Predator population size structure alters consumption of prey from epigeic and grazing food webs

Numerous studies have found that predators can suppress prey densities and thereby impact important ecosystem processes such as plant productivity and decomposition. However, prey suppression by spiders can be highly variable. Unlike predators that feed on prey within a single energy channel, spiders often consume prey from asynchronous energy channels, such as grazing (live plant) and epigeic (soil surface) channels. Spiders undergo few life cycle changes and thus appear to be ideally suited to link energy channels, but ontogenetic diet shifts in spiders have received little attention. For example, spider use of different food channels may be highly specialized in different life stages and thus a species may be a multichannel omnivore only when we consider all life stages. Using stable isotopes, we investigated whether wolf spider (Pardosa littoralis, henceforth Pardosa) prey consumption is driven by changes in spider size. Small spiders obtained > 80% of their prey from the epigeic channel, whereas larger spiders used grazing and epigeic prey almost equally. Changes in prey consumption were not driven by changes in prey density, but by changes in prey use by different spider size classes. Thus, because the population size structure of Pardosa changes dramatically over the growing season, changes in spider size may have important implications for the strength of trophic cascades. Our research demonstrates that life history can be an important component of predator diet, which may in turn affect community- and ecosystem-level processes.


Introduction
Across a wide array of terrestrial and aquatic ecosystems, predators suppress prey densities, which can in turn affect important ecosystem services (Tscharntke et al. 2008). The ability of generalist predators to consume more than one prey resource and switch among prey can stabilize prey populations, and indirectly affect primary production and decomposition. In many food webs, predators link the plant-based, grazing energy channel with the detritus-based, soil-surface epigeic channel by consuming prey from both channels (Polis and Strong 1996;Ward et al. 2015;Wimp et al. 2013;Wolkovich et al. 2014). Here we use the term grazing channel to refer to herbivores that feed on live plants and the term epigeic channel to refer to consumers of plant detritus, its associated microbes, and algae found at the soil surface. Theoretical models show that such channel linkage can stabilize food webs when predators shift between channels in response to changes in relative prey abundance (Rooney et al. 2006;Ward et al. 2015;Wolkovich et al. 2014). However, the stabilizing effect may be considerably weaker if some predator size classes can shift between channels and respond to changes in prey availability, while other size classes cannot, potentially due to negative interactions among size classes of predators (Rudolf 2006). When predators can not only shift between different prey species, but also between prey from different food webs (multichannel omnivory), these different food webs can sustain predator populations at different times of the year, which can increase predator population density and stability (Settle et al. 1996).
Multichannel omnivory by generalist predators can lead to prey suppression in both food webs, with cascading effects on primary producers (Oelbermann and Scheu 2002;Polis and Strong 1996;von Berg et al. 2010) and the detrital food web (Leroux and Loreau 2010;Wise et al. 1999). However, the ability of such multichannel omnivores to effectively suppress prey populations depends on their ability to switch to a new prey resource as it becomes abundant, and constraints such as age or size may hinder their ability to shift. Previous research has often treated related species or even members of the same broad trophic group (such as herbivores or predators) as functionally equivalent. However, enormous diversity exists in how predators respond to prey within a trophic group, among closely related species, and even within the same species (Wimp et al. 2019). Indeed, directly examining such variability may explain why predator initiation of top-down effects and trophic cascades is so variable from one system to the next and at different points in time.
Development-related diet shifts are common among consumers with complex life cycles, such as amphibians, marine invertebrates, and holometabolous insects. Fish and reptiles have simple life cycles, but are frequently gape limited, and their diets change as their mouths enlarge (e.g. Scharf et al. 2000). Spiders are important predators in many arthropod food webs, and external digestion frees them somewhat from gape limitation; spider prey can range from a small fraction of the spider's size to several times its size (e.g. Huseynov 2006). Spiders frequently eat prey from multiple food webs Settle et al. 1996;Perkins et al. 2018), and would seem to be ideally suited to link energy channels as envisioned by theoretical models. However, little is known about the way in which diet changes with spider size (but see Wise 2006 (and references therein); Wise et al. 2006;Oelbermann et al. 2008;Bartos 2011). Life-history diet shifts have rarely been studied in spiders, and even fewer size-related shifts between energy channels have been reported (but see Shimazaki and Miyashita 2005).
A predator's diet can be affected by factors other than the ability to capture desired prey. A predator may evade its own predators, including conspecific cannibals, by modifying its activity schedule, foraging activity, or habitat use (Rudolf 2007). We previously showed that a wolf spider common in salt marshes, Pardosa littoralis (henceforth Pardosa), feeds from both grazing and epigeic energy channels (Wimp et al. 2013). However, whether Pardosa consumption from the two food webs remains constant throughout the season, or changes with spider size, remains unknown. Pardosa are univoltine, with reproduction beginning in early summer. As a result, size structure of the Pardosa population changes substantially over the course of the season (personal observations). While Pardosa are not gape limited, spider size may affect behavior and prey consumption. For example, Pardosa is cannibalistic, with the smallest individuals most likely to become victims (Langellotto and Denno 2006). The layer of dead leaves at the base of marsh grass, known as thatch, offers some refuge from cannibalism (Langellotto and Denno 2006) and any small spiders spending time there would encounter epigeic prey almost exclusively, since grazing herbivores generally occur above the thatch layer . Small Pardosa might, therefore, be unlikely to increase consumption of herbivores in response to an herbivore outbreak and thus not suppress herbivore prey populations, but instead continue to consume epigeic prey even if those prey densities are declining. Here, we address the following research question: does the reliance of Pardosa on grazing and epigeic energy channels change as Pardosa grow? We used stable isotopes of carbon and nitrogen to examine shifts in Pardosa prey use with ontogenetic changes in body size. While we measured nitrogen isotopes to confirm trophic status, carbon isotopes have proven useful in determining the relative importance of dual food webs in the diets of generalist predators (e.g. Wise et al. 2006), because carbon isotopic signatures remain relatively unchanged from food source to consumer (Fry 2006).

Study site and organisms
We conducted this study at a salt marsh near Tuckerton, New Jersey, USA (39° 30.8′ N, 74° 19.0′ W) that is dominated by the cordgrass Spartina alterniflora , now reclassified as Sporobolus alterniflorus). We focused on the dominant species in the live plant and epigeic (algal and detrital) food webs; relationships among these dominant arthropods have been well characterized , 2003Wimp et al. 2013). Pardosa hatch and grow during the summer, overwinter as late-instar juveniles, mature to adults in the spring and begin reproduction in early summer (Dobel et al. 1990). The grazing herbivore assemblage in S. alterniflora is dominated by two congeneric planthopper species, Prokelisia dolus and P. marginata, which constitute upwards of 80% of herbivore biomass (Denno et al. 2000). The planthopper Delphacodes penedetecta and the mirid bug Trigonotylus uhleri rank as the next most abundant herbivores. The mirid bug Tytthus vagus consumes planthopper eggs exclusively, and frequently falls prey to Pardosa.
We focused on three species to represent the marsh epigeic food web because of their abundance at the study marsh, similar to the high relative abundances of the herbivore species listed above: Orchestia grillus, Venezillo parvus, and Saldula interstitialis. The amphipod O. grillus is abundant and feeds on surface epiphyton as well as Spartina detritus (Agnew et al. 2003;Lopez et al. 1977); although amphipods may not be part of a typical terrestrial detrital food web, they are commonly found in systems with a terrestrial-aquatic interface, both freshwater and marine. The isopod V. parvus is less abundant than O. grillus at this marsh, but ingests substantial amounts of Spartina litter (Zimmer et al. 2004). Finally, the saldid bug S. interstitialis is saprophagous in early instars, while older nymphs and adults are active predators feeding on surface and subsurface arthropods, including amphipods (Griesinger and Bauer 1990).

Arthropod samples
Four times during the summer of 2009 we sampled marsh arthropods from six blocks, each composed of three 10-m 2 sample plots. Within each block, the three plots were located in different habitats. One plot was located near the upland border of S. alterniflora with S. patens (this border occurs at mean high water level), a second plot was located in the center of the S. alterniflora meadow, and the third in intermediate form S. alterniflora near a tidal creek. We sampled the entire arthropod community on 16 Jun 2009, 20 Jul 2009, 18 Aug 2009 Sept 2009 using a D-vac suction sampler that we placed in ten different locations within each plot for 3-s periods following the methods of Wimp et al. (2013) and Murphy et al. (2012). We immediately placed the collected arthropods into closed containers with ethyl acetate, then transported them to the laboratory in coolers with dry ice and stored in a − 20 °C freezer until they were processed.
We sorted arthropod samples by species, counted the abundance of each species, cleaned them of foreign debris, and dried them at 60 °C for 3 days. Pardosa were then weighed and assigned to one of six size classes by dry weight (Appendix A). Size classes were chosen to divide the collected Pardosa into roughly equally represented groups.

Stable isotope analysis
We used stable isotopes of carbon to assess the level of multichannel omnivory by Pardosa over the course of the summer, by measuring δ 13 C values of Pardosa and their grazing and epigeic prey. While δ 15 N values are useful for determining the trophic level of an organism and we measured them in this study for this purpose, δ 13 C values reflect changes in food web use. Carbon isotopes have proven useful in determining the relative importance of dual food webs in the diets of generalist predators (Newsome et al. 2007;Wimp et al. 2013;Wise et al. 2006), because carbon isotopic signatures, δ 13 C, remain relatively unchanged from food source to consumer (Fry 2006). Therefore, if the δ 13 C values of the bases of two food webs are sufficiently distinct, a predator's δ 13 C value reflects its relative consumption from the two webs even when intermediate feeding links are unclear (DeNiro and Epstein 1981;Post 2002). In previous work, we found that δ 13 C values in salt marsh grazing and epigeic webs differed substantially for at least part of the summer (Wimp et al. 2013). In the present study, we examined both δ 13 C to examine whether Pardosa diet changes across different size classes, and δ 15 N to confirm trophic position. We had to combine individuals of the smaller size classes to obtain sufficient tissue for stable isotope analysis, but only individuals collected in the same plot on the same day were combined. We then ground samples and obtained their isotopic values as described in Wimp et al. (2013).

Statistical analyses
We performed an ANOVA on Pardosa δ 15 N values across the different collection months to confirm predator status and see how this status might change during the season or depending on the S. alterniflora habitat type (center, intermediate, edge; Appendix B). In all ANOVA and regressions involving Pardosa isotopes, we treated all Pardosa recovered from a plot as repeated measures from that plot. We also examined how Pardosa δ 15 N changed with spider size using a correlation analysis (Appendix B).
To examine how multichannel omnivory was related to the size of Pardosa, we used carbon stable isotopes. Initially, we collected Pardosa from different habitats, but we wanted to determine whether carbon isotopes differed according to habitat or could be combined. Thus, we performed a repeated measures ANOVA by SAS proc mixed (SAS 2002), which showed that habitat within block had no effect on the δ 13 C of grazing prey (F 2,17.1 = 0.03, P = 0.97), epigeic prey (F 2,14.3 = 2.45, P = 0.12), or Pardosa (F 2,12 = 1.32, P = 0.30), and that habitat had no effect on Pardosa size (F 2,15.5 = 1.24, P = 0.32). Thus, the different habitats in each of our transects did not affect δ 13 C isotopic ratios and were combined for further analysis.
We were able to collect a large range of Pardosa sizes only in June and July (Appendix A), so we examined the relationship between δ 13 C and size only during those months. Because we had to combine individuals of the smaller size classes (1-3) to obtain sufficient tissue for stable isotope analysis (explained above), we were not able to collect enough individuals of the smaller size classes to obtain even a single sample for the months of August and September, and we only had a single sample in size class 4 (Appendix A, Figs. A1, A2). We performed separate ANOVA analyses in June and July with size class as a discrete explanatory variable and δ 13 C as the response. We included only size classes for which we had at least five measurements in a given month. Variances were not equal across size classes, so the ANOVA was performed using separate variance estimates for each size class with SAS proc mixed.
We used the IsoError mixing model (Phillips and Gregg 2001) to estimate percent consumption from the two food webs in each of the 4 months, based on δ 13 C values of Pardosa and their grazing and epigeic prey. Separate estimates were made for small Pardosa (size classes 1, 2 and 3), and large Pardosa (size classes 5 and 6). We omitted size class 4 to maximize the contrast between the two size categories and also because this size class had the fewest samples (it was absent or had a single sample in June, August, and September). These models assumed that Pardosa δ 13 C values were 0.5‰ higher than those of their prey, based on studies of congeneric Pardosa species by Oelbermann and Scheu (2002) and Mellbrand and Hamback (2010). IsoError takes into account the amount of data available and variation within those data to calculate confidence intervals for diet estimates.
To examine how prey density and the size of Pardosa might influence their reliance on the grazing and epigeic food webs, we first examined whether the densities of grazing and epigeic prey and the grazer/epigeic prey ratio changed according to month using a one-way ANOVA.
To assess the effect of prey densities and Pardosa size on Pardosa diet (δ 13 C value), we performed regressions separately on data from June and July, the 2 months in which a wide range of Pardosa size classes coexisted. Our response variable was Pardosa δ 13 C, and explanatory variables consisted of log-transformed Pardosa dry weight, grazer density, and epigeic feeder density, as well as all interactions. We removed non-significant terms from the model in a stepwise fashion until only significant terms remained.

Trophic position of Pardosa
Pardosa δ 15 N values changed according to month of study, but not in a consistent way (F 3,155=12.77 , P < 0.0001, Appendix B1). We also found that the δ 15 N values for Pardosa changed according to habitat and were greatest in the intermediate S. alterniflora along tidal creeks relative to S. alterniflora in the center of a patch or along the highelevation habitat edge (F 2,10.3 = 17.93, P = 0.0004, values for other taxa in Appendix B2). Finally, we found that Pardosa δ 15 N values increased with an increase in spider weight (ρ = 0.572, P < 0.0001, Appendix B3).

Multichannel omnivory by Pardosa of different sizes
We found that large Pardosa consume substantial amounts of both grazing and epigeic prey, while small Pardosa primarily consume epigeic prey. The largest Pardosa individuals, those with dry weights greater than 4 mg, had carbon isotopic values generally intermediate between those of grazing and epigeic prey during all 4 months (Fig. 1) indicating that large Pardosa are consistent multichannel omnivores throughout the season. The δ 13 C values for small Pardosa were lower than large Pardosa in both June (F 2,35 = 11.19, P = 0.0002) and July (F 4,50 = 6.20, P = 0.0004), the only months when large and small Pardosa coexisted (Fig. 2). During both June and July, some small Pardosa had δ 13 C values in the same range as large Pardosa, but others had δ 13 C values much lower than those of any large individual (Fig. 1), indicating a much heavier reliance on epigeic prey than grazing prey. Mixing models also confirmed that large and small Pardosa consumed very different diets. Using mixing models, we estimated that diets of large Pardosa (size classes 5 and 6) contained roughly twice as much grazing prey as small Pardosa (size classes 1, 2, and 3). Models estimated that in June, 41% of the carbon in tissues of large Pardosa came from the grazing food channel, while only 19% of carbon in small Pardosa came from that channel (Fig. 3). The 95% confidence intervals for those estimates overlapped by only two percent. Similarly, in July, we estimated that the diets of large Pardosa contained roughly three times as much grazing prey as small Pardosa. Models estimated that in July, 43% of the carbon in tissues of large Pardosa came from the grazing channel, while only 15% of the carbon in small Pardosa came from this channel, and confidence intervals overlapped by less than one percent. In August and September, the fraction of grazing prey in the diets of large Pardosa increased to 60% (Fig. 3). There were too few small Pardosa in those later months to estimate diets.
These changes in Pardosa diet happened even though prey carbon isotopic values were relatively consistent throughout the season. Mean δ 13 C of prey in the grazing channel varied less than 0.33‰ throughout the study. Prey values in the epigeic channel were also very similar in June, August, and September, differing by at most 0.43‰, but δ 13 C in July was 0.73‰ higher than in any other month (Fig. 1). These results contrast with our earlier study (Wimp et al. 2013) in which δ 13 C values for epigeic feeders changed by 3.07‰ over the course of the season and changed in a consistent way; epigeic prey fed primarily on algae early in the season and then switched to the Spartina web later in the season. However, in the present study, epigeic prey continued to use algal resources throughout the growing season (Fig. 1). Such variation in the use of algal detrital webs (e.g. between our last study and this current study) may be driven by the enormous annual variation in benthic microalgal abundance (Van Raalte et al. 1976). Within each sample date, δ 13 C of epigeic prey varied much more than δ 13 C of grazing prey (Fig. 1), possibly reflecting the greater variety of resources at the base of the epigeic web, but did not become more enriched as the season progressed.

Impact of prey density and Pardosa size on multichannel omnivory
Densities of grazers, epigeic feeders, and their log ratio differed significantly among months (grazers F 3, 35.3 = 4.02, P = 0.015; epigeic feeders F 3, 34.9 = 3.03, P = 0.042; ratio F 3, 35.2 = 7.18, P = 0.0007, Appendix C). When grazing and epigeic prey were examined separately in June, Pardosa δ 13 C was significantly correlated with both Pardosa weight (F 1,36.6 = 18.04, P = 0.0001) and grazer density (F 1,12.4 = 17.62, P = 0.0012), but not epigeic feeder density. When examined as a ratio, we found that the ratio of grazing to epigeic prey did not affect Pardosa δ 13 C in June (F 1,11.9 = 2.76, P = 0.12).When grazing and epigeic prey were examined separately in July, Fig. 1 Carbon isotopic values, δ 13 C, of Pardosa (black circles) during each of the 4 months a June, b July, c August, and d September. Gray bands show mean ± standard deviation of δ 13 C for prey taxa, with letters identifying taxa described in the figure legend. Pardosa dry weights represent median values for size classes (described in Appendix A) except for individuals > 4.0 mg (size class 6), in which actual dry weights are used because each data point represents a single Pardosa individual  IsoError mixing model estimates and 95% confidence intervals for percent of carbon in Pardosa tissue obtained from prey in the grazing food web by small (≤ 0.6 mg, size classes 1-3, black triangles) and large (> 1.3 mg, size classes 5-6, black circles) Pardosa during each of the four sample months. Small Pardosa were numerous only in June and July Pardosa δ 13 C was significantly correlated with Pardosa weight (F 1,40.1 = 32.71, P < 0.0001), epigeic density (F 1,7.3 = 13.05, P = 0.008), and their interaction (F 1,40.4 = 27.02, P < 0.0001), but not with grazer density. When examined as a ratio in July, we found that Pardosa δ 13 C was significantly correlated with the grazerto-epigeic ratio (F 1,10.2 = 10.19, P = 0.0093), log Pardosa weight (F 1,44.9 = 6.07, P = 0.018), and their interaction (F 1,43.7 = 13.65, P = 0.0006; Fig. 4). This interaction in July occurred because small Pardosa were relatively unaffected by the grazer/epigeic prey ratio, but the δ 13 C value for larger Pardosa increased with an increase in grazing prey density ( Fig. 4; for each Pardosa individual collected in July, the figure shows an individual's weight (mg), δ 13 C value, and the grazer/epigeic ratio for prey found in the plot where that Pardosa individual was collected). Thus, the larger Pardosa were better able to track changes in prey density than small Pardosa.

Discussion
Spiders are abundant predators that have been shown to link grazing and epigeic webs in natural and managed ecosystems Settle et al. 1996;Perkins et al. 2018), and early season feeding from the epigeic web can increase prey suppression on the grazing food web later in the season (Rypstra et al. 1999;Rypstra and Marshall 2005;Sunderland 1999;Wise et al. 1999). However, the impacts of spiders on top-down control of the grazing food web can be temporally variable (DeBach and Rosen 1991). We found that Pardosa diet changes as individuals grow larger during the season, and that these changes in diet are not due solely to variability in prey population densities.
We found that multichannel omnivory is influenced by spider size; small spiders rely more heavily on the epigeic food web, whereas large spiders remain multichannel omnivores throughout the season yet increase their use of grazing prey. Importantly, these results were not driven by changes in δ 13 C values for grazing or epigeic prey, which remained relatively consistent across months. Additionally, the ratio of grazing to epigeic prey did not change in a consistent way across the season (e.g., more grazing prey as the season progressed), yet large Pardosa obtained more of their carbon from the grazing food web as the season progressed. Our findings here may help to explain why spider initiation of trophic cascades can be temporally variable, since spider size affects their ability to consume grazing prey. During June and July, months when large and small Pardosa coexisted, small Pardosa had lower δ 13 C values than did large Pardosa, indicating heavier dependence on epigeic prey. Mixing models estimated that 40% to 60% of the carbon in tissues of large Pardosa came from the grazing channel, whereas small Pardosa obtained less than 20% of their carbon from that channel. Thus, size structure of the Pardosa population may affect suppression of grazing prey in June and July.
The reliance of small Pardosa on epigeic prey means that the density of epigeic prey early in the growing season could impact the strength of trophic cascades later in the season via changes in Pardosa density. For example, abundant epigeic prey in June and July could lead to especially high densities of small Pardosa, which could exert strong predation pressure on grazing prey as Pardosa grow, thereby increasing chances for trophic cascades later in the season as envisioned by Polis and Strong (1996) and shown in other systems (e.g. Settle et al. 1996). However, a shortage of epigeic prey early in the growing season could depress Pardosa density later in the growing season, which could dampen a late-season trophic cascade even if grazing prey is abundant. Numerous other studies have found variation in the top-down effects of spiders (consumptive or non-consumptive) on herbivores that indirectly affect primary producers across different studies (spiders affect herbivory with subsequent impacts on primary producers Hlivko and Rypstra 2003;Rypstra et al. 1999;Rypstra and Marshall 2005;Sunderland 1999;Wise et al. 1999) (spiders dampen trophic cascades Birkhofer et al. 2008;Finke and Denno 2004;Halaj and Wise 2002;Hogg and Sanders et al. 2011), and also within the same study (e.g. Gratton and Denno 2003;Murphy et al. 2012;Snyder and Wise 2001). Size structure of the spider population and the abundance of early season epigeic prey may thus help us to understand the variable ability of spiders to initiate trophic cascades in the grazing food web.
Why small Pardosa rely more heavily on the epigeic food web is unknown, but there are several possible mechanisms. First, small Pardosa may be physically unable to capture grazing prey, but the wide range of δ 13 C values exhibited by small Pardosa suggests that this is unlikely. Second, there may be a nutritional advantage in epigeic prey for small Pardosa. Other Pardosa species are capable of selecting prey based on nutritional needs (Mayntz et al. 2005) and those needs can change over time (Bressendorff and Toft 2011). However, a nutritional need for epigeic prey would not likely produce the wide range of diets that we observed in small Pardosa. Third, and we believe the most likely explanation, is that the ecology of fear explains why small Pardosa avoid habitats with grazing prey. Grazing prey and epigeic prey are vertically stratified in the Spartina marsh with epigeic prey found near the soil beneath the thatch layer and grazing prey found above the thatch (Denno 1977). Thatch is known to protect small Pardosa not only from cannibalism by larger Pardosa, but also cannibalism by small Pardosa (Langellotto and Denno 2006). Our results show that large Pardosa feed on grazing prey more than small Pardosa and thus we suggest that small Pardosa may avoid the habitat where grazing prey and large Pardosa are found due to the risk of cannibalism. However, behaviors of small Pardosa may be variable, with some bold individuals venturing above the thatch layer, while others remain hidden in the thatch.
Intimidation of potential prey is known to have as great an effect on trophic cascades as direct consumption of prey in many predator-prey systems (Preisser et al. 2005). A number of studies have examined the impacts of nonconsumptive effects on herbivores (reviewed by Finke and Snyder 2010) and detritivores (reviewed by Sitvarin et al. 2016); while we did not investigate non-consumptive effects directly, our results suggest that they may lead to changes in resource use by conspecific predators and that this should be investigated. In both June and July, some small Pardosa had very low δ 13 C, but others had δ 13 C in the same range as large individuals, indicating the same mixture of grazing and epigeic prey. This overlap with large Pardosa could not occur if small individuals were unable to capture grazing prey, but would occur if some especially bold small individuals ventured into the more dangerous habitat above the thatch, or if grazing prey were driven into the thatch to avoid predators. However, small Pardosa may pay a price for increased herbivore consumption in the form of increased cannibalism, as has been shown for other Pardosa species (Buddle et al. 2003;Rickers and Scheu 2005). For example, Samu et al. (1999) found that cannibalism between individuals could be explained by differences in body size for Pardosa agretis; large spiders could cannibalize smaller spiders if they were 2× larger, and if large spiders were 4× larger, then cannibalism was almost always the outcome. In June, our largest spiders (size class 6) were more than 30× larger than the smallest spiders (size class 1), so behavioral modifications to avoid cannibalism seem likely. More importantly, previous studies have found that larger spiders (conspecific or different species) can impact the spatial distribution of smaller spiders, because smaller spiders respond to cues from larger spiders by avoiding those areas (Persons and Rypstra 2001;Folz et al. 2006;Rypstra et al. 2007). If we consider only data from July when we have adequate representation of both large and small spiders from the same plots, we found that the δ 13 C values of large spiders reflected increased use of the more abundant grazing food channel, whereas small spiders were unaffected by this change in prey densities. Thus, unlike the larger spiders, smaller spiders are less capable of shifting to the most abundant prey and we suggest that this may be related to the ecology of fear, specifically fear of becoming prey for larger spiders.
Studies of predator diversity and its effects on prey suppression have largely focused on species richness (Finke and Snyder 2010), but intraspecific interactions, including cannibalism, can alter prey suppression (Takizawa and Snyder 2011, and references therein;Toscano and Griffen 2012). Cannibalism is common among spiders (Wise 2006), thus dietary shifts to avoid cannibalism similar to that of Pardosa may be widespread and have important implications for both food web theory and biological control of pest herbivores. The importance of size structure on diet has been frequently studied in aquatic systems, especially fish, where morphological constraints such as gape limitation alter prey usage (e.g. Scharf et al. 2000). But there is a need to understand how intraspecific interactions may affect the diet choice of predators more broadly to understand when and how predators initiate trophic cascades (Miller and Rudolf 2011). Here we show that size structure can affect diet even when morphological constraints are not present; thus, the ability of multichannel omnivores to fully use different prey resources may be constrained by the size of the predator and not simply the abundance of prey.