2005. The Journal of Arachnology 33:390±397 SIZE DEPENDENT INTRAGUILD PREDATION AND CANNIBALISM IN COEXISTING WOLF SPIDERS (ARANEAE, LYCOSIDAE) Ann L. Rypstra: Department of Zoology, Miami University, 1601 Peck Blvd. Hamilton, Ohio 45011 USA. E-mail: [email protected] Ferenc Samu: Department of Zoology, Plant Protection Institute, Hungarian Academy of Sciences, PO Box 102, Budapest, H-1525 Hungary ABSTRACT. Two species of wolf spider, Hogna helluo (Walckenaer 1837) and Pardosa milvina Hentz 1844 dominate the predatory community on the soil surface of agroecosystems in eastern North America. Although as adults they are very different in size, differences in phenology ensure that they overlap in size at various times during the year. In a laboratory experiment, we explored the propensity of each species to attack and kill the other wolf spider species (intraguild predation), conspeci®cs (cannibalism) or crickets (ordinary predation). Both spiders attacked and killed a broader size range of crickets more quickly than they approached other spiders. We found no differences in Hogna foraging on conspeci®cs or Pardosa, but Pardosa attacked and killed Hogna more readily than conspeci®cs. Because Hogna was so slow in attacking other spiders, their impact as an intraguild predator may be quite small, especially if their approach to crickets is an indication of their predatory tendencies with insects. On the other hand, Pardosa attacked and killed small Hogna as readily as crickets, which suggests they may have an in¯uence on Hogna populations if Hogna young emerge coincident with large juvenile or adult Pardosa. Keywords: Cannibalism, intraguild predation, agrobiont spiders, predator-prey Cannibalism and intraguild predation (IGP) tory tendencies of two species of wolf spider are important to spider communities and have (Araneae, Lycosidae) that coexist on the soil the potential to affect population sizes and/or surface in agricultural ®elds across the eastern species diversity of spiders as well as that of portion of North America. Because the species potential insect prey (Wagner & Wise 1996; differ in size, activity, and phenology, we Hodge 1999; Samu et al. 1999; Finke & Den- wanted to characterize the circumstances un- no 2002; Matasumura et al. 2004; Denno et der which these spiders engaged in cannibal- al. 2004). Predation is a dynamic process, the ism or intraguild predation and compare those outcome of which depends on the relative siz- predatory interactions to attacks on insect es of the predator and prey, their physiological prey. Under controlled laboratory conditions, state, attack strategy and inherent aggressive- we paired a wide size range of individuals ness (Walker et al. 1999; Persons et al. 2001; with conspeci®cs, the other species of spider, Buddle 2002; Balfour et al. 2003; Buddle et or crickets and documented the outcome and al. 2003; Mayntz et al. 2005). Many of these timing of predation. In this way, we hoped to gain a better understanding of the speci®c factors will shift over time both with age and predatory strategy of each of the spider spe- recent experience and thus the relative impor- cies and the relative in¯uence that these spe- tance of cannibalism and/or IGP to foraging cies have on their insect prey, which would individuals, population structure and commu- help us to gain insight into the nature of their nity composition will shift as well (Wagner & co-existence. Wise 1996; Balfour et al. 2003; Buddle et al. 2003). For spiders that coexist, an understand- METHODS ing of the situations under which cannibalism Study species.ÐHogna helluo (Walckenaer and IGP occur is critical to understanding how 1837) and Pardosa milvina Hentz 1844 co- and when they can persist in the same habitat. exist on the soil surface in disturbed riparian In the present study we explore the preda- habitats and agroecosystems throughout the 390 RYPSTRA & SAMUÐPREDATION & CANNIBALISM IN WOLF SPIDERS 391 sity Ecology Research Center (Oxford, Butler County, Ohio, USA) and held in the labora- tory or reared from animals collected at that site. When not involved in experimentation, spiders were housed individually in translu- cent plastic cylindrical containers 8 cm in di- ameter with 5 cm walls with 1±2 cm of damp peat moss covering the bottom. Spiders were watered and fed once or twice weekly on a diet of crickets (Acheta domesticus), fruit ¯ies (Drosophila spp.) and or meal worms (Tene- brio spp.). Containers with spiders were held in an environmental chamber between 23Ð25 8C on a 12:12 L:D cycle at 60±75% humidity. Experimental protocol.ÐSpiders were Figure 1.ÐMean prey to predator mass ratio randomly selected from the laboratory popu- (PPR 6 S.E.) for captured prey vs. the time (min lation and brought to standard hunger levels 6 S.E.) it took the prey to be captured. Trials where by feeding them ad libitum with Drosophila Hogna was predator are indicated by solid squares melanogaster for 2 days. Spiders were then and those where Pardosa was predator are indicated held for 7 days before testing to ensure that by open squares. Speci®c prey types are listed with they were similarly hungry. Spiders were ran- an arrow pointing to the data for that treatment. domly assigned to be paired with conspeci®cs (to monitor cannibalism), heterospeci®cs (to eastern portion of North America (Dondale & monitor intraguild predation) or crickets (to Redner 1990; Marshall & Rypstra 1999; Mar- monitor ordinary predation). Those assigned shall et al. 2002). Pardosa is small (20 mg), to be paired with conspeci®cs were marked active, and can be found at high densities (10± with a drop of acrylic paint on the abdomen 15 per m2) whereas Hogna is large (800 mg), or cephalothorax so that we could identify in- less active, and found at relatively low den- dividuals. All spiders and crickets were sities (1±2 per m2) in soybean ®elds in the weighed and then introduced into a testing midwestern section of the United States (Mar- arena simultaneously. The arenas consisted of shall & Rypstra 1999; Walker et al. 1999; 14 cm diameter Petri dishes with a base of Marshall et al. 2002). Pardosa is an annual dampened plaster of Paris (as in Samu et al. species with a mid-July population peak. Ex- 1999). Animals were allowed to interact in the cept for a relatively short period during which arena for 24 h during which time we recorded the adults and spiderlings co-occur, the size if and when predation occurred. Experiments distribution of Pardosa individuals active in were run in groups that included representa- the ®elds at any given time is fairly narrow tives of all treatments between July 1998 and (Marshall et al. 2002). On the other hand, July 2001. Hogna seems to have a two-year life cycle Statistical analysis.ÐIn order to determine with more stages occurring in the ®elds at the how similar the spiders and insects used in same time (Marshall et al. 2002). Although each treatment were, we compared the mass Hogna are usually larger than Pardosa, be- of predators and prey across all treatments in cause of the variability in their life cycle, it is ANOVAs. In addition, we calculated prey/ possible for large subadult or adult Pardosa predator mass ratio (PPR) by dividing the to coexist with early stages of Hogna. Previ- mass of the prey by the mass of the predator. ous studies have revealed that each species In cases where there was no predation, we readily consumes smaller individuals of the randomly assigned one of the spiders as prey other in the laboratory (Persons et al. 2001; and the other as predator using a coin toss Balfour et al. 2003). Here we explore those algorithm. In order to ensure pairings were predatory interactions systematically across a similar across treatments, PPRs were also broad range of size ratios. compared in an ANOVA. The effects of pred- Both species of spiders were collected from ator species, prey type, and PPR on the fre- corn and soybean ®elds at the Miami Univer- quency of predation were compared using a 392 THE JOURNAL OF ARACHNOLOGY Figure 2.ÐFifty percent lethal mass ratio (LR50) (6 95% con®dence interval) for Hogna (on the left) and Pardosa (on the right). Prey type is on the X-axis. Where treatments are indicated with the same small letter in the middle of each ®gure, the overlap of the 95% con®dence ranges suggests that there was no signi®cant difference. logistic regression analysis. From the logistic between prey types. We de®ned PPRcritical as regression, we determined the PPR at which the maximum PPR value where the differenc- there was a 50% likelihood of a predatory es between predation on two prey types were event (LR50). Differences in LR50s across no longer signi®cantly different (P 5 0.05). treatments were evaluated by comparison of In order to ®nd the PPRcritical, we started by the 95% con®dence intervals. The effects of removing the sample with the highest PPR the same factors (predator species, prey type, and rerunning the statistical test, if it was still and PPR) on the time until predation were signi®cant, we removed the sample with the evaluated using a parametric survival analysis next highest PPR, and ran the test again. We using the Proportional Hazards model. In this continued this process until the P-value as- case pairwise comparisons were made using sociated with any difference was equal to the Bonferroni test with an overall P-value of 0.05. 0.05. Both the logistic regression and survival models were run initially with all interactions RESULTS included. The non-signi®cant interactions Overall there were no differences in the were removed after the ®rst run and the mod- mass of the Hogna, Pardosa, or crickets used els were run again.