DOT Project Number:  90-00-LRTF-926, 927, 007

Fiscal Year:  1999-2000

Award:  $50,000

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

Summary Report:

THE EFFECTS OF SEPTEMBER AND DECEMBER HAYING, SPRING FIRE, AND A SPRING GLYPHOSATE APPLICATION FROM 2000-2002 ON THE I-35 PRAIRIE RECONSTRUCTION

When the prairie reconstruction on I-35 reached its fifth year of growth in 2000 and there was no evidence that Department of Transportation officials were planning to conduct prescribed burns, concern developed that continued fuel accumulation without litter management jeopardized the safety of interstate users and the success of prairie reconstructions along federal and state highways. An automobile accident or careless motorist could provide a source of ignition that potentially might burn several miles through continuous and heavy litter. The fact that this could happen under dry and windy conditions creates a very serious concern and a lucid need for a litter management plan. This research addressed this concern by implementing fall haying treatments.

Enhancement of the prairie reconstruction with additional species was also an interest investigated in this research. The interseeding of species missing or in low abundance is a goal many prairie managers are likely to encounter. In a five year old prairie reconstruction, much of the space for colonization is preempted by established species. Therefore success in interseeding will depend in part on the creation of colonization space, or in ecological terms, the occurrence of a disturbance or perturbation. Five species were selected for interseeding; they included Potentilla arguta (prairie cinquefoil), Astragalus crassicarpus (ground plum), Pedicularis canadensis (lousewort), Gentiana andrewsii (bottle gentian), and Veronicastrum virginicum (Culver’s root). All of the species but ground plum were included in seed mixes used on the I-35 project area, but none were observed to be established on the study sites. All five species were purchased from Prairie Moon Nursery in Winona, Minnesota.

Together these goals of controlling litter and creating colonization opportunities guided formulation of the treatments applied to plots in this research. Haying is a late summer or fall application and could be considered a perturbation or mild disturbance. It was applied twice (consecutively) to two of the four plots. Spring fire can function as a perturbation or mild disturbance and was applied once to one plot. A spring herbicide application of glyphosate may result in either a moderate or strong disturbance; it was applied once to one of the fall haying plots. The control plot was retained and represented no management activity.

Field Methods
After a year of rest from treatments (July 1999 to July 2000), the plots were inventoried in August 2000 to obtain pretreatment baseline data. The fall haying treatments began the fall of 2000. September haying occurred between September 13 and 17 and December haying was completed on December 9. The plots were cut at about 2 inches height with a power trimmer fitted with a metal grass-cutting blade. The plots were raked clean and the biomass scattered in the nearby prairie away from the treatment plots. Both of the hay plots were interseeded with all five species in mid December 2000.

On April 24, 2001, the plot hayed in September was sprayed with glyphosate in a hand sprayer at the rate of 2 fluid ounces per 100 ft2. One gallon of water was used per plot. For each plot, the forb species that were present and actively growing were recorded.

About a week later, on May 1, spring fire was applied to the plots treated with fall fire in earlier studies. The first plots burned in the morning were somewhat damp and did not burn completely, but by afternoon relative humidity was in the 40-50% range and the wind was over 15 mph. The drier conditions resulted in good burns. An effort to burn off the remaining litter on site 598N, the first plot burned in the morning, without the help of the Story County IRVM crew resulted in an accidental fire which burned an area encompassing all of 598N and 601N. Consequently the treatments on these sites were compromised and could not be included in the this study. Inventories were not performed on these two sites in 2001 and 2002. Four days following the prescribed burn, the spring burn and control plots were interseeded.

Interseeding was done the same way for all plots. The total number of seeds available for interseeding was calculated by determining the average seed weight from a sample of 500 seeds and calculating number of seeds ordered based on the total weight. Using the total number of seeds available, the number of seeds available for each of the 60 plots was calculated. Ground plum was seeded at a rate of 100 seeds per treatment plot, lousewort was seeded at 240 seeds per plot, and the remaining three species were seeded at 750 seeds per plot. In the fall of 2000, 60 lots of each species were weighed out and stored in a refrigerator inside a plastic bag with sand. The sand was used as a filler to help scatter the seeds evenly over the treatment plots by hand.

A LI-COR line quantum sensor (LI-191SA) was used to measure Photosynthetic Active Radiation (PAR) at the ground surface on all the plots during the first growing season (2001). It integrates PAR over a sensor 1 meter long and about 1 cm wide. Three measurements were made in each plot (pseudoreplicates) and averaged to obtain a single measurement for the plot. One ambient light measurement was made before and after the measurements were made on the plots at each site. These measurements were averaged to provide an ambient light comparison for the site. Measurements were made under full sun during the middle part of the day, generally between 1100 and 1400 hours. PAR measurements were made on three dates in 2001 – May 18, June 11, and July 14.

In August 2001 species inventories were completed before the second round of treatments was began in the fall of 2001. Both of the hay plots were cut and raked in September in the same way as in 2000. However, since a goal of the treatments was to establish new species already seeded on the plots, neither the herbicide nor prescribed fire treatments were applied in the spring of 2002 because both could be stressful on young plants establishing from seed. Measurements of PAR were made once during 2002 on June 8. The final species inventory was made on the study sites in August and September 2002.

Analytical Methods
The same analytical approach used in previous I-35 prairie reconstruction studies was used to evaluate treatment effects in this study. The density of a species from two inventories, one a pretreatment year and the other a post-treatment year, were used to calculate the per capita growth rate (r). Per capita growth rates were then compared among treatments with ANOVA. Three ANOVAs were performed, one evaluating treatment effects after one treatment application (2000 to 2001), one evaluating treatment effects after the second application of haying and two growing seasons (2000 to 2002), and one evaluating treatment effects over all treatments applied (1997 to 2002). Only the species common enough to have 33% nonzero observations in the inventories were selected for evaluation.

A direct evaluation of the susceptibility of species to glyphosate was made for those species observed at the time of the herbicide application. The densities and frequencies of these species on the herbicide plots in the 2000, 2001 and 2002 inventories were compared with repeated measures ANOVA using the sites as spatial blocks. Likewise, PAR was compared among ambient levels and treatments with repeated measures ANOVA. Multiple comparison tests were completed when necessary with Tukey’s test.

Results
None of the five species interseeded on the plots was observed in the species inventories. It is possible that they may have been present as an unidentified forb seedling. However, concern for this possibility during the inventories impelled careful inspection of forb seedlings with the five interseeded species in mind. Successful establishment of interseeded species is not realized until young (identifiable) plants are present. Since that did not happen, no direct assessment can be made of the treatment’s value for interseeding.

An indirect assessment of the treatments can be made by comparing the light measurements. Assuming that germination and establishment will be enhanced with favorable light levels, and that a result of disturbance or perturbation is increased light, the effect the treatments have on creating favorable environments for interseeded species can be evaluated. Two treatments, the combination of fall haying and spring glyphosate and the spring burn, resulted in light levels that were about 80% of the ambient light early in the growing season and significantly greater than fall haying alone and the control. PAR decreased in all the treatments during the rest of the growing season, but only the fall hay/spring glyphosate plots remained above 1000 umol/m2/s-1 one month later in June. Growth on the spring burn plots reduced surface PAR to levels equivalent with the fall hay plots and significantly lower than PAR on the fall hay/spring glyphosate plots. At mid-summer, only the fall hay/spring glyphosate plots exhibited PAR significantly higher than the control plots. Two years of repeated fall haying provided higher light levels that the single spring burn and the control.

These results suggest that the greatest disturbance effect and concomitant light availability in established prairie was achieved with a combination of fall haying and spring glyphosate. While the glyphosate probably provided most of that effect, the importance of the fall haying prior to application should not be overlooked. The removal of litter by haying would encourage earlier growth of plants in the spring and allow better contact of the herbicide with foliage. Each of the treatments resulted in different levels of disturbance or perturbation. The fall haying and spring glyphosate treatment produced a medium disturbance while the spring burn resulted in a minor disturbance. Two years of successive fall haying produced a perturbation during the second growing season, while the single fall haying was no different than the control. It appears that for interseeding to be successful in an established prairie, small-scale medium to severe disturbances need to occur and that an herbicide such as glyphosate can help accomplish that.

The use of glyphosate as a disturbance mechanism and haying as a litter control technique are management options prairie managers should consider, but not without some knowledge of effects that either will have on the community composition of prairie. The results of the analysis of treatment effects on species growth rates will provide some helpful information.

Grasses
Treatment effects were observed for only two native grasses – big bluestem and Canada wildrye. All the plots of big bluestem incurred negative growth from 2001 to 2002, but the growth on the control and spring burn plots was lower than that on the fall hay and herbicide plots. This effect carried over into the 2000 to 2002 period in that the fall hay/spring herbicide plots had marginally higher growth rates than the control plots. The wildrye species exhibited lower growth rates due to spring burning than on control plots. When treatements are considered together over the period from 1997 to 2002, the treatment consisting of two spring burns and then two fall hayings was beneficial for big bluestem (increased growth rate compared to control) and harmful for wildrye (decreased growth rate compared to control).

Indiangrass, little bluestem and side-oats grama exhibited similar growth rates for all treatments and comparisons. Indiangrass hinted at better growth rates on the control and spring fire plots and lower growth due to fall haying, but none of these were statistically significant. Little bluestem also showed numerically higher growth rates on spring burn plots, while side-oats grama indicated a bit lower growth on the fall hay and herbicide plots, but none of these were significantly different from growth on the other plots.

Among the non-native grasses, there were very clear treatment effects. All three cool-season perennials – smooth brome, tall fescue, and Kentucky bluegrass – were drastically reduced by the combination of fall haying and spring glyphosate. The three species exhibited a similar pattern of significant decrease the first growing season, followed by a significant increase in growth the second growing season, and ending the two year period with a significant decrease in growth rates. Although they made some recovery the second growing season, their tiller densities were still much lower in 2002 than in 2000, thus the glyphosate provided control over a two year period. There were no differences among any of the other treatments. Smooth brome growth was a bit lower in the spring burn, but there was not a significant decrease. This may be because the burn was conducted too early. Use of a late spring burn is an important factor in exerting stress on smooth brome. Fall haying did not indicate any signs of controlling smooth brome or Kentucky bluegrass, but there was a hint of stress on tall fescue after the December haying. Although not significant, it is interesting to note the negative growth rate compared to the control corroborates the significant negative effect fall fire had on tall fescue in treatments for earlier studies. Often it is not so much the nature of the stress that’s important as it is the timing of the stress.

Foxtail species exhibited a response to the treatments somewhat opposite in effect to the exotic cool-season perennial grasses. The fall haying/spring glyphosate plots exhibited growth rates for foxtail that increased significantly the first growing season, were neutral during the second growing season, and ended the two year period with a significant increase. Foxtail most likely responded to the habitat availability created by the decrease in cool-season perennial grasses. Foxtail also experienced increased growth rates due to spring fire. Fall haying tended to favor foxtail growth compared to the control, but there was no significant difference.

The graminoid seedlings are most likely represented by exotic species such as foxtail, crabgrass, and witchgrass. Very young unidentifiable tillers are also counted in this category. Fall haying was beneficial for them and significantly increased their growth compared to the control plots during the first growing season. By the second growing season that effect was gone, but the fall hay/spring glyphosate plots exhibited marginally better growth over the control plots and spring fire (decreases in growth were smaller). For some reason, herbicide application did not promote graminoid seedlings as much as it did foxtail. Lower level disturbance or perturbation from fall haying was more effective.

The exotic grasses response to the combination of treatments over the period from 1997 to 2002 seemed to be most influenced by the glyphosate application. The biggest effect (a significant decrease in growth rates) occurred in the treatment consisting of summer mow, fall hay and spring glyphosate. Smooth brome was also secondarily reduced by the combination of two spring fires and fall haying. For foxtail, the clear effects seen separately in individual studies seemed to cancel out over the entire 1997 to 2002 period. Growth rates were mostly slightly negative with a marginal difference in the slightly higher growth in the treatment consisting of two spring fires and fall hay treatment relative to the control. Density of graminoid seedlings decreased in all treatment plots over the 1997 to 2002 period, perhaps as a consequence of prairie maturation over time. Fall haying as well as spring and fall fires were the most beneficial and permitted graminoid seedlings to nearly maintain stable population growth.

Prairie Forbs
The effects of September and December haying, spring fire, a spring glysophate application were observed for 7 of the 12 prairie forbs analyzed. Butterfly milkweed exhibited a marginal increase in growth rates on the fall hay/spring glyphosate plots relative to the December hay plots. However, a switch occurred the second growing season with lower growth rates on the fall hay/spring glyphosate plots relative to the fall hay plots. Since both of these plots had the same treatment (September haying) in fall 2001, the switch must be due to indirect affects. In fact, no butterfly milkweed plants were observed at the time of glyphosate application, so indirect effects are the only possibility. Perhaps the removal of cool-season grasses the first growing season was initially beneficial to butterfly milkweed, but by the second growing season competition from ruderals colonizing the disturbance was detrimental.

New England aster experienced significantly lower growth rates on the fall hay/spring glyphosate plots than the control and fall hay plots. The plants rebounded the second growing season with significantly higher growth rates than the control, which made their growth over the two year period 2000 to 2002 equivalent to the control. Overall, two years of fall haying was the most beneficial to New England aster.

Frost aster exhibited a variety of effects. During the first growing season, frost aster growth was significantly decreased by the spring fire treatment and further significantly lowered by the fall hay/spring glyphosate treatment. At the same time and relative to these decreases, the December haying plots exhibited an increase in growth. During the second growing season, frost aster plants rebounded on the both the spring burn plots and the fall hay/spring glyphosate plots with significantly higher growth than on the control. The net effect over two years was a detrimental effect from fall hay/spring glyphosate relative to a beneficial effect from fall haying. The two-year negative growth rates observe for spring fire and fall hay/spring glyphosate were equivalent to the control plots.

The prairie clovers were favored by fall haying; they displayed higher growth rates on the December hay plots relative to the control plots. This effect was retained marginally over the two year period from 2000 to 2002. The glyphosate application did not affect prairie clover density.

Five prairie forb species were not significantly affected by any of the treatments. Showy tick trefoil, ox-eye, blazingstar, sweet coneflower and rigid goldenrod experienced similar growth among all treatments. The treatments that numerically resulted in the best growth over the treatment period were two years of fall haying for showy tick trefoil, ox-eye and rigid goldenrod, spring fire or control for blazingstar, and spring fire or two years of fall haying for sweet coneflower. While glyphosate did not boost the population growth of these species, it did not hinder it either.

Wild bergamot experienced a marginal effect from the treatments. The first growing season growth rates were marginally decreased in the spring fire and the fall hay/spring glyphosate plots relative to the control. By the second growing season, and also over both growing seasons, these slight differences were gone and bergamot growth was equivalent among treatments.

Grey-headed coneflower was drastically impacted by the fall hay/spring glyphosate treatment. Population growth on the fall hay/spring glyphosate plots was significantly lower than on all other plots during the first growing season. The plants on the fall hay/spring glyphosate plots rebounded some during the second growing season with a marginally higher growth rate than the control plots. This regrowth was sufficient to produce equitable (although negative) growth rates over both growing seasons for the control and the fall hay/spring glyphosate plots, but not enough to match the slightly positive growth rates observed in the plots hayed two years in the fall.

During the first growing season, black-eyed susan exhibited significant positive growth from spring fire relative to its significant negative growth after December haying. By the second growing season, the fall hay plots and the fall hay/spring glyphosate plots experienced significantly higher growth rates than the control plots. The overall effect of both growing seasons was significantly higher black-eyed susan growth on all of the treatments relative to the control. Any type of disturbance or perturbation was apparently beneficial to black-eyed susans.

Effects of combined treatments over the 1997 to 2002 period on prairie forbs were very uncommon. For 10 of the 12 species, there were no effects on growth rates despite a variety and range of treatments. Only frost aster and grey-headed coneflower were significantly affected. For frost aster, the combination of two spring burns followed with two fall hayings provided significantly higher growth rates than did either the spring burn or the fall hay/spring glyphosate treatments. For grey-headed coneflower, the most highly disturbed plots – the combination of summer mowing, fall fire, and spring glyphosate – produced significantly lower growth rates than all other treatments.

Native Ruderal and Exotic Forbs
Significant effects of September and December haying, spring fire, a spring glysophate application were observed for all 5 weedy forbs analyzed and for the unidentified forb seedlings. As would be expected, due to the ruderal nature of these species, most of the significant effects were associated with the greatest disturbance – fall hay/spring glyphosate. Wild lettuce species (including exotic prickly lettuce and native yellow lettuce) increased dramatically on the fall hay/spring glyphosate plots, but not until the second growing season. Mean stem density on fall hay/spring glyphosate plots was nearly 35/m2, while all other plots had mean stem density less than 7/m2. Since there was no effect seen the first growing season, the response over both growing seasons was the same as the second growing season.

Dandelion was the only species in this group to not respond to the fall hay/spring glyphosate treatment. Its growth was fairly equivalent among all treatments for all periods, except for a marginally significant increase after two years of fall haying compared to the control and spring fire plots. Basal-leaved species such as dandelion might grow the best after fall haying removes competition and litter thereby providing a window of growth for several weeks until frost occurs.

Horseweed exhibited a response to the treatments identical to wild lettuce. An impressive and significant increase in growth occurred during the second growing season on the fall hay/spring glyphosate plots. Mean stem density on fall hay/spring glyphosate plots was nearly 50/m2, while all other plots had mean stem density less than 10/m2. As with wild lettuce, no effects were observed the first growing season thus the response over both growing seasons was equivalent to the effects during the second growing season.

Growth of bird’s foot trefoil increased significantly in response to both fall haying and the combination of fall hay/spring glyphosate. Once again the increase on the fall hay/spring glyphosate plots did not occur until the second growing season. The overall growth for both growing seasons on the fall hay/spring glyphosate plots was not significantly different from the control, due to a negative growth trend the first growing season. The increase on fall haying plots was only marginally significant and only manifest for the two-year period encompassing both growing seasons (i.e., accrued over both seasons).

Yellow wood sorrel response was similar to wild lettuce and horseweed. During the first growing season, spring fire and fall haying stimulated a marginal increase in growth relative to the fall hay/spring glyphosate plots. But by the second growing season, growth of yellow wood sorrel flourished on the fall hay/spring glyphosate plots while growth on the other plots was negative. Fall haying was probably more favorable than the spring burn because haying occurred prior to the second growing season but spring burning did not. In the second growing season mean stem densities were about 27/m2 for fall hay/spring glyphosate plots, 6/m2 for fall hay plots, and less than 1/m2 for spring fire and control. The more intense disturbance caused by fall hay/spring glyphosate produce the best growth, followed by the consistent disturbance of two consecutive years of fall haying.

Individuals counted as forb seedlings were likely both native and exotic ruderal species, although it is possible some seeded species could also be represented. The effects of the treatments were very clear for forb seedlings. In the first growing season, a single fall haying in December resulted in significantly higher growth than in all other treatments. By the second growing season, growth rates had dropped to negative values in all plots except for a positive and significantly higher growth rate in the fall hay/spring glyphosate plots. For both growing seasons combined, growth was significantly higher on the fall hay plots and the fall hay/spring glyphosate plots than the control and spring fire plots. Although none of the interseeded species were observed, the positive effects of fall haying on forb seedling density suggest that it may be a better approach for use in interseeding than spring fire.

Effects of combined treatments over the 1997 to 2002 period on ruderal and exotic forbs occurred for most of the species and mostly reflect the effects produced by September and December haying, spring fire, a spring glysophate application. Wild lettuce, horseweed and yellow wood sorrel all exhibited the significantly higher growth in the plots with the greatest disturbance – the combination of summer mowing, fall haying, and spring glyphosate – than in the control. Growth of yellow wood sorrel was also enhanced by the combination of spring fire and fall haying. There were no long-term effects shown by dandelion and bird’s foot trefoil. Growth of dandelion has increased similarly on all plots over the 5 years and growth of bird’s foot trefoil has decreased similarly over the 5 years.

Species Sprayed with Glyphosate
When the glyphosate was applied to the plots, those forb species present and known to have been sprayed were recorded. This provided an opportunity to investigate the direct effects of glyphosate on a greater number of species. For some of these species, the direct effects of glyphosate corroborate the effects already described. For example, during the first growing season the mean density of frost aster on the fall hay/spring glyphosate plots dropped significantly to 4.5% of its density prior to glyphosate application in 2000. During the second growing season, frost aster regained enough stems to reach about 35% of the original density, but that was statistically equivalent to its density in the first growing season and still significantly lower than its density prior to glyphosate application. In terms of frequency, frost aster exhibited the same pattern, with the only difference being that during its rebound the second growing season frequency was significantly increased from the low point exhibited the first season after glyphosate application. Although native and included in the seed mix, frost aster is a common old-field species that can achieve high abundance. A spring glyphosate application will effectively thin it and make room for more desirable species.

The direct effects of glyphosate on grey-headed coneflower also corroborate the treatment effects described earlier. Density and frequency of grey-headed coneflower decreased significantly from glyphosate and did not exhibit a rebound the second growing season as did frost aster. Two additional species not included in the treatment analyses were directly affected by glyphosate. Smooth aster frequency was significantly decreased for both growing seasons after glyphosate application. Its stem density exhibited a decrease and then a rebound, but these means were statistically equivalent. Golden alexander was also susceptible to glyphosate, exhibiting significant decreases for both density and frequency. A few other species appear susceptible to glyphosate, but due to low replication lack significant effects – for example reed canary grass, black-eyed susan, and bird’s foot trefoil .

Many species present when the glyphosate was applied did not exhibit susceptibility. Some of these were well replicated, like wild bergamot, dandelion and sweet coneflower. Other species, despite having only one or two observations, appeared to be tolerant (azure aster, Canada thistle, three-finger coreopsis, and spiderwort) or even positively affected by glyphosate (Virginia anemone, common wormwood, yellow rocket, decurrent false aster, and ox-eye). Some species, especially ruderals, appeared to be controlled by glyphosate the first growing season but then rebounded the second growing season to an abundance higher than initially via a strong positive response to the disturbance created (wild parsnip, wild lettuce).

Summary
Both haying and glyphosate applications have some merit for consideration by prairie managers. Haying is well documented in research as simulating fire effects. The removal of litter is the key mechanism in the ecological effects of fire, and haying accomplishes the same thing although usually less effectively and with the result of greater nutrient removal. Haying can also simulate the disturbance or perturbation effects of fire, probably even better than fire since haying can be applied at any time and in particular at times when fire cannot be applied due to low flammability. Although not demonstrated in this research, there is other research that suggests fall haying can minimize dominance by warm-season native grasses and clonal, late summer forbs (e.g., sunflowers). An additional advantage of haying over fire is that it can also provide an economic return. Prairie hay has value as forage or, if cut in the fall, as a source of mulch and seed.

The use of glyphosate in prairie restoration is most commonly done as a pre-plant herbicide treatment to remove cool-season grasses and other non-native plants. A glyphosate application prior to planting was utilized in the I-35 reconstruction, and early research documented a strong recovery of exotic cool-season grasses during the first few years. Subsequent glyphosate application in this research clearly shows a strong inhibitory affect on cool-season grasses. Time will tell if the glyphosate treatment in this research was more effective than the initial burn down. It seems likely that the lack of competition in the first years of a reconstruction would favor a strong regrowth by top-killed cool-season grasses. An advantage of using an early spring application of glyphosate after the prairie is established is the competitive environment created by established prairie plants. Regrowth of cool-season perennial grasses already weakened from top-killing by the glyphosate will be compromised by competitive interactions from healthy prairie species just beginning to produce seasonal growth.

None of the prairie species susceptible to glyphosate in this study were completely eliminated. Their abundance was reduced, but some plants persisted. Exposure of prairie plant populations to glyphosate can be minimized by timing the application to early spring or late fall, when prairie species are dormant. Spring application has the advantage of combining the herbicide stress with competitive stress, while the fall application combines herbicide stress with winter stress. Tall fescue appears to be better controlled by stresses that weaken plants going into winter. Fall burning was more effective than spring burning, so a fall application of glyphosate may be more effective than spring glyphosate, although this research documented reduction of tillers from 150/m2 to less than 5/m2 after a spring application of glyphosate.

Exposure could be further minimized by careful selective application of glyphosate in situations when a hand sprayer can be used. In any case, seeding of prairie species into the areas of glyphosate application is advised to encourage the establishment of desirable species. Glyphosate creates a more intense and longer disturbance than fall haying or spring burning and consequently is more likely to result in establishment of ruderal and exotic species recruited from the seed bank. If seeding can be accomplished (and even if it isn’t), mowing the patches of glyphosate disturbance with either tractor or hand equipment will help control weedy growth and favor establishment of slower growing natives.