DOT Project Number:  90-00-LRTF-831

Fiscal Year:  1998

Award:  $27,500.00

Principal Investigator:  Dr. Thomas Rosburg, Department of Biology, Drake University, thomas.rosburg@drake.edu

Summary Report:

THE STORY COUNTY INTERSTATE 35 PRAIRIE RECONSTRUCTION: MONITORING COMMUNITY ESTABLISHMENT AND EVALUATING FIRE AND MOW TREATMENTS, 1996 TO 1998

The first data analyses comparing pretreatment density with post-treatment density to rigorously assess the effects of phase 1 treatments (spring and fall burning, summer mowing) were presented in this report. These data describe the effect of one application of a treatment on species abundance. The report also presents preliminary information on the effect of an early spring fire on the survival, growth and reproduction of certain prairie forbs. Finally an evaluation of the quality of the prairie reconstruction through the first three years is presented.

Fire and Mow Treatments
The first fall burn was completed on November 22, 1997 and the spring burn was completed on May 11, 1998. Personnel from Story County IRVM Program and the Iowa Department of Transportation provided assistance and equipment. Fire lanes were constructed and metal fire barriers were used to prevent fire encroachment into adjacent plots. All of the burn plots were successfully burned, although the amount of litter present varied among the plots and affected the intensity of the fire. A propane torch was used to achieve a homogenous burn on each plot. It was used primarily during the spring burn on a few plots with low levels of fuel (more cool-season grass biomass than warm-season grass biomass). The torch was used to simulate a burn in areas that didn’t burn freely by singeing the plant material present.

Summer mow treatments in 1998 were applied with a rotary mower between June 10 and 25 by personnel from the maintenance division of the Iowa Department of Transportation. The mowing height was about 10 cm.

Effect of Treatments on the Density of Species
In order to integrate both the pretreatment and post-treatment densities in the assessment of treatment effects, the annual per capita population growth rate (r) was determined for the period between pretreatment density (summer 1997) to post treatment density (summer 1998).
The per capita growth rate is the number of stems, tillers, or caudices added to or lost from the population per individual stem, tiller, or caudex in the population. In this sense, population growth in this study is ramet based rather than genet based. Increases in density, for example, can occur by either sexual reproduction (seed) or asexual reproduction (vegetative growth). The amount of absolute change in abundance of a species due to either vegetative growth or seed production is a function of the current ramet population rather than the number of genets. Many prairie plant species increase in abundance through vegetative growth, which can occur continuously through the growing season as a function of higher physiological vigor and increasing age. These characteristics make per capita growth rate (r), which is a continuous model, especially appropriate for assessing change in abundance in this study. A value of 1 was added to both the 1997 and 1998 densities to allow calculation when density was 0. Positive values of (r) indicate an increasing population; negative values indicate a decreasing population. This analysis is looking for differences in the mean (r) among treatments.

The values of (r) for a species were analyzed with analysis of variance (ANOVA) using the following model. Potential sources of variation in (r) were: 1) treatments (which represents variation due to treatment effects), 2) seed mix (which represents effects caused by different environmental conditions on dry-mesic vs. mesic sites), and 3) site within seed mix (which represents variation due to differences in site characteristics such as biotic or historical factors). Multiple means comparisons were made with the student’s t-test.

Since a legitimate response from a species cannot be expected unless it has a reasonable presence on a given study site, sites were selected for inclusion only if there were three or more non-zero observations of the species among the eight possible observations at a study site (at least 37% of the 8 observations derived from 4 plots measured for 2 years are nonzero). About 30 of the most common species among the following six groups were selected for analyses: 1) prairie grasses, 2) exotic cool-season grasses, 3) prairie forbs, 4) native ruderal forbs, 5) exotic forbs and annual grasses, and 6) seedlings. Inclusion of a study site represents one replication for a species.

The results of the first summer mowing in 1997 were presented in report 2, therefore this summary will focus on the effects observed from one fall burn in November 1997, one spring burn in May 1998, and one summer mow with a rotary mower between June 10 and 25 in 1998. The disc mower was not available in 1998, thus the comparison between disc and rotary mowing was eliminated from the research and all mow plots were mown with a rotary mower.

Treatment effects were not especially common in this first year of phase 1. Among 28 taxa analyzed, only seven demonstrated an effect on growth rate due to treatments, and three of these were at the marginal level of p<0.10. Differences in growth rates attributable to seed mix (i.e., mesic vs. dry-mesic environments) were equally uncommon, with only five significant differences. On the other hand, differences due to variation in site within seed mix occurred in over 50% of the species analyzed, indicating that local site conditions and history are very important factors affecting growth rates and likely mediating the affect of treatments (i.e., management tools).

Among the native grasses, Canada wildrye exhibited marginal differences among the treatments; the highest growth rates were observed on control and fall fire plots. Both spring fire and summer mowing resulted in a decrease in population growth. Little bluestem experienced significantly higher growth on the spring fire and summer mow plots than on the control plots. These two species exemplify the differences in responses between native cool-season grasses such as Canada wildrye and native warm-season grasses such as little bluestem. A single management approach is not favorable to both species. Cool-season species are likely to be susceptible to spring fire while warm season species are likely go be stimulated by spring fire. A similar difference was observed in the non-native cool-season grasses. Smooth brome was very effectively reduced by spring fire, while fall fire was the only treatment to significantly reduce the growth of tall fescue. For both of these species, which are troublesome invaders or residents of prairie, no action and summer mowing allowed significantly more growth than did burning. Grey-headed coneflower was the only seeded prairie forb to exhibit a significant effect of the treatments, and it demonstrated a preference for fall fire over spring fire and summer mowing. Spring fire was the only treatment that limited growth of birds-foot trefoil, an aggressive non-native legume. Fall fire and summer mowing did not provide control and actually resulted in significantly higher growth rates for birds-foot trefoil than did spring burning.

Foxtail and forb seedlings responded very favorably to the fire treatments, both spring and fall burn plots exhibited significantly greater growth than did the summer mow. This suggests that summer mowing is a good control technique for foxtail if densities are high. Mowing did not have an effect on foxtail and forb seedlings relative to the control. These results demonstrate the positive effect that fire generally has on seed germination. In some cases this may be positive for the reconstruction if forb seedlings are seeded desirable species, or it may be a negative outcome if the seedlings are ruderal natives or exotic weeds.

Among the five species demonstrating an effect of seed mix (dry-mesic or mesic) on population growth, three are seeded prairie species and two are undesirable residents. The overriding pattern in these results is the greater growth rates on dry-mesic sites compared to mesic sites for four of the five species. Wood sorrel is the only species in which the mesic sites had higher growth rates than dry-mesic sites, and this result is probably an artifact of the much higher beginning densities of wood sorrel on the dry-mesic sites in 1997. Wood sorrel decreased to virtually nothing on all plots in 1998, but the decrease was larger on the dry-mesic plots because they had higher densities to begin with. The pattern suggests that for these species, four or which might be described as early successional, the increased openness and lower competition for light on a dry site might be more important for establishment than the favorable moisture at a mesic site. Since the productivity on dry-mesic sites was about 61% of the productivity on mesic sites (report 2), competitive stresses may have been considerably less on the dry-mesic sites.

Effect of Spring Fire on Prairie Forbs
Prior to the burn on May 11, 1998, individual genets of prairie species were flagged on the spring burn plot and two measurements were made on each genet: 1) the number of ramets, and 2) the height of each ramet. Individual genets of the same species were located on the control plots and likewise flagged and measured. Genets were selected that were relatively isolated to facilitate separation of the genet from other individuals. Ramets were recognized as either individual stems or individual stem caudices for basal-leaved species. Height was measured from the ground to the node associated with the highest leaf, or when flowers were present, to the base of the flower. The stem caudices of basal-leaved species were assigned a minimum height of 1 cm.

After the burn, the flagged plants on the spring burn plots were monitored weekly for two months (May and June), and then biweekly during the month of July. The plants on the control plots were monitored biweekly from May through July. Monitoring involved making measurements of the number of ramets per genet, the height of individual ramets, and recording the presence of either flowers or evidence of vertebrate herbivory on individual ramets. Altogether there were 120 individuals monitored on the spring burn plots representing 20 different species. There were 75 individuals monitored on the control plots representing 17 of the 20 species marked on the fire plots.

Ramet frequency and height were statistically compared between plants on spring burn plots and control plots for 13 of the 21 species monitored. The other eight species could not be compared because of a lack of replication on either the burn or control plots. Data that were normally distributed were compared with two-sample T-tests. If data were not normally distributed the Wilcoxon Rank test was used. The populations of marked plants were used to determine survival of the genets at two times – at one month and at one year after the date of the spring burn. Reproductive effort was assessed by calculating both the percentage of genets in flower and the percentage of ramets in flower. The requirement for a reproductive genet was at least one flower on the genet. Likewise, a reproductive ramet needed to have at least one flower to be reproductive. Incidence of herbivory was assessed at the ramet level. Herbivory was restricted to vertebrate consumption and was recognized when parts of stems or leaves were missing. These variables were compared between the plants on burn and control plots using the 17 species that were in common as replicates. In most cases the statistical comparisons were made with the Wilcoxon Signed Rank test due to strong non-normality in the observations.

Median survival for 17 species of marked prairie forbs following the 1998 spring fire was 100%, which was not significantly different from the 100% survival of the species on the control plots (Wilcoxon Signed Rank test, p=0.13, Table 8). Among the 20 species monitored on the burn plots, 17 had 100% survival one month after the burn. Those species that seem most susceptible to late spring fire were common wormwood and sweet-scented coneflower. Since common wormwood is a non-native, accidental species in the reconstruction, its mortality is a beneficial result. Among those genets that survived the spring fire, their survival during the following winter was virtually equal for plants on the burn and control plots (for May 9, 1999, Wilcoxon Signed Rank test, p=0.63). There was no evidence that a single late spring burn weakened the plants and put them in a stressful situation for winter survival.

Reproductive effort was lower for plants on the spring burn plots compared to plants on the control plots. The median percentage of genets in flower for burn plants and the median percentage of ramets in flower for burn plants were both 0% and significantly lower than for plants of the same species on control plots (Wilcoxon Signed Rank test, p=0.004).

Herbivory was significantly higher on plants on the burn plots than the control plots. Over 50% of the species on the burn plots exhibited evidence of herbivory, while less than 25% of the same species were grazed on control plots. The mean proportion of ramets exhibiting herbivory on the burn plots (11.2%) was significantly greater than the same species on the control plots (1.8%) (paired t-test, p=0.037). The species hit the hardest were the asters, including azure aster, smooth aster, and New England aster. Ox-eye and purple prairie clover were also selected by the herbivores. Apparently the fire made these species more appealing.

A comparison of growth (ramet frequency and height) between plants on burn and control plots was performed at four times during the 1998 growing season for 13 species. The most consistent pattern was a decrease in ramet height in the plants on the spring burn plots relative to the control plots. All 13 species exhibited significantly lower ramet height on at least one date, and more often on three or four dates, which was attributable to the late spring burn. The average height of ramets of plants stressed by the spring burn was typically only 50% (sometimes 25%) of the height of ramets on unburned controls. For strongly basal-leaved species like grey-headed coneflower and common wormwood ramet height was reduced even more. For the most part, the average ramet height of all 13 species of burned plants was not able to recover enough growth by the end of July to make them an equivalent height to the unburned plants.

In contrast to ramet height, ramet frequency was greater on spring burn plots than on control plots in most cases. This was not always a significant difference, but for three species – ox-eye, frost aster, and Canada milkvetch – the late spring burn did significantly increase ramet or stem density. For these species, the plants on the spring fire plots had nearly twice as many ramets as plants on control plots. The late spring burn caused a significant decrease in ramet frequency, at least for a short time, in three species. The ramet frequency of wild bergamot, common wormwood, and azure aster was significantly lower on burn plots than on control plots within the first two weeks after the burn. The ramet frequency of wormwood was decreased by about 50% and was maintained through the end of July. The decrease in ramet frequency was more temporary in azure aster and wild bergamot, they recovered to levels similar to the control plants by one month and two weeks, respectively, after the burn.

Monitoring and Evaluation of Prairie Quality
Over the first three growing seasons of the I-35 reconstruction the relative abundance of seeded species (RASS) ranged from 25 to 73% and from 11 to 68% for the mesic seed mix and the dry-mesic seed mix respectively. The percentage of seeded species present (PSSP) ranged from 10 to 44% for the mesic seed mix and from 15 to 49% for the dry-mesic seed mix. Thus there was considerable site-specific variation in the success and progress of the reconstruction. One of the sites with the lowest success (low RASS and PSSP) is 739N. This site experienced excessive flooding shortly after seeding in 1996, which likely caused relatively low germination of seeded species and adversely affecting quality.
Another site with very low success was 598N. This site lies on the top very sandy ridge of soil, and although the slope is very gradual it probably experiences fairly dry soil conditions. Because of its gentle slope, it was seeded with the mesic seed mix. Therefore success may be low on this site because the species in the seed mix are less suited for the dry environment. A third site with low success was 691S. This site has a
unique history relative to the other sites. It was accidentally burned by a very early spring burn the year of seeding. During the first year there was abundant growth of ruderal, weedy species that may have been a consequence of the disturbance that resulted from the early burn combined with the spring herbicide treatment.
The small consecutive increases in RASS each of the second and third years have significantly improved the quality of the reconstruction, as the mean RASS in 1998 is significantly greater than the mean RASS in 1996 for both seed mixes. This provides quantitative evidence for the axiom that time is a requirement for a successful reconstruction. However, RASS is still below 50% for both seed mixes, suggesting that there is still room for improvement.
On the mesic sites, PSSP has increased significantly from 25.2% (1996) to 33.8% (1998), most of which occurred during the second year (1997 growing season). Likewise, on the dry-mesic sites PSSP increased significantly during the second year and slightly the third year to its current level of 35.7%. On the mesic sites, the increase in PSSP since seeding in 1996 is equivalent to a little over 5 additional species. On the dry-mesic sites, the increase in PSSP represents about 8 new species. Clearly the germination and establishment of all seeded prairie species does not happen during the first year. The Cumulative Site PSSP (CS-PSSP) counts species presence over all the plots on the 15 sites rather than for just one plot at one site. CS-PSSP is nearly twice as high as mean PSSP. Due to the spatial dispersion of individual plants, a 3x5 m sample plot can only include a few of the many species that occur on the project site. On the dry-mesic sites, CS-PSSP has increased over 20 percentage points since 1996 and now is approximately 68%. On the mesic plots the increase from 1996 has only been about 10 percentage points, due in part to a slight decrease between 1997 and 1998. PSSP could be affected by site productivity and competition, thus the dry-mesic plots could have higher PSSP due to lower competition and shading than on the mesic sites.

Patterns of establishment for seeded prairie species over the first three years are similar to those presented in report 2 (for 1996 and 1997), and need not be discussed here. Specific information is available in the original report.

Combining both seed mixes, there were 76 species that were seeded on the study sites in the project area. Eighteen of these species, or 24%, have not been observed to have established on the plots or elsewhere on the project area through germination of seeds. Two of these species – Pedicularis lanceolata and Potentilla arguta – were observed in the study plots in 1996 by Troy Siefert but have not been observed since then by myself in the same plots or elsewhere. It is possible they were misidentifications. Two more species, Virginia windflower (Anenome virginiana) and narrowleaf mountain mint (Pycnanthemum tenuifolium), are most likely accidental replacements for the intended species Anenome cylindrica and Pycnanthemum virginianum. Both A. virginiana and P. tenuifolium have been observed on several study plots, while A. cylindrica and P. virginianum have not been observed. In addition to the two forb species identified in report 2 that are not native to central Iowa – common wormwood and false aster – five additional species included in the seed mix are not known to be indigenous to central Iowa and should not have been included in the project. White goldenrod (Solidago ptarmicoides) is native to northwestern Iowa gravel hill prairies. It was in the mesic seed mix and has been observed in study plots at three sites. Because of its habitat preference, its success would have been higher if it was included the dry-mesic mix. Prairie milkweed (Asclepias hirtella) is principally a species of the southern half of Iowa, northern blazingstar (Liatris ligulistylis) is a species that ranges across north central and northeast Iowa, dotted blazingstar (Liatris puntata) is a western species with an Iowa distribution primarily on the Loess Hills and gravel prairies, and prairie ragwort (Senecio paupercaulus) is more typical of wet mesic prairies in eastern Iowa. The first two species were included in both seed mixes, while the second two were in the dry-mesic mix. None of these have been observed on the project area through 1998.
Over the first three years, from 1996 to 1998, there have been four non-resident, prairie species observed on the study sites or project area that were not listed for either the mesic or dry-mesic seed mixes, nor any of the other numerous mixes used in the project area. All four are apparently accidental species in the seed mix. Two species, blue-eyed grass (Sisyrinchium campestre) and winged loosestrife (Lythrum alatum), are central Iowa species and acceptable for Story County, although it is unlikely they will persist due to competitive pressure. The other two species, ashy sunflower (Helianthus mollis) and blanket flower (Gaillardia pulchella), are not native to Iowa and detract from the quality of the reconstruction. Ashy sunflower is native to areas south and east of Iowa and blanket flower’s natural range is a south and west of Iowa.

On June 27, 1998 the bulk of the project area (both right-of-ways between 13th Street and E29, and the areas within the interchange at Highway 30) was surveyed in order to locate species that were relatively rare and did not occur at any of the study sites. This resulted in finding five species in the project area that were seeded on the study sites but had not yet been observed on the plots by the 1998 inventory.
Assuming that Anenome cylindrica and Pycnanthemum virginianum were not actually in the seed mixes and that Pedicularis lanceolata and Potentilla arguta are not established on the project area, then there are 16 species that have not established at all, and five species with low success (not on plots, but present on project area). Several explanations exist for these species low establishment, none of which can be verified in the current study. Seed viability could have been low, seed mortality due to pathogens or predators could have been high, environmental cues for breaking seed dormancy may have been absent, proper germination conditions may have been absent, or seedling survival could have been low. Some may be present in very small amounts or currently in an inconspicuous growth stage. Species like Allium canadense, Carex bicknelli, and Lysimachia quadriflora may be missed due to their lack of showiness or prominent stature. Annual surveys of the project area will continue to be performed so that the status of these species can be determined.

The occurrence of species non-native to prairies in Story County raises some concerns about reconstruction design and planting. Despite the work of several professionals at various stages of the reconstruction, there were still nine plant species seeded on the project that do not represent Story County prairie. Four of these species (nearly half) are not native to Iowa, and one of those is not native to North America. The other five species represent species planted outside their Iowa range. Of the nine species, six are known to be established. Is this a problem, and how much concern should there be with movement of species to areas outside their indigenous range? If the goal is to recreate prairie landscapes through reconstruction that exemplify the native landscapes, then as much effort as possible should be made to insure the integrity of the seed mix. There are many reconstructions in progress by many different agencies and individuals, and it is the amalgamation of all of these that truly makes the rebuilding of the prairie landscape a possibility. There must be a heightened sense of responsibility on the part of each reconstructionist to take steps that will help increase the integrity and value of their project’s contribution to the larger goal of rebuilding the landscape. In the case of the I-35 reconstruction, it seems that more attention to ecology could have been used since six of the ten species were purposefully included. Accidental species will always be a problem, but there are methods that will minimize them. Primary responsibility is to know the source of the seed and the nature of its harvest. Seeds from remnant prairies are more likely to be a mixture (which may include non-appropriate accidentals) especially if a bulk harvesting method was used. Species-specific harvests, such as hand collections, should be less likely to contain unwanted accidentals.

On the bright side, it appears that often these “accidental” species either do not establish well or may not be capable of maintaining themselves in the long-term. On the I-35 project, three of the nine species appear to have not established, four are currently present in small amounts, and only two are abundant enough to pose a serious concern. These two species – common wormwood (Artemisia absinthium) and decurrent false aster (Boltonia decurrens) – are not currently exhibiting evidence of increasing or becoming aggressive.