Regional Genetic Differences in Forest Herbaceous Species

The appropriate collection zone for seeds and transplants of native shrubs, grasses, and forbs is a key concern for plant restoration ecology, and has been the subject of a number of extensive reviews (Hufford and Mazer 2003, McKay et al. 2005). This issue was also the topic of a 2014 workshop that included academics, practioners, and nursery growers (Herman et al. 2014). “How local is local?” is a succinct way to summarize the issue of uncertainly about the size of collection zones (McKay et al. 2005).

If natural selection produces ecotypic variants that are better suited to the local environment compared to non-local genotypes, this is often termed home-site advantage. As a result, there are multiple concerns about the potential failure of non-local genotypes in restoration, including outbreeding depression, founder effects, and introduction of maladapted genotypes (Handel et al. 1994, Hufford and Mazer 2003, McKay et al. 2005). At the opposite end of the spectrum, it is also possible that highly successful non-local genotypes could swamp local populations numerically or in fitness advantage, resulting in species-wide loss of genetic diversity (Hufford and Mazer 2003).

In contrast, a high degree of gene flow, which connects populations genetically through the movement of pollen or seeds, can hinder among population genetic differentiation (Loveless and Hamrick 1984). Breeding system, floral morphology, pollinator system, seed dispersal, and life cycle are related to degree of gene flow and therefore to among population genetic structure (reviewed by Loveless and Hamrick 1984).

Despite the questions outlined above, we lack information about among population genetic variation for most native shrub, grass and forb species that would be candidates for restoration (McKay et al. 2005, Rice and Emery 2003, Johnson et al. 2004). This is especially the case for forest herbaceous species, which historically have been less widely used in restoration by conservation professionals, and less known about and supplied by nursery professionals (Altrichter et al. 2017).

My goals in this study were to: 1) determine whether there was genetically based regional variation in traits for forest herbaceous species that would be candidates for forest ground layer restoration, and 2) assess whether any among population genetic variation observed could be related to whether species were primarily self pollinated or outcrossed. My goal was not to delineate collection zones per se, but to evaluate a very practical issue related to restoring woodland groundlayer species: most species of interest are not available for purchase locally, and practioners must look to regional nurseries to supply plant material (Altrichter et al. 2017). Thus, knowledge of regional genetic differences are needed to help guide practical decisions about where to purchase plants.

To test for among population genetic variation at the regional level, I compared populations of forest herbaceous species between central Iowa and northeast Iowa/ southeast Minnesota, approximately 320–400 kilometers apart. I chose to investigate at the regional level because this is the appropriate scale for common, widespread species (Montalvo and Ellstrand 2001), and because in this region there are few nurseries that supply shade tolerant species, making it necessary to obtain plant material regionally rather than locally (Altrichter et al. 2017).

Specifically, I chose 10 species across six families that are offered by native plant nurseries in the Midwest region, USA (Table 1). These species, while not rare, are likely to be included in woodland and savanna restoration due to their beauty, resilience, and nutrient uptake capacity (Mottl et al. 2006, Mabry et al. 2008, Gerken Golay et al. 2013). Species were also selected to represent a range mating systems (Table 1), which may influence gene flow, and thus influence the potential for among population genetic differences (Loveless and Hamrick 1984). Nomenclature follows USDA Plants Database (USDA, NRCS 2016).

I obtained non-local seeds of each species from Prairie Moon Nursery, Winona, MN, with seed source originally from Filmore, Winona, Brown, Pine, and Houston Counties in Minnesota; Clayton and Allamakee Counties in Iowa, and Rock and Pierce Counties in Wisconsin. Local central Iowa seeds of each species were hand collected from natural populations in Polk County, Iowa, for all species except Aquilegia canadensis (wild columbine), which was collected in Jasper County, Iowa, 50 kilometers east of Polk County.

Plants were grown in greenhouses at Iowa State University, Ames, IA, representing a common garden. Because greenhouse bench space was limited, and the process of growing and measuring the plants was time intensive, the study took place over a span of four years (2004, 2006– 2007, and 2008). The methods were the same for each round, with the exception of sample sizes. For the plants grown in 2004, sample size was 24 plants for most traits (8 plants arrayed over 3 blocks that corresponded to separate greenhouse benches), and in 2006–2007 and 2008, sample size for each species was 18 plants (six plants arrayed over the three blocks). Due to variation in reproduction and mortality, however, actual sample sizes varied by trait and are given with Tables 2–3. All seeds were cold-stratified for between 2–3 months, then sowed in flats in the greenhouse. When seedlings had approximately four leaves, they were transplanted to individual pots (approximately 4 cm). If this pot size was inadequate to prevent root binding, plants were later moved to bigger pots.

Although some trials took place during different years, both the local and non-local populations within a species were compared in the same year, and therefore subject to identical conditions. Comparisons between northeast Iowa/southeast Minnesota (non-local) and Iowa (local) populations were therefore not confounded by different experimental conditions and protocols. In addition, all procedures, from stratification, sowing, potting, measuring, and final harvest were identical between years.

Because plant traits act independently, a range of characters were measured for each species (Via et al. 1995). All species were measured at regular intervals until they reached reproductive maturity, but for simplicity only the final measurements are presented. In addition, at one sample interval, leaf chlorophyll content was measured for each species. I used a SPAD 502™ chlorophyll meter (Spectrum Technologies, Inc., Aurora, IL) to measure chlorophyll content on three separate leaves of each plant. These values were then averaged to obtain a mean for each plant.

Four species, A. canadensis, Elymus villosus (downy wild rye), Festuca subverticillata (woodland fescue), and Solidago flexicaulis (zig zag goldenrod) were the subject of two, rather than one, greenhouse trial, both so that I could test methodology, and to test whether the greenhouse would produce consistent results. Seeds were purchased or collected separately for the two trials and represent independent subsamples of the population. For all species, differences in population means were evaluated using 95% confidence intervals.

I also coded each species as primarily autogamous (selfing), primarily outcrossing, or mixed/facultative outcrossing based on a review of the literature (Table 1): Anemone virginiana (thimbleweed) (Molano-Flores and Hendrix 1998), A. canadensis (Routley et al. 1999), Campanula americana (tall bellflower) (Galloway et al. 2003), E. villosus (Gould and Shaw 1983, L. Clark, Iowa State University, pers. comm.), F. subverticillata (Gould and Shaw 1983, L. Clark, Iowa State University, pers. comm.), Lobelia inflata (Indian tobacco) (Simon and Johnston 2000), Polygonum virginianum (jump seed) (Cruden and Lyon 1985), Silphium perfoliatum (cup plant) (Loewe and Boe 2001), Solidago flexicaulis (zig zag goldenrod) (pers. obs), and Teucrium canadense (American germander) (Caddell 2014).

I found evidence for regional genetic variation in morphology for seven of the 10 species examined in this study (Table 2). Previous research has revealed evidence of differentiation for two of four additional forest perennial herbaceous species that our research team has examined (Mabry and Caruso 2006, Gerken Golay et al. 2013), including an additional self pollinating species, Asarum canadense (wild ginger) (Gerken Golay et al. 2013). The results for the species that were subject to two green-house trials, A. canadensis, E. villosus, F. subverticillata and S. flexicaulis, were consistent across the two trials (Tables 2 and 3), strengthening the conclusion that the subsamples of each population were representative of the overall regional genetic differences among populations (Richards et al. 2006).

In addition, five of the 10 species measured for chlorophyll content differed, suggesting that there are genetically based physiological differences as well (Table 3). The three species that were similar morphologically, C. americana, S. perfoliatum and S. flexicaulis, also had similar chlorophyll content (Tables 2 and 3) and were therefore consistent for morphological traits and the physiological trait. Both of the two primarily self pollinating species, A. canadensis and L.inflata, exhibited differentiation based on morphology and leaf chlorophyll content (Tables 1–3). In contrast, no consistent patterns emerged between mating system and among population genetic differentiation for species that outcross or have mixed mating systems, either in this or in our previous research (Tables 1–3; Mabry and Caruso 2006, Gerken Golay et al. 2013).

The results for the three self pollinated species in this and in our previous work matched the prediction that regional genetic differences would occur in these species. This suggests that a more conservative collection zone be considered for species that primarily self-pollinate. Because we lack specific data on the what the size that zone should be, practioners who have concerns about introducing inappropriate genotypes may want to consider collecting self-pollinating species locally.

One point of uncertainty about the ecological implications of regional genetic differences is that local ecotypes may be less important with global climate change (Handel 2013). As climate change and associated changes in weather patterns become a reality, restoration, rather than being concerned about disrupting local ecotypes, may require new genotypes obtained from broad locales that can tolerate these new conditions (Handel 2013). In fact, non-local sources may be better adapted if the climate is changing faster than traits evolve through natural selection (Johnson et al. 2004). Thus, under a scenerio of changing climate it may be necessary to include collection zones large enough to ensure that new genotypes are obtained.