Phenotypic flexibility of metabolic rate and evaporative water loss does not vary across a climatic gradient in an Afrotropical passerine bird

Global variation in avian thermal physiology is correlated with climate, with several analyses identifying air temperature (T_a) as the major correlate of mass-independent variation in avian metabolic rates (Jetz et al., 2008; Stager et al., 2016; White et al., 2007). Avian metabolic rates tend to be lower in tropical latitudes than in temperate-zone regions (Hail, 1983; Londoño et al., 2015; Stager et al., 2016; Weathers, 1979; Wiersma et al., 2007), and both metabolic rates and evaporative water loss (EWL) are reduced in arid-zone compared with mesic-zone birds (Tieleman et al., 2002, 2003; Tieleman and Williams, 2000). There is evidence of similar patterns of metabolic variation within species (e.g. MacMillen and Hinds, 1998; Sabat et al., 2006), supporting the idea that the thermal physiology of endotherms is more flexible than previously thought (Angilletta et al., 2010).

Understanding the sources of intraspecific variation in avian thermal physiology is particularly important when interpreting thermoregulatory responses in the context of a changing climate (Boyles et al., 2011; Huey et al., 2012; Williams et al., 2008). Physiological differences among birds can arise from genotypic adaptation (hard-wired differences in response to different selection pressures) or phenotypic plasticity (Angilletta et al., 2010; Piersma and Drent, 2003), although the distinction between these categories may be blurred by epigenetic transmission and the associated transgenerational effects (e.g. Mariette and Buchanan, 2016). In recent years, increasing attention has been given to the role of phenotypic plasticity, which includes developmental plasticity (ontogenetic variation that becomes fixed on reaching maturity) and phenotypic flexibility (reversible changes throughout an individual's lifetime; Piersma and Drent, 2003; Pigliucci, 2001). Phenotypic flexibility is an important component of acclimatization and acclimation – adjustments in response to environmental conditions in the field or laboratory, respectively (Piersma and Drent, 2003; Schlichting and Pigliucci, 1998).

A frequently-cited example of phenotypic flexibility is seasonal metabolic acclimatization in small north-temperate birds, typically involving winter increases in basal metabolic rate (BMR; minimum resting metabolic rate in post-absorptive, non-reproductive individuals) and summit metabolism (M_sum; maximum metabolic rate during acute cold exposure; reviewed by McKechnie et al., 2015; Swanson, 2010). In particular, a higher M_sum is associated with enhanced cold tolerance during the cold winters typical of higher latitudes (Cooper, 2002; Petit et al., 2017; Swanson, 2001; Swanson and Liknes, 2006), and both BMR and M_sum have been reported as flexible in response to short-term (days to weeks) fluctuations in T_a in several temperate species (Broggi et al., 2007; Dubois et al., 2016; Petit and Vézina, 2014; Swanson and Olmstead, 1999). Adjustments in BMR are typically associated with changes in the mass and metabolic intensity of the digestive and excretory organs, whereas adjustments in M_sum are associated with changes in the mass and metabolic intensity of the skeletal muscles, heart mass and supply of substrates and oxygen to muscles (Milbergue et al., 2018; Swanson, 2010; Swanson and Vézina, 2015; Zhang et al., 2015a).

Only relatively recently have workers investigated seasonal adjustments in BMR and M_sum of birds inhabiting lower latitudes with milder winters and hotter summers, but the limited available data suggest more variability in the magnitude and direction of metabolic adjustments compared with that of their high-latitude counterparts (reviewed by McKechnie et al., 2015). Intraspecific variation in seasonal metabolic acclimatization has also been reported among populations of two subtropical species, Euplectes orix (van de Ven et al., 2013) and Plocepasser mahali (Noakes et al., 2017; Smit and McKechnie, 2010). A continuum between selection for cold tolerance or energy conservation has been suggested to drive patterns of avian metabolic adjustments in response to fluctuations in minimum T_a and food availability, respectively (Smit and McKechnie, 2010), and it is possible that the milder winters at lower latitudes permit greater flexibility in the direction and magnitude of seasonal metabolic acclimatization (McKechnie et al., 2015; Noakes et al., 2017). This does not necessarily mean subtropical and tropical birds have an inherently greater physiological flexibility than their temperate-zone counterparts, but rather that very low winter T_a at high latitudes results in demands for enhanced cold tolerance that simply overwhelm other factors affecting metabolic adjustments (Noakes et al., 2017).

Short-term thermal acclimation experiments are a powerful tool for quantifying the reaction norms (i.e. range of phenotypic traits a single genotype can produce) of phenotypically flexible traits, and can provide information on how phenotypic flexibility varies among and within species. Numerous acclimation studies have investigated flexibility in avian BMR and M_sum in temperate-zone species (McKechnie and Swanson, 2010), with higher metabolic rates typically reported for birds from colder acclimation T_a (T_accl) treatments (generally: T_accl≤5°C; e.g. Barceló et al., 2017; Milbergue et al., 2018; Swanson et al., 2014; Vézina et al., 2017). Fewer studies have investigated metabolic flexibility in lower-latitude birds, with these tending to focus on flexibility of BMR and thermoneutral EWL in response to relatively moderate T_accl treatments (often 15°C versus 30°C) and typically reporting lower BMR and EWL in birds acclimated to higher T_a (e.g. Cavieres and Sabat, 2008; Maldonado et al., 2009; Tieleman et al., 2003).

It has also been hypothesized that greater thermoregulatory flexibility confers adaptive benefits in unpredictable or variable environmental conditions (i.e. temperature, rainfall and/or food abundance; Cavieres and Sabat, 2008; Tieleman et al., 2003). Support for this pattern at the intraspecific level was reported for Zonotrichia capensis from Chile, as the magnitude of flexibility in BMR and EWL in response to thermal acclimation varied among three populations in relation to environmental variability (Cavieres and Sabat, 2008). To the best of our knowledge, no acclimation studies have investigated metabolic flexibility in response to relatively low T_accl (i.e. <10°C) or in M_sum among subtropical birds.

In light of the paucity of studies investigating metabolic flexibility in response to thermal acclimation in lower-latitude birds, and the apparently greater flexibility in patterns of seasonal acclimatization in subtropical taxa, we investigated whether metabolic reaction norms vary across a climatic gradient in a subtropical passerine bird, the white-browed sparrow-weaver (Plocepasser mahali, hereafter sparrow-weaver). Sparrow-weavers were caught from three populations that differ in patterns of seasonal acclimatization in BMR and M_sum (Noakes et al., 2017), although it remains unclear whether these differences reflect interpopulation variation in metabolic reaction norms or simply acclimatization to local environmental conditions.

As typically reported for north-temperate species (McKechnie and Swanson, 2010), we predicted that BMR and M_sum would be highest in sparrow-weavers from the coldest acclimation treatment (T_accl=5°C). We also expected lower BMR and EWL in birds acclimated to the hottest treatment (T_accl=35°C), as reported for other subtropical species (e.g. Cavieres and Sabat, 2008; Maldonado et al., 2009; Tieleman et al., 2003). We predicted that metabolic reaction norms vary among populations according to the environmental variability of local climates (sensu Cavieres and Sabat, 2008), with greater flexibility in sparrow-weavers from the Kalahari Desert site (Askham) where there is more seasonal variation in temperature, as well as less predictable rainfall (Table 1).

We received permission to conduct research from the University of Pretoria Animal Ethics Committee (EC-054-15) and the permitting bodies for the Gauteng (CPF6-000206), Limpopo (ZA/LP/H0/2355), Northern Cape (FAUNA 929/2/2015, FAUNA 1389/2015) and Free State (JM 3286/2018) Provinces of South Africa. We captured adult sparrow-weavers, Plocepasser mahali Smith 1836, during the austral summer (10–20 February 2018) at three sites in South Africa across a climatic gradient: one arid site near Askham in the Kalahari Desert (Northern Cape Province) and two more mesic sites at Frankfort (Free State Province) and Polokwane (Limpopo Province; Table 1). We obtained climate data for each site from the closest weather stations of the South African Weather Service to quantify variation in seasonal temperature extremes and rainfall among these sites (Table 1). Sparrow-weavers were captured (34 birds over two nights at each site) and temporarily housed at capture sites as described by Noakes and McKechnie (2019). We transported birds by road in modified pet carriers back to the Small Animal Physiological Research Facility at the University of Pretoria (25°45′S, 28°15′E). Birds from Frankfort and Polokwane spent about 3 h in transit (∼200 km and 260 km to Pretoria, respectively), whereas the trip from Askham to Pretoria occurred over 2 days (∼6 h travelling per day, ∼1000 km in total).

At the Small Animal Physiological Research Facility, we allocated sparrow-weavers to one of three climate-controlled rooms (∼11 birds from each site per room), all initially set to T_a=25°C and a photoperiod of 12 h:12 h light:dark cycle. We housed birds in cages (600×400×400 mm), typically with two birds per cage (a male and female from the same site, sex determined by bill colour; du Plessis, 2005) but with six of the 102 birds housed individually. Birds were provided with water and wild bird seed mix ad libitum, as well as mealworms and superworms (∼4 of each per bird per day) during the entire period in captivity. We monitored body mass (M_b) of individuals regularly using a Scout Pro Balance scale (SP602US, Ohaus, Pine Brook, NJ, USA) throughout the study period to ensure birds maintained similar values to capture M_b. Sparrow-weavers were released at their capture sites after the completion of experiments.

Three days after the arrival of the birds from the final capture site, we set rooms to T_a=20°C so that all birds were first exposed to the intermediate T_accl regime for 4 days. Following this initial period, we adjusted T_a in two of the rooms to experimental T_accl values of 5 and 35°C over a 4 day period, changing T_a in 5°C increments every 2 days. An acclimation period of 28 days was allowed after experimental T_accl values were set in all rooms, after which post-acclimation metabolic measurements occurred over 63 days. Therefore, the duration of acclimation actually varied among individuals from 28 to 91 days, with 28 days near the upper end of the range of acclimation periods used in previous studies (e.g. 30 days: Cavieres and Sabat, 2008; 21 days: McKechnie et al., 2007; 17 days: McKechnie and Wolf, 2004; 28 days: Tieleman et al., 2003). Actual T_a in each room fluctuated by only a small amount around the setpoint T_accl values over the course of experiments (Table 2).

Open flow-through respirometry was used to measure O₂ consumption (V̇_O2) and CO₂ production (V̇_CO2) in sparrow-weavers, using the same experimental setup and calibration procedures described by Minnaar et al. (2014) and Noakes et al. (2017), with modifications listed below. We used clear plastic containers as metabolic chambers, with volumes of 2.6 l (ClipFresh, Hong Kong, China; incurrent flow rate ∼1 l min⁻¹) and 4 l (Lock & Lock, Seoul, South Korea; incurrent flow rate ∼2 l min⁻¹) for BMR and M_sum measurements, respectively. We placed metabolic chambers in a temperature-controlled cabinet (KMF 720, Binder, Tuttlingen, Germany) for BMR measurements, and in a modified fridge-freezer (40 l; ARB, Kilsyth, VIC, Australia) for M_sum measurements.

We measured T_a in the metabolic chambers using thermistor probes (TC-100, Sable Systems, Las Vegas, NV, USA) during BMR measurements, and a Cu-Cn thermocouple and thermocouple reader (RDXL12SD, OMEGA Engineering, Norwalk, CT, USA) during M_sum measurements, inserted through a small hole in the chamber walls sealed with a rubber grommet. We measured core body temperature (T_b) of sparrow-weavers using temperature-sensitive passive integrated transponder (PIT) tags (Destron Fearing, St Paul, MN, USA), and the tags were calibrated and implanted into the peritoneal cavity of individuals as described by Noakes et al. (2017). We monitored T_b using a racket antenna positioned next to metabolic chambers and connected to PIT tag readers (HPR Plus, Biomark, Boise, ID, USA) during BMR measurements, and a handheld PIT tag scanner (DTR-4, Destron Fearing) during M_sum measurements.

In an attempt to control for the effects of habituation to metabolic chambers after repeated measurements (Jacobs and McKechnie, 2014), we measured M_sum within 12 h before BMR measurements for half of the individuals and during the 12 h after BMR measurements for the rest of the birds. For three individuals we could not measure BMR and M_sum within 12 h of each other, and thus did not calculate metabolic expansibility (ME) for these individuals. We randomized the order in which birds from different populations and acclimation treatments were measured. Our maximum sample sizes were 11 birds from each population per T_accl treatment for BMR and M_sum measurements; however, because of equipment malfunction, our actual sample sizes were often lower (Table 2). Sample sizes for each sex were approximately equal within populations and T_accl treatment (overall n=91 birds, 47 males and 44 females).

We quantified BMR and thermoneutral EWL by measuring gas exchange values for four birds simultaneously at T_a≈30°C (actual T_a=30.1±0.5°C) throughout the night, as this T_a has been reported to be in the thermoneutral zone of sparrow-weavers (Noakes et al., 2017; Smit and McKechnie, 2010). Thermoneutral T_b was concurrently monitored during BMR measurements, but sample sizes for T_b were lower as a result of intermittent PIT tag reception (Table 2). Birds were placed into individual metabolic chambers just before sunset and removed after sunrise, with M_b measured on entry into and removal from the chambers. We sequentially subsampled air from the baseline channel and respective chambers, repeating a cycle lasting 100 min throughout the night: baseline air for 10 min, followed by air from chambers in sequence for 20 min each, and baseline air again for the last 10 min.

We used the sliding cold exposure method in a helox environment (Swanson et al., 1996) to measure M_sum, and the protocol described by Minnaar et al. (2014) and Noakes et al. (2017). We visually monitored O₂ and CO₂ traces throughout measurements for a plateau in increasing metabolic rate indicating that M_sum had been obtained, and terminated measurements when a distinct decrease in metabolic rate away from this plateau was observed. The T_b of birds was measured on removal from chambers to confirm hypothermia, and all individuals decreased T_b by ≥3.7°C from initial values upon entering the chamber (mean T_b decrease=6.2±1.5°C).

We calculated whole-animal metabolic rates (BMR and M_sum), EWL and ME values as described by Noakes et al. (2017). The BMR of each individual represents the lowest 5 min average metabolic rate during overnight measurements (average EWL and T_b were calculated for the same time period), and M_sum is the highest 5 min average metabolic rate during sliding cold exposure measurements. Respiratory exchange ratio (RER; V̇_CO2/V̇_O2) during measurements was often below the expected range of 0.71 to 1.00 (average RER=0.63±0.16), and as thermal equivalence data are not available at RER<0.71 (Withers, 1992), RER=0.71 was assumed to calculate metabolic rates in watts for measurements below this value. To confirm that this assumption was not a source of error in our results, we repeated statistical analyses with V̇_O2 and V̇_CO2 as the response variable. The helox temperature at which M_sum was reached was used as an estimate of the cold limit temperature for each bird (T_CL). All values are presented as means±s.d.

We fitted linear models to our data using R 3.5.2 (http://www.R-project.org/), and checked the assumptions of all models by testing for multicollinearity between continuous predictor variables and inspecting model residual plots (normal Q–Q, residuals versus fitted, scale-location and residuals versus leverage plots). The ‘lm’ base function was used to investigate variation in capture M_b among sexes and sites, as well as variation in post-acclimation M_b, BMR, thermoneutral EWL and T_b, M_sum, ME and T_CL with predictor variables: T_accl treatment, population, M_b (except when M_b was the response variable) and Julian date (to account for varying lengths of acclimation period). The ‘dredge’ function from the MuMIn package (https://CRAN.R-project.org/package=MuMIn) was used to determine the combination of predictor variables (and interactions between them) that produced linear models that best explained variation in the respective response variables, and if multiple top models had a similar fit (Akaike information criterion differed by <2), we chose the most parsimonious model. Sex was included as a predictor variable in initial models, but as it was never included in the top models explaining variation in thermoregulatory variables and as Noakes et al. (2017) also reported no sex differences in this species, sex was excluded from further analyses. If T_accl and/or population was a significant predictor of a response variable, we used post hoc tests of multiple comparisons of means (Tukey contrasts for linear models, multcomp package; Hothorn et al., 2008) to investigate variation among specific T_accl treatments (for M_b, BMR and EWL) and/or populations (for M_b and M_sum).

The initial M_b of sparrow-weavers (i.e. at capture) was significantly higher for males (41.9±3.2 g) than for females (40.4±3.0 g, F_1,88=6.413, P=0.013) across all populations, and varied significantly among populations (F_2,88=7.964, P<0.001). Initial M_b of birds from Askham (39.7±2.9 g) was significantly lower than that of birds from both Frankfort (42.5±3.1 g, P<0.001) and Polokwane (41.4±3.1 g, P=0.034), but did not differ between the last two populations (P=0.315).

Following acclimation to T_a of 5, 20 or 35°C (i.e. T_accl), the M_b of most individuals increased compared with values measured at capture (7.2±8.6% higher at the time of each individual's metabolic measurements), and ranged from a minimum of 37.8 g (an Askham bird at T_accl=35°C) to a maximum of 53.0 g (a Polokwane bird at T_accl=5°C). Post-acclimation M_b did not significantly vary among sexes (F_1,85=1.957, P=0.166), but did vary among populations and acclimation treatments (Table 3). The M_b of birds from Askham was significantly lower than that of birds from Frankfort after acclimation (∼6% lower), but Polokwane sparrow-weavers did not differ significantly from Askham or Frankfort birds (Table 4). Birds acclimated to T_accl=35°C had significantly lower M_b than those acclimated to T_accl=5 and 20°C (∼8% and 7% lower, respectively), but M_b did not significantly differ between the last two treatments (Table 4).

The BMR of sparrow-weavers after acclimation ranged from a minimum of 0.3 W (a Polokwane bird at T_accl=35°C) to a maximum of 0.7 W (an Askham bird at T_accl=35°C). The top model explaining variation in BMR included acclimation treatment and duration of acclimation for each individual, with BMR significantly decreasing at a rate of 3.5 mW per day after the start of the acclimation period (Table 3). BMR was significantly lower in birds acclimated to T_accl=35°C than in those acclimated to T_accl=5°C regardless of population (∼12% lower), but birds at T_accl=20°C did not significantly differ from either of the other acclimation treatments (Fig. 1, Table 4).

Thermoneutral EWL significantly varied among acclimation treatments regardless of population (Table 3), with lower EWL in birds from T_accl=35°C than in those from both T_accl=5 and 20°C (∼25% and 20% lower, respectively), but no significant difference between the last two treatments (Table 4). In contrast, the top model explaining variation in thermoneutral T_b was the null hypothesis (i.e. no predictor variables included; Tables 3 and 4). In summary, thermoregulatory responses measured at thermoneutrality did not vary significantly among populations after acclimation, with BMR and EWL lower in all birds from the hottest acclimation treatment compared with the cooler treatments. Repeating analyses using V̇_O2 and V̇_CO2 instead of BMR (W) as the response variable revealed the same patterns of variation.

The M_sum of sparrow-weavers after acclimation ranged from a minimum of 2.0 W to a maximum of 3.8 W (both Askham individuals at T_accl=20°C). The top model explaining variation in M_sum­ included population and M_b, with M_sum increasing with increasing M_b at a rate of 30 mW g⁻¹ (Table 3). Post-acclimation M_sum was significantly higher in Askham than in both Frankfort and Polokwane birds (∼5% and 8% higher, respectively), but did not significantly differ between the last two populations (Fig. 1, Table 4).

The null hypothesis was the top model explaining variation in the T_CL (i.e. helox temperature at which M_sum was reached), and ME significantly increased with increasing length of acclimation at a rate of 4.0×10⁻² per day (i.e. opposite pattern to that observed in BMR; Tables 3 and 4). In summary, responses during cold exposure did not vary significantly among acclimation treatments, but M_sum­ was higher in Askham birds than in the other populations after acclimation. Repeating analyses using V̇_O2 and V̇_CO2 instead of M_sum (W) as the response variable revealed the same patterns of variation.

Sparrow-weavers showed phenotypic flexibility in M_b, BMR and EWL in response to thermal acclimation, but not in M_sum. In partial support of our predictions, BMR and EWL were lower in birds from the hottest acclimation treatment (T_accl=35°C versus 5 and 20°C), but M_sum did not significantly vary in response to acclimation. In contrast to our predictions, there were similar reaction norms for BMR and EWL in the three populations, although post-acclimation M_sum was higher in Askham birds regardless of T_accl treatment. These findings suggest the considerable variation previously reported for BMR of sparrow-weavers among populations and seasons is the result of phenotypic flexibility to local conditions (Noakes et al., 2017). The consistently higher M_sum of Askham birds, however, suggests that this trait could be fixed in this population as the result of genotypic adaptation, developmental plasticity or epigenetics (Noakes et al., 2017).

We found patterns of among-population variation in initial (i.e. capture) M_b similar to those previously reported for these sparrow-weaver populations (lowest in Askham, intermediate in Polokwane and highest in Frankfort birds; Noakes et al., 2016, 2017). After acclimation, however, we found no significant difference in M_b between Askham and Polokwane sparrow-weavers, suggesting M_b­ variation between these populations is the result of phenotypic flexibility (sensu Noakes and McKechnie, 2019), although higher M_b in Frankfort birds persisted regardless of T_accl treatment (Table 4). Frankfort is one of the coldest locations in the distribution of sparrow-weavers (daily winter minimum T_a regularly <0°C; Table 1) and thus a larger M_b may have an adaptive value for this population following Bergmann's rule (Bergmann, 1847).

Post-acclimation M_b was significantly lower in the hottest treatment (T_accl=35°C) than in the cooler treatments (T_accl=5 and 20°C; Table 4), a similar pattern to that reported among sparrow-weavers acclimated to three different T_a (i.e. lower M_b in T_accl=42°C versus 30 and 36°C; Noakes and McKechnie, 2019). Previous acclimation studies have typically interpreted avian M_b variation with T_accl as responses associated with adjustment to colder T_a (e.g. Barceló et al., 2017; Tieleman et al., 2003; Vézina et al., 2017; Williams and Tieleman, 2000), whereas lower M_b could also provide benefits for birds at hotter T_a (e.g. increased surface area available for passive heat dissipation relative to volume). It is unlikely that M_b variation is a major source of thermoregulatory differences in the present study, as M_b was only a significant predictor of M_sum, which did not significantly vary with T_accl (Table 3).

An unexpected finding was that BMR of individuals significantly decreased with time since the onset of the acclimation period (post-acclimation measurements occurred over 63 days), despite an initial acclimation period (28 days) towards the upper end of the range typically used in previous studies (∼17–30 days; e.g. Cavieres and Sabat, 2008; McKechnie et al., 2007; McKechnie and Wolf, 2004; Tieleman et al., 2003). This finding questions the assumption that avian metabolic rates stabilize after a set period of time at a fixed T_accl, and many previous studies do not consider or provide the duration of post-acclimation measurement periods (e.g. Barceló et al., 2017; Cavieres and Sabat, 2008; Maldonado et al., 2009; Williams and Tieleman, 2000). To the best of our knowledge, only one previous study has included acclimation duration as a continuous variable in analyses of avian metabolic variation, finding no significant change in BMR or M_sum of Poecile atricapillus over 45 days of measurements (initial acclimation period=39 days; Milbergue et al., 2018). We are not certain why BMR decreased with acclimation duration in sparrow-weavers, although it does highlight the limited understanding of the rapidity of avian metabolic adjustments in the literature (McKechnie and Swanson, 2010). We consider it unlikely this trend influenced patterns of variation among T_accl treatments or populations in the present study, as the rate of decreasing BMR with time was consistent among all birds.

Post-acclimation BMR (Table 4) was similar to values measured in sparrow-weavers within 60 h of capture from the same populations during summer and winter (range of mean BMR: 0.38–0.64 W; Noakes et al., 2017). Mean BMR of each population and T_accl treatment was consistently higher than values predicted using allometric equations for tropical (29–42% higher) and passerine (10–18% higher) birds (Table 5; Londoño et al., 2015). Higher than predicted metabolic rates appear to be typical in sparrow-weavers (Noakes et al., 2017; Noakes and McKechnie, 2020), which is unexpected as birds from lower latitudes are generally reported to have a ‘slower pace of life’ than higher-latitude counterparts (Hail, 1983; Weathers, 1979; Wiersma et al., 2007; Londoño et al., 2015).

Similar to findings in temperate-zone species (e.g. Barceló et al., 2017; Milbergue et al., 2018; Vézina et al., 2017), BMR was higher in birds from the coldest compared with the hottest acclimation treatment (i.e. T_accl=5°C versus 35°C); however, it did not significantly vary between the two cooler treatments (i.e. T_accl=5 versus 20°C; Fig. 1, Table 4). In support of our predictions, BMR and thermoneutral EWL were lower in birds at T_accl=35°C compared with cooler treatments (Fig. 1, Table 4), which is consistent with studies on lower-latitude species (Cavieres and Sabat, 2008; Tieleman et al., 2003; Williams and Tieleman, 2000). For example, Z. capensis acclimated to 30°C had lower BMR and EWL compared with individuals acclimated to 15°C (∼16% and 17% lower, respectively), but there was no seasonal variation in BMR or EWL in field-acclimatized birds from the same population in central Chile (Maldonado et al., 2009). In contrast, field-acclimatized sparrow-weavers displayed greater flexibility in BMR (ranging among populations from no seasonal change to ∼59% higher in winter), compared with variation among T_accl treatments (∼12% lower BMR at T_accl=35 versus 5°C; Noakes et al., 2017).

We report no interpopulation variation in post-acclimation BMR or EWL, suggesting different patterns of seasonal BMR acclimatization among populations of sparrow-weavers reflects phenotypic flexibility in response to local environmental conditions (Noakes et al., 2017). This also supports the idea of no clear pattern of BMR and EWL variation among arid versus mesic populations of sparrow-weavers (Noakes et al., 2017), which contrasts with previous studies reporting lower BMR and/or EWL in arid compared with mesic populations of Z. capensis and Haemorhous mexicanus (Cavieres and Sabat, 2008; MacMillen and Hinds, 1998; Sabat et al., 2006). Reduced BMR and EWL in arid-zone birds has been suggested to confer adaptive value by minimizing heat production and conserving water in hot, dry environments (Tieleman et al., 2002; Tieleman and Williams, 2000), and the reduced BMR and EWL in sparrow-weavers from the hottest T_accl treatment probably provided similar benefits.

Post-acclimation M_sum and T_CL were similar to values reported for field-acclimatized sparrow-weavers from the same populations during summer and winter (range of mean M_sum: 2.40–3.86 W; Noakes et al., 2017), and ME was within the range typically reported for birds (∼3–8; Table 4; Swanson, 2010). As was the case for BMR and as reported previously in sparrow-weavers (Noakes et al., 2017; Noakes and McKechnie, 2020), mean M_sum values were higher than predicted using allometric equations for tropical birds (18–24% higher; Table 5; Wiersma et al., 2007). However, post-acclimation M_sum was generally similar to values predicted for oscine passerines (Table 5; Swanson and Bozinovic, 2011).

In contrast to what we expected, M_sum was not higher in sparrow-weavers at the coldest T_accl, and M_sum, T_CL and ME did not vary in response to thermal acclimation (Fig. 1, Table 4). Higher BMR and M_sum have been reported in temperate birds from colder acclimation treatments (McKechnie and Swanson, 2010), but the coldest treatment in the current study (T_accl=5°C) is on the warmer end of the range typically used in previous experiments (cold T_accl range from −10 to 5°C; Barceló et al., 2017; Milbergue et al., 2018; Swanson et al., 2014; Vézina et al., 2017). However, two temperate-zone species (Junco hyemalis and Calidris canutus islandica) from cold-acclimation treatments similar to that of the current study (T_accl=3 and 5°C, respectively) had ∼16–32% higher BMR and/or M_sum than conspecifics from milder treatments (T_accl=24 and 25°C, respectively; Swanson et al., 2014; Vézina et al., 2017). To the best of our knowledge, no previous acclimation studies have investigated flexibility in M_sum, or metabolic values in response to T_accl <10°C, in subtropical birds.

Increases in M_sum are associated with higher cold tolerance in small, north-temperate birds (Cooper, 2002; Petit et al., 2017; Swanson, 2001; Swanson and Liknes, 2006), and M_sum has been reported as flexible in response to short-term (days to weeks) fluctuations in minimum T_a in several higher-latitude species (Dubois et al., 2016; Petit and Vézina, 2014; Swanson and Olmstead, 1999). The lack of flexibility of M_sum in sparrow-weavers in response to thermal acclimation is consistent with the idea that enhancing cold tolerance is of less importance at lower latitudes with comparatively milder winters. Metabolic values of Askham sparrow-weavers measured during the summers and winters over a 4 year period were never related to minimum T_a, but were significantly lower in seasons with lower food abundance (Noakes and McKechnie, 2020). This suggests that fluctuations in food availability in relation to energy conservation are an important driver of metabolic flexibility in subtropical birds, and constant food availability could explain the lack of significant M_sum variation among T_accl­ treatments. Moreover, M_sum can vary as a consequence of other constraints on muscle function; for example, M_sum is typically higher in birds with migratory disposition (Swanson, 2010), and exercise-training of Passer dosmesticus resulted in elevations in both maximal metabolic rates and M_sum (∼19.7% and 15.5% higher, respectively; Zhang et al., 2015b). Another potential explanation for the lack of significant M_sum variation with T_accl in sparrow-weavers could be that M_sum is more coupled with activity levels rather than minimum T_a in this subtropical species.

Post-acclimation M_sum was higher in Askham birds than in the other populations regardless of T_accl treatment (Fig. 1, Table 4), which is consistent with interpopulation variation in field-acclimatized sparrow-weavers (Noakes et al., 2017). Higher M_sum could be a fixed trait in the Askham population associated with cold winters at this desert site (although Frankfort winters are colder; Table 1); however, we cannot identify whether this pattern is the result of genotypic adaptation, developmental plasticity or epigenetics. Moreover, higher M_sum in Askham birds was not associated with higher T_CL compared with the other populations, and variation in M_sum has never been directly associated with T_CL variation in sparrow-weavers (Noakes et al., 2017; Noakes and McKechnie, 2020), suggesting metabolic variation is not primarily associated with enhancing cold tolerance in this subtropical species.

It has been hypothesized that birds from environments with higher variability and/or unpredictability in climates may have greater flexibility in their thermal physiology (Schlichting and Pigliucci, 1998; Tieleman et al., 2003). Support for this pattern was reported at the intraspecific level among three populations of Z. capensis, as flexibility in BMR and EWL in response to T_accl=15 and 30°C varied in relation to variability of local climates (Cavieres and Sabat, 2008). We thus predicted greater physiological flexibility in sparrow-weavers from Askham on account of considerable seasonality in T_a and less predictable rainfall (Table 1), but found similar reaction norms for BMR and EWL among populations and no M_sum variation with T_accl. The occurrence of similar reaction norms among sparrow-weaver populations was also reported for their capacity to cope with high T_a (i.e. thermoregulatory responses at ∼38°C≤T_a≤54°C) during an acclimation study at different T_accl (30, 36 and 42°C; Noakes and McKechnie, 2019). However, reaction norms could vary in a manner outside the scope of what was measured during these studies, such as the upper/lower limits or the rate at which individuals adjust their physiology (McKechnie, 2008). Moreover, thermoregulatory reaction norms are not necessarily fixed in individuals; for example, flexibility in mammalian BMR has been reported to vary among and within individuals between seasons in Phodopus sungorus (Boratyński et al., 2016, 2017).

Little is known about the shape of avian physiological reaction norms as the majority of acclimation studies have only included two T_accl treatments (McKechnie, 2008). The reaction norm for BMR has been reported as approximately linear in Streptopelia senegalensis across T_accl=10, 22 and 35°C (McKechnie et al., 2007), as well as in field-acclimatized Poecile atricapillus at minimum T_a ranging from −20 to 30°C (Petit and Vézina, 2014). In contrast, the BMR reaction norm of sparrow-weavers does not appear to be precisely linear, as BMR was similar in birds from T_accl=5 and 20°C, but lower in individuals from T_accl=35°C (Table 4, Fig. 1). Petit and Vézina (2014) reported the reaction norm of M_sum in field-acclimatized Poecile atricapillus was approximately linear between lower and upper limits (i.e. −10 and 24°C, respectively), whereas the present study gives no information on the shape of the M_sum reaction norm of sparrow-weavers (Table 4, Fig. 1).

We found phenotypic flexibility in BMR and thermoneutral EWL of sparrow-weavers in response to short-term thermal acclimation, with similar reaction norms for BMR and EWL among populations along a climatic gradient. BMR and EWL were lower in birds from the hottest T_accl treatment as previously reported for lower-latitude species (Cavieres and Sabat, 2008; Maldonado et al., 2009; Tieleman et al., 2003). In contrast to studies on higher-latitude birds (McKechnie and Swanson, 2010), M_sum did not vary in response to thermal acclimation. This suggests that factors other than minimum T_a and enhancing cold tolerance may drive patterns of metabolic variation in this subtropical species (e.g. food availability; Noakes and McKechnie, 2020). To the best of our knowledge, this is the first acclimation experiment investigating phenotypic flexibility of M_sum, as well as metabolic rates in response to low T_accl (i.e. <10°C), in a subtropical bird, and more studies are required to determine how avian metabolic flexibility varies with latitude.

We thank the Rossouw and de Klerk families for allowing us to catch birds on their properties. We are grateful to Sekgwari Malematja, Gabriel Foley and Mervyn Uys for assistance in catching birds, and Janca Nortjé and Chuma Mateza for help maintaining birds. We also thank the South African Weather Service for providing climate data.