Rangewide responses of Mimulus cardinalis to an extreme heat event

Abstract

Premise

Extreme events are an understudied aspect of ongoing anthropogenic climate change that could play a disproportionate role in the threat that rapid environmental shifts pose to natural populations.

Methods

We exposed plants originating from seeds that were harvested before (ancestors) and after (descendants) multiple extreme heat events from six populations across the range of Mimulus cardinalis (Phyrmaceae) to a short‐term heat‐wave treatment in controlled growth chamber environments. We assessed physiological, performance, and functional responses (stomatal conductance, leaf temperature deficit, photosystem II efficiency, relative growth rate, specific leaf area, and leaf dry matter content) to the heat‐wave treatment, along with evolutionary responses (differences between ancestors and descendants) of M. cardinalis populations to the recent natural extreme heat event.

Results

Plants in the heat‐wave treatment increased their overall performance, and the magnitude of increase was generally greatest among trailing‐edge populations. Despite limited overall trait differences between ancestors and descendants, there was some evidence of divergent evolutionary responses among regions to the natural extreme heat event. However, we did not find evidence of adaptive evolution that affected how M. cardinalis populations responded to the heat‐wave treatment.

Conclusions

These results demonstrate that many M. cardinalis populations may reside in environments that are below their optimum average temperature, revealing potential resiliency to future warming. However, limited evolutionary responses in M. cardinalis to the recent extreme heat wave could still indicate potential for future vulnerability to extreme climate events of increased intensity, frequency, and duration.

Ongoing climate change is associated with an increase in the frequency, duration, and intensity of extreme events such as floods, storms, droughts, and heat waves (IPCC, ²⁰²²). Extreme climate events (i.e., those with a ≤5% probability of occurring; Smith, ²⁰¹¹) are recognized as critical components of climate change and can be important selective forces for natural populations (Gutschick and BassiriRad, 2003; Grant et al., ²⁰¹⁷; Baeckens and Donihue, ²⁰²⁵). For example, temperature extremes are expected to have severe and widespread ecological impacts (Smith, 2011; Ruthrof et al., ²⁰¹⁸; IPCC, ²⁰²²), yet a disproportionate number of previous studies have focused solely on impacts of the mean temperature increases associated with climate change (Breshears et al., 2021). In terrestrial systems, such temperature extremes can be defined as “heat waves” when three or more consecutive days experience maximum temperatures above the 90th percentile for a particular location at a particular time (Perkins and Alexander, 2013). Global climate models project not only an increase in absolute maximum temperatures, but also increases in the intensity, frequency, duration, and geographic extent of heat waves (Meehl and Tebaldi, 2004; Seneviratne et al., ²⁰¹²; Coumou and Robinson, ²⁰¹³; Coumou et al., ²⁰¹³; Perkins‐Kirkpatrick and Gibson, ²⁰¹⁷; Guerreiro et al., ²⁰¹⁸; Dahl et al., ²⁰¹⁹; IPCC, ²⁰²²). Heat waves in North America have already become more extreme between 1961 and 2021, with increases in average frequency (from two to six per year), duration (from 3 to 4 days), and intensity (from 2.0°C to 2.3°C; NOAA, ²⁰²⁴). Therefore, both past data and future projections point to an urgent need to improve our understanding of population vulnerability to heat waves and the potential for adaptive evolution in traits that promote heat tolerance in response to extreme climate events.

In general, populations exhibit a breadth of thermal performance (i.e., a range of temperatures at which they are considered to be reasonably successful) and a thermal optimum (i.e., a specific temperature at which they are most successful; Huey and Stevenson, ¹⁹⁷⁹; Huey and Kingsolver, ¹⁹⁸⁹). Populations exposed to temperatures outside of their thermal performance breadth should experience a reduction in performance and fitness (Huey and Stevenson, 1979; Huey and Kingsolver, ¹⁹⁸⁹). High temperatures exceeding the thermal performance breadth can result in detrimental effects on numerous physiological processes, including disruption of plant water balance (stomatal closure or excessive water use), inhibition of photosynthetic machinery, accumulation of reactive oxygen species, and slowing of nitrogen metabolism (Hasanuzzaman et al., 2013; Feller and Vaseva, ²⁰¹⁴; Teskey et al., ²⁰¹⁵; Aparecido et al., ²⁰²⁰; dos Santos et al., ²⁰²²). Such extreme heat conditions can negatively impact plant survival and reproduction to the extent that populations may be unable to persist in the face of ongoing climate change (Gutschick and BassiriRad, 2003; Teskey et al., ²⁰¹⁵). Plants can use a variety of strategies to resist these negative impacts, which includes escaping the most extreme periods of high temperature (e.g., by shifting to faster life history; Boyko et al., ²⁰²³; Geissler et al., ²⁰²³; Collins et al., ²⁰²⁵), avoiding detrimental effects of extreme heat (e.g., by altering leaf angle or water use to reduce surface temperature; Farquhar and Sharkey, ¹⁹⁸²; Teskey et al., ²⁰¹⁵; Griffani et al., ²⁰²⁴), or tolerating heat stress (e.g., by possessing molecular heat tolerance mechanisms like scavenging of reactive oxygen species; Hasanuzzaman et al., ²⁰¹³; Feller and Vaseva, ²⁰¹⁴; dos Santos et al., ²⁰²²). As such, extreme heat exerts strong selection on plant populations, with individuals that are more successful in escaping, avoiding, or tolerating the detrimental effects of extreme heat being more likely to survive and reproduce (Gutschick and BassiriRad, 2003; Teskey et al., ²⁰¹⁵; Grant et al., ²⁰¹⁷). Genetic variation in mechanisms of resistance to extreme heat within a plant population therefore presents the possibility of adaptive evolution as an overarching means of population persistence in the face of increasingly detrimental effects of extreme heat events (Orsenigo et al., 2014; Lancaster and Humphreys, ²⁰²⁰; Martin et al., ²⁰²³).

While adaptive evolution can be a key mechanism by which populations cope with climate extremes (Grant et al., 2017; Baeckens and Donihue, ²⁰²⁵), we lack a clear understanding of whether populations across the range of a species can adapt fast enough to persist in the face of more frequent and intense heat waves. Populations across species ranges may vary in their ability to adapt to extreme climate events based on range limit theory (Peischl et al., 2015; Sexton et al., ²⁰¹⁶; Lancaster and Humphreys, ²⁰²⁰). While trailing‐edge populations could better avoid or tolerate extreme heat due to greater historical exposure to higher temperatures (Angert et al., 2011; King et al., ²⁰¹⁹; Chiono and Paul, ²⁰²³), climate warming could reduce population sizes and adaptive potential (Jump et al., 2006, ²⁰¹⁰; but see Sheth and Angert, ²⁰¹⁶). Conversely, under equilibrial range limit theory, populations at the range center are expected to harbor the greatest genetic variation and adaptive potential (Jump and Peñuelas, 2005; Guo, ²⁰¹²; Kremer et al., ²⁰¹²; Latron et al., ²⁰²⁰). An influx of range‐center migrants at the leading edge as the climate warms could therefore boost leading‐edge genetic variation and adaptive potential (Jump and Peñuelas, 2005; Guo, ²⁰¹²; Kremer et al., ²⁰¹²; Latron et al., ²⁰²⁰). However, conditions of range limit theory are generally not in equilibrium due to shifting environments and populations, and recent theory suggests more nuanced (and potentially species‐specific) trends of genetic variation, population connectivity, and adaptive potential across species ranges (Polechová, 2025). For example, key environmental response traits could have low heritability, limiting their adaptive potential (Ahrens et al., 2020; Sheth et al., ²⁰²⁶), or range‐edge populations may be locally adapted to more variable climate conditions and therefore could be integral to a species’ ability to withstand the effects of climate change across its range (Stevens, 1989; Rehm et al., ²⁰¹⁵; King et al., ²⁰¹⁹; Preston et al., ²⁰²²). Comparing responses to a historic heat wave among populations across a species range can therefore provide geographical context to our understanding of how natural populations respond to extreme climate events and their ecological consequences.

A resurrection study is a powerful approach to assess evolutionary responses in natural populations, by simultaneously growing and comparing seeds harvested at different times (i.e., ancestors and descendants; Franks et al., ²⁰⁰⁷, ²⁰⁰⁸, ²⁰¹⁸). We implemented a resurrection experiment in growth chambers and compared responses to a heat‐wave treatment between ancestral (2010) and descendant (2017) cohorts from populations across the geographic range of the scarlet monkeyflower, Mimulus cardinalis Douglas ex Benth (Phyrmaceae; syn. Erythranthe cardinalis; Lowry et al., ²⁰¹⁹; Sheth et al., ²⁰²⁶). We used seeds collected from six populations (two leading‐edge, two range‐center, and two trailing‐edge) before (ancestors) and after (descendants) a multiyear period of extreme heat and drought in western North America (2012–2016; Diffenbaugh et al., ²⁰¹⁵; Robeson, ²⁰¹⁵; Ullrich et al., ²⁰¹⁸). Populations across the range of M. cardinalis may vary in their ability to respond to heat waves based on not only the magnitude of temperature extremes they have experienced, but also their possession of mechanisms to withstand extreme heat (Angert et al., 2011; King et al., ²⁰¹⁹; Lancaster and Humphreys, ²⁰²⁰; Kitudom et al., ²⁰²²; Chiono and Paul, ²⁰²³). Previous work found greater responses to artificial selection on flowering phenology in trailing‐edge populations compared to range‐center or leading‐edge populations and correlated changes in leaf functional traits, which is contrary to expectations based on range limit theory (Sheth and Angert, 2016). Additionally, when measured in controlled greenhouse environments, broad‐sense heritability estimates in several floral and leaf functional and physiological traits in M. cardinalis are relatively high (>0.2; Nelson et al., ²⁰²¹), suggesting that some populations have the capacity to adapt to environmental change. In the present study, we were interested in testing the effects of heat waves on physiological, performance, and functional traits. For example, stomatal conductance of water (g _sw) is an important measure of the rate of gas exchange in plant tissue; high g _sw suggests the potential for high rates of photosynthesis and transpiration, and low g _sw suggests plants may be conserving water and photosynthesizing more slowly (Turner, 1991; Urban et al., ²⁰¹⁷). Stomatal conductance is related to leaf temperature deficit (leaf temperature − air temperature), which indicates not only a physical cooling effect based on heat loss through transpiration, but also a potential adaptive response to avoid heat stress by maintaining leaves at a temperature that is more ideal for photosynthesis (Farquhar and Sharkey, 1982; Teskey et al., ²⁰¹⁵; Griffani et al., ²⁰²⁴). Similarly, high photosystem II efficiency (Φ_PSII, or the proportion of light absorbed by photosystem II that is used for photochemistry) indicates a healthy plant that is conducting photosynthesis efficiently and has the potential for a high relative growth rate, while low Φ_PSII indicates potential plant stress and likely lower relative growth rate (Bilger et al., 1995; Neri et al., ²⁰²⁴). Specific leaf area (SLA) and leaf dry matter content (LDMC) are widely used metrics of plant response to environmental conditions (Pérez‐Harguindeguy et al., 2013). In general, plants in hotter, drier environments produce leaves with lower SLA and higher LDMC and have been reported to recover photosynthetic activity more readily following heat stress (Knight and Ackerly, 2003; Reich, ²⁰¹⁴; Leigh et al., ²⁰¹⁷; Liu et al., ²⁰²³). Using these traits to compare responses of resurrected populations to a heat‐wave treatment can therefore elucidate potential patterns of evolution to extreme climate across the species range.

Our objectives were to (1) quantify responses in plant physiological, performance, and functional traits to a heat‐wave treatment and determine whether those responses varied among populations from across the range of M. cardinalis, and (2) characterize differences in response to a heat‐wave treatment between 2010 ancestors and 2017 descendants, indicative of an evolutionary response to the recent extreme heat and drought event experienced by natural M. cardinalis populations. To address these objectives, we quantified numerous traits expected to be involved in plant response to the surrounding environment, namely g _sw, leaf temperature deficit, Φ_PSII, relative growth rate in leaf number (RGR), SLA, and LDMC (Pérez‐Harguindeguy et al., 2013; Reich, ²⁰¹⁴). For objective 1, if plants are under heat stress, we predicted the greatest reduction in performance in leading‐edge populations, based on their lesser historical exposure and adaptation to high temperatures. Conversely, plant performance could increase under the heat‐wave treatment, reflective of potential escape or avoidance strategies, with young plants experiencing high g _sw to facilitate leaf cooling and high RGR to enable faster life history. For objective 2, if adaptive evolution has occurred due to the recent heat and drought event, descendants should exhibit shifts in plant physiological, performance, and functional traits compared to ancestors. Under a scenario of adaptive evolution in response to stress due to extreme heat, we expected descendants to exhibit shallower negative responses between the heat‐wave treatment and the control than ancestors, indicating greater resistance to high temperatures. Furthermore, because leading‐edge and range‐center populations experienced the greatest magnitude of mean and/or maximum temperature increase between historical conditions and the 2010 to 2017 resurrection timeframe, we expected the smallest reduction in performance (descendants relative to ancestors) to occur in these populations compared to trailing‐edge populations. Conversely, if extreme heat events instead facilitate an escape or avoidance evolutionary response, we would expect descendants (particularly those at the trailing edge) to experience steeper performance increases in response to the heat‐wave treatment. Our results contribute to our understanding of how plant populations adapt to changing environments over space and time and whether evolution can rescue populations that are declining due to climate change and the associated increases in extreme climate events.

MATERIALS AND METHODS

Study system and source populations

Mimulus cardinalis is a perennial herb ranging from southern Oregon, United States (leading range edge) to northern Baja California, Mexico (trailing range edge; Figure 1). It occurs in seasonally wet riparian zones below 2400 m a.s.l. throughout this region. The subject of numerous previous studies, M. cardinalis is a model system for the study of evolutionary, physiological, and demographic responses to climate change (Angert and Schemske, 2005; Angert et al., ²⁰¹¹; Muir and Angert, ²⁰¹⁷; Sheth and Angert, ²⁰¹⁶, ²⁰¹⁸; Anstett et al., ²⁰²¹, ²⁰²⁴; Branch et al., ²⁰²⁴; Kooyers et al., ²⁰²⁵; Sheth et al., ²⁰²⁶). All plant material used for this study originated from M. cardinalis seeds that were previously harvested from each of two leading‐edge (N1 and N2), two range‐center (C1 and C2), and two trailing‐edge (S1 and S2) populations, as described by Sheth and Angert (2016) and Wooliver et al. (2020). From 2012 to 2016, the range of M. cardinalis experienced severe drought and warming (Diffenbaugh et al., 2015; Robeson, ²⁰¹⁵; Ullrich et al., ²⁰¹⁸). In general, leading‐edge and range‐center source populations experienced the greatest increase in mean and/or maximum temperatures compared to historical averages while temperatures at the trailing edge became less variable, with all source populations experiencing periods that satisfy the definition of a heat wave (Figure 2; Appendix S1: Table S1; Perkins and Alexander, ²⁰¹³; Wooliver et al., ²⁰²⁰). Rangewide population growth rates declined from 2010 to 2014 (including N1, C2, and S1 in this study), with the lowest probability of survival at the leading and trailing range edges (Sheth and Angert, 2018). According to demographic data collected from 2010 to 2018, trailing‐edge populations have shorter generation times than range‐center and leading‐edge populations (Anstett et al., 2024), suggesting that trailing‐edge populations have the highest potential for rapid evolutionary change. Previous genetic work including the northern half of the geographic range of M. cardinalis documented a net influx of migrants from warmer environments into leading‐edge populations, which could potentially boost genetic variation in thermal traits and contribute to adaptation to increased heat (Paul et al., 2011). To study these effects, seeds were collected from 80 to 200 individuals from each focal population (representing unique maternal families) in the fall of 2010 (ancestral cohort) and in the fall of 2017 (descendant cohort). A refresher generation was created by crossing plants within each population and cohort, generating 18 full‐sib seed families per population and cohort combination to reduce maternal and seed age effects, as described by Wooliver et al. (2020).

Figure 1.

Map of western United States showing the six focal populations of Mimulus cardinalis used in this study. Populations were sourced from the leading edge (N1 and N2), range center (C1 and C2), and trailing edge (S1 and S2) of the range of M. cardinalis and are set on a backdrop of grey‐scale extreme maximum temperature across a 30‐year period from 1981 to 2010; darker shades represent higher temperatures and lighter shades represent lower temperatures. Extreme maximum temperature gridded data were provided by AdaptWest Project (2022) and developed with the ClimateNA v7.30 software (Wang et al., 2016).

Figure 2.

Historical and recent heat‐wave temperatures for the six focal Mimulus cardinalis populations used in this study. Thick lines for each of the two leading‐edge (N1 and N2), two range‐center (C1 and C2), and two trailing‐edge (S1 and S2) source M. cardinalis populations represent average historical (1990–2010) daily maximum temperatures from 15 May to 15 June each year. Thin lines represent daily maximum temperatures from 15 May to 15 June of each year from 2010 (ancestral cohort) through 2017 (descendant cohort), including all individual years of recent extreme heat events (2012 to 2016). The most extreme temperature across all years and populations (population S2; 4 June 2016/Day 156; 37.7°C) guided the temperature regime for the heat‐wave treatment in our study. All climate data were extracted from PRISM (PRISM Climate Data; PRISM Group, ²⁰¹⁴).

Experimental design

To determine whether M. cardinalis has evolved in response to recent extreme events as a result of climate change, we conducted a resurrection experiment in growth chambers using the refresher generations of ancestral and descendant seed families from the six source populations. We planted seeds into 72‐cell plug trays (one seed per cell; 24 trays) filled with autoclaved Sunshine Mix #4 Aggregate Plus (Sun Gro Horticulture, Agawam, MA, USA), topped with a thin layer of autoclaved cement sand (Southern Products & Silica Co., Hoffman, NC, USA) and thoroughly misted with water. Planting was staggered by tray and occurred from 21 February through 6 March 2024. To reduce the effect of competitive interactions arising from among‐region variation in germination time and growth rate, seeds from each region were planted in separate trays. Each tray contained one plant from each of the 18 seed families per population and cohort combination (18 seed families × 2 populations per region × 2 cohorts per population = 72 plants per tray, arranged randomly). Seedlings were established in a common environment in a walk‐in growth chamber under fluorescent lights (14 h light/10 h dark), with a 20°C day/15°C night temperature regime. We allowed 4 weeks of establishment for populations C1, C2, S1, and S2, and 5 weeks of establishment for populations N1 and N2 due to differences in germination time and growth rate between populations (R. Wooliver and S. N. Sheth, North Carolina State University, personal observations). Trays were sub‐irrigated with a general nutrient solution (containing NPK + 13 micronutrients) three times per week (Mondays, Wednesdays, and Fridays). Twice per week (Tuesdays and Thursdays), we emptied any excess water out of the trays to limit the growth of fungus, algae, and gnat larvae in the soil and rotated trays to minimize positional effects.

The 24 trays were then divided into four groups of six (with each group containing two trays from each of the three regions). Each group of six trays was then transferred into one of four reach‐in growth chambers (Percival LT‐105X, Percival Scientific, Perry, IA, USA) for the heat‐wave experiment. Two chambers were assigned as the heat‐wave treatment, and two chambers were assigned as controls, with the six trays positioned randomly within each chamber. This design ensured that each seed family within each population and cohort combination was replicated four times in each temperature regime, for a total of 1728 plants in the full experiment. Given the early growth stage of the plants used in the experiment (~4 to 5 weeks from planting), temperature regimes mimicked the natural climate from early in the growing season (~15 May to 15 June), with the aim of simulating the most extreme heat wave that occurred between ancestral and descendant cohorts among all source populations. Examination of absolute maximum temperature and extreme daily temperature (95th percentile of maximum) data from the six focal populations from 1991 to 2023 and from 2011 to 2017 (PRISM Climate Data; PRISM Group, 2014) guided our design of control and heat‐wave temperature regimes, respectively (Perkins and Alexander, 2013; Breshears et al., ²⁰²¹; Figure 2). Control chambers were set to 22°C day/8°C night for the entire 8‐day duration of the experiment. From 2011 to 2017, the duration of temperatures above the 95th percentile of maximum temperature in these populations ranged from 1 to 9 days, leading to the selection of a 5‐day heat‐wave treatment, while the absolute maximum temperature from population S2 (37.7°C; 4 June 2016) guided the temperature regime for the heat‐wave treatment. Heat‐wave chambers were set to 22°C day/8°C night for the first 3 days of the experiment to allow for plant acclimation before the 5‐day heat‐wave treatment was applied (38°C day/14°C night). Temperature was ramped up and ramped down over the first 4 h and last 4 h of the day period, at a rate of 2.8°C/h for the 22°C day/8°C night temperature regime and 4.8°C/h for the 38°C day/14°C night temperature regime. All chambers were set to a 14 h light/10 h dark photoperiod (light intensity of ~477 µmol m^–2 s^–1) and light intensity was ramped up and ramped down over the first 2 h and last 2 h of the day period at a rate of ~20% of the daytime value per hour. The experiment was conducted between 25 March to 11 April 2024, with some staggering between chambers that matched the staggered planting dates of each group of six trays, which ensured a consistent developmental stage across all trays within each growth chamber run. During the heat‐wave experiment, trays were sub‐irrigated with reverse osmosis water rather than nutrient solution to eliminate confounding effects of greater soil nutrient uptake if higher temperatures affected plant water use.

Focal traits

We measured six traits as important indicators of the ability of plants to respond to extreme heat while maintaining vital performance: g _sw, leaf temperature deficit, Φ_PSII, RGR in leaf number (hereafter, RGR), SLA, and LDMC. On the final day of the heat wave (day 8), we used a handheld porometer/fluorometer (LI‐600, LI‐COR Biosciences, Lincoln, NE, USA) to measure g _sw, leaf temperature deficit, and Φ_PSII, using the youngest fully expanded leaf at the second or third node and excluding plants that were too small to fill the aperture of the porometer. Comparing g _sw (measured in mmol m⁻² s⁻¹) between control and heat‐wave chambers could confound our results because g _sw increases with temperature due to the temperature dependence of diffusivity of water vapor through stomata (Turner, 1991; Urban et al., ²⁰¹⁷; Griffani et al., ²⁰²⁴; Mills et al., ²⁰²⁴). Therefore, we also calculated the diffusion coefficient of water vapor (D _wv) for each stomatal conductance measurement based on the air temperature in the growth chamber at the time of measurement (to the nearest minute) and assuming constant pressure as:

$$D_{\text{wv}}=2.13 \times {10}^{-5} m^2 s^{-1}\times {\left(\frac{T_{\text{chamber}}}{273.15}\right)}^{1.8}$$

using 2.13 × 10⁻⁵ m² s⁻¹ as the baseline diffusion coefficient for water vapor at 0°C and 1 atm of pressure (Nobel, 2020). For each control plant, we were then able to calculate the expected increase in g _sw based solely on the air temperature increase from the control chamber (22°C) to the heat‐wave chamber (38°C) as:

$$g_{\text{sw},\text{heat wave}}=g_{\text{sw},\text{control}}\times \left(\frac{2.69 \times {10}^{-5} m^2 s^{-1}}{D_{\text{wv}}}\right)$$

using 2.69 × 10⁻⁵ m² s⁻¹ as the diffusion coefficient for water vapor at 38°C and 1 atm of pressure (Nobel, 2020), the diffusion coefficient for water vapor for each control plant calculated above, and the actual g _sw data for each control plant. We calculated the leaf temperature deficit as T _leaf – T _air (i.e., more extreme negative values represent plants that have more drastically reduced surface temperatures compared to the air temperature in the growth chamber), where T _air was based on the internal sensors of the growth chamber and T _leaf was measured by a noncontact infrared thermometer within the porometer. Clamping the porometer onto the leaf has minimal effect on temperature because the focal leaf area is exposed to ambient T _air, light, and humidity. The measurements are fast (10 to 20 s); hence, the stomatal kinetic responses are limited. We measured Φ_PSII at ambient temperature during a rectangular flash using the light‐adapted configuration as: Φ_PSII = ($F_{\mathrm{m}}^{\prime }$ – F _s)/$F_{\mathrm{m}}^{\prime }$, where $F_{\mathrm{m}}^{\prime }$ is the maximal fluorescence and F _s is the steady‐state fluorescence.

To measure RGR, we counted all true leaves >1 mm in length on day 3, 1 day before initiation of the heat‐wave treatment (no. leaves_initial), and on day 9, the day after the heat‐wave treatment concluded (no. leaves_final). To quantify plant performance as the relative change in leaf number during the heat‐wave treatment, we calculated RGR as (no. leaves_final – no. leaves_initial)/(no. leaves_initial × no. days). Although RGR is not a direct measure of reproductive output (i.e. fitness), size is positively correlated with fruit number in natural M. cardinalis populations (Sheth and Angert, 2018) and RGR has been used in past studies as a metric of relative performance (Wooliver et al., 2020). The day after the heat‐wave treatment concluded (day 9), we also harvested the youngest fully expanded leaf at the second or third node from each plant. We measured leaf wet mass with a microbalance (MyWeigh iBalance 311, HBI International, Phoenix, AZ, USA) and leaf area with a bench‐top leaf area meter (LI‐3100C, LI‐COR). Leaves were then dried at 65°C for a minimum of 1 week and weighed with a microbalance (XSR104, Mettler‐Toledo, Columbus, OH, USA) to determine dry mass. We calculated SLA (in m² kg⁻¹) as wet leaf area/dry leaf mass and LDMC (in mg g⁻¹) as dry leaf mass/wet leaf mass. In addition, we measured the length of the longest leaf before initiation of the heat‐wave treatment as a covariate to account for differences in initial size at the start of the experiment. However, length of the longest leaf had no impact on how any independent variable affected any of the measured traits, so it was excluded from the final models.

Statistical analyses

All statistical analyses were conducted in R Version 4.5.0 (R Core Team, ²⁰²⁵) and R Studio Version 2025.05.1 (Posit Team, 2025). The six traits described above served as the response variables of interest. Data for SLA were log‐transformed and data for LDMC were square‐root‐transformed to satisfy assumptions of normality and homogeneity of variance. We ran all general linear mixed effects models using the glmmTMB function in the R package glmmTMB (version 1.1.5; Brooks et al., ²⁰¹⁷). All test statistics and P‐values were calculated using the Anova function in the R package car (version 3.1−2; Fox and Weisberg, ²⁰¹⁹) with Type II sums of squares. During initial analysis, we detected significant temporal residual autocorrelation in the model for leaf temperature deficit using the gls function in the R package nlme (version 3.1−168; Pinheiro et al., ²⁰²⁵). Therefore, we accounted for temporal autocorrelation, assuming exponential decay through time, using the dispformula argument in the glmmTMB function. For all significant interactions, estimated marginal means, custom contrasts, effect sizes, and standard errors used to compare slopes between groups were generated using the emmeans and contrast functions in the R package emmeans (version 1.10.4; Lenth, ²⁰²⁵). Statistical significance between groups was assessed at α = 0.05. Estimated marginal means for SLA and LDMC were back‐transformed to avoid artificial shrinking of effect sizes between groups. All figures were created using the R package ggplot2 (version 3.5.2; Wickham, ²⁰¹⁶).

To analyze physiological traits (g _sw, leaf temperature deficit, and Φ_PSII), we constructed a model including fixed effects of cohort, region, heat‐wave treatment, and all two‐ and three‐way interactions. Family was included as a random effect to account for the non‐independence of replicates with each family. We included time of day as a covariate because of its expected effects on these three physiological traits of interest and because of inconsistencies in the order in which individuals were selected for physiological trait measurement, which could result in confounding of our independent variables of interest. Despite all measurements occurring at the moment growth chamber light conditions reached daytime maximums or later, these inconsistencies in measurement order necessitated a less robust interpretation of the results for leaf temperature deficit and resulted in large standard errors for groups from which many individuals were measured early in the day. We also included two‐ and three‐way interactions between time of day and region and/or heat‐wave treatment to further account for this potential for confounding. However, we avoid direct interpretations of the main effect of time of day or any interactions involving time of day, and we do not anticipate any confounding of independent variables of interest based on time of day to affect our conclusions. To analyze RGR, SLA, and LDMC, we constructured similar models to the one described above, but only including the fixed effects of heat‐wave treatment, cohort, region, and all two‐ and three‐way interactions, with solely family as a random effect because time of day is not expected to affect these variables.

A two‐way interaction between region (leading edge versus range center versus trailing edge) and the heat‐wave treatment (heat wave versus control) would indicate variation in physiological, performance, and functional trait response to extreme heat across the range of M. cardinalis (objective 1). A two‐way interaction between cohort (ancestors versus descendants) and the heat‐wave treatment would indicate an evolutionary response to the recent extreme heat event that affects how M. cardinalis responds to high temperatures (objective 2). A two‐way interaction between cohort and region would indicate divergent evolutionary responses to the recent extreme heat event across the range of M. cardinalis (objectives 1 and 2). A three‐way interaction between cohort, region, and the heat‐wave treatment would indicate divergent evolutionary responses across the range of M. cardinalis that affect how different populations across the range might respond to extreme heat events. Sample sizes used in the models to interpret these interactions are provided in Appendix S1: Table S2. We also included similar models, replacing region with population, in Appendix S1 for completeness (see Appendix S1: Table S3). In general, the two source populations within each region responded in relatively similar ways to the heat‐wave treatment and between cohorts (Appendix S1: Figures S1, S2).

RESULTS

Region, heat‐wave treatment, and their two‐way interaction affected g _sw (Table 1). Heat‐wave plants exhibited greater g _sw than control plants (100 ± 12%). When exposed to the heat‐wave treatment, trailing‐edge plants experienced greater increases in g _sw than both range‐center and leading‐edge plants (151 ± 15%, 86 ± 8%, and 67 ± 35% increases, respectively; Tables 1 and 2, Figure 3A). However, g _sw did not differ between cohorts, and cohort did not exhibit any two‐ or three‐way interactions with region and/or heat‐wave treatment to affect g _sw (Table 1).

Table 1.

Results of mixed effects models assessing how cohort (2010 ancestors or 2017 descendants), region (leading edge, range center, or trailing edge), and heat‐wave treatment (treated or control) and their two‐ and three‐way interactions affected Mimulus cardinalis functional traits. Measured traits include stomatal conductance (g _sw), leaf temperature (T _leaf) deficit, photosystem II efficiency (Φ_PSII), relative growth rate (RGR) in number of leaves, specific leaf area (SLA), and leaf dry matter content (LDMC). For g _sw, T _leaf deficit, and Φ_PSII, we include time of day as a covariate along with select two‐ and three‐way interactions between time of day and region and/or heat‐wave treatment to account for potential confounding due to diurnal variation in these variables. Significant effects (P < 0.05) are listed in bold.

		g sw		T leaf Deficit		ΦPSII
Predictor	df	χ 2	P	χ 2	P	χ 2	P
Cohort (C)	1	2.344	0.126	0.051	0.821	0.046	0.830
Region (R)	2	37.529	<0.001	1832.654	<0.001	58.183	<0.001
Heatwave (H)	1	430.297	<0.001	5701.898	<0.001	447.035	<0.001
Time of day (T)	1	1.371	0.242	302.324	<0.001	29.583	<0.001
C × R	2	0.306	0.858	1.991	0.370	0.886	0.642
C × H	1	3.011	0.083	5.705	0.017	0.322	0.571
R × H	2	16.704	<0.001	198.774	<0.001	51.360	<0.001
R × T	2	2.114	0.348	94.469	<0.001	4.031	0.133
H × T	1	9.322	0.002	166.083	<0.001	30.150	<0.001
C × R × H	2	2.346	0.309	0.812	0.666	0.785	0.675
R × H × T	2	1.751	0.417	9.213	0.010	50.191	<0.001

		RGR in Leaf Number		Specific Leaf Area		LDMC
Predictor	df	χ 2	P	χ 2	P	χ 2	P
Cohort (C)	1	0.281	0.596	1.572	0.210	1.780	0.182
Region (R)	2	70.691	<0.001	44.150	<0.001	34.864	<0.001
Heatwave (H)	1	130.600	<0.001	0.933	0.334	79.103	<0.001
C × R	2	3.639	0.162	19.789	<0.001	12.489	0.002
C × H	1	0.002	0.966	1.137	0.286	0.474	0.491
R × H	2	18.996	<0.001	24.315	<0.001	27.236	<0.001
C × R × H	2	0.016	0.992	0.077	0.962	0.506	0.776

Table 2.

Post‐hoc slope contrasts describing statistical tests of differences in slope for all measured traits among regions when comparing ancestors versus descendants (for significant Cohort × Region interactions only – i.e., for SLA and LDMC) or comparing control plants versus heat‐wave plants (for significant Region × Heat wave interactions), with positive or negative slopes being in reference to the first half of the listed contrast. Significant slope contrasts (P < 0.05) are listed in bold.

	g sw			T leaf Deficit			ΦPSII
Pairwise Slope Contrasts	Slope	SE	P	Slope	SE	P	Slope	SE	P
Region × Heatwave
Leading vs. Center	−0.105	0.111	0.346	−4.450	0.373	<0.001	−0.025	0.039	0.521
Leading vs. Trailing	−0.240	0.116	0.039	−3.940	0.471	<0.001	0.027	0.041	0.510
Center vs. Trailing	−0.135	0.053	0.012	0.511	0.387	0.187	0.052	0.019	0.006

	RGR in Leaf Number			Specific Leaf Area			LDMC
Pairwise Slope Contrasts	Slope	SE	P	Slope	SE	P	Slope	SE	P
Cohort × Region
Leading vs. Center	—	—	—	−0.020	0.039	0.602	−0.016	0.197	0.934
Leading vs. Trailing	—	—	—	0.125	0.038	0.001	−0.570	0.194	0.003
Center vs. Trailing	—	—	—	0.145	0.035	<0.001	−0.554	0.180	0.002
Region × Heatwave
Leading vs. Center	0.021	0.019	0.277	−0.172	0.039	<0.001	0.958	0.187	<0.001
Leading vs. Trailing	−0.052	0.019	0.005	−0.164	0.038	<0.001	0.697	0.183	<0.001
Center vs. Trailing	−0.073	0.017	<0.001	0.007	0.035	0.834	−0.261	0.169	0.123

Figure 3.

The effect of the heat‐wave treatment on physiological, performance, and functional traits of Mimulus cardinalis plants from three regions (leading edge, range center, and trailing edge) within its range. Measured traits include (A) stomatal conductance (g _sw), (B) leaf temperature deficit, (C) photosystem II efficiency (Φ_PSII), (D) relative growth rate (RGR), (E) specific leaf area, and (F) leaf dry matter content. Error bars are ±1 SE around each estimated marginal mean.

Region, heat‐wave treatment, their two‐way interaction, and a two‐way interaction between cohort and heat‐wave treatment affected leaf temperature deficit (T _leaf – T _air; Table 1). Heat‐wave plants exhibited lower leaf temperatures than air temperatures (5.3 ± 0.1°C lower, on average), while control plants exhibited higher leaf temperatures than air temperatures (1.0 ± 0.1°C higher, on average). Overall, leaf temperature deficit was strongly affected by the earlier measurement time of day for many leading‐edge control plants, so it is difficult to make meaningful conclusions from slope contrasts based on the interactions of cohort × heat wave and region × heat wave, so we limit our interpretation to only heat‐wave plants. When exposed to the heat‐wave treatment, ancestors and descendants exhibited similar leaf temperature deficits (5.1 ± 0.2°C, and 5.4 ± 0.1°C, respectively), while trailing‐edge plants exhibited larger average leaf temperature deficits than leading‐edge and range‐center plants (raw means of 6.0 ± 0.3°C, 5.2 ± 0.2°C, and 4.7 ± 0.2°C, respectively; Tables 1 and 2, Figure 3B). Leaf temperature deficits were similar between cohorts, and cohort did not exhibit a two‐way interaction with region or a three‐way interaction with region and heat‐wave treatment to affect leaf temperature deficit (Table 1).

Region, heat‐wave treatment, and their two‐way interaction affected Φ_PSII (Table 1). Heat‐wave plants exhibited greater Φ_PSII than control plants (8 ± 2%). When exposed to the heat‐wave treatment, range‐center plants experienced greater increases in Φ_PSII than both leading‐edge and trailing‐edge plants (13 ± 2%, 8 ± 6%, and 4 ± 2% increases, respectively; Tables 1 and 2, Figure 3C). There were no differences in Φ_PSII between cohorts, and cohort did not exhibit any two‐ or three‐way interactions with region and/or heat‐wave treatment to affect Φ_PSII (Table 1).

Region, heat‐wave treatment, and their interaction affected RGR (Table 1). Heat‐wave plants exhibited greater RGR than control plants (41 ± 4%). When exposed to the heat‐wave treatment, trailing‐edge plants experienced larger increases in RGR than leading‐edge and range‐center plants (61 ± 6%, 46 ± 9%, and 22 ± 5% increases, respectively; Tables 1 and 2, Figure 3D). RGR did not differ between cohorts, and cohort did not exhibit any two‐ or three‐way interactions with region and/or heat‐wave treatment to affect RGR (Table 1).

Region and two‐way interactions between region and either cohort or heat‐wave treatment affected SLA (Table 1). For leading‐edge and range‐center plants, descendants had higher SLA than ancestors (6 ± 3% and 8 ± 3%, respectively), while for trailing‐edge plants, descendants had lower SLA than ancestors (6 ± 2%; Tables 1 and 2, Figure 4A). When exposed to the heat‐wave treatment, range‐center and trailing‐edge plants increased in SLA (both 6 ± 3%), while leading‐edge plants decreased (10 ± 3%; Tables 1 and 2, Figure 3E). There were no differences in SLA between cohorts or heat‐wave treatments, and heat‐wave treatment did not interact with cohort or both cohort and region to affect SLA (Table 1).

Figure 4.

The effect of cohort (ancestors or descendants) on functional traits of Mimulus cardinalis plants from three regions (leading edge, range center, and trailing edge) within its range. Measured traits include (A) specific leaf area and (B) leaf dry matter content. Error bars are ±1 SE around each estimated marginal mean.

Region, heat‐wave treatment, and two‐way interactions between region and either cohort or heat‐wave treatment affected LDMC (Table 1). Heat‐wave plants exhibited lower LDMC than control plants (10 ± 1%). For leading edge and range‐center plants, descendants had lower LDMC than ancestors (both 5 ± 2%), while for trailing‐edge plants, descendants had higher LDMC than ancestors (5 ± 2%; Tables 1 and 2, Figure 4B). When exposed to the heat‐wave treatment, range‐center and trailing‐edge plants exhibited decreases in LDMC (15 ± 2%, and 12 ± 2%, respectively), while leading‐edge plants were not affected (Tables 1 and 2, Figure 3F). There were no differences in LDMC between cohorts, and heat‐wave treatment did not interact with cohort or both cohort and region to affect LDMC (Table 1).

DISCUSSION

We investigated responses to multiple extreme heat events in populations from across the range of M. cardinalis by testing for changes in plant physiological, performance, and functional traits of resurrected ancestor (before the natural heat wave) and descendant (after the natural heat wave) populations exposed to a heat‐wave treatment. Heat‐wave plants demonstrated greater physiological performance (increased g _sw and Φ_PSII; Figure 3A, C) and faster development of cheaper leaves (increased RGR and decreased LDMC; Figure 3D, F) compared to control plants, which is indicative of a lack of heat stress, potentially due to a shift toward escape and/or avoidance strategies. Responses to the heat‐wave treatment also varied among regions, with populations from the trailing edge of the range of M. cardinalis exhibiting the greatest positive response for g _sw and RGR (Figure 3A, D), but not for Φ_PSII (Figure 3C), while SLA and LDMC diverged among regions in response to extreme heat (Figure 3E, F). Ancestors and descendants did not differ in any measured traits, overall (Table 1), potentially indicating a lack of evolutionary response to the recent heat wave in western North America. However, among regions, there was some evidence of divergent responses between ancestors and descendants. Namely, leading‐edge and range‐center descendants developed thinner and cheaper leaves per unit area, while trailing‐edge descendants developed thicker and more expensive leaves per unit area, relative to ancestors (Figure 4). Ancestral and descendant cohorts did not differ in the magnitude of their response to the heat‐wave treatment (Table 1), however, indicating that despite divergent evolutionary responses between regions, descendants did not adapt in a manner that is likely to increase their ability to withstand future heat waves. Here we discuss our results with respect to both plant‐level trait responses and population‐level evolutionary responses to extreme heat events.

Plant trait responses to heat‐wave treatment

We observed a strong and positive response to the heat‐wave treatment in terms of increased stomatal conductance and photosystem II efficiency, along with increased growth rate of cheaper leaves. In general, heat stress causes stomatal closure and inhibits photosynthetic enzymes, so it is possible that the heat‐wave treatment was not severe enough to elicit a stress response (Hasanuzzaman et al., 2013; Feller and Vaseva, ²⁰¹⁴; Teskey et al., ²⁰¹⁵; dos Santos et al., ²⁰²²). Instead, in the absence of stress‐induced stomatal closure, g _sw will increase with temperature due to the physical effects of temperature‐dependent diffusivity, as well as plant‐level responses to a rise in temperature, such as increased stomatal density or stomatal opening (Turner, 1991; Urban et al., ²⁰¹⁷; Griffani et al., ²⁰²⁴; Mills et al., ²⁰²⁴). Overall, we calculated the physical effects of temperature‐dependent diffusivity to be responsible for ~13% of the increase in g _sw for heat‐wave plants (Appendix S1: Figure S3), demonstrating that plant physiological responses to the temperature increase are predominantly responsible for the increased performance for heat‐wave plants in terms of gas exchange (Hasanuzzaman et al., 2013; dos Santos et al., ²⁰²²). Increased g _sw also represents a heat avoidance strategy by providing a cooling effect to leaves that could prevent heat stress from occurring (Farquhar and Sharkey, 1982; Teskey et al., ²⁰¹⁵; Griffani et al., ²⁰²⁴). In our experiment, control growth chambers were set to a daytime temperature of 22°C, and the plants therein had an average leaf temperature of 21.4°C, while heat‐wave chambers were set to a daytime temperature of 38°C, and the plants therein had an average leaf temperature of 31.6°C, demonstrating a substantial cooling effect. It is therefore possible that leaf cooling through stomatal conductance (i.e., latent heat loss through transpiration) allowed heat‐wave plants to avoid negative physiological effects of high temperatures, while still allowing photosynthetic machinery to operate at a faster rate compared to control plants that were maintained at a temperature that was below optimal in controlled conditions (ambient CO₂ and sufficient light; Farquhar and Sharkey, ¹⁹⁸²; Teskey et al., ²⁰¹⁵; Griffani et al., ²⁰²⁴).

Within the general trend of increased performance for heat‐wave plants, individuals from the trailing edge of the range of M. cardinalis often experienced the most substantial performance increase relative to control plants. The relatively high increase in g _sw and RGR in trailing‐edge plants is consistent with the natural environment in which these source populations are found. In general, trailing‐edge populations (in our study, population S2, specifically; Appendix S1: Table S1) have experienced historic and recent climate conditions that are warmer than those of range‐center and leading‐edge populations, which led to the prediction that these populations would exhibit greater resistance to the heat‐wave treatment (Angert et al., 2011; King et al., ²⁰¹⁹; Chiono and Paul, ²⁰²³). Conversely, leading‐edge populations exhibit greater plasticity of thermally regulated gene expression, which could reduce their vulnerability to future climate fluctuations (Preston et al., 2022). In addition, trailing‐edge populations of M. cardinalis often exhibit a more annualized life history strategy, with rapid, early growth and overall greater vegetative investment, which could be further amplified if plants adopt a faster life history strategy to escape heat stress (Muir and Angert, 2017; Nelson et al., ²⁰²¹). By contrast, heat waves in southern California often occur late in the growing season, possibly having a substantial impact on reproductive output, which may not have been captured by our study of responses to a heat wave by plants at an early development stage. A future study involving experimental plants grown for an entire reproductive cycle (including a measure of reproductive fitness) and subjected to a heat wave of greater intensity or duration would provide more definitive insight into divergent responses to extreme heat in leading‐edge, range‐center, and trailing‐edge populations.

Environment‐specific evolutionary responses to a recent heat wave

Across all measured traits, we observed a lack of overall differences between ancestors and descendants. These negligible evolutionary differences are not surprising given that heritability estimates for the same traits in the same six M. cardinalis source populations are low when measured in a realistic field setting, indicating limited evolutionary potential (Sheth et al., 2026). Additionally, a recent meta‐analysis of resurrection studies found lesser trait differences in perennials (like M. cardinalis) compared to annuals (Pennington et al., 2025), which is consistent with our findings. However, it is important to highlight that an overall effect of cohort on SLA and LDMC could have been masked by descendants from each region responding in different directions compared to ancestors, with trailing‐edge descendants trending toward more expensive leaves and leading‐edge and range‐center descendants trending toward cheaper leaves. In general, more expensive leaves (typically smaller and thicker) tend to be selected for in hotter environments (i.e., the trailing edge), while populations that are less adapted to high temperatures may opt for more conservative resource investment strategies, which is in line with our results (Knight and Ackerly, 2003; Reich, ²⁰¹⁴; Leigh et al., ²⁰¹⁷; Liu et al., ²⁰²³). However, the focal M. cardinalis populations all experienced not only extreme heat, but also extreme drought between 2012 and 2016, which could have substantially affected selection on plant traits, resource investment, and thermal tolerance (Feller and Vaseva, 2014; Teskey et al., ²⁰¹⁵; Anstett et al., ²⁰²¹; Cook et al., ²⁰²¹; dos Santos et al., ²⁰²²). All experimental plants in this study were kept well‐watered, so it is difficult to identify the relative contributions of extreme temperature and soil moisture conditions on plant traits in order to make broad conclusions regarding the evolutionary response to extreme climate events in this system.

The heat‐wave treatment was also unsuccessful in eliciting a heat stress response in our experimental plants, which could have masked our ability to detect an evolutionary response between ancestor and descendant cohorts (i.e., both cohorts experienced substantially higher performance in the heat‐wave treatment). In fact, thermal performance curves for the same source populations as the present study predict greater performance (in terms of RGR) at 38°C than at 22°C for all populations (Wooliver et al., 2020). This result is consistent with classic thermal performance curve hypotheses, which posit that organisms are optimally placed at “suboptimal” temperatures on a thermal performance curve, due to its left‐skewed asymmetry, which results in greater risk of large declines in fitness at relatively modest temperature increases compared to lower risk at similarly modest temperature decreases (Martin and Huey, 2008). With only a single heat‐wave treatment, we could only assess linear responses of plant traits to temperature in the current study. However, our results together with M. cardinalis thermal performance curves can provide insight into how our traits of interest might respond to fluctuations in temperature nonlinearly. Specifically, thermal performance curves demonstrate that some ancestral cohorts of M. cardinalis had thermal optima that were warmer than their historic maximum July temperatures (Wooliver et al., 2020), indicating that M. cardinalis populations may have already adapted to high yearly average and extreme temperatures before 2010 (Appendix S1: Table S1). Additionally, recent heat waves may not have exposed M. cardinalis populations to temperatures that extend above the upper range of their thermal performance breadth, reducing the likelihood of a strong evolutionary response to those climate extremes. However, as temperatures continue to rise, the thermal performance breadth of M. cardinalis populations could become narrower in the absence of adaptation to warming, indicating the potential for future vulnerability to high temperatures (i.e., a lack of evolutionary rescue in response to climate change; Wooliver et al., ²⁰²⁰; Kitudom et al., ²⁰²²). Alternatively, if substantial adaptation to warming (or extreme heat) occurs in response to both recent and future heat waves, thermal performance breadths of plant populations could expand, increasing the likelihood of evolutionary rescue (Ahrens et al., 2021).

Thermal optima in controlled conditions may not be representative of the climate optima for populations in their natural environments, which are likely to be less predictable due to natural environmental variability and the involvement of numerous additional environmental variables. For example, water availability for riparian species like M. cardinalis is likely to decline throughout the growing season, while plants in this experiment remained well‐watered throughout the entire heat‐wave treatment, removing the potential for negative effects of stomatal opening on water loss under drought conditions (Cook et al., 2021; Posch et al., ²⁰²⁴). In natural populations, heat waves are likely coupled with low soil moisture (Ullrich et al., 2018; Maraun et al., ²⁰²⁵), and a follow‐up study that crosses heat‐wave and drought treatments would enable a more thorough investigation of the interactive effects of extreme temperature and soil moisture on plant performance. Additionally, the heat‐wave treatment may have been mistimed with what is likely to be experienced by plants at the specific developmental stage of those used in the experiment, so we may not have fully captured how selection due to rare, episodic extreme climate events acts on natural populations. Therefore, a future experiment including a more comprehensive array of heat‐wave and/or drought treatments and plant developmental stages could further elucidate how populations will respond to realistic extreme climate events that are likely to increase in intensity, frequency, duration, and geographic extent over time (Meehl and Tebaldi, 2004; Seneviratne et al., ²⁰¹²; Coumou and Robinson, ²⁰¹³; Coumou et al., ²⁰¹³; Perkins‐Kirkpatrick and Gibson, ²⁰¹⁷; Guerreiro et al., ²⁰¹⁸; Dahl et al., ²⁰¹⁹; IPCC, ²⁰²²).

CONCLUSIONS

Extreme climate events can play an important role in imposing selection on natural populations and could increasingly threaten the maintenance of biodiversity if populations are unable to cope with their frequency, duration, and intensity in the future. Overall, we found limited differences between ancestral and descendant cohorts of M. cardinalis harvested before and after multiple extreme heat and drought events in western North America, restricting our ability to make predictions about how M. cardinalis populations will respond to future extreme heat events. However, we did find divergent trends in leaf functional traits between cohorts for populations from different areas of the range of M. cardinalis, indicating that evolutionary responses to extreme climate likely depend on the environment. We also found that across numerous traits, plant performance increased for heat‐wave plants and often increased most substantially in plants from the warmest local environments at the trailing edge of the range. This result indicates that either recent heat waves have not been extreme enough to impose strong selection on M. cardinalis populations, potentially mediated by historical adaptive evolution to extreme heat and/or heat resistance strategies, such as leaf cooling through stomatal conductance, leaf angle, or various molecular mechanisms (Hasanuzzaman et al., 2013; dos Santos et al., ²⁰²²). Many M. cardinalis populations therefore likely reside in natural environments that are below their optimum average temperature to avoid modest temperature increases posing a substantial threat to survival and fitness. However, the expanding intensity, frequency, and duration of heat waves, compounded with associated effects of drought, could result in sustained and/or repeated high temperatures that pose an increasing threat to performance and fitness if populations are unable to undergo evolutionary rescue, highlighting the importance of continued study of the effects of extreme climate events on natural populations.

AUTHOR CONTRIBUTIONS

R.A.B. and S.N.S. conceptualized and designed the project; R.A.B., S.C., C.F.L., and C.P. performed the experiment and collected all data; L.J.A., C.D.M., and S.N.S. analyzed the data; L.J.A., R.A.B., and S.N.S. wrote the manuscript; L.J.A., C.D.M., and S.N.S. edited and revised the manuscript with input from all coauthors.

Supporting information

Appendix S1. Supplementary tables and figures.

Figure S1. The effect of the heat‐wave treatment on the physiological, performance and functional traits of Mimulus cardinalis plants from the two leading‐edge (N1 and N2), two range‐center (C1 and C2), and two trailing‐edge (S1 and S2) source populations.

Figure S2. The effect of cohort (ancestors or descendants) on functional traits of Mimulus cardinalis plants from the two leading‐edge (N1 and N2), two range‐center (C1 and C2), and two trailing‐edge (S1 and S2) source populations.

Figure S3. The effect of the heat‐wave treatment on stomatal conductance (g _sw) of Mimulus cardinalis plants from the three regions within its range.

Table S1. Location and climate data for the six focal Mimulus cardinalis populations used in this study.

Table S2. Sample sizes for each of the six response variables, grouped by region, cohort, and heat‐wave treatment used in this study.

Table S3. Results of mixed effects models assessing how cohort, population, and heat‐wave treatment and their interactions affected Mimulus cardinalis physiological, performance, and functional traits.

Albano, L. J. , Bingham R. A., Correa S., Laufenberg C. G., Payst C., Muir C. D., and Sheth S. N.. 2026. Rangewide responses of Mimulus cardinalis to an extreme heat event. American Journal of Botany 113(2): e70145. 10.1002/ajb2.70145

DATA AVAILABILITY STATEMENT

The data set generated and analyzed during this study and the associated code for statistical analyses are in the Dryad Digital Repository (https://doi.org/10.5061/dryad.905qfttx8).