DOT Project Number:  90-00-LRTF-613

Fiscal Year:  2006

Award:  $28,244

Principal Investigators:  Andrea Blong and Dr. Brian Wilsey, Department of Ecology, Evolution and Organismal Biology, Iowa State University, bwilsey@iastate.edu

Summary Report:

Native cover crop effects on prairie establishment

Introduction

Assembly rule theory predicts that communities form through deterministic processes such that community members regulate the identity of future colonists. Diamond (1975) first proposed that assembly rules exist after observing the distribution of bird species on a series of islands. He discovered that certain species never occurred together on the same island, in what he termed “forbidden combinations.” Gotelli and McCabe (2002) conducted a meta-analysis of 96 species presence-absence matrices across several taxonomic groups. They found that in most communities, species co-occurred less than expected by chance, supporting Diamond’s findings. However, they did not suggest any specific assembly rule to account for such patterns of co-occurrence.

One rule that may be applied to multiple taxa is the guild (or functional group) assembly rule proposed by Fox (1983). The guild assembly rule predicts that the presence of guild members can prevent colonization by other species of that guild. Presence of guild members would lead to the greater colonization of a species from the other guilds. Although some consider abiotic filtration of species to be an assembly rule (Weiher and Keddy 1999), for the purposes of this paper only plant-plant interactions that have the potential to alter community composition (e.g., competition or facilitation) shall be considered for possible community assembly rules.

Although the guild assembly rule was developed for mammals (Fox 1983), efforts have been made to test it with plant functional groups (plant species that perform similar functions in the community, e.g. fix nitrogen). Wilson and Whittaker (1995) found evidence for functional groups in a salt marsh community that was correlated with leaf morphology. Fargione et al. (2003) found evidence for the functional group assembly rule using the perennial grassland functional groups of C₃ graminoid, C₄ graminoid, leguminous forb, and non-leguminous forb in a sand prairie system. However, they tested this assembly rule using the biomass of only one species from each functional group, not the total biomass of all the members of each group. Fargione and Tilman (2005) found that plots containing higher proportions of a dominant, shallow-rooted C₄ prairie grass contained fewer non-planted species possessing shallow roots and mid-season phenology, again supporting the assembly rule proposed by Fox (1983).

An early-emerging species is one that is among the first to germinate during restoration. Early-emerging plant species can potentially have large effects on later colonists, such as by limiting water or nutrient availability. Assuming that species do vary in important functional traits, one can test for the presence of assembly rules by experimentally controlling the functional group identity of the dominant early-emerging species in a developing community. Important traits could include longevity, phenology, and differences in productivity.

Annual species often dominate early successions/restorations, and may prevent invasion of perennial species. Polley et al. (2006) showed in Texas that winter annual invaders significantly decreased soil moisture early in the growing season, thus significantly lowering the growth of perennial prairie species later in the year. Biennial species and C₃ graminoids grow primarily during the cooler parts of the growing season, which could prevent other cool-season species from invading the community. A C₄ graminoid, Schizachyrium scoparium, was shown to exclude other species possessing a mid-season phenology (Tilman and Fargione 2005). Grasses tend to produce large quantities of long-lasting litter (Facelli and Facelli 1993), which would lead to invasion by shade-tolerant species. Leguminous forbs could increase soil nitrogen content, allowing nitrophilic species a competitive edge over other species.

Chase (2003) predicts that high productivity sites are more likely to experience the effects of assembly rules than low productivity sites, thus leading to diverging community compositions. Low productivity sites are predicted to attain a single community composition, as abiotic forces more strongly structure the community than plant-plant interactions. Furthermore, plant interactions, and therefore assembly rules, may also be contingent upon the abiotic environment. Bertness and Callaway (1994) demonstrated increasing positive plant interactions as abiotic stress increased, and increasing negative plant interactions at less abiotically taxing sites. By replicating an early-emerging species experiment at two sites that differ in productivity, we can investigate the interactions of abiotic conditions with assembly rules.

Assembly rule theory also has practical applications, especially in restoration projects. A cover or nurse crop (Shirley 1994; Perry and Galatowitsch 2003; Pywell et al. 2002) is a plant species that, as used in restorations, is hypothesized to improve recruitment and establishment of desirable species in one of two ways: directly, through the amelioration of the relatively harsh abiotic conditions (e.g., high light, high evaporation) associated with newly planted restorations; or indirectly, by suppressing competitive weeds (Figure 1), which is where assembly rule theory would be useful. Assembly rule theory could assist in choosing a cover crop whose traits would exclude weeds while simultaneously allowing establishment by desirable species. While Shirley (1994) provides anecdotal evidence for the efficacy of cover crops, experimental evidence has shown little to no support for their use (Pywell et al. 2002; Perry and Galatowitsch 2003). Even if cover crops do not prevent weed invasion, however, their utility may lie in other areas. If cover crops regulate invader identity according to assembly rules, they could be utilized to increase β diversity in prairie restorations. By creating a patchy mosaic of various cover crops, restorations could attain high β diversity similar to that found in unplowed prairie remnants (Martin et al. 2005, Wilsey et al. 2005).

With these factors in mind, we addressed the following questions.  First, how does productivity affect plant-plant interactions?  Evidence shows that species are more likely to compete at high community productivity levels (Bertness and Callaway 1994; Callaway et al.2002); competition-driven communities may therefore feel the effects of assembly rules more strongly.  Conversely, are the abiotic conditions so harsh at low productivity sites such that we find evidence for facilitation (positive plant interactions)?  Second, do early-emerging species differ significantly in important traits such as period of peak growth, aboveground biomass, and fire intensity when burned?  Third, do differences among early-emerging species result in divergent communities?  Additionally, if communities do diverge, are the colonists from a different or the same functional group as the early-emerging species?  Answers to these questions will help us determine if native early-emerging species could serve as an appropriate cover crop.

Materials and Methods

Study sites

Experimental plots were established in 2004 at each of two locations differing in  productivity.  The high productivity site (August average biomass 424 g/m²; see main plot experimental design) is located at the Iowa State University Horticulture Farm, north of Ames, Iowa, USA.  The low productivity site (August average biomass 187 g/m²) is on the Iowa State University Western Research Farm, near Castana, Iowa, USA.  The plots were located in abandoned pastures on west- to southwest-facing hillsides previously dominated by the non-native C₃ grass, smooth brome (Bromus inermis).  The Ames site had not been grazed since the establishment of the Horticulture Farm in the 1960's; the Castana site was grazed by cattle until 2001.

Experimental Design

Main Plots

Cover crop (early-emerging) species effects on invader composition

At each of the two sites, replicated 5 x 5 m plots were disked to remove the brome and  then seeded with one of five, functionally different, early-emerging species at the rate of 11 kg/ha or kept as non-vegetated controls (Table 1).

Each treatment had 6 replicate plots for a total of 36 plots at each site. Treatments were assigned to plots using a completely randomized design.  Corridors of 3 m between the plots were periodically mowed.  For our treatments, we chose species that are among the first to germinate and become established in tallgrass prairie restorations, and included some which have been suggested as possible cover crop species (Packard and Mutel 1997).  Dornbush and Wilsey (unpublished data) found in a prairie restoration experiment started from seed that Chamaecrista fasciculata, Desmanthus illinoensis, and Rudbeckia hirta had the most numerous seedlings in the first year. Bouteloua curtipendula was the earliest emerging C₄ grass and Elymus canadensis the earliest emerging C₃ grass in several previous experiments (Wilsey unpublished data, Dickson and Wilsey, unpublished data).  The C₃ grass E. canadensis has also been used previously as a cover crop (Packard and Mutel 1997, Martin et al. 2005).

Each treatment was applied by adding seeds on bare ground a week after disking occurred in April 2004. Plots were then culti-packed to prevent seed movement and provide good contact between the seeds and the soil.  Rhizobium species commonly associated with native legumes were added in equal amounts to each plot, in the form of live Rhizobia in peat moss mixed with local soil.  This was done to ensure that specific Rhizobium were available to the native legumes.  Non-planted species (invaders) were removed for two months to ensure the full establishment of treatment species, by spot-treating with herbicide (Roundup) prior to germination of treatment species, or by hand weeding or mowing after germination. Once treatment species achieved canopy heights of 15-30 cm in late June 2004, we discontinued invader management measures. Invaders were allowed to enter the plots freely thereafter.  In control plots, invaders were clipped to the soil surface twice during the first growing season (in June and October), and large, persistent biennial plants were spot-treated with herbicide (Roundup) once at the end of the first growing season at the Ames site. Invader management in control plots was also discontinued when the prairie seed mix was first added in November 2004.

Since late-emerging prairie species were our target for an ongoing test of assembly rules (Figure 1), we added a seed mix of 28 late-emerging species (Table 2) to each plot during the winters of 2005, 2006 and 2007 using an equal number of seeds per species.

One of our main objectives was to estimate how much each community diverges due to the invasion of non-planted (weed) species. Weeds can hinder native species establishment (Pywell et al. 2002) by quickly dominating open ground, and limiting light, water, or nutrients. By measuring invader composition in the initial year of community establishment, we had an early test of whether the treatment species can lead to divergent invader species compositions. In July 2004, we estimated relative cover of all species in two 30 x 30 cm quadrats in each plot in order to compare species compositions among treatments.

At the Ames site, we decided to investigate community development more thoroughly. In September 2004, we sampled aboveground biomass within two 50 x 50 cm quadrats randomly placed in each treatment plot. We clipped the biomass at approximately 1 cm above ground, and also collected litter. Control plots were not sampled, since they had been clipped for invader control during the growing season. We sorted the biomass by species using taxonomic guides (Eilers and Roosa 1994) and dichotomous keys (Barkley et al. 1986, Pohl 1954). Biomass was then oven-dried for 48 hours at 60 °C and weighed.

In 2005, we collected biomass in one 50 x 50 cm quadrat in each plot twice in the growing season, once each in June and August.  All biomass samples were sorted to species using taxonomic guides (Eilers and Roosa 1994) and dichotomous keys (Pohl 1954; Barkley et al. 1986), oven-dried at 60̊C for 48 hours, and weighed.

Tests of Fox’s assembly rule in invader communities using functional group sets

We chose two functional group sets to test Fox’s assembly rule. We investigated two functional group sets, in order to best ascertain which characteristic of an early-emerging species was the most important determinant of species composition.  One set defined a species according to its grass-forb-legume status, which employs the functional groups used in perennial grassland studies, based on mode of photosynthesis, grass-forb status, and the ability to fix nitrogen.  These groups were C₃ graminoids, C₄ graminoids, non-leguminous forbs, and leguminous forbs functional groups. The second functional group set was defined by the longevity of a species.  Annuals and biennials were combined into one group, and perennial species into another.  This functional group set is commonly used in annual and disturbed grasslands.  Early successional systems are typically dominated by annual/biennial species for a number of years (Bazzaz 1996).  Their dominance in early successional sites may be due to the wide seed dispersal ability of annual and biennial species (Fargione et al. 2003) or due to their superior competitive ability in the harsh conditions of early succession (Pacala and Rees 1998).

Transplant Sub-Plot Treatments: cover crop (early-emerging) species effects on late-emerging native seedlings

To study the direct effects of a cover crop (early-emerging) species on late-emergent native prairie species, we established weeded sub-plots within the main plots and planted seedlings of eight, functionally different species. While early-emerging species may directly compete with late-emerging species, they may also have an indirect positive effect on establishment through invader suppression (Figure 1). By removing the weed component from the community, we could study only the direct effects of the early-emerging species treatments.

We chose native prairie species to represent the grassland functional groups (C₃ graminoids, C₄ graminoids, leguminous forbs, and non-leguminous forbs) that had functionally similar counterparts among the early-emerging species treatments (Table 1). To avoid misrepresenting a species-specific response as being characteristic to its entire functional group, we used two representative species from each functional group. All transplant species are ones commonly used in prairie restorations.

Transplant species

1)   Dalea purpurea, a perennial N-fixing legume

2)   Lespedeza capitata, a perennial N-fixing legume

3)   Dicanthelium oligosanthes, a perennial C₃ grass

4)   Stipa spartea, a perennial C₃ grass

5)   Schizachyrium scoparium, a perennial C₄ grass

6)   Andropogon gerardii, a perennial C₄ grass

7)   Monarda fistulosa, a perennial C₃ forb

8)   Ratibida pinnata, a perennial C₃ forb

D. purpurea, L. capitata, S. scoparium, A. gerardii, and M. fistulosa were started from seed in a greenhouse in March 2004. Due to lack of seed availability, we purchased seedlings of R. pinnata and S. spartea from a supplier in May 2004, and trimmed them to a similar size as those species started from seed. Only one seedling of S. spartea was planted in each sub-plot due to restricted supplies. D. oligosanthes was unavailable in either seed or seedling form; plants of this species were dug from each study site in May 2004 and trimmed to a similar size as the other transplant species. All trimmed species were allowed to recover from trimming before being transplanted outdoors.

In June 2004, two 1 x 1 m subplots, at least 20 cm from the main plot edge, were cleared of all invaders (Figure 2). Two months after the the early-emerging species had been sown, two seedlings of each study species (except S. spartea) were added into each subplot in late June/early July 2004. Seedling transplants were placed in each subplot in random locations, with the restriction of having at least 10 cm between transplants. When the random location landed upon an early-emerging species plant, transplants were placed directly to the side of the plant. Subplots were weeded monthly throughout each growing season, so that plant-plant interactions would only occur between the early-emerging species and transplants.

At the end of each growing season (2004 and 2005), a random subplot was harvested from each plot before the first killing frost of the season. This was done at the Ames site in October 2004 and September 2005, and at the Castana site in November 2004 and October 2005. At the Ames site, roots of one transplant of each species were excavated to a depth of 15 cm with a 5 cm diameter coring tool, and the aboveground biomass of each transplant species and the early-emerging species were collected. At the Castana site, due to its longer distance from Ames, only aboveground biomass and crowns of the target and early-emerging species were harvested.

Statistical Analysis

Main Plots

For the 2004 relative cover data, we averaged the two sub-samples within each plot due to their lack of independence. This average was then ln(y+1) transformed in order to improve normality. A two-way ANOVA (site x treatment) was performed on the transformed relative cover of the early-emerging species. In order to interpret the site x treatment interaction, we performed a 'slice' procedure on the results.  This procedure independently tests the strength of the interactions among the multiple levels of a class (such as site) and a treatment without physically separating the data by class level, which would reduce the degrees of freedom for the treatment effects, and thus the power of the analysis (Littel et al. 2002).

To analyze the 2004 Ames site biomass data, we averaged the treatment species biomass of the two sub-samples of each plot. For the 2004 Ames data and 2005 data for both sites, we ln(y+1) transformed the biomass of the treatment species to improve normality. We then performed a one-way repeated measures ANOVA on the Ames site data (2004, June 2005, August 2005), and a two-way (site x treatment) repeated measures ANOVA on the 2005 biomass data for both sites. In the two-way analysis, we again sliced the data by site.

Invader species composition: Year 1

For the Ames site investigation, species biomass per quadrat was ln(y+1) transformed to improve normality. We then averaged the two sub-samples of each plot, and performed a one-way (treatment) ANOVA analysis for each guild set, using contrasts appropriate to the functional group set we tested.

Invader species composition: Year 2

To test if significant differences existed among the treatments, we utilized the multi-response permutation procedure (MRPP), a non-parametric test (McCune and Grace, 2002). Each site and collection time were analyzed individually, using the relative biomass of the harvested invader species.

When the MRPP indicated that differences in invader composition existed amongst the treatments, we performed a nonmetric multidimensional scaling (NMDS) (McCune and Grace, 2002) ordination, using the relative invader biomass of each species. This was done to determine which invader species were driving differences amongst treatments. Each site was ordinated separately, using both harvest times in each ordination.

Spring burning of main plots

Each plot was burned in April, 2006 to determine how cover crop species might affect fire intensity.  Fire intensity was quantified by measuring soil surface temperatures with paints that melt at known temperatures (Omega Corp.).  The proportion of each plot that had burned was quantified by noting the proportion of 16 locations (grid with placements one meter apart) that were either burnt (black charcoal present) or not (charcoal not present).

Transplant Sub-Plots

Aboveground biomass

Because we collected the aboveground biomass of the transplants at both sites, we could compare the effects of site upon the seedling biomass production.  Aboveground biomass of each species was ln(y+1)-transformed to improve normality. We then performed two-way (site x treatment) repeated measures ANOVA on the ln-transformed aboveground seedling biomass. Also using aboveground biomass, we analyzed species diversity (Simpson’s Diversity = 1/'p_i²) of the transplant communities using transplant biomass.  In 2004, two subplots from the Ames site were lost (an E. canadensis and a B. curtipendula subplot), while in 2005 a control plot at the Castana site was lost; both sub-plots from these plots were therefore dropped from all transplant analyses.

Tests for species composition differences

We used several measurements for differences in species composition among transplants. We first used a multi-response permutation procedure (MRPP) to test for differences in species composition among treatments using relative total biomass of the transplants. Total transplant species biomass per sub-plot was calculated using the following formula:

Total Biomass = (Aboveground biomass) + [(# surviving transplants) * (belowground biomass of transplant)]

Because the belowground biomass of seedlings was collected differently at each site, we analyzed the sites separately.

Second, we analyzed the functional group community composition and how it changed over time. We found the relative biomass of each transplant species, separated the species into grassland functional groups, and rank-transformed the functional groups to maintain normality. We then performed repeated measures ANOVA on the rank-transformed grassland functional group BIOMASS for each site separately, as root biomass was collected differently at each site.

Main Plots: Effect of cover-crop species on invaders

Early-emerging species establishment and biomass

We found differences between sites in early-emerging species relative cover (Figure 3). In the July 2004 relative cover sample, relative cover of treatment species was significantly higher at the Ames site (Figure 3, p < 0.05). There were also significant differences in the relative cover among the early-emerging species at the Ames site (Table 3). The two annual/biennial treatments, C. fasciculata and R. hirta, possessed higher relative covers than the three perennial treatments in Ames (Figure 3). C. fasciculata had an average relative cover of 66%, while R. hirta had an average of 72%. Average cover of the other three treatment species ranged from 23 to 46%.

In 2005, relative biomass of the early-emerging species significantly differed between sites and among treatments, with a significant site x treatment interaction (Table 3). R. hirta showed marked decreases in relative biomass from June to August at both sites. The relative biomass of this species peaked early in the growing season, with average relative biomasses of 43 and 53% at Ames and Castana, respectively. At the Castana site, the other treatment species showed increases in relative biomass over time. The treatment B. curtipendula showed the greatest increases in relative biomass, from 5% in June to 37% in August.

At the Ames site, B. curtipendula and C. fasciculata significantly (Table 4) decreased in relative biomass from September 2004 (55% and 77%, respectively) to August 2005 (28% and 3%, respectively), while the other treatments showed no significant changes over time.

Composition of invaders at the Ames site, 2004

Cover crop treatments varied in their amount of invader biomass accumulation, but trends were not always consistent with Fox's assembly rule. We found that C₃ graminoids were significantly higher within the C₃ treatment, E. canadensis (contrast p < 0.05). C₃ graminoid invaders were also significantly lower within the forb treatment, R. hirta (contrast p < 0.001).  E. canadensis also contained more C₄ graminoids than than other treatments (contrast p < 0.05).  No other contrasts were found to be significant.

Within the annual/biennial and perennial functional group sets, perennial treatments contained significantly higher amounts of perennial invaders (contrast p < 0.001) than annual/biennial treatments. Annual/biennial invasion was not significantly different among treatments. Additionally, perennial treatments contained more invader biomass than annual/biennial treatments (Figure 4) in year 1.

The biomass of the early-emerging species was an important predictor of invader biomass. Because biomass is correlated with light, water, and nutrient uptake, we performed an ANCOVA that included the biomass of the early-emerging species treatment. Biomass of invaders decreased as early-emerging species biomass increased (Figure 5). After we accounted for biomass of the early-emerging species, there was no other effect of early-emerging species.

Composition of invaders at Ames and Castana, 2005
 

The MRPP analysis of the Ames site 2005 relative invader biomass indicated significant differences in composition among treatments for both collection periods (June A = 0.08,  p < 0.02; August A = 0.10, p < 0.01). Treatments at the Castana site were not significantly different for either collection time (June A = -0.07, p < 0.94; August A = -0.03, p < 0.93).

The NMDS ordination of the Ames 2005 biomass data showed that differences in community compositions were strongly driven by B. inermis, an exotic perennial C₃ grass, and Coronilla varia, an exotic perennial legume (Figure 6). The ordination graph indicates that B. inermis and C. varia tend to be negatively correlated, such that treatments with high B. inermis biomass tended to have low C. varia biomass, and vice versa. E. canadensis treatments tended towards high B. inermis biomass and low C. varia biomass, whereas C. fasciculata treatments tended otherwise (Figure 6). Where neither species was abundant, the ordination indicated that treatment plots contained higher amounts of the native annual forb Artemisia artemisiifolia, the exotic annual C4 grass Setaria glauca, and the native annual forb Polygonum ramosissimum (Figure 6).  Control plots tended to have low relative biomasses of both B. inermis and C. varia.

At the Castana site, the biomasses of B. inermis and the exotic perennial legumes Trifolium pratense and T. repens were negatively correlated (Figure 7).  Plots containing lower relative biomasses of the above species either contained more of the exotic legumes Melilotus spp., or the native annual forb, Conyza canadensis. However, none of the treatments appeared to group strongly along the above species' axes. (Figure 7).

Fire Intensity

Fire temperatures were much higher in plots dominated by Bouteloua curtipendula (side-oats grama), and to a lesser extent Elymus canadensis (Canada wildrye), than in plots dominated by non-grass species (P < 0.05).  This effect was found at both study sites.  The proportion of the plot that burned followed the same trend as fire temperature, with a greater proportion of the plots dominated by Bouteloua curtipendula being burned compared to other species.  Burns in plots dominated by forb cover crops were patchy and cooler compared to burns in the grass dominated plots.

Transplant Sub-Plots

Aboveground biomass of target seedlings

Aboveground biomass of the transplants (Fig. 8) increased significantly (Table 5) from year 1 to year 2 as plants fully established. Transplant aboveground biomass differed significantly between sites, with the Ames site possessing higher aboveground transplant biomass (Figure 8). At the Ames site there was a significant effect of early-emerging species presence (p < 0.001 for both years), indicating that early-emerging species were competing with the late-emerging target species. Control and Leguminous forb treatments had significantly higher aboveground transplant biomasses, while the Forb (R. hirta) treatment had significantly lower aboveground transplant biomass in Ames (Table 5). There was no effect of early-emergings species presence at the Castana site, indicating that early-emerging species were neither competing with or facilitating late-emerging species at this site.

Transplant species diversity also showed a site x treatment interaction (Table 5). This treatment effect was significant at only the Ames site for both years (p < 0.001). Ames Control and D. illinoensis plots resulted in significantly higher transplant diversity (Figure 9); however, this effect did not persist in the D. illinoensis plots. Transplant diversity generally increased between years among the other treatments.

Transplant species composition

There were no differences in species composition among treatments (MRPP, p > 0.05) The MRPP analyses of the relative total transplant biomass by site and time showed no significant differences in species composition (Figure 10).

Functional group biomasses at the Ames site showed significant differences among treatments (Table 6).  Except for the Forb transplants however, these differences were driven solely by the Control treatments, where all functional groups had the highest biomass.  The Forb transplants had significantly higher biomass in the Leguminous Forb treatments (particularily D. illinoensis), and significantly lower biomass in the Forb treatment.  At the Castana site, we found no significant differences among treatments for any of the functional groups (Table 7).

Discussion

The results of the transplant experiment showed that competition took place between the cover crop (early-emerging species) and the target prairie plants, and that competition was strongest at the higher productivity site.  Plant-plant interactions at the low productivity site tended to be more neutral.  These results suggest that these species are not acting as nurse plants.

Cover crop (early-emerging) species competed strongly at the seedling stage of the transplants, with differences largely disappearing by the end of the second growing season. This suggests that once an establishing species survives the seedling establishment phase, early-emerging species might have few long-term effects on its survival and growth.  Although our experiment studied the invasion of native species into a native-dominated community, our data are similar to those of Yurkonis et al. (2005), who found that exotic species extirpated native species by limiting their seedling recruitment and not by out-competing adult native plants. In both cases, invasion appears to be limited by seedling recruitment, prevented by adult plants in the communities.     

While neutral theory (Hubbell 2005) predicts that plant species are equivalent in traits such as biomass or resource capture, our findings show that plant species are not equivalent. Using biomass as a surrogate for resource capture and timing of growth, we found significant differences among the early-emerging species in relation to peak biomass and timing of the peak growth period.  Perhaps most importantly, species varied in their fire temperature and proportion burned in spring 2006.   Hooper and Vitousek (1997) and Hooper (1998) demonstrated that California serpentine grassland species differed greatly in their germination, period of peak growth, and seed set.  Fargione and Tilman (2005) showed that a prairie grass, S. scoparium, differed from other grassland species in its rooting depth and phenology. Zavaleta and Hulvey (2004) and Losure et al. (2006) have shown that timing of growth matters greatly to invader suppression.  Differences that we found in fire intensities might have long term effects on  colonization in future years, and this will be monitored in the future of our study. 

The identity of the cover crop (early-emerging) species also affected invader colonization and composition, but this occurred only at the high productivity site.  This implies that highly productive communities are structured more by competition than by the abiotic environment, as described in the theoretical paper by Chase (2003).  Highly productive communities are predicted to be more strongly shaped by community assembly rules, suggesting that assembly rules should be found operating at the Ames site.

The transplant and biomass data show mixed support for Fox's assembly rule. Although at the Ames site the forb transplants did support this assembly rule to some extent, the other transplants did not conform to the rule.  Analysis of the 2004 Ames invader biomass data using the grassland functional group set revealed that both C₃ and C₄ graminoid invaders preferentially invaded the E. canadensis (C₃ grass) treatment. One reason that the graminoids were able to invade may be due a combination of their relative tolerance for shade and litter build-up in comparison to other functional groups.  This combination of traits would tend to favor invasion by graminoids.

Results from the 2004 invader analysis using the longevity functional groups showed no support for Fox's assembly rule.  Annual-biennial early-emerging species had greater suppression of perennial invaders than did perennial treatments.  Our data do support the niche theory of succession put forth by Pacala and Rees (1998), where annual and biennial species dominate early successional communities due to their specialization for the stressful abiotic conditions found with disturbance.  The high biomasses of the annual-biennial treatment species show that they were better competitors for resources than their perennial invaders.  Conversely, the lower biomasses typical of the perennial treatments indicate that more resources were available to invaders in these communities, allowing for more invasion by the less competitive (in comparison to annual-biennial species) perennial invaders in the initial year of colonization.

Use of early-emerging species as cover crops

A successful cover crop should either directly facilitate native species growth through the amelioration of the harsh abiotic conditions of bare ground or have the ability to suppress non-planted species invasion, thereby indirectly encouraging native species recruitment (Figure 1).  Our seedling transplant study found that early-emerging species do not facilitate the growth of established late-emerging prairie species when those species are added after the cover crop has fully established.  Competitive interactions were strong at the Ames (high productivity) site during the initial year of establishment.  However, once transplants were fully established in the plots (year 2), differences in transplant biomass between the controls and treatments were smaller.  These data suggest that at high productivity sites, the presence of an early-emerging species cover crop would not have long-term negative effects on adult later-emerging species once they became established.  At the Castana (low productivity) site, early- and late-emerging species experienced little to no competition or facilitation, suggesting that communities at that site are structured more strongly by the abiotic environment (Chase 2003).  Our second cover crop experiment, which varied timing of planting (spring vs. fall) and adds the seed mix either with the cover crop or in the growing season after establishment, found a large decrease in prairie species establishment when the mix was added the year after establishment (as was done here).  These results will be reported on in the 2008 annual report.

Unwanted species invasion can be a serious problem in restorations, as the invaders can sometimes out-compete the target species and dominate the restoration.  Some of the cover crop (early-emerging) species suppressed invasion more than others. The annual and biennial treatments, C. fasciculata and R. hirta, respectively, had lower total invader biomass than the perennial treatments. If a reduction in invader biomass is all that is desired, planting an annual or biennial cover crop is suggested. However, because the tested annual and biennial species did not persist over time (personal observations), the native perennial species would have to establish very early for the restoration to be successful.

As noted above, the most dramatic patterns of suppression were found using the longevity-based functional group set.  In addition to suppressing perennial invaders, the annual and biennial cover crops could impede establishment of by native perennial species, thus slowing down the restoration.  Perennial cover crops could have the opposite effect and allow native perennial establishment. Because annual-biennial invader biomass was not influenced by cover crop identity, it may be futile to use cover crops to prevent annual and biennial species invasion.

Our data suggest that perennial invaders, and not annual-biennial invaders, may pose the greatest threats to perennial grassland restorations. At the Ames site, the restoration communities tended to contain high relative biomass of either one of the exotic perennials, C. varia, a legume, or the C₃ grass B. inermis. At the Castana site, communities were strongly associated with B. inermis, and the exotic perennial legumes Medicago lupulina, Trifolium pratense, and T. repens. Although Melilotus spp. does not make up a significant proportion of the biomass, the other species attain average relative biomasses (including the treatment species) of 10-57 %. Such community dominance by exotic species could prevent native seedlings from invading the community, as demonstrated by Yurkonis et al. (2005).  The second cover crop experiment tends to support this notion, because there has been much greater establishment of native species when prairie mix seeds were added with the cover crop seed than when seeds were added the year after.

Although native cover crops do suppress invasion, and do compete to some extent with established later-emerging native species, whether they negatively affect native perennial seedling establishment remains an important unanswered question. This can be tested with longer-term data.

Conclusions

Our study of the invader species composition yielded several important results. First, cover crop-caused early divergences in community composition appear to depend on the productivity of the site.  However, plots at the high productivity site may eventually converge on one community composition as the early-emerging species decline in abundance; this will be tested with longer-term data.  Secondly, the main plot study revealed that while plant traits do regulate invader identity, such regulation does not explicitly follow Fox's assembly rule, and some traits appear to be stronger community regulators than others.  Fire intensity, longevity of a species and time of peak growth (Losure 2006, Polley et al. 2006, Fargione and Tilman 2005) may be the coarsest trait 'filters' of tallgrass community regulation, while other traits act as finer filters.

The overall implications of our experiments for tallgrass prairie restorations would be that high productivity sites may be easier and faster to restore than low productivity sites.  Adding prairie seed mix in the year following cover crop establishment does not appear to work well in establishing prairie species.  Also, perennial species are more likely to cause problems in the long run than annual-biennial species, particularly the perennial exotics B. inermis and C. varia.  Evidence from Losure (2006) suggests that C. varia is exceptionally difficult to eradicate once it has become established in a community.     

Acknowledgments

Thanks to the Iowa Department of Transportation Living Roadway Trust Fund, which provided the primary funds for this project.   The Nebraska Nature Conservancy, Prairie Biotic Research, Inc., and Iowa State University also provided funding towards the project.

Hooper, D.U. 1998. The role of complementarity and competition in ecosystem response to variation in plant diversity. Ecology 79: 704-719.

Hooper, D.U. and P.M. Vitousek. 1997. The effects of plant composition and diversity on ecosystem processes. Science 277: 1302-1305.

Hubbell, S.P. 2005. Neutral theory in community ecology and the hypothesis of functional equivalence. Ecology 19: 166-172.

Jones, C.G., J.H. Lawton, and M. Shachak. 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78: 1946-1957.

Lockwood, J.L., R.D. Powell, M.P. Nott, and S.L. Pimm. 1997. Assembling ecological communities in time and space. Oikos 80: 549-553.

Lockwood, J.L. and C.L. Samuels. 2004. Assembly models and the practice of restoration. In: Temperton, V.M., R.J. Hobbs, T. Nuttle, and S. Halle (eds). Assembly rules and restoration ecology. Island Press, 55-70.

Losure, D.A. 2006. Patterns and processes of plant community invasibility. Master's thesis, Iowa State University.

Martin, L.M., K.A. Moloney, and B.J. Wilsey. 2005. An assessment of grassland restoration success using species diversity components. Journal of Applied Ecology 2: 327-336.

McCune, B. and J.B. Grace. 2002. Analysis of Ecological Communities. MJM Press.

Nuñez, C.I., M.A. Aizen, and C. Ezcurra. 1999. Species association and nurse plant effects in patches of high-Andean vegetation. Journal of Vegetation Science 10: 357-364.

Pacala, S.W. and M. Rees. 1998. Models suggesting field experiments to test two hypotheses explaining successional diversity. American Naturalist 152: 729-737.

Packard, S. and C.F. Mutel. 1997. The tallgrass restoration handbook: For prairies, savannas, and woodlands. Island Press, Washington D.C.

Padillia, F.M. and F.I. Pugnaire. 2006. The role of nurse plants in the restoration of degraded environments. Frontiers in Ecology and the Environment 4: 196-202.

Paine, R. 1969. Food web complexity and species diversity. American Naturalist 100: 65-75.

Perry, L.G. and S.M. Galatowitsch. 2003. A test of two annual cover crops for controlling Phalaris arundinacea invasion in restored sedge meadow wetlands. Restoration Ecology 11: 297-307.

Pohl, R.W. 1954. How to know the grasses, 3rd ed. Wm. C. Brown Company Publishers.

Polley, H.W., B.J. Wilsey, J.D. Derner, H.B. Johnson, and J. SanaBria. 2006. Early-successional plants regulate grassland productivity and species composition: a removal experiment. Oikos 00: 1-9.

Pywell, R.F., J.M. Bullock, A. Hopkins, K.J. Walker, T.H. Sparks, M.J.W. Burke, S. Peel. 2002. Restoration of species-rich grassland on arable land: assessing the limiting processes using a multi-site experiment. Journal of Applied Ecology 39: 294-309.

Schroeder, A., L. Persson, and A.M. De Roos. 2005. Direct experimental evidence for alternative stable states: a review. Oikos 110: 3-19.  

Shirley, S. 1994. Restoring the tallgrass prairie. An illustrated manual for Iowa and the midwest. University of Iowa Press, Iowa City.

Suding, K.N., K.L. Gross, and G.R. Houseman. 2004. Alternative states and positive feedbacks in restoration ecology. Trends in Ecology and Evolution 19: 46-53.

Valiente-Banuet, A., and E. Ezcurra. 1991. Shade as a cause of the association between the cactus Neobuxbaumia tetetzo and the nurse plant Mimosa luisana in the Tehuacan Valley, Mexico. Journal of Ecology 79: 961-971.

Von Holle, B., and D. Simberloff. 2003. Testing Fox's assembly rule: does plant invasion depend on recipienct community structure? Oikos 105: 551-563.

Wilsey, B. and G. Stirling.  2007.  Species richness and evenness respond in a different manner to propagule density in developing prairie microcosm communities.  Plant Ecology in press.

Wilsey, B.J., L.M. Martin, and H.W. Polley. 2005. Predicting plant extinction based on species-area curves in prairie fragments with high beta richness. Conservation Biology 19: 1835-1841.

Wilson, J.B., and R.J. Whittaker. 1995. Assembly rules demonstrated in a saltmarsh community. Journal of Ecology 83: 801-807.

Yurkonis, K.A., S.J. Meiners, and B.E. Wachholder. 2005. Invasion impacts diversity through altered community dynamics. Journal of Ecology 93: 1053-1061.

Zavaleta, E., and K. Hulvey. 2004. Realistic species losses disproportionately reduce grassland resistance to biological invaders. Science 306: 1175-1177.