Full text

White-tailed deer (Odocoileus virginianus) and other large herbivores have a substantial influence on vegetation diversity (Alverson et al. 1988; McNaughton et al. 1988; Pastor et al. 1988; Anderson 1994a; Collins et al. 1998; Knapp et al. 1999). However, most studies considered the effects of deer browsing on forest trees or understory vegetation (Beals et al. 1960; Anderson and Loucks 1979; Frelich and Lorimer 1985; Alverson et al. 1988; Anderson and Katz 1993; Anderson 1994a, 1997; Balgooyen and Walter 1995; Rooney 1997; Waller and Alverson 1997; Augustine and Frelich 1998; Williams et al. 1999; Webster and Parker 2000). In contrast, there is relatively little published literature on effects of deer browsing on prairie plants (England and Meyer 1986; Ritchie and Tilman 1995).

Several studies have demonstrated that there is a continuum of plant response to herbivory that is affected by availability of resources (e.g., levels of inorganic nutrients and irradiance), competition, life span (e.g., annuals, biennials, woody perennials), seasonality of browsing (e.g., early growing season vs. late growing season), plant growth form and other factors. Generally, plants occurring in environments with abundant resources with high growth rates and large underground storage capabilities have the greatest capacity for compensatory response following herbivory (Whitham et al. 1991). Because prairie plants receive higher levels of photosynthetically-active radiation than forest understory plants and have a large allocation of net production to belowground structures (Risser et al. 1981), it might be expected that prairie plants may have a greater capacity than forest understory species for regrowth or overcompensation following defoliation (Maschinski and Whitham 1989; Whitham et al. 1991).

Englund and Meyer (1986) reported that in a newly planted prairie restoration site, mortality of forbs protected from deer browse was not significantly lower than among unprotected plants. However, unprotected plants were smaller and had fewer flowering heads. Ritchie and Tilman (1995) found that protection from browsing increased abundances of legumes as a group, but this was primarily due to the increased abundance of Lathyrus venosus. Two other species of legumes studied, Lespedza capitata and Amorpha canescens, were not significantly affected by herbivore exclosure (Ritchie and Tilman 1995).

Despite the dearth of published literature on the influence of deer browsing on prairie plants, the issue is of concern to resource managers. The purpose of this study was to monitor and characterize patterns of browsing on prairie species under conditions of unusually high deer densities. Specifically, we sought to determine: (1) which species of prairie plants were browsed by deer and if they were selectively browsed, (2) the intensity of browsing, (3) seasonal browsing patterns, (4) the influence of browsing on reproduction of prairie plants, and (5) indicator species that could be used to characterize browsing intensity in prairies.

Methods. STUDY AREA. Goose Lake Prairie State Park (GLPSP) is located in northeastern Illinois (Grundy County) about 70 km southwest of Chicago. The State Park encompasses 1,027 ha and includes an Illinois Nature Preserve (612 ha) of which approximately 21% is mesic tallgrass prairie. The majority of the area (51%) is mixed grass (Calamagrostis candensis, Spartina pectinata) and sedge (Carex stricta) meadows (Nelson and Anderson 1983). Preliminary field observations on this site conducted by Illinois Department of Natural Resources (IDNR) personnel in the spring of 1992 indicated that 95% of all prairie forb species were browsed. The State Park has low forb diversity that resulted from agricultural activities, especially on the mesic prairie sites, prior to its acquisition by the state of Illinois.

Deer hunting was restricted on the State Park until 1997. However, hunting occurred on the adjacent 218 ha Heidecke Lake Fish and Wildlife Area (HLFWA) between 1983 and 1992 with an average of 29 deer being removed from the site annually. Aerial deer censuses of the GLPSP and HLFWA complex, which are not separated by barriers to deer movement, were conducted annually between 1991 and 1994. Censuses were made from a helicopter when snow cover was on the ground. Deer densities ranged from 30 to 42 deer km^sup -2^. These deer densities were likely above those associated with historic conditions in the Midwest (4 to 19 deer km^sup -2^) (Dahlberg and Guettinger 1954; McCabe and McCabe 1984) and might be sufficiently high to cause negative effects on the prairie community (Alverson et al. 1988; Tilghman 1989; Anderson 1994a, Healy 1997).

Even though the Nature Preserve is 21% mesic prairie all of it was subjected to grazing and has low diversity and abundance of forbs. It would have been impractical to establish enough deer exclosures to adequately document patterns of deer browsing. Under these conditions, the deer may browse species irrespective of preferences, because there is little choice available. Consequently, in May 1992, we chose to establish a 33 m X 48 rn study area in mesic tallgrass prairie adjacent to the west side of GLPSP in the HLFWA, because of its high diversity and abundance of prairie forbs. The one hundred species of forbs noted in this area gave us the opportunity to determine the selective patterns of deer browsing when a large number of species were available. The experimental design was a complete block design with two replicate plots for each experimental condition. The experimental area encompassed nearly all the high quality remnant. Vegetation on the remnant study sites is similar in floristic composition to the vegetation that would be expected to occur on mesic sites on Goose Lake Prairie in the absence of disturbance (Nelson and Anderson 1983; Anderson et al. 1984). The HLFWA site was burned in the spring of 1992 but not during the rest of the study.

We divided the study area into four 24 m X 16.5 rn quadrants (quarters). Three parallel transect lines extended the full length of each quadrant. Along each transect line, we located 12 points beginning 1 meter from the starting point and then at two meter intervals. From each point, we randomly located the center of a 25 cm X 25 cm quadrat to sample the vegetation.

In December 1992, IDNR personnel fenced two of the above quadrants to exclude deer but not smaller animals, in order to compare the influence of all herbivores on prairie forbs with herbivory on prairie forbs without deer browsing. We chose quadrants to exclose or leave open (reference plots) to minimize differences in species abundance and composition between the exclosed and reference (unexclosed) quadrants based on quantitative data on prairie forbs obtained during summer 1992. The percent similarity (([2 * (Sigma)shared abundance of species between samples]/[Sigma species abundance in the two samples]) * 100) (Gauch 1983) between the two areas based on forb species composition and abundance was 83%. This is a high similarity value as two random samples of the same area usually yield similarity values of about 82-85% (Bray and Curtis 1957; Beals 1960). Based on this index of similarity the exclosed and reference areas were quite similar-almost identical-before the changes that followed the experimental treatment. We sampled reference and exclosure areas in 1993 and 1994.

We frequently observed groups of 2-5 deer as well as single animals moving through the study area. It is likely that groups of deer at least this large used the areas during the entire period described here, but we did not specifically estimate number of deer using this site. A corn field (about 10 ha) located 40 m south of the study area likely encouraged deer activity near the site. Moreover, Nixon et al. (1991) reported that deer are territorial in central Illinois during the summer, but home ranges of mothers and daughters overlap in the summer when mothers and yearlings share about 63% of their home ranges. During the period of parturition, the overlap in home ranges between yearlings and mothers was nearly 100 percent (Nixon et al. 1991).

QUANTIFYING BROWSING ON FORBS AND GRAMINOIDS. In each quadrat, we counted the number of browseable units. Browseable units varied depending upon the morphology of the plant and the quantifiable portion of the plant deer consumed. For example, violet leaves occur in loose basal rosettes and deer often browse individual leaves so these were counted. The upright stems and attached leaves of species such as sunflowers (Helianthus spp.), leadplant (Amorpha canescens), and culver's root (Veronicastrum virginicum) are browsed units so stems and attached leaves were counted. Other species such as rattlesnake master (Eryngium yuccifolium) and wild quinine (Parthenium integrifolium) have tight clusters of basal leaves that often are browsed as a unit and we counted the clusters. A listing of how "browseable unit" was defined for specific species is available on the senior author's web page (http://www.bio.ilstu. edu/BEES/anders.htm). Each browseable unit (hereafter referred to as stem) was placed into one of five categories (1) recently deerbrowsed, (2) non-deer foraged, (3) damaged, (4) old deer browsed, and (5) unbrowsed. Deer browsed stems were characterized by tearing, because deer lack upper incisors and therefore tear plants during feeding. Deer often consume the whole forb stem or a large portion of it as they browse. In contrast, rodents and rabbits (Sylvilagus floridanus) have upper and lower incisors and leave a sharply severed stem as they browse, and they consume a small portion of each plant they browse. Consumption of forbs by rabbits, rodents, insects, and by a nesting mallard duck (Anas platyrhnchos) in one exclosed plot, was included among non-deer foraging. Damaged stems included those that were mechanically injured due to trampling, and stems for which the source of the injury was unknown. Old deer browse characteristically had the portion of the stem below the cut point appearing dried, and, for some species, regrowth of the stem indicated that the stem was browsed some time in the past.

Because of the clumped growth form of graminoids (grasses and grass-like plants [sedges]) with numerous leaves arising from a single clump, it was not practical to count leaves. Consequently, we recorded the presence of grasses by species and sedges by genus in the quadrats and noted if plants were browsed or unbrowsed. However, a single browsed leaf of a species in a quadrat would be scored the same as having all of the leaves in the quadrat browsed. To increase the ability to discriminate variation in browsing among species, we divided the 25 cm x 25 cm quadrats into quarters (quadrants) and sampled grasses and sedges as described above in each quadrant. This sampling occurred during July 1993 and effectively increased the number of quadrats sampled four-fold (from 144 to 576). From these data, we determined the frequency of browsed and unbrowsed graminoids.

For each sampling date in 1992, we analyzed forb count data using the Pearson chi-square test to determine if the proportion of recently deer browsed stems differed significantly among the species in the sample as a whole. If this statistic was significant (p < 0.05), then we tested each species separately using Pearson chi-square with a continuity correction to determine if the proportion of stems of a species browsed differed significantly from the proportion of stems browsed in the population as a whole. Species whose stems were browsed in proportions significantly greater than, less than, or not different from the population as a whole were termed selected, low use, and proportional use species, respectively. For each species, we calculated percent browsed, relative use, and relative abundance for each sampling date using the formulae given below. In all of these calculations, browsed stems were recently deer browsed stems and total stems include all five categories of stems previously described.

1. % Browsed = number of browsed stems of a species/total number of stems of a species x 100, 2. Relative Use = number of browsed stems of a species/number of browsed stems for all species x 100, and

3. Relative Abundance = total number stems of a species/total number of stems of all species x 100.

Data collected in 1993 and 1994 were analyzed in a similar manner to the 1992 data, but because only two of the four quadrants now excluded deer, we compared browsing patterns between non-deer foraging in the exclosed and reference plots and deer browsing in the reference plots. Relationships between foraging patterns of deer and non-deer foragers in exclosed and reference areas were explored using linear regression.

QUANTIFYING EFFECTS OF DEER ON FORB REPRODUCTION. For species with high abundances and in flower at the time of sampling, we counted flowering stems (stems with fruits, flowers, and/or buds) that occurred within a meter on either side of the three 24 m transect lines established per study plot. These data were obtained for all sampling dates in 1993 and 1994. We tested for differences in mean number of flowering stems per transect by species between reference and exclosed plots using t-tests. Alpha was corrected for multiple tests using a Bonferonni correction (014) so that statistical significance was accepted was accepted at ot < 0.0125 (Sokal and Rohlf 1995). For the species used, the total stem counts in the 25 X 25 cm quadrats for reference and exclosed plots, respectively, were about the same in 1992 prior to exclosure: Phlox pilosa (33 vs. 31), Helianthus mollis (13 vs. 13), Veronicastrum virginicum (49 vs. 67).

The growth form of Baptisia leucantha, a single annual bushy stem with a terminal flowering stalk, makes it highly susceptible to damage and reduced reproductive output if deer browse it. Deer heavily browsed Baptisia. However, it had a low density in our study area and was rarely included in a sample. To document the effect of deer browsing on the reproductive effort of this species, all B. leucantha (29) occurring in our study area during July 1992 were categorized as being browsed or unbrowsed. The number of flower buds, flowers, and fruits was counted for each plant. Reproductive data for Baptisia were analyzed using a Kruskal-Wallis test to determine if browsed and unbrowsed plants differed in numbers of total reproductive units (buds, flowers, and fruits) per plant.

In 1992, IDNR personnel established an additional deer exclosure (12.1 m X 11.5 m) on Goose Lake Prairie State Park and they burned the site in spring 1993. This burn appeared to stimulate the flowering of Amorpha canescens. To document the influence of deer browsing on this species, in July 1993, we sampled the density of flowering stems, and browsed and unbrowsed stems of Amorpha canescens using m2 quadrats. Using a stratified random sampling procedure, 20 and 48 ml quadrats were used to sample the exclosure and an adjacent reference area (13 m X 24 m), respectively.

There were no flowering Amorpha stems outside the exclosure or browsed stems inside the exclosure. Only differences in total stem counts (browsed plus unbrowsed stems) were compared statistically between exclosed and reference areas using the non-parametric Kruskal-Wallis test, because the data were not normally distributed and could not be transformed to achieve normality.

Except where noted otherwise, we accepted statistical significance at alpha < 0.05.

Results. BROWSING ON PRAIRIE FORBS AND GRASSES. The percentage of sampled forb species that were browsed varied among sampling periods from 30.2% to 67.4% (Table 1). The higher total stem counts in 1992 compared to counts in the two subsequent years resulted from all plots being unprotected in 1992, while only two of the four plots were available for deer browsing in 1993 and 1994. The number of species sampled was relatively constant for the 1992 and 1993 samples, ranging from 42 to 46. However, in 1994 the numbers of species sampled in July (37) and August (31) were lower than in the previous two years. The number of stems sampled in these two months also declined. Browsing intensity varied seasonally, with the browsing intensity increasing from May through mid-to-late June and then decreasing from late-July into August (Table 1).

In May 1992, six species were browsed at a significantly greater proportion than the population of forbs as a whole (Table 2). The percentage of all stems browsed increased from 10.9% in May to 18.9% in July (Table 1). Seven species were browsed at a higher proportion than the population as a whole for the July sample. During August, seven species were browsed in a greater proportion than the population as a whole, but the percentage of all stems browsed decreased to 3.5% (Table 1, Table 2). One species (Pycnanthemum flexuosum) was selected in all three months, and only two species (Eryngium yuccifolium, and Stachys palustris) were selected in two months. All other species were selected only in a single month. Eryngium, unlike the other species, tended to have only the tips of its leaves browsed early in the growing season. Apparently, the high fiber content in the mature foliage tends to limit browsing to the tips of growing leaves. However, occasionally the above-ground portion of Eryngium was completely consumed, and browsing also occurred on flowering stems in July. Four forbs (Aster azureus, Desmodium illinoense, Euphorbia corollata, and Solidago rigida) had 100% of their stems browsed. However, all of these species occurred in low relative abundance, ranging from 0.1 to 0.5% (Table 2).

Table 1.

Table 2.

Proportional use species usually had higher relative abundances, but lower percentages of their stems browsed than selected species. However, because of high stem abundances, some proportional use species had higher relative use than selected species that had higher percentages of stems browsed but fewer total stems. For example, in July Parthenium integrifolium and Silphium integrifolium accounted for 39% of the browsed stems, but the two species also contributed 33.3% of the total stems sampled. Low use species always had higher relative abundances than relative use values, but the reverse was the case for selected species that always had higher relative use than relative abundance.

Comparison of selected species across all sampling dates indicated that species selected by deer vary yearly as well as seasonally (Table 3). Of 100 forbs sampled in this study, deer selected 26 in at least one of nine sampling periods. The number of species selected varied from 5 to 7 per sampling period, except for July and August 1994 when only a single species was selected in each month.

Amorpha canescens, Stachys palustris, and Veronicastrum virginicum were the most frequently selected species, each being selected in 4 of 9 samples. Pycnanthemum flexuosum and Helianthus mollis were selected in 3 samples, five species were selected twice, and 17 species were selected once.

Table 3.

Table 4.

Deer and other foragers tended to select different browse species (Table 4). In 1993 and 1994, only three of 15 species selected by deer (Parthenium integrifolium, Rudbeckia subtomentosa, and Silphium terebinthinaceum) were selected by other foragers. Deer also tended to select more species-generally 5 to 6 per sample date-compared to 1 to 3 per sample selected by other foragers (Table 4). Commandra umbellata was heavily browsed in the exclosure and reference areas. We think that other foragers were largely responsible for this browsing, although it was difficult to determine if deer or other foragers browsed the rather wire-like stem of Commandra. The abundance of Commandra declined from over two hundred stems sampled on the study plots in May 1992 (237 stems) and in June 1993 (247 stems) to 57 stems in June 1994. A seasonal decline in Commandra stem counts also occurred. For example, samples taken in July and August 1993 had 109 stems and 81 stems for the two months, respectively, compared to 247 in June of the same year (Anderson 1994b).

Conclusions. Deer browsed selectively on prairie forbs but not on prairie grasses or sedges. Forb species selected by deer tended to have low relative abundances compared to low and proportional use species, but the selected species had a higher percentage of their stems browsed. Species selected by deer varied somewhat seasonally and yearly. Deer selected 26% of the 100 species of forbs sampled in at least 1 of 9 samples obtained over 3 growing seasons. However, a majority of the species (17 of the 26 selected species) was selected in only 1 sample. Highest browsing intensity occurred during late June and early July. The data suggest niche separation between deer and other foragers in the species they select for browsing.

Historically there may have been foraging niche separation between bison, which fed primarily on grasses (Vinton et al. 1993; Steuter 1997; Collins et al. 1998; Knapp et al. 1999), and deer (McCullough 1980), which selectively browse forbs and consume little, if any, prairie graminoids. Under conditions of high deer density, plant diversity of prairies may decline as forbs are placed at a competitive disadvantage with grasses.