L. A. Donovan, Department of Plant Biology, 2502 Plant Sciences Building, University of Georgia, Athens 30602, Georgia, USA. Tel.: +1 706 542 2969; fax: +1 706 542 1805; e-mail: firstname.lastname@example.org
Hybrid speciation is thought to be facilitated by escape of early generation hybrids into new habitats, subsequent environmental selection and adaptation. Here, we ask whether two homoploid hybrid plant species (Helianthus anomalus, H. deserticola) diverged sufficiently from their ancestral parent species (H. annuus, H. petiolaris) during hybrid speciation so that they are more fit than the parent species in hybrid species habitats. Hybrid and parental species were reciprocally transplanted into hybrid and parental habitats. Helianthus anomalus was more fit than parental species in the H. anomalus actively moving desert dune habitat. The abilities to tolerate burial and excavation and to obtain nutrients appear to be important for success in the H. anomalus habitat. In contrast, H. deserticola failed to outperform the parental species in the H. deserticola stabilized desert dune habitat, and several possible explanations are discussed. The home site advantage of H. anomalus is consistent with environmental selection having been a mechanism for adaptive divergence and hybrid speciation and supports the use of H. anomalus as a valuable system for further assessment of environmental selection and adaptive traits.
Hybridization is receiving renewed attention as an important process in speciation (Arnold, 1997; Rieseberg, 1997; Barton, 2001; Baack & Rieseberg, 2007). For homoploid hybridization in plants, where chromosome number remains the same, models and empirical evidence suggest that both fertility selection (i.e. endogenous, genetic or genomic selection) and ecological selection (i.e. environmental or environment-dependent selection) play large roles in the speciation process (Rieseberg et al., 2003; Lexer & Fay, 2005; Karrenberg et al., 2007). Initial hybridization events can reveal cryptic variation via transgressive segregation, creating new phenotypes that can allow them to escape to habitats where the hybrid traits are more successful than parental traits (Rieseberg et al., 1999, 2003). Reproductive isolation of hybrids from parentals is then likely because of several mechanisms (incompatibility of genomes, assortative mating, spatial isolation) and facilitates the potential for further environmental selection in the hybrid habitat (Buerkle et al., 2000; Mavarez et al., 2006; Buerkle & Rieseberg, 2007; Hendry et al., 2007; Lowry et al., 2008a,b). This leads to the expectation that in the hybrid habitats, the hybrid species should have higher fitness than parental species, with the caveat that habitats may have changed since speciation occurred. We test this expectation for two homoploid hybrid Helianthus species.
Helianthus anomalus (sand sunflower) and H. deserticola (desert sunflower) are diploid species resulting from the hybridization of the same two ancestral parental species, Helianthus annuus (common sunflower) and Helianthus petiolaris (prairie sunflower) (Rieseberg, 1991; Rieseberg et al., 1996). Both hybrid species appear to have originated within the last 200,000 years and potentially have multiple origins (Schwarzbach & Rieseberg, 2002; Gross et al., 2003, 2007), although much earlier dates of origin are currently being evaluated (L. H. Rieseberg, unpublished data). The parental species H. annuus and H. petiolaris are widespread and occur throughout the central and western United States in disturbed habitats. Helianthus annuus occurs on mesic clay-based soils, whereas H. petiolaris occurs on relatively drier and sandier soils. The hybrid species are restricted to the semi-arid Great Basin and Colorado Plateau regions of the western United States and are endemic to desert dune habitats that appear to be more extreme than the parental habitats (Comstock & Ehleringer, 1992; Schwarzbach et al., 2001; Gross et al., 2003). Thus, despite overlap in geographic range, the hybrid species habitats are generally spatially isolated from the parent species, with little seed dispersal among the habitats. The hybrid species are also morphologically distinct from the parental species, with a mixture of parent-like and transgressive traits (Schwarzbach et al., 2001; Rosenthal et al., 2002, 2005a; Rieseberg et al., 2003; Karrenberg et al., 2007).
Helianthus anomalus is restricted to actively moving desert sand dunes in Utah and northern Arizona, occupying unstable substrates that have low plant cover and low soil nitrogen (N), but tend to store more water than adjacent stabilized areas with finer soil texture (Pavlik, 1980; Hamerlynck et al., 2004; Rosenthal et al., 2005b; Grigg et al., 2008). Previous studies contrasting H. anomalus morphology to that of parental species have noted likely adaptations to the active desert dune habitats such as large seeds with reserves for substantial root production, succulent leaves that may enhance water status or resist abrasion and lower relative growth rate that conserves nitrogen (Schwarzbach et al., 2001; Brouillette et al., 2006). Phenotypic selection analyses of leaf ecophysiological traits indicate that the ability to maintain a high leaf N is an important adaptation in this habitat (Donovan et al., 2009). Helianthus anomalus does have higher N use efficiency than its parental species, facilitated by longer leaf lifetime and contributing to greater tolerance of low nutrient stress (Aerts & Chapin, 2000; L. C. Brouillette & L. A. Donovan, unpublished data).
Helianthus deserticola occurs on stabilized sandy soils on the desert floor in Nevada, Utah and Arizona. The stabilized dune habitat is characterized by more plant cover and higher soil N than that of H. anomalus, but lower plant water availability as the growing season progresses because of soil texture and competition from other plants (Rosenthal et al., 2005b; Donovan et al., 2007). Previous studies contrasting H. deserticola to parental species have noted likely adaptations to the stabilized desert dune habitats such as smaller leaves that reduce leaf temperatures and transpirational water loss, and earlier flowering that increases the likelihood of reproduction as water availability declines through the season (Rieseberg et al., 2003; Gross et al., 2004). Phenotypic selection analyses reinforce the importance of flowering time in this habitat and additionally suggest that phosphorus and boron nutrition maybe more important than N (Gross et al., 2004; Donovan et al., 2009).
The objective of this study was to test whether two hybrid species (H. anomalus and H. deserticola) are more fit than the parental species (H. annuus and H. petiolaris) in the hybrid species habitats. Previous studies in 2002 provided partial data for addressing this question. In those studies, H. anomalus seedling transplants had greater survival and biomass than those of parental species in the H. anomalus habitat (Ludwig et al., 2004), but H. deserticola seedling transplants failed to outperform the parental species in the H. deserticola habitat (Gross et al., 2004). However, seedling growth and survival is only one component of plant fitness and may not always reflect lifetime fitness (Campbell & Waser, 2007). In this study, we used a combination of seed plot experiments and seedling transplant experiments to look at germination success, seedling growth and survival, and reproductive output to provide a more complete assessment of fitness. The performance of the hybrid and parental species in the parental habitats is also assessed. We additionally measured plant leaf ecophysiological traits related to the carbon gain and water use to determine how they differed across species and habitats, possibly reflecting adaptive traits.
Materials and methods
The study sites were located in Utah, in the western United States. For each study species (H. anomalus, H. deserticola, H. annuus and H. petiolaris), an experimental study area was selected at a location where that species naturally occurred. These experimental areas are designated as the ANO, DES, ANN and PET habitats, respectively. The ANO habitat was located near Jericho Picnic area at Little Sahara Recreation Area (LSRA) managed the Bureau of Land Management. This is an active sand dune habitat with a very low species cover (Rosenthal et al., 2005b). The DES habitat was also located at LSRA approximately 5 km from the ANO habitat on a site with stabilized sandy soils dominated by bunch grasses, the annual invasive grass Bromus tectorum and a few Artemisia tridentata shrubs (Rosenthal et al., 2005b). The ANN habitat was located in Tintic Valley, approximately 40 km from LSRA, on highway 6 just west of Eureka, UT, in a habitat dominated by A. tridentata and Chrysothamnus nauseosus (Donovan & Ehleringer, 1994). The PET habitat was located in southern Utah 2 km east of Zion NP along highway 9 and dominated by grasses and annual herbs. All four experimental areas have a prior history of grazing by cattle or sheep. The experimental areas were fenced to deter cattle, and the plots were weeded to remove native vegetation prior to and during the experiments.
Seeds of the four species were collected during 2002 in Utah and Northern Arizona. For H. anomalus, H. deserticola and H. annuus, the seeds came from LSRA. For H. petiolaris, the seeds came from central Utah along interstate 15 near the exit to highway 20. Seeds were stored in a cold room until use.
Seed plot experiments
Achenes (one seeded fruits, hereafter called seeds) of all four species were planted into all four species habitats. Within each habitat, we established 10 seed plots, each set up as a pentagon 1 m to a side and divided into five triangular subplots. The four species were randomly assigned to four of the subplots, 60 seeds per subplot, for a total number of 600 seeds per species and 2400 seeds per habitat. The remaining fifth subplot was left unplanted as a control to assess volunteer seedlings. Before planting, seeds were cold stratified (4 °C on moist filter paper) for four weeks to approximate the effect of overwintering in the natural environment and to promote germination. Seeds were planted into the seed plots, 5 cm below the soil surface, 21–25 April 2003.
After planting, seed plots were checked weekly, and emergent seedlings marked individually with toothpicks so the total number of seedlings germinating and dying could be followed throughout the season. At the end of the growing season, 3–9 September, subplots were assessed for the number of reproductive units (buds, flowers and seed heads) and then harvested for aboveground biomass. Plants were sorted into vegetative biomass (stem and leaves) and reproductive biomass (buds, flower and seed heads), dried at 60 °C and weighed.
Differences in per cent germination, biomass and numbers of reproductive units were tested with a mixed model analysis of variance (anova, SAS, 2001) with species and habitat as fixed factors, plot as a random factor nested with habitat and degrees of freedom determined with Satterthwaite method. Differences between species and habitats were tested with a post hoc LSmeans procedure. Germination data were arcsine transformed and biomass data were log transformed before statistical analyses to better meet anova assumptions.
Seedling transplant experiment
Seed germination was initiated 3–7 April 2003 at the University of Georgia, following the protocols of Schwarzbach et al. (2001) and Gross et al. (2004). After germination and initial seedling growth, seedlings were transported from Georgia to Utah by truck 8–10 May and then maintained outside in Utah to acclimatize to local temperature, humidity and UV conditions. Seedlings were transplanted into the seedling plots on 15–18 May and watered approximately every other day until 10 days after planting. Plants that died prior to 22 May, because of transplant shock, frost damage or burial, were excluded from the data analyses. Seedlings of all four species were transplanted into all four habitats, except that H. anomalus was not transplanted into the DES habitat because not enough individuals were available. For the ANN, PET and DES habitats, seedlings were transplanted into 6 blocks for a total of 60 H. anomalus, 60 H. deserticola, 60 H. annuus and 60 H. deserticola seedlings in each habitat (except no H. anomalus seedlings in DES garden). For the ANO habitat, seedlings were transplanted into 8 blocks, but one block was destructively harvested mid-season and excluded from this study, leaving 7 blocks and total of 182 H. anomalus, 70 H. deserticola, 126 H. annuus and 126 H. deserticola seedlings in the ANO habitat. We used more individuals in the ANO habitat because this study was combined with a phenotypic selection experiment with artificial hybrids. The results of the selection analysis for H. anomalus are presented in Donovan et al. (2009).
After study initiation on 22 May, plants were checked for survival and presence of the first flower every week until final harvest. Plants were considered dead when all leaves were wilted or plants were completely buried in the sand. Aboveground biomass was collected for plants that died. Mature seed heads were collected every three weeks to prevent them from falling on the ground and to be able to determine total number of seed heads produced per plant. Because many seeds matured and dispersed between collections, it was not possible to determine number of seeds for all seed heads and plants. However, maturing seed heads were collected at a more frequent interval (2–3 days) as they matured, 12–19 August in the ANO, ANN and DES habitats and assessed for number of seeds per seed head, seed biomass and seed head biomass. In the PET habitat, it was not possible to collect seed heads this frequently because of the distance from the other habitats.
Leaf traits were assessed 28 June–3 July for each live plant with more than four leaves and at least one fully expanded leaf for sampling: leaf size (individual leaf area), succulence, N concentration and water-use efficiency (WUE, ratio of photosynthetic carbon gain per unit transpirational water loss). The youngest fully expanded leaf was collected from each plant between 6 and 8 am (when maximally hydrated), placed in Ziploc bags to maintain turgor and subsequently measured for leaf wet biomass and leaf area (CID, Inc., Pullman, WA, USA). Leaves were then dried at 60°C and weighed. Leaf succulence was calculated as: (wet weight – dry weight)/leaf area (Jennings, 1976). The leaves were then individually ground and analysed for N concentration (per dry leaf biomass) (Carbo Erba NA 1500 elemental analyzer, Milan, Italy) and leaf carbon isotopic composition (δ13C, Finnegan mass spectrometer, Breman Germany). Leaf δ13C provides a time-integrated measure of leaf intercellular CO2 concentration (ci). Integrated ci is, in turn, a relative measure of integrated instantaneous WUE, provided leaf temperatures are similar (Farquhar et al., 1989; Ehleringer, 1993). Greater (less negative) leaf δ13C reflects greater WUE.
At the end of the growing season, 3–9 September, all surviving plants were assessed for the number of reproductive units (buds, flowers and seed heads) and then harvested for aboveground biomass. Plants were sorted into vegetative biomass (stem and leaves) and reproductive biomass (buds, flower and seed heads), dried at 60 °C and weighed.
Differences in biomass, reproductive organs produced per plant and leaf traits were tested with a mixed model anova with species and habitat as fixed factors, block as a random factor within habitat and degrees of freedom determined with Satterthwaite method. Differences between species and habitats were tested with a post hoc LSmeans procedure. Biomass data were log transformed before statistical analyses to better meet anova assumptions.
Lifetime fitness for each species and habitat was then calculated as the product of three stages: (i) per cent germination from seed plots adjusted to account for species differences in maximum germination across habitats (i.e. multiplied by a factor of 1.48, 4.86 and 2.94 for H. anomalus, H. deserticola and H. petiolaris, respectively), (ii) per cent survival of transplanted seedling to first flower (from seedling plots) and (iii) average reproductive output (seed number) for plants that flowered (from seedling plots). Based on a subsampling of seed heads as they matured over short interval in August, the number of seeds per seed head was assessed for each species in the ANO, ANN and DES habitats. To estimate the number of seeds per seed heads for each species in the PET garden, we first used the data from the ANN garden because that were the most similar to the PET garden for many traits, but we also calculated an average for each species from all other gardens. We present the former, but the fitness calculated by the methods is correlated (r2 = 0.97, P < 0.001, n = 11, i.e. all available species*plot combinations), and resulted in the same species rankings for fitness in the PET habitat. From the number of seeds per seed head and the number of reproductive units, we estimated the total number of seeds for each flowering plant.
Soil nutrient analysis
Soils were sampled 25–27 August for soil nutrient analysis. Five soil cores were taken per habitat, with soils sampled at 0, 25, 50 and 75 cm depths. Soils from all depths were dried at 60 °C, ground with a ball mill. Soil P was estimated from acid persulfate extracts (Nelson, 1987) using Alpkem continuous-flow colorimetry. Soil organic matter was determined by loss on ignition (Schulte & Hopkins, 1996). Soils from 0 and 25 cm were additionally assessed for pH and electrical conductivity (EC, a common measure of soil salinity) (Robertson et al., 1999). Samples were also submitted for analysis of N, but the samples were all below the detection limit for %N and there was insufficient soil remaining to reanalyse with a more sensitive method. Differences in soil characteristics were tested with an anova with habitat as a fixed factor and soil depth as a continuous factor followed by an LSmeans test for differences between habitats.
Seed plot experiment
For all four species and habitats, the germinating seedlings generally emerged within a week of planting, peaked in number alive during the first few weeks and then declined to relatively stable numbers through the remainder of the growing season (Fig. 1), except for July and August mortality in the DES habitat. The per cent germination (based on marked seedlings as they emerged) differed by species and habitat, and there were significant species by habitat interactions (Table 1). Looking first at species differences in maximum germination across habitats, H. annuus had the highest germination rates (40.3 ± 3.9% in ANN habitat), followed by H. anomalus (27.3 ± 4.3% in ANN habitat), H. petiolaris (13.7 ± 1.9% in PET habitat) and H. deserticola (8.3 ± 1.5% in ANN habitat). Seedling germination, survival, biomass and reproduction on a subplot basis all differed by species and habitat, with significant species by habitat interactions (Table 1).
Table 1. Seed plot performance of two hybrid species (Helianthus anomalus, H. deserticola) and their ancestral parent species (H. annuus, H. petiolaris) planted into each of the four species habitats (ANO, DES, ANN, PET, respectively): germination, survival for germinated seedlings, biomass and number of reproductive unit per plots.
Total Biomass (g)
Reproductive Biomass (g)
Number reproductive units
For F statistics, *P < 0.5, **P < 0.01, ***P < 0.001.
Data are means (SE) of 10 plots per garden. In each plot, 60 seeds of each species were planted. For each trait, means with the same letter are not significantly different as tested with LSmeans procedure after mixed model anova.
In the ANO habitat, a few seedlings of each species emerged, but germination and survival were highest for H. anomalus (Fig. 1, Table 1). Only H. anomalus seedlings survived to reproduction in the seed plots in the ANO habitat. For H. anomalus, germination, biomass and number of reproductive units per subplot were lower in the ANO habitat when compared to the parental habitats (ANN and PET). However, in the parental habitats, the parental species had greater biomass and more reproductive units than H. anomalus.
In the DES habitat, germination and survival were generally higher than in the ANO habitat (Fig. 1, Table 1). However, H. deserticola did not do better than the parental species in the DES habitat for germination, survival, biomass or reproductive units per subplot. Helianthus deserticola had similar performance in its own habitat (DES) when compared to the ANN habitat, but generally did better in the PET habitat. In the parental habitats, the parental species generally had greater biomass and more reproductive units than H. deserticola.
In both the ANN and PET habitats, H. annuus had higher germination and biomass than H. petiolaris per subplot (Fig. 1, Table 1). In the PET habitat, H. petiolaris produced more reproductive units.
Transplant seedling survival, biomass and reproduction
Seedlings of all four species were transplanted into all four habitats except that H. anomalus seedlings were not transplanted into in the DES habitat. Seedling survival across seedling transplant plots was not statistically compared by species and habitat, because there was only one seedling transplant plot per habitat. However, seedling survival at final harvest, by species and habitat (Fig. 2), roughly paralleled survival observed in the seedling plots: H. anomalus seedling survival was highest in its own habitat, but this was not true for H. deserticola (Table 2). The declining survival of plants through the growing season reflected both preflowering mortality and naturally occurring senescence after flowering, particularly for H. deserticola and H. petiolaris. Therefore, survival of transplanted seedlings to flowering was separated out as a specific component to be included in our estimates of lifetime fitness (Table 2). Helianthus anomalus seedling survival to flowering was highest in its own habitat, but again, this was not the case for H. deserticola.
Table 2. Seedling performance of two hybrid species (Helianthus anomalus, H. deserticola) and their ancestral parent species (H. annuus, H. petiolaris) planted into each of the four species habitats (ANO, DES, ANN, PET, respectively): date of first flower, total biomass, reproductive biomass, number of reproductive units, leaf size, leaf succulence, leaf N and leaf δ13C. Data are means (S.E.). For each trait, means with the same letter are not significantly different as tested with LSmeans procedure after mixed model ANOVA.
Survival first flower (%)
Date of first flower
Total biomass (g)
Reproductive biomass (g)
Seeds per flowering plant
Leaf size (cm2)
Leaf succulence (mg cm-2)
Leaf N (%)
Leaf δ13C (‰)
For F statistics, *P < 0.5; **P < 0.01; ***P < 0.001.
For transplanted seedlings, flowering generally initiated earliest for H. deserticola and H. petiolaris in the hybrid species habitats (ANO and DES) and latest for H. annuus and H. anomalus in the ANN and ANO habitats (approximately 2–4 week later). Biomass and reproduction on an individual seedling basis roughly paralleled that of the combined seedlings in seed plots. In the ANO habitat, H. anomalus had greater biomass and more reproductive units than the other species (Table 2). Helianthus anomalus also had greater biomass and more reproductive units in its own habitat when compared to parental ANN and PET habitats. In the DES habitat, H. deserticola did not have greater biomass or more reproductive units than the other species. Helianthus deserticola also did not have greater biomass and more reproductive units in its own habitat when compared to parental ANN and PET habitats.
In both the ANN and PET habitats, H. annuus plants had greater biomass than H. petiolaris. Helianthus petiolaris produced more total reproductive units in the PET habitat.
For transplanted seedlings, leaf traits early in the growing season differed by species and habitat, with significant species by habitat interactions (Table 2). The species differences evident across all habitats were that leaf succulence was highest for H. anomalus, leaf size was largest for H. annuus and leaf size tended to be smallest for H. deserticola. Comparing habitats, leaf N was generally highest for plants in the ANN habitat, followed by the DES and then the ANO and PET habitats. Leaf WUE was lowest for plants in the ANO habitat. Comparing species, H. anomalus had a lower leaf N concentration than the parental species in the parental habitats, but in its home habitat, it maintained a leaf N concentration similar to or greater than the parental species.
Lifetime fitness was estimated by integrating data from the seed plots and the seedling plots (Fig. 3). Lifetime fitness for each species and habitat was calculated as the product of three stages: (i) per cent germination (from seed plots, Fig. 1, Table 1) adjusted to account for species differences in maximum germination across habitats, (ii) per cent survival of transplanted seedlings to flowering (from seedling plots, Fig. 2, Table 2) and (iii) average seed production for plants that flowered (from seedling plots, Table 2). Helianthus anomalus had the highest lifetime fitness in its own habitat, as did each of the parental species. Helianthus deserticola did not have the highest fitness in its home habitat.
The habitats differed for soil characteristics. Soil organic matter content differed by habitat and depth (habitat F3,76 = 605.1, P < 0.001; depth F3,76= 10.3, P < 0.001; habitat*depth F9,76= 15.4, P < 0.001). The ANN habitat had the highest soil organic content, followed by successively lower amounts in the PET, DES and ANO habitats (Fig. 4). Soil P also differed by habitat and depth (habitat F3,78= 97.3, P < 0.001; depth F3,78= 23.5, P < 0.001; habitat*depth F9,78= 4.0, P = 0.004). The ANN and PET habitats had higher soil total P than the DES and ANO habitats.
Soil pH differed by habitat and depth (habitat F3,37= 160.9, P < 0.001; depth F1,37 = 46.9, P < 0.001; habitat*depth F3,37 = 10.9, P < 0.001). The ANN, PET and habitats generally had a lower soil pH than the DES and ANO habitats. Soil EC also differed by habitat, although not as dramatically as for other soil characteristics (habitat F3,37 = 5.4, P < 0.004; depth F1,37 = 2.6, P = 0.120; habitat*depth F3,37 = 2.3, P = 0.097 The ANN, PET and DES habitats generally had a higher EC than the ANO habitat.
We tested whether the hybrid species H. anomalus and H. deserticola were each more fit than their parental species in the hybrid species habitat, which would be consistent with the hypothesized role of environmental selection in the hybrid speciation process. Although these two hybrid species are closely related (from same ancestral parent cross) and both endemic to desert sand dune habitats, the two hybrid species yielded very different results in this study.
Helianthus anomalus was the most fit in its own habitat, based on seed germination and transplanted seedling performance. For the seed plots, H. anomalus had much higher germination and survival than parental species in the ANO habitat. This indicates strong selection at the recruitment stage, probably because of the need to survive burial and excavation in actively moving sand dune habitat. In the ANO habitat, we observed spring storms to add or remove as much as 10 cm of the sandy soil at some locations, with the amount and location of change being highly variable and dependent on wind speed and direction. Helianthus anomalus has much larger seeds and thus potential reserves to achieve necessary growth in response to burial or excavation (Chen & Maun, 1999; Schwarzbach et al., 2001). It would be interesting to test whether H. anomalus also has greater ability to tolerate burial, reallocate biomass to shoot growth when buried, elongate stems when buried, or suberize roots when exposed (Brown, 1997).
For the seedling transplants, H. anomalus survival to flowering was higher than parental species in the ANO habitat, consistent with results from the previous year (Ludwig et al., 2004). The higher survival of H. anomalus relative to parental species in the ANO habitat was not obviously because of any single factor. Rather, parental species failed to thrive, possibly because of ANO habitat soil characteristics of lower fertility and EC, higher pH and extremely sandy soil texture that affect rooting architecture, soil cation exchange capacity and water retention (Pavlik, 1980; Hamerlynck et al., 2004; Rosenthal et al., 2005b; Grigg et al., 2008). Additionally, sand abrasion of leaves and other unknown factors likely played a role. The higher survival of H. anomalus seedlings in the ANO habitat did not appear to be because of differences in plant water status, based on a lack of species differences in plant predawn water potential measurements at mid-season (L. A. Donovan, unpublished data). The estimated number of seeds produced per flowering plant did not differ by species within the ANO habitat, although the low sample sizes limited statistical power. Nevertheless, the individual seed germination and seedling transplant studies, as well as the resulting estimate of integrated lifetime fitness, all demonstrate a clear home site advantage for H. anomalus.
The greater success of H. anomalus in its own habitat came at the expense of its performance in parental habitats. Helianthus anomalus had higher or similar germination in the parental habitats compared to its home habitat, but seedling survival to flowering and seed production per flowering plant was lower in the parental habitats. The lower inherent relative growth rate of H. anomalus may reflect a lower competitive ability, which might be a factor in its lack of success in the higher plant cover of the parental habitats (Grime, 1977; Brouillette et al., 2006). However, in our experiments, the lower performance of H. anomalus in the parental habitats likely was not because of competition because the plants in the seedling plots were spaced at least 30 cm apart, excavated roots did not appear to be intertwined and all volunteer plants were periodically removed. We hypothesize that the rooting and nutrient-related traits that allow for H. anomalus success in its unstable sandy dune habitat come at the cost of success in the parental habitat soils that are stable and higher in clay content.
Helianthus deserticola was not most fit in its own habitat in this study. For the seed plots, relatively few H. deserticola seeds emerged in any habitat and very few of the emerged plants survived until reproduction. Seedlings of H. deserticola transplanted into the DES habitat did better than the seeds, but the parental species still had a higher survival and produced more flowers and reproductive biomass than H. deserticola. These results are consistent with the seedling transplant data from the previous year (Gross et al., 2004). Thus, none of the results for individual fitness components, or integrated lifetime fitness, indicate that H. deserticola is more fit than the parental species in this DES habitat.
We can offer several possible interpretations of this unexpected result. One interpretation is that in general (across all sites and years), H. deserticola is not more fit than parental species in DES habitats, suggesting that environmental selection did not play a large role in speciation. Although this is a possibility, we think that this interpretation is not yet warranted given the strong modelling and genetic evidence of the importance of environmental selection in hybrid speciation (Karrenberg et al., 2007) and that this study assessed only one H. deserticola site and one year. One obvious way to further address this question would be to repeat the study with more sites and seed sources for each of the species and to capture genetic and environmental variation among populations (Schwarzbach & Rieseberg, 2002; Gross et al., 2007). This would strengthen the inference of the study as a whole, but was not possible in the current study because of the extensive investment of time and resources in the seed and seedling studies at the four chosen sites. Until more data are available, we offer several other possibilities for consideration.
One possibility is that the H. deserticola advantage in DES habitat is only apparent in extreme years that were not represented this study year with average precipitation. Desert habitats are known for their extreme annual variability in climate, particularly rainfall (Noy-Meir, 1973; Comstock & Ehleringer, 1992), and the early flowering of H. deserticola may be particularly advantageous in drought years when soil water is depleted early in the growing season. However, the seedling transplant growth and survival data from this study are consistent with those in 2002 (Gross et al., 2004), which was the driest year out of the 28 years for which precipitation data are available for this area (1980–2007). Thus, although possible, we currently have no support for this hypothesis.
A second possibility is that H. deserticola does not have a home site advantage at this site and time because hybrid speciation and population establishment occurred when conditions were different. This site is at the northern edge of the current range, and conditions have been changing in terms of longer term climate and shorter term invasions of exotic species. The climate of the region where H. deserticola is found (Great Basin and adjacent Colorado Plateau) has fluctuated since the species is estimated to have arisen. Paleoclimate data for the last 50 000 years indicate that precipitation and temperatures were cooler and wetter during the North America glacial maxima, followed by a drying and warming trend over the last 13 000 years (Wharton et al., 1990). However, records from pollen and woodrat middens suggest that over this time interval, varied topography (elevation, aspect) and microhabitat diversity provided local refugia, allowing major vegetation associations to persist despite climate change (Wharton et al., 1990; Nowak et al., 1994). Given the complexity of potential plant and climate interactions, it is not likely that we will be able to reconstruct the long-term success of the hybrid species when compared to parental species at this particular site.
On a shorter timescale, invasive plants have modified much of the region where H. deserticola is found. For example, since the late 1900s, the non-native annual grass Bromus tectorum (cheatgrass) has invaded the Great Basin and Colorado Plateau regions, facilitated by domestic livestock grazing, and reduced the number of native grasses and annual forbs (Knapp, 1996). Although not found on the unstable dunes where H. anomalus thrives, B. tectorum is often the dominant herbaceous plant on the stabilized substrates where H. deserticola is located. Bromus tectorum tends to out-compete other annual species by depleting soil moisture early in the growing season (Knapp, 1996). This might explain the pattern that we observed while collecting seeds of H. deserticola: it occurs only in areas where B. tectorum is either locally absent or in low density. Additionally, B. tectorum alters soil nutrient and carbon dynamics and biota (microarthropods, nematodes, arbuscular mycorrhizal fungi), facilitates dispersal of some pathogens and may change the herbivores in the community (Sperry et al., 2006; Meyer et al., 2008; Rowe & Brown, 2008). In the DES habitats study plots, we removed the B. tectorum before planting and weeded out any volunteers, so any effects of B. tectorum were not because of direct competition. However, it is possible that an invasive plant like B. tectorum, or some other recent change at this DES site, made the environment less favourable for H. deserticola when compared to parental species. It might be possible to test this hypothesis if H. deserticola sites can be found that vary in the extent and duration of B. tectorum invasion.
Another related possibility is that there is a local site disadvantage because of local adaptation of pathogens, as has been found for California dwarf flax and the rust fungus Melampsore lini (Springer, 2007). Our experiments confounded habitat site with seed collection site for the two hybrid species. Again, reciprocal transplants among multiple species sites and seed sources would be needed to evaluate this possibility.
Both of the parental species had a home site advantage based on the lifetime fitness estimate, although this advantage was not apparent in all of the seed and seedling stages. This is consistent with the finding that parental species’ cytoplasms were strongly locally adapted for hybrids of known parentage transplanted into each species habitat in New Mexico (Sambatti et al., 2008).
Why did the hybrid species differ in their results? One likely reason is that the ANO habitat differed from the parental habitats more than the DES habitat. In the ANO habitat, the disturbance of the moving sand and the low fertility apparently imposes stronger selection than in the DES habitat. The ranking of soil organic content, a rough measure of nutrient availability, was ANO < DES < PET < ANN habitats in this study. This is consistent with previous reports of lower soil organic content and lower soil N in ANO than in DES habitats (Rosenthal et al., 2005b; Donovan et al., 2007) and nutrient limitation of H. anomalus growth in the ANO habitat (Ludwig et al., 2006). The only soil N data available for the parental habitats come from seed collection trips in 1999 and 2000 (n = 5–6 sites per habitat and 2–4 replicates per site, L. A. Donovan, unpublished data). For those collection sites, soil N was lower for the ANO when compared to the ANN and DES sites (0.017 ± 0.004, 0.060 ± 0.011, 0.052 ± 0.01 mg g−1, respectively, P < 0.003). Thus, the two parental habitats were again more similar to each, and ANO habitat most extreme. The greater similarity between the DES and parental habitats may have made the species rankings in this hybrid habitat more labile in response to climate change, invasive species and/or local habitat adaptation.
The extreme nature of the ANO habitat is also reflected in the leaf trait data. Generally, plants in the ANO habitat exhibited among the lowest WUE, as expected based on greater water availability, and among the lowest leaf N, as expected based on lowest N availability. The leaf trait patterns can also be related to phenotypic selection results within the hybrid species habitats. Although all species tended to have higher succulence in the ANO habitat, and H. anomalus consistently had the highest leaf succulence in each habitat, phenotypic selection analysis has not demonstrated any adaptive advantage for higher leaf succulence in the ANO habitat (Donovan et al., 2009). The relatively high ranking of H. anomalus in the ANO habitat for leaf N, compared to its lower ranking in the parental habitats, is consistent with greater leaf N and associated traits being important for fitness in the ANO habitat (Donovan et al., 2009).
Overall, one of the hybrid species and both parental species were more fit in their home habits than the other habitats. This is consistent with a home site advantage and thus local adaptation, as has been found in many reciprocal transplant studies of populations and species (e.g. Joshi et al., 2001; Campbell & Waser, 2007; Schemske & Bierzychudek, 2007; Lowry et al., 2008a,b; Hereford, 2009; but see Kimball et al., 2008). In this case, the known evolutionary relationship between the species, and the models of hybrid speciation mechanisms, allow us to interpret the clear home site advantage for H. anomalus as consistent with environmental selection having been a mechanism for adaptive divergence and hybrid speciation. This makes H. anomalus a valuable system for further assessment of environmental selection and adaptive traits. For H. deserticola, the jury is still out.
We appreciate the help with the experiments from Jennifer Lance, Jill Johnston, Briana Gross, Nolan Kane and Christian Lexer. We also thank Ferris Clegg, the Bureau of Land Management, and Little Sahara Recreation Area for use of the field site and Utah State University for use of Tintic field station. This project was funded by National Science Foundation grants 0131078 and 0614739 to LAD and National Institute of Health grant GM59065 to LHR.