Climate‐related variation of metabolic rate across the distribution of a broadly tolerant invasive forest pest

Metabolic rate is a widely studied physiological species trait related to energetics, climate, and geographical distributions. Hypotheses have been proposed to explain variation in metabolic rate, but evidence has been mixed due to the limited sampling scope of intraspecific studies. Successful biological invasions offer a unique opportunity to examine the development of intraspecific physiological variation and how it relates to climate, invasive spread and species range limits. The aim of this study was to determine if metabolic rate variation across the spongy moth invasive range was consistent with a pattern of adaptation to local climate.

While macrophysiology has been an important lens for understanding physiological variation across large geographical extents, much of the literature has focused on one physiological trait in particular, the breadth of thermal tolerance (difference between upper and lower limits of thermal tolerance).These studies show that thermal tolerance breadth generally increases towards higher latitudes where climates become more variable (Calosi et al., 2010;Deutsch et al., 2008;Sheth & Angert, 2014;Sunday et al., 2011) and that species with a broader thermal tolerance tend to have larger geographical ranges (Bonino et al., 2015;Calosi et al., 2010;Sheth & Angert, 2014;Slatyer et al., 2013).However, these studies also show that thermal tolerance limits are often mismatched with climatic extremes of species' realized latitudinal ranges for terrestrial ectotherms (Sunday et al., 2012).Given that physiological tolerance is a complex multivariate trait, it is likely that other traits, in addition to thermal tolerance breadth play an important role in determining species range dynamics (Agosta et al., 2013;Gaston, 2003;Sexton et al., 2009).Thus, macrophysiological studies that examine other physiological variables, especially those related to energetics, are needed to expand our understanding of the linkages between climate and physiology.
Studies that examine geographical variation in basal or resting metabolic rates can help fill this gap because they represent the energetic cost to fuel the maintenance of biological processes for an organism.These costs can significantly impact the energy available for growth and reproduction which will impact individual fitness and therefore species distributional limits (Dunham et al., 1989;Martinez et al., 2017).Numerous macro-scale comparative studies have shown that variation in metabolic rate can be related to climate and geography (e.g.Addo-Bediako et al., 2002;Chown & Gaston, 1999;DeLong et al., 2018).From these studies, two main hypotheses have been proposed to explain this variation.The metabolic cold adaptation hypothesis (MCA) predicts that species or populations from colder climates have elevated metabolic rates relative to species or populations from warmer climates when assayed at the same temperature (Addo-Bediako et al., 2002;Clarke, 1991Clarke, , 1993;;DeLong et al., 2018;Krogh, 1916).
The 'hotter is better' hypothesis (HIB) predicts that the performance of physiological traits of warm-adapted species or populations is driven by thermodynamic effects which results in a higher metabolic rate (Bennett, 1987;DeLong et al., 2018;Frazier et al., 2006;Kingsolver & Huey, 2008).
To address these limitations, more comprehensive studies of intraspecific variation that span several degrees of latitude and a sufficient number of populations are needed.
Biological invasions offer an especially unique opportunity to study macro-scale intraspecific physiological variation in response to climate variation and how this relates to geographical distributions.Successful invasive species often exhibit a high degree of phenotypic plasticity including a eurythermal physiology (Hendry, 2016;Kelley, 2014).As a result, invasive species typically have a broad thermal tolerance (Kelley, 2014) which is also characteristic of many other wide-spread species (Calosi et al., 2010;Sheth & Angert, 2014;Slatyer et al., 2013;Sunday et al., 2011).The advantages of eurythermal physiology include the capacity to perform over a broad array of environmental temperatures, which facilitates spread and expansion towards both relatively cold and warm climates (Kelley, 2014;Thompson et al., 2021).In addition, there is evidence that climaterelated physiological and life-history traits of invasive species are capable of rapid evolution as they encounter increasingly divergent climates (Hellmann et al., 2008;Moran & Alexander, 2014;Sexton et al., 2009), facilitating invasion and more generally, species expansion.Together, these features make biological invasions good model systems for understanding the linkages between species range dynamics, climate variation, and physiology and providing insight into how species respond to changing climates.
The spongy moth (Lymantria dispar dispar, Lepidoptera: Erebidae) invasion in North America has become an important model system for understanding these linkages.Previous work has shown that a broad thermal tolerance combined with a broad diversity of potential food plants has allowed L. d. dispar to spread rapidly across 11 degrees of latitude in approximately 150 years and provides strong evidence that this spread is related to climate (Thompson et al., 2021).At the northern end of their range, cold temperatures limit expansion via winter mortality and short growing seasons with fewer degree days for development (Denlinger et al., 1992;Gray, 2004;Madrid & Stewart, 1981).At the southern end of their range, growth and development of larvae appears less constrained by temperature, but their spread south may be limited by insufficient cold temperatures to complete diapause (Gray, 2004).More recent work suggests that expansion at the southern range has stalled and/ or contracted possibly due to prolonged exposure of larvae to supraoptimal temperatures (Tobin et al., 2014).This is consistent with additional work that shows supraoptimal temperatures negatively impact L. d. dispar growth, development and survival (Banahene et al., 2018;Faske et al., 2019;Thompson et al., 2017).Together, these studies establish that climate variation and thermal physiology have played a significant role in L. d. dispar range dynamics.
The aim of our study was to explore the influence of divergent climates on physiological trait variation across the geographical range of an eurythermal invasive species and to determine if that variation is consistent with a pattern of adaptation to local climates.
In this study, we addressed the following questions: (1) do the energetics (whole-organism metabolic rate) of L. d. dispar vary across its invasive range?(2) Is this variation related to local climate?(3) How do the observed patterns compare to other intraspecific studies?We accomplished this by conducting a classic comparative experiment using L. d. dispar larvae from 14 populations across its invasive range and assayed their metabolic rate at three ecologically relevant temperatures (i.e.temperatures that occur normally across the L. d. dispar range during larval development).We then surveyed the available macrophysiology literature for intraspecific metabolic rate studies focusing on ectotherms (Table 1).We hypothesized that metabolic rate of L. d. dispar has locally adapted to climate as it has expanded its invasive range over the past 150 years.If supported, we predict variation in population-level metabolic rates will be correlated with local climate and geographical variables.To our knowledge, this study represents the most comprehensive macrophysiological analysis of population-level variation in whole-organism metabolism along a latitudinal gradient for a single species (Table 1).
It was introduced from France to Medford, MA in 1869 and has since spread over 900,000 km 2 (Grayson & Johnson, 2018;Tobin et al., 2012).The current range of L. d. dispar encompasses eastern Canada, south into southern Virginia and northern North Carolina and westward into Wisconsin and Minnesota (Figure 1).Due to efforts over the past 34 years by state and federal agencies to manage the expansion of L. d. dispar, it is likely one of the most well-documented biological invasions in the world (Grayson & Johnson, 2018).

| Study populations
This study was conducted in parallel with a study on growth and development by Thompson et al. (2021) in 2018 using the same L. d. dispar populations.These populations were sourced from 14 locations across the current range (Figure 1).Eleven populations were located along the invasion front from areas of active expansion and the three remaining populations were collected from interior portions of its range where it has long been established.At each source location, multiple egg masses (n = 11-60) were collected for each population.For invasion front (i.e.range margin) populations, egg masses were collected from inherently low-density populations, but with densities high enough for sampling.Some populations (MA1, MA2, NY, NC1, NC2) were collected 1-2 years prior to the experiment (2016-2017) and were reared under common garden conditions for 1-2 generations (Thompson et al., 2021).The remaining populations were collected during fall of 2017 and, along with the previously collected populations, were overwintered in the egg stage in the same common garden conditions prior to the experiment.

| Rearing L. d. dispar larvae
For each population, we used an admixture of eggs made from multiple egg masses in an effort to randomly sample the gene pool.We hatched eggs at 25°C in an environmental chamber (model I-22VL, Percival Scientific, Inc.) over a 6-week period.Once hatched, we took larvae from each population and group reared (n = 10) them in 178 mL cups with 20-25 mL of artificial diet (USDA APHIS formulation).We placed these cups in a single environmental chamber and at a constant 25°C on a 14 h light, 10 h dark cycle.We monitored cups daily and until larvae reached the third instar, the midpoint of larval development.

| Routine metabolic rate
We assayed routine metabolic rate (RMR) at three ecologically relevant temperatures (15, 25 and 30°C).Sample sizes ranged from 6 to 13 individuals per population per temperature treatment.We broadly defined our measurements as routine (as opposed to resting or basal) metabolic rates to account for absorptive processes and diel locomotor activity of larvae during respirometry trials.
However, based on previous work (May et al., 2018) we are confident that individuals were minimally active during the trials.RMR of larvae was measured within 24-72 h after transitioning to the 3rd instar.We measured RMR using a push-mode, stop-flow respirometry system (Field Metabolic System [FMS], model 2, Sable Systems International, Las Vegas, NV) connected to two 8-channel multiplexers (model RM-8, Sable Systems International).Each channel was fitted with a 35-mL chamber.We housed the system in a homemade walk-in temperature control chamber (TCC) described by Martinez and Agosta (2016) and temperature was monitored by the FMS with a temperature probe that was inserted into one of the baseline chambers.At the end of each respirometry trial, larvae were euthanized in a −20°C freezer (i.e.different larvae were measured for each assay temperature).

TA B L E 1
Literature survey results for studies (N = 63) examining population-level intraspecific variation in metabolic rate (standard, routine or resting) among different populations (≥2) located at different latitudes in ectotherms.Note: For the latitudinal cline, 'positive' indicates that the metabolic rate increased with increasing latitude and 'negative' indicates that the metabolic rate decreased with increasing latitude.Species in bold are invasive species.
TA B L E 1 (Continued) Our respirometry protocol was similar to a previous study (May et al., 2018), with minor modifications.We isolated and held larvae without food for at least 1 h and then weighed (0.001 g) each individual.During each respirometry trial, air inside each chamber was flushed at a flow rate of 100 mL/min for 3 min each hour for 4 h.
During hour 1, we allowed larvae to acclimate to the assay temperature and during hours 2-4, respirometry measurements were taken once per hour.Incurrent air was scrubbed for water vapour and CO 2 using three sequential columns (silica gel, soda lime, Drierite®) before being pushed through each selected chamber.The excurrent air from the selected chamber was routed through a water vapour sensor and then scrubbed for water vapour using magnesium perchlorate.This dry air was then routed through a CO 2 sensor and subsequently scrubbed for CO 2 using Ascarite®.Finally, the dry, consumed by integrating the area under each M S O 2 peak.We then divided the sample volume by the enclosure time to find the O 2 consumption rate (VO 2 μL/h) for each individual.

| Statistical analysis of RMR
We analysed RMR data using a linear mixed effects model using the 'lme4' package (Bates et al., 2015(Bates et al., , 2021) ) in R version 4.1.0(R Core Team, 2021): Population, assay temperature and body mass were fixed effects.The population was treated as a fixed effect so we could test for differences among populations and relate these differences to geographical location.Both assay temperature and body mass, which are known to have a strong influence on metabolic rate, were treated as continuous covariates.The population-by-temperature interaction was included to test for differences in the temperature sensitivity of RMR (i.e.slope of RMR-temperature relationship) among populations.Sample hour was included as a random effect.
Using this model, we computed the estimated marginal mean for each population using the 'emmeans' package in R (Lenth et al., 2021).This estimate represents the RMR for each population that has been adjusted to the same body mass across all populations (grand mean).We then checked for significant differences in RMR among populations using multiple comparisons with a Šidák correction at each assay temperature.

| Statistical analysis of RMR and climate
Population estimated marginal means for RMR were analysed as a function of climate and geographical variables.Climate data were retrieved from the PRISM Climate Group (2012) using the geographical coordinates for each population.From PRISM, we obtained 30-year climate normals (1981-2010) for temperature (mean, minimum, maximum and range) and precipitation for each season.To reduce dimensionality and quantify local climate for each population, we performed a principal components analysis (PCA) with these climate and geographical data.The PCA was done using base R (R Core Team, 2021).
From these principal components (PC), we used PC regression to determine if larval RMR was significantly related to climate.Finally, we calculated the temperature coefficient (Q 10 ) for larval RMR by plotting log-transformed RMRs as a function of temperature.Using the slope of this relationship, we calculated the Q 10 of RMR (Lighton, 2008): where m is the slope of the RMR-assay temperature relationship.We then used PC regression in R (R Core Team, 2021) to determine if the sensitivity of RMR to temperature (Q 10 ) was related to climate.Specifically, we searched for empirical studies making intraspecific comparisons among different populations (≥2) that were located at different latitudes and measured standard, routine or resting metabolic rates.Studies that did not meet these criteria were removed from our search results.

| RMR of L. d. dispar populations
We found that population, temperature, body mass and the population by temperature interaction all had significant effects on RMR (Table 2).Overall, the linear mixed effects model explained 70% of the total variance in RMR with 49.3% explained by the fixed effects (Table S1).Effect sizes for body mass (ω 2 = 0.31) and assay temperature (ω 2 = 0.51) show these predictors had the largest influence on TA B L E 2 ANOVA table for the linear mixed effects model.2).Effect sizes for population (ω 2 = 0.08) and its interaction with temperature (ω 2 = 0.07) were relatively small.Estimated marginal means from this model show that RMR increased as assay temperature increased and that population RMR increased from southern to northern latitudes across assay temperatures (Figure 2).

| Population RMR and climate
Correlation matrices for each season show that RMR was negatively related to measures of temperature and precipitation and positively related to latitude of the population source location (Figure S1).As expected, many of these climate and geographical variables showed moderate to strong correlations with one another.The nature of these correlations was consistent across seasons.Several of these correlations also show moderate to strong relationships between RMR and local climate and geographical variables (Figure S1).A PCA of these variables showed 86.5% of the cumulative variance was explained by PC1 (λ = 15.26,σ 2 = 66.4%) and PC2 (λ = 4.62, σ 2 = 20.1%).A PCA biplot revealed that populations grouped by climate similarity and geographical location (Figure 3).Specifically, the angle, magnitude, and direction of loadings characterizes a latitudinal gradient with cold and relatively dry climates at the northern range margin and warm and relatively wetter climates at the southern range margin.
For PC1, regression showed RMR declined from colder to warmer climates (Figure 4).While the relationship with RMR was R 2 = 0.01, p = 0.307) or the relationship was weak and not significant (30°C: Adj.R 2 = 0.13, p = 0.115) (Figure S2).Q 10 of RMR showed the opposite trend and increased from warmer to colder climates after the removal of a single outlier (Figure 5).However, this relationship was weak and not significant (Adj.R 2 = 0.12, p = 0.131).

| Comparison of L. d. dispar RMR variation to other studies
Our literature search queries found 1454 unique studies.These were mostly studies that either did not focus on ectotherms, made interspecific comparisons, or were reviews.Removal of these studies yielded a total of 63 studies that were intraspecific and measured the metabolism of an ectotherm in relation to location and/or climate (Table 1).The latitudinal span (terrestrial: x = 11.2°,range = 4-25°; (24 terrestrial, 29 aquatic), including seven invasive species (5 terrestrial, 2 aquatic).Within each ecosystem, over half of the studies were from a single class (terrestrial: 13 insect studies with 13 species; aquatic: 14 fish studies with 13 species).
Most studies in the survey found no relationship between metabolic rate and latitude (13 terrestrial, 14 aquatic).This was followed closely by studies showing a positive relationship (10 terrestrial, 14 aquatic); studies showing a negative relationship were the least common (8 terrestrial, 4 aquatic).The majority of studies used latitude as a proxy for climate and statistical comparisons were often based on population location.We found few studies (2 terrestrial, 2 aquatic) that made direct comparisons of metabolic rate with local climate and these studies only examined one variable (temperature).

| DISCUSS ION
Across the current invasive range of L. d. dispar in North America, larval metabolic rate showed the expected positive relationship with body mass and assay temperature for an ectotherm (Gillooly et al., 2001) which explained the majority of variance in the data (Table S1).We did detect a significant population-by-temperature interaction; however, the effect was small with no systematic differences among populations in the sensitivity (slope) of metabolic rate across the assayed temperature range.This was consistent with a non-significant relationship between Q 10 and climate among these populations (Figure 5).Population also exhibited a small, but significant effect and revealed a clinal pattern of increasing metabolic rate from south-to-north (Figure 2).Further examination using PCs of local climate and geographical variables revealed a significant cline in metabolic rate that persisted across assay temperatures (Figure 4).To our knowledge, this study is one of the most comprehensive macrophysiological analyses of intraspecific variation in metabolic rate and its relation to climate and invasive spread for an ectotherm to date (Table 1).Only a few studies that we found in our literature four previous studies that we found included similar data on local climates in their analysis (Table 1).
A primary objective in examining macroscale climate-related clines is determining the contribution of phenotypic plasticity and genetically based local adaptation to these patterns.The latitudinal cline in population-level RMR we observed is consistent with predictions of the MCA hypothesis, which posits that cooler climates select for the evolution of elevated metabolic rates.Definitive evidence of MCA requires a trans-generational acclimation treatment and comparative genetic analysis among populations to determine the contribution of phenotypic plasticity to interpopulation variation in thermal physiology (Terblanche et al., 2009).Previous work by our group has tested for the presence of MCA in L. d. dispar by using two pairwise comparisons with populations from divergent climates and different latitudes (May et al., 2018).In these comparisons, there was no evidence that larval metabolic rate was elevated in northern populations (i.e.no MCA).There was some evidence of an acclimation response, but it was not consistent among populations.While the current study did not test for an acclimation response, we did observe that RMRs assayed at 25°C (rearing temperature for larvae) were less variable (Figure 2) and showed a stronger relationship with climate relative to the other assay temperatures (Figure 4).Along with thermal acclimation by larvae, our results could also have been influenced by non-genetic parental effects since we were unable to rear all populations under common conditions for an entire generation prior to the study (but note that all were overwintered for the same period of time in common conditions during the egg stage prior to the experiment).Previous work has demonstrated that maternal provisioning and maternal diet can influence larval development in L. d. dispar (Rossiter, 1991;Rossiter et al., 1993), and possibly their thermal performance.Despite this, we find it unlikely that thermal acclimation of larvae during rearing and/or transgenerational parental effects explain the climate-related latitudinal cline in RMR we observed for two reasons.First, we found that the different rearing conditions among populations (Table S3) had little to no effect on RMR (Table S4).Specifically, when this was included as a random effect in the linear mixed effects model, rearing conditions accounted for <2% of the left over variance, resulted in a higher AIC score, and did not change our results (Table S4).Second, d. dispar has spread across the landscape of North America from its point of introduction in 1869.
The climate-related differences in larval metabolic rate found among the populations in this study have also been observed for other L. d. dispar traits.For example, thermal reaction norms show survival decreases and pupal mass increases from north-to-south in these same populations (Thompson et al., 2021).Larger body size in the south could be related to season length, with longer growing seasons supporting longer development times (Friedline et al., 2019).However, the opposing trend with survival suggests there is a climate-related evolutionary trade-off between body size and survival in this system (Thompson et al., 2021).One common explanation for latitudinal or climate-related body size clines in insects is that they arise from varying levels of stress resistance to starvation or desiccation (Chown & Gaston, 1999;Cushman et al., 1993).
However, this seems unlikely in this system given that L. d. dispar are generally not food limited except during outbreaks (Wittman & Aukema, 2019) and the water content of leaves is relatively high (45%-75%) (Scriber & Slansky, 1981).A more plausible explanation is that larger body size leads to more efficient energy use.These same populations have an inverse relationship between larval metabolic rate and body size, meaning that larger individuals have a proportionally lower metabolic rate (Blanckenhorn et al., 2007;Kleiber, 1932;Reim et al., 2006).In the north, where there are colder climates and shorter growing seasons, higher larval metabolic rates could facilitate faster growth and maturation at a smaller body size (i.e.MCA).
Conversely, larger body sizes in the south may help conserve energy by reducing mass-specific larval metabolic rates in response to prolonged exposure to warmer climates with longer growing seasons.
One limitation of this study is that our conclusions about RMR are based on data from a single instar midway through larval development.Because complex life cycles can result in variable responses across developmental stages (Bowler & Terblanche, 2008;Sinclair et al., 2016;Spicer & Gaston, 1999), it is important to note that metabolic rate in L. dispar (Pincebourde & Woods, 2020).Together, this work would be very useful for predicting L. d. dispar spread in response to climate change.Additionally, while changes in whole-organism metabolic rate provide a general sense of energetic trade-offs, they offer little information on net energy available for growth, storage and reproduction.One way to address this is by examining mitochondrial physiology to determine how metabolic substrates are being used to produce cellular energy (i.e.ATP).Previous work from our group revealed that the shape and thermal optimum of mitochondrial performance was similar to the thermal performance of growth but did not match whole-organism MR-T in Manduca sexta larvae (Martinez et al., 2017).This mismatch between whole-organism and mitochondrial physiology may be a potential mechanism for producing intraspecific variation in RMR.If this same mismatch exists in L. d. dispar, it could help explain the clinal variation in larval RMR observed in this study.Finally, one outstanding question that remains to be addressed in this system is the degree to which the trait variation expressed in the invasive range represents a conserved response from the ancestral range (Huey et al., 2005), or something evolutionarily novel.A phylogeographical study of L. d. dispar that includes an

CO 2
free air was routed through an O 2 cell to measure the fractional concentration of O 2 .A minimum of one chamber was kept empty to use as a baseline for each trial.We analysed O 2 consumption traces using ExpeData software (Sable Systems International).After drift, lag and baseline corrections, all O 2 fractional concentration values were transformed to sample volumes of O 2 (Lighton, 2008): where FR is the excurrent air flow rate corrected for standard temperature and pressure.We calculated the total sample volume of O 2 M S O 2 = O 2 × FR ∕ 1 − 0.2095 − O 2 F I G U R E 1 Map of L. d. dispar populations that egg masses were collected from.Dark grey shaded region represents the quarantined areas in the United States.Colour scale of points represents the latitude of each population.
Estimated marginal means for population RMRs at each assay temperature.The colour scale represents the latitude of each population.Error bars represent the 95% confidence interval.Black bars represent multiple comparisons with a Šidák correction.Estimated marginal means with black bars that do not overlap with other estimated marginal means are significantly different from one another.Numbers in parentheses represent the sample size for each population at each assay temperature.weakerat 15°C (Adj.R 2 = 0.18, p = 0.076) (Figure4), it was strong and significant at the warmer temperatures (25°C: Adj.R 2 = 0.71, p < 0.001; 30°C: Adj.R 2 = 0.46, p = 0.005).For PC2, we found no relationship with RMR (15°C: Adj.R 2 = −0.07,p = 0.725; 25°C: Adj.

F
I G U R E 3 Biplot with climate PC1 and PC2.(a) Coordinates for climate PC1 and PC2 and the amount of variance they explain.(b) Loadings on climate PC1 and PC2.Each point represents a single population and the colour scale represents the latitude of each population (a).Arrows represent the direction and magnitude of loadings for each climate and geographical variable use in the PCA.Arrows are labelled by season (Wi = winter, Sp = spring, Su = summer, Fa = fall) and the corresponding variable (Elev, elevation; Prcp, Precipitation; T max , maximum temperature; T mean , mean temperature; T min , minimum temperature; T range , temperature range [T max -T min ].).F I G U R E 4 PC regression with estimated marginal means of population RMR as a function of PC 1 at each assay temperature.Each point represents a single population, and the colour scale represents the latitude of each population.Regression lines were fitted to each plot to determine if differences in larval RMR among populations were related to climate.These relationships were considered significant at α = 0.05.aquatic: x = 10.7°,range = 2-26°) and number of populations (terrestrial: x = 4, range = 2-9; aquatic: x = 4, range = 2-10) was similar between terrestrial and aquatic studies.There was a relatively even number of terrestrial (n = 31) and aquatic (n = 32) studies with 14 taxonomic classes (6 terrestrial, 11 aquatic, 3 shared) and 53 species These results are consistent with our hypothesis that metabolic rate variation among L. d. dispar populations follows a pattern of local adaptation in response to divergent climates.Specifically, they indicate that L. d. dispar physiology at the whole-organism level has responded to climate through adjustments in metabolic rate across ecologically relevant temperatures (although not sensitivity to these temperatures) as their range has expanded across eastern North America in the past 150 years.
search (7 out of 63, 15%) focused on invasive species experiencing recent or ongoing range expansion.Further, while our study covers similar degrees of latitude compared to the average previous study (~11), it uses significantly more populations (14 vs. mean of 4 with 27% of studies comparing only two populations and 32% comparing 3), covering the length of the invasion front from north to south.Finally, while all other studies used geographical location as a predictor variable (i.e. a correlate of climate), we included multiple aspects of local climate (different measures of environmental temperature, precipitation and season) for each population in our analysis.Only While our results are suggestive of MCA, we cannot rule out the influence of F I G U R E 5 Temperature coefficients (Q 10 ) for RMR for each population plotted as a function of PC1.Each point represents a single population, and the colour scale represents the latitude of each population.Linear regression was used to determine if temperature sensitivity of larval RMR was related to climate among these L. d. dispar populations.thermal acclimation on larval metabolic rate variation among these L. d. dispar populations in producing climate-related patterns.
a previous genetic analysis of L. d. dispar populations along the same latitudinal gradient has demonstrated that northern and southern populations have experienced spatially divergent selection pressures on temperature-dependent life history traits (larval development time and pupal mass)(Friedline et al., 2019).Together, these data support the idea that the patterns we observed in metabolic rate are genetically based.Regardless of the mechanism (evolved differences, expressed plasticity, or both) the observed climaterelated latitudinal cline in RMR contributes to growing evidence that, after an initial period of rapid geographical expansion fuelled by broad thermal tolerance and abundant food availability, divergent selection pressures related to climate variation have played a significant role in physiological and life history adaptation (see below) as L.
d. dispar is expected to vary during larval development since their body mass increases by 10-30 fold.Previous work has shown that whole-organism metabolic rate generally increases with body mass and each successive larval instar (Blossman-Myer & Burggren, 2010; Callier & Nijhout, 2012).While the slope of this relationship may change throughout development, any changes in metabolic rate across different instars would likely result in a proportional increase or decrease in metabolic rate (Blossman-Myer & Burggren, 2010; Callier & Nijhout, 2012), which would not alter the overall trends observed in our study.It is worth noting that unpublished data from our group with the OTIS laboratory strain of L. d. dispar confirm these patterns in RMR with instar and assay temperature and also show that temperature sensitivity of RMR is highest during the third instar (Powers et al. unpublished).To obtain a more complete picture of metabolic performance in L. d. dispar, future work should examine metabolic rate variation across different stages of larval ontogeny and/or assay metabolic rate across a wider range of ecologically relevant temperatures.This work should also consider how well the macro-scale climate data used in this study reflect the microclimates utilized by L. d.