Ficha Catalográfica

The central theme of this thesis is the concept of behavioral fever, which has been defined as the increase of body temperature by effecting the change in preferred temperatures due to the recognition by the body of an infection or pathogen. The thesis is composed of three parts. In the General Introduction, behavioral fever is presented around seven fundamental points to the understanding of this response within Ecophysiology, starting with the definition and the laboratory and field research until now, to the ecological limitations of individuals and the implications of this theme in conservation. In the first chapter, with scientific text format, we present the research in which we studied the behavior and thermal preference of Proceratophrys boiei species under experimental conditions in individuals injected with lipopolysaccharides (LPS), to simulate an infection, and in intact individuals (injected with Saline, a control group). In this research we considered two alternatives of behavioral responses, as discussed in the General Introduction: a) behavioral fever, which is characterized by an increase in the individuals body temperature by changing the preferred temperatures within a thermal landscape; B) patient behavior, which, in the context of experimental design, would be recognized by the decrease in the activity of individuals. Thus, we recorded the following treatments for 24 hours with a thermographic camera: 1) intact individuals in the thermal gradient switched off, 2) intact individuals in the connected thermal gradient, 3) individuals injected with saline, in the bound thermal gradient 4) individuals injected with LPS in the thermal gradient on. For each of the treatments it was recorded the locomotion distance and the thermal preferences, along with other details of the behavior and the thermal preferences. From our results, we conclude that the individuals of P. boiei present a patient's behavior as a dominant response when injected with LPS and their thermal preferences are a consequence of patient behavior and not behavioral thermoregulation. Finally, the general discussion explains how chapter 1 contributes to the discussion of each of the seven points highlighted in the general introduction attempting to propose a complete methodology and studies to maintain the dialogue between the physiology and the ecology of individuals in the context of infection and disease. Abstract Behavioral fever in infected individuals is the increase of body temperature mediated by modified thermal preferences, when the selection of higher temperatures is possible. In ectothermic vertebrates, behavioral fever is often studied through an injection of endotoxins of gram-negative bacteria, followed by protocols registering body temperatures in a thermal gradient. Usually, the results of such tests are compared with those of a control (e.g. Saline injections) to determine body temperatures under “normal” and “febrile” situations. This technique necessarily involves ambiguity in determining fever temperatures, even more given the limits of information on tropical anuran lineages. Within tropical anurans, species from open environments (and presumably naturally exposed to diverse thermal environments) have received most attention, whereas forest species have been neglected. Hence herein we tested the hypothesis that individuals of the species Proceratophrys boiei (Odontophrynidae) from the Atlantic forest of Brazil, respond behaviorally to simulated infection. There are two possible response: fever and sick behavior (characterized by reduced mobility and low thermal variance). To test this hypothesis, P. boiei’s individuals were placed in a thermal gradient and their body temperature (assumed to be preferred) was recorded for 24h in each experiment. To simulate an infection we injected systemically lipopolysaccharide (LPS) injection with a dose of 2 mg/kg in P. boiei’s individuals. Our results suggest that individual P. boiei reduce activity after LPS injections, and that they prefer the extremes of a gradient (perhaps as a refuge), with no thermal preferences. Thus, sickness behavior is a dominant response, and given the overall results, the presence of individuals in the hot extreme of the gradient, when evident, resulted from inactivity and not from thermoregulation. fever response. This may be a trend in other anuran species, possibly more common in forest forms.


Abstract
The central theme of this thesis is the concept of behavioral fever, which has been defined as the increase of body temperature by effecting the change in preferred temperatures due to the recognition by the body of an infection or pathogen. The thesis is composed of three parts. In the General Introduction, behavioral fever is presented around seven fundamental points to the understanding of this response within   (Kluger, 1991;Bicego et al., 2000, Guyton andHall, 2006;.

Abstract
Behavioral fever in infected individuals is the increase of body temperature mediated by modified thermal preferences, when the selection of higher temperatures is possible.
In ectothermic vertebrates, behavioral fever is often studied through an injection of endotoxins of gram-negative bacteria, followed by protocols registering body temperatures in a thermal gradient. Usually, the results of such tests are compared with those of a control (e.g. Saline injections) to determine body temperatures under "normal" and "febrile" situations. This technique necessarily involves ambiguity in determining fever temperatures, even more given the limits of information on tropical anuran lineages. Within tropical anurans, species from open environments (and presumably naturally exposed to diverse thermal environments) have received most attention, whereas forest species have been neglected. Hence herein we tested the hypothesis that individuals of the species Proceratophrys boiei (Odontophrynidae) from the Atlantic forest of Brazil, respond behaviorally to simulated infection. There are two possible response: fever and sick behavior (characterized by reduced mobility and low thermal variance). To test this hypothesis, P. boiei's individuals were placed in a thermal gradient and their body temperature (assumed to be preferred) was recorded for 24h in each experiment. To simulate an infection we injected systemically lipopolysaccharide (LPS) injection with a dose of 2 mg/kg in P. boiei's individuals.
Our results suggest that individual P. boiei reduce activity after LPS injections, and that they prefer the extremes of a gradient (perhaps as a refuge), with no thermal preferences. Thus, sickness behavior is a dominant response, and given the overall results, the presence of individuals in the hot extreme of the gradient, when evident, resulted from inactivity and not from thermoregulation. A diverse thermal landscape theoretically allowing for thermoregulation was not sufficient to elicit a behavioral fever response. This may be a trend in other anuran species, possibly more common in forest forms.

Introduction
Fever is the regulated increase of body temperature associated with a change in thermoregulatory set point (Kluger, 1991). The thermoregulatory activity during a fever is driven either by metabolic physiology or by behavior. In the case of physiologically-driven fever, relevant processes vary across lineages, but generally encompass increased metabolic and heart rate and reduced loss of body heat, mediated by signaling cascades involving thyroid hormones . In contrast, the behavioral course for fever exploits the thermal diversity of the environment (if available) under modified patterns of neural physiology. During pathogen infection, real or simulated by an endotoxin injection, several ectothermic animal species increase body temperature by means of such modified behavior, which is called behavioral fever Kluger, 1977;Bicego et al., 2000;Bicego and Branco, 2002;Demas and Nelson, 2011;Bolltaña et al., 2013). Behavioral fever can be elicited under experimental conditions that make thermal gradients available for treated animals.
Under such circumstances, behavioral fever is defined as the selection of higher than normal temperatures, mediated by shifts in temperature preferences (Firth et al., 1980;Muchlinski, 1985).
Interpreting fever as a behavioral trait is not obvious because preferred temperatures may be naturally variable (Brattstrom 1979;Lillywhite, 1970;Navas, 1997;Sanabria et al., 2012). This is of importance for a body temperature "higher than normal" makes sense only when the "normal" temperature is unambiguously defined. This is a complex issue, though, dominant behavioral inclinations in the field may vary for many reasons, and thermal preferences do vary in nature in the absence of infection and due to shifts in physiological states. Anuran body temperature is influenced by time of year, photoperiod and time of day, and temperature and humidity of the environment (Schwarzkopf & Alford, 1996;Seebacher et al., 2002). However, whereas some amphibian species thermoregulate behaviorally (Brattstrom, 1963) and exploit thermal resources in their microhabitat (Tracy & Christian, 1986), others thermoconform (Iturra-Cid et al., 2014;Rodriguez et al., 2016), prioritize hydroregulation (Cruz & Galindo, 2017;Navas et al., 2007) or even develop hypothermia under dry conditions. Additional complications are that thermoregulation in the laboratory not necessarily reflects field thermal ecology (Feder, 1982), and that thermal preferences may vary substantially. In experimental systems, animals with no manipulation at all are very rarely explored, so that punctured controls are considered as a normal thermoregulation pattern, or that such pattern must come from the literature. In the latter case, studies may differ in the thermoregulatory responses elicited (Sherman et al., 1991).
Behavioral fever was originally reported for amphibians under experimental frameworks. Early treatments involved pathogens (Gram-negative bacteria, endotoxin LPS, fungus, etc.) or pyrogenic agents (e.g. prostaglandins, PGE, a known activator of fever) and some of them elicited behavioral fever (Hutchinson & Erskine;1981).
Studies also focused on thermoregulatory responses to drugs (Capsaicin, prostaglandin E1, melatonin, and chlorpromazine), that is believed to stimulate regulation of temperature through behavioral responses (Hutchison, 1981;Hutchison & Erskine, 1981;Hutchison & Spriestersbach, 1986). Salamanders were the target of groundbreaking studies that demonstrated, for example, that Necturus maculosus increases body temperature (behavioral fever) when injected with prostaglandin E1 into the third ventricle of the brain, although nowadays anurans receive most attention regarding behavioral fever (Bicego & Branco, 2002 see Table 1). However, fever is not the only possible response to infection in animals, and anurans are no exception.
Sickness behavior is technically not mutually exclusive with fever, and is a syndromic response to pathogenic states usually characterized by lethargy and anorexia (Hart, 1988;Larson & Dunn, 2001;Johnson, 2002). In ectothermic tetrapods this behavior has received less attention than it deserves, for behavioral fever requires active thermoregulation. Therefore, the activity involved in behavioral thermoregulation may indeed be mutually exclusive with lethargy, so that behavioral fever and sickness behavior are unlikely to be dominant trends simultaneously. It is of particular significance that sickness behavior and not behavioral fever may dictate patterns of activity in infected individuals of some anuran species, even those known to thermoregulate in thermal gradients.
A key example of complex responses occurs in the toad Rhinella marina, a species able of multifaceted regulation of body temperature through behavior (Malvin &Wood, 1991, Sievert, 1991). Yet, despite this ability, individuals injected with lipopolysaccharide (LPS) may either display or not a thermoregulation response, apparently via perception of the thermal environment. Therefore, the observed thermoregulatory response to pyrogens depends on the initial position of toads in a thermal gradient, and animals may abandon thermoregulation and displaying sickness behavior as a dominant response (Llewellyn et al., 2011). Consequently, behavioral fever may help individual amphibians against infection, but needs not to be a universal response, and dominant responses may be influenced by the thermal settings of environments. This is a key issue linking amphibian disease and conservation, for emergent disease is a putative cause of amphibian extinction in numerous contexts (e.g. Batracochytrium dendrobatidis, Bd, Woodhams et al., 2003;Berger et al., 2004;Pitrowski et al., 2004;Forrest et al., 2011;Karavlan & Venesky, 2016 A corollary of the previous question is that sickness behavior, if evident, would manifest in reduced activity of infected frogs. If shifts in activity do occur after infection we ask 2) whether these shifts are affected by the structure of thermal landscapes. Our expectation was that diverse thermal landscapes could favor febrile responses whereas flatter thermal landscapes would favor sickness behavior.

Species information
Proceratophrys boiei is an anuran amphibian species endemic to the Atlantic Forest of Brazil, occurring from the state of Espírito Santo to the state of Santa Catarina and can inhabit the Cerrado transactional areas above 1200m in the states of Minas Gerais, São Paulo, and Rio de Janeiro. This species has cryptozoic habits, and when inactive remains hidden in excavated or natural cavities in the soil (Giaretta et al., 1999). They are regarded as nocturnal (Haddad et al., 2013), but can be active at the end of the afternoon (17h) and during the day. The reproductive season runs from September to January, with a higher presence of males vocalizing at the end of the rainy season (Bertoluci, 1998;Haddad et al., 2013). The males call mostly at night (Pombal, 1997) but also in the morning (CC pers. obs) on the ground near small streams or lakes (Bertolucci & Rodrigues, 2002). Females lay eggs in swamps or in the stream (AmphibiaWeb, 2015). Tadpoles are benthic (Izecksohn et al. 1979), and metamorphs are more commonly seen in February . Typical body size is 40-62 mm SVL in adult males and 40-74 mm SVL in females and the mean body mass is 9.53 g AmphibiaWeb, 2015).

Maintenance of Individuals in captivity
Frogs were kept in groups of two to five individuals and maintained in glass terrariums provided with moist litter for hiding. Filtered water was supplied and changed twice a week. Frogs were fed cockroaches ad libitum. Maintenance temperature was influenced by environmental temperature, under a 12-hour photoperiod.

Measurement of body temperature in the system (preferred Tb)
To evaluate body temperature in the laboratory we used a thermal system composed by a water bath, with the aluminum surface of 1.10 m x 0.30 m x 0.50 m. We established the extreme temperatures as 5°C and 35°C, which did not harm the animals, for some frogs may die when approaching drastic gradient extremes (CA Navas pers. obs). The soil of the system was covered with Petri dishes full of water and by a 5 mm layer of vermiculite to offer proper refuge to the animals as recommended in other studies (Spotila, 1982). This layer was also important for lethargic behaviors to be elicited. We We are aware that captivity itself can be a confounding variable, and designed a reciprocal treatment control as to maximize the amount of information collected from the number of animals available. For control, we split all animals into two groups, one to be injected with LPS and another with saline solution (early season) and measured thermal parameters. After two weeks, the converse treatments were performed, that is, each group received the reciprocal treatment. Individuals were fasted 48 hours before the experiments to avoid any influence of feeding on thermoregulation.
It turned out that preliminary analyses pointed out to subtle yet relevant differences in the results among experiments performed in April and May, so we decided not to pool this data. It must be clear that animals were studied (but not collected) in different seasons, and the differences here highlighted apply to two phases of the same set of animals measured at different time of year. For example, in our reciprocal design, Late Season individuals injected with LPS, in Early Season received a saline injection, and vice versa (see details below). The groups generated are hereafter referred to as Early Season (1) and Late Season (2) frogs. We name these groups after season for this species has proven very resilient to captivity, but the timing of captivity in this experiment encompassed the most relevant climatic seasonal transition in nature, which occurs from autumn to early winter (Databank, hydro-meteorological -Climatological station at Parque Estadual Intervales). In addition, we had learned from field observations that this frog shifts behavioral patterns between the reproductive (males call near water) and post-reproductive seasons (individuals found far from water sources), and therefore field behavior changes from April to May.
We quantified behavior on a flat thermal landscape (thermal gradient off, room temperature, Early season only) and a complex thermal landscape (thermal gradient on, both seasons). Under these two conditions, we compared intact individuals, individuals injected with saline and individuals injected with LPS. We did not measure actual distances moved to limit disturbance to animals, but deduced activity from changes in position along the gradient, which was split in five areas 1 (warm) to 5 (cold), so that area 3 was in the middle of the gradient. We quantified distances moved based on the location of individual frogs along the image sequences during the procedure. Each treatment started in Area 3 at 18h and then frogs were left free to move during 24 hours.
We thermographed the gradient every 10 minutes during those 24 hours (in previous testing, we showed that individual frogs could be observed from the substrate in thermographs). We produced various measures of central tendency and dispersion of body temperature and quantified movement with what we called "time series" variables. These variables were associated to changes across time and included both movement (central tendency and dispersion) and direction of shifts in body temperature (increase, decrease or none). Details of this pool of variables are available in Table 2.

Calculations and statistical analysis
To detect eventual preferences for any position in the gradient (each of the five areas) we contrasted frequencies of location in a flat thermal landscape (room temperature) against a uniform distribution (equal frequencies among areas, i.e., no preferences) using a Chi-square test (Zar, 1998).
We combined all data on body temperature (central tendency and dispersion) and activity with Principal Component Analyses (PCA) composed in all cases by the 13 variables related in Table 2. These PCAs were performed emphasizing the best possible data for each experimental contrast; given that the data set had gaps (e.g. Early season animals only) (see Table 3). The conditions and amount of data differed slightly across treatments, so that a global analysis was not possible. These PCA´s differed in details, but globally the three main components displayed highest loads for similar arrangements of variables (example in Table 4). In all cases the data generated a first To evaluate if there was a decrease or increase in distance moved that is associated with sickness behavior and behavioral fever, we contrasted treatments with main effects ANOVA over total distance moved.

Movement and exploration in a flat and complex thermal landscapes (Component 3)
Intact frogs moved more than any other experimental group (Fig. 3). Among

Frogs with LPS injection in a complex thermal landscape (L) in Early and Late Season
Early

Distance moved in the treatments
All groups differed in distance movement around the gradient (F (3, 0.05) = 26.208, p = 0.00001). Intact frogs in a flat thermal landscape moved more compared to intact frogs, saline injected frogs, and LPS injected frogs in a complex thermal landscape. Overall, frogs injected with saline and LPS moved less (Fig. 6).

Discussion
Most studies aiming behavioral responses to infection in amphibians focus on body temperature, especially both mean and variance data (Kluger, 1977;Bicego et al., 2002). In this article, we used an experimental design contrasting two alternative responses, fever and sick behavior, which conveys a more complex dimension of the problem. In the laboratory, a typical fever response involves moving in a gradient as to raise body temperature compared to indicators typical of non-infected animals Hunt et al., 2011). Under laboratory conditions, such response is unambiguously clear in some anuran species (Kluger, 1977;Bicego et al., 2002;), but does not need to be a universal. Indeed, we show that P.boiei injected with LPS decreases movement, and that the dominant response to infection is lethargy (but not prostration), not fever. LPS reduced movement and exploration, and to some degree thermal variance, with impact of time of year. Therefore, the exploratory behavior across thermal landscapes may vary with the climatic transitions that occur through the year in the area they inhabit.
Anyhow, we suppose that our results indicate more altered movement patterns (a correlate of thermal variance in complex thermal landscapes) and not an active thermoregulatory response in the strict sense. Overall, under our experimental conditions P. boiei did not behave as a thermoregulating species but as a thermoconformers, and fever was not a dominant behavioral response.
Given that behavioral fever in ectothermic tetrapods is seen as a state of altered thermoregulation, and that thermoregulation was not evident in our study, sickness behavior was the dominant pattern. This behavioral response has been considered as a first line defense (Hart, 1988) before other physiological responses of the immune system (Braga, 2013;Llewellyn et al., 2011). Fever and Sickness behavior are not mutually exclusive in endothermic tetrapods; even more for these two responses have the same activation pathways and mediators (e.g. cytokines ;Hart, 1988;Dantzer, 2001). In birds for example, after LPS injection, individuals became feverish and reduced activity being somnolent and increasing anorexia (Johnson et al., 1993).
However, in ectothermic tetrapod, like anurans, these responses may fail to manifest simultaneously, or may emerge only under some circumstances according to perception of thermal landscapes. Llewellyn and colleagues (2011) showed that toads reduce activity after simulated infection, apparently also responding with sickness behavior more than behavioral fever. In that study, shifts in mean temperature seem a consequence of reduced activity like in P. boiei. Indeed, in the experimental designed by Llewellyn and colleagues the body temperature of cane toads depends on the place of release, so that thermoregulation, if occurred at all, was an opportunistic response.
An important contrast between our study and that by Llewellyn and colleagues is that not infected cane toads are known to thermoregulate in thermal gradient (Malvin & Wood, 1991;Sievert, 1991) whereas P. boiei does not seem to do so under any tested circumstance. Rather, P. boiei explore and move around when not infected (e.g. flat and complex thermal landscape). Another contrast between cane toad and P. boiei is that the latter species inhabits transition and fully-grown forests (Haddad et al., 2013;Bertoluci & Rodrigues, 2002), thus the evolution of their thermoregulatory behavior may be influenced by lack of opportunity for thermoregulation, as in the case of some lizards (Ruibal, 1961).
The infection-induced lethargy observed by both Llewellyn and colleagues (2011) and us is characteristic of sickness behavior (Hart,1988;Inui, 2001), a syndrome encompassing inactivity, anorexia, and change in the diel cycle of body temperatures (Hart, 1988), which is common in vertebrates as part of the immune system response (Hart, 1988;Inui, 2001) and may concur with refuge seeking (Llewellyn et al., 2011).
We interpreted the preferences for the narrower borders of the system as a refugeseeking behavior emerging under our experimental conditions, and this is something to be tested formally in future studies. A similar behavioral response was reported in anuran tadpoles, which swim to gradient borders and remain there even if no established gradient exist or any other treatment, and there remain for several hours (the same experimental period) (Lucas & Reynolds, 1967).
A typical protocol in the study of fever is to use saline injection as a control for LPS injection (Casterlin & Reynolds, 1977;Bicego et al., 2002;Llewellyn et al., 2011).
Our results demonstrated that this may not be an adequate procedure is some cases as P. boiei, because the impact of the injection is perceivable in terms of activity and position, and regarding this aspect produces results closer than those with LPS treatment. This result of change in behavior because of saline injection points to a refuge-seeking behavior as mentioned above. Also, whereas controls to analyze behavior in the system are rare, they are possibly important. In the absence of such controls, it would have been difficult for us to tell apart preferences for a position within a system, and preferred temperatures, thus leading to misinterpreted results.
Although we report here dominant behavioral trends across the study, we must point out the high inter-individual variability that characterizes results across the study.
Such variability has implications. On one hand, responses across individuals are not homogeneous, and the lack of an obvious thermoregulatory trend likely enhanced observed thermal variances. On the other hand, under these conditions is becomes impossible to define a "normal" thermal situation to anchor a definition of behavioral fever. This observation, often overlooked, is not new. Kluger (1977) had noticed this for the anuran Hyla cinerea, and commented on the high variability in the body temperature present in both, control and treatment groups, pointing out that this could be a characteristic of amphibians perhaps related to alternative behavioral drives to regulate salt and water. Deen & Hutchinson (2001) reported a similar response on juveniles of the specie Iguana iguana with LPS injection displaying two different thermoregulatory responses, fever and hypothermia. The inter-individual variability of the two behavioral responses was apparently affected by acclimation temperature and magnitude of energetic reserves.
Regarding season, we emphasize again that the differences of the data collected at different times of year came as an unexpected result, and do not constitute a formal study of seasonality. However, as turned out, the data do support the idea that dominant behavioral responses to infection could vary according physiological differences by time of year. We do not know of studies testing these hypotheses, but it seems plausible for thermoregulatory patterns may change across the year in anurans (Sievert, 1991;Bicego et al., 2001;Noronha de Souza et al., 2016). Seasonality may be important because in nature seasons affect the average temperatures and the structure of thermal landscapes (Ortega & Navas, in preparation). Thus, individuals, due to ecological restrictions, physiological shifts, or both, may change thermal preferences including context disease (Schwaner, 1989). Finally, differences in fever response relate to the dose, site injection and kind of endotoxin applied (Kluger, 1991), so we cannot postulate that P. boiei will respond as here described under any sort of natural or experimental infection. However, we show unambiguously that behavioral fever may not be a dominant response under some circumstances that we interpret as compatible with ecology. On the same lines, behavioral responses to infection could vary with the time of day at which endotoxins are applied. In birds, for example, variation in the behavioral response is context-dependent and regulate according to the needs and time of the infection (Skold et al., 2015). In anurans, variation in preferred body temperatures occur across seasons (Bicego et al., 2001;Noronha de Souza et al., 2016).
More studies about this topic need to corroborate this hypothesis for anurans.

Conclusions
Proceratophrys boiei did not develop behavioral fever when LPS was injected. In individuals of P. boiei, LPS injection produces a change in behavior with a reduction in activity and decrease in body temperature variation, characteristics of sickness behavior. This response was dominant behavior in all individuals. Our study present sickness behavior as an alternative response for a tropical species giving us a better understanding of the potential of thermoregulatory and behavioral responses of forest species when infected.

Acknowledgments
We thank Jesus Ortega-Chinchilla for the technical support in the use of the equipment during the experiments, Jessica Citadini, Ananda Brito and Renata Vaz for helping to collect and maintain the animals during this project. To Renata Vaz for the comments and suggestions in the manuscript. This study was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).
SPOTILA, J.R. Role of temperature and water in the ecology of lungless salamanders.

Mean (°C)
Mean of the temperature measured for each individual in each treatment.

Median (°C)
Median of the temperature measured for each individual in each treatment.

Moda (°C)
Moda of the temperature measured for each individual in each treatment.

Variance
Variance of the temperature measured for each individual in each treatment.

Standard Deviation
The standard deviation of the temperature measured for each individual in each treatment.

Minimum (°C)
Minimum of the temperature measured for each individual in each treatment.

Maximum (°C)
Maximum of the temperature measured for each individual in each treatment.

Range (°C)
Minimum minus maximum value for temperature measured in each individual for each treatment.

Mean Change (°C)
Mean of the degrees change (difference between previous and next temperature registration) of the temperature measured for each individual in each treatment.

Variance Change
Variance of the mean change.

Mean Direction
Mean of the direction (0, did not change; 1 increase temperature; -1 decrease temperature) of the temperature measured for each individual in each treatment.

Mean Movement
Mean of the moves done between areas of the system for each individual in each treatment.

Variance Movement
Variance of the mean movement for each individual in each treatment.