Growth, stress, and acclimation responses to fluctuating temperatures in field and domesticated populations of Manduca sexta

Abstract Diurnal fluctuations in temperature are ubiquitous in terrestrial environments, and insects and other ectotherms have evolved to tolerate or acclimate to such fluctuations. Few studies have examined whether ectotherms acclimate to diurnal temperature fluctuations, or how natural and domesticated populations differ in their responses to diurnal fluctuations. We examine how diurnally fluctuating temperatures during development affect growth, acclimation, and stress responses for two populations of Manduca sexta: a field population that typically experiences wide variation in mean and fluctuations in temperature, and a laboratory population that has been domesticated in nearly constant temperatures for more than 300 generations. Laboratory experiments showed that diurnal fluctuations throughout larval development reduced pupal mass for the laboratory but not the field population. The differing effects of diurnal fluctuations were greatest at higher mean temperature (30°C): Here diurnal fluctuations reduced pupal mass and increased pupal development time for the laboratory population, but had little effect for the field population. We also evaluated how mean and fluctuations in temperature during early larval development affected growth rate during the final larval instar as a function of test temperature. At an intermediate (25°C) mean temperature, both the laboratory and field population showed a positive acclimation response to diurnal fluctuations, in which subsequent growth rate was significantly higher at most test temperatures. In contrast at higher mean temperature (30°C), diurnal fluctuations significantly reduced subsequent growth rate at most test temperatures for the laboratory population, but not for the field population. These results suggest that during domestication in constant temperatures, the laboratory population has lost the capacity to tolerate or acclimate to high and fluctuating temperatures. Population differences in acclimation capacity in response to temperature fluctuations have not been previously demonstrated, but they may be important for understanding the evolution of reaction norms and performance curves.


| INTRODUC TI ON
Temperature and other key environmental factors vary at daily, seasonal and annual time scales, particularly in terrestrial environments.
As a result, an ectothermic organism may experience a wide range of environmental and body temperatures during its lifetime, and different life stages or seasonal generations may experience different temperature conditions. Because temperature strongly influences physiological rate processes (e.g., feeding or metabolic rate), life-history traits (age or size at maturity), and fitness components (survival or reproduction) in most ectotherms, organisms exhibit a diversity of fixed and plastic responses to variable thermal environments (Cossins & Bowler, 1987;Huey & Bennett, 1990). There are two broad categories of plastic responses to temperature: thermal reaction norms and thermal performance curves (Beaman et al., 2016;Huey & Kingsolver, 1989). A thermal reaction norm represents the phenotypic trait value for some trait of interest as a function of the previous body temperature(s) experienced by the organism or genotype.
For example, temperatures experienced throughout development can determine life-history traits such as final body size and age at maturity. Alternatively, a thermal performance curve (TPC) represents the performance of some rate or trait of an organism-for example, the rate of feeding, metabolism, growth, survival, or reproduction-as a function of its current body temperature..
Responses to current temperatures may also be altered by previous temperatures experienced during development, which we will term time-dependent effects (Kellermann et al., 2017;Kingsolver et al., 2015). Numerous studies have demonstrated how prior thermal history can reversibly or irreversibly alter subsequent physiological responses to temperature, including heat or cold tolerance, thermal preference, and the thermal sensitivity of metabolic rate (Bowler, 2005). Such time-dependent effects may have either positive (thermal acclimation) or negative (thermal stress) consequences for performance and fitness (Bowler, 2005, MacLean et al., 2019Metzger & Schulte, 2017;Zeh et al., 2014).
Many recent studies have emphasized the importance of diurnal fluctuations in temperature for ectotherm performance and fitness in nature (Colinet et al., 2015;Kern et al., 2015). It is important to distinguish three distinct consequences of diurnal variation for organisms. First, because thermal performance curves are nonlinear, mean performance is generally different in constant and fluctuating environments with the same mean temperature. This effect is wellknown and is routinely incorporated into physiological and ecological models of mean performance in variable environments, including changing climates (Deutsch et al., 2008;Ruel & Ayres, 1999). Second, diurnal fluctuations can produce extreme temperatures that exceed lower or upper thermal limits, leading to stress, damage, or death.
The negative effects of short-term exposure to extreme temperatures have been widely documented (Cossins & Bowler, 1987). Third, previous exposure to diurnal fluctuating temperatures earlier in development could alter subsequent responses to temperature-that is, positive (acclimation) or negative (stress) time-dependent effects (Kellermann et al., 2017;Kingsolver et al., 2015). Whereas numerous studies have demonstrated stress or acclimation responses to prior exposure to constant temperatures or to single heat or cold shocks (Cossins & Bowler, 1987;Sinclair & Chown, 2005), few studies have explored the consequences of diurnal fluctuations for such time-dependent effects (Cavieres et al., 2018;Kern et al., 2015;Kingsolver et al., 2015). For example, a previous study with domesticated Manduca sexta showed that diurnal temperature fluctuations during early larval development increased subsequent larval growth rates at high temperatures, relative to larvae reared at constant temperatures (Kingsolver et al., 2015).
Ectotherms can vary widely in their responses to thermal environments in ways that reflect their evolutionary histories (Angilletta, 2009). Population and species differences in thermal reaction norms, thermal limits, and thermal performance curves have been documented in many taxa, and these differences contribute to adaptation to local environmental temperatures (Frazier et al., 2006;Huey & Bennett, 1987;MacLean et al., 2019;Sunday et al., 2011).
However, few studies have evaluated evolutionary differences in organismal responses to diurnal fluctuations, in part because thermal means and fluctuations are often confounded in natural environments (Kellermann et al., 2017).
In this study, we explore the effects of mean temperature and diurnal temperature fluctuations on larval growth, stress, and accli- Carolina that shares a common ancestry with the laboratory population dating to the 1960s. By comparing populations that differ in their recent evolutionary experience to fluctuating thermal environments, our experiments test two predictions: (a) diurnal fluctuations will have fewer negative impacts on thermal reaction norms of life-history traits in the field than the laboratory population; (b) diurnal fluctuations will generate stronger acclimation or weaker stress responses in TPCs for the field than the laboratory population. Our K E Y W O R D S acclimation, fluctuating temperatures, growth, laboratory adaptation, reaction norms, stress, thermal performance curves results provide partial support for these predictions, but the effects of fluctuations depend strongly on mean temperatures.

| Study system
The Tobacco Hornworm, Manduca sexta, occurs across northern South America, Central America, and southern North America. The adult hawkmoths are nectar feeders that can be highly dispersive (Madden & Chamberlin, 1945). The herbivorous larvae feed on a variety of hostplants, primarily in the Solanaceae family, and they are an important agricultural pest on domesticated tobacco in the southeastern United States. The larvae of M. sexta typically have 5 instars; toward the end of the final instar, larvae stop feeding, wander off the hostplant to burrow in the soil, and pupate below the soil surface.
Because of its rapid growth and development, large body size and successful maintenance on artificial diets, M. sexta has been a model system for insect physiology, development, and ecology for more  (D'Amico et al., 2001, Diamond & Kingsolver, 2011, Kingsolver, 2007, Kingsolver & Nagle, 2007. Therefore, these populations could also differ in their time-dependent responses to diurnal fluctuations. Many aspects of the thermal biology of Manduca larvae have been explored. Field measurements show that M. sexta larvae experience a wide range of environmental and body temperatures within and between seasonal generations (Case y, 1976 18 and 34°C. In short-term (24-48 hr) feeding trials, they can maintain positive growth rates for temperatures between 10 and 42°C, with maximum growth rates near 35°C (Kingsolver & Woods, 1997); and they can survive 24 hr heat shocks of 42-43°C (Case y, 1977).
Several previous studies with M. sexta have explored the effects of diurnal temperature fluctuations, demonstrating that mean larval growth and developmental rates differ between constant and fluctuating conditions with the same mean temperature (Kingsolver et al., 2009, Stamp, 1994. For example, our experiments with laboratory M. sexta show that diurnal temperature fluctuations increase mean growth and development rates at low mean temperatures (20°C), but decrease these rates at high mean temperatures (30°C) (Kingsolver et al., 2015). These analyses indicate that the effects of fluctuations cannot be fully accounted for by the nonlinearity of thermal performance curves for growth and development: Time-dependent effects such as stress or acclimation also contribute to these responses, especially at intermediate and higher temperatures (Kingsolver et al., 2015). Similarly, at intermediate mean temperatures, diurnal fluctuations during early larval development can increase the optimal temperature and maximal growth rate in laboratory M. sexta (Kingsolver et al., 2015). The effects of high and fluctuating temperatures, and the roles of stress and acclimation responses, are largely unknown for field populations of M. sexta; these are the primary focus of the experiments reported here.

| Experiments
The basic rearing protocol is similar in both experiments described here. Eggs were placed in petri dishes in a humidified environmental chamber at constant 25°C and 14L:10D photocycle until hatching.
Newly hatched larvae (5-10/dish) were randomly assigned to experimental treatments (see below) with abundant food, and the food in each dish was changed regularly as needed. After molting into the 3rd instar, each larva was transferred to an individual dish and maintained individually for the rest of the experiment. Laboratory population larvae were fed on a standard artificial diet; larvae from the field population were fed on an artificial diet with the addition of dried tobacco (Nicotiana tabacum) leaves (8.2% dry mass) to serve as a feeding stimulus. Previous studies show that larvae from the laboratory population feed and grow very similarly on diets with and without tobacco , Kingsolver, 2007. Newly hatched larvae were assigned to experimental treatments and reared as described above. Starting with the 3rd instar, survival, age (day), and mass (Mettler AT and XSE Toledo microbalances) for each larva were recorded at the start of each subsequent instar, at wandering, and at pupation. Sample sizes were N = 516 (field population) and N = 273 (laboratory population).

| Experiment 1: Effects of temperature regimes on thermal reaction norms for lifehistory traits
Our analyses focus on survival, development time, and mass at pupation. Because the experiments with the laboratory and field populations were done at different times, we analyzed the laboratory and field data separately. All analyses were done using R (version 3.5.0). Pupal development time and pupal mass were modeled using linear models (function lm) with MT and DFT as factors; pupal survival was modeled as a binomial response using generalized linear models (function glm) with a logit link function.  where m f = final mass at the end of the trial, m i was initial mass at the start of the trial, and t was the duration of the trial. This represents the relative rate of growth, relative to the initial mass during the trial. Trial duration was set at 24h for all larvae, but for practical reasons, there was some variation in the actual duration between the initial and final measurements for each individual (mean = 23.8h, SD = 1.1 h). Small size at the start of the 5th instar is an indicator of poor condition: Small size at this stage frequently results in additional larval instars and reduces survival to wandering and pupation (Kingsolver, 2007). This is particularly true for the field population, which is more poorly adapted to the artificial diet and shows greater variation in growth rate and size , Kingsolver, 2007, Kingsolver & Nagle, 2007. To account for this, we excluded larvae that were less than 250 mg (field population) or 600 mg (laboratory population) at the start of the 5th instar from the analyses (Kingsolver, 2007, D'Amico et al., 2001. The different thresholds reflect the large difference in size between final instar larvae and pupae of field compare with laboratory populations , Kingsolver, 2007, Kingsolver & Nagle, 2007. Larvae that did not feed or lost weight during the trial were also excluded. The final sample sizes were N = 180 (field population) and N = 505 (laboratory population).

| Experiment 1: Effects of temperature regimes on thermal reaction norms for life-history traits
For the field population, both mean temperature (MT) and diurnally fluctuating temperature (DFT) during development significantly affected survival to pupation, but there was no significant interaction between MT and DFT. Mean survival at MT = 30°C was lower than at 25°C, especially with larger diurnal fluctuations.
Overall mean survival across all rearing treatments was 0.48, because field larvae are not well-adapted to artificial diet, even when tobacco is incorporated into the diet (Kingsolver, 2007, Kingsolver et al., 2009. Pupal mass for the field population was significantly affected by MT and DFT, with no significant interaction between MT and DFT (Table 1a). The magnitude of these effects was relatively modest (Figure 1). For example, mean pupal mass was 21% smaller at a mean temperature of 30°C than at 25°C, and 5% smaller in fluctuating (±10°C) than in constant (±0°C) rearing conditions. Development time to pupation was significantly affected by mean temperature but not diurnal fluctuations; there was a marginally significant (p = .0504) interaction between MT and DFT (Table 1)

| Experiment 2: Effects of temperature regimes on later thermal performance curves for larval growth
For the field population reared at a mean temperature of 25°C, final mass at the end of the growth trial was significantly affected by test temperature, diurnal fluctuations, and their interaction, indicating that diurnal fluctuations during development affected thermal sensitivity of subsequent larval growth (Table 2). Initial mass and its interaction with test temperature also significantly affected final mass. Mean relative growth rate (day −1 ) was greatest For the laboratory population reared at a mean temperature of 30°C, final mass was significantly affected by rearing DFT, test temperature, initial mass, and the interaction between rearing DFT and test temperature (Table 2)  Note: Analyses for the field (A) and laboratory (B) population are given separately.

| Life-history responses to mean and fluctuating temperatures
Fluctuating temperatures during development have a wide range of effects on performance and life-history traits in ectotherms (Colinet et al., 2015;Deutsch et al., 2008;Kern et al., 2015;Ruel & Ayres, 1999;Sinclair et al., 2016). Because rates of growth, development, and other biological processes vary nonlinearly with temperature, the effects of diurnal fluctuations on mean performance can change with mean temperature, especially when diurnal high temperatures exceed the optimal temperature for performance (Colinet et al., 2015;Ruel & Ayres, 1999). For example, diurnal fluctuations can increase mean growth rates at low mean temperatures but decrease them at high mean temperatures, because of the time-dependent stress effects from repeated exposure to high temperatures (Kingsolver et al., 2015). The consequences of repeated exposure to such stressful, sublethal conditions have now been documented in a number of insect systems (Colinet et al., 2015;Sgro et al., 2016;Xing et al., 2014;Zhang et al., 2015).  (Figure 1). The field population also shows less plasticity to mean temperature (shallower reaction norm slopes) than the laboratory population for these life-history traits.
Previous studies have documented substantial evolutionary changes in this and other M. sexta populations during laboratory domestication, including increases in larval growth rate and final body size, reduced immune responses, and reduced tolerance to novel hostplants , Diamond & Kingsolver, 2011, Kingsolver, 2007, Kingsolver & Nagle, 2007, D'Amico et al., 2001. The present results suggest that evolution of the laboratory population under constant temperatures has resulted in lower tolerance and greater sensitivity to high, fluctuating temperatures during development. This is consistent with the finding that larval survival at high, constant temperatures (35°C) is lower in the laboratory than the field population (Kingsolver & Nagle, 2007).
In contrast, patterns of HSP gene expression in response to heat shocks are similar in the two populations (Alston et al., 2020).
Given the lack of replicate populations in our study, the evolutionary mechanisms and relative roles of selection and drift in these patterns are unknown. Studies with five strains of Drosophila melanogaster showed no changes in mean heat tolerance with domestication over 55 generations (Krebs et al., 2001). Similarly, comparisons between laboratory and field populations for nine

| Stress, acclimation, and selection history
Thermal performance curves provide a useful way to characterize the thermal sensitivity of an individual, genotype or population over a range of body temperatures (Huey & Kingsolver, 1989;Huey & Stevenson, 1979). Many studies have evaluated the potential for thermal acclimation of performance curves, whereby constant temperatures during development or an acclimation period alter subsequent performance (Angilletta, 2009, Beaman et al., 2016Huey et al., 1999). Beneficial acclimation to constant temperatures has been documented in many ectotherms, especially in aquatic systems (Angilletta, 2009, Seebacher & Grigaltchik, 2014, but few studies have explored acclimation to fluctuating temperatures. For example, a recent study with D. melanogaster showed that both mean and diurnal fluctuations in rearing temperature affected thermal sensitivity of adult walking speed (Cavieres et al., 2016). In particular, mean optimal temperature for walking speed increased with diurnal fluctuations at both low and high mean temperatures. By contrast, mean maximum performance increased with diurnal fluctuations at low mean temperature, but decreased with fluctuations at high mean temperatures. This finding suggests that the effects of fluctuations on acclimation and stress responses may depend on mean developmental temperatures (Cavieres et al., 2016;Kingsolver et al., 2015).
Our results for the laboratory population of M. sexta support this suggestion ( Figure 2). Diurnal fluctuations at lower mean temperature (25°C) during development increased the optimal temperature and maximum performance for larval growth rate, the classic signature of beneficial acclimation. Conversely, fluctuations at higher mean temperature (30°C) decreased growth rate at most test temperatures, confirming that high, fluctuating temperatures during development are stressful for this population (Figure 1; (Kingsolver et al., 2015). As a consequence, plastic responses to diurnal fluctuations during development can generate beneficial acclimation or stress-induced reductions in performance depending on mean temperature conditions.
Note that diurnal fluctuations had little effect on performance at the lowest or highest test temperatures: The acclimation and stress responses we observed were largely restricted to intermediate temperatures ( Figure 2). Recent studies in a variety of ectotherms have suggested that there is limited plasticity in upper thermal limits (e.g., critical thermal maximum temperature, CT max ) (Hoffmann et al., 2013;Kellermann et al., 2017;Sorensen et al., 2016 rates of larval growth and development (Kingsolver et al., 2015(Kingsolver et al., , 2016Kingsolver & Nagle, 2007;Kingsolver & Woods, 1997). As the result, the potential for acclimation to high or fluctuating temperature may be quite different for upper thermal limits than for other components of thermal sensitivity.
The field population of M. sexta also had a positive acclimatory (i.e., increased mean growth rate) response to diurnal fluctuations at a mean temperature of 25°C, but the laboratory and field population had different responses at the high mean rearing temperature (30°C) (Figure 2). In particular, diurnal fluctuations during rearing significantly reduced subsequent larval growth rates for the laboratory population, but not for the field population. The genetic and physiological bases for these differing responses are unknown, but they are consistent with the hypothesis that, during laboratory domestication at constant temperatures, the laboratory population has lost the capacity to acclimate to high, fluctuating temperatures during larval development.
The field and laboratory populations used in this study also differ their adaptation to larval food: whereas laboratory M. sexta perform well on both artificial diet and tobacco (their most common natural host in the Southeast US), field M. sexta have slower growth, development, and survival on diet than on tobacco . There is also greater variability in growth and development rates in field larvae on diet, and some field larvae have additional instars when reared on diet and other lower quality food resources (Kingsolver, 2007. Larvae that are below a "critical weight" at the start of the 5th instar are likely to have additional instars, and this differs between field and laboratory populations (D'Amico et al., 2001, Davidowitz et al., 2003; this was the rationale for excluding small larvae from the study (see Section 2). However, the effects of the lower resource quality of diet for field M.sexta are unlikely to explain the population differences in thermal tolerance and acclimation capacity at high, fluctuating temperatures that we show here. Models and several empirical studies suggest that reduced nutrient availability will reduce maximal growth rate, optimal temperature and upper thermal limits (Huey & Kingsolver, 2019;Thomas et al., 2012). In contrast, we find that laboratory but not field M. sexta suffered reduced growth when reared at high and fluctuating temperatures (Figure 2).
Numerous studies have documented evolutionary reductions in environmental tolerance or in adaptive plasticity as a result of relaxed selection (Lahti et al., 2009;Stoks et al., 2016).
Experimental evolution studies in the laboratory can yield either increased or decreased plasticity in the traits that are under selection (Garland & Kelly, 2006 Kingsolver et al. (2015) and are included for comparison produced strong directional selection for increasing larval growth rate and large final size, and relaxed selection for hostplant defensive chemicals, immune response and heat tolerance (D'Amico et al., 2001;Diamond & Kingsolver, 2011;Kingsolver & Nagle, 2007). Studies of laboratory domestication provide a useful tool for understanding how changes in the direction and strength of selection affect the evolution of plasticity (Garland & Kelly, 2006). Population differences in acclimation capacity in response to temperature fluctuations have not been previously demonstrated, but may be important for understanding the evolution of thermal reaction norms and thermal performance curves.

ACK N OWLED G EM ENTS
We thank Emily McGuirt, Matthew Nielsen, Anna Pearson, and Rachel Rice for help with the experiments. This work supported in part by the US National Science Foundation [IOS 155559].

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest. Writing-review & editing (supporting).

DATA AVA I L A B I L I T Y S TAT E M E N T
Data presented in the paper are available on Dryad at the time of publication at https://doi.org/10.5061/dryad.15dv4 1nw3.