237

Chapter 11 Responses of a Tallgrass Prairie (Araneae) Community to Various Burn Seasons and Its Importance to Tallgrass Prairie Management

David J. Wade and Robert E. Roughley1 Department of Entomology, University of Manitoba Winnipeg, Manitoba, Canada R3T 2N2

Abstract. A four-year study was conducted in which were used as bioindicators to determine if there was an optimal season to burn tallgrass prairie. The burn seasons examined were spring, summer, and fall. Overall, the species diversity of the spider community in the burned areas was not significantly different from the unburned areas for any burn season, although some significant differences occurred within the first year following the burns. Spider abundance decreased following each burn season treatment and the recovery period varied depending on the burn season. Burn season also affected the structure of the spider communities in that the summer and fall burn treatments became dominated by distincta (Blackwall) rather than the initially dominant P. moesta Banks. The use of a mosaic of burn season treatments is recommended to maximize the diversity of spiders within the tallgrass prairie habitat.

Résumé. Nous avons réalisé une étude de quatre ans afin de déterminer s’il existe une saison optimale pour les brûlis en prairies à herbes hautes en utilisant des araignées en guise de bioindicateurs. Les brûlis ont été effectués au printemps, en été ou en automne. Dans l’ensemble, la diversité des espèces d’araignées mesurée dans les zones brûlées n’était pas significativement différente de celle mesurée dans les zones non brûlées, sans égard à la saison choisie pour les brûlis, même si certaines différences significatives s’observaient au cours de la première année suivant les brûlis. L’abondance des araignées a diminué après chaque saison de brûlis, et la période de rétablissement a varié en fonction de la saison choisie. Le choix de la saison des brûlis a également influé sur la structure des communautés d’arachnides : dans les zones soumises à des brûlis d’été ou d’automne, le Pardosa distincta (Blackwall) est devenu dominant et a remplacé l’espèce qui dominait auparavant, le P. moesta Banks. On recommande de recourir à une mosaïque de traitements saisonniers afin de maximiser la diversité des araignées dans l’habitat de la prairie à herbes hautes.

Introduction Approximately 300 years ago, tallgrass prairie covered 570,000 km2 of North America (Howe 1994), extending from southern Saskatchewan and Manitoba, south to Texas, and eastward to Iowa (Robertson et al. 1997). Only 1% of this habitat remains in North America and less than 1% in Manitoba (Morgan 1994; Robertson et al. 1997; Nature Conservancy of Canada 2000). The main tallgrass prairie sites in Manitoba are the Tall Grass Prairie Preserve (1,820 ha), the St. Charles Rifle Range (SCRR; 192 ha), and the Living Prairie Museum (12 ha). Tallgrass prairie is adapted to disturbance, and active management requires regular disturbances such as fire, grazing, or mowing (Howe 1994; Schwartz and Hermann 1997). Prior to European settlement, tallgrass prairie was regularly burned by lightning strikes in

1 Deceased.

Wade, D. J. and R. E. Roughley. 2010. Responses of a Tallgrass Prairie Spider (Araneae) Community to Various Burn Seasons and Its Importance to Tallgrass Prairie Management. In of Canadian Grasslands (Volume 1): Ecology and Interactions in Grassland Habitats. Edited by J. D. Shorthouse and K. D. Floate. Biological Survey of Canada. pp. 237-249. © 2010 Biological Survey of Canada. ISBN 978-0-9689321-4-8 doi:10.3752/9780968932148.ch11 238 D. J. Wade and R. E. Roughley

late summer or early autumn and by Aboriginal people purposely igniting the habitat in the early spring or autumn (Warren et al. 1987; Collins 1990; Howe 1994; Schwartz and Hermann 1997). Fire is currently the most widely used management technique, with spring burns being most common (Schwartz and Hermann 1997; Collins and Steinauer 1998). Spring burns favour late season grasses such as big bluestem and other key tallgrass prairie grass species (Schwartz and Hermann 1997), although summer burns may maximize plant diversity (Howe 1994). Without regular disturbances, the plant community becomes dominated by few species that reduce species richness and productivity (Reed 1997; Schwartz and Hermann 1997; Collins and Steinauer 1998). However, the disturbance regime necessary to maintain high diversity and productivity has not yet been determined (Howe 1994; Collins and Steinauer 1998). The effects of fire on the spider (Araneae) fauna of prairie habitats are not well-known (see reviews by Warren et al. 1987; Reed 1997; Bell et al. 2001). Spiders are important predators and can survive the physical effects of fire by finding protection in places such as cracks in the soil (Warren et al. 1987). However, because of the upper lethal temperature of most spiders, biologists hypothesize that the majority of spiders are not able to survive a fire incident and that they recolonize following the burn (Bell et al. 2001). Spider survival after a fire also depends on the availability of prey, both its density and diversity, in the burned area (Warren et al. 1987). The pioneer spider fauna in grasslands following fire includes certain species of linyphiids, theridiids, and lycosids (Riechert and Reeder 1972; Bell et al. 2001). These species prefer bare ground, are more tolerant to microclimate changes, and are relatively less dependent on vegetation for web construction (especially lycosids) (Bell et al. 2001). These pioneering species often decline in numbers over time as the habitat recovers from the effects of fire and as old-growth species increase in abundance (Bellet al. 2001). In general, species diversity increases over time following fire in grasslands, but the short-term responses can vary (Bell et al. 2001). Bell et al. (2001) recommended that more studies be done to examine the effect of different burn regimes on spider communities. They also recommended that burning be conducted on large connected habitats with a rotation of regimes to conserve the highest species richness and range of stand ages. Maintaining refugia and keeping the inter-burn period long enough are also important so that the spider fauna can recover (Harper et al. 2000). Although no studies have been done of the impact of fire on spiders in tallgrass prairie in Manitoba, some have been conducted in similar tallgrass prairie habitats in Illinois (Rice 1932; Harper et al. 2000), Wisconsin (Riechert and Reeder 1972), and Kansas (Nagel 1973). Harper et al. (2000) found that spider abundance significantly decreased in the 10 weeks following a spring burn. Spider abundance was more negatively affected in the enclosure study sites (i.e., closed system), suggesting that recolonization from unburned areas is important for spiders. Rice (1932) found that spider numbers were lower in burned versus unburned subclimax tallgrass prairie following a spring burn. Riechert and Reeder (1972) saw the same trend on two separate prairie plots. On a subclimax prairie site, spider abundance recovered within a week and species composition favoured vagrant species. However, on a climax site, spider abundance did not recover after 45 days, albeit species composition was unaffected. These researchers also observed that spiders moved from burned areas to unburned areas. The long-term effect of burning on the spider community was inconclusive in their study. In the year following the burn, spider abundance on the climax prairie site had returned to pre-burn levels, whereas levels at the subclimax prairie site remained below pre-burn levels. Similarly, Nagel (1973) found that spider abundance was lower in the burned versus unburned prairie following a spring burn on a prairie in Response of spiders to fire in tallgrass prairie 239

Kansas. Johnson (1995) found that spider abundance and density was higher in annually burned Spartina pectinata Link wetlands, but species composition was similar. This increase in spider abundance was correlated with increases in insect prey. Howe (1994) predicted that fire season would affect the resultant tallgrass plant community from observations that species (1) differ in response to thatch removal and soil warming, (2) have different seeding phenologies and germination requirements, and (3) differ in patterns of rhizome recruitment. His prediction was validated for the tallgrass prairie plant community on the SCRR (Sveinson 2001). By extension, fire season also is predicted to affect spiders. Spider species have different phenologies and peak activity periods (Aitchison 1984) and differentially migrate into burned areas (Riechert and Reeder 1972; Harper et al. 2000). The objectives of this study were to determine (1) the effect of burn season on the spider community, (2) the optimal burn cycle interval, and (3) the optimal burn season for tallgrass prairie management. This research was part of a multidisciplinary study that also used the plant and ground beetle (Coleoptera: Carabidae) communities as bioindicators to evaluate the effectiveness of burn season as a tool for tallgrass prairie management. The results for these other bioindicators are discussed in Sveinson (2001), Roughley (2001), and Roughley et al. (see Chapter 10). A secondary component of the study was to determine the uniqueness of the tallgrass prairie spider and ground beetle fauna compared with the adjacent aspen forest fauna, the results of which are discussed in Roughley et al. (2006).

Study Area The study was conducted at the SCRR, located just to the northwest of Winnipeg, Manitoba (49°54′35.1″ N, 97°20′47.3″ W). The 192 ha property is owned by the Canadian Department of National Defence and includes 47.9 ha of high-quality tallgrass prairie (Morgan 1994). The area has remained undisturbed for at least 100 years, but a portion of the study area may have been cultivated until 50 years ago. This area was coined the “go-back” prairie by Morgan (1994). The plant and ground beetle diversity of the SCRR is well-known, with over 112 plant species and 103 ground beetle species being present (Sveinson 2001; see Chapter 10).

Experimental Design To test for an effect of burn season, we established four replicate blocks (A, B, C, D) of five plots (50 × 50 m per plot) at SCRR (Fig. 1). Each block contained three burn treatment or burn season plots (spring, summer, and fall) and a “standard” plot that abutted a central, unburned “refuge” plot. The standard plot was used only for botanical analyses (Sveinson 2001). Treatment burns were applied only in 1997, on 6 June (spring burn), 5 August (summer burn), and 9 September (fall burn). Unburned areas within treatment plots after the passage of fire were left intact. Placement of the blocks allowed for maximum coverage of the study area, which included placing one block (block D) in the go-back prairie. In 1998, two control plots (X and Y, Fig. 1) were incorporated into the experimental design to ensure that the refuge treatment was not trapped out. The sampling protocol of these two controls was the same as the other blocks except that sampling did not begin until 1 May 1998. Pitfall traps (n = 16) in a 10 m grid were operated in each plot (Fig. 1). The traps were plastic yellow containers with an outside top diameter of 11.5 cm and a depth of 6.5 cm. They were placed into the ground so that the lip of the container was flush with the surface 240 D. J. Wade and R. E. Roughley

Fig. 1. Schematic diagram of experimental design for the St. Charles Rifle Range tallgrass prairie project. spr = spring treatment; sum = summer treatment; fall = fall treatment; ref = refuge treatment; con = control treatment. Gray squares represent the botanical standard (modified from Roughley 2001).

of the ground. A rain cover consisting of a square piece of plywood with nails as supports was placed over each trap to prevent flooding. Traps were half filled with a saturated salt solution with a drop of dish detergent. Traps were emptied weekly and the contents of the traps were passed through a strainer with a mesh size of 1 mm. Collections within plots were pooled across traps each week and stored in 70% ethanol. Traps were operated in 1997 from 28 May to 7 November, in 1998 from 3 April to 10 November, in 1999 from 20 April to 12 November, and in 2000 from 31 March to 10 November. Adult spiders recovered from traps were identified to species by the senior author from the catalogue by Platnick (2008). Because species identification was not always possible, juvenile spiders were excluded from data analyses. Voucher specimens are stored in the J.B. Wallis Museum at the University of Manitoba.

Data Analysis The data were analyzed by using the software programs SYSTAT and Bio-DAP. The analyses were based on the yearly total per treatment square for each species. Various measures were used to examine alpha diversity, including species richness, Shannon- Wiener index, log series alpha index, Simpson index, Berger-Parker index, and log series alpha evenness. These indices were chosen so that the data could be examined thoroughly with a combination of diversity- and dominance-based measures. Beta diversity was Response of spiders to fire in tallgrass prairie 241

examined by using the Jaccard coefficient and Morisita-Horn index. These two measures were chosen to provide a balanced examination of community similarity with a qualitative (Jaccard) and quantitative (Morisita-Horn) measure. The values from these indices and the abundance data were analyzed by means of the general linear model analysis of variance function in SYSTAT (version 9). Multiple pairwise comparisons were done by the Tukey HSD method, with a significance level of 0.05, focusing on within-year comparisons. We analyzed the short-term effects of burn season on spider abundance by breaking the sampling weeks of 1997 into four burn periods. Period 1 (N1) was all weeks prior to the spring burn, period 2 (N2) was the weeks between the spring and summer burns, period 3 (N3) was the weeks between the summer and fall burns, and period 4 (N4) was the weeks after the fall burn. Between-year comparisons were not analyzed statistically because there were too many year-to-year differences in sampling intensity that would have made data standardization difficult and statistical differences hard to interpret. These year-to-year differences included the number of sampling weeks per year, an apparent sorting bias, and varying weather conditions. The apparent sorting bias and varying weather conditions were assumed to be equal among treatments and blocks for each year.

Spider Abundance A total of 54,396 adult spiders were collected over the duration of the study, representing 126 species in 17 families. Adult spider abundances were different among treatments in 1997 (P = 0.001, df = 3, F = 13.288), 1998 (P = 0.001, df = 4, F = 26.133), 1999 (P = 0.012, df = 4, F = 5.965), and 2000 (P = 0.030, df = 4, F = 4.735) (Table 1). The refuge treatment had the highest abundance in all four years and the fall treatment had the lowest abundance from 1998 to 2000. The abundance levels increased steadily in the spring treatment from 1997 to 2000. The summer and fall treatments had their lowest abundances in 1998 and increased steadily from 1998 to 2000. When 1997 was broken down by burn period, the abundance tended to decrease for each burn season treatment following its burn time (Table 2). These decreases were significant compared with the refuge following the spring and summer burns but not for that following the fall burn.

Table 1. The effect of burn season on the natural log abundance (mean ± SD) of adult spiders collected on the St. Charles Rifle Range. Values with the same letter within a given year are not significantly different at the 5% confidence level based on Tukey-HSD comparisons.

Treatment Year 1997 1998 1999 2000 Spring 5.96 ± 0.33 a 6.66 ± 0.38 a 6.71 ± 0.34 a 6.96 ± 0.47 a,b Summer 6.23 ± 0.16 a 6.06 ± 0.18 b 6.48 ± 0.26 b 6.87 ± 0.32 a,b Fall 6.23 ± 0.16 a 5.96 ± 0.21 b 6.39 ± 0.24 b 6.72 ± 0.27 a Refuge 6.55 ± 0.20 b 7.04 ± 0.34 a 6.87 ± 0.21 a 7.05 ± 0.28 b Control * 6.44 ± 0.19 a,b 6.73 ± 0.04 a,b 7.09 ± 0.17 a,b

* not applicable (traps not installed until 1998). 242 D. J. Wade and R. E. Roughley

Table 2. The effect of burn season on the natural log abundance (mean ± SD) of adult spiders collected by period* in 1997 on the St. Charles Rifle Range. Values with the same letter within a given period are not significantly different at the 5% confidence level based on Tukey-HSD comparisons.

Period* Treatment N1 N2 N3 N4 Spring 3.70 ± 0.58 a 5.49 ± 0.23 a 3.94 ± 0.33 a,b 3.92 ± 0.42 a Summer 3.52 ± 0.43 a 5.98 ± 0.20 b,c 3.37 ± 0.52 a 3.50 ± 0.43 a Fall 3.51 ± 0.29 a 5.87 ± 0.14 b 4.27 ± 0.23 a,b 3.57 ± 0.77 a Refuge 3.42 ± 0.46 a 6.18 ± 0.18 c 4.48 ± 0.16 b 4.46 ± 0.33 a

* N1 = pre-spring burn period; N2 = spring burn to summer burn period; N3 = summer burn to fall burn period; N4 = post-fall burn period.

Pardosa moesta Banks and P. distincta (Blackwall) constituted 31.4 and 21.7%, respectively, of the adult spiders caught (Table 3). Other abundant species included Alopecosa aculeata (Clerck) (6.3%), pratensis Emerton (3.6%), Hogna frondicola (Emerton) (3.5%), and Zelotes fratris Chamberlin (3.1%) (Table 3). The abundance of P. moesta differed among treatments for all four years (1997: F3= 7.913, P = 0.007; 1998:

F4 = 72.135, P < 0.001; 1999: F4 = 10.148, P = 0.003; 2000: F4 = 17.858, P < 0.001), with the refuge treatment having the highest abundance each year (Table 4). The abundance of P. distincta did not differ among treatments in any year (Table 4). In 1999 and 2000, P. distincta was most abundant in the summer treatment and had the lowest abundance in the spring treatment all four years. In general, P. moesta was the dominant species for all treatments in 1997 except for Block D, where P. distincta was dominant. By 2000, P. moesta was dominant only in the spring and refuge treatments for Blocks A–C, whereas P. distincta was the dominant species in the remaining treatments and Block D. Although the refuge treatment had the highest total abundance, some species were more abundant in the summer or fall treatments than they were in the refuge treatment. These species included P. distincta; H. frondicola; Xysticus ampullatus Turnbull, Dondale and Redner; Arctosa rubicunda (Keyserling); and Enoplognatha marmorata Hentz (in order of abundance in Table 3).

Diversity Indices There were no differences in species richness among treatments following the burns in 1997 or in any subsequent year (1997: F = 0.979 df = 3; 1998: F = 1.094, df = 4; 1999: F = 1.380, df = 4; 2000: F = 1.000, df = 4) (Table 5). The diversity of spiders, as measured by the Shannon-Wiener index, differed among treatments in 1997 (P = 0.015, df = 3, F = 6.242) and 1998 (P = 0.039, df = 4, F = 4.259) but not in 1999 (df = 4, F = 1.658) or 2000 (df = 4, F = 0.410) (Table 5). Spider diversity, as measured by the log series alpha index, differed among treatments in 1997 (P = 0.030, df = 3, F = 4.714) and 1998 (P = 0.008, df = 4, F = 6.799) but not in 1999 or 2000 (1999: P = 0.202, df = 4, F = 1.860; 2000: P = 0.062, df = 4, F = 3.532) (Table 5). Response of spiders to fire in tallgrass prairie 243

Table 3. Abundance of adults by burn season treatment of the 25 most abundant spider species collected on St. Charles Rifle Range from 1997 to 2000. Note: Controls are not directly comparable to other treatments because of differences in sampling intensity.

Treatment* % of All Species Total Spr Sum Fall Ref Con Species Pardosa moesta Banks 4675 2286 1831 6890 1390 17072 31.38 Pardosa distincta (Blackwall) 1990 2987 2770 2773 1281 11801 21.69 Alopecosa aculeata (Clerck) 843 563 550 920 565 3441 6.32 Agroeca pratensis Emerton 444 358 300 611 219 1932 3.55 Hogna frondicola (Emerton) 394 514 542 346 85 1881 3.46 Zelotes fratris Chamberlin 433 238 220 574 240 1705 3.13 Xysticus ampullatus Turnbull, 250 291 355 257 114 1267 2.33 Dondale & Redner Ozyptila conspurcata Thorell 279 130 240 331 173 1153 2.12 Centromerus sylvaticus (Blackwall) 287 199 171 344 139 1140 2.10 Xysticus ferox (Hentz) 221 196 232 272 123 1044 1.92 Trochosa terricola Thorell 343 132 104 279 121 979 1.80 Goneatara nasutus (Barrows) 237 260 158 218 27 900 1.65 Pirata minutus Emerton 173 236 94 144 37 684 1.26 Pardosa modica (Blackwall) 208 124 52 202 48 634 1.17 Thanatus striatus C.L. Koch 173 141 108 157 36 615 1.13 Arctosa rubicunda (Keyserling) 97 149 157 90 12 505 0.93 Schizocosa crassipalpata Roewer 101 81 87 167 41 477 0.88 Xysticus emertoni Keyserling 108 52 61 130 105 456 0.84 Cicurina arcuata Keyserling 69 89 125 55 24 362 0.67 Gnaphosa parvula Banks 76 48 31 147 56 358 0.66 Agroeca ornata Banks 90 61 92 93 16 352 0.65 Euryopis funebris (Hentz) 87 69 76 74 37 343 0.63 Ceraticelus laetus (O.P.-Cambridge) 137 21 21 121 24 324 0.60 plumosa (Emerton) 90 135 35 47 17 324 0.60 Enoplognatha marmorata (Hentz) 57 117 106 29 9 318 0.58

* Spr = spring; Sum = summer; Fall = fall; Ref = refuge; Con = control.

The diversity of spiders in the burn treatments as measured by the Simpson index were not significantly different among treatments in any year (1997: F = 2.928; 1998: F = 2.164; 1999: F = 0.311; 2000: F = 0.110), but the refuge treatment had the highest values in all four years (Table 5). No difference was found in diversity among treatments in any year based on the Berger-Parker index (1997: F = 2.346; 1998: F = 0.959; 1999: F = 0.354; 2000: F = 0.080) (Table 5). The alpha evenness values of the burn treatments differed among treatments in 1998 (P = 0.043, df = 3, F = 4.118) and 2000 (P = 0.029, df = 4, F = 4.488) but not in 1997 or 1999 (Table 5). 244 D. J. Wade and R. E. Roughley

Table 4. The effect of burn season on the natural log abundance (mean ± SD) of adult Pardosa distincta and P. moesta collected on the St. Charles Rifle Range. Values with the same letter within a given year arenot significantly different at the 5% confidence level based on Tukey-HSD comparisons.

Year Species Treatment 1997 1998 1999 2000 Pardosa Spring 4.37 ± 0.55 a 4.68 ± 0.46 a 5.01 ± 0.29 a 4.83 ± 0.58 a distincta Summer 4.52 ± 0.56 a 5.00 ± 0.21 a 5.44 ± 0.13 a 5.54 ± 0.21 a Fall 4.61 ± 0.59 a 4.91 ± 0.16 a 5.34 ± 0.11 a 5.43 ± 0.18 a Refuge 4.71 ± 0.38 a 5.30 ± 0.16 a 5.17 ± 0.20 a 5.22 ± 0.38 a Control N/A* 5.04 ± 0.60 a 5.24 ± 0.65 a 5.44 ± 0.73 a

Pardosa Spring 4.59 ± 1.00 a 5.80 ± 0.65 a,c 5.57 ± 0.64 a,b 5.63 ± 0.90 a moesta Summer 5.22 ± 0.70 a,b 3.95 ± 0.66 a 4.61 ± 1.08 a 4.84 ± 0.98 b Fall 5.17 ± 0.67 a,b 3.21 ± 0.92 b 4.62 ± 0.54 a 4.45 ± 1.01 b Refuge 5.92 ± 0.36 b 6.21 ± 0.66 a,c 5.85 ± 0.45 b 5.75 ± 0.63 a Control N/A 5.30 ± 0.21 a 5.45 ± 0.35 a 5.32 ± 0.96 a

* not applicable (traps not installed until 1998)

Similarity Indices

Jaccard Index

The spring and refuge treatments in 1997 were most similar (Cj = 0.634) and the summer and fall treatments had the next highest similarity value (Cj = 0.605) (Table 6). In 1998, summer and fall treatments were most similar (Cj = 0.737). In 1999, spring and fall treatments had the highest similarity (Cj = 0.736) and in 2000, the spring and refuge treatments were again the most similar (Cj = 0.758) (Table 6). In 2000, the similarity values among treatments were fairly similar, ranging from 0.630 (summer and control) to 0.758 (spring and refuge). The similarity values among the refuge and control treatments between years remained fairly constant, but the values for the other treatments varied (Table 6).

Morisita-Horn Index The similarity values in 1997 between treatments ranged from 0.924 (spring and refuge) to 0.996 (summer and fall) (Table 6). In 1998, the values diverged, with spring and refuge

(Cmh = 0.993) and summer and fall (Cmh = 0.983) having the two highest similarity values. The high similarity values between the spring and refuge treatments and the summer and fall treatments continued in 1999 and 2000 (Table 6). The similarity values for the refuge and control treatments between years were constant and relatively high (Table 6). The values for the spring treatment between years also remained constant and relatively high. The values for the summer and fall treatments had greater variation but also had high values for the later-year comparisons (Table 6). Response of spiders to fire in tallgrass prairie 245 * N/A N/A N/A N/A N/A N/A Control 36.0 ± 1.0 a 53.5 ± 1.5 a 44.0 ± 3.0 a 2.34 ± 0.05 a 0.180 ± 0.006 a 0.355 ± 0.010 a 9.877 ± 0.781 a 2.225 ± 0.045 a,b 8.294 ± 0.023 a 0.204 ± 0.020 a 0.150 ± 0.020 a 0.376 ± 0.047 a 2.625 ± 0.065 a 0.330 ± 0.032 a -0.110 ± 0.015 a -0.110 -0.129 ± 0.001 b -0.131 ± 0.004 b 11.488 ± 0.044 a,b 11.488 Refuge 45.8 ± 3.2 a 53.8 ± 3.9 a 47.5 ± 3.2 a 33.8 ± 3.9 a 9.556 ± 0.369 a 2.038 ± 0.329 a 0.148 ± 0.036 a 0.346 ± 0.078 a 0.546 ± 0.096 a 0.460 ± 0.135 a 0.318 ± 0.064 a 0.374 ± 0.075 a 2.653 ± 0.162 a 0.200 ± 0.041 a 2.323 ± 0.125 a 7.443 ± 1.124 b 1.698 ± 0.234 b 0.278 ± 0.111 a 0.278 ± 0.111 -0.108 ± 0.009 a -0.092 ± 0.001 a -0.141 ± 0.017 a -0.101 ± 0.006 a 11.673 ± 0.506 b 11.673 10.478 ± 0.595 a Fall 42.2 ± 4.1 a 53.8 ± 3.8 a 48.0 ± 4.9 a 31.2 ± 3.7 a Treatment 0.166 ± 0.027 a 0.131 ± 0.055 a 0.264 ± 0.045 a 0.474 ± 0.049 a 0.352 ± 0.050 a 0.289 ± 0.101 a 0.358 ± 0.075 a 2.818 ± 0.221 a 0.180 ± 0.035 a 2.495 ± 0.127 a 7.404 ± 1.163 b 1.990 ± 0.190 b 2.523 ± 0.100 b -0.092 ± 0.008 a -0.084 ± 0.007 a -0.155 ± 0.031 a -0.090 ± 0.009 a 12.290 ± 1.029 a 12.831 ± 0.659 a,b 12.031 ± 0.918 b Summer 58.0 ± 3.7 a 46.8 ± 3.5 a 42.2 ± 1.9 a 34.0 ± 1.6 a 0.148 ± 0.079 a 0.291 ± 0.136 a 0.358 ± 0.074 a 2.740 ± 0.296 a 0.189 ± 0.032 a 2.430 ± 0.073 a 0.167 ± 0.026 a 0.349 ± 0.046 a 0.269 ± 0.048 a 0.483 ± 0.056 a 2.503 ± 0.146 b 8.245 ± 0.752 b 2.035 ± 0.173 b -0.080 ± 0.004 a -0.093 ± 0.003 a -0.096 ± 0.007 a -0.131 ± 0.011 a -0.131 ± 0.011 11.511 ± 0.521 a 11.511 13.563 ± 0.417 a 11.648 ± 0.869 a,b 11.648 Spring 42.5 ± 4.2 a 55.5 ± 8.3 a 51.2 ± 5.1 a 33.8 ± 1.9 a 9.640 ± 0.658 a 0.241 ± 0.070 a 0.134 ± 0.019 a 0.227 ± 0.120 a 0.397 ± 0.149 a 2.173 ± 0.200 a 0.451 ± 0.109 a 0.307 ± 0.045 a 0.342 ± 0.068 a 2.758 ± 0.038 a 0.171 ± 0.034 a 8.966 ± 1.067 a 2.518 ± 0.126 a 2.203 ± 0.341 a -0.110 ± 0.010 a -0.110 -0.087 ± 0.009 a -0.127 ± 0.016 a -0.091 ± 0.011 a -0.091 ± 0.011 12.159 ± 1.388 a 12.378 ± 1.164 a,b Log series alpha Alpha evenness Simpson Alpha evenness Simpson Alpha evenness Simpson Berger-Parker Species richness Shannon-Wiener Berger-Parker Berger-Parker Species richness Berger-Parker Shannon-Wiener Simpson Log series alpha Alpha evenness Log series alpha Shannon-Wiener Species richness Species richness Log series alpha Shannon-Wiener Diversity Index The effect of burn season on the diversity (mean ± SD) of spiders present on the St. Charles Rifle Range. For each diversity index, treatment values with the same letter not applicable (traps not installed until 1998) not installed until not applicable (traps 1998

2000 1999 Year 1997 are not significantly different at the 5% confidence level based on Tukey-HSD comparisons. at the 5% confidence level based on are not significantly different Table 5. Table * 246 D. J. Wade and R. E. Roughley Co00 Co00 0.655 0.699 0.932 0.630 0.935 0.930 0.693 0.901

Re00 Re00 0.724 0.820 0.713 0.868 0.758 0.991

Fa00 Fa00 0.710 0.985 0.717 0.765

Su00 Su00 0.724 0.824

Co99 Co99 0.605 0.545 0.658 0.939 0.925 0.955 0.595 0.969

Re99 Re99 0.674 0.648 0.831 0.809 0.682 0.994

Fa99 Fa99 0.711 0.995 0.736 0.834

Su99 Su99 0.727 0.854

Co98 Co98 0.586 0.620 0.741 0.907 0.652 0.831 0.594 0.908

Re98 Re98 0.725 0.494 0.679 0.609 0.671 0.993

Fa98 Fa98 0.737 0.983 0.727 0.506

Su98 Su98 0.724 0.618

from 1997 to 2000 based on (a) Jaccard and (b) Morisita-Horn indices. *

Re97 Re97 0.556 0.573 0.958 0.941 0.634 0.924

Fa97 Fa97 0.605 0.996 0.600 0.993

Su97 Su97 0.558 0.988

Fa00 Fa00 Su00 Su00 Fa99 Sp00 Fa99 Sp00 Su99 Su99 Fa98 Fa98 Sp99 Sp99 Fa97 Sp98 Fa97 Sp98 Su98 Su98 Su97 Su97 Sp97 Sp97 Re00 Re99 Re99 Re98 Re98 Re00 Sp = spring; Su = summer; Fa = fall; Re = refuge; Co control. Sp = spring; Su summer; Fa fall; Re (a) Jaccard (b) Morisita-Horn Table 6. Similarity values between burn season treatments Table * Response of spiders to fire in tallgrass prairie 247

Significance of Findings The results from this study agree with the trends seen in previous studies (Rice 1932; Riechert and Reeder 1972; Nagel 1973). Spider abundance was negatively affected, whereas species diversity was generally unaffected following each burn treatment. The negative effect of the burn treatments on the abundance of spiders was observed in the period immediately following the burn treatment in 1997, although the effect of the fall burn was not significant. The abundance of spiders recovered within one year of the spring treatment but required an additional two years to recover after the summer treatment. The abundance of spiders after the fall burn treatment never did fully recover during the project. We assumed that the refuge treatment represented the pre-burn status of the tallgrass prairie habitat at SCRR. Recolonization into a burned area is an important factor in determining recovery rates following a fire incident (Harper et al. 2000). The majority of spider species collected in this study have peak abundance levels in the spring (Aitchison 1984, pers. obs.). Therefore, recolonization rates would have been higher following the spring treatment and allowed for a quicker recovery. The spring treatment had a significant increase in species diversity in 1997 that could have also affected recolonization rates. Summer and fall burn season treatments also increased species diversity but not until the following year. Overall, the abundance and diversity data strongly suggest that four years is an appropriate recovery period for the spider community. Some of the negative effects that burn season had on spider abundance were due to its effect on the two most abundant species, P. moesta and P. distincta. Burn season did not affect the abundance of P. distincta but did have a negative effect on P. moesta. This resulted in the initially dominant P. moesta (except in Block D) being replaced by P. distincta as the dominant species in the summer and fall treatments (and all treatments in Block D). The reason that this trend was not seen in the spring treatment was because the spring burn occurred before the peak activity period of P. moesta (Aitchison 1984, pers. obs.), allowing it to rapidly recolonize the burned area. The summer and fall burns occurred when the females of P. distincta were more active than P. moesta (pers. obs.); thus, P. distincta recolonized at a faster rate. This difference in phenology resulted in the formation of P. distincta-dominated versus P. moesta-dominated spider communities. An important consideration when deciding which burn season to use is community stability (Collins 2000). A community that has high stability will resist disturbances to maintain functionality (Holling 1973). An important aspect of community stability is resilience, which is the rate at which a system returns to its former state following a disturbance (Holling 1973). If fire were to become a predominant management technique, then the community that is most resilient to the effects of fire would be preferred, which in this study was the P. distincta-dominated community. This was evident in Block D, where in 1997 the spring treatment was dominated by P. distincta. In 1998, P. moesta became the dominant species, but P. distincta became the dominant species in 1999 and maintained this dominance in 2000. In comparison, P. moesta could not regain its dominance in the summer or fall treatments in Blocks A, B, and C. The P. distincta-dominated community also exhibited resistance, another important component of stability. The P. distincta-dominated communities in the summer and fall treatments in Blocks A, B, and C were maintained even though the adjacent refuge treatment had a relatively high abundance of P. moesta. Based on these results, the prediction can be made that once the spider community shifts to the P. distincta-dominated community, it will remain as such as long as fire is a regular disturbance. How regular that fire disturbance needs to be remains to be investigated. 248 D. J. Wade and R. E. Roughley

From a forest transect study that was conducted concurrently with this study, P. distincta is a tallgrass prairie specialist on the SCRR (Roughley et al. 2006). Other tallgrass prairie specialists include H. frondicola and A. rubicunda. The increased abundance of these species in response to the summer and fall burn treatments is important, because biologists theorize that what is beneficial to the habitat specialists should be good for the habitat as a whole (Bell et al. 2001). This trend is important to prairie conservation because summer and fall burn seasons have not been thoroughly examined or implemented as management options (Howe 1994; Collins et al. 1998). Our recommendation to prairie managers is to incorporate all three burn seasons into the management strategy, with a four- to five-year burn rotation. All three burn treatments had a slightly different outcome. Therefore, by incorporating all three burn seasons, the diversity of the entire site should be maximized. Refugia are still required to protect species from the physical effects of fire and to ensure species restricted to the tallgrass prairie (e.g., P. distincta) are not extirpated. The spider communities should be monitored by the use of pitfall traps to determine the long-term effects of burn management. A detailed burn management strategy for the SCRR was developed by Roughley (2001) and is fully supported by this study. Future studies are needed to examine factors that were not tested in this study but that are relevant when our recommendations are implemented on a larger scale. These factors include the effects of burn size, burn timing, fire frequency, and fire rotation sequence on the spider community.

Acknowledgements The project was funded by the Canadian Department of National Defence, a postgraduate scholarship from NSERC (DJW), a Discovery Grant from NSERC (RER), and a Duff Roblin Fellowship from the University of Manitoba (DJW). We thank numerous people who assisted with the project over the years (see Roughley 2001 for a complete list) and we appreciate all their help. In particular, the senior author thanks Jackie Le Gal for helping sort spiders during the summer of 2001. We would also like to thank Dr. N.J. Holliday for his statistical advice.

References

Aitchison, C.W. 1984. The phenology of winter-active spiders. The Journal of Arachnology, 12: 249–272. Bell, J.R., Wheater, C.P., and Cullen, W.R. 2001. The implications of grassland and heathland management for the conservation of spider communities: a review. Journal of Zoology, 255: 377–387. Collins, S.L. 1990. Introduction: fire as a natural disturbance in tallgrass prairie ecosystems.In Fire in North American Tallgrass Prairies. Edited by S.L. Collins and L.L. Wallace. University of Oklahoma Press, Norman. pp. 3–7. Collins, S.L. 2000. Disturbance frequency and community stability in native tallgrass prairie. The American Naturalist, 155: 311–325. Collins, S.L., Knapp, A.K., Hartnett, D.C., and Briggs, J.M. 1998. The dynamic tallgrass prairie: synthesis and research opportunities. In Grassland Dynamics: Long-Term Ecological Research in Tallgrass Prairie. Edited by A.K. Knapp, J.M. Briggs, D.C. Hartnett, and S.L. Collins. Oxford University Press, New York. pp. 301–315. Collins, S.L., and Steinauer, E.M. 1998. Disturbance, diversity, and species interactions in tallgrass prairie. In Grassland Dynamics: Long-Term Ecological Research in Tallgrass Prairie. Edited by A.K. Knapp, J.M. Briggs, D.C. Hartnett, and S.L. Collins. Oxford University Press, New York. pp. 140–156. Harper, M.G., Dietrich, C.H., Larimore, R.L., and Tessene, P.A. 2000. Effects of prescribed fire on prairie arthropods: an enclosure study. Natural Areas Journal, 20: 325–335. Response of spiders to fire in tallgrass prairie 249

Holling, C.S. 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics, 4: 1–23. Howe, H.F. 1994. Managing species diversity in tallgrass prairie: assumptions and implications. Conservation Biology, 8: 691–704. Johnson, S.R. 1995. Spider communities in the canopies of annually burned and long-term unburned Spartina pectinata wetlands. Environmental Entomology, 24: 832–834. Morgan, J. 1994. Prototype Vegetation Management Plan: Prairie Habitats. Canadian Forces Base-Winnipeg, Winnipeg, Manitoba. Nagel, H.G. 1973. Effect of spring prairie burning on herbivorous and non-herbivorous populations. Journal of the Kansas Entomological Society, 46: 485–496. Nature Conservancy of Canada. 2000. Field Guide to Manitoba’s Tall Grass Prairie Preserve. The Nature Conservancy of Canada, Winnipeg, Manitoba. Platnick, N.I. 2008. The World Spider Catalog, Version 8.5. American Museum of Natural History [online]. Available from http://research.amnh.org/entomology/spiders/catalog/ [accessed 16 March 2008]. Reed, C.C. 1997. Responses of prairie insects and other arthropods to prescription burns. Natural Areas Journal, 17: 380–385. Rice, L.A. 1932. The effect of fire on the prairie communities. Ecology,13 : 392–401. Riechert, S.E., and Reeder, W.G. 1972. Effects of fire on spider distribution in southwestern Wisconsin prairies. In Proceedings of the 2nd Midwest Prairie Conference. Edited by J.H. Zimmerman. University of Wisconsin, Madison. pp. 73–90. Robertson, K.R., Anderson, R.C., and Schwartz, M.W. 1997. The tallgrass prairie mosaic. In Conservation in Highly Fragmented Landscapes. Edited by M.W. Schwartz. Chapman & Hall, New York. pp. 55–87. Roughley, R.E. 2001. St. Charles Rifle Range Tallgrass Prairie: An Endangered Habitat. The Use of Fire as a Biodiversity and Conservation Management Tool. Final unpublished report to 1 Canadian Air Division, Department of National Defence, Winnipeg, Manitoba. Roughley, R.E., Pollock, D.A., and Wade, D.J. 2006. Biodiversity of ground beetles (Coleoptera: Carabidae) and spiders (Araneae) across a tallgrass prairie: aspen forest ecotone in southern Manitoba. The Canadian Entomologist, 138: 545–567. Schwartz, M.W., and Hermann, S.M. 1997. Midwestern fire management: prescribing a natural process in an unnatural landscape. In Conservation in Highly Fragmented Landscapes. Edited by M.W. Schwartz. Chapman & Hall, New York. pp. 213–233. Sveinson, J. 2001. Botany. In St. Charles Rifle Range Tallgrass Prairie: An Endangered Habitat. The Use of Fire as a Biodiversity and Conservation Management Tool. R.E. Roughley. University of Manitoba, Winnipeg. pp. 38–49. Warren, S.D., Scifres, C.J., and Teel, P.D. 1987. Response of grassland arthropods to burning: a review. Agriculture, Ecosystems & Environment, 19: 105–130.