DOT Project Number:  90-00-LRTF-416

Fiscal Year:  2004

Award:  $9,000.00

Principal Investigator:  Kirk Moloney and David Losure, Department of Ecology, Evolution and Organismal Biology, Iowa State University, kmoloney@iastate.edu

Research Report:

IMPACT OF PLANT COMMUNITY COMPOSITION ON THE SPREAD OF THE INVASIVE PLANT SPECIES CROWN VETCH

Abstract

Plant invasions have been hypothesized to proceed at the local, i.e. individual patch or stand, scale according to one of several distinct spatial patterns. However, few studies have attempted to reconstruct the patterns of perennial herbaceous plant invasions at local scales due to difficulty in determining the age of individuals. We used herb chronology to determine the ages of roots within several crown vetch (Coronilla varia L.) patches in order to characterize the spatial age structure of these patches. Additionally, we examined both sexual and vegetative crown vetch reproduction, with regard to potential impacts on local spread and persistence, through seed bank sampling and greenhouse experiments. We found little distinct spatial age structuring in crown vetch patches, perhaps due to a lack of older roots caused by rapid ramet turnover within patches. We also found no support for the hypothesis, proposed by several land managers, that crown vetch builds up a large seed bank. However, we did find that even small fragments of crown vetch plants are capable of vegetative regeneration, which may be important in explaining this species' persistence in spite of control measures.

Introduction

As the spread of exotic organisms has generated increasing environmental concern (Vitousek et al. 1996), researchers have worked to document the patterns of past and present invasions (Pysek et al. 1998) and to generate models capable of predicting future ones (Hastings 1996). These patterns, and likely the processes causing them, have proven to be highly scale dependent (Collingham et al. 2000, Levin 1992).

It has been possible to reconstruct patterns of invasion at large geographic scales from herbarium records and other sources (Weber 1998). The rate at which many invaders spread can often be accurately modeled at regional scales using simple reaction-diffusion models that assume dispersal and movement are random (Skellam 1951, Andow et al. 1990, van den Bosch et al. 1992). However, when applied at more localized scales these models have yielded mixed results. Frappier et al. (2003) found that a diffusion model accurately described the spread of a stand of buckthorn (Rhamnus frangula L.) in New Hampshire, but Lonsdale (1993) found that simple diffusion models were inadequate when examining the spread of Mimosa pigra L. from a wetland in Australia. Both of these studies focused on invading woody plants in situations where the progress of the invasion could be reconstructed by aging stem cross sections or observing aerial photographs. Few studies have attempted to reconstruct the patterns of perennial herbaceous plant invasions at local, i.e., individual patch or stand, scales due to difficulty in determining the age of individuals (Dietz 2002).

If the spatial spread of invasives across a landscape can be predicted accurately by models that assume movement is random, then the models are useful without further complication (Andow et al. 1990). However, movement is clearly not random. Plant dispersal and recruitment are greatly effected by micro site variation (Kadmon and Shimda 1990) and neighbor effects (Barton 1993). Understanding the movement of invaders at a finer scale may help us understand the processes that govern invasions. Clonal plants may be particularly useful study organisms in examining localized patterns of invasion, as they have the ability to respond to local conditions by selectively placing ramets in favorable sites (Evans and Cain 1995, Van Kluenen and Fischer 2001).

Lovett Doust (1981) proposed two possible contrasting invasion patterns for clonal species. She described species that would advance as tightly packed fronts, a “phalanx pattern,” as one extreme, and species that would spread out into the surrounding vegetation to minimize intraspecific contact, a “guerilla pattern,” as another. Wilson and Lee (1989) expanded the concept of guerilla and phalanx invasion patterns beyond clonal species to include all plant invasions and added the term “infiltration invasion” to describe a pattern where both short distance dispersal (typical of a phalanx pattern) and long distance dispersal (typical of a more guerilla like pattern) occur simultaneously. These patterns could potentially occur at either the scale of individual populations or at much larger landscape scales. Identifying by which, if any, of these patterns an individual population has expanded could provide valuable insight into the processes that prevent or allow a species to spread locally (Dietz 2002). Additionally, if the pattern by which a population is expanding is close to the phalanx or infiltration pattern, it should be possible to estimate the rate of spread of the population from the spatial-age structure (Dietz 2002).

Study System

THE PLANT

Crown vetch (Coronilla varia L.)  is a herbaceous perennial legume native to the Mediterranean region. Its trailing stems form dense patches in which very few other species are found. Crown vetch spreads by seed and asexually, and is now widespread in the United States, being found in all lower 48 states except Louisiana, North Dakota and California according to the USDA’s plant distribution maps (plants.usda.gov). It has been widely used as a ground cover and for erosion control, and was heavily planted along roadsides between the 1950s and 1980s. However, it has fallen out of favor as its effectiveness in erosion control has come into question (USDA 2002). Crown vetch also competes with more desirable native vegetation (Walck et al. 1999, Symstad 2004) and can spread away from roadsides and into natural areas (Solecki 1997). It is difficult to eradicate once established, and prairie reconstruction near or on roadsides can be made difficult or impossible in areas with large crown vetch populations (Shirley 1994).

Boone

The Boone site is approximately a half-acre of un-maintained land adjacent to the Des Moines River and the city of Boone’s water purification plant. Vegetation at the site is a diverse mix of native and introduced species. While introduced forage grasses such as Bromus inermis Leyss. and Poa pratensis L. are prevalent, there are also many prairie grasses such as Andropogon gerardii Vitman, Sorghastrum nutans (L.) Nash, Schizachyrium scoparium (Michx.) Nash, and Panicum virgatum L. at the site. The forb community includes some native genera including Silphium L., Verbena L., Solidago L. and Helianthus L., as well as a wide variety of weeds and introduced species such as Cirsium arvense (L.) Scop., Rosa multiflora Thunb., Lotus corniculatus L., and Ambrosia L.. The site also contains extensive patches of crown vetch (Fig.1). It is unlikely that crown vetch was seeded directly into the site, but it was probably seeded along a county road that borders the site. These intentionally planted populations are the most likely original propagule source for the patches now invading this site.

Western Research Farm

Iowa State’s western research farm is located in the loess hills region of Iowa. The crown vetch patches used in seed bank sampling are in a pasture located adjacent to a roadside ditch containing crown vetch. Prior to 2002, the pasture was dominated by smooth brome (Bromus inermis Leyss.). In 2002 the area was treated with herbicide, plowed under, and planted with prairie species as part of a restoration study. Many weeds came into the pasture after it was planted, but crown vetch was not one of the early invaders and by the end of 2003 was still present only in trace amounts. However, in 2004 crown vetch became a major problem. By May 2005, dense patches of crown vetch had formed so that it comprised 46% of the vegetative cover in the restoration experiment. Smooth brome had also become quite prevalent, and these invaders forced the abandonment of the experiment (B. Wilsey unpublished data).

Methods

SPATIAL-AGE STRUCTURE

We characterized the spatial-age structure of three crown vetch patches at the Boone site. We hypothesized that roots of similar ages might be clustered together in relatively distinct regions of the patches, and that locating and delineating these regions would provide insight into the pattern of spread that had led to the current patch dimensions. As we had no reasonable way of forming a priori hypotheses about the relative sizes or locations of these regions within patches, we tried different sampling methods in each patch in order to get an idea of what type of sampling design would be most appropriate for future work.

The smallest patch sampled (patch 1) measured roughly 3 x 3.5 m. Seventy-three 0.25 m² quadrats were sampled within this patch. Quadrats were placed directly adjacent to one another, for complete coverage of about 2/3 of the total patch. All roots were dug up within each quadrat. The age of each root was determined by counting annual growth rings in cross sections of the root, using the methodology of Dietz and Ullman (1997).

The second crown vetch patch (patch 2) sampled was roughly twice as large as the first. A 6 m transect was run across the center of the patch. Five transects were run across the patch perpendicular to the 6 m main transect, crossing it at 1 m intervals. All transects were sampled every 0.5 m. The sampling procedure consisted of removing one shovel full of soil and extracting all crown vetch roots from it. The largest of these roots were then aged.

The third patch (patch 3) sampled was quite large. A 22 m long transect was run across the center of the patch, and a second 20 m long transect was run perpendicular to this transect so that it crossed at the midpoint in the center of the patch. The sampling procedure was the same as for the second patch, except that the sampling interval was 1 m instead of 0.5 m.

PATCH BOUNDARY MAPPING

We used a Trimble GeoXT GPS unit with sub-meter accuracy to map the boundaries of all crown vetch patches at the Boone site on August 31, 2004, and again on August 29, 2005. On each of these days we walked through the site attempting to find all crown vetch patches. At each patch, we slowly walked around the patch outline while recording our path with the GPS unit. These data were recorded as shape files and then imported into ArcView 9.0. In ArcView, we calculated the area of each shape file. These areas were summed in order to obtain the total area covered by crown vetch patches in each year. Our goal was to examine year-to-year variation in total patch size in order to see if crown vetch was actively taking over more of the site.

SEXUAL REPRODUCTION

Seed Depth and Emergence

In order to examine the ability of crown vetch seedlings to emerge from different soil depths, we conducted a greenhouse study using commercially purchased crown vetch seed from Nature’s Own (Manteno, Ill.). We seeded 30 pots with 25 crown vetch seeds per pot. These 30 pots were divided into three treatment groups of 10 pots. In the first treatment, the seeds were placed with tweezers approximately 1-2 cm below the potting soil surface. In the second treatment the pots were filled with soil to a point 4 cm below the lip of the pot, the seeds were added, and then the pots were filled with packed potting soil up to the lip. In the third treatment the seeds were 8 cm below the lip of the pot. Pots were placed in random order on a greenhouse table, kept watered, and monitored for seedling emergence for 8 weeks. Seedlings were removed as they emerged to avoid any reduction in emergence due to intraspecific competition. Seedling emergence was compared between treatments with SAS ANOVA and Tukey-adjusted comparisons of means.

Seed Bank

Two studies were conducted to examine the distribution of crown vetch seeds in the soil seed bank at the Boone site and at the Western Research Farm.

At the Boone site, on May 19, 2005, one 41 m long transect was placed through a large patch of crown vetch and out into an area with smaller scattered crown vetch patches. We sampled at 22 points along this transect. All sampled points were at least 0.5 m apart, and points were selected so that samples were taken within and on the borders of the large and smaller patches of crown vetch, as well as in areas with other vegetation. At each sampling point two 2.5 cm diameter soil cores were taken to a depth of 15 cm, for a total of 44 soil cores. These cores were separated into 4 layers, the surface layer from 0-2 cm deep, and subsequent layers from 2-6, 6-10 and 10-15 cm deep. These were then bagged and taken directly to an Iowa State greenhouse. Each of the 176 samples (44 cores x 4 layers/core) was spread out in a 4 inch pot on top of sterilized potting soil. All pots were kept watered and seed germination was recorded weekly. All seedlings were removed when they could be identified to the generic level. Seedlings that became large enough to inhibit the germination of other seeds in the same pot, but that could not yet be identified, were transplanted into separate pots and allowed to continue growing. These seedlings were kept until they could be positively identified, or at least until we could be certain that they were not crown vetch. We ran the experiment for 17 weeks.

The procedure followed at the Western Research Farm pasture was quite similar to the one used at the Boone site, except for the layout of the transects. On June 21, 2005, a 4 m transect was run through a patch of crown vetch. At the 1 and 3 m points along this transect, 3 m long side transects were placed at right angles to the main transect across the patch and out into the adjacent grassland. Each transect was sampled at 0.5 m intervals, with two soil cores being removed at each sampling point. Soil cores were taken to a depth of 10 cm, and divided into three layers 0-2, 2-6 and 6-10 cm deep. A total of 42 soil cores were taken. Each layer from each core was placed in its own pot for a total of 126 pots, and the same procedure described above was followed to determine the number and composition of the viable propagule supply in the soil.

VEGETATIVE REPRODUCTION

Because we found no evidence to support the hypothesis that crown vetch builds up a large seed bank (see results), and because mowing has been shown to be ineffective in crown vetch control (Symstad 2002), we examined the ability of crown vetch to regenerate from vegetative fragments. We clipped plants at ground level and brought them to the lab. Plant sections in the following categories were cut from the plants: leaflet only, 5-10 cm section of leaf, 2-10 cm long section of stem without a node, and 2-10 cm long section of stem with a node. These sections were then placed in pots on the surface of packed potting soil. Three sections, all of the same type, were placed in each pot. Ten pots were given each section type, for a total of 40 pots. These pots were divided into two sets of twenty pots (5 pots per section type). One set was watered 2 to 3 times per week until the soil was well saturated, the other set was only watered once per week. All pots were on the same table in the greenhouse, but the low and high water treatments were slightly separated in order to avoid incidental watering of the low water treatment.  All pots were monitored for one month.

Results

SPATIAL-AGE STRUCTURE

We aged a total of 287 roots from the three crown vetch patches. The majority of these (66%) were only 2 or 3 years old. The root ages from each patch, and the spatial distribution of the oldest roots are shown in figures 2, 3 and 4. There was a weak tendency for roots of similar age to occur close to one another (Figure 5). However, all significant spatial structuring quickly disappears at larger distances (Figure 5). This makes it impossible to discern any clear pattern by which these crown vetch patches may have spread.

PATCH BOUNDARY MAPPING

The total area of all crown vetch patches at the Boone site was 3060 m² in 2004. One year later, this area had increased to 3630 m² (Figure 1). This increase was due to the appearance of several new, small patches as well as the expansion of existing patches. However, not all of the patches expanded between 2004 and 2005. Some of them contracted, and others shifted location slightly by shrinking on one side while expanding on another.

SEXUAL REPRODUCTION

Seed Depth and Emergence

There were significant differences in seedling emergence among the planting depth treatments (F_2,27 = 8.32 p = 0.002). Seedling emergence from a planting depth of 4 cm was significantly higher than from both the surface layer and 8 cm (Tukey adjusted comparisons alpha <0.05) (Figure 6). Seedling emergence in all treatments was rapid, with 88% of seedlings emerging within the first 14 days after planting.

Seed Bank

The viable seed bank at the Boone site contained an estimated 13,150 seeds/m², while the western research farm site seed bank had an estimated 9,200 seeds/m². Both of these viable seed densities are within the range reported for Great Plains grasslands by Lippert and Hopkins (1950). No crown vetch seedlings emerged from any of the soil samples at either site. Other data on the composition of the seed bank at each site is reported in the appendix.

VEGETATIVE REPRODUCTION    

In the high water treatment, all 15 of the stem sections that contained a node survived and produced new growth by sending up shoots from the node. Some of this new growth was visible as early as three days after the experiment began. None of the stem sections without nodes or the leaf sections produced any new growth, but they were still green at the end of the month.

In the low water treatment, several of the stem sections that contained nodes produced new growth, but none of them survived past three weeks. All of the plant sections in the low water treatment were clearly dead by the end of the month.

Discussion

Crown vetch is quite difficult to eradicate once established. One explanation frequently given for the resilience of crown vetch is that it is thought to build up a large and persistent seed bank (The Nature Conservancy 2003). This belief is likely based on the fact that the plants flower and produce seed for nearly the entire growing season. It is usually possible to find both newly opened flowers and mature fruits within a few feet of each other within a crown vetch patch. However, in a study examining interactions between patches of crown vetch and tall fescue along a roadside embankment, Luken (1987) noted that successful recruitment from seed was rare in established crown vetch patches. To our knowledge, our seed-bank sampling represents the first attempt to actually collect data on crown vetch in the seed bank. We were unable to find any viable crown vetch seed in soil samples taken in and around two established crown vetch patches. This result may surprise some land managers, but it is perhaps not unusual given the low rates of sexual reproduction by many clonal species (Harper 1977, Erikkson 1989). We did find extensive systems of underground rhizomes in the patches we sampled, indicating a great deal of vegetative reproduction was taking place. Through our greenhouse study, we also documented crown vetch’s ability to propagate from fragments of the above-ground stems. Although it is not clear to what extent this actually takes place in the field, it could be the mechanism behind other experimental results showing that mowing is ineffective as a crown vetch control method (Symstad 2002).

While many crown vetch invasions are probably initiated by seed, as it has been reported to invade sites not directly adjacent to plantings (The Nature Conservancy 2003), asexual reproduction appears to be much more important than sexual reproduction in explaining the persistence and resilience of established patches. Any stem section that contains a node is capable of generating a new plant, provided adequate moisture is present. Keeping crown vetch patches mowed may help prevent the production of seed that could disperse to start new invasions but is unlikely to help eradicate existing patches. Additionally, because very small stem fragments are capable of regeneration, it is highly unlikely that any contact herbicide could kill enough of the stems to prevent a patch from coming back, and even herbicides that translocate throughout the plant may not reach all the plants in a given patch. Repeated control measures in combination with close monitoring will be needed to remove crown vetch from areas where it is undesirable.

There are several patterns by which clonal species have been hypothesized to invade (Lovett Doust 1981). If crown vetch patches spread in distinct patterns, understanding these patterns would be very useful in developing methods for controlling this species. However, we found no such patterns in the spatial age structure of crown vetch patches. We found some weak spatial autocorrelation at very short distances, meaning that roots of similar age were more likely than would be expected by chance to be found close to one another, but this autocorrelation disappeared very quickly as distance increased. Dietz (2002) found strong spatial-age structure in a patch of spotted knapweed (Centaurea maculosa Lam.) when he first sampled in 1999. However, when the patch was re-sampled in 2003 it had not expanded beyond the 1999 boundaries and this structure had disappeared (Dietz 2004). The crown vetch patch boundary mapping we did in 2004 and repeated in 2005 showed that the total area of all crown vetch patches at the Boone site had increased. There did not appear to be a consistent pattern of increase, a few new patches had appeared, and some old patches had expanded while a few had shrunk slightly or shifted. Some of this increase may have been due to sampling error, as it is possible that we did not find every small patch present at the site in 2004. However, it seems unlikely that sampling error could explain all of the increase, especially since some of the larger patches have clearly expanded quite a bit. If the older crown vetch patches have reached some sort of dynamic equilibrium with their surroundings and only expand and contract in response to annual variation in local conditions, then natural population turnover as older individuals within the patch die and are replaced could destroy any spatial-age structure that may have been present as the patches were initially expanding. This would imply that models of invasive spread based on simple patterns or random movement and the rate of population increase may give accurate predictions initially (Andow et al. 1990), but that these predictions may become less and less accurate as the invasion progresses and interactions between new invaders, established members of the invasive population and the surrounding habitat increase. However, the lack of distinct spatial patterns in root ages in these crown vetch patches may simply result from the fact that the majority of roots sampled (66%) were only 2-3 years old. Ramet turnover within crown vetch populations my simply be too fast for much spatial-age structuring to develop.

Acknowledgements

We are grateful to the Iowa Department of Transportation for providing funding. We also thank the following people for all of their assistance: Kevin Day, Heidi Marttila-Losure, Drew Rayburn, Renae Schmitt, Wayne Roush and the staff of the Western Research Farm.

References

Andow, D., Kareiva, P. Levin, S. and A. Okubo. 1990. Spread of invading organisms. Landscape Ecology 4:177-188

Barton, A.M. 1993. Factors controlling plant distributions: drought competition and fire in montane pines in Arizona. Ecological Monographs 63:367-397.

Collingham, Y. Wadsworth, R., Huntley, B. and P. Hulme. 2000. Predicting the spatial distribution of  non-indigenous riparian weeds: issues of spatial scale and extent. Journal of Applied Ecology 37(Suppl 1):13-27.

Dietz, H. and I. Ullmann. 1997. Age-determination of dicotyledonous herbaceous perennials by means of annual rings: exception or rule? Annals of Botany 80:377-379

Dietz, H. 2002. Plant invasion patches – reconstructing pattern and process by means of herb-chronology. Biological Invasions 4:211-222.

Dietz, H. 2004. Plant invasion patches revisited-changes in development. Weed Technology 18:1381-1385.

Erikkson, O. 1989. Seedling dynamics and life histories in clonal plants. Oikos 55:231-238.

Evans, J.P. and M.L. Cain. 1995. A spatially explicit test of foraging behavior in a clonal plant. Ecology 76:1147-1155.

Frappier, B. Lee, T. Olson, K. and R. Eckert. 2003. Small-scale invasion pattern, spread rate and lag-phase behavior of Rhamnus frangula L. Forest Ecology and Management 186:1-6

Harper, J. L. 1978 The demography of plants with clonal growth. In A. H. J. Freysen and J. W. Woldendorp [eds.], Structure and functioning of plant populations, 27–48. North-Holland, Amsterdam, The Netherlands.

Hastings, A. 1996. Models of spatial spread: a synthesis. Biological Conservation 78:143-148.

Kadmon, R. and A. Shimda. 1990. Spatiotemporal demographic processes in plant populations: an approach and a case study. American Naturalist 135:382-397.

Levin, S. 1992. The problem of pattern and scale in ecology. Ecology 73:1943-1967.

Lippert, R.D. and H.H. Hopkins. 1950. Study of viable seeds in various habitats in mixed grass prairies. Trans. Kansas Acad. Science. 53: 355-364.

Lonsdale, W.M. 1993. Rates of spread of an invading species- Mimosa pigra in Northern Australia. Journal of Ecology 81:513-522.

Lovett Doust, L. 1981. Population dynamics and local specialization in a clonal perennial (Ranunculus repens): I the dynamics of ramets in contrasting habitats. Journal of Ecology 69:743-75.

Luken, J.O. 1987. Potential patch dynamics on a roadside embankment: interactions between crownvetch and kentucky-31 tall fescue. Reclamation and Revegetation Research 6:177-186.

Pysek, P. Prach, K. and Mandak, B. 1998. Invasions of alien plants into habitats of central European landscape: an historical pattern. In: Edwards, K. Kowarik, I, Starfinger, U. and M. Williamson (eds.) Plant Invasions-Ecological Mechanisms and Human Responses, pp 23-32. Backhuys Publishers, Leiden, The Netherlands.

Shirley, S. 1994. Restoring the Tallgrass Prairie. University of Iowa Press. Iowa City, IA

Skellam, J.G. 1951. Random dispersal in theoretical populations. Biometry 38: 196-218.

Solecki, M.K. 1997. Controlling Invasive Plants. In: S. Packard and C.F. Mutel (eds.)The Tallgrass Restoration Handbook: For Prairies Savannas and Woodlands. Island Press. Washington D.C.

Symstad, A.J. 2002. Mowing does not increase effectiveness of herbicide on crown vetch in tallgrass prairie (Illinois). Ecological Restoration 20:287-288

Symstad, A.J. 2004 Secondary Invasion Following the Reduction of Coronilla varia in Sand Prairie. American Midland Naturalist 152:183-189.

THE NATURE CONSERVANCY. 2003. Element stewardship abstract for Coronilla varia L. TNC Wildland Invasive Species Team. http://tncweeds.ucdavis.edu/esadocs/documnts/corovar.html

USDA NRCS. 2002. Plant fact sheet for Crown vetch. USDA NRCS Plant Materials program. http://plants.nrcs.usda.gov/factsheet/pdf/fs_cova2.pdf

van den Bosch, F. Hengeveld, R. and J. Metz. 1992. Analysing the velocity of animal range expansion. Journal of Biogeography 19:135-150.

Van Kluenen, M and M. Fischer. 2001. Adaptive evolution of plastic foraging responses in a clonal plant. Ecology 82:3309-3319.

Vitousek, P.M., D’Antonio, C.M., Loope, L.L. & Westbrooks, R. 1996. Biological invasions as global environmental change. American Scientist 84: 468-487.

Walck, J.L., Baskin, J.M. and C.C. Baskin. 1999. Effects of competition from introduced plants on establishment, survival, growth and reproduction of the rare plant Solidago shortii (Asteraceae). Biological Conservation 88:213-219.

Weber, E. 1998. The dynamics of plant invasions: a case study of three exotic goldenrod species (Solidago L.) in Europe. Journal of Biogeography 25:147-154

Wilson, J.B. and W.G. Lee. 1989. Infiltration invasion. Functional Ecology 3: 379-380.