COMMUNITY AND ECOSYSTEM ECOLOGY
Small-Scale Spatial Pattern of Web-Building Spiders (Araneae) in Alfalfa: Relationship to Disturbance from Cutting, Prey Availability, and Intraguild Interactions KLAUS BIRKHOFER,1,2 STEFAN SCHEU,1
AND
DAVID H. WISE3,4
Environ. Entomol. 36(4): 801Ð810 (2007)
ABSTRACT Understanding the development of spatial patterns in generalist predators will improve our ability to incorporate them into biological control programs. We studied the small-scale spatial patterns of spider webs in alfalfa by analyzing the relationship between web locations over distances ranging from 4 to 66 cm. Using a coordinate-based spatial statistic (O-ring) and assuming a heterogeneous distribution of suitable web sites, we analyzed the impact of cutting and changes in spider abundance on web distribution. We analyzed the inßuence of small-scale variation in prey availability by comparing web distributions to the pattern of sticky-trap captures of Aphididae and Diptera described by a count-based spatial statistic (SADIE). Cutting of alfalfa reduced the overall density of web-building spiders but had no immediate impact on the spatial distribution of their webs. Availability of aphids was highest before the alfalfa was cut and was clumped at a scale of 66 cm. Spider webs, however, were not clumped at any scale or date. In contrast, webs were regularly distributed at smaller distances (⬍20 cm) immediately before and after cutting. Because cursorial and web-building spiders were most active during this period, we hypothesize that the development of small-scale regularity in web locations was driven by intraguild interactions. Our results suggest that intraguild interactions contribute to the development of small-scale spatial patterns of spider webs in alfalfa. Variation in prey availability may have more of an inßuence on web distribution in crops with a different vegetation structure or if patterns are studied at larger spatial scales. KEY WORDS generalist predators, predatorÐprey association, biological control
Spiders can act as biological control agents in different agroecosystems (for review, see Riechert and Lockley 1984, Wise 1993, Symondson et al. 2002, Nyffeler and Sunderland 2003), with web-building species preying on such economically important pest taxa as Aphididae and Auchenorrhyncha (Nyffeler et al. 1994). Alfalfa agroecosystems have several pests, with the potato leaf hopper, Empoasca fabae (Harris) (Auchenorrhyncha), and virus-transmitting aphid species being among the most destructive (Pons et al. 2005). Web-building spiders are abundant in alfalfa Þelds (Wheeler 1973) and may play an important role in limiting the population growth of aphids in agroecosystems (Marc et al. 1999). Effective incorporation of web-building spiders in pest management strategies, such as conservation biological control, requires knowledge of factors such as prey distribution, spiderÐ 1 University of Technology Darmstadt, Department of Animal Ecology, Schnittspahnstrasse 3, 64287 Darmstadt, Germany. 2 Corresponding author: Justus-Liebig-University Giessen, Department of Animal Ecology, Heinrich-Buff-Ring 26-32, 35392 Giessen, Germany (e-mail:
[email protected]). 3 University of Kentucky, Department of Entomology, S-225 Ag Sci Bldg.ÐNorth, Lexington, KY 40546-0091. 4 Present address: University of Illinois at Chicago, Department of Biological Sciences & Institute for Environmental Science & Policy, 845 W. Taylor St. (M/C 066), Chicago, IL 60607-7060.
spider interactions, and farming practices that limit spider density and determine the distribution of spiders within arable Þelds, at both large and small scales. Spider populations in agroecosystems are generally disturbed by agricultural management practices, with known detrimental effects from tillage (Thorbek and ¨ berg and Ekbom 2006), and Bilde 2004), sowing (O insecticide application (Pekar 1999). Effects of cutting are compared with catastrophic events by Thomas and Jepson (1997), who documented a 56 Ð 89% reduction in the population size of linyphiid spiders in recently cut grass lays. Despite this immediate negative effect, spider populations recovered quickly through aerial recolonization (ballooning). In contrast, Gudleifsson and Bjarnadottir (2004) did not measure any impact of cutting in hay Þelds, but instead attributed ßuctuations of spider population size to temperature changes over the season. Howell and Pienkowski (1971) report a rapid decline of linyphiid abundance after cut alfalfa was removed from the Þeld, but the population recovered to preharvest densities within a week. In addition to studying the numerical response to management practices and increased pest abundance (Elliott et al. 2002), ecologists have recently given considerable attention to the large-scale spatial association between prey species and generalist predators.
0046-225X/07/0801Ð0810$04.00/0 䉷 2007 Entomological Society of America
802
ENVIRONMENTAL ENTOMOLOGY
Carabid beetles and ground running spiders are among the most intensely studied groups. Results are contradictory, revealing species-speciÞc and temporally unstable associations or even the absence of a spatial relationship with prey at the Þeld scale (Bryan and Wratten 1984, Warner et al. 2000, Winder et al. 2005, Pearce and Zalucki 2006). Harwood et al. (2001, 2003) found that web-building spiders (Linyphiidae) in winter wheat preferentially located their webs in patches with greater prey availability. Because many pest species (e.g., aphids) colonize crops early in the growing season, any delay between pest arrival and an increase in predator numbers would reduce their potential to suppress pest populations. A rapidly forming small-scale spatial association between aphid prey and predators could limit aphid population growth before the spider exhibited a numerical response at the larger scale of the entire Þeld. Such a change in small-scale spatial association might be of major importance for aphid control, because aphids may be most successfully suppressed under certain threshold densities and when distribution is still limited to a few patches within crops (Marc et al. 1999, Lang 2003). Intraguild interactions (competition and predation) are known to affect the abundance, distribution, and community structure of generalist predators (Rosenheim et al. 1995). Samu et al. (1996) experimentally increased the availability of web sites for a linyphiid spider in wheat. Adult females apparently competed over these artiÞcial web sites, with heavier individuals winning contests more frequently. Harwood and Obrycki (2005) showed in a laboratory study that competition for space inßuenced web size and web-site tenacity for three linyphiid species. Mortality did not differ in intra- and interspeciÞc competition treatments, indicating comparable rates of cannibalism and intraguild predation. Further studies in nonagricultural systems have uncovered intra- and interspeciÞc competition for web-site space (Spiller 1984, Toft 1987, Wise 1993, Herberstein 1998). Few studies of solitary web-building spiders have analyzed small-scale spatial patterns over a range of distances (studies of burrow-living spiders are discussed elsewhere; Marshall 1997, Birkhofer et al. 2006). Small-scale aggregation in web-building spiders is often limited to juvenile life stages, with individuals later dispersing from the vicinity of the maternal web (for review, see Burgess and Uetz 1982). Web locations of the funnel web spider Agelenopsis aperta (Gertsch 1934) reßect the distribution of suitable microhabitat sites at large scales, resulting in aggregations (Riechert 1976). Block-size analysis of variance (ANOVA) revealed regularity at smaller scales that resulted from territorial behavior. Hodge and StorferIsser (1997) documented small-scale aggregation as a result of conspeciÞc cuing; individuals searching for web sites favored patches with webs over web-free areas, leading to clusters of spiders. This study is an analysis of small-scale spatial patterns of spider webs in alfalfa, based on the complete mapping of webs before and after alfalfa was cut. Data on the dispersion patterns of potential prey (aphids
Vol. 36, no. 4
and ßies) and the activity-densities of a major category of nonweb spider (lycosids) were also collected before and after cutting. Thus, we were able to infer the possible impacts of available prey, cutting, and intraguild interactions (e.g., competition, cannibalism, and/or intraguild predation) on the small-scale distribution of spider webs. Materials and Methods Study Site and Sampling. During Þve 2-d survey periods between 14 September and 7 November 2005, 1,224 spider webs were located within 2 by 2-m plots (14 September and 4 October, N ⫽ 5; other dates, N ⫽ 4) inside an alfalfa Þeld (University of Kentucky Spindletop Research Farm, Fayette County, KY). Intact vacant webs were included in the analysis because abandoned sheet webs degenerate quickly (Toft et al. 1995, Schmidt and Tscharntke 2005). Survey periods were restricted to 2 d so that spatial patterns observed during a single survey period would likely result from a consistent combination of processes. The plot size selected permitted a focus on small-scale processes, such as neighborÐneighbor conßicts and aggregation of spiders in response to small-scale variation in prey availability. The plot size of 2 by 2 m is also close to the maximum possible without causing excessive disturbances to spiders in yet-unsampled areas. Excluding the sampling date immediately after cutting, there were 69 ⫾ 7 web locations per 4-m2 plot, a density sufÞcient for our point pattern analysis. New plots were randomly chosen on each survey date from a 3 by 8-plot grid, which was at least 20 m from the Þeld edge and 6 m from other three by 8-plot grids. Each search for webs within a 4-m2 plot lasted 3Ð 4 h. Use of a water vaporizer facilitated web detection. Spiders (N ⫽ 684) were captured from occupied webs with an aspirator. Immature spiders were determined to family level, adults (N ⫽ 252) to species using keys of Kaston (1948) and Ubick et al. (2005). Website characteristics (size, stratum, and spider family) were recorded, and locations were marked with a labeled ßag. Five randomly located measurements of vegetation height were taken from all grids before Cartesian coordinates of all web locations were recorded. A wooden 2 by 2-m frame enclosed the survey area, with measurement tape Þxed to two parallel sides indicating the Y coordinate. A right angled, 1-m ruler was used to record the X coordinate after the area had been divided into two 2 by 1-m segments. The meadow was cut on 1 October and alfalfa was removed on 3 October. The Þrst web survey was conducted 16 d before cutting, followed by a survey the day after alfalfa removal. Two further surveys were conducted weekly thereafter, and a last survey was conducted 35 d after cutting. The dominant web-building species in North American alfalfa Þelds belong to four families: Linyphiidae, Theridiidae, Tetragnathidae, and Araneidae (Culin and Yeargan 1983a, b). Most species from these families build webs either on the soil surface or attached to low vegetation. In our study plots, the vegetation
August 2007
BIRKHOFER ET AL.: SPIDER WEB SPATIAL PATTERN IN ALFALFA
layer was dominated by two long-jawed orb weavers (Tetragnathidae), Tetragnatha laboriosa Hentz, 1850 and Glenognatha foxi (McCook, 1894), accounting for 8 and 10%, respectively, of all adults; the orb weaver (Araneidae) Gea heptagon (Hentz, 1850) (9% of all adults); and the sheetweb spider (Linyphiidae) Florinda coccinea (Hentz, 1850) (13%). The ground layer was dominated by the linyphiid species Meioneta unimaculata (Banks, 1892), Tennesseellum formica (Emerton, 1882) and Erigone autumnalis Emerton, 1882, accounting for 13, 36, and 5%, respectively. The above percentages are based on the total number of adult spiders collected in both strata. Analyses of spatial patterns were based on webs of all species pooled, with the exception of one analysis in which webs were divided into two size classes. Pooling was necessary because most individuals were immature stages of the smaller linyphiids that could not be identiÞed to species. Activity-density of wolf spiders (Lycosidae) was estimated using pitfall traps (white plastic cups Þlled with water, Ø ⫽ 9.0 cm) in a neighboring part of the same alfalfa Þeld. Six pitfalls were placed in three rows at 10, 20, and 30 m from the Þeld edge with 6 m between traps. Traps were opened for 2 d during each web survey period. Spatial Analysis. Spatial patterns of point locations (in our case, web locations) can be assigned to one of three categories: aggregated, random, or regular. Because the category of spatial pattern exhibited can be a function of spatial scale (Wiens 1989), we chose a spatial-analysis technique, the O-ring statistic (Programita Software, Wiegand and Moloney 2004), that enabled us to vary the scale at which the pattern of web locations was analyzed, within the limits of our 2 by 2-m plots. This technique requires knowing the coordinates of all points comprising the observed pattern. The distribution pattern of available prey (Aphididae and Diptera) was analyzed differently, because coordinate mapping is impractical for nonstationary animals. Prey distribution was analyzed with a countbased spatial statistic (SADIE; Perry 1998). Choice of methods was based on the guidelines of Fortin et al. (2002) and Perry et al. (2002). The following section brießy describes the techniques and provides the rationale behind speciÞc analyses applied to our data. The intensity of a pattern is the number of points (N) relative to the area mapped (A). Spatial-analysis software can redistribute the points of known intensity and pattern according to different null-model distributions. Any point, either observed or part of the pattern generated by a particular null model, can be analyzed for the spatial relationship to all other points within this pattern. The O-ring statistic uses each point within a pattern as a focal location and calculates the number of points in rings of deÞned width around those focal points (ring width in this study ⫽ 10 cm). Ring distances to focal points increase stepwise, providing the O-ring statistic over scales with increasing distance. Results of O-ring analysis for an observed pattern do not by themselves provide information about the category of the pattern at different dis-
803
Fig. 1. Example of a null model of a spider web distribution pattern generated by a heterogeneous Poisson process applied to the observed distribution of webs (F) inside a 4-m2 grid on 15 September. Dark areas, areas with a lower probability of web occurrence; brighter areas, higher probability of occurrence.
tances. To make inferences about the category of the observed pattern, a simulated reference distribution based on a null-model with identical intensity can be used to create conÞdence bounds for the observed pattern. Because of the documented correlation between vegetation complexity and spider web location (Janetos 1986, Riechert and Gillespie 1986, Uetz 1991, Dennis et al. 1998, Samu et al. 1999, McNett and Rypstra 2000), for our null model of spider web distribution, we chose a heterogeneous Poisson process, which models heterogeneity of plant distribution (Wildova 2004, Wijesinghe et al. 2005). An example of areas with different probabilities of web occurrence based on a mapped pattern of spider webs is given in Fig. 1. To estimate web-site probability areas, a window of 30 cm radius is located around each observed location; the magnitude of overlap between such circles deÞnes web-site probability. The relatively large window size is a conservative proxy for vegetation-caused habitat heterogeneity. Based on this probability distribution, the heterogeneous Poisson null model redistributes points 100 times and calculates the O-ring statistic for all replicates to estimate the 95% conÞdence envelopes (Þfth highest and Þfth lowest simulation value; Stoyan and Stoyan 1994). When comparing O-ring statistics of observed and reference patterns, deviation from randomness at a given distance is indicated by the O-ring function crossing the conÞdence envelopeÑregularity when crossing the lower boundary, aggregation when crossing the upper limit. In addition to a univariate analysis, in which all webs are assigned to the same pattern independent of website or web-owner characteristics, we also performed a bivariate analysis, which differentiates between two
804
ENVIRONMENTAL ENTOMOLOGY
types of points. The bivariate analysis uncovered possible spatial relationships between webs of larger (⬎5 mm) and smaller (⬍3 mm) species. Webs of larger species (⬎5 mm body length; F. coccinea, T. laboriosa, and G. heptagon) were distinguishable from webs of smaller species (⬍3 mm body length; G. foxi, M. unimaculata, T. formica, E. autumnalis, and immature and unidentiÞed linyphiids of the soil surface) on the basis of architectural features and/or stratum. The univariate null model tests for spatial randomness between points, whereas the bivariate null model tests for independent occurrence of the two different point patterns. Crossing envelopes in the bivariate analysis indicates repulsion between webs when crossing the lower limit and attraction when crossing the upper envelope. Only data from 14 September (before cutting) had sufÞcient numbers of webs of larger species for a bivariate analysis. To correct for edge effects, data were analyzed for distances only up to one third of the shortest extent of the study area (Fortin and Dale 2005), with distances ⬍4 cm being omitted because of the web size itself. Because measuring coordinates for nonstationary objects is impossible, spatial patterns of Aphididae and Diptera, insects that are potential prey for web-building spiders, were estimated using a 2 by 2-m grid of 16 regularly spaced, horizontally orientated sticky traps (simulating spider webs). Each trap consisted of 8 by 8-cm pieces of uncolored wire mesh that was coated with an insect banding compound (Tanglefoot Company, Grand Rapids, MI; see also Harwood et al. 2001) and placed slightly above soil level in the vegetation. Three grids were randomly located along a 150-m transect 30 m from the edge of the alfalfa Þeld during each 2-d web survey period, with grid positions being used only once. Aphids and ßies stuck to the traps were counted under a stereomicroscope. SADIE was used to analyze the spatial pattern in the form of ranked count data (for details, see Perry 1998). SADIE techniques have been used widely to study predator and pest distributions at larger scales (Ferguson et al. 2000, 2003, Warner et al. 2003, Winder et al. 2005). Because those studies rely on trap-capture data, results depend on the chosen trap distance (predeÞned scale) and only provide information about patterns at that scale. Nonspatial Statistics. Spider web density and vegetation height were log10-transformed to reduce variance heteroscedasticity (after transformation, Levene tests were F ⫽ 1.497, df ⫽ 4,17; P ⫽ 0.247 [web density]; F ⫽ 2.413; df ⫽ 4,17; P ⫽ 0.089 [vegetation height]). Effects of factors on these variables were analyzed by one-way ANOVA followed by least signiÞcant difference (LSD) post hoc tests. Lycosid activity densities, which did not satisfy the assumption of homogeneous variances after being log-transformed, were analyzed by Kruskal-Wallis ANOVA with subsequent multiple comparisons of ranks. Rates of spider recolonization and plant growth after cutting were calculated as the difference between average spider density or vegetation height within each grid at a speciÞc date and the average at the Þrst date divided
Vol. 36, no. 4
Fig. 2. Availability (total number trapped per grid of sticky traps, mean ⫾ SE) of (a) Aphididae, (b) Diptera, and (c) wolf spider activity density (median ⫾ interquartiles) on each survey date. Different letters indicate statistically signiÞcant differences between dates (LSD for aphid and dipteran prey, multiple comparisons of ranks for wolf spiders). Dotted line indicates cutting date (1 October).
by the number of elapsed days. The relationship between variables was estimated by calculating the Spearman rank-order correlation coefÞcient. StatistiTable 1. Numbers of aphids trapped (N), SADIE index of aggregation (Ia), and probability that Ia differs significantly from randomness (P) on three 4 by 4 grids of horizontally oriented sticky traps each 66 cm apart 14 September
18 October
7 November
Grid no.
N
Ia
P
N
Ia
P
N
Ia
P
1 2 3
13 56 34
1.423 1.359 1.246
0.017 0.022 0.078
39 32 21
0.856 0.950 0.824
0.781 0.559 0.894
20 15 30
0.759 0.989 0.966
0.956 0.466 0.519
Ia ⬎ 1 indicates aggregation; Ia ⫽ 1 indicates a random pattern; Ia ⬍ 1 indicates regularity.
August 2007
BIRKHOFER ET AL.: SPIDER WEB SPATIAL PATTERN IN ALFALFA
805
Table 2. Differences between the five sampling dates in the distribution patterns of spider webs and potential prey (Aphididae and Diptera), activity density of lycosids (potential predators), plant characteristics (height and growth rate), and the rate of web recolonization
Web pattern Aphididae pattern Diptera pattern Web density Lycosid activity density Vegetation ht Web recolonization rate Plant growth rate
14 September
4 October
11 October
18 October
7 November
Regular Aggregated Aggregated/random a ab a -----
------d a d -----
Regular Random Random c bc c a a
Random Random Random c c bc b b
Random Random Random b c b b b
The horizontal dotted line indicates the date that the alfalfa was cut (1 Oct.). The letter a indicates the highest and d the lowest mean for each variable.
cal analyses were performed with Statistica 7.1 (StatSoft, Tulsa, OK). Values are means ⫾ SE; medians are shown with interquartiles. Results Prey Availability and Vegetation Height. Availability of aphid prey (activity density based on sticky traps) differed between dates (F ⫽ 3.86; df ⫽ 4,10; P ⫽ 0.038), with signiÞcantly more aphids being trapped before than immediately after cutting (Fig. 2a; LSD, P ⬍ 0.001). Aphid numbers had recovered to precutting levels 2 wk later (Fig. 2a; LSD, P ⫽ 0.692). Aphids were aggregated before cutting, but were randomly distributed thereafter (Table 1). Availability of Diptera prey also differed between dates (F ⫽ 3.56; df ⫽ 4,10; P ⫽ 0.047), with the highest Diptera activity density occurring 2 wk after cutting (Fig. 2b). The
aggregation of Diptera before cutting was weaker than that of aphids, with only one grid showing a signiÞcant nonrandom distribution (N ⫽ 96, Ia ⫽ 1.465, P ⫽ 0.010). As with aphids, the distribution of available Diptera prey did not deviate signiÞcantly from spatial randomness at all other dates (Table 2). Vegetation height differed between dates (F ⫽ 71.44; df ⫽ 4,17; P ⬍ 0.001), with alfalfa plants not recovering to preharvest heights after 35 d (Fig. 3a; LSD, P ⬍ 0.001). Plant growth rates were highest in the 10 d after cutting, with signiÞcantly slower growth later on (Fig. 3b; LSD, P ⬍ 0.001). Spiders. Web densities differed between most dates (Fig. 3c; F ⫽ 81.93; df ⫽ 4,17; P ⬍ 0.001). Cutting caused a signiÞcant decline, but web densities were higher 1 wk later (LSD, P ⬍ 0.001), remaining at about this level for 2 more wk. Web density had not recovered to preharvest levels 35 d after cutting (LSD, P ⫽
Fig. 3. (a) Vegetation height, (b) plant growth rate after cutting, (c) density of spider webs, and (d) spider recolonization rate after cutting (mean ⫾ SE) on each survey date. Different letters indicate signiÞcant differences between dates (LSD). Dotted line shows cutting date (1 October).
806
ENVIRONMENTAL ENTOMOLOGY
Vol. 36, no. 4
Fig. 4. Univariate O-ring statistic (F) analyzing the spatial pattern of all spider webs up to distances of 66 cm. 95% conÞdence envelopes (E) are calculated by simulations of a heterogeneous Poisson process based on the intensity of the observed pattern. O-ring statistics crossing these envelopes indicate signiÞcant deviation from spatial randomness, with values below the lower envelope indicating spatial regularity and values above the upper envelope showing aggregation. N ⫽ number of replicated 4-m2 plots analyzed on each date. (a) 14 September (N ⫽ 5); (b) 11 October (N ⫽ 4); (c) 18 October (N ⫽ 4); and (d) 7 November (N ⫽ 4).
0.025). Spider recolonization rate differed between dates (F ⫽ 45.33; df ⫽ 2,9; P ⬍ 0.001), being signiÞcantly higher during the 2 wk after cutting than a week later (Fig. 3d; LSD, P ⬍ 0.001) or in November (LSD, P ⬍ 0.001). Plant growth and spider recolonization rates were positively correlated (N ⫽ 12, rs ⫽ 0.87, P ⬍ 0.001). Wolf spider activity density differed between dates, with maximum activity densities before and immediately after cutting (Kruskal-Wallis ANOVA; H ⫽ 45.11; N ⫽ 90; P ⬍ 0.001; Fig. 2b). Pitfall catches of lycosids declined signiÞcantly between 4 and 11 October (multiple comparisons of ranks, P ⫽ 0.005) and never recovered to preharvest densities (multiple comparisons of ranks, P ⫽ 0.001). The univariate spatial analysis of all recorded web locations independent of the size of the web owner indicated regularity at small distances before, and 11 d after, cutting (Fig. 4a and b). Web owners had signiÞcantly fewer neighbors than expected under the null model of a heterogeneous Poisson process at distances up to 18 cm before cutting, and up to 8 cm 11 d after cutting. The distribution did not deviate from randomness at any distance during later survey dates (Fig. 4c and d). A bivariate analysis using a null model with larger species (⬎5 mm) as Þxed locations, and randomly simulating the web locations of smaller spe-
cies (⬍3 mm) around those points, indicated repulsion of smaller web owners by larger individuals at most web distances up to 32 cm (Fig. 5). Discussion Effects of Cutting Alfalfa and Prey Availability on Spider Web Spatial Patterns. Cutting reduced the overall density of web-building spiders, but had no immediate impact on the spatial distribution of webs. In contrast to results from some other studies (Howell and Pienkowski 1971, Thomas and Jepson 1997), webbuilding spider densities had not returned to precutting levels a month after cutting. This reßects the Þnding by Bell et al. (2002) that the date of cutting determines the time necessary for recovery. Our study was carried out in the end of the growing season, when temperatures are declining and after the main peaks of spider abundance in spring and late summer (Culin and Yeargan 1983a, b). The availability of aphids, which are preyed on by web-building spiders, was highest before cutting, when the density of web-building spiders was also highest. Before the alfalfa was cut, aphid availability, measured as activity density by sticky traps, was clumped at a scale of 66 cm (the distance between
August 2007
BIRKHOFER ET AL.: SPIDER WEB SPATIAL PATTERN IN ALFALFA
Fig. 5. Bivariate O-ring statistic (F) on 14 September analyzing the spatial pattern of webs of small species with respect to the distribution of large species, up to distances of 66 cm. This analysis was based on the bivariate null model of independence between two patterns. In this model, the pattern of the large species (⬎5 cm) is Þxed, and the pattern of the small species (⬍3 cm) is assumed to vary at random with respect to the distribution pattern of the large species. 95% conÞdence envelopes (E) are calculated from simulations of a heterogeneous Poisson process based on the intensity of both patterns in each grid. O-ring statistics crossing these envelopes indicate signiÞcant deviation from randomness of the smaller species with respect to the larger, with values below the lower envelope indicating repulsion of smaller individuals through larger neighbors, and values above the upper envelope indicating attraction between the two size classes.
sticky traps). Spider web locations, however, were not aggregated at this scale (the web distribution pattern was measured with the O-ring statistic up to a distance of 66 cm). At 66 cm, webs were distributed at random with respect to each other, and in further contrast to the aggregated pattern exhibited by aphids, webs were regularly dispersed at small distances (⬍20 cm) just before and after cutting. This absence of a correlation between the spatial pattern of aphid prey and spider web distribution differs from the Þnding of Harwood et al. (2003), that web-building spiders were spatially associated with prey in wheat Þelds. If web builders in our alfalfa Þeld were spatially associated with areas of increased prey availability, the spatial scale was larger than that analyzed in our study. A recent molecular study indicates a strong feeding link between decomposer prey (Collembola) and web builders (Linyphiidae) (Agusti et al. 2003). Collembola may be of higher food quality than aphids, which are poor-quality prey for several spider species (Toft 2005). Although it is possible that webs in our alfalfa Þeld were associated with nonaggregated prey from the decomposer subsystem, several studies indicate that mites and springtails aggregate at small scales in response to microhabitat heterogeneity (Rebek et al. 2002, Harwood et al. 2003, Ducarme and Lebrun 2004, Grear and Schmitz 2005). Absence of evidence for congruent small-scale spatial patterns of spider webs and available prey in our study is in line with results from a large-scale spatial
807
analysis on cursorial-predator distribution in alfalfa and soybean agroecosystems (Pearce and Zalucki 2006). The authors found no association between surface- and foliage-dwelling predators and prey at the single analyzed scale (SADIE statistic). We suggest that agroecosystems such as alfalfa, which exhibit high plant coverage and a relatively uniform microclimate, differ from crops with regular seed stands by not showing intense prey aggregations in distinct patches. Microhabitats that are favorable in terms of climate and vegetation structure are distributed more randomly in alfalfa than in wheat Þelds, in which highquality microhabitats display a regular dispersion pattern (Harwood et al. 2001). Effects of Intraguild Interactions on Web Spatial Patterns. Wolf spiders had the highest activity density after cutting, possibly reßecting elevated activity in response to reduced vegetation coverage. Langellotto and Denno (2006) documented that increased habitat complexity reduced rates of intraspeciÞc antagonistic interactions and cannibalism in a wolf spider. In our study, webs were more regularly spaced at small distances than expected under the null model of a heterogeneous Poisson distribution before, and 11 d after, cutting. This pattern was observed in times of high web spider abundance, during and immediately after highest wolf spider activity density, and when recolonization rates of web-building spiders were highest. Small-scale regularity after cutting suggests that reduced habitat complexity in response to cutting contributes to increased rates of intraguild interactions. Plant growth and web-building spider recolonization rates were positively correlated, but regular distributions of web locations were only observed at dates with the most contrasting differences in plant characteristics (i.e., before and after cutting of the alfalfa). Our spatial analysis showed no deviation from randomness later in the season, when web density and vegetation height became more similar between dates (Table 2). We therefore suggest that small-scale spatial regularity is not a feature caused by distinct vegetation characteristics. Because we could not Þnd any small-scale aggregation of webs that resembled the distribution of available prey and because we used a null model that incorporated environmental heterogeneity for spider web analysis, we propose that smallscale regularity resulted from intra- and interspeciÞc interactions within the guild of generalist predators. Because some webs were vacant and many individuals were immature, species determination was often not possible; therefore, we cannot distinguish inter- from intraspeciÞc interactions. Two types of interactions could explain this regular spacing. First, individuals searching for web sites might perceive established webs and avoid building webs nearby. This avoidance behavior may be advantageous if it reduces competition and enhances prey availability for established web owners (Spiller 1986). Second, direct encounters between spiders, including web builders and wolf spiders, could have produced the regularity observed before and after cutting. The assumption that intraguild interactions caused spatial
808
ENVIRONMENTAL ENTOMOLOGY
regularity is supported by the observation that larger web owners had fewer smaller neighbors than expected under spatial randomness at radii up to 32 cm around their webs. Larger web-building linyphiids have a competitive advantage over smaller neighbors (Harwood et al. 2001). From our spatial analysis, we cannot determine whether the observed small-scale regularity is a consequence of predation or avoidance. A mesocosm experiment showed that heterospeciÞc web-building linyphiids from alfalfa Þelds preyed more frequently on spiders under increasing spider density (Harwood and Obrycki 2005). Such intraguild predation in web-building spiders is probably limited to species that forage in excess of their webs (Wise 2006). Our spider community includes some of the species studied by Harwood and Obrycki (2005) and was dominated by linyphiids (67% of all adults), several species of which are known to forage in excess of their webs (Alderweireldt 1994). Therefore, we suggest that spatial regularity at least in part developed though intraguild predation. Denno et al. (2004) documented intraguild predation involving lycosids feeding on a small sheet-web species (Linyphiidae). In our study, spider webs showed small-scale regularity only in times of highest wolf spider activity density, suggesting that predation by lycosids may have inßuenced pattern formation. However, interference competition and predation, including cannibalism among some of the smaller linyphiids (Wise 2006), may also have contributed to the pattern. Samu et al. (1996) increased the density of webs occupied by a linyphiid species by artiÞcially enhancing the number of web sites in a wheat Þeld. The investigators found indications of interference competition for webs at high spider densities and documented tolerance between neighbors in areas with higher numbers of artiÞcial web sites. Our analysis uncovered web distribution patterns not often found in studies conducted at larger spatial scales. Harwood et al. (2001) point out the importance of small-scale analysis of web-building spider distribution, a recommendation supported by our results. Spatial analysis using grid-based count data on a larger scale would have missed the small-scale regularity uncovered in our study. A short temporal scale is also critical for documenting distribution patterns in spiders, particularly smaller species such as many of the linyphiids in our system. Extending the survey period beyond 2 d to 1Ð2 wk would have obscured different pattern-forming processes, because our results suggest that patterns sometimes change within 1 wk. We suggest that the paucity of documented regular dispersion patterns for web-building spiders might at least in part come from spatial analyses that do not use small spatial scales or short sampling periods. Conclusions. This study suggests that intraguild interactions contribute more than prey-mediated aggregation to small-scale spatial patterns of spider webs in alfalfa. We documented regularity at small distances with no similarities between the distribution of spider webs and available prey before cutting. Cutting of alfalfa did not directly inßuence web distribution, be-
Vol. 36, no. 4
cause webs showed regularity before, and 11 d after, cutting. Nevertheless, small-scale regularity disappeared as web densities recovered in the weeks after cutting, an observation deserving future study. Analyzing pattern development over a range of spatial scales, in addition to uncovering factors affecting overall web spider density, is crucial for further reÞnement of conservation biological control strategies involving web-building spiders. Knowledge of spatial requirements at large and small scales and the development of measures to minimize intraguild interactions may increase survival rates of generalist predators and hence their effectiveness in pest control (Provencher and Vickery 1988, Scheu 2001).
Acknowledgments We thank M. Bostrom, J. R. Lensing, and R. T. Bessin for Þeld assistance and/or suggestions during the study; T. Wiegand for assistance with the spatial analysis; and the subject editor and two referees for helpful comments on an earlier version of this manuscript. This research was supported by a grant from the German Research Foundation to K. B. and Kentucky Agricultural Experiment Station Hatch Project KY-008005. This is publication 07-08-046 of the Kentucky Agricultural Experiment Station.
References Cited Agusti, N., S. P. Shayler, J. D. Harwood, I. P. Vaughan, K. D. Sunderland, and W.O.C. Symondson. 2003. Collembola as alternative prey sustaining spiders in arable ecosystems: prey detection within predators using molecular markers. Mol. Ecol. 12: 3467Ð3475. Alderweireldt, M. 1994. Prey selection and prey capture strategies of linyphiid spiders in high-input agricultural Þelds. Bull. Br. Arachnol. Soc. 9: 300 Ð308. Bell, J. R., P. J. Johnson, C. Hambler, A. J. Haughton, H. Smith, R. E. Feber, F. H. Tattersall, B. H. Hart, W. Manley, and D. W. Macdonald. 2002. Manipulating the abundance of Lepthyphantes tenuis (Araneae: Linyphiidae) by Þeld margin management. Agric. Ecosyst. Environ. 93: 295Ð304. Birkhofer, K., J. R. Henschel, and S. Scheu. 2006. Spatial pattern analysis in a territorial spider: evidence for multiscale effects. Ecography 29: 641Ð 648. Bryan, K. M., and S. D. Wratten. 1984. The responses of polyphagous predators to prey spatial heterogeneityÑ aggregation by carabid and staphylinid beetles to their cereal aphid prey. Ecol. Entomol. 9: 251Ð259. Burgess, J. W., and G. W. Uetz. 1982. Social spacing strategies in spiders, pp. 317Ð351. In P. N. Witt and J. S. Rovner (eds.), Spider communication. Princeton University Press, Princeton, NJ. Culin, J. D., and K. V. Yeargan. 1983a. Comparative study of spider communities in alfalfa and soybean ecosystemsÑ foliage-dwelling spiders. Ann. Entomol. Soc. Am. 76: 825Ð 831. Culin, J. D., and K. V. Yeargan. 1983b. Comparative study of spider communities in alfalfa and soybean ecosystemsÑ ground-surface spiders. Ann. Entomol. Soc. Am. 76: 832Ð 838. Dennis, P., M. R. Young, and I. J. Gordon. 1998. Distribution and abundance of small insects and arachnids in relation to structural heterogeneity of grazed, indigenous grasslands. Ecol. Entomol. 23: 253Ð264.
August 2007
BIRKHOFER ET AL.: SPIDER WEB SPATIAL PATTERN IN ALFALFA
Denno, R. F., M. S. Mitter, G. A. Langellotto, C. Gratton, and D. L. Finke. 2004. Interactions between a hunting spider and a web-builder: Consequences of intraguild predation and cannibalism for prey suppression. Ecol. Entomol. 29: 566 Ð577. Ducarme, X., and P. Lebrun. 2004. Spatial microdistribution of mites and organic matter in soils and caves. Biol. Fert. Soils. 39: 457Ð 466. Elliott, N. C., R. W. Kieckhefer, G. J. Michels, Jr., and K. L. Giles. 2002. Predator abundance in alfalfa Þelds in relation to aphids, within-Þeld vegetation, and landscape matrix. Environ. Entomol. 31: 253Ð260. Ferguson, A. W., Z. Klukowski, B. Walczak, J. N. Perry, M. A. Mugglestone, S. J. Clark, and I. H. Williams. 2000. The spatio-temporal distribution of adult Ceutorhynchus assimilis in a crop of winter oilseed rape in relation to the distribution of their larvae and that of the parasitoid Trichomalus perfectus. Entomol Exp. Appl. 95:161Ð171. Ferguson, A. W., Z. Klukowski, B. Walczak, S. J. Clark, M. A. Mugglestone, J. N. Perry, and I. H. Williams. 2003. Spatial distribution of pest insects in oilseed rape: Implications for integrated pest management. Agric. Ecosys. Environ 95: 509 Ð521. Fortin, M. J., and M.R.T. Dale. 2005. Spatial analysis: a guide for ecologists. Cambridge University Press, Cambridge, UK. Fortin, M. J., M.R.T. Dale, and J. Hoef. 2002. Spatial analysis in ecology, pp. 2051Ð2058. In A. El-Shaarawi and W. W. Piegorsch (eds.), Encyclopedia of environmetics. Wiley, Chichester, UK. Grear, J. S., and O. J. Schmitz. 2005. Effects of grouping behavior and predators on the spatial distribution of a forest ßoor arthropod. Ecology 86: 960 Ð971. Gudleifsson, B. E., and B. Bjarnadottir. 2004. Spider (Araneae) populations in hayÞelds and pastures in northern Iceland. J. Appl. Entomol. 128: 284 Ð291. Harwood, J. D., and J. J. Obrycki. 2005. Web-construction behavior of linyphiid spiders (Araneae, Linyphiidae): competition and co-existence within a generalist predator guild. J. Insect Behav. 18: 593Ð 607. Harwood, J. D., K. D. Sunderland, and W.O.C. Symondson. 2001. Living where the food is: web location by linyphiid spiders in relation to prey availability in winter wheat. J. Appl. Ecol. 38: 88 Ð99. Harwood, J. D., K. D. Sunderland, and W.O.C. Symondson. 2003. Web-location by linyphiid spiders: prey-speciÞc aggregation and foraging strategies. J. Anim. Ecol. 72: 745Ð756. Herberstein, M. E. 1998. Web placement in sympatric linyphiid spiders (Arachnida, Araneae): individual foraging decisions reveal inter-speciÞc competition. Acta Oecol. 19: 67Ð71. Hodge, M. A., and A. Storfer-Isser. 1997. ConspeciÞc and heterospeciÞc attraction: a mechanism of web-site selection leading to aggregation formation by web-building spiders. Ethology 103: 815Ð 826. Howell, J. O., and R. L. Pienkowski. 1971. Spider populations in alfalfa, with notes on spider prey and effect of harvest. J. Econ. Entomol. 64: 163Ð168. Janetos, A. C. 1986. Web-site selection: are we asking the right questions?, pp. 9 Ð22. In W. A. Shear (ed.), Spider webs, behavior, and evolution. Stanford University Press, Stanford, CA. Kaston, B. J. 1948. Spiders of Connecticut. Bull. Conn. Geol. Nat. Hist. Survey 70: 1Ð 874. Lang, A. 2003. Intraguild interference and biocontrol effects of generalist predators in a winter wheat Þeld. Oecologia (Berl.) 134: 144 Ð153.
809
Langellotto, G. A., and R. F. Denno. 2006. Refuge from cannibalism in complex-structured habitats: implications for the accumulation of invertebrate predators. Ecol. Entomol. 31: 575Ð581. Marc, P., A. Canard, and F. Ysnel. 1999. Spiders (Araneae) useful for pest limitation and bioindication. Agric. Ecosys. Environ. 74: 229 Ð273. Marshall, S. D. 1997. The ecological determinants of space use by a burrowing wolf spider in a xeric shrubland ecosystem. J. Arid Environ. 37: 379 Ð393. McNett, B. J., and A. L. Rypstra. 2000. Habitat selection in a large orb-weaving spider: vegetational complexity determines site selection and distribution. Ecol. Entomol. 25: 423Ð 432. Nyffeler, M., and K. D. Sunderland. 2003. Composition, abundance and pest control potential of spider communities in agroecosystems: a comparison of European and US Studies. Agric. Ecosys. Environ. 95: 579 Ð 612. Nyffeler, M., W. L. Sterling, and D. A. Dean. 1994. How spiders make a living. Environ. Entomol. 23: 1358 Ð1367. ¨ berg, S., and B. Ekbom. 2006. Recolonisation and distriO bution of spiders and carabids in cereal Þelds after spring sowing. Ann. Appl. Biol. 149:203Ð211. Pearce, S., and M. P. Zalucki. 2006. Do predators aggregate in response to pest density in agroecosystems? Assessing within-Þeld spatial patterns. J. Appl. Ecol. 43: 128 Ð140. Pekar, S. 1999. Side-effect of integrated pest management and conventional spraying on the composition of epigeic spiders and harvestmen in an apple orchard (Araneae, Opiliones). J. Appl. Entomol. 123: 115Ð120. Perry, J. N. 1998. Measures of spatial pattern for counts. Ecology 79: 1008 Ð1017. Perry, J. N., A. M. Liebhold, M. S. Rosenberg, J. Dungan, M. Miriti, A. Jakomulska, and S. Citron-Pousty. 2002. Illustrations and guidelines for selecting statistical methods for quantifying spatial pattern in ecological data. Ecography 25: 578 Ð 600. Pons, X., E. Nunez, B. Lumbierres, and R. Albajes. 2005. Epigeal aphidophagous predators and the role of alfalfa as a reservoir of aphid predators for arable crops. Eur. J. Entomol. 102: 519 Ð525. Provencher, L., and W. Vickery. 1988. Territoriality, vegetation complexity, and biological control: the case for spiders. Am. Nat. 132: 257Ð267. Rebek, E. J., D. B. Hogg, and K. D. Young. 2002. Effect of four cropping systems in the abundance and diversity of epedaphic springtails (Hexapoda: Parainsecta: Collembola) in southern Wisconsin. Environ. Entomol. 31: 37Ð46. Riechert, S. E. 1976. Web-site selection in a desert spider, Agelenopsis aperta (Gertsch). Oikos 27: 311Ð315. Riechert, S. E., and T. Lockley. 1984. Spiders as biologicalcontrol agents. Annu. Rev. Entomol. 29: 299 Ð320. Riechert, S. E., and R. G. Gillespie. 1986. Habitat choice and utilization in web-building spiders, pp. 23Ð 48. In W.A. Shear (ed.), Spider webs, behavior, and evolution. Stanford University Press, Stanford, CA. Rosenheim, J. A., H. K. Kaya, L. E. Ehler, J. J. Marois, and B. A. Jaffee. 1995. Intraguild predation among biological-control agentsÑtheory and evidence. Biol. Control 5: 303Ð 335. Samu, F., K. D. Sunderland, C. J. Topping, and J. S. Fenlon. 1996. A spider population in ßux: selection and abandonment of artiÞcial web-sites and the importance of intraspeciÞc interactions in Lepthyphantes tenuis (Araneae: Linyphiidae) in wheat. Oecologia (Berl.) 106: 228 Ð239. Samu, F., K. D. Sunderland, and C. Szinetar. 1999. Scaledependant dispersal and distribution patterns of spiders in agricultural systems: a review. J. Arachnol. 27: 325Ð332.
810
ENVIRONMENTAL ENTOMOLOGY
Scheu, S. 2001. Plants and generalist predators as links between the below-ground andabove-ground system. Basic Appl. Ecol. 2: 3Ð13. Schmidt, M. H., and T. Tscharntke. 2005. Landscape context of sheetweb spider (Araneae: Linyphiidae) abundance in cereal Þelds. J. Biogeogr. 32: 467Ð473. Spiller, D. A. 1984. Competition between two spider species: experimental Þeld study. Ecology 65: 909Ð919. Spiller, D. A. 1986. InterspeciÞc competition between spiders and its relevance to biological-control by general predators. Environ. Entomol. 15: 177Ð181. Stoyan, D., and H. Stoyan. 1994. Fractals, random shapes and point Þelds. Methods of geometrical statistics. Wiley, Chichester, UK. Symondson, W.O.C., K. D. Sunderland, and M. H. Greenstone. 2002. Can generalist predators be effective biocontrol agents? Annu. Rev. Entomol. 47: 561Ð594. Thomas, C.F.G., and P. C. Jepson. 1997. Field-scale effects of farming practices on linyphiid spider populations in grass and cereals. Entomol. Exp. Appl. 84: 59 Ð 69. Thorbek, P., and T. Bilde. 2004. Reduced numbers of generalist arthropod predators after crop management. J. Appl. Ecol. 41: 526 Ð538. Toft, S. 1987. Microhabitat identity of two species of sheetweb spidersÑÞeld experimental demonstration. Oecologia (Berl.) 72: 216 Ð220. Toft, S. 2005. The quality of aphids as food for generalist predators: implications for natural control of aphids. Eur. J. Entomol. 102: 371Ð383. Toft, S., C. Vangsgaard, and H. Goldschmidt. 1995. Distance methods used to estimate densities of web spiders in cereal Þelds. Acta Jutl. 70: 33Ð 45. Ubick, D. P., P. Paquin, P. E. Cushing, and V. Roth. 2005. Spiders of North America: an identiÞcation manual. American Arachnological Society. Uetz, G. W. 1991. Habitat structure and spider foraging, pp. 325Ð348. In S. Bell, E. D. McCoy, and H. R. Mushinsky (eds.), Habitat structure: the physical arrangement of objects in space. Chapman & Hall, London, UK.
Vol. 36, no. 4
Warner, D. J., L. J. Allen-Williams, A. W. Ferguson, and I. H. Williams. 2000. Pest-predator spatial relationships in winter rape: implications for integrated crop management. Pest Manag. Sci. 56: 977Ð982. Warner, D. J., L. J. Allen-Williams, S. Warrington, A. W. Ferguson, and I. H. Williams. 2003. Mapping, characterisation, and comparison of the spatio-temporal distributions of cabbage stem ßea beetle (Psylliodes chrysocephala), carabids, and Collembola in a crop of winter oilseed rape (Brassica napus). Entomol. Exp. Appl. 109: 225Ð234. Wheeler, A. G. 1973. Studies on Arthropod fauna of alfalfa: 5. Spiders (Araneida). Can. Entomol. 105: 425Ð 432. Wiegand, T., and K. A. Moloney. 2004. Rings, circles, and null-models for point pattern analysis in ecology. Oikos 104: 209 Ð229. Wiens, J. A. 1989. Spatial scaling in ecology. Funct. Ecol. 3: 385Ð397. Wijesinghe, D. K., E. A. John, and M. J. Hutchings. 2005. Does pattern of soil resource heterogeneity determine plant community structure? An experimental investigation. J. Ecol. 93: 99 Ð112. Wildova, R. 2004. Below-ground spatial pattern of rhizomes in a grassland community and its relevance to aboveground spatial pattern. Plant Ecol. 174: 319 Ð336. Winder, L., C. J. Alexander, J. M. Holland, W.O.C. Symondson, J. N. Perry, and C. Woolley. 2005. Predatory activity and spatial pattern: the response of generalist carabids to their aphid prey. J. Anim. Ecol. 74: 443Ð 454. Wise, D. H. 1993. Spiders in ecological webs. Cambridge University Press, Cambridge, UK. Wise, D. H. 2006. Cannibalism, food limitation, intraspeciÞc competition and the regulation of spider populations. Annu. Rev. Entomol. 51: 441Ð 465. Received for publication 21 February 2007; accepted 7 May 2007.
