Printer Friendly

Differing selection on plant physiological traits in response to environmental water availability: a test of adaptive hypotheses.

The evolutionary response to selection can be predicted by the phenotypic selection in one generation and the genetic variances and covariances (Lande and Arnold 1983). Phenotypic selection studies, which describe the relationship between traits and fitness, can be used not only to predict the evolution of the traits studied but also as an empirical test of adaptive hypotheses. To support the hypothesis that a trait is adaptive in a given environment, demonstrating that the trait is correlated with fitness in that environment is not sufficient. The correlation of trait and fitness must also be lessened or absent in an environment where the trait is not expected to be adaptive (Wade and Kalisz 1990). Many studies have demonstrated strong natural selection in the wild (Endler 1986), but few studies have attempted to test evolutionary predictions and adaptive hypotheses. Here I describe a test of adaptive hypotheses for plant physiological traits that affect plant carbon uptake and water loss. A large body of functional analyses and comparative studies provides hypotheses on how selection on these traits should depend on environmental water availability (i.e., Ehleringer 1975; Givnish 1986, Cowan 1986). In this paper, I describe a field study of selection on leaf size and water-use efficiency in wet and dry environments and compare the results with the predictions of adaptive hypotheses. In a companion paper (Dudley 1996), I measure the genetic differentiation in these traits between populations from wet and dry environments as well as the genetic covariance matrix and compare these results with the field studies of selection.

The adaptive hypotheses I examine in this paper are based on the physiology of photosynthesis. When plants open their stomates (pores in the leaf epidermis) to allow carbon dioxide to diffuse into the leaf for photosynthesis, water diffuses out rapidly. Transpirational water loss is potentially costly for plants. Transpiring away more water than the plant can acquire may cause drought stress, which can compromise the ability to grow and to acquire carbon, and in the extremity is fatal (Givnish 1986; Schulze 1982). Thus, although increased net carbon acquisition gives plants more resources to compete, grow, and reproduce in all environments (Ehleringer 1975), the evolution of traits that affect carbon acquisition may be constrained in drier habitats because carbon gain is accomplished at the cost of water loss. One way that plants adapt to dry environments is by reducing water loss (Givnish 1986). Water loss can be reduced by having smaller leaves to reduce the transpiring leaf surface (Givnish 1979) and in some cases to reduce leaf temperature and thus lower the water potential at the leaf surface (Nobel 1991). Another potential adaptation is to change the relative rates of gas exchange to maximize the carbon assimilation to water-loss ratio, defined as the water-use efficiency (Cohen 1970, Cowan 1986).

These adaptive hypotheses are supported by comparative studies showing higher water-use efficiencies in populations and species from drier environments (Gurevitch et al. 1986; Kalisz and Teeri 1986, Ehleringer and Cooper 1988). Studies of phenotypic selection in Xanthium strumarium grown in a cultivated field (Farris and Lechowicz 1990), Prunella vulgaris grown in the greenhouse in high and low light environment (Winn and Evans 1991), and Plantago lanceolata grown in an old field (Tonsor, manuscript) have found selection for higher water-use efficiency. The relationship between biomass and water-use efficiency (measured as carbon isotope ratio) has been explored in both well-watered and water-limited environments in a common garden study on the desert shrub Chrysothamnus nauseosus (Donovan and Ehleringer 1994). However, a further test of the hypothesis that higher water-use efficiency and smaller leaves are selected because they are adaptations to drought stress is to compare the strength of natural selection on water-use efficiency and leaf size in environments that differ only in water availability (Wade and Kalisz 1990). Measures of natural selection cannot prove a causal relationship between traits and fitness (Mitchell-Olds and Shaw 1987; Rausher 1992), but finding that the associations between traits and fitness change as predicted between environments would provide strong support for their hypothesized adaptive value (Wade and Kalisz 1990).

Here I present a study of natural selection for leaf and gas exchange traits in Cakile edentula var. lacustris (Brassicaceae), a succulent [C.sub.3] beach annual found on sandy beaches along the Great Lakes of the North America (Rodman 1974). This species is considered to be subject to edaphic drought because its beach sand substrate does not retain water, and Lake Michigan populations of C. edentula vat. lacustris may experience several days or weeks without rainfall. Populations of C. edentula vat. lacustris around Lake Michigan grow in sites that differ greatly in water availability because of rainfall, topography, seeps, and other sources of groundwater. To test the adaptive hypotheses suggested by the physiological literature, I measured phenotypic selection on instantaneous water-use efficiency and leaf size for plants of known ancestry placed into natural environments contrasting in water availability. These measurements allow me to ask the following questions: (1) Does selection favor greater water-use efficiency in the dry environment than in the wet environment? (2) Does selection favor smaller leaf size in the dry environment than in the wet environment?


Study Species and Sites

The study plant, C. edentula vat. lacustris, belongs to a genus native to the sandy beaches of the North Atlantic Ocean and adjoining bodies of water. All members of the genus are succulent herbaceous annuals. Leaves of Cakile species have been reported to range from 0.5-1.5 mm in thickness, have stomates on both leaf surfaces, and have undifferentiated parenchyma (Rodman 1974). Physiological measurements made on greenhouse-grown Cakile maritima from northern California indicate relatively high photosynthetic capacities with maximum photosynthetic rates of 37 [[micro]molar]-[m.sup.-2]-[s.sup.-1] and conductances of 12 mm-[s.sup.-1] (De Jong 1978). Cakile edentula var. lacustris is an indeterminately flowering annual with considerable variation in flowering schedule. Breeding systems of Cakile species range from predominantly self-pollinating to predominantly outcrossing. Flowers are born on elongating terminal racemes. Fruits consist of two segments, the upper being deciduous and the lower remaining attached to the raceme. Each segment contains usually one seed, with upper seeds being slightly larger than lower seeds. Cakile edentula vat. lacustris is native to the Great Lakes and is the only member of the genus found on freshwater beaches. It is regarded as a primarily autogamous taxon, with small flowers and high rates of spontaneous fruit set (Rodman 1974).

Seeds were collected from the Indiana Dunes National Lakeshore from 23 randomly chosen naturally pollinated plants, 12 from a dry site (West Beach), and 11 from a wet site (Mount Baldy) approximately a 26-km distance from one another. These two populations are further characterized in Dudley (1996). Both sites lay along the base of dunes parallel to the beach, facing northwest. Both sites had only sparsely distributed annual vegetation, and consequently plants received full sunlight. Environmental water availability was determined by observation. The wet site contained a permanent seep of water along the base of the dune. Water was observed trickling from the seep throughout the summer, and plants near the seep grew in a slurry of sand and water. The dry site lacked any water source other than the lake itself. Natural populations of plants also indicated differences in favorableness between sites, with very large plants found at the wet site and small plants at the dry site, suggesting that the plants do not reach the water table. Nutrient measurements indicated that levels were similar and very low for the two sites. The two populations from contrasting habitats were pooled into the experimental population used in this study to increase the expected phenotypic variation and ensure that the experimental population was not more adapted to one of the experimental environments.

Because the breeding system is believed to be predominantly self-pollinating, progeny for experiments were generated by allowing plants to self-fertilize. Maternal sibships from the field-collected plants were raised in the greenhouse and allowed to self pollinate. One plant from each maternal sibship was randomly selected to be the parent of the experimental generation. Approximately 10 offspring from each of the 23 original families were assigned to each of two treatments: wet and dry.

Field Environments

The field experiment took advantage of the natural patchiness in water availability found at one site. At the Mount Baldy site, a natural seep has created a narrow strip (from 1-5 m in width) parallel to the beach of extremely wet sand providing the wet environment. There was abundant available water during the entire season near the seep. One transect was laid out along the length of the seep for 50 m. The slope approximately 4 m above the seep provided the dry environment. The sand on the slope was well drained, and it received water only from rainfall. The wet and dry treatments both had soil that was 99.9 % coarse and fine sand, 0.1% silt and clay particles, and had similar, low soil nutrients. In the dry environment, two parallel transects (each of 25 m) were laid out, one approximately 2 m above the other running parallel to the beach and the other transect. In both environments, transects were placed in areas away from perennials and shrubs growing on the seep such that the experimental plants received full sunlight. Limiting the experiment to only two sites was necessary because of practical considerations but did result a lack of replication of the environmental water availability treatment. However, although the sites differed markedly in water availability, they were similar in nutrient levels and in light availability.

Seeds of the experimental population were induced to germinate by soaking them overnight and removing their seed coats. Seeds were planted May 7, 1990 into 50:50 promix Turface in plug trays and lightly covered with soil mix. The plug trays were kept on dry heat under natural lighting for 1 wk and then in the greenhouse for the 2-day week. All seeds germinated within 4-8 days. The seedlings were then transferred to a cold frame to acclimate. Later, they were labeled with bird-band tags and transplanted to the field on June 12, 1990. At that time, they had two to six true leaves. The seedlings were planted in randomized order in the transects every 20 cm. Rather than planting a straight line of plants, the seedlings were alternately offset either 5 cm above or 5 cm below the transect to present a slightly more natural appearance.


Leaf size and gas-exchange rates were measured in late summer on surviving plants, which were producing flowers and fruits. For each plant, one recently produced fully expanded leaf was sampled, and measures made of net photosynthetic rate (assimilation of C[O.sub.2] in [[micro]molar] C[O.sub.2]-[m.sup.-2][s.sup.-1]), the rate at which the leaf surface acquires carbon for photosynthesis; stomatal conductance (moles [H.sub.2]O-[m.sup.-2][s.sup.-1]), the leaf surface's influence on the rate of water loss; and the area and weight of the dried, pressed, leaf. Instantaneous water-use efficiency was then calculated as the ratio of photosynthetic rate to stomatal conductance. Such single-leaf measures of gas exchange and leaf size are heritable (Dudley 1996). Because gas-exchange rates were measured on a dry-leaf area basis which is extrapolated to give wet-leaf area measures (see below), these results are more accurate as comparative measures among plants rather than as estimates of field gas-exchange rates. Because few plants died after measurement, the data were used for analysis of fecundity selection.

Measurements of photosynthesis and stomatal conductance were done on August 23 and 24, 1990, using an Analytical Development Company LCA-2 portable photosynthesis system (Long and Hallgren 1985; Field et al. 1989). Measurements were made under ambient light; both days were clear with occasional clouds. The mean light level was 990 photons-[m.sup.-2]-[s.sup.-1]. Though light levels varied from 330-1280 [[micro]molar] photons-[m.sup.-2]-[s.sup.-1], 90% of measures were made at light levels above 810 [[micro]molar] photon-[m.sup.-2]-[s.sup.-1]. Cuvette temperatures varied from 25-33 [degrees] C. Plants were measured between 10:00 A.M. and 3:00 P.M. in a predetermined order that was random with respect to environment, position within environment, and the plant's parentage. Gas-exchange traits could not be measured on 50% of the surviving dry-site plants because they were moribund or defoliated. Wet-environment plants that were temporarily inaccessible because of sand movement (N = 38) were not measured. Measurements were made on all dry-environment plants that possessed leaves (N = 62) and on 102 wet-environment plants. For each plant, one recently expanded new leaf was detached and measured within 90 sec. Detached leaves of this species maintain constant gas-exchange ratios for more than 5 min (Dudley, pers. obs.). The sample leaves were subsequently pressed and dried to provide a measure of the size of a recent fully expanded leaf and leaf dry mass. Areas of dry leaves were estimated with a LI-COR leaf-area meter. Wet and dry leaf areas from leaves sampled from several greenhouse grown plants provided a conversion factor for calculating gas-exchange rates on a wet-leaf area basis (wet area = 1.90 dry area in chamber).

The LCA-2 was factory calibrated for the C[O.sub.2] and relative humidity spans. The relative humidity of 0 was verified using dried air and the PPM C[O.sub.2] was verified by measuring air passed through soda lime to remove C[O.sub.2]. Leaf temperatures in the chamber was estimated from the energy-balance equation (Parkinson 1984; Analytical Development Company 1992). Boundary-layer resistances for Cakile leaves in a Parkinson leaf chamber were estimated on filter-paper mimics (Parkinson 1985; Analytical Development Company 1992). Because boundary-layer resistance was correlated with the size of the "leaf," a regression equation estimating boundary-layer resistance as a function of leaf size was used in subsequent calculations.

Photosynthetic rates and stomatal conductances were calculated according to von Caemmerer and Farquhar (1981). Measures of gas exchange were covariate corrected for date and time (Winn and Evans 1991; Harris 1975; Farris and Lechowicz 1990). Light level and temperature were not found to explain any significant variance when date and time were included in the model. Plants that appeared to be wilted or to have extremely aberrant water-use efficiencies were omitted from subsequent analysis (N = 3).

A sensitivity analysis conducted on the data found that the conclusions about the relative differences among individuals and populations were robust to the potential error in calibration of the equipment or estimation of the boundary-layer conductance.

I harvested plants on September 15, 1990 to avoid loss of the study population to burial by sand, waves from large storms, or movement of beach sand offshore by changes in currents. At this time, most of the plants in the dry environment had ceased flowering, but the plants in the wet environment varied from mostly flowering to senescent. The measurements I obtained estimate plant fitnesses given an early end to the growing season. For an indeterminately flowering annual, the ranking of fitnesses in a population may depend on season length because high early reproduction may come at the cost of lowered vegetative growth reducing resources for later reproduction (Geber 1990).

The collected plants were air dried at room temperature. Sand was scrubbed off the plants when necessary. The air-dried plants were partitioned into vegetative biomass (stems and large roots) and reproductive biomass (fruits) and then weighed.

I used the reproductive biomass allocated to fruits as a fitness measure because it is the hypothesized causal link between carbon acquisition traits and fitness [ILLUSTRATION FOR FIGURE 1 OMITTED]. Most fruits contain two seeds (Rodman 1974); and because Cakile flowers indeterminately, higher reproductive biomass is largely the result of more fruits rather than larger fruits. Using fruit biomass for fitness measures maternal investment in reproduction, but ignores the potential for variance in biomass allocated per fruit and per seed.

Data Analysis

Means and standard deviations were calculated for leaf size, photosynthetic rate, stomatal conductance, water-use efficiency, vegetative biomass, and fruit biomass in the wet and the dry environments for the subset of plants on which I measured gas exchange (PROC MEANS, SAS Institute 1989). The difference in means between the study populations grown in the two environments is presented analogously to the genetic differences between populations (Lande and Arnold 1983) as the difference in means for a trait between the two environments, divided by the pooled phenotypic standard deviation for each environment.

Standardized selection differentials, which include both direct and indirect selection on a trait, were calculated for each environment as the covariance between the standardized trait and fitness (PROC CORR) (Lande and Arnold 1983). Selection differentials were considered significantly different from 0 if the Pearson correlation coefficient between the trait and relative fitness differed significantly from 0 (Lande and Arnold 1983). Selection differentials for the same trait in the two environments were considered significantly different from each other if the z-transformed Pearson correlation coefficients between the trait and fitness differed significantly from each other (Sokal and Rohlf 1981). Phenotypic correlations among traits were obtained from the Pearson correlation matrix of traits (PROC CORR).

To test the adaptive hypotheses about direct selection on leaf size and water-use efficiency, I calculated linear and nonlinear (stabilizing and correlational) selection gradients following Lande and Arnold (1983). Directional selection gradients were obtained from a linear regression of relative fitness on the traits, and quadratic selection gradients were obtained from then regressing relative fitness on linear and quadratic terms (PROC GLM in SAS). I estimated selection gradients only for water-use efficiency and leaf size to avoid collinearity from "spurious correlations" (Sokal and Rohlf 1981) caused by including traits that are mathematical functions of each other. Thus, photosynthetic rate and stomatal conductance could not be included in the same regression as water-use efficiency, and because leaf sizes were smaller than the cuvette sampling area, the calculations of photosynthetic rate and stomatal conductance included leaf size as a divisor. However, water-use efficiency is estimated independently of leaf size. I tested for differences between environments in standardized selection gradients through a test of heterogeneity of slopes in an analysis of covariance (ANCOVA) (PROC GLM). Because the residuals from the regression analysis were not normally distributed, I also calculated the significance levels of the standardized regression coefficients from jackknife estimates of the variances (FreeStat, Mitchell-Olds 1989). I found that significance changed slightly in both directions but the overall trend was similar, suggesting that the parametric analysis was robust to the lack of normality of the residuals in this data set. I therefore present the significance levels from the parametric analysis because of the ease in testing for differences in regression coefficients between environments.

One data point in the wet environment with extremely high fitness was found to have a disproportionately large effect on the regression analysis and was omitted as an outlier following the criterion of Sokal and Rohlf (1981). In the dry environment, the nonlinear selection analysis was run with and without one outlying data point with very large leaf size, and the stabilizing and correlational selection gradients remained significant.

I performed a path analysis to examine the linear components of the functional dependency between the water-use efficiency and leaf size and fitness (Arnold 1983). Path diagrams can be used to factor linear selection gradients on phenotypic traits into the effect of a trait on some aspect of performance (e.g., leaf traits on vegetative biomass), and the effect of performance traits on fitness (e.g., vegetative biomass on fruit biomass) (Arnold 1983). I used a path model that allowed for leaf size and water-use efficiency to affect fitness through vegetative biomass and that allowed for leaf size and water-use efficiency to affect fitness via an alternate path ([ILLUSTRATION FOR FIGURE 1 OMITTED]; Ehleringer and Clark 1988). Path coefficients were estimated using PROC GLM to obtain the appropriate regression coefficients on the standardized variables and relative fitness. The sum of all paths leading from the leaf traits to relative fitness equals the directional selection gradients obtained above (Li 1975; Arnold 1983).


The plants grown in the wet environment had greater fruit biomass than did plants in the dry environment. The plants grew much bigger in the wet environment and had very high fruit production (Table 1). In the wet site, 21% of the original plants were washed away because of shifts in sand location and storms. Of the remaining wet-site plants (N = 183), an additional 23% died, primarily from human trampling. In the dry site, 50% of the plants died, either from apparent drought stress, insect damage, or from human trampling. Although the sites were near each other, the presence of insects appeared to vary between the sites. More herbivores were observed in the dry site, and possible pollinators were more prevalent in the wet site. Such differences may be the result of differences in the plants rather than the sites themselves. Because mortality did not differ significantly among families or between populations either in the dry environment (P = 0.22) or in the wet environment (P = 0.58), differences between environments in traits indicate the average phenotypic plasticity for the population.


In these results, extrapolated from dry leaf areas (Table 1), and in greenhouse measures with fresh leaf areas (Dudley, pers. obs.), C. edentula var. lacustris exhibited moderately high photosynthetic rates and very high stomatal conductances (Korner et al. 1979). High gas-exchange rates are thought to be typical for amphistomatous thick leaves found in high light environments (Mott et al. 1982). As predicted by functional hypotheses for adaptive plasticity, plants grown in the dry environment had smaller leaves than in the wet environment (Table 1). Contrary to adaptive-plasticity predictions, stomatal conductances were much higher in the dry environment. Though photosynthetic rates were also somewhat higher in the dry environment, the water-use efficiency was lower in the dry environment. Specific leaf weight (ratio of leaf weight to leaf area) was higher in the dry site (dry = 0.0123g/[cm.sup.2], wet = 0.0074 g/[cm.sup.2], P [less than] 0.0001), and leaves were noticeably thicker in the dry environment.

The fecundity selection differentials, which include direct and indirect selection, differed significantly between environments for all traits but photosynthetic rate (Table 2). In both environments, vegetative biomass, a long-term measure of carbon uptake, and photosynthetic rate, a short-term measure of carbon uptake, were positively associated with fitness, a result predicted by the functional hypotheses. The associations between stomatal conductance and fitness differed between environments as predicted by functional hypotheses: plants with greater stomatal conductance and thus higher rates of water loss per unit leaf area were less fit in the dry environment, and there was no association of stomatal conductance and fitness in wet environment (Table 2). The associations between water-use efficiency and fitness differed between environments as predicted by functional hypotheses: plants with higher water efficiency, which gained more carbon for the same water loss, were more fit in the dry environment, but no association of water-use efficiency and fitness existed in the wet environment. Unexpectedly, larger leaf sizes were not associated with higher fitness in the wet environment but were associated with higher fitness in the dry environment (Table 2). This result was not predicted by the hypothesis that reduction of water loss through having smaller leaves is adaptive in drier environments.
TABLE 2. Standardized directional selection differentials in each
environment. Significance differences from 0 in each environment
and the t-test of differences between selection differentials on
the same trait in the different environments are taken from
significance levels for the corresponding correlation coefficient
between the trait and relative fitness. Dry environment N = 57,
wet environment N = 99.

                                     Standardized selection

                              Dry          Wet
Trait                     environment   environment    t-value

Photosynthetic rate       0.41 (*)        0.18(*)      0.666
Stomatal conductance     -0.33(*)         0.10         2.24(*)
Water-use efficiency      0.66(***)       0.01         3.23(**)
Leaf size                 0.54(***)       0.00         2.56(*)
Vegetative biomass        0.89(***)       0.39(***)    2.06(*)

* P [less than] 0.05; ** P [less than] 0.01; *** P [less than]

The phenotypic correlations among traits were strong and depended on the environment (Table 3, [ILLUSTRATION FOR FIGURE 2 OMITTED]). Photosynthetic rate and stomatal conductance showed a strong positive correlation in the wet environment, but their relationship was weaker in the dry environment. Stomatal conductance was negatively correlated with leaf size in the dry environment. Photosynthetic rates were not correlated with leaf size. Leaf size was positively correlated with water-use efficiency in both environments (Table 3).

I performed a multivariate fecundity selection analysis on water-use efficiency and leaf size in each environment to estimate the direct selection on each trait (Table 4, [ILLUSTRATION FOR FIGURE 3 OMITTED]). Because of the high mortality before traits and fitness were measured, the power to test for the significance of selection gradients was low. In the wet environment, the only significant selection gradient was weak stabilizing selection on water-use efficiency. In the dry environment, there was strong directional selection for increased water-use efficiency, stabilizing selection for an intermediate leaf size, and positive [TABULAR DATA FOR TABLE 3 OMITTED] correlational selection between water-use efficiency and leaf size (Table 4, [ILLUSTRATION FOR FIGURE 3 OMITTED]). As the fitness surface [ILLUSTRATION FOR FIGURE 3 OMITTED] shows, the correlational selection caused the optimum leaf size to increase with increased water-use efficiency. There was no significant directional selection gradient in the dry environment for leaf size. The significant directional selection differential for leaf size in the dry site (Table 2) appears to be an indirect response to selection on water-use efficiency and to the correlational selection on leaf size and water-use efficiency.

Because of the small sample size, the power to test for differences between environments in selection on water-use efficiency and leaf size was low. Nonetheless, the differences in directional selection on water-use efficiency, stabilizing selection for an intermediate leaf size, and positive correlational selection between water-use efficiency and leaf size between the wet and dry environments were all highly significant (Table 4, [ILLUSTRATION FOR FIGURE 4 OMITTED]). These differences are consistent with functional hypotheses that suggest that environmental water availability causes the selection on leaf size and water-use efficiency.

Path-analysis models using leaf size and water-use efficiency as carbon uptake traits were used to analyze the directional selection gradients in the two environments [ILLUSTRATION FOR FIGURE 5 OMITTED]. The results agreed with the expectation that leaf traits affect fitness through their effect on vegetative biomass. Greater leaf size increased vegetative biomass in the wet environment. In the dry environment, higher water-use efficiency led to greater vegetative biomass. The analyses did show a marginally significant direct effect of water-use efficiency on fruit biomass in the dry site. In both environments, higher vegetative biomass caused increased fitness, but the increase in fruit biomass with vegetative biomass was significantly greater (P [less than] 0.05, analysis of covariance) in the dry environment than in the wet environment [ILLUSTRATION FOR FIGURE 5 OMITTED].


This study examined the differences in fecundity selection on leaf traits between wet and dry environments. Significant differences in selection between environments were found despite the relatively small sample sizes. The results, as predicted by physiology of carbon acquisition and water loss, demonstrate the importance of net carbon uptake for plant fecundity and show that fecundity selection on traits that affect water loss differed as predicted between wet and dry environments. As predicted, selection for greater water-use efficiency was found in the dry environment compared with the wet environment. Selection favored an intermediate leaf size in the dry environment, whereas larger leaves were associated with greater vegetative biomass in the wet environment. This study compared selection in two strongly contrasting environments for one species but found results that strongly support adaptive hypotheses. Further studies in other species, replicated across contrasting environments, are necessary to see if these results can be generalized. In interpreting these results, I consider three questions: the support of the results for the hypothesized causal relationships between traits and fitness; the implications for the evolution of the traits in the different environments; and the description of the adaptive value of the traits in each environment.

Causal Interpretation of Selection Analyses

The difference in selection observed between wet and dry environments provides direct evidence for the selective impact of water availability on carbon acquisition traits (Wade and Kalisz 1990). The path analysis of the selection gradients provides a test of the hypotheses about why water-use efficiency and leaf size are selected (Arnold 1983). It supported the functional hypothesis that water-use efficiency and leaf size affect fitness because they affect net carbon acquisition. The stronger selection on water-use efficiency in the dry [TABULAR DATA FOR TABLE 4 OMITTED] environment was caused by greater effects of water-use efficiency on vegetative biomass, a greater effect of vegetative biomass on fitness, and a marginally significant direct effect of water-use efficiency on fitness compared with the wet environment. The direct effect may be explained by plants with higher water-use efficiency rates being more successful in provisioning fruits. In a common garden study on the desert shrub Chrysothamnus nauseosus, Donovan and Ehleringer (1994) found, contrary to the results in this study, that the phenotypic correlation between water-use efficiency (measured as carbon isotope ratio) and vegetative biomass was positive in well-watered treatments and 0 in water-limited treatments, though the genetic correlation in each treatment was positive. They suggest that unmeasured, correlated traits such as carbon allocation may explain their results.

The path analysis also provides a way of evaluating potential selection in longer seasons. Selection on a trait that affects vegetative biomass depends on the strength of selection on vegetative biomass, as the results here demonstrate. For one indeterminately flowering annual, Polygonum arunatum, early reproductive success was found to be largely determined by timing of allocation to reproduction, but later reproductive success was strongly associated with vegetative growth (Geber 1990). If Cakile follows a similar pattern, then in years with a long growing season, selection for larger leaf size would be predicted in the wet environment, but selection on water-use efficiency would still be expected to be greater in the dry environment than the wet environment.

Predicting Evolution

In this experiment, siblings from 23 families from two populations were planted in both environments. Though the experimental plants suffered high mortality in each environment, mortality did not differ between source populations or among families. Therefore, this study provides standardized fecundity selection gradients on the same population in different environments and can be used to predict how that population would evolve in response to selection. One caveat in predicting the response to selection is that high mortality occurred before the physiological traits were measured. The mortality, which may have been an episode of survivorship selection, was greatest in the dry environment. As a consequence, the statistical power for testing the hypotheses was low. The extreme contrast in water availability between the environments did increase the ability to find differences in selection gradients. However, this study did not have the power to identify smaller selection gradients and differences in selection between environments, which can have important evolutionary implications.

The observed selection gradients indicate that populations in each environment will be under different selection regimes. The significant difference in directional selection on water-use efficiency suggests that for the same population water-use efficiency will be selected to increase in the dry site and to remain near the mean in the wet site. The selection gradient analysis shows that leaf size was selected to be at an intermediate optimum in the dry site. The lack of significant directional selection on leaf size in the dry environment implies that the optimum leaf size in the dry environment does not differ significantly from the mean leaf size in the dry environment. In contrast, though there was no significant selection on leaf size in the wet environment, the path analysis suggests that because increased leaf size was associated with increased vegetative biomass, larger leaf size may be selected in some years. These results suggest that for the same population leaf size will be selected to remain near the mean in the dry site and potentially to increase in the wet site.

The intermediate optimum for leaf size in the dry environment depended on water-use efficiency. Plants with lower water-use efficiency were more fit if they had smaller leaves, and plants with higher water-use efficiency were more fit if they had larger leaves. Such correlational selection may act as a force of selection on the genetic variances and covariances that constrain evolution (Phillips and Arnold 1989). These results suggest that selection was acting at the level of the leaf on the balance between leaf carbon uptake and water loss. The correlational selection may reflect the dependence of whole-plant carbon acquisition on multiple interdependent leaf traits, which has been long recognized by physiologists (e.g., Ehleringer and Clark 1988). Thus, this result further supports the argument that studies of natural selection must be informed by underlying biology of the study traits.

The differences between selection gradients in the wet and dry environment were consistent with a genetic analysis of differentiation between the wet site and dry site populations. A companion study (Dudley 1996) found that the population from the dry site had higher water-use efficiency and smaller leaves compared with the population from the wet site. The genetic architecture of the traits suggests that differentiation between populations found in the differing environments is constrained by the lower genetic variation for water-use efficiency and the positive genetic correlation between water-use efficiency and leaf size (Dudley 1996).

Describing Adaptation

Measuring the adaptive value of traits requires comparing the fitness surfaces for the unstandardized traits in each environment, rather than the standardized selection gradients. However, because plasticity between the two environments of all traits but photosynthesis was extreme, the measured selection gradients are for plants showing very different phenotypes for water-use efficiency and leaf size [ILLUSTRATION FOR FIGURE 2 OMITTED] and different correlations between the traits (Table 3). Therefore, the fitness surfaces cannot be statistically compared because they are valid over different ranges of the traits, but some inferences may be made.

The lack of significant directional selection on leaf size in the dry environment, coupled with stabilizing selection, implies that the optimum leaf size in the dry environment did not differ significantly from the mean leaf size in the dry environment. Leaves of the dry environment plants were significantly smaller than in the wet environment. Though leaf size did not affect fitness in the wet environment, larger leaves led to increased vegetative biomass. These results are consistent with the hypothesis that a smaller leaf size is adaptive in the dry site, and a larger leaf adaptive in the wet site, in agreement with functional predictions.

The finding that water-use efficiency was under strong positive directional selection in the dry site and weak stabilizing selection in the wet site agrees with functional hypotheses that high water-use efficiency is more critical when water is limiting. However, the difference in average water-use efficiency between environments suggests an alternate hypothesis, that the environments have a common selection curve for water-use efficiency, with the wet-site plants having the optimum values and the dry-site plants having values lower than the optimum. Therefore the results do not conclusively support the hypothesis that higher water-use efficiency is more adaptive in the dry site than in the wet site. A conclusive test would require experimental manipulation to extend the range of phenotypes in each environment.

The lower water-use efficiency exhibited by plants in the dry environment compared with the wet environment is very puzzling. Commonly plants show greater water-use efficiency in drier environments (e.g., Ehleringer and Cooper 1988; Donoran and Ehleringer 1994) though decreases in water-use efficiency with seasonal increases in drought stress have been observed in some species (DePuit and Caldwell 1975; Smedley et al. 1991). In this study, the functional arguments, the field-selection results, and the genetic differences between populations all suggest that Cakile exhibited a mal-adaptive plastic response to the environment. Such an apparently maladaptive plastic response suggests either that the evolution of adaptive plasticity in water-use efficiency may be constrained by genetic correlations or that the expression of adaptive trait combinations may be constrained by the phenotypic correlations among traits (Scheiner 1993).

One phenotypic correlation that may constrain the adaptive plasticity of physiological traits is that between water-use efficiency and leaf size. Water-use efficiency was positively associated with leaf size, because stomatal conductance was negatively associated with leaf size. As Figure 2 shows, the water-use efficiencies and leaf sizes of plants in both environments appear to follow a curvilinear relationship. Whether this correlation reflects a functional relationship is not known (Bhagsari and Brown 1986). It is plausible that such a correlation could be the result of the positive correlational selection for water-use efficiency and leaf size (Philips and Arnold 1989; Arnold 1992). Nonetheless, for a given leaf size, the dry-site plants had a lower water-use efficiency than the wet-site plants [ILLUSTRATION FOR FIGURE 2 OMITTED], suggesting that this correlation is not sufficient to explain the lower water-use efficiency in the dry-environment plants.

Another hypothesis for the greater stomatal conductance and higher photosynthetic rates seen in dry-site plants is suggested by the significantly higher specific leaf weight and the observation that leaves were thicker and more succulent in the plants grown in the dry site. Specific leaf weight has often been observed to be positively correlated with photosynthetic rate (Bhagsari and Brown 1986). A biophysical constraint on thicker leaves is that greater conductance is needed to permit C[O.sub.2] diffusion to the interior of thicker leaves (Mott et al. 1982). Greater succulence in the dry-site plants may permit long-term water storage and thus be an adaptation to drought, but this hypothesis requires further studies of the physiology and ecology of succulent leaves in [C.sub.3] species.

A weakness of selection analyses is that they are based upon correlations among traits (Mitchell-Olds and Shaw 1987; Wade and Kalisz 1990). Such correlations may result from a causal relationship between the traits and fitness, but they may also result from environmentally induced covariance between the traits and fitness or from indirect selection on unmeasured, correlated phenotypic traits (Rausher 1992). These possibilities cannot be analyzed from the selection statistics but must be answered through independent studies. The extensive literature on plant water use suggests that the results I observed are more consistent with a causal relationship between traits and fitness than with environmentally induced covariance. Variation within the dry treatment in environmental water availability would be expected to cause a positive correlation between stomatal conductance and fitness if wetter sites permit more growth and greater water loss but not the observed negative correlation. Variation within the dry treatment in environmental water availability could also be expected to cause a positive correlation between leaf size and fitness if wetter sites permit both larger leaves and more growth but not the observed intermediate optimum for leaf size. The significant path between water-use efficiency and vegetative biomass in the dry environment suggests that, if selection on unmeasured correlated characters is responsible for the significant selection on water-use efficiency, then the unmeasured characters must be associated with carbon uptake.

The selection study found considerable agreement between natural selection in the field and predictions from functional analysis of the traits. This agreement is evidence for the predictive power of both functional studies and studies of natural selection, despite the complexity of possible interactions among correlated traits in determining fitness. This study shows that the overriding importance of the benefits of carbon uptake and costs of water use in predicting adaptations among species in different environments (Horn 1979) may be reflected in the microevolutionary processes within species.


I would like to thank M. Newton for excellent technical assistance and S. Yamins, S. Suwanski, and J. Zdenek for expert greenhouse care of the plants. I thank K. Black, L. Birch, K. Karoly, K. Donohue, M. Wade, and P. Chu for their help with field work. This paper has benefited from the comments of M. Liebold, S. Arnold, E. Simms, J. Schmitt, J. Teeri, S. DeWalt, the reviewers, and especially from my dissertation advisor, M. Wade.


Analytical Development Company. 1992. Instruction manual, type LCA-3 carbon dioxide leaf chamber analysis system. Analytical Development, Hoddesdon, England.

Arnold, S. J. 1983. Morphology, performance, and fitness. American Zoologist 23:347-361.

-----. 1992. Constraints on phenotypic evolution. American Naturalist 140:S85-S107.

Bhagsari, A. A., and R. H. Brown. 1986. Leaf photosynthesis and its correlation with leaf area. Crop Science 26:127-132.

Caemmerer, S. von, and G. D. Farquhar. 1981. Some relationships between the biochemistry of photosynthesis and the gas exchange of leaves. Planta 153:376-387.

Cohen, D. 1970. The expected efficiency of water utilization in plants under different competition and selection regimes. Israel Journal of Botany 19:50-54.

Cowan, I. R. 1986. Economics of carbon fixation in higher plants. Pp. 133-170 in T. J. Givnish, ed. On the economy of plant form and function. Cambridge University Press, New York.

De Jong, T. M. 1978. Comparative gas exchange of four California beach taxa. Oecologia 34:343-351.

DePuit, E. J., and M. M. Caldwell. 1975. Stem and leaf gas exchange of two arid land shrubs. American Journal of Botany 62: 954-961.

Donovan, L. A., and J. R. Ehleringer. 1994. Potential for selection on plants for water-use efficiency as estimated by carbon isotope discrimination. American Journal of Botany 81:927-935.

Dudley, S. A. 1996. The response to differing selection on plant physiological traits: evidence for local adaptation. Evolution 50: 103-110.

Ehleringer, J. R. 1975. Annuals and perennials of warm deserts. Pp. 162-180 in B. F. Chabot and H. A. Mooney, eds. Physiological ecology of North American plant communities. Chapman and Hall, New York.

Ehleringer, J. R., and C. Clark. 1988. Evolution and adaptation in Encelia (Asteraceae). Pp. 221-248 in L. D. Gottlieb and S. K. Jain, eds. Plant evolutionary biology. Chapman and Hall, New York.

Ehleringer, J. R., and T. A. Cooper. 1988. Correlations between carbon isotope ratio and microhabitat in desert plants. Oecologia 76:562-566.

Endler, J. 1986. Natural selection in the wild. Princeton University Press, Princeton, NJ.

Farris, M. A., and M. J. Lechowicz. 1990. Functional interactions among traits that determine reproductive success in a native annual plant. Ecology 71:548-557.

Field, C. B, J. T. Ball, and J. A Berry. 1989. Photosynthesis: principles and field techniques. Pp. 209-254 in R. W. Pearcy, J. Ehleringer, H. A. Mooney, and P. W. Rundel, eds. Plant physiological ecology field methods and instrumentation. Chapman and Hall, New York.

Geber, M. A. 1990. The cost of meristem limitation in Polygonum arenastrum: Negative genetic correlations between fecundity and growth. Evolution 44:799-819.

Givnish, T. J. 1979. On the adaptive significance of leaf form. Pp. 375-407 in O. T. Solbrig, S. Jain, G. B. Johnson, and P. H. Raven, eds. Topics in plant population biology. Columbia University Press, New York.

-----. 1986. Optimal stomatal conductance, allocation of energy between leaves and roots, and the marginal cost of transpiration. Pp. 171-214 in T. J. Givnish, ed. On the economy of plant form and function. Cambridge University Press, New York.

Gurevitch, J., J. A. Teeri, and A. M. Wood. 1986. Differentiation among populations of Sedum wrightii (Crassulaceae) in response to limited water availability: Water relations, C[O.sub.2] assimilation, growth and survivorship. Oecologia, Berlin 70:198-204.

Harris, R. J. 1975. A primer of multivariate statistics. Academic Press, New York.

Horn, H. S. 1979. Adaptation from the perspective of optimality. Pp. 48-62 in O. T. Solbrig, S. Jain, G. B. Johnson, and P. H. Raven, eds. Topics in plant population biology. Columbia University Press, New York.

Kalisz, S., and J. A. Teeri. 1986. Population-level variation in photosynthetic metabolism and growth in Sedum wrightii. Ecology 67:20-26.

Korner, C. H., J. A. Scheel, and H. Bauer. 1979. Maximum leaf diffusive conductance in vascular plants. Photosynthetica 13:45-82.

Lande, R., and S. J. Arnold. 1983. The measurement of selection on correlated characters. Evolution 37:1210-1226.

Li, C. C. 1975. Path analysis - a primer. Boxwood Press, Pacific Grove, CA.

Long, S. P., and J-E. Hallgren. 1985. The measurement of C[O.sub.2] assimilation by plants in the field and in the laboratory. Pp. 62-94 in J. Coombs, D. O. Hall, S. P. Long, and J. M. O. Scurlock, eds. Techniques in bioproductivity and photosynthesis, 2d ed. Pergamon Press, New York.

Mitchell-Olds, T. 1989. Free-stat users manual. Technical Bulletin 101. Division of Biological Sciences, University of Montana, Missoula.

Mitchell-Olds, T., and R. G. Shaw. 1987. Regression analysis of natural selection: Statistical inference and biological interpretation. Evolution 31:1149-1161.

Mott, K., A. C. Gibson, and J. W. O'Leary. 1982. The adaptive significance of amphistomatous leaves. Plant, Cell, and Environment 5:455-460.

Nobel, P. S. 1991. Physicochemical and environmental plant physiology. Academic Press, New York.

Parkinson, K. J. 1984. A simple method for determining the boundary layer resistance in leaf cuvettes. Plant, Cell, and Environment 8:223-226.

-----. 1985. Porometry. Pp. 171-191 in B. Marshall and F. I. Woodward, eds. Instrumentation for environmental physiology. Cambridge University Press, Cambridge.

Philips, P. C., and S. J. Arnold. 1989. Visualizing multivariate selection. Evolution 43:1209-1222.

Rausher, M. D. 1992. The measurement of selection on quantitative traits: Biases due to environmental covariance between traits and fitness. Evolution 46:616-626.

Rodman, J. E. 1974. Systematics and evolution of the genus Cakile (Cruciferae). Contributions from the Gray Herbarium, Harvard University 205:3-146.

SAS Institute, Inc. 1989. SAS/STAT user's guide, Version 6, 4th ed. SAS Institute, Cary, NC.

Scheiner, S. M. 1993. Genetics and evolution of phenotypic plasticity. Annual Review of Ecology and Systematics 24:35-68.

Schulze, E.-D. 1982. Plant life forms and their carbon, water, and nutrient relations. Pp. 615-676 in O. L. Lange, P. S. Nobel, C. B. Osmond, and H. Ziegler [eds.], Physiological plant ecology II. Water relations and carbon assimilation. Springer, New York.

Smedley, M.P., T. E. Dawson, J.P. Comstock, L. A. Donovan, D. E. Sherrill, C. S. Cook, and J. R. Ehleringer. 1991. Seasonal carbon isotope discrimination in a grassland community. Oecologia 85:314-320.

Sokal, R. R., and F. J. Rohlf. 1981. Biometry. Freeman, New York.

Wade, M. J., and S. Kalisz. 1990. The causes of natural selection. Evolution 44:1947-1955.

Winn, A. A., and A. S. Evans. 1991. Variation among populations of Prunella vulgaris L. in plastic responses to light. Functional Ecology 5:562-571.
COPYRIGHT 1996 Society for the Study of Evolution
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Dudley, Susan A.
Date:Feb 1, 1996
Previous Article:Developmental stability in leaves of Clarkia tembloriensis (Onagraceae) as related to population outcrossing rates and heterozygosity.
Next Article:The response to differing selection on plant physiological traits: evidence for local adaptation.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters