PREMISE: The study of phenotypic divergence of, and selection on, functional traits in closely related taxa provides the opportunity to detect the role of natural selection in driving diversification. Where selection in field populations differs between taxa in a pattern that is consistent with the phenotypic difference between them, this provides evidence that natural selection reinforces the divergence. Few studies have sought evidence for such concordance for physiological traits.

METHODS: Herbarium specimen records were used to detect phenological differences between sister taxa independent of the effects on flowering time of long-term variation in the climate across collection sites. In the field, physiological divergence in photosynthetic rate, transpiration rate, and instantaneous water use efficiency were recorded during vegetative growth and flowering in 13 field populations of two taxon pairs of Clarkia, each comprised of a self-pollinating and a outcrossing taxon.

RESULTS: Historically, each selfing taxon flowered earlier than its outcrossing sister taxon, independent of the effects of local long-term climatic conditions. Sister taxa differed in all focal traits, but the degree and (in one case) the direction of divergence depended on life stage. In general, self-pollinating taxa exhibited higher gas exchange rates, consistent with their earlier maturation. In 6 of 18 comparisons, patterns of selection were concordant with the phenotypic divergence (or lack thereof) between sister taxa.

CONCLUSIONS: Patterns of selection on physiological traits measured in heterogeneous conditions do not reliably reflect divergence between sister taxa, underscoring the need for replicated studies of the direction of selection within and among taxa.

Study species—Clarkia unguiculata Lindley and C. xantiana ssp. xantiana A.Gray are self-compatible winter annuals that flower in late spring, are endemic to California, and occupy oak and pine woodlands, grasslands, and roadside habitats, California. While C. xantiana ssp. xantiana is restricted to the southern Sierra Nevada and Transverse ranges to the south, the geographic range of C. unguiculata is much more widespread, occupying both the Sierra Nevada and the Coastal and Transverse Ranges (Lewis and Lewis, 1955). In the Lake Isabella region of the southern Sierra Nevada (Kern and Tulare counties), both taxa may occur alone or in sympatry with their predominantly selfing sister taxa (C. exilis and C. xantiana ssp. parviflora, respectively). Where sister taxa co-occur, reproductive isolation is reinforced by differences between sister taxa in flowering phenology and floral morphology, with selfers flowering earlier and showing greater levels of herkogamy and dichogamy than their bee-pollinated outcrossing counterparts (Eckhart and Geber, 1999; Dudley et al., 2007).

            In the southern Sierra Nevada, the sister taxa studied here differ with respect to habitat. Although C. exilis is sometimes encountered at the same sites as C. unguiculata, when these species are in sympatry, C. unguiculata occupies a broader range of habitats, growing in woodlands and encroaching into grasslands. By contrast, C. exilis has a more patchy distribution; it is generally restricted to relatively cool canyons, stream beds, and ravines, or is found near boulders or rocky outcrops that provide some shade and/or soil moisture (Vasek, 1958). The join distributions of C. xantiana ssp. xantiana and ssp. parviflora follow a west to east gradient, with xantiana occurring in populations to the west and parviflora occurring towards the east, nearly reaching the Mojave desert. Although there is a narrow central zone of contact between these subspecies (Eckhart et al., 2004; Geber and Eckhart, 2005), sites occupied solely by ssp. parviflora receive less precipitation than those occupied solely by ssp. xantiana.

Historical climatic and phenological differences between sister taxa: herbarium-based study—Occurrence data and the GPS coordinates of all specimen records available for each of the focal taxa were downloaded from the Consortium of California Herbarium portal (https://cch2.org) on July 31, 2021 (Fig.1). The date of specimen collection was converted to day-of-year (DOY: 1-365; 366 for leap years). The GPS coordinates of each collection site were then used to extract from the climate database, ClimateNA v.7.01 (Wang et al., 2016) the mean values of 23 annual climatic variables estimated from 1951-1980, the 30-year period preceding the point at which climate warming is considered to have begun to accelerate (IPCC, 2013). The climate variables estimated are defined in Appendix S1 (see Supplemental Data with this article).  To estimate climatic parameters, ClimateNA uses interpolated values from the PRISM climate database (PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu) to generate scale-free estimates of local climatic conditions at a resolution of 4km x 4km.

The DOY of specimen collection does not provide a precise record of the flowering date of a specimen because an individual plant collected in flower may have been collected at any time after flowering has begun, and the duration of an individual’s flowering period may be several weeks or longer. Nevertheless, the date of specimen collection has been found to be a reliable proxy for flowering time (Davis et al., 2015), and has been used in many studies to detect the factors influencing flowering phenology (Davis et al., 2015; Matthews and Mazer, 2016; Willis et al., 2017; Ferris and Willis, 2018; Park and Mazer, 2018, 2019; Ellwood et al., 2019; Love et al., 2019; Park et al., 2019; Mazer et al., 2021; Park, Breckheimer, et al., 2021; Park, Ramirez‐Parada, et al., 2021; Pearson et al., 2021). Among the focal taxa of this study, images were available from CCH2 on 25 December 2021 for 26, 679, 10, and 33 specimens of C. exilis, C. unguiculata, C. xantiana ssp. parviflora, and C. xantiana ssp. xantiana, respectively. Each image was visually inspected to determine whether the individual plant(s) on each sheet bore one or more open flowers. Of these, 24 (92%), 659 (97%), 8 (80%), and 30 (91%) specimens bore one or more open flowers, indicating that the DOY of specimen collection for these taxa generally reflects the specimen’s flowering time at a given site.

Statistical analysis of herbarium-derived data and climate—To assess whether sister taxa differ with respect to mean flowering time independent of the effects of local climatic conditions on flowering phenology due to evolutionary and/or short-term plastic responses to climate, we controlled statistically for climatic variation among collection sites when testing for differences between taxa in mean flowering time. To characterize each collection site using a multivariate index of its climate, principal components (PCs) for the 24 annual climatic variables were estimated using the pooled data for each pair of sister taxa. In both pairs of taxa, the first two principal components (PC1 and PC2) for the 24 climatic parameters together account for >75% of the variance in climate (Appendix S2). In the pooled data for Clarkia unguiculata and C. exilis, PC1 and PC2 account for 46% and 31% of the variance, respectively; in Clarkia xantiana ssp. xantiana and ssp. parviflora, PC1 and PC2 explain 62% and 21% of the variance, respectively (Appendix S2). In both taxon pairs, across their geographic distribution, the loadings of PC1 indicate that it represents an axis dominated by temperature, while PC2 represents an axis dominated by humidity (Appendix S3).

For each data set, the first two PCs were used to determine whether sister taxa differ with respect to their flowering phenology (controlling for site-specific climatic conditions) and in the chronic climatic conditions to which they are exposed. The function prcomp() was used to estimate PCs, following centering and scaling of the data to unit variance; the function fviz_pca_var() in the package factoextra (v.1.0.7) was used to visualize the contributions of each variable in each PC. These and all subsequent analyses were conducted in RStudio (v. 1.4.1717) with R v. 4.1.1.

Phenological differences between sister taxa—To determine whether sister taxa differ in their mean flowering date while controlling for the effects of geographic variation in local chronic climatic conditions (which influence DOY; see Results), we conducted four linear regressions on each sister pair. In all of these models, the day of year of specimen collection (DOY) was the response variable (Table 1).  In neither taxon pair were either of the two-way interactions (Taxon x PC1 or Taxon x PC2) statistically significant. Below, we present the results for the model with the lowest AIC value. Type III sums of squares were used to detect a significant difference between taxa in their mean date of collection, controlling for variation among collection sites with respect to the covariates. The lm function with the default setting for contrasts and the lme4 package (v. 1.1-27.1) were used.

Climatic differences between sister taxa —To determine whether sister taxa differ with respect to the chronic climatic conditions at their respective herbarium specimen collection sites, we used linear models to detect statistically significant differences between taxa with respect to the mean values of the first and the second PCs.  In these analyses, either PC1 or PC2 was the response variable, and taxon was the predictor variable. The lm function (using the default setting for contrasts: c[contr.treatment, contr.poly]) in the lme4 package (v. 1.1-27.1) was used.

 

Contemporary differences between sister taxa: phenotypic divergence in physiological traits—Sampling of extant populations: sample sizes and life stages—Naturally occurring field populations were surveyed in the region of Lake Isabella, California in the southern Sierra Nevada, where we sampled populations at elevations ranging from 457 to 1628 m. In spring 2010, 13 sites were identified that, at the onset of spring, contained a population of one or more of our focal taxa with a sufficient number of healthy juvenile plants that we judged we would be able to survey at least 90 individuals for their physiological performance during both vegetative and flowering stages (Fig.2; see Appendices S4 and S5 for a summary of each population’s GPS coordinates, elevation, sampling dates, and sample sizes). Each population was surveyed once during vegetative growth and once during flowering in spring 2010; a different group of individuals was sampled at each life stage. For each survey, 90 individuals were selected at random positions along one or more transects traversing the population, with the restriction that a surveyed plant had to bear at least one leaf long enough to sample with a LiCoR-6400 gas exchange analyzer (leaves that were too short could not cross the leaf chamber when it was closed). Sampled plants were typically at least 1m apart from one another. In total, we measured the physiological performance of  >1800 individuals. We analyzed and compared these traits while controlling statistically for leaf position on the primary stem and for leaf temperature at the time of measurement, both of which can influence measures of physiological performance (Mazer et al., 2010). 

Physiological survey—Gas exchange measurements were recorded on leaves produced by the primary branch on each plant. Measured leaves remained attached to the stem while using a portable infrared gas exchange analyzer (IRGA; LI-COR 6400, Lincoln, Nebraska, USA). We recorded photosynthetic rate (A; μmol CO₂ m^-2 s^-1), transpiration rate (E; mol H₂O m^-2 s^-1), and the derived parameter, A/E, instantaneous water use efficiency (WUE_i;  μmol_CO2 mol_H2O^-1 x 0.0001). In addition, the position of each measured leaf, recorded as the number of the node that bore it, was recorded; the node just above the cotyledons was referred to as node 1, with subsequently produced nodes counted in sequence. The temperature of the leaf (°C) at the time of measurement was also recorded. Node number and leaf temperature were included as covariates in our statistical models in order to control for their potential effects on physiological rates, as observed by Mazer et al. (2010). The following LI-COR settings were used: light source, 6400-40 fluorometer; stability, values of CO₂ and H₂O remained stable for at least 15 s with a change in slope of <1; stomatal ratio = 1; flow = 500 μmol s^-1; PAR_i (PAR in the leaf chamber) = 1500 mol m^-2 s^-1. CO₂ concentration in the IRGA reference chamber was 400 ppm.  

For leaves too narrow to fill the entire 2cm² circular area exposed within the LI-COR 6400’s leaf chamber, the actual leaf area exposed was calculated and then the LI-COR 6400’s Simulator Software was used to adjust those physiological parameters that were influenced by leaf area. To measure the leaf area exposed within the chamber, we used an ink pad to apply black ink to the foam gasket of the chamber; when the chamber was then closed upon the leaf, the gasket stamped with ink the outer boundary of the leaf’s area that was exposed within the chamber. Following gas exchange measurements, each leaf that failed to fill the chamber’s circular opening was removed, taped to a piece of paper with transparent tape, and subsequently scanned using a digital scanner. We then used the image analysis program ImageJ (http://rsb.info.nih.gov/ij/index.html) to measure the area of each leaf that had been exposed within the LI-COR chamber during gas exchange measurement.     

Instantaneous gas exchange measurements and WUE_i provide point estimates of the physiological performance of individual plants and are not equal to the same parameters measured over the entire lifespan of individuals (Zelitch, 1982; Serichol-Escobar et al., 2016).  In the current study, however, the large sample sizes obtained per taxon, the replication of each taxon within and among 2-4 populations, and the statistical control of variation in other plant attributes that are correlated with instantaneous gas exchange rates (e.g., leaf position on the stem, leaf temperature at the time of measurement, and the life cycle stage at which plants are measured) helped to account for the variance in physiological performance that can obscure differences between taxa. Numerous other recent studies have compared taxa, experimental treatments, or life history stages based on instantaneous gas exchange measurements (Mazer et al., 2010; Franks, 2011; Dudley et al., 2012, 2015; Gorai et al., 2015; Cheng et al., 2017; Gulias et al., 2018; Velikova et al., 2018; Wang et al., 2019; Scalon et al., 2021).

In late June to mid-July 2010, all surviving plants were surveyed and, when flower production had ceased, the total number of full-sized or mature fruits borne by each plant was counted to provide an estimate of lifetime reproductive fitness. Many plants, however, suffered from frugivory by the larvae of Hyles lineata (the white-lined sphinx moth: Sphingidae) and, at some locations, by vertebrates, so total fruit production could not be accurately measured.  Consequently, as plants fully senesced and dried, we collected all above-ground parts (including fruits) from each plant, air-dried each skeleton in a paper bag, and returned them to the lab for weighing. All fruits retained by each plant were removed and counted, and the above-ground stem biomass (in mg, without any leaves, flowers, or ovaries) of each plant was recorded. To assess whether above-ground stem biomass is a reasonable proxy for lifetime reproductive success in these Clarkia taxa, we measured the relationship between above-ground stem biomass and total fruit production among individuals that were sampled during flowering and that suffered no frugivory. In all taxa, the correlation coefficient between these log-transformed variables was > 0.60 (C. unguiculata, r = 0.61, n = 151, p < 0.0001; C. exilis, r = 0. 90, n = 60, p < 0.0001; C. xantiana ssp. xantiana, r = 0.76, n = 173, p < 0. 0001; C. xantiana ssp. parviflora, r = 0.84, n= 81, p < 0.0001). In the analyses of phenotypic selection described below, we used above-ground stem biomass as the response variable (i.e., the estimate of individual fitness).

Climate data — To assess the influence of recent climatic conditions on the physiological rates of populations and taxa, monthly climatic data (cumulative precipitation; mean minimum, maximum and mean temperature; and vapor pressure deficit [VPD]) was extracted from the PRISM climate database (PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu) for November 2007-June 2008 (corresponding to the growing season surveyed by Mazer et al., 2010) and for November 2009-June 2010 (the growing season of the current study). These data were used to determine whether differences between years in the magnitude or direction of physiological divergence between sister taxa was associated with year-specific climate.  In addition, these climate data were used to examine whether physiological differences between sister taxa observed during flowering in the current study were associated with the mean climatic conditions to which their sampled populations were exposed in the month when they were surveyed (Tables S3, S4). When comparing C. exilis and C. unguiculata, climatic conditions for the months of April and May 2010, respectively, were used; these corresponded most closely to the months when each species were surveyed. When comparing the subspecies of C. xantiana, climatic conditions for May 2010 were used for ssp. xantiana, except for the Democrat Springs and Saw Mill Road sites, for which conditions in June 2010 were used due to their mid-June survey dates (Appendix S6).  For C. xantiana ssp. parviflora, climatic conditions for May 2010 were used, except for the Long Valley population, for which June 2010 conditions were used.

This README.txt file was generated on 2022-02-01 by Susan Mazer and David Hunter

GENERAL INFORMATION

1. Title of Dataset: 

2. Author Information

    A. Principal Investigator Contact Information

        Name: Susan J. Mazer

        Institution: University of California, Santa Barbara

        Address: Department of Ecology, Evolution & Marine Biology, UCSB, Santa Barbara, CA, 93106

        Email: sjmazer@ucsb.edu

    B. Associate or Co-investigator Contact Information

        Name: David J. Hunter

        Institution: Westmont College, Santa Barbara, California, 93108

        Address: Department of Mathematics and Computer Science

        Email: dhunter@westmont.edu

    C. Alternate Contact Information

        Name: Alisa A. Hove

        Institution: Warren Wilson College

        Address: Biology Department, P.O. Box 9000, Asheville, North Carolina, 28815

        Email: ahove@warren-wilson.edu

    C. Alternate Contact Information

        Name: Leah Dudley

        Institution: East Central University

        Address: Department of Biological and Environmental Sciences, East Central University, Ada, Oklahoma, 74820

        Email: ldudley@ecok.edu

3. Date of data collection (single date, range, approximate date) <suggested format YYYY-MM-DD>: 

Field data were collected from 2010-04-01 to 2010-06-15, with subsequent specimen processing (to measure above-ground stem biomass and to estimate fruit production) occurring approximately from 2010-07-01 to 2010-12-01

4. Geographic location of data collection <latitude, longitude, or city/region, State, Country, as appropriate>: 

Names and GPS coordinates of populations surveyed for physiological traits and total above-ground stem biomass in 2010.

Locality Name                                   Taxon               Elevation (m)    Latitude (N)    Longitude (W)

Willow Springs                                  exilis                365                        35.670        -118.902

Stark Creek                                        exilis                457                        35.474        -118.726

Stark Creek                                        unguiculata      457                        35.474        -118.726

Live Oak                                            unguiculata      430                        35.479        -118.749

Democrat Springs Campground        unguiculata      590                        35.531        -118.663

Democrat Springs Campground        xantiana           590                        35.531        -118.663

China Gardens                                   unguiculata      641                        35.533        -118.646

Borel Road                                        xantiana           707                         35.584        -118.522

Camp 3                                             xantiana            896                        35.812         -118.453

Fay Ranch Road                               parviflora          927                        35.700         -118.304

Sawmill Road                                  parviflora         1003                        35.678       -118.481

Sawmill Road                                  xantiana           1217                        35.675       -118.510

Long Valley                                     parviflora         1628                        35.815        -118.091

5. Information about funding sources that supported the collection of the data: 

Funding for this research was awarded by the National Science Foundation (award IOS-0718227) to L.S.D. and S.J.M.  A.A.H. received funding from the Luce Foundation, the Ralph M. Parsons Foundation, and the California Native Plant Society. 

SHARING/ACCESS INFORMATION

1. Licenses/restrictions placed on the data: None

2. Links to publications that cite or use the data:  Manuscript under review in the American Journal of Botany

3. Links to other publicly accessible locations of the data: N/A

4. Links/relationships to ancillary data sets: N/A

5. Was data derived from another source? No

    A. If yes, list source(s): 

6. Recommended citation for this dataset: 

Mazer, S.J., D. J. Hunter, A. A. Hove, and L. S. Dudley. 2022. Data used in manuscript, "Context-dependent concordance between physiological divergence and phenotypic selection in sister taxa with contrasting phenology and mating systems" (in review in the American Journal of Botany"

DATA & FILE OVERVIEW

1. File List: 

Three data sets are available. 
 

1) CUCE_herbarium_Normals_1951_1980_na.rm.csv:   These are the occurrence records and climatic parameters associated with the herbarium specimens of Clarkia unguiculata and Clarkia exilis analyzed in the current manuscript.
 

2) CXXCXP_herbarium_Normals_1951_1980_na.rm.csv:   These are the occurrence records and climatic parameters associated with the herbarium specimens of Clarkia xantiana ssp. xantiana and Clarkia xantiana ssp. parviflora analyzed in the current manuscript.
 

The R code used to analyze these data sets and to create the figures included in the manuscript is presented, along with the output, in the following html file (also uploaded into Dryad):

2022-01-17-Analyses_of_Herbarium_Records-Dryad

3) ClarkiaFieldData2010.csv: This data set contains all of the information for the field-collected data recorded in the wild populations of all four taxa (Clarkia unguiculata, C. exilis, C. xantiana ssp. xantiana and C. xantiana ssp. parviflora).  These data were used in the analyses of total and direct phenotypic selection on photosynthetic rate, transpiration rate, and instantaneous water use efficiency presented in the manuscript.  In addition, these data were used to detect physiological divergence between sister taxa.  The R code used to analyze this data set and to create the figures included in the manuscript is presented, along with the output, in the following html files (also uploaded into Dryad):

Gas_Exchange_Least_Squares_Means_Table.html

Gas_Exchange_Confidence_Intervals.html

Gas_Exchange_Summary.html

Gas_Exchange_Means_Figure.html

Total_Selection_Gradient_Models.html

Total_Selection_Figure.html

Direct_Selection_Gradient_Models.html

Direct_Selection_Figure.html

loadClarkiaData.R

2. Relationship between files, if important:

The file loadClarkiaData.R is a helper script used by all of the other scripts.

3. Additional related data collected that was not included in the current data package: N/A

4. Are there multiple versions of the dataset? No

    A. If yes, name of file(s) that was updated: 

        i. Why was the file updated? 

        ii. When was the file updated? 

METHODOLOGICAL INFORMATION

1. Description of methods used for collection/generation of data: 
Study species—Clarkia unguiculata Lindley and C. xantiana ssp. xantiana A.Gray are self-compatible winter annuals that flower in late spring, are endemic to California, and occupy oak and pine woodlands, grasslands, and roadside habitats, California. While C. xantiana ssp. xantiana is restricted to the southern Sierra Nevada and Transverse ranges to the south, the geographic range of C. unguiculata is much more widespread, occupying both the Sierra Nevada and the Coastal and Transverse Ranges (Lewis and Lewis, 1955). In the Lake Isabella region of the southern Sierra Nevada (Kern and Tulare counties), both taxa may occur alone or in sympatry with their predominantly selfing sister taxa (C. exilis and C. xantiana ssp. parviflora, respectively). Where sister taxa co-occur, reproductive isolation is reinforced by differences between sister taxa in flowering phenology and floral morphology, with selfers flowering earlier and showing greater levels of herkogamy and dichogamy than their bee-pollinated outcrossing counterparts (Eckhart and Geber, 1999; Dudley et al., 2007).

 In the southern Sierra Nevada, the sister taxa studied here differ with respect to habitat. Although C. exilis is sometimes encountered at the same sites as C. unguiculata, when these species are in sympatry, C. unguiculata occupies a broader range of habitats, growing in woodlands and encroaching into grasslands. By contrast, C. exilis has a more patchy distribution; it is generally restricted to relatively cool canyons, stream beds, and ravines, or is found near boulders or rocky outcrops that provide some shade and/or soil moisture (Vasek, 1958). The join distributions of C. xantiana ssp. xantiana and ssp. parviflora follow a west to east gradient, with xantiana occurring in populations to the west and parviflora occurring towards the east, nearly reaching the Mojave desert. Although there is a narrow central zone of contact between these subspecies (Eckhart et al., 2004; Geber and Eckhart, 2005), sites occupied solely by ssp. parviflora receive less precipitation than those occupied solely by ssp. xantiana.

Historical climatic and phenological differences between sister taxa: herbarium-based study—Occurrence data and the GPS coordinates of all specimen records available for each of the focal taxa were downloaded from the Consortium of California Herbarium portal (https://cch2.org) on July 31, 2021 (Fig.1). The date of specimen collection was converted to day-of-year (DOY: 1-365; 366 for leap years). The GPS coordinates of each collection site were then used to extract from the climate database, ClimateNA v.7.01 (Wang et al., 2016) the mean values of 23 annual climatic variables estimated from 1951-1980, the 30-year period preceding the point at which climate warming is considered to have begun to accelerate (IPCC, 2013). The climate variables estimated are defined in Appendix S1 (see Supplemental Data with this article).  To estimate climatic parameters, ClimateNA uses interpolated values from the PRISM climate database (PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu) to generate scale-free estimates of local climatic conditions at a resolution of 4km x 4km.

The DOY of specimen collection does not provide a precise record of the flowering date of a specimen because an individual plant collected in flower may have been collected at any time after flowering has begun, and the duration of an individual’s flowering period may be several weeks or longer. Nevertheless, the date of specimen collection has been found to be a reliable proxy for flowering time (Davis et al., 2015), and has been used in many studies to detect the factors influencing flowering phenology (Davis et al., 2015; Matthews and Mazer, 2016; Willis et al., 2017; Ferris and Willis, 2018; Park and Mazer, 2018, 2019; Ellwood et al., 2019; Love et al., 2019; Park et al., 2019; Mazer et al., 2021; Park, Breckheimer, et al., 2021; Park, Ramirez‐Parada, et al., 2021; Pearson et al., 2021). Among the focal taxa of this study, images were available from CCH2 on 25 December 2021 for 26, 679, 10, and 33 specimens of C. exilis, C. unguiculata, C. xantiana ssp. parviflora, and C. xantiana ssp. xantiana, respectively. Each image was visually inspected to determine whether the individual plant(s) on each sheet bore one or more open flowers. Of these, 24 (92%), 659 (97%), 8 (80%), and 30 (91%) specimens bore one or more open flowers, indicating that the DOY of specimen collection for these taxa generally reflects the specimen’s flowering time at a given site.

Statistical analysis of herbarium-derived data and climate—To assess whether sister taxa differ with respect to mean flowering time independent of the effects of local climatic conditions on flowering phenology due to evolutionary and/or short-term plastic responses to climate, we controlled statistically for climatic variation among collection sites when testing for differences between taxa in mean flowering time. To characterize each collection site using a multivariate index of its climate, principal components (PCs) for the 24 annual climatic variables were estimated using the pooled data for each pair of sister taxa. In both pairs of taxa, the first two principal components (PC1 and PC2) for the 24 climatic parameters together account for >75% of the variance in climate (Appendix S2). In the pooled data for Clarkia unguiculata and C. exilis, PC1 and PC2 account for 46% and 31% of the variance, respectively; in Clarkia xantiana ssp. xantiana and ssp. parviflora, PC1 and PC2 explain 62% and 21% of the variance, respectively (Appendix S2). In both taxon pairs, across their geographic distribution, the loadings of PC1 indicate that it represents an axis dominated by temperature, while PC2 represents an axis dominated by humidity (Appendix S3).

For each data set, the first two PCs were used to determine whether sister taxa differ with respect to their flowering phenology (controlling for site-specific climatic conditions) and in the chronic climatic conditions to which they are exposed. The function prcomp() was used to estimate PCs, following centering and scaling of the data to unit variance; the function fviz_pca_var() in the package factoextra (v.1.0.7) was used to visualize the contributions of each variable in each PC. These and all subsequent analyses were conducted in RStudio (v. 1.4.1717) with R v. 4.1.1.

Phenological differences between sister taxa—To determine whether sister taxa differ in their mean flowering date while controlling for the effects of geographic variation in local chronic climatic conditions (which influence DOY; see Results), we conducted four linear regressions on each sister pair. In all of these models, the day of year of specimen collection (DOY) was the response variable (Table 1).  In neither taxon pair were either of the two-way interactions (Taxon x PC1 or Taxon x PC2) statistically significant. Below, we present the results for the model with the lowest AIC value. Type III sums of squares were used to detect a significant difference between taxa in their mean date of collection, controlling for variation among collection sites with respect to the covariates. The lm function with the default setting for contrasts and the lme4 package (v. 1.1-27.1) were used.

Climatic differences between sister taxa —To determine whether sister taxa differ with respect to the chronic climatic conditions at their respective herbarium specimen collection sites, we used linear models to detect statistically significant differences between taxa with respect to the mean values of the first and the second PCs.  In these analyses, either PC1 or PC2 was the response variable, and taxon was the predictor variable. The lm function (using the default setting for contrasts: c[contr.treatment, contr.poly]) in the lme4 package (v. 1.1-27.1) was used.

Contemporary differences between sister taxa: phenotypic divergence in physiological traits—Sampling of extant populations: sample sizes and life stages—Naturally occurring field populations were surveyed in the region of Lake Isabella, California in the southern Sierra Nevada, where we sampled populations at elevations ranging from 457 to 1628 m. In spring 2010, 13 sites were identified that, at the onset of spring, contained a population of one or more of our focal taxa with a sufficient number of healthy juvenile plants that we judged we would be able to survey at least 90 individuals for their physiological performance during both vegetative and flowering stages (Fig.2; see Appendices S4 and S5 for a summary of each population’s GPS coordinates, elevation, sampling dates, and sample sizes). Each population was surveyed once during vegetative growth and once during flowering in spring 2010; a different group of individuals was sampled at each life stage. For each survey, 90 individuals were selected at random positions along one or more transects traversing the population, with the restriction that a surveyed plant had to bear at least one leaf long enough to sample with a LiCoR-6400 gas exchange analyzer (leaves that were too short could not cross the leaf chamber when it was closed). Sampled plants were typically at least 1m apart from one another. In total, we measured the physiological performance of  >1800 individuals. We analyzed and compared these traits while controlling statistically for leaf position on the primary stem and for leaf temperature at the time of measurement, both of which can influence measures of physiological performance (Mazer et al., 2010). 

Physiological survey—Gas exchange measurements were recorded on leaves produced by the primary branch on each plant. Measured leaves remained attached to the stem while using a portable infrared gas exchange analyzer (IRGA; LI-COR 6400, Lincoln, Nebraska, USA). We recorded photosynthetic rate (A; μmol CO2 m-2 s-1), transpiration rate (E; mol H2O m-2 s-1), and the derived parameter, A/E, instantaneous water use efficiency (WUEi;  μmolCO2 molH2O-1 x 0.0001). In addition, the position of each measured leaf, recorded as the number of the node that bore it, was recorded; the node just above the cotyledons was referred to as node 1, with subsequently produced nodes counted in sequence. The temperature of the leaf (°C) at the time of measurement was also recorded. Node number and leaf temperature were included as covariates in our statistical models in order to control for their potential effects on physiological rates, as observed by Mazer et al. (2010). The following LI-COR settings were used: light source, 6400-40 fluorometer; stability, values of CO2 and H2O remained stable for at least 15 s with a change in slope of <1; stomatal ratio = 1; flow = 500 μmol s-1; PARi (PAR in the leaf chamber) = 1500 mol m-2 s-1. CO2 concentration in the IRGA reference chamber was 400 ppm.  

For leaves too narrow to fill the entire 2cm2 circular area exposed within the LI-COR 6400’s leaf chamber, the actual leaf area exposed was calculated and then the LI-COR 6400’s Simulator Software was used to adjust those physiological parameters that were influenced by leaf area. To measure the leaf area exposed within the chamber, we used an ink pad to apply black ink to the foam gasket of the chamber; when the chamber was then closed upon the leaf, the gasket stamped with ink the outer boundary of the leaf’s area that was exposed within the chamber. Following gas exchange measurements, each leaf that failed to fill the chamber’s circular opening was removed, taped to a piece of paper with transparent tape, and subsequently scanned using a digital scanner. We then used the image analysis program ImageJ (http://rsb.info.nih.gov/ij/index.html) to measure the area of each leaf that had been exposed within the LI-COR chamber during gas exchange measurement.     

Instantaneous gas exchange measurements and WUEi provide point estimates of the physiological performance of individual plants and are not equal to the same parameters measured over the entire lifespan of individuals (Zelitch, 1982; Serichol-Escobar et al., 2016).  In the current study, however, the large sample sizes obtained per taxon, the replication of each taxon within and among 2-4 populations, and the statistical control of variation in other plant attributes that are correlated with instantaneous gas exchange rates (e.g., leaf position on the stem, leaf temperature at the time of measurement, and the life cycle stage at which plants are measured) helped to account for the variance in physiological performance that can obscure differences between taxa. Numerous other recent studies have compared taxa, experimental treatments, or life history stages based on instantaneous gas exchange measurements (Mazer et al., 2010; Franks, 2011; Dudley et al., 2012, 2015; Gorai et al., 2015; Cheng et al., 2017; Gulias et al., 2018; Velikova et al., 2018; Wang et al., 2019; Scalon et al., 2021).

In late June to mid-July 2010, all surviving plants were surveyed and, when flower production had ceased, the total number of full-sized or mature fruits borne by each plant was counted to provide an estimate of lifetime reproductive fitness. Many plants, however, suffered from frugivory by the larvae of Hyles lineata (the white-lined sphinx moth: Sphingidae) and, at some locations, by vertebrates, so total fruit production could not be accurately measured.  Consequently, as plants fully senesced and dried, we collected all above-ground parts (including fruits) from each plant, air-dried each skeleton in a paper bag, and returned them to the lab for weighing. All fruits retained by each plant were removed and counted, and the above-ground stem biomass (in mg, without any leaves, flowers, or ovaries) of each plant was recorded. To assess whether above-ground stem biomass is a reasonable proxy for lifetime reproductive success in these Clarkia taxa, we measured the relationship between above-ground stem biomass and total fruit production among individuals that were sampled during flowering and that suffered no frugivory. In all taxa, the correlation coefficient between these log-transformed variables was > 0.60 (C. unguiculata, r = 0.61, n = 151, p < 0.0001; C. exilis, r = 0. 90, n = 60, p < 0.0001; C. xantiana ssp. xantiana, r = 0.76, n = 173, p < 0. 0001; C. xantiana ssp. parviflora, r = 0.84, n= 81, p < 0.0001). In the analyses of phenotypic selection described below, we used above-ground stem biomass as the response variable (i.e., the estimate of individual fitness).

Climate data — To assess the influence of recent climatic conditions on the physiological rates of populations and taxa, monthly climatic data (cumulative precipitation; mean minimum, maximum and mean temperature; and vapor pressure deficit [VPD]) was extracted from the PRISM climate database (PRISM Climate Group, Oregon State University, http://prism.oregonstate.edu) for November 2007-June 2008 (corresponding to the growing season surveyed by Mazer et al., 2010) and for November 2009-June 2010 (the growing season of the current study). These data were used to determine whether differences between years in the magnitude or direction of physiological divergence between sister taxa was associated with year-specific climate.  In addition, these climate data were used to examine whether physiological differences between sister taxa observed during flowering in the current study were associated with the mean climatic conditions to which their sampled populations were exposed in the month when they were surveyed (Tables S3, S4). When comparing C. exilis and C. unguiculata, climatic conditions for the months of April and May 2010, respectively, were used; these corresponded most closely to the months when each species were surveyed. When comparing the subspecies of C. xantiana, climatic conditions for May 2010 were used for ssp. xantiana, except for the Democrat Springs and Saw Mill Road sites, for which conditions in June 2010 were used due to their mid-June survey dates (Appendix S6).  For C. xantiana ssp. parviflora, climatic conditions for May 2010 were used, except for the Long Valley population, for which June 2010 conditions were used.

2. Methods for processing the data: 

The submitted data reflect the data recorded in the field.  

3. Instrument- or software-specific information needed to interpret the data: 

R Packages and libraries used to analyze the data are included in the description of the Methods provided above.

4. Standards and calibration information, if appropriate:  N/A

5. Environmental/experimental conditions: N/A

6. Describe any quality-assurance procedures performed on the data: 

See R code for cases in which data were filtered to exclude clear outliers.

7. People involved with sample collection, processing, analysis and/or submission: 

SJM and LSD were awarded funding from the National Science Foundation to conduct the work described here; SJM, LSD, and AAH designed and conducted the field work; all authors contributed to planning the statistical analyses; DJH and SJM conducted the analyses; all authors discussed and reached consensus on the interpretation of the results; SJM wrote the manuscript, with editorial contributions from all co-authors.

S. K. Emms participated in the sampling design and recording of field data.  We are grateful to H. Guo for field assistance and to UCSB undergraduates K. Law, M. Echt, and S. Ricceri for assistance with measuring lifetime fruit production.

DATA-SPECIFIC INFORMATION FOR: 

CUCE_herbarium_Normals_1951_1980_na.rm.csv   

1. Number of variables: 30

2. Number of cases/rows: 778

3. Variable List: Note that the climate variables in this data set refer to 30-year annual means from 1951-1980

Specimen ID: reference number of no particular use

Taxon = subspecific epithet: "unguiculata" refers to Clarkia unguiculata, and "exilis" refers to Clarkia exilis

DOY = Day of Year of specimen collection
Year = Year of specimen collection
Latitude, in decimal degrees
Longitude, in decimal degrees
Mean_Ann_Temp = MAT
Mean_Warmest_Month_Temp = MWMT, Mean daily temperature of the warmest month, in degrees Centigrade
Mean_Coldest_Month_Temp = MCMT, Mean daily temperature of the coldest month, in degrees Centigrade
MWMT-MCMT = the difference between the mean temperature of the warmest month and the mean temperature of the coldest month, in degrees Centigrade
Mean_Ann_PPT_mm = MAP, or Mean Annual Precipitation
Mean_Summer_PPT_mm = May to September precipitation, in millimeters
Ann_Heat_Moisture_Index = Annual Heat Moisture Index, or AHM
Summer_Heat_Moisture_Index = SHM
Deg_Days_Below_O = DD < 0 degrees Centigrade
Deg_Days_Above_5 = DD > 5 degrees Centigrade
Deg_Days_Below_18 = DD < 18 degrees Centigrade
Deg_Days_Above_18 = DD > 18 degrees Centigrade
NFFD = number of frost free days in a given year
bFFP = the day of the year (from 1-365) on which the frost-free period begins
eFFP = the day of the year (from 1-365) on which the frost-free period end
FFP = the length of the frost-free period (typically eFFP - bFFP)
PPT_as_Snow_mm = precipitation as snow (mm).  For individual years, it covers the period between August in the previous year and July in the current year.
EMT = extreme minimum temperature over 30 years (degrees Centigrade)
EXT = extreme maximum temperature over 30 years (degrees Centigrade)
Eref = Hargreaves reference evaporation (mm)
Hargreaves_CMD_mm =  Hargreaves climatic moisture deficit (mm)
Rel_Humidity = mean annual relative humidity (%) 
CMI = Hogg’s climate moisture index (mm)
DD1040 = Degree-days above 10°C and below 40°C

CXXCXP_herbarium_Normals_1951_1980_na.rm.csv  

1. Number of variables: 30

2. Number of cases/rows: 65

3. Variable List: the climate variables in this data set refer to 30-year annual means from 1951-1980

Specimen ID: reference number of no particular use

Taxon = subspecific epithet: "xantiana" refers to Clarkia xantiana ssp. xantiana, and "parviflora" refers to Clarkia xantiana ssp. parviflora

DOY = Day of Year of specimen collection
Year
Latitude, in decimal degrees
Longitude, in decimal degrees
Mean_Ann_Temp = MAT
Mean_Warmest_Month_Temp = MWMT, Mean daily temperature of the warmest month, in degrees Centigrade
Mean_Coldest_Month_Temp = MCMT, Mean daily temperature of the coldest month, in degrees Centigrade
MWMT-MCMT = the difference between the mean temperature of the warmest month and the mean temperature of the coldest month, in degrees Centigrade
Mean_Ann_PPT_mm = MAP, or Mean Annual Precipitation
Mean_Summer_PPT_mm = May to September precipitation, in millimeters
Ann_Heat_Moisture_Index = Annual Heat Moisture Index, or AHM
Summer_Heat_Moisture_Index = SHM
Deg_Days_Below_O = DD < 0 degrees Centigrade
Deg_Days_Above_5 = DD > 5 degrees Centigrade
Deg_Days_Below_18 = DD < 18 degrees Centigrade
Deg_Days_Above_18 = DD > 18 degrees Centigrade
NFFD = number of frost free days in a given year
bFFP = the day of the year (from 1-365) on which the frost-free period begins
eFFP = the day of the year (from 1-365) on which the frost-free period end
FFP = the length of the frost-free period (typically eFFP - bFFP)
PPT_as_Snow_mm = precipitation as snow (mm).  For individual years, it covers the period between August in the previous year and July in the current year.
EMT = extreme minimum temperature over 30 years (degrees Centigrade)
EXT = extreme maximum temperature over 30 years (degrees Centigrade)
Eref = Hargreaves reference evaporation (mm)
Hargreaves_CMD_mm =  Hargreaves climatic moisture deficit (mm)
Rel_Humidity = mean annual relative humidity (%) 
CMI = Hogg’s climate moisture index (mm)
DD1040 = Degree-days above 10°C and below 40°C

4. Missing data codes: N/A

5. Specialized formats or other abbreviations used: N/A

********************

ClarkiaFieldData2010.csv

1. Number of variables: 11

2. Number of cases/rows: 2350

3. Variable List:  

Taxon = Which of the four focal taxa a given record (row) represents 
Population = The name of the field population in which the focal individual was observed
Reproductive_status = The life history stage of the population on the date of observation: Vegetative vs. Reproductive (flowering stage)
Biomass = Above ground stem biomass (in grams), measured at senescence, excluding leaves and fruits
DayGE_Node = Node number of the leaf on which gas exchange traits were measured
Photo_DayGE = Photosynthetic rate, umol/m2/s   
Trans_DayGE = Transpiration rate, mol/m2/s
WUE_DayGE = Instantaneous water use efficiency, [(umol CO2/m2/sec)/(mol H2O/m2/sec)]*1E4              
Tleaf_DayGE = Temperature of the leaf on which gas exchange rates were measured, at the time of measurement        
Cond_DayGE = Stomatal conductance, mol/m2/s          
Shoot_height_cm = Height of the plant in centimeters, recorded in each population 0-2 days before gas exchange traits were measured, with one exception.  The Willow Springs population of Clarkia exilis was measured for gas exchange traits during flowering on April 26, 2010, but the height of these plants was measured on April 14.         

4. Missing data codes: 

NA

5. Specialized formats or other abbreviations used: 

N/A