DOT Project Number:  90-00-LRTF-614

Fiscal Year:  2006

Award:  $23,976.00

Principal Investigator:  Dr. Laura Jackson, Department of Biology, University of Northern Iowa, Laura.L.Jackson@uni.edu

Other Project Participants:  Craig Hemsath, Department of Biology, University of Northern Iowa

Research Report:

ADDING WILDFLOWER DIVERSITY TO SPECIES-POOR GRASSLAND: REQUIREMENTS OF CONSERVATIVE SPECIES

Summary:

            Investigations into the special establishment requirements of two showy, conservative spring wildflowers has led to the conclusion that both are susceptible to heavy seed predation.  Seed predators such as mice have great potential to eat the seeds of prairie plantings, increasing the cost of prairie restoration and reducing forb diversity.  We tested the effectiveness of a feeding deterrent (capsaicin) and a feeding supplement (black sunflower seeds added to the prairie seed mix) to reduce seed predation and boost forb establishment.  Preliminary results show that over just 5 days, 50% of seeds were eaten by mice and another 22% by ants and other invertebrates.  Dusting capsaicin on seeds had no effect for most species, but it did nearly triple the germination of midland shooting star.  In a late Spring planting, adding extra seed to the research plot had the most  success, cutting seed predation in half (15% predation  vs. 30% predation in the control treatment over 11 days).  In a Fall planting, however, adding extra seed had no significant effect.  These results have been presented at the North American Prairie Conference in Kearney, Nebraska, the Iowa Academy of Sciences Conference in Pella, Iowa, and the Iowa Prairie Conference in Sioux City, Iowa.

            LRTF Funds have supported field work in summer and fall 2006, and spring-summer 2007 by graduate student Craig Hemsath.  The funds leveraged a full-time undergraduate field assistant each summer, and $3000 in other grants.  LRTF funds made it possible not only to conduct field experiments but also to analyze the data and share this information with roadside managers and other restoration practitioners.   One of us (Laura Jackson) will be presenting results of this research to the Winterfest event for the Iowa Association of County Conservation Boards in January 2008.  At least one manuscript is being drafted for publication in Restoration Ecology.

Background:      

            Since 1998, research at the University of Northern Iowa Prairie Preserve has been devoted to forb enhancement of species-poor grasslands.   Species-poor grasslands are prairie plantings dominated by warm and cool season grasses, with very little abundance or diversity of wildflowers.  These grasslands, which are common throughout the state along roadsides and CRP fields, are inferior at excluding exotic and invasive weeds, provide poor wildlife habitat, and do not bestow the same aesthetic qualities of a successful diverse prairie reconstruction.

            A study was begun in spring 2005 planting prairie phlox and shooting star at a seed density of 15,000 seeds/m² (1,400 seeds/ft²) to study effects of four different seed treatments on seed germination. We wished to improve success at planting these two expensive, hard-to-establish species. We found very poor germination of both species during the first growing season. Upon further investigation we discovered many of the seeds were absent from the soil during subsequent seed recovery attempts. This led to a few interesting questions about the fate of those seeds:

• Were those seeds consumed by seed predators such as rodents and ants?
• Were the seeds dormant, and possibly entered the seed bank?
• Have the seeds died in soil and will never germinate?

            After an extensive literature search, we discovered that seed predators appeared to be a possible reason for the seed disappearance and resulting poor germination rates. Previous work has shown rodents are capable of removing 50-95% of seeds (Heithaus, 1981; Sullivan, 1982; Nolte, 2000). This amount of granivory can drastically reduce the amount of new plant recruitment in a revegetation project. With money from LRTF we were able to expand on this problem and study possible means to reduce the amount of damage seed predators can have on a seeding project. The study was conducted in two sets of experiments.

Part I - Reduction of Granivory in a Prairie Reconstruction

The following questions were addressed in several experiments conducted in Spring 2006-Summer 2007:

•          How much seed predation are rodents responsible for in a prairie reconstruction?

•          Is it possible to use a feeding deterrent such as the chemical capsaicin to keep rodents from eating broadcast prairie seeds, or that would be cheap enough and effective enough to warrant its use?

•          Could we use an extra source of food that is cost-efficient enough with our broadcast seed to distract the seed predators from eating our prairie seeds?

•          What role do invertebrates play in seed loss and would capsaicin be an effective deterrent?

•          Would rain wash off or reduce the effectiveness of a chemical deterrent such as capsaicin?

•         Does the amount of granivory influence the density of seedlings the following growing season?

Part II - Addition of Conservative, Showy, and Expensive Seeds

The following questions were addressed in four experiments, begun initially by Amy Carolan in Spring 2005, and repeated with the same conservative species and planted in Spring and Fall 2006.

•          Does planting time affect the success of more conservative species?

•          Do pre-planting seed treatments affect germination in conservative species?

Would the addition of capsaicin reduce the amount of seed predation and thus increase the number of seedlings?

Methods

Part I - Reduction of Granivory in a Prairie Reconstruction

Experimental Design

            All  experiments were conducted on the UNI Tallgrass Prairie, a species-poor reconstructed prairie on the University of Northern Iowa campus.  The first experiment, conducted in June-July 2006, was a randomized block design consisting of two, 25x50-m blocks, each containing eight, 5x5-m plots with 5-m between each plot and a 5m buffer around the plot grid. Each block was mowed one week before the beginning of each feeding trial to a height of 15-cm. The mowing kept the area uniform between trials throughout the growing season. Within each of the 16 whole plots, seed cards were randomly placed allowing us to measure rates of seed loss. Half of the seeds used were coated with a capsaicin containing powder. Four of the eight plots, in each block, were randomly chosen to receive the extra food treatment. Black-oil sunflower seeds were broadcast on top of the assigned 5x5-m plots at a rate of 430seeds/m² (40seeds/ft²).    

            In November of 2006 another trial was conducted which consisted of a randomized block design with three treatments at the whole plot level (capsaicin, alternative food, untreated control) and three treatments at the within-plot level (Silphium integrifolium, Dodecatheon meadii, and Phlox pilosa). Each block consisted of nine, 5x5-m plots. The plots were then randomly assigned one of the three plot-level treatments. The treatments consisted of addition of the alternative food source (Helianthus annus), capsaicin treated seeds on all cards, or a control receiving neither the alternative food nor capsaicin. Within each plot, there were 15 total cards, five cards each of Silphium integrifolium, Phlox pilosa, or Dodecatheon meadii seeds.

            Along with the seed removal cards, we broadcast additional seed of the three species onto all the plots. This was done to allow seedling census to occur during the spring of 2007. Each of the eighteen plots received the same amount of seeds. In addition, the seeds for the capsaicin treated plots did receive the capsaicin treatment. We used the following seeding rates to broadcast the three species: Silphium integrifolium broadcast at 100 seeds/m², Phlox pilosa at 250 seeds/m², and Dodecatheon meadii at 300 seeds/m². Silphium integrifolium was seeded at a lower rate since this species is considerably large seed, we did not want to overwhelm the plots with very high densities of one species. Due to lack of availability, the rates used for Phlox pilosa and Dodecatheon meadii were the maximum density we could use.

Determining Rate of Seed Loss

            To measure seed loss we used a method developed by Paula Westerman and Matt Liebman at Iowa State University.  They used seed cards to measure seed loss in agricultural systems. The seed card consisted of an 11x14-cm (4.25x5.5-in) piece of sandpaper with 30 rosinweed (Silphium integrifolium) seeds glued to the sandpaper using 3M spray adhesive (Figure 1). Rosinweed is a large aster seed, common in native roadside plantings, similar to sunflower, and previous research has suggested rosinweed is a highly desirable food item for rodents. Seven of the 14 cards in each plot contained seeds previously treated with Squirrel Away^®, a commercial feeding deterrent containing the chemical capsaicin. The treated seeds were treated according to the manufacturer’s recommendations. The remaining seven cards remained untreated. Each card was identified by a metal tag allowing us to follow each card individually.

            For the first week after the commencement of each feeding trial the amount of seeds remaining on each card were counted and recorded. Starting the second week, scoring occurred on an every other day basis with a final score taken 18 days after the trial began. Data were analyzed with a repeated measures ANOVA with the capsaicin treatment as the first factor and the extra food treatment as the second.

            The fall trial lasted six weeks, with a day of data collection once about every ten days. The cards were not removed as snow cover made locating the cards very difficult. The procedure placing the cards and recording the seeds remained the same from the summer trial.

Determining Granivory’s Influence on Seedling Emergence

            From May 23^rd to June 5^th, 2007, we counted the number of seedlings present from the Fall 2006 trial. Three 0.5-m transects were randomly selected across each 5x5-m plot. All seedlings from the three species seeded in Nov. 2006 were identified and recorded.

Granivore Survey

            Throughout each feeding trial we trapped small mammals to determine granivores responsible for the seed loss. Early morning bird counts were also attempted to reveal any seed loss due to birds. A Sherman live trap was placed in each plot three consecutive days each week during the trial. The traps were baited with the same sunflower seeds used for the extra food treatment. The traps were opened each evening and closed each morning. The identification of each captured animal was recorded along with the location within the block.

Role of Invertebrates in Prairie Seed Loss

            Using 1.2-cm (1/2-in) hardware cloth, 30 exclosures measuring 15x7.5x30-cm (6x3x12-in) were built. Fifteen were closed on all six sides, completely closing the exclosure off to rodents or birds, whereas the other 15 left two sides open allowing the entrance of vertebrates. Within each exclosure two new seed cards were placed. Each of these seed cards were constructed in a manner similar as above, and then attached to the inside of a 10-cm (4-in) Petri dish (Figure 2). The seed species used included: purple prairie clover (Dalea purpurea), Indian grass (Sorghastrum nutans), and grey headed coneflower (Ratibida pinnata). We had previously determined the palatability of several native seed species. Of these trials we used the three species we concluded were most palatable to invertebrates. One tray contained seeds treated with the capsaicin powder as above, and the other tray the seeds remained untreated. The finished exclosures were placed at 15 randomly selected sites in the UNI Campus prairie preserve. Each site held one open and one closed exclosure approximately two meters apart.

            Recording followed the same pattern as with the Rosinweed experiment. We counted seed loss daily for a week and extended the collection period over several weeks counting on a more infrequent basis. The groups were compared with a three-way ANOVA to test for significance.

Simulated Rain on Capsaicin Effectiveness

            To test if capsaicin has the ability to persist on the seeds as a viable feeding deterrent, we simulated two different rainfall events on capsaicin treated seeds. For this experiment raw, whole, shelled sunflower seeds were treated with capsaicin as before. We used four different treatments: a no rain/full strength capsaicin, seeds subjected to 0.76-mm rain event, seeds subjected to 7.6-mm rain event (mean rainfall event for June in Cedar Falls), and a no rain/no capsaicin control. The simulated rain was determined by calculating the rate of water being poured out of a watering can. Then we determined the length of time needed to reach the desired volume of water.

            Seed cards were constructed in similar fashion as the Silphium integrifolium experiment with 30 seeds on each card. Fifteen replicate plots (5x5-m) were marked out in two transects with 8-m between each plot. Inside each plot two cards of each treatment were randomly placed. Seed loss was counted daily until all seeds were removed. Partially eaten seeds were counted as consumed and then removed from the card. A one-way ANOVA was used to test for differences among groups.

Part II - Addition of Conservative, Showy, and Expensive Seeds

Experimental Design

            This experiment was a randomized block design.  A 15x20-m area was set aside on the UNI campus prairie where there was a high density of warm season grasses.  In the fall of 2002 the area was burned in preparation for the addition of forb seed.  Twenty, 3x5-m blocks were then established in the area.  Half of the blocks were designated for the addition of Midland shooting star (Dodecatheon meadii) while the other half was designated for Prairie phlox (Phlox pilosa).  Both species are considered to be conservative and showy as well as expensive.  We tried to obtain Hoary Puccoon (Lithospermum canascens), but was unavailable at any price. Within each of the blocks four, 50x225-cm plots were permanently fixed and assigned to one of four treatments.

Treatments  

            The treatments for this experiment include two different planting times and two different pre-planting seed treatments for the 2005 seeding.  The planting times included an early planting (March 12^th, 2005) or a late planting (May 23^rd, 2005).  The pre-seeding seed treatments included stratified (cold/wet treatment for four weeks) or unstratifed.  All plots were seeded at a rate of 1.3 seeds/cm² (13,000seeds/m²).  All plots were mowed throughout the growing season to keep the established vegetation at a height of 15-cm.  Plots were scouted for germination until seedling emergence was detected (June 9^th) at that time seedling censuses began and were conducted every two weeks until seedling numbers began to decline.

            During 2006, the treatments were modified to reflect and supplement concurrent experiments with seed predation. The treatments for the 2006 seeding include two different planting times and two different pre-planting seed treatments. The planting times include a spring planting (April 5^th, 2006) and will include a fall planting (projected October 2006). The pre-planting seed treatments include capsaicin treatment (Squirrel Away^®) or untreated.

Beginning in May 2006 seedling censuses began on all planted plots (2005 and 2006 plantings). Using a 50x25-cm quadrat broken into a grid of 5x10-cm rectangles the entire plot of phlox was counted. Due to the high number of shooting star seedlings we randomly chose five grid points from each quadrat to sample. Effectively we sampled 20% of the plot.

            For the 2005 planting we analyzed the data using an ANOVA looking for differences among the planting time and seed treatment. The spring 2006 planting we only used ANOVA for the differences between seed treatments.

Results

Part I - Reduction of Granivory in a Prairie Reconstruction

            Adding an extra source of food for seed predators did significantly reduce the amount of seed predation on Silphium integrifolium (p=0.0017) in Spring 2006, whereas the capsaicin had no effect on the rate of seed loss (p>0.05) (Figure 3). The extra food treatment provided a window of about eight days for the seed to germinate or bury into the soil to escape predation. In this study we observed predation in the plots ranging from 20% in the extra food treatment to over 90% in the control plots.

            In the Fall 2006 trial, significant differences in the means (p < 0.0001) as well as the rate over time (p < 0.0001) were observed between the two blocks. Block 1 had a higher rate of predation with an average loss of 1.58% per day compared to Block 2’s 1.37% per day. This resulted in Block 1 having an average of 30.14% survival at the end of the trial, whereas Block 2 held a 44.88% average survival. Much of the difference in blocks comes from the large difference in Silphium integrifolium predation between the two blocks  which had  an average survival of 48.2% ±3.2% for Block 1, vs.  79.6% ±2.06% survival in block 2. The difference in blocks along with differences in seed preferences led to a block x species interaction (p < 0.0001) and an interaction over time (p < 0.0001).

            Granivore seed preference made a significant (p < 0.0001) difference in rates of predation (Figure 4). Phlox pilosa and Dodecatheon meadia had 95.7% ±2.1% and 83.7% ±2.07% seed loss 14 days into the trial, compared to 12.4% ±1.1% loss for Silphium integrifolium over the same time..

            None of the three seed treatments showed any difference (p = 0.13) in predation. This was also true for the treatments over time (p = 0.26). However, there was a significant (p = 0.014) treatment x species interaction for their effect over time. The data did indicate a trend of the sacrifice food treated plots having slightly higher rates of predation (Figure 4). We also recorded 71.68-mm of rain during the month of November, with 50.4-mm falling before November 14^th.

Granivore Survey

            The most common seed predator captured was the meadow vole (Microtus pennsylvanicus), consisting of all but one captured animal. Along with the voles, one thirteen-lined ground squirrel (Spermophilus tridecemlineatus) was caught. We did make several day-time observations of ground squirrel tunnels inside the plots. Ground squirrels were also seen running through and even eating seeds from the seed card. Early morning bird surveys revealed no birds preying upon the experiment, but we cannot exclude the possibility of birds playing a minor role in seed predation.

Determining Granivory’s Influence on Seedling Emergence

            Seedling emergence in spring 2007 partially reflected the observed seed predation the previous Fall.  There were significant (p<0.0001) species differences occurring between Phlox pilosa, Dodecatheon meadia and Silphium integrifolium. Phlox pilosa, which experienced the heaviest seed predation,  averaged 0.59 seedlings/1000 pure live seed (PLS) compared to 2.28 and 2.82 seedlings/1000PLS for Dodecatheon meadia and Silphium integrifolium, respectively. There were no differences between any of the seed treatments or their interactions with species or block.

            There was a significant correlation (r²=0.59, p=0.01) between Dodecatheon meadia predation recorded in the Fall Trial and the number of seedlings present (Figure 5). Silphium integrifolium did not have a significant correlation (r²=0.24, p=0.34). Predation and seedling numbers were too low to produce a correlation (Figure 5).

Role of Invertebrates in Prairie Seed Loss

            Both vertebrate and invertebrate seed predators are selective in their seed choice. In open exclosures, seed predators which included both invertebrates and vertebrates selected R. pinnata over the other two species with nearly 100% eaten after five days. S. nutans was the second choice (66% eaten) and D. purpurea the third (45% eaten). In closed exclosures, seed predators would have been mainly invertebrates.  R. pinnata was the most preferred with 34% eaten; D. purpurea was similar with 24% eaten. S. nutans was the least preferred, with 10% eaten. All of this was after five days of predation.

Simulated Rain on Capsaicin Effectiveness

            The capsaicin powder did offer protection in predation rates compared to the no capsaicin control, 85% predation compared to 97% for the control (Figure 6). However, the simulated rain at both levels did diminish the effectiveness of capsaicin as seen in the increased levels of predation, 92% and 93% The rain treatments fell in between the two controls, no rain/full capsaicin and no rain/no capsaicin, and were not statistically different from either of the two. This study may have been limited by using sunflower kernels, a very attractive bait. It appears the use of sunflowers was too good a food source for granivores and the seed was eaten no matter what.

Part II - Addition of Conservative, Showy, and Expensive Seeds

2005 Plantings- Seed stratification and Planting time

            Both the stratification and planting time had no difference in the germination of shooting star (p>0.05) (Figure 7). All four of the treatments averaged from 211-311 seedlings/m². Phlox on the other hand showed a very strong negative reaction to stratification (p=0.002). The stratified phlox plots had seedlings ranging from 0-6 seedlings/m² compared to the unstratified ranges of 119-214 seedlings/m² (Figure 7). The planting time for phlox showed no significant effect for either of the stratification treatments (p>0.05). Even for the best plots the germination rate is still 3.8% (577seedlings/15,000seeds) of the seeds and overall the average is still under 1% germination.

2006 Plantings- Capsaicin treatment and planting time

            Capsaicin treatment had a very strong positive effect on the germination of shooting star (p=0.002). The untreated seeds had a mean number of seedlings of 39.8seedlings/m² compared to the 80.8seedlings/m² recorded in the capsaicin treated plots (Figure 80). The phlox showed no difference between the capsaicin treatments (p>0.05).

            In May of 2007, seedlings from the spring 2006 and fall 2006 planting were counted again. Capsaicin treated Dodecatheon meadia plots had significantly (p=0.005) more seedlings than the control. The capsaicin treated plots averaged 833.8 seedlings/m², compared to 274.7 seedlings/m² for the untreated control (Figure 80). Planting time had no affect on the germination of Dodecatheon meadia. Spring planted plots averaged 556.4 seedlings/m² and fall planted plots averaged 552.0 seedlings/m². Planting time did significantly (p<0.0001) affect the germination of Phlox pilosa. Spring planted plots averaged 113.6 seedlings/m², whereas fall planted plots averaged 4.09 seedlings/m² (Figure 80). Treating the seeds with capsaicin had no effect on seed germination. Capsaicin treated plots averaged 68.2 seedlings/m² compared to control plots averaging 49.51 seedlings/m².

Discussion/Conclusions

Part I - Reduction of Granivory in a Prairie Reconstruction

            The scientific literature has shown rodents can be responsible for up to 90% seed loss, but very little has been done to document these losses in prairie reconstructions. Our experiments have confirmed this as we documented similar seed losses over just a three week period. If this amount of seed loss is consistent with a restoration effort, much of seed we broadcast is being used as mouse and insect food, and not increasing the floral diversity or forb density.   Seed predation is affected by species of seed, site, and time of year.

            Capsaicin looks like it has potential as a feeding deterrent for certain species. Capsaicin worked for deterring granivory on shooting star (see Part II below) and also whole sunflower kernels, but not rosinweed or Phlox. At the cost of $0.26/lb of seed capsaicin may still be a cheap insurance to slow down seed predation.  A weakness of capsaicin is that it is quickly washed off of seeds by rain, based on our simulated rain experiments. It would be valuable to investigate more lasting capsaicin coatings, or other possible feeding deterrents such as Thiram.

            Broadcasting sunflower seeds with prairie seeds (“sacrifice seed”) reduced seed predation in our spring experiment, but had no effect in the fall trial.  In  spring 2006 there were 20% more seeds remaining in the extra food plots compared to the control plots. This protection may provide an extended window for the prairie seeds to become buried in the soil and out of the way of seed predators. This treatment cost about $20/acre for the extra sunflower seeds. Although a little more expensive than the capsaicin powder, the extra food did clearly provide protection against granivory, at least in the spring trial.  We can speculate that in the Fall, mice populations are high and it would take much more extra sunflower seeds to saturate these populations.  They also may be caching seed rather than eating it immediately, leading to complete disappearance of all broadcast seeds.

            Invertebrates appear to play a crucial role in seed loss. Due to their size we believe they will focus on smaller seeded species, but nonetheless are still capable of removing a large proportion (perhaps 25%) of seeds. Both invertebrates and vertebrates appear to be highly selective in their choice of seeds to consume. This may partially explain why some species are much easier to get established than others if their seeds are distasteful to granivores. 

            Finally, we discovered that another 50% or so of newly emerged seedlings will be lost throughout the summer due to seedling predation.  In a pilot study of caged versus uncaged seedlings, UNI undergraduate Ryan Boswell found significant losses in the uncaged control plots.  Camera traps revealed that voles were consuming these seedlings.

            This study produced evidence that seed and seedling predation are significant, affecting total number of seedlings the following year.  Seed predation is affected by site, season of planting, and species of seed, and can be reduced in some cases.   We are aware of no other studies in the Midwest to investigate seed and seedling predation in this detail the context of prairie reconstruction. 

Part II - Addition of Conservative, Showy, and Expensive Seeds

            There are a lot of prairie species about whose life histories and germination requirements are poorly known. Some species may respond very differently to different planting times and stratification.  Shooting star and prairie phlox are early to mid-spring flowering species that are notoriously difficult to establish in prairie reconstructions.  Shooting Star appeared to establish slightly better if planted in early spring (March) while Phlox established better if planted in May (Figure 7).  Prairie phlox apparently requires both a summer and winter in the soil in order to germinate.  Seeds planted in May, 2006 were much more likely to germinate in spring 2007 than those planted in November 2006 (Figure 8).

            It is common to stratify seeds for spring planting, to simulate a winter dormancy requirement of many species.  Cold moist stratification can have major, positive impacts on germination success. However, we observed that the phlox treated with a cold-wet stratification resulted in only a few seedlings compared to hundreds in the unstratified treatment (Figure 7). Like the planting time, effectiveness of cold-moist stratification appears to be highly species specific. 

            Capsaicin was highly successful in protecting shooting star seeds from predation, nearly tripling the number of seedlings in treated plots (Figure 8).  Both Spring and Fall plantings of capsacin showed this effect.  Neither planting was accompanied by significant rainfall, probably preserving the efficacy of capsaicin for a longer period of time than was found in other  capsaicin experiments.

            Based on our experiments, we feel confident in recommending that for maximum establishment of Phlox pilosa, it is important to plant in mid to late spring, without previous stratification, and to use a sacrifice food source or partial seed burial to help reduce seed predation.  Delayed germination of a year should be expected; therefore, competition should be limited in the reconstruction by mowing.  For shooting star, we highly recommend use of capsaicin powder at a time when rain is not forecast for several days.  Early planting time (March) may be preferable to May   Seedlings remain very small and are green for only part of the growing season; therefore it is important to reduce competition for light for at least three seasons.

Figure 1: Seed card in the field. Thirty Silphium integrifolium seeds glued to sandpaper and held to the ground with two roofing nails in the corner. Note the metal tag identifying the individual card’s number.
 

Figure 2: Seed card used in the Vertebrate exclosure experiment. Four inch, plastic Petri dish with sandpaper attached inside. Three species used from the top going clockwise are: Grey headed Coneflower, Indian grass, and Purple Prairie Clover.
 

Figure 3: Influence of sacrifice food and capsaicin on disappearance of Silphium integrifolium seed in June 2006.
 

Figure 4: Mean and standard error of percentage of seeds remaining on seed cards in Fall 2006. There were three species and three seed treatments. Individual block means are not shown.
 

Figure 5: Correlation between seed card predation in fall with seedling emergence (seedlings emerged per 100 pure live seeds planted) in the subsequent spring. There was a significant correlation in D. meadia (r² = 0.59, p < 0.01) but neither of the other two species was significant.
 

Figure 6: Affect of simulated rainfall survival of sunflower kernels after two days of exposure.
 

Figure 7: Emergence of prairie phlox and shooting star seedlings in June 2006 from 2005 plantings in March and May. In half of replicate plots, seeds were cold-moist stratified prior to planting.
 

Figure 8: Emergence of prairie phlox and shooting star seedling planted in spring and fall 2006, and counted in spring 2007. In half of replicate plots, seeds were treated with capsaicin.