 hour roam ing da ta . S i m p l e l inear r e g r e s s i o n s we re ca l cu la ted for e a c h l izard at e a c h tempera ture , a s wel l a s for poo led da ta at e a c h tempera tu re . V a l u e s in bold indicate that the re lat ionship w a s signi f icant. 3 hour roaming experiments 17°C, Fasting 27°C, Fasting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 3.28 18 0.017 0.040 0.659 <0.0001 13 0.004 0.236 0.306 0.050 LB 2.58 15 0.011 0.043 0.481 0.004 15 0.007 0.022 0.787 <0.0001 Lefty 3.34 17 0.016 -0.001 0.522 0.001 16 0.004 0.056 0.815 <0.0001 Mu 3.88 17 0.014 0.011 0.820 <0.0001 9 0.007 -0.026 0.912 <0.0001 Pi 2.22 17 0.012 0.074 0.427 0.005 14 0.009 0.223 0.813 <0.0001 Rho 3.22 17 0.015 -0.037 0.542 0.001 11 0.004 0.179 0.347 0.056 Pooled 101 0.014 0.021 0.632 <0.0001 78 0.004 0.184 0.455 <0.0001 37°C, Fasting 37°C, Digesting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 p Gimpy 3.28 18 0.000 0.573 0.003 0.842 16 0.004 0.326 0.630 0.000 LB 2.58 17 0.008 0.119 0.862 <0.0001 18 0.010 -0.044 0.955 <0.0001 Lefty 3.34 17 0.002 0.464 0.311 0.020 15 0.010 -0.063 0.404 0.011 Mu 3.88 16 0.005 0.103 0.589 0.001 17 0.006 -0.013 0.670 <0.0001 Pi 2.22 16 0.005 0.263 0.608 0.000 18 0.005 0.310 0.443 0.003 Rho 3.22 16 0.009 0.105 0.385 0.010 17 0.004 0.288 0.411 0.006 Pooled 100 0.003 0.362 0.302 <0.0001 101 0.004 0.242 0.452 <0.0001 T a b l e 9 : Hear t rate v s . m a s s - s p e c i f i c 0 2 consump t i on reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 3 hour roam ing da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h activi ty a s wel l a s for poo led da ta at e a c h activi ty. V a l u e s in bold indicate that the re lat ionship w a s s igni f icant . 3 hour roaming experiments Sleep Quiet Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 3.28 3 0.021 -0.041 0.181 0.721 17 0.008 0.151 0.618 0.000 LB 2.58 0 24 0.009 0.033 0.568 <0.0001 Lefty 3.34 2 1.200 -8.040 1.000 N/A 32 0.009 0.022 0.811 <0.0001 Mu 3.88 0 12 0.004 0.061 0.927 <0.0001 Pi 2.22 0 30 0.005 0.219 0.597 <0.0001 Rho 3.22 10 0 .005 0.025 0.036 0.601 28 0.012 0.052 0.735 <0.0001 Pooled 15 0.005 0.037 0.022 0.600 143 0.007 0.121 0.617 <0.0001 Alert Moving Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 p Gimpy 3.28 16 0.005 0.331 0.341 0.018 29 0.004 0.238 0.800 <0.0001 LB 2.58 19 0.007 0.109 0.764 <0.0001 22 0.008 0.064 0.890 <0.0001 Lefty 3.34 19 0.007 0.147 0.622 <0.0001 12 0.006 0.020 0.671 0.001 Mu 3.88 14 0.004 0.096 0.604 0.001 33 0.004 0.162 0.828 <0.0001 Pi 2.22 16 0.007 0.274 0.621 0.000 19 0.006 0.215 0.866 <0.0001 Rho 3.22 16 0.013 0.020 0.712 <0.0001 7 0.007 0.037 0.898 0.001 Pooled 100 0.007 0.174 0.474 <0.0001 122 0.005 0.139 0.664 <0.0001 76 T a b l e 10: Hear t rate v s . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 3 hour roam ing da ta . S i m p l e l inear r eg ress i ons we re ca l cu la ted for poo led un t rans fo rmed a n d poo led t rans fo rmed da ta for 3 hour expe r imen ts . V a l u e s in bold indicate that the re lat ionship w a s s igni f icant. 3 hour roaming experiments N x y Slope Intercept r2 P Overall Pooled 380 f» sV02 In fH sVQ2 In fH In sVn 0.006 0.158 0.639 <0.0001 0.192 -0.265 0.682 <0.0001 0.679 -3.441 0.739 <0.0001 Oj pulse As described in Chapter 2, mass-specific 0 2 pulse was calculated and tested with a 2-way repeated measures A N O V A followed by post-hoc Holm-Sidak tests to quantify the extent to which the variation observed in both 24 hour confined and 3 hour roaming experiments could be due to changes in mass-specific 0 2 pulse. Figure 11 illustrates the relationship between mean mass-specific 0 2 pulse, temperature, activity and digestive state for the 24 hour confined experiments. To recap the relevant results presented in Chapter 2, in fasting animals, 0 2 pulse was significantly affected by both temperature and activity. Specifically, 0 2 pulse was greater at i7°C than at 27°C or 37°C, but did not significantly change from 27°C to 37°C (see Figure 11A). Moreover, for any given temperature, mass-specific 0 2 pulse was significantly lower during movement than when animals were alert, quiet or asleep, but otherwise remained constant for the other activity levels. At 37°C, digestion did not significantly affect mean 0 2 pulse, which changed with activity level during the 37°C/digesting experiments in the same manner as it did during the 37°C/fasting experiments (Figure 11B). Interestingly, these qualitative trends also emerged from the 3 hour roaming experiments (Figure 12), but the values are significantly lower in all cases. 77 A B 17°C 27°C 37°C Figure 12: M e a n m a s s - s p e c i f i c 0 2 pu l se for 3 hour roaming exper imen ts , a s a funct ion of A ) t empera tu re a n d activity, a n d B ) d igest ive state and activi ty. A x e s a re s t anda rd i zed to t hose of F igure 11 in o rder to faci l i tate c o m p a r i s o n . 78 Breathing frequency vs. Oz consumption The same regression analysis was performed on breathing frequency data instead of heart rate, and the results are presented in Tables n to 18. Generally, these regressions were as variable as, or more variable than the heart rate regressions. 24 hour confined experiments When regressions were performed on 24 hour confined data grouped by tegu, treatment and activity level (Table 11), most relationships were not significant (60 of 90 regressions had P>o.o5), and r2 values ranged from 0.001 to 0.97 (excluding regressions based on only 2 data points). Data were pooled by treatment and activity, but still ten of sixteen regressions proved to be not significant. Of the significant pooled regressions, r2 values were low, ranging from 0.06 to 0.32. Thus at this grouping level, the best regression only accounted for 32% of the variability in the data. When the data were grouped by tegu and temperature (Table 12), most regressions were significant (only 6 of 24 regressions had P>o.05), but r2 values were variable between lizards, ranging from 0.12 to 0.74 among the significant regressions. Slopes and intercepts were significantly different between individuals within each treatment, so the data were pooled. These pooled regressions were all significant, but r2 values were very low, ranging from 0.04 to 0.30. Thus the best model only explained 30% of the variability in the data. This variability is apparent in Figure 13A. When analyzing the same data using activity as a grouping factor (Table 13), individual regressions were mostly significant during the sleep and quiet activity 79 states (only i of 1 6 regressions had P>o.o5), but were almost never significant when the lizards were alert or moving ( 1 5 of 1 6 regressions had P>o.o5). Of the significant regressions, for any given activity, slopes and intercepts were different between lizards. So, once again data were pooled, producing regressions that were significant for lizards when they were sleeping and quiet, but not when lizards were alert or moving. Additionally, each regression was an extremely poor fit to the data, with r2 values around 0 . 0 6 for the significant regressions. Figure 1 3 B illustrates the variability in the breathing frequency data when activity was taken into account. As with the heart rate regressions, transformations on breathing frequency and oxygen consumption data (In fa vs. sVD , and ln/i? vs. lnsV0 ), did not improve the fit of the regressions. The results of regressions performed on pooled data (ignoring treatment and activity) are shown in Table 1 4 . All regressions were significant. In all cases, regressions were poor compared with regressions using heart rate data (see Table 6 ) , but as before, transforming the data produced better regressions. For the untransformed data, only 140/0 of the variability in oxygen consumption was explained by breathing frequency. When breathing frequency was natural log-transformed, 250 /0 of the variability was accounted for. When both breathing frequency and oxygen consumption were natural log-transformed, the regression was slightly improved, explaining 29% of the variability in oxygen consumption. Therefore when temperature, digestive state and activity were ignored, and when the data were transformed, at least 719/0 of the variability in oxygen consumption was still left unexplained. 80 Table 11: B rea th ing f r equency v s . mass -spec i f i c 0 2 consumpt ion reg ress ion var iab les and est imat ion of fit (r 2) for 24 hour con f ined da ta . S i m p l e l inear reg ress ions w e r e ca lcu la ted for e a c h l izard at e a c h tempera ture a n d activity, as wel l a s for poo led da ta at e a c h temperature and activity. V a l u e s in bo ld ind icate that the re la t ionship w a s signi f icant ly different f rom the m e a n m a s s -spec i f i c 0 2 c o n s u m p t i o n . W h e n N=0 or 1, reg ress ion ana lys i s cou ld not be per fo rmed. Simi lar ly , w h e n N=2, stat ist ical s ign i f i cance cou ld not be tested. 24 hour confined experiments 17°C, Fasti ng 27°C, Fasti ng Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 2.98 55 0.006 0.262 0.686 <0.0001 28 0.164 0.127 0.465 <0.0001 LB 2.63 35 0.075 0.175 0.421 <0.0001 23 0.155 0.110 0.585 <0.0001 Lefty 3.26 35 0.050 0.157 0.217 0.005 28 0.044 0.144 0.303 0.002 Mu 3.85 41 0.014 0.221 0.087 0.061 26 0.024 0.250 0.197 0.023 Pi 2.09 33 0.033 0.186 0.060 0.168 25 -0.008 0.523 0.018 0.527 Rho 3.18 42 0.020 0.244 0.043 0.187 11 0.057 0.396 0.087 0.379 Quiet Gimpy 2.98 1 3 0.049 0.380 0.944 0.151 LB 2.63 11 0.097 -0.006 0.755 0.001 12 0.074 0.221 0.126 0.258 Lefty 3.26 14 0.040 0.046 0.608 0.001 32 0.029 0.154 0.346 0.000 Mu 3.85 10 -0.002 0.348 0.009 0.792 22 0.016 0.314 0.281 0.011 Pi 2.09 19 0.058 0.238 0.344 0.008 10 0.012 0.478 0.131 0.304 Rho 3.18 6 0.015 0.294 0.095 0.552 6 0.012 0.349 0.183 0.397 Alert Gimpy 2.98 2 -0.004 0.491 1.000 N/A 4 0.100 -0.198 0.788 0.112 LB 2.63 7 -0.007 0.920 0.032 0.702 5 0.125 -0.175 0.970 0.002 Lefty 3.26 7 0.092 -0.510 0.530 0.064 2 0.011 0.383 1.000 N/A Mu 3.85 7 -0.008 0.673 0.184 0.337 3 0.085 -0.737 1.000 0.003 Pi 2.09 1 5 -0.012 1.708 0.105 0.595 Rho 3.18 5 -0.002 0.412 0.018 0.831 2 0.000 0.684 1.000 N/A Moving Gimpy 2.98 1 9 0.000 1.100 0.000 0.980 LB 2.63 10 0.008 0.220 0.142 0.284 1 Lefty 3.26 2 0.009 -0.043 1.000 N/A 1 Mu 3.85 2 -0.033 2.218 1.000 N/A 3 0.003 0.170 0.093 0.803 Pi 2.09 2 0.064 -1.394 1.000 N/A 2 -0.132 9.154 1.000 N/A Rho 3.18 6 0.014 0.051 0.507 0.113 3 0.004 1.110 0.065 0.836 Sleep Pooled 241 0.007 0.255 0.245 <0.0001 141 0.005 0.392 0.003 0.554 Quiet Pooled 61 0.014 0.304 0.095 0.016 85 0.025 0.258 0.322 <0.0001 Alert Pooled 29 -0.006 0.773 0.031 0.364 21 0.011 0.718 0.069 0.249 Moving Pooled 23 0.005 0.374 0.029 0.441 19 -0.005 1.249 0.018 0.580 81 Table 11: Continued. 24 hour confined experiments 37°C, Fasting 37°C, Digesting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 2.98 23 0.110 0.144 0.564 <0.0001 37 0.066 0.330 0.435 <0.0001 LB 2.63 27 0.066 0.349 0.173 0.031 15 0.104 0.278 0.333 0.024 Lefty 3.26 27 0.033 0.304 0.294 0.004 35 0.065 0.158 0.304 0.001 Mu 3.85 33 0.011 0.428 0.062 0.162 28 0.003 0.675 0.001 0.898 Pi 2.09 26 0.128 0.354 0.321 0.003 18 0.373 0.006 0.406 0.005 Rho 3.18 36 0.104 0.314 0.326 0.000 26 0.021 0.531 0.072 0.185 Quiet Gimpy 2.98 29 0.014 0.431 0.034 0.336 11 0.076 0.343 0.619 0.004 LB 2.63 24 0.006 0.485 0.063 0.236 38 0.023 0.548 0.071 0.105 Lefty 3.26 23 0.018 0.328 0.223 0.023 15 0.006 0.555 0.035 0.504 Mu 3.85 6 0.045 0.146 0.583 0.077 16 0.002 0.521 0.008 0.742 Pi 2.09 21 0.011 0.565 0.154 0.078 27 0.011 0.710 0.095 0.117 Rho 3.18 17 0.023 0.472 0.242 0.045 25 0.022 0.582 0.155 0.051 Alert Gimpy 2.98 5 0.006 0.643 0.208 0.441 4 -0.124 2.652 0.805 0.103 LB 2.63 7 0.009 0.501 0.243 0.261 4 0.038 0.255 0.879 0.062 Lefty 3.26 4 0.054 -0.110 0.525 0.276 5 0.020 0.129 0.667 0.092 Mu 3.85 11 0.066 -0.705 0.542 0.010 9 0.054 -0.431 0.331 0.105 Pi 2.09 12 0.034 0.574 0.231 0.113 7 0.048 0.453 0.666 0.025 Rho 3.18 4 -0.006 0.756 0.015 0.879 4 -0.004 0.716 0.010 0.900 Moving Gimpy 2.98 4 0.011 -0.034 0.750 0.134 5 0.000 0.717 0.002 0.947 LB 2.63 1 2 0.023 -0.471 1.000 N/A Lefty 3.26 0 2 0.019 -0.602 1.000 N/A Mu 3.85 6 -0.002 0.746 0.094 0.554 6 -0.007 0.967 0.412 0.169 Pi 2.09 5 0.012 0.279 0.127 0.556 10 -0.001 0.984 0.007 0.816 Rho 3.18 0 5 0.007 0.486 0.052 0.712 Sleep Pooled 172 0.013 0.457 0.061 0.001 159 0.023 0.523 0.093 <0.0001 Quiet Pooled 120 0.009 0.487 0.075 0.003 132 0.003 0.652 0.009 0.286 Alert Pooled 43 0.004 0.680 0.020 0.362 33 -0.001 0.911 0.000 0.944 Moving Pooled 16 0.003 0.517 0.049 0.410 30 -0.001 0.818 0.003 0.774 82 . • f o • • «•• • • • • 17°C, fasting • 27°C, fasting • 37°C, fasting • 37°C, digesting 20 40 60 80 Breathing Frequency (breaths/min) B <b o o o 0 ° « D<5b *o 0 0 o * o 0 0 ° « b 0 S B * _ # « o O O J o O o o o o 20 40 60 80 Breathing Frequency (breaths/min) Figure 13: P o o l e d breath ing f r e q u e n c y and m a s s - s p e c i f i c 0 2 consump t i on da ta f rom 24 hour con f i ned expe r imen ts , co lou r c o d e d by A ) t reatment, a n d B) activi ty. 83 Table 12: B rea th ing f r equency vs . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab l es a n d es t imat ion of fit (r 2) for 24 hour con f ined da ta . S i m p l e l inear r eg ress i ons we re ca l cu la ted for e a c h l izard at e a c h tempera tu re , a s wel l a s for poo led da ta at e a c h tempera tu re . V a l u e s in bold indicate that the re lat ionship w a s s igni f icant . 24 hour confined experiments 17°C, Fasting 27°C, Fasting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 2.98 59 0.005 0.267 0.598 <0.0001 44 0.012 0.518 0.355 <0.0001 LB 2.63 63 0.008 0.321 0.182 0.001 41 0.047 0.287 0.587 <0.0001 Lefty 3.26 58 0.013 0.272 0.201 0.000 63 0.006 0.292 0.141 0.002 Mu 3.85 60 0.003 0.285 0.063 0.052 54 0.002 0.391 0.024 0.262 Pi 2.09 55 0.033 0.242 0.579 <0.0001 42 0.013 0.519 0.285 0.000 Rho 3.18 59 0.008 0.274 0.250 <0.0001 22 0.018 0.426 0.739 O.0001 Pooled 354 0.008 0.288 0.186 <0.0001 266 0.013 0.382 0.299 <0.0001 37°C, Fasting 37°C, Digesting Tegu Mass (kg) N Slope Intercept r2 p N Slope Intercept r2 p Gimpy 2.98 61 0.006 0.446 0.159 0.002 57 0.005 0.649 0.029 0.206 LB 2.63 59 0.009 0.464 0.191 0.001 59 0.010 0.568 0.251 <0.0001 Lefty 3.26 54 0.022 0.346 0.294 <0.0001 57 0.001 0.630 0.002 0.752 Mu 3.85 56 0.003 0.492 0.189 0.001 59 -0.001 0.662 0.006 0.568 Pi 2.09 64 0.008 0.599 0.117 0.006 62 0.004 0.793 0.057 0.063 Rho 3.18 57 0.023 0.473 0.212 0.000 60 0.005 0.617 0.108 0.010 Pooled 351 0.006 0.510 0.101 <0.0001 354 0.004 0.647 0.041 0.000 Table 13: B rea th ing f r equency vs . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 24 hour con f i ned da ta . S i m p l e l inear r e g r e s s i o n s we re ca l cu la ted for e a c h l izard at e a c h activi ty a s wel l a s for poo led da ta at e a c h activity. V a l u e s in bold indicate that the re lat ionship w a s signi f icant. 24 hour confined experiments Sleep Quiet Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 2.98 143 0.005 0.408 0.046 0.010 44 0.012 0.525 0.091 0.047 LB 2.63 100 0.112 0.192 0.389 <0.0001 85 0.009 0.485 0.051 0.037 Lefty 3.26 125 0.063 0.120 0.675 <0.0001 84 0.031 0.173 0.469 <0.0001 Mu 3.85 128 0.054 0.175 0.465 <0.0001 54 0.008 0.392 0.109 0.015 Pi 2.09 102 0.026 0.436 0.023 0.126 77 0.016 0.549 0.115 0.003 Rho 3.18 115 0.070 0.303 0.197 <0.0001 54 0.018 0.505 0.089 0.028 Pooled 713 0.011 0.398 0.065 <0.0001 398 0.011 0.469 0.063 <0.0001 Alert Moving Tegu Mass (kg) N Slope Intercept r2 p N Slope Intercept r2 P Gimpy 2.98 15 -0.006 1.068 0.027 0.557 19 0.004 0.670 0.032 0.467 LB 2.63 23 0.013 0.605 0.089 0.167 14 0.005 0.478 0.036 0.517 Lefty 3.26 18 0.008 0.622 0.049 0.378 5 0.013 -0.248 0.903 0.013 Mu 3.85 30 -0.002 0.661 0.004 0.737 17 -0.003 0.705 0.037 0.462 Pi 2.09 25 0.008 0.919 0.052 0.272 19 0.006 0.669 0.061 0.309 Rho 3.18 15 -0.001 0.593 0.001 0.917 14 0.017 0.126 0.246 0.072 Pooled 126 0.003 0.741 0.007 0.368 88 0.003 0.600 0.018 0.218 84 T a b l e 14 : B rea th ing f r e q u e n c y vs . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 24 hour con f i ned da ta . S i m p l e l inear r eg ress i ons we re ca l cu la ted for poo led un t rans fo rmed a n d poo led t rans fo rmed da ta for 24 hour expe r imen ts . V a l u e s in bo ld indicate that the re lat ionship w a s s igni f icant . 24 hour confined experiments N x y Slope Intercept r2 P Overall Pooled 1325 In fR sV02 In fR In sVn 0.008 0.453 0.136 <0.0001 0.129 0.329 0.247 <0.0001 0.260 -1.172 0.287 <0.0001 3 hour roaming experiments As with heart rate, breathing frequency appeared to be less variable during 3 hour roaming experiments than during 24 hour confined experiments (see Figure 14). Table 15 shows the results of regression analysis performed on each lizard at each activity level and treatment. However, at this grouping level only 8 of 67 regressions proved to be significant with highly variable r2 values even for those regressions that were significant (ranging from 0.54 to 0.96). Pooling the data produced only 2 significant regressions, both with low r2 values (0.12 and 0.33). Table 16 shows the results of regression analyses performed on the breathing frequency data grouped by treatment. Although most regressions were significant (only 7 of 24 regressions had P>o.o5), variability was still an issue, with r2 values for significant regressions ranging between 0.30 to 0.88. When differences in slopes and intercepts were tested, slopes proved to be not significantly different between lizards during the i7°C/fasting trials, but intercepts were significantly different in all treatments. Thus the data were pooled; these regressions were all significant, but r2 values were still low, ranging from 0.20 to 0.43. Figure 14A shows this data with respect to treatment. 85 The next analyses involved pooling the data by activity type (Table 17), and although sample sizes were too low for "sleep" regressions, most other regressions were significant (only 4 of 18 regressions had P>o.o5). There was still much variation in the r2 values for significant regressions, with a range of 0.29 to 0.96. For any given activity, slopes were significantly different between lizards except in the "alert" category. Intercepts were significantly different in all cases. Accordingly, pooled regressions were calculated. These were all significant (except for "sleep"), but r2 values were all in the range of 0.33 to 0.35. Hence the best model only explained 35% of the variation in the relationship between breathing frequency and oxygen consumption. Figure 14B shows these data with respect to activity type. As before, transformations on breathing frequency and oxygen consumption data (ln/R vs. sVQ , and ln/j; vs. lnsV0 ), did not improve the fit of the regressions. The final level of analysis involved pooling all 3 hour roaming data. The results of these regressions are shown in Table 18. All regressions were significant, but in all cases, r2 values were lower compared with regressions using heart rate data (see Table 10). Nevertheless, transforming the data produced better regressions. 42% of the variability in oxygen consumption was explained by breathing frequency for untransformed data. When breathing frequency was natural log-transformed, 53% of the variability in oxygen consumption was due to changes in breathing frequency. When both breathing frequency and oxygen consumption were natural log-transformed, the regression was improved further, explaining 63% of the variability in oxygen consumption. Therefore when temperature, digestive state and activity are ignored, and when the data are transformed, only 379/0 of the variability in oxygen consumption is left unexplained by changes in breathing frequency. 86 Table 15: Brea th ing f r equency vs . mass -spec i f i c 0 2 consumpt ion reg ress ion var iab les and est imat ion of fit (r 2) for 3 hour roaming da ta . S i m p l e l inear r e g r e s s i o n s were ca lcu la ted for e a c h l izard at e a c h tempera ture and activity, a s well as for poo led da ta at e a c h temperature a n d activity. V a l u e s in bold ind icate that the re la t ionship w a s s igni f icant ly different f rom the m e a n m a s s - s p e c i f i c 0 2 c o n s u m p t i o n . W h e n N=0 or 1, regress ion ana l ys i s cou ld not be per formed. S imi lar ly , w h e n N=2, stat ist ical s ign i f i cance cou ld not be tes ted . 3 hour roaming experiments 17°C, Fasting 27°C, Fasting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 3.28 3 0.008 0.062 0.103 0.792 0 LB 2.58 0 0 Lefty 3.34 2 0.062 0.016 1.000 N/A 0 Mu 3.88 0 0 Pi 2.22 0 0 Rho 3.22 10 0.009 0.051 0.042 0.571 0 Quiet Gimpy 3.28 9 0.024 0.088 0.232 0.189 0 LB 2.58 3 0.002 0.117 0.006 0.952 2 -0.012 0.364 1.000 N/A Lefty 3.34 7 0.020 0.056 0.388 0.135 6 0.005 0.080 0.402 0.176 Mu 3.88 10 0.012 0.044 0.544 0.015 0 Pi 2.22 8 0.055 0.065 0.289 0.170 9 0.017 0.254 0.284 0.140 Rho 3.22 3 0.012 0.032 0.963 0.123 8 0.031 0.109 0.790 0.003 Alert Gimpy 3.28 2 -0.026 0.417 1.000 N/A 0 LB 2.58 6 0.002 0.241 0.073 0.605 5 0.015 0.120 0.863 0.022 Lefty 3.34 6 0.060 -0.077 0.869 0.007 2 -0.004 0.259 1.000 N/A Mu 3.88 4 0.011 0.059 0.683 0.173 2 0.001 0.192 1.000 N/A Pi 2.22 3 0.303 -0.693 0.776 0.314 4 -0.007 0.542 0.187 0.568 Rho 3.22 2 -0.006 0.250 1.000 N/A 1 Moving Gimpy 3.28 4 0.006 0.164 0.236 0.514 13 0.002 0.376 0.279 0.064 LB 2.58 6 0.000 0.294 0.002 0.941 8 0.005 0.219 0.111 0.420 Lefty 3.34 2 0.051 -0.889 1.000 N/A 8 -0.005 0.546 0.406 0.089 Mu 3.88 3 -0.004 0.331 0.082 0.816 7 0.003 0.218 0.224 0.284 Pi 2.22 6 0.003 0.267 0.018 0.800 1 Rho 3.22 2 -0.004 0.245 1.000 N/A 2 -0.003 0.492 1.000 N/A Sleep Pooled 15 0.023 0.044 0.220 0.078 0 Quiet Pooled 40 0.022 0.071 0.124 0.026 25 0.011 0.186 0.066 0.213 Alert Pooled 23 0.001 0.216 0.002 0.851 14 0.006 0.238 0.045 0.465 Moving Pooled 23 0.002 0.230 0.013 0.599 39 0.005 0.175 0.137 0.020 87 Table 15: C o n t i n u e d . 3 hour roaming experiments 37°C, Fasting 37°C, Digesting Activity Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Sleep Gimpy 3.28 0 0 LB 2.58 0 0 Lefty 3.34 0 0 Mu 3.88 0 0 Pi 2.22 0 0 Rho 3.22 0 0 Quiet Gimpy 3.28 5 0.000 0.544 0.000 0.991 3 -0.007 0.816 0.869 0.235 LB 2.58 10 -0.004 0.464 0.109 0.351 9 0.008 0.276 0.224 0.198 Lefty 3.34 8 0.008 0.424 0.544 0.037 11 0.006 0.355 0.234 0.131 Mu 3.88 0 2 0.050 -0.490 1.000 N/A Pi 2.22 5 0.010 0.307 0.914 0.011 8 0.003 0.434 0.058 0.565 Rho 3.22 9 0.004 0.349 0.138 0.325 8 -0.002 0.394 0.005 0.867 Alert Gimpy 3.28 8 0.007 0.392 0.339 0.130 6 -0.003 0.674 0.544 0.094 LB 2.58 4 0.005 0.353 0.330 0.425 4 -0.001 0.588 0.015 0.879 Lefty 3.34 7 -0.002 0.632 0.026 0.732 4 0.005 0.396 0.535 0.269 Mu 3.88 3 0.010 0.190 0.762 0.325 5 -0.016 0.638 0.179 0.478 Pi 2.22 5 0.011 0.437 0.106 0.592 4 -0.014 1.212 0.961 0.020 Rho 3.22 6 -0.017 0.982 0.309 0.252 7 -0.003 0.466 0.080 0.539 Moving Gimpy 3.28 5 -0.001 0.637 0.022 0.813 7 0.012 0.026 0.620 0.036 LB 2.58 3 0.023 -0.389 0.498 0.501 5 0.000 0.857 0.000 0.988 Lefty 3.34 2 0.011 0.104 1.000 N/A 0 Mu 3.88 13 0.000 0.526 0.000 0.948 10 0.001 0.441 0.010 0.785 Pi 2.22 6 -0.003 0.939 0.115 0.511 6 0.001 0.670 0.021 0.784 Rho 3.22 1 2 0.066 -1.897 1.000 N/A Sleep Pooled 0 0 Quiet Pooled 37 0.006 0.372 0.331 0.000 41 0.001 0.395 0.014 0.462 Alert Pooled 33 0.002 0.517 0.008 0.613 30 0.003 0.409 0.076 0.139 Moving Pooled 30 -0.004 0.843 0.067 0.166 30 0.001 0.621 0.001 0.867 88 1.0 c o -#—• CL E 00 c - — -O c n o ^ O | — E o S-o CL oo • 00 00 co 0.8 0.6 0.4 0.2 H 0.0 20 40 60 80 Breathing Frequency (breaths/min) 17°C, fasting 27°C, fasting 37°C, fasting 37°C, digesting B c o a. E Z! 00 c . — . O O ) o ^ - T 1 C O | — E a> CL 00 • 00 00 C O 1.0 0.8 0.6 0.4 0.2 0.0 c o o o 0 ^ o <3 o o o„ 0 Q ) ° o 9 . * » t <fi ° 0 ° * 1 t f e % c O * o o o 20 40 60 80 Breathing Frequency (breaths/min) Figure 14: Pooled breathing frequency and mass-specific 0 2 consumption data from 3 hour roaming experiments, colour coded by A) treatment, and B) activity. 89 T a b l e 1 6 : Brea th ing f requency vs . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r iab les a n d es t imat ion of fit (r 2) for 3 hour roaming da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h tempera tu re , a s wel l a s for poo led da ta at e a c h tempera tu re . V a l u e s in bold indicate that the re lat ionship w a s s igni f icant . 3 hour roaming experiments 17°C, Fasting 27°C, Fasting Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 3.28 18 0.008 0.125 0.602 0.000 13 0.002 0.376 0.279 0.064 LB 2.58 15 0.004 0.180 0.413 0.010 15 0.006 0.216 0.744 <0.0001 Lefty 3.34 17 0.007 0.113 0.296 0.024 16 0.004 0.111 0.757 <0.0001 Mu 3.88 17 0.006 0.074 0.700 <0.0001 9 0.005 0.128 0.876 0.000 Pi 2.22 17 0.004 0.233 0.163 0.109 14 0.012 0.285 0.840 <0.0001 Rho 3.22 17 0.003 0.061 0.368 0.010 11 0.004 0.212 0.310 0.075 Pooled 101 0.006 0.121 0.357 <0.0001 78 0.004 0.233 0.426 <0.0001 37°C, Fasting 37°C, Digesting Tegu Mass (kg) N Slope Intercept r2 p N Slope Intercept r2 p Gimpy 3.28 18 0.002 0.528 0.061 0.323 16 0.001 0.553 0.012 0.691 LB 2.58 17 0.008 0.313 0.628 0.000 18 0.008 0.264 0.739 <0.0001 Lefty 3.34 17 0.004 0.483 0.566 0.001 15 0.007 0.344 0.680 0.000 Mu 3.88 16 0.003 0.324 0.492 0.003 17 0.004 0.248 0.681 <0.0001 Pi 2.22 16 0.006 0.460 0.556 0.001 18 0.005 0.480 0.439 0.003 Rho 3.22 16 0.001 0.414 0.003 0.841 17 0.003 0.358 0.203 0.069 Pooled 100 0.003 0.442 0.198 <0.0001 101 0.004 0.369 0.362 <0.0001 T a b l e 1 7 : Brea th ing f r e q u e n c y v s . m a s s - s p e c i f i c Q 2 c o n s u m p t i o n reg ress ion va r iab les a n d es t imat ion of fit (r 2) for 3 hour roaming da ta . S i m p l e l inear r eg ress i ons we re ca lcu la ted for e a c h l izard at e a c h act iv i ty a s wel l a s for poo led da ta at e a c h activity. V a l u e s in bold indicate that the re lat ionship w a s signi f icant. 3 hour roaming experiments Sleep Quiet Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 P Gimpy 3.28 3 0.008 0.062 0.103 0.792 17 0.006 0.225 0.349 0.013 LB 2.58 0 24 0.008 0.258 0.146 0.065 Lefty 3.34 2 0.062 0.016 1.000 N/A 32 0.029 0.046 0.719 <0.0001 Mu 3.88 0 12 0.017 0.031 0.957 <0.0001 Pi 2.22 0 30 0.011 0.284 0.550 <0.0001 Rho 3.22 10 0.009 0.051 0.042 0.571 28 0.020 0.159 0.549 <0.0001 Pooled 15 0.023 0.044 0.220 0.078 143 0.010 0.222 0.340 <0.0001 Alert Moving Tegu Mass (kg) N Slope Intercept r2 P N Slope Intercept r2 p Gimpy 3.28 16 0.008 0.315 0.402 0.008 29 0.006 0.254 0.369 0.001 LB 2.58 19 0.007 0.221 0.646 <0.0001 22 0.017 -0.211 0.699 <0.0001 Lefty 3.34 19 0.014 0.161 0.743 <0.0001 12 0.005 0.116 0.140 0.230 Mu 3.88 14 0.012 0.071 0.660 0.000 33 0.007 0.060 0.677 <0.0001 Pi 2.22 16 0.010 0.394 0.288 0.032 19 0.011 0.118 0.750 <0.0001 Rho 3.22 16 0.001 0.374 0.004 0.814 7 0.018 -0.281 0.499 0.076 Pooled 100 0.009 0.246 0.329 <0.0001 122 0.008 0.121 0.352 <0.0001 90 T a b l e 1 8 : B rea th ing f r equency v s . m a s s - s p e c i f i c 0 2 c o n s u m p t i o n reg ress ion va r i ab les a n d es t imat ion of fit (r 2) for 3 hour roam ing da ta . S i m p l e l inear r eg ress i ons we re ca l cu la ted for poo led un t rans fo rmed a n d poo led t rans fo rmed da ta for 3 hour expe r imen ts . V a l u e s in bo ld indicate that the re lat ionship w a s s igni f icant . 3 hour roaming experiments N x y Slope Intercept r2 P Overall Pooled 380 In fR SV02 In fR In sV0 0.006 0.243 0.419 <0.0001 0.121 0.073 0.525 <0.0001 0.449 -2.316 0.627 <0.0001 91 D I S C U S S I O N The correlation between heart rate and metabolic rate Any regression equation can serve as a predictive tool, but unless the coefficient of determination (r2) is close to i, the confidence in the accuracy of that prediction is reduced. Most studies in which the Heart Rate Method has been validated have achieved r2 values around 0.8 and higher; that is, around 80% or more of the variation in oxygen consumption was explained by variations in heart rate. The data collected in this study was grouped using every combination possible, and regression analyses on almost every level failed to produce a model with a consistently high r2. Contrary to our hypothesis, only when temperature, digestive state and activity were disregarded was the predictive power of heart rate maximized. This result was unexpected in light of other calibration studies which emphasize the need to take temperature (e.g. Butler et al., 2002), activity (e.g. Bevan et al., 1995b; Froget et al., 2002; Hawkins et al., 2000; Nolet et al., 1992), and/or digestion (e.g. Froget et al., 2001; McPhee et al., 2003) into account. Nevertheless, it suggests that much of the variation in the correlation between heart rate and oxygen consumption can be averaged out only when the maximum scope of metabolic states is included. Accounting for the variability in the heart rate and oxygen consumption data As will be discussed later (see pages 97-98), any factor that influences the 0 2 pulse will alter the relationship between heart rate and oxygen consumption. This, in turn, will be manifested as increased scatter around the regression, unless 0 2 pulse is kept constant by controlling for the influencing factor. If such measures are 92 successful, then this will be represented by regressions with little scatter, but with differing slopes and/or intercepts under different conditions. Temperature, digestive state and activity level were accounted for in these experiments, however there are other potential influences which were not controlled in the current study. Indeed, there are several possible explanations for the extreme variation seen in the data collected from the 24 hour confined experiments. Methodologically, confining the lizards to a box may have caused them to be unusually stressed. Also, the length of the sampling interval (1-2 minutes) was less than ideal (4-6 minutes). However, some variation may have been real; these experiments were carried out over the course of 24 hours and will include any circadian rhythms that may occur in heart rate and oxygen consumption. To my knowledge, no other calibration study thus far has considered this to be a significant confounding factor. The two variables may not vary proportionately, however, and thus many calibration and validation experiments reported in the literature that were performed exclusively during the day (e.g. Butler et al., 2002), may not accurately predict nighttime metabolic rates and this may partially explain some of the variability in our 24 hour results. In order to determine if the predictive value of the Heart Rate Method could be improved in these animals, I carried out the second series of calibration experiments during the daytime only and refined the methods. Two potential methodological influences were addressed in these 3 hour roaming experiments. First, the issue of stress was at least partially alleviated by allowing the lizards much more room in which to move and exercise under more natural conditions. The assumption was that, at any given temperature and activity, 0 2 pulse was not constant under the influence of stress. Indeed, this was confirmed in Chapter 2 and is obvious from Figures 11 and 12. Secondly, the issue of sample duration was explored by reanalyzing the same data for each lizard at 27°C ten times, using a different sample 93 duration each time (see the following section for details). When the "optimum" sample duration was identified as approximately 5 minutes in length, the rest of the data were analyzed using this optimum sample duration. Presumably, instantaneous changes in 0 2 pulse were averaged out over longer sample durations. Indeed, when these measures were taken, the variability in the data was drastically reduced, as can be seen when comparing the scales of Figures 8 and 10. However, considerable variability remained in the data collected from the 3 hour roaming experiments, and even though the regressions for these data generally explained more of the variability than did the regressions for the 24 hour confined data, they still had low r2 values for data grouped by temperature, activity or digestive state. It was for this reason that the idea of taking temperature, activity and/or digestive state into account was abandoned. Visual inspection of the data suggested that the effect of temperature and digestion was to extend the relationship along a single curve. Indeed, once the data was pooled, regression analysis confirmed that a single equation explained more of the variability than did separate equations per treatment. Thus the hypothesis that heart rate would only be tightly associated with oxygen consumption if temperature and digestive state were taken into account proved to be incorrect. The equation which best described the relationship in unstressed animals is given by: Ins V 0 i =-3.441+0 .679 l n / R . (6) The data indicate that both within- and between-individual variability exists in this group of animals. An example of the disparity between individual responses can be seen in Figure 15, which depicts the regressions of two different animals under the same conditions (37°C/digesting, 3 hour roaming experiments). These regressions were taken from individuals of similar size and same sex, and were based on the 94 same sample size with a similar number of data points falling in each activity category. The first regression has an extremely high r2 and tight confidence intervals, while the other is characterized by much more scatter and wide confidence intervals. Slopes and intercepts are also different. Note that these regressions represent one of the best grouping options (i.e. by treatment). What isn't illustrated in Figure 15 is the within-individual variability that caused the relationship within each lizard to be inconsistent over time; that is, although one animal may have had a remarkably tight relationship for one treatment, the same animal could produce a poor regression for another treatment (see r2 values for any given individual in Tables 4 and 8). 95 F i g u r e 1 5 : R e g r e s s i o n s of two dif ferent l izards under the s a m e cond i t ions (37°C/d iges t ing , 3 hour roam ing expe r imen ts ) . T h e da ta for A w a s co l lec ted f rom " L B " a n d for B w a s co l lec ted f rom " P i . " C o n f i d e n c e intervals for e a c h r eg ress i on are ind ica ted by a d a s h e d l ine. 96 Sample duration To date, although many other studies have tested the applicability of the Heart Rate Method on several different species, none have explored how sample duration may affect the regression model (with the possible exception of Green et al. (2001), who, in validation studies, compared measured sV0 with estimated sVQ based on 30 minute averages and 24 hour averages). Of course, it is not surprising that in animals where the regression is poor, much of the short-term variability will be averaged out if longer samples are taken. Nevertheless, it is noteworthy that despite the fact that one of the most emphasized benefits of the Heart Rate Method is the potential to predict short-term metabolic rate, in some animals the best results are achieved not from instantaneous data, but from data averaged over longer times. In fact, to my knowledge only one published study has successfully calibrated the Heart Rate Method using samples less than 5 min in duration (i.e. 30-60 second samples, Butler et al., 2002). Thus, in most cases, the metabolic costs of activities that last less than 5 minutes may not be accurately estimated. O, pulse In these animals, 74% of the variability in oxygen consumption could be accounted for by changes in heart rate, so 260/0 of the variation in oxygen consumption must have been mediated by something else. According to Fick's equation, the only other factor that could have contributed to changes in oxygen consumption was the O z pulse (either via changes in stroke volume or oxygen extraction at the tissues). Indeed, mean 0 2 pulse did vary depending on activity state and temperature. In terms of activity, mean 0 2 pulse was significantly lower during movement. Based on this, one might be tempted to conclude that excluding the "Moving" data would produce regressions that explained more of the variability. However when "Moving" data were omitted from regression analyses for any given temperature, it 97 was interesting to note that r2 values generally decreased. This was likely due to the fact that there was still plenty of variability on a sample-by-sample basis as is evident from the wide error bars for each mean value (see Figures 5 and 6 in Chapter 2). In terms of temperature, mean 0 2 pulse was significantly higher at IJ°C. If not for the fact that error bars around the means for any given temperature were also wide, this would not have affected the r2 value calculated for any given experiment since temperature was held constant. In terms of digestive state, there was no statistical difference between mean 0 2 pulses when fasting and when digesting, which is consistent with the observation that r2 values were similar between 37°C/fasting and 37°C/digesting experiments. However, there was still some variability around each mean value. With these facts in mind, it is certain that 0 2 pulse was not constant for all pooled data. This should not come as a surprise, in light of the fact that heart rate and stroke volume (S V) are related to blood pressure (BP): BP = fHxSVxTPR (7) where TPR is the total peripheral resistance. Blood pressure is tightly regulated, so changes in heart rate that are independent of changes in TPR must be accompanied by reciprocal changes in stroke volume, and therefore 0 2 pulse. If stroke volume does not change consistently in proportion with heart rate, 0 2 pulse will vary in an unpredictable manner. Other factors If some of the variation in oxygen consumption was not caused by changing 0 2 pulse, heart rate must be uncoupled from oxygen consumption via some other mechanism. Other studies have identified circumstances under which this 98 uncoupling probably occurs, at least temporarily. Given that blood pressure is related to heart rate, stroke volume and total peripheral resistance, any event that causes a change in blood pressure via a change in peripheral resistance but not stroke volume will cause a change in heart rate independent of oxygen consumption and 0 2 pulse. The most obvious example is the hysteresis that occurs in reptiles when rapidly heated or cooled. For example, when bearded dragons are rapidly heated from below their preferred body temperature, heart rate is higher at any given body temperature (and metabolic rate) than it is during cooling (Bartholomew & Tucker, 1963; Grigg & Seebacher, 1999). These physiological adjustments occur in order to facilitate thermoregulation. If the ambient temperature increases, peripheral resistance drops as vessels vasodilate and heart rate increases. Thus, blood pressure is maintained, and warmed peripheral blood is more efficiently circulated to the cooler core with a higher heart rate. On the other hand, if the ambient temperature drops, heat loss from the periphery is slowed by vasoconstriction and by decreasing the heart rate. Therefore, one can assume that oxygen consumption is more closely related to body temperature than it is to heart rate during these periods of thermoregulation. This hysteresis has been identified in many reptiles. Indeed, a hysteresis has been noted in tegu lizards kept in semi-wild conditions. In the morning, their heart rate increases dramatically well before body temperature has begun to increase, and in the evening, the opposite occurs (Sanders et al., unpublished data). One might expect that if the trigger for this hysteresis is changing ambient temperature, then it should not occur in our experiments since the animals were acclimated to constant temperature. However it has also been noted that in semi-wild conditions, on days when lizards stay inactive in their burrow due to poor weather, a small hysteresis is still seen in the mornings despite no change in burrow temperature (Sanders et al., 99 unpublished data). Thus, if this hysteresis is a circadian event regulated by intrinsic cues, this could be a source of variability in the 24 hour confined experiments. However, it is less likely to occur in the 3 hour roaming experiments since they were all carried out during the day. Other physiological situations that may require changes in heart rate without changing metabolism or 0 2 pulse are digestion and stress. During digestion, vessels dilate around the gut allowing a local increase in blood flow to facilitate absorption. Heart rate may compensate in order to maintain blood pressure. In our animals there was little difference in variability surrounding the heart rate regressions of fasting and digesting lizards at 37°C, so this effect was probably minimal. Had the lizards eaten much more than 4 0 / 0 of their body weight, perhaps an increase in blood flow to the gut would have caused greater variation during the 37°C/digesting trials. Similarly, stress is also associated with changes in total peripheral resistance, and heart rate may act to offset these changes. Indeed, there was far more variability in the 24 hour confined dataset when lizards were presumed to be under more stress. Shunting Shunting of blood back to the systemic circulation may also be a source of the remaining variation in oxygen consumption. Tegu lizards have relatively good pressure separation in the ventricle and exhibit no left-right (L-R) shunt. However, they are capable of an intracardiac right-left (R-L) shunt from the cavum pulmonale to the left aorta during periods of intermittent breathing, i.e. during the apnea between episodes of breathing (Johansen et al., 1987). Recall Fick's equation for the delivery of oxygen to the tissues: 100 Vo. = / „ x V s ( C a o - C „ J (0 Because flow rate of blood to the tissues (Q_) is equal to heart rate multiplied by stroke volume, this equation can also be described as follows: v 0 i =6cao - 6cvo (8) If a R-L shunt exists such that deoxygenated blood is recirculated systemically, arterial O z levels will decrease. The amount of oxygen delivered in the arterial blood will be related to the O z content and flow rate of the blood passing through the lungs (C p u i o and Q_pul respectively) and the 0 2 content and flow rate of the blood bypassing the lungs (C s I l and Q_shuM respectively). Thus, Fick's equation in the presence of a R-L shunt is: (Note that when the R-L shunt is eliminated, Q_shunt equals zero so equation 9 is equal to equation 8.) The relationship between oxygen consumption and heart rate will not be constant if the degree of shunting varies (e.g. if Q_shuM increases and Q_pul decreases), because the oxygen delivery to the tissues will not be constant. The ratio of Q_shunt to Q_pul is determined not just by heart rate and stroke volume, but also by other variable factors such as resistance to flow through these different routes. Our lizards exhibited intermittent breathing during sleep and, to a lesser extent, when they were quiet. Therefore, if the tendency is to shunt during periods of apnea, as was observed by Johansen et al. (1987), shunting should only affect data collected when lizards were sleeping, or possibly quiet. This would be consistent 101 with models which suggest that a R-L shunt limits metabolic rate by decreasing arterial 0 2 levels (Wang & Hicks, 1996). It stands to reason that shunting would need to be reduced during more active states in order to meet the increased 0 2 demand. In fact, under normoxic conditions, reducing the degree of a R-L shunt should be more effective than increasing ventilation when increased O z delivery is required (e.g. during digestion or exercise) (Wang & Hicks, 1996). The upshot is that shunting could be relatively high during sleep and quiet states, and could be reduced or eliminated when tegus are alert, moving, and/or digesting. The combined effect of a) varying C>2 pulse, b) transient changes in blood pressure independent of oxygen consumption, and c) shunting during the "Sleep" and "Quiet" activity categories, may account for much of the variability in the data. Further studies would need to be performed in order to confirm this. Nevertheless, it is interesting to note that the Heart Rate Method may still be applied to this species regardless of such sources of variation. As long as variation in oxygen consumption is accommodated primarily by changes in heart rate, and secondarily by stroke volume and/or changes in the degree of shunt, the Heart Rate Method may still be the most practical predictor of metabolic rate. The correlation between breathing frequency and metabolic rate Since Fick's equation can also be described in terms of breathing frequency: Vo, = / R X V T 1 \ " F H J (5) 102 it seemed reasonable to consider whether this might be a better proxy for metabolic rate than heart rate. The chances of this being true were thought to be low even before the regression analyses were performed, because tidal volume is known to be quite variable at any given temperature and activity state (see Results in Chapter 2). Consequently, the term VT \Flo - F E q J (the equivalent of 0 2 pulse in this equation) was probably not constant. Nonetheless the regressions were carried out in order to verify this theory. As suspected, these regressions turned out to be less reliable than the heart rate regressions. Accordingly, while use of the correlation between breathing frequency and metabolic rate may not be as accurate as the correlation between heart rate and metabolic rate, it may be a useful alternative method. Summary and future directions for research This study confirms the potential for using heart rate as a proxy for metabolic rate in lizards. Efforts were made to exclude potential circadian changes from the calibration. Experiments were conducted in larger enclosures in order to reduce stress and to allow voluntary activity rather than forced activity (i.e. running on a treadmill). Different sample durations were tested to determine if there was an "ideal" sample duration for calibration purposes. Data were pooled in every combination possible for regression analysis. After all these efforts, there was still substantial within- and between-individual variability. However this variability was averaged out once all grouping factors were disregarded, and as much as 749/0 of the variability in oxygen consumption in these animals could be explained by changes in heart rate. The relationship in unstressed animals was best described by the following equation: lnsV0 =-3.441+0.679 ln/R . 103 The two different protocols generated extremely different results, with a maximum of only 41% of the variability in oxygen consumption being accounted for by heart rate during the 24 hour confined experiments. Thus 330 /0 of the variability in these experiments was eliminated by implementing the 3 hour roaming protocol. This difference was presumably due to differences in stress, and would suggest the need to conduct 24 hour roaming studies to obtain a more complete dataset, with the inclusion of sleep values. Any increase in variability in such studies may be partly attributable to the potential inclusion of circadian rhythms, rather than stress. 104 C H A P T E R 4 : C o n c l u s i o n In order to confirm that the calibration equation derived in Chapter 3 could be used in tegu lizards in general, validation experiments would need to be performed using a different population of lizards, preferably under conditions similar to those expected to be employed in future studies. However, plans for such validation studies were abandoned after inspecting the 24 hour confined dataset when it appeared that the variability in the relationship between heart rate and oxygen consumption was too high to be of predictive value. Although these validation experiments were not strictly carried out, the Heart Rate Method was briefly tested against the Time Energy Budget method using the existing data. First, the calibration equation derived in Chapter 3 (based on pooled and doubly-transformed data from the 3 hour roaming experiments, see Table 10) was used to predict total oxygen consumption for 3 hour data sets in two different animals (note that these are the same two animals used in Figure 15). Next, oxygen consumption was calculated using the energy budget from Chapter 2 (mean values of oxygen consumption per activity level and treatment are defined in Table 21 in the Appendix, and were multiplied by the time spent at each activity level). Finally, each of these two predicted values of oxygen consumption was compared with the total measured oxygen consumption for the 3 hour period. The accuracy of the Heart Rate Method was variable from one data set to the next; this is not surprising since the Heart Rate Method does not generally give accurate predictions for individual animals (Nolet et al., 1992). However, in most cases the predicted value generated by the Heart Rate Method was at least as accurate as that generated by the Time Energy Budget method (see Table 19). This is of particular interest since the time energy budget is the favoured method to predict metabolic rate when DLW is too expensive to carry out. It would seem that the Heart Rate Method could be used 105 to predict metabolic rate with the same accuracy, but with much less effort on the part of the investigator, particularly in light of the fact that neither temperature, nor activity state, nor digestive state need be known. Still, proper validation studies must be performed before the Heart Rate Method can be formally used. T a b l e 1 9 : A c o m p a r i s o n of the predic t ion a c c u r a c i e s of metabo l i c rate of two different me thods in two different ind iv idua ls . Fo r a n y g iven c o m p a r i s o n , the predic t ion va lue c l oses t to the ac tua l va lue of metabo l i c rate is ind ica ted in bo ld . No te that the two 37°C/d iges t ing t rea tments c o r r e s p o n d to those in F igure 15 . S e e text for deta i ls . Prediction accuracy (% difference of actual value) Tegu Treatment Time Energy Budget Heart Rate Method L B 17°C, Fasting -21.1 -4.1 L B 27°C, Fasting -16.4 6.4 L B 37°C, Fast ing 1.9 -14.0 L B 37°C, Digesting -11.6 -9.9 Pi 17°C, Fast ing -36.6 -15.0 Pi 27°C, Fast ing -41.8 -46.7 Pi 37°C, Fast ing -14.3 -11.1 Pi 37°C, Digesting -19.5 -17.7 It has been mentioned previously that the Heart Rate Method is best used as a predictor of mean population metabolic rate. This study is particularly instructive as an example of this principle. Although there is too much within- and between-individual variability to derive an accurate calibration equation for single animals, a sufficiently powerful model is obtained from the pooled data of several individuals. With this in mind, it is to be expected that this calibration equation could only be validated for population means. 106 R E F E R E N C E S Abe AS. 1983. Observations on dormancy in tegu lizards - Tupinambis teguixim (Repulia, Teiidae). Naturalia 8: 235-239. Abe AS. 1995. Estivation in South American amphibians and reptiles. Brazilian Journal of Medical and Biological Research 28: 1241-1247. Andersen JB, and Wang T. 2003. Cardiorespiratory effects of forced activity and digestion in toads. Physiological and Biochemical Zoology 76: 459-470. Andrade D, and Abe A. 1999a. Physiology of torpor in the tegu lizard. Comparative Biochemistry and Physiology - Part A: Molecular <& Integrative Physiology 124: S51. Andrade DV, and Abe AS. 1999b. Gas exchange and ventilation during dormancy in the tegu lizard Tupinambis merianae. The Journal of Experimental Biology 202: 3677-3685. Andrade DV, Brito SP, Toledo LF, and Abe AS. 2004a. Seasonal changes in blood oxygen transport and acid-base status in the tegu lizard, Tupinambis merianae. Respiratory Physiology & Neurobiology 140: 197-208. Andrade DV, Cruz-Neto AP, and Abe AS. 1997. Meal size and specific dynamic action in the rattlesnake Crotalus durissus (Serpentes: Viperidae). Herpetologica 53: 485-493. Andrade DV, Sanders C, Milsom WK, and Abe AS. 2004b. Overwintering in Tegu Lizards. In???, ed. Life in the Cold: Evolution, Mechanisms, Adaptation, and Application. Twelfth International Hibernation Symposium. Institute of Arctic Biology, University of Alaska Fairbanks, Alaska, USA: Biological Papers of the University of Alaska, 13-22. Bartholomew GA, and Tucker VA. 1963. Control of changes in body temperature, metabolism, and circulation by the agamid lizard, Amphibolurus barbatus. Physiological Zoology 36: 199-218. Bennett AF, and Dawson WR. 1976. Metabolism. In: Gans C and Dawson WR, eds. Biology of the Reptilia. London: Academic Press. 127-223. Bennett AF, and Hicks JW. 2001. Postprandial exercise: prioritization or additivity of the metabolic responses? Journal of Experimental Biology204: 2127-2132. 107 Bevan R, Woakes A, Butler P, and Boyd 1.1994. The use of heart rate to estimate oxygen consumption of free-ranging black-browed albatrosses, Diomedea melanophrys. Journal of Experimental Biology 193: 119-137. Bevan RM, and Buder PJ. 1992. The effects of temperature on the oxygen consumption, heart rate and deep body temperature during diving in the tufted duck Aythyafuligula. Journal of Experimental'Biology 163: 139-151. Bevan RM, Keijer E , and Butler PJ. 1992. A method for controlling the feeding behaviour of aquatic birds: Heart rate and oxygen consumption during dives of different duration. Journal of Experimental Biology 162: 91-106. Bevan RM, Speakman JR, and Butler PJ. 1995a. Daily energy expenditure of tufted ducks: A comparison between indirect calorimetry, doubly labelled water and heart rate. Functional Ecology 9: 40-47. Bevan RM, Woakes AJ, Butler PJ, and Croxall JP. 1995b. Heart rate and oxygen consumption of exercising gentoo penguins. Physiological Zoology 68: 855-877. Boutilier RG. 2001. Mechanisms of cell survival in hypoxia and hypothermia, journal of Experimental Biology 204: 3171-3181. Boyd IL, Bevan RM, Woakes AJ, and Butler PJ. 1999. Heart rate and behavior of fur seals: implications for measurement of field energetics. American Journal of Physiology: Pleart and Circulatory Physiology 276: H844-857. Burggren WW. 1987. Form and function in reptilian circulations. American Zoologist 27: 5-19. Buder P, Woakes A, Boyd I, and Kanatous S. 1992. Relationship between heart rate and oxygen consumption during steady-state swimming in California sea lions. Journal of Experimental Biology 170: 35-42. Butler PJ. 1993. To what extent can heart rate be used as an indicator of metabolic rate in free-living marine mammals. Symp Zool Socl^ond 66: 317-332. Butler PJ, Frappell PB, Wang T, and Wikelski M . 2002. The relationship between heart rate and rate of oxygen consumption in Galapagos marine iguanas LAmblyrhynchus cristatus) at two different temperatures. Journal of Experimental Biology 205: 1917-1924. Butler PJ, Woakes AJ, Bevan RM, and Stephenson R. 2000. Heart rate and rate of oxygen consumption during flight of the barnacle goose, Branta leucopsis. Comparative Biochemistry and Physiology - Part A: Molecular <& Integrative Physiology 126: 379-385. 108 Clark T D , Butler PJ, and Frappell PB. 2003. The relationship between body temperature, heart rate and rate of oxygen consumption in Rosenberg's goanna (Varanus rosenbergi) at various levels of activity is affected by time in captivity (abstract). Comparative Biochemistry and P/ysiology - Part A: Molecular & Integrative Physiology 134: S22. de Souza SCR, de Carvalho JE, Abe AS, Bicudo JEPW, and Bianconcini MSC. 2004. Seasonal metabolic depression, substrate utilisation and changes in scaling patterns during the first year cycle of tegu lizards (Tupinambis merianae). Journal of Experimental Biology 207: 307-318. Fitzgerald LA, Cook JA, and Aquino AL. 1999. Molecular phylogenetics and conservation of Tupinambis (Sauria: Teiidae). Copeia 1999: 894-905. Frappell PB, and Daniels CB. 1991. Ventilation and oxygen consumption in agamid lizards. Physiological Zoology 64: 985-1001. Froget G, Butler P, Handrich Y, and Woakes A. 2001. Heart rate as an indicator of oxygen consumption: influence of body condition in the king penguin. Journal of Experimental Biology 204: 2133-2144. Froget G, Handrich Y, Maho YL, Rouanet J-L, Woakes A, and Butler P. 2002. The heart rate/oxygen consumption relationship during cold exposure of the king penguin: a comparison with that during exercise. Journal of Experimental Biology 205: 2511-2517. Glass M L , Wood SC, and Johansen K. 1978. The application of pneumotachography on small unrestrained animals. Comparative Biochemistry &Physiology 59A: 425-427. Green J, Buder P, Woakes A, Boyd I, and Holder R. 2001. Heart rate and rate of oxygen consumption of exercising macaroni penguins. Journal of Experimental Biology 204: 673-684. Grigg GC, and Seebacher F. 1999. Field test of a paradigm: hysteresis of heart rate in thermoregulation by a free-ranging lizard (Pogona barbatd). Proceedings: Biological Sdences 266: 1291-1297. Hawkins P, Butler P, Woakes A, and Speakman J. 2000. Estimation of the rate of oxygen consumption of the common eider duck (Somateria mollissimd), with some measurements of heart rate during voluntary dives. Journal of Experimental Biology 203: 2819-2832. Hicks JW. 2002. The physiological and evolutionary significance of cardiovascular shunting patterns in reptiles. News in Physiological Sdences 17: 241-245. Hicks JW, and Wang T. 1996. Functional role af cardiac shunts in reptiles. The Journal of Experimental Zoology 275: 204-216. 109 Johansen K, Abe AS, and Andresen JH. 1987. Intracardiac shunting revealed by angiocardiography in the lizard Tupinamhis teguixin. Journal of Experimental Biology 130: 1-12. Klein W, Andrade DV, Abe AS, and Perry SF. 2003. Role of the post-hepatic septum on breathing during locomotion in Tupinamhis merianae (Reptilia: Teiidae). Journal of Experimental Biology 206: 2135-2143. Klein W, Perry SF, Abe AS, and Andrade DV. 2005, In review. Metabolic response to feeding in Tupinamhis merianae. Circadian rhythm and a possible respiratory constraint. Lefrancois C, Claireaux G, and Lagardere J-P. 1998. Fleart rate telemetry to study environmental influences on fish metabolic expenditure. Hydrohiologia 371/372: 215-224. Lilrywhite HB, Zippel KC, and Farrell AP. 1999. Resting and maximal heart rates in ectotherrruc vertebrates. Comparative Biochemistry and Physiology - Part A: Molecular & Integrative Physiology 124: 369-382. McNab BK. 1999. On the comparative ecological and evolutionary significance of total and mass-specific rates of metabolism. Physiological and Biochemical Zoology 72: 642-644. McPhee JM, Rosen DAS, Andrews RD, and Trites AW. 2003. Predicting metabolic rate from heart rate in juvenile Steller sea lions Eumetopiasjubatus. Journal of Experimental Biology 206: 1941-1951. Milsom WK, Andrade DV, Brito SP, Toledo LF, Wang T, and Abe AS. 2005, Submitted. Seasonal changes in daily metabolic patterns of tegu lizards (Tupinamhis merianae) placed in the cold (17°C) and dark. Physiological and Biochemical Zoology. Milstead WW. 1961. Notes on Teiid lizards in southern Brazil. Copeia 1961: 493-495. Nilssen KJ, Johnsen HK, Rognmo A, and Blix AS. 1984. Heart rate and energy expenditure in resting and running Svalbard and Norwegian reindeer. American journal of Physiology -Regulatory, Integrative and Comparative Physiology. R963-967. Nolet BA, Buder PJ, Masman D, and Woakes AJ. 1992. Estimation of daily energy expenditure from heart rate and doubly labeled water in exercising geese. Physiological Zoology65: 1188-1216. Packard GC, and Boardman TJ. 1999. The use of percentages and size-specific indices to normalize physiological data for variation in body size: wasted time, wasted effort? Comparative Biochemistry and Physiology - Part A: Molecular <&'Integrative Physiology 122: 37-44. 110 Palacios R, Villalba R, and Fitzgerald LA. 1997. Tupinambis merianae (common tegu): Predation. Herpetologicai'Review 28: 204-205. Skovgaard N , and Wang T. 2004. Cost of ventilation and effect of digestive state on the ventilatory response of the tegu lizard. Respiratory Physiology <& Neurobiology 141: 85-97. Storey KB, and Storey JM. 1990. Metabolic Rate Depression and Biochemical Adaptation in Anaerobiosis, Hibernation and Estivation. Quarterly Review of Biology 65: 145-174. Storey KB, and Storey JM. 2004. Metabolic rate depression in animals: transcriptional and translational controls. BiologicaiReviews 79: 207-233. Thorarensen H , Gallaugher PE, and Farrell AP. 1996. The limitations of heart rate as a predictor of metabolic rate in fish. Journal of Fish Biology 49: 226-236. Toledo LF, Brito SP, Milsom WK, Abe AS, and Andrade DV. 2005, Submitted. Effects of changes in season, temperature and body mass on the standard metabolic rate of tegu lizards {Tupinambis merianae). van Sluys M , and Rocha CFD. 1999. Tupinambis merianae (common tegu): Activity. Herpetological Review 30:42-43. Wang T, Altimiras J, Klein W, and Axelsson M . 2003. Ventricular haemodynamics in Python molurur. separation of pulmonary and systemic pressures. Journal of Experimental Biology 206: 4241-4245. Wang T, and Hicks J. 1996. The interaction of pulmonary ventilation and the right-left shunt on arterial oxygen levels. Journal oj Experimental Biology 199: 2121-2129. Wang T, Krosniunas E H , and Hicks JW. 1997. The role of cardiac shunts in the regulation of arterial blood gases. American Zoologist il: 12-22. Wang T, Warburton S, Abe A, and Taylor T. 2001. Vagal control of heart rate and cardiac shunts in reptiles: Relation to metabolic state. Experimental Physiology 86: 777-784. Wang T, and Warburton SJ. 1995. Breathing partem and cost of ventilation in the American alligator. Respiration Physiology 102: 29-37. Withers PC. 1977. Measurement of V D 2 , V C 0 2 , and evaporative water loss with a flow-through mask. Journal of Applied Physiology 42: 120-123. I l l Wood SC, Glass ML, Andersen NA, and Heisler N. 1987. Respiratory properties of blood and arterial blood gases in the tegu lizard: effects of temperature and hypercapnia. Experimental Biology 47: 27-31. Zar JH. 1999. BiostatisticalAnalysis. Prentice-Hall Inc., Upper Saddle River, NJ. 112 APPENDIX T a b l e 20: M e a n s of raw da ta a n d de r i ved va r iab les for 24 hour con f i ned expe r imen ts Variable Activity 17°C 27°C 37°C 37°C+food Mass-specific 0 2 consumption (ml/min/kg) Sleep Quiet Alert Moving 0.27 ± 0.01 0.39 ± 0.04 0.71 ± 0.15 0.58 ± 0.15 0.41 ± 0.04 0.53 ± 0.10 0.83 + 0.12 0.96 + 0.18 0.50 ± 0.02 0.55 ± 0.03 0.73 ± 0.04 0.69 ± 0.05 0.63 ± 0.04 0.67 ± 0.04 0.92 ± 0.12 0.74 ± 0.06 Mass-specific C 0 2 production (ml/min/kg) Sleep Quiet Alert Moving 0.19 ± 0.02 0.28 ± 0.02 0.60 ± 0.11 0.55 + 0.10 0.27 ± 0.01 0.38 ± 0.06 0.67 ± 0.10 0.83 ± 0.17 0.35 ± 0.02 0.41 ± 0.03 0.59 ± 0.05 0.69 ± 0.06 0.46 ± 0.02 0.45 ± 0.02 0.71 ± 0.08 0.74 ± 0.09 RER Sleep Quiet Alert Moving 0.71 ± 0.03 0.76 ± 0.06 0.90 ± 0.07 1.04 ± 0.07 0.67 ± 0.04 0.76 + 0.04 0.86 ± 0.04 0.88 ± 0.08 0.70 ± 0.03 0.76 ± 0.03 0.82 ± 0.05 1.01 ± 0.05 0.74 + 0.02 0.69 ± 0.02 0.80 ± 0.04 1.00 ± 0.05 Heart Rate (beats/min) Sleep Quiet Alert Moving 9.1 + 1.2 12.5 ± 1.2 23.3 ± 4.9 29.1 ± 4.1 14.9 ± 1.5 19.8 ± 2.2 37.8 + 5.1 56.5 ± 3.7 25.5 + 4.4 28.2 ± 4.1 41.5 ± 6.4 55.3 ± 9.6 31.7 ± 2.9 34.6 ± 3.5 56.7 ± 8.6 73.1 ± 5.9 s 0 2 pulse (ml/beat/kg) Sleep Quiet Alert Moving 0.032 ± 0.003 0.032 ± 0.004 0.035 ± 0.009 0.023 + 0.006 0.030 ± 0.004 0.028 ± 0.004 0.023 ± 0.004 0.017 ± 0.003 0.022 ± 0.003 0.023 ± 0.004 0.022 ± 0.004 0.013 ± 0.002 0.021 ± 0.003 0.022 ± 0.003 0.020 ± 0.006 0.011 ± 0.002 Tidal Volume (ml) Sleep Quiet Alert Moving 15.3 ± 2.2 13.6 ± 2.9 20.8 ± 6.4 16.8 ± 3.0 20.0 ± 2.4 14.4 ± 2.1 20.9 ± 5.6 18.4 ± 3.3 23.0 ± 3.8 17.1 ± 2.4 19.8 ± 5.0 19.8 ± 5.8 22.5 ± 3.7 19.0 ± 2.6 24.7 + 5.9 16.7 ± 3.0 Breathing Frequency (breaths/min) Sleep Quiet Alert Moving 2.9 ± 1.1 11.8 ± 6.4 20.3 ± 3.5 47.6 ± 5.9 2.5 ± 0.4 7.4 ± 1.5 16.2 ± 4.1 48.6 ± 3.9 3.7 ± 1.1 7.3 ± 0.7 16.4 + 2.1 43.4 ± 6.9 4.2 ± 1.0 7.5 ± 1.6 16.9 ± 2.9 50.9 ± 3.2 % 0 2 extracted (%) Sleep Quiet Alert Moving 5.4 ± 0.6 2.7 ± 0.5 1.6 ± 0.4 0.5 ± 0.1 6.3 ± 1.9 3.6 ± 0.8 2.2 ± 0.6 0.5 ± 0.1 4.8 ± 1.1 3.5 ± 0.9 2.1 ± 0.6 0.6 ± 0.1 5.1 ± 1.7 4.2 ± 1.2 2.8 ± 1.4 0.6 ± 0.1 Ventilation (ml/min) Sleep Quiet Alert Moving 34 ± 5 82+7 314 ± 78 728 ± 107 44 ± 8 87 ± 16 326 ± 96 842 ± 121 74 ± 20 127 ± 24 321 ±111 777 ± 241 90 ± 23 138 ± 37 471 ± 167 829 ± 146 Air Convection Sleep Quiet Alert Moving 120 ± 12 235 + 36 523 ± 121 1589 ± 367 126 ± 30 193 ± 44 461 ± 161 1019 ± 169 150 ± 40 236 ± 47 486 ± 200 1159 ± 362 149 ± 34 235 ± 79 653 + 256 1227 ± 279 Rate of 0 2 delivery to the lung (ml/min/kg) Sleep Quiet Alert Moving 7 ± 1 17 ± 2 66 ± 16 153 ± 22 9 ± 2 18 ± 3 68 ± 20 177 ± 25 16 ± 4 27 ± 5 67 ± 23 163 ± 51 19 ± 5 29 ± 8 99 ± 35 174 ± 31 113 Table 21: M e a n s of raw da ta a n d de r i ved va r i ab les for 3 hour roaming expe r imen ts Variable Activity 17°C 27°C 37"C 37°C+food Mass-specific Sleep 0 2 consumption Quiet (ml/min/kg) Alert Moving 0.08 ± 0.01 0.12 ± 0.02 0.22 ± 0.03 0.25 ± 0.03 N/A 0.23 ± 0.05 0.27 ± 0.05 0.48 ± 0.08 N/A 0.48 + 0.03 0.54 + 0.04 0.63 ± 0.06 N/A 0.40 ± 0.03 0.52 ± 0.06 0.66 ± 0.07 Mass-specific Sleep C 0 2 production Quiet (ml/min/kg) Alert Moving 0.05 ± 0.01 0.10 ± 0.02 0.18 ± 0.02 0.28 ± 0.03 N/A 0.20 ± 0.04 0.28 ± 0.05 0.47 ± 0.06 N/A 0.34 ± 0.04 0.41 ± 0.03 0.57 ± 0.04 N/A 0.29 ± 0.03 0.40 ± 0.05 0.56 ± 0.05 RER Sleep Quiet Alert Moving 0.68 ± 0 . 1 0 0.83 ± 0.03 0.87 ± 0.03 1.14 ± 0.07 N/A 0.90 + 0.16 1.08 ± 0.16 1.03 ± 0.06 N/A 0.70 ± 0.06 0.77 ± 0.01 0.92 + 0.06 N/A 0.74 + 0.03 0.77 ± 0.01 0.86 ± 0.04 Heart Rate Sleep (beats/min) Quiet Alert Moving 6.2 ± 0.4 8.2 ± 1.0 11.6 ± 1.8 18.4 ± 1.8 N/A 19.3 ± 4.6 27.3 ± 4.6 60.3 ± 3.8 N/A 40.5 ± 5.6 49.7 ± 3.6 87.4 ± 5.3 N/A 45 .3 ± 4.7 56.2 ± 6.0 89.4 ± 7.1 s 0 2 pulse Sleep (ml/beat/kg) Quiet Alert Moving 0.012 ± 0.002 0.015 ± 0.002 0.020 ± 0.002 0.014 ± 0.001 N/A 0.014 ± 0.004 0.012 ± 0.003 0.008 ± 0.001 N/A 0.013 ± 0.001 0.011 ± 0.001 0.007 + 0.000 N/A 0.010 ± 0.001 0.010 ± 0.001 0.007 ± 0.001 Tidal Volume Sleep (ml) Quiet Alert Moving 19.2 ± 3.6 16.8 ± 2.0 18.0 ± 1.6 15.3 ± 1.3 N/A 13.6 ± 0.5 17.4 ± 4.3 20.4 ± 2.3 N/A 12.2 ± 1.4 12.9 ± 1.9 17.2 ± 1.4 N/A 13.6 ± 1.5 12.2 ± 2.5 16.0 ± 2.5 Breathing Sleep Frequency Quiet (breaths/min) Alert Moving 1.1 ± 0.1 2.6 ± 0.2 7.8 ± 1.6 25.9 ± 1.1 N/A 7.0 ± 1.6 13.4 ± 1.1 48.9 ± 3.1 N/A 16.6 ± 2.8 26.5 ± 1.8 51.9 ± 3.4 N/A 17.6 ± 6.2 29.6 ± 4.2 56.5 ± 5.2 % 0 2 extracted Sleep (%) Quiet Alert Moving 2.1 ± 0.6 1.7 ± 0.4 1.2 ± 0.3 0.3 ± 0.1 N/A 1.5 ± 0.5 0.8 ± 0.3 0.3 ± 0.1 N/A 1.5 ± 0.3 0.9 ± 0.1 0.4 ± 0.1 N/A 1.3 ± 0.3 0.9 ± 0.2 0.4 ± 0.1 Ventilation Sleep (ml/min) Quiet Alert Moving 20 ± 4 4 4 + 7 144 ± 34 399 ± 43 N/A 96 ± 25 242 ± 78 1008 ± 158 N/A 200 ± 46 327 ± 39 878 ± 77 N/A 204 ± 48 319 ± 46 866 ± 116 Air Convection Sleep Quiet Alert Moving 294 ± 71 394 ± 72 793 ± 238 1857 ± 497 N/A 478 ± 148 1058 ± 408 2355 ± 392 N/A 398 ± 68 640 ± 107 1495 ± 226 N/A 540 ± 137 709 ± 188 1422 ± 286 Rate of 0 2 delivery Sleep to the lung Quiet (ml/min/kg) Alert Moving 4 ± 1 9 ± 1 30 ± 7 84 ± 9 N/A 20 ± 5 51 ± 16 212 ± 33 N/A 42 ± 10 69 ± 8 184 ± 16 N/A 4 3 + 10 67 ± 10 182 ± 24 114 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0092147/manifest

Comment

Related Items