Competition between introduced and native spiders (Araneae: Linyphiidae)
- First Online:
- Cite this article as:
- Houser, J.D., Ginsberg, H. & Jakob, E.M. Biol Invasions (2014) 16: 2479. doi:10.1007/s10530-014-0679-0
The European sheet-web spider Linyphia triangularis (Araneae: Linyphiidae) has become established in Maine, where it often reaches very high densities. Two lines of evidence from previous work suggest that L. triangularis affects populations of the native linyphiid spider Frontinella communis. First, F. communis individuals are relatively scarce in both forest and coastal habitat where L. triangularis is common, but more common where L. triangularis is at low density. Second, in field experiments, F. communis species are less likely to settle in experimental plots when L. triangularis is present, and F. communis disappears from study plots when L. triangularis is introduced. Here we test two mechanisms that may underlie these patterns. First, we tested whether L. triangularis invades and usurps the webs of F. communis. When spiders were released onto webs of heterospecifics, L. triangularis was more likely to take over or share webs of F. communis than the reverse. We also observed natural takeovers of F. communis webs. Second, we explored the hypothesis that L. triangularis reduces prey availability for native species. We sampled flying prey in areas with L. triangularis and those where it had been removed, and found no effect of spider presence on measured prey density. We also found no effect of prey supplementation on web tenacity in F. communis, suggesting that F. communis movements are not highly dependent on prey availability. We conclude that web takeover is likely more important than prey reduction in driving negative effects of L. triangularis on F. communis.
KeywordsSpidersWeb invasionInterference competitionInvasiveLinyphia triangularis
Invasive predators may harm native predators in a myriad of ways, such as by competing for prey or other resources or by preying upon the native. These effects may be severe: in a thorough review of invasive arthropod predators, Snyder and Evans (2006) note the striking number of cases in which an invasive species led to the precipitous decline or even elimination of native species.
A diverse array of spider species from numerous families have been found outside of their native ranges (e.g., Burger et al. 2001; Garb et al. 2004; Hann 1990; Hogg and Daane 2011a, b; Jakob 1991; Johnson et al. 2012; Kleinteich and Schneider 2011; Paquin et al. 2008; Platnick et al. 2012; Simó et al. 2011; Vink and Dupérré 2011). In many regions, the number of non-native spiders is surprisingly high: for example, non-native spiders were found on 51 of 60 sampling sites in California, and comprised more than 50 % of spiders on eight sites (Burger et al. 2001). It is likely that many additional introduced species remain undocumented, given the paucity of spider experts and the uncharacterized status of the spider fauna of many regions (Kobelt and Nentwig 2008).
One might expect that native spiders are at great risk from introduced spiders, but in fact the nature of these interactions vary widely across species. In some cases, the number of native spiders are positively correlated with introduced spiders, most likely because both native and introduced species flourish under similar ecological conditions. For example, numbers of both native and non-native spiders increased with urban habitat fragmentation in southern California (Bolger et al. 2008), and non-natives occurred where native spiders were most abundant and most diverse (Burger et al. 2001).
In other cases, introduced spiders may harm native spiders through a number of non-exclusive mechanisms. First, a number of spider species invade heterospecific webs (e.g., Cangialosi 1997; Gonzaga et al. 1998; Kerr 2005). Because spider webs are metabolically costly to produce and time consuming to construct (e.g., Ford 1977; Jakob 1991), web sharing and usurpation may represent substantial costs to the original resident of the web. Furthermore, productive web sites may be difficult to find, and the evicted resident relinquishes any information about web-site quality that it may have gathered over multiple days of prey capture (e.g., Jakob 2004; McNett and Rypstra 1997). Whether web invasion is common in introduced spiders is not known. However, the anthrophilic pholcid Holocnemus pluchei, introduced to California from the Mediterranean, is extremely successful and has reached high population densities, and invades webs of heterospecifics such as black widows (Latrodectus mactans) (EMJ, pers. obs.). In contrast, the invasive linyphiid Mermessus trilobatus is not superior to native spiders in its ability to take over webs (Eichenberger et al. 2009).
Second, intraguild predation, where an invasive predator preys on native predators, is well known in other taxa (e.g., Katsanis et al. 2013). Spiders frequently prey upon one another (e.g., Burley et al. 2006; Hodge and Marshall 1996; Polis et al. 1989). The invasive yellow sac spider Cheiracanthiummildei (Miturgidae) leads to the decline of several spider species, most likely through intraguild predation rather than competition for food (Hogg and Daane 2011a).
Third, invasive spiders may compete with natives for prey. However, while spiders can reduce the number of insects in the environment (e.g., Nyffeler and Sunderland 2003), studies of spider competition have often failed to yield clear evidence of food competition (e.g., Horton and Wise 1983; Riechert and Cady 1983). In an influential review, Wise (1993) suggests that while spiders are frequently food limited at the individual level, other factors such as predation, interference competition, and intraguild predation often prevent spider populations from reaching sufficient densities to significantly reduce insect densities. In other cases, spiders that appear to be possible competitors exhibit niche partitioning (Novak et al. 2010). One of several exceptions is Spiller’s (1984a, b) study of the orb weavers Metepeira grinnelli and Cyclosa turbinata, which provided convincing evidence that these species compete for food. Thus, competition for prey is a plausible, albeit unlikely, means by which invasive spiders might harm natives.
Our study species is the European hammock spider, Linyphia triangularis (Araneae: Linyphiidae). First collected in Maine in 1983, it has been reported from every Maine county save one (Jennings et al. 2002). In some areas in Acadia National Park, it reaches densities of up to 12 per m2 (unpublished data). These high densities, coupled with the relatively large size of L. triangularis webs (average 25 × 33 cm; unpublished data), mean that in some places, webs of the invader cover a large portion of the habitat suitable for web-building spiders. Both correlational and experimental evidence suggests that L. triangularis has caused a decline of native linyphiid spiders in coastal Maine. In 4 years of line-transect surveys at Acadia National Park (2003–2006), we found that in two coastal sites and a forested site, the density of L. triangularis was negatively correlated with the density of three native linyphiid species, Frontinella communis, Pityohyphantes subarcticus, and Neriene radiata (Jakob et al. 2011a). Two field experiments suggest that this negative relationship was, at least in part, the result of interactions between L. triangularis and native species. Following experimental release, Pityohyphantes subarcticus were less likely to remain on plots containing high densities of L. triangularis than on plots from which the invader had been removed (Houser 2007). In a second study, Frontinella communis were more likely to settle in plots from which L. triangularis were removed, and established F. communis were more likely to leave plots to which L. triangularis were added (Bednarski et al. 2010). However, the mechanisms underlying these effects were not explored.
Here, we first addressed the hypothesis that web invasion is a mechanism for the apparent decline of the native F. communis at Acadia National Park. In its native range in Europe, L. triangularis is aggressive toward heterospecific linyphiids, invading and usurping their webs and even occasionally preying upon some species (Toft 1987, 1988, 1990). We staged web invasions in the field by releasing L. triangularis and F. communis directly onto heterospecific webs and observing subsequent interactions. We also marked webs of F. communis in an area where L. triangularis were common and observed whether takeover by L. triangularis occurred naturally. In both studies, we looked for occurrences of intraguild predation.
Next, we addressed the hypothesis that L. triangularis affected F. communis through competition for food. We tested whether L. triangularis caused a local reduction in aerial insects by removing spiders from some areas and not others and then measuring insect abundance with sticky traps. In addition, we tested whether F. communis is behaviorally responsive to changes in prey by comparing web-site tenacity in F. communis that received food supplements versus those that did not. Food-limited spiders often follow a win-stay/lose-shift strategy, and stay longer in webs where they have successfully captured prey (reviewed in Jakob et al. 2011b). Increased tenacity in supplemented spiders would provide evidence in accord with the hypothesis that F. communis is sensitive to local prey availability.
Materials and methods
Study area and study species
We conducted our research at the Schoodic Point Section of Acadia National Park, Hancock County, Maine (44.337°N, 68.057°W). Our field sites were along forest margins, bordering the Atlantic coast, or in a powerline cut. Vegetation at our sites included small conifer trees (red and white spruce, Picea rubens and P. glauca, and balsam fir, Abies balsamea) as well as a variety of other evergreen and herbaceous species (e.g., Juniperus spp., Solidago spp., and Vaccinium spp.). Field sites contained high densities of L. triangularis and relatively few large native linyphiids, although native species are abundant in similar habitat at Cobscook Bay State Park (Washington County, Maine), about 70 km NE of our site, where L. triangularis occurs in much lower numbers, and were more abundant in nearby forests away from edges (Jakob et al. 2011a).
Typical of linyphiid spiders, webs of our study species consist of both a horizontal sheet, under which the spider sits, and support strands above, below, and to the side of the sheet anchoring it to the vegetation. The support strands above the web also serve as a barrier that deflects flying insects down onto the horizontal sheet. The webs of the aptly-named bowl-and-doily spider (F. communis) have a bowl-shaped capture sheet with a flat sheet underneath. The catching sheets of L. triangularis webs are usually curved gently downward, but are sometimes convex or saddle-shaped.
We staged interactions in the field by releasing spiders on heterospecific webs during July and August of 2002 and 2003. We collected spiders and used them on the same day at the same field site to reduce handling stress. We estimated the relative size of the released spider (hereafter, the intruder) and the original resident of the web prior to the introduction. We chose a diversity of size classes, representative of those common at our field sites at the time of the experiment. While mature L. triangularis are larger than mature F. communis, in July and August the species overlap in size (Houser 2007). We staged 113 web introductions between L. triangularis and F. communis. Because these spiders are quite delicate and we wanted to minimize handling, we estimated relative size of spiders by eye, classifying one spider as larger than the other if they differed in body length by approximately 10 %, and as the same size if the difference was smaller than 10 %. Spiders did not differ by more than approximately 20 %. Linyphia triangularis was the larger spider in 69 trials, smaller in 23, and similar in 21.
We marked webs with label tape affixed to the foliage. We released spiders onto the underside of the edge of the horizontal sheet of the web and then conducted focal observations of both the intruder and resident for 15 min or until one spider left the vicinity. We chose 15 min because pilot observations showed that most activity occurred within the first 5 min of release, followed by either the departure of one or both of the spiders or long periods of quiescence. We also recorded fighting, defined as physical contact such as grappling with chelicerae or first legs (Jakob 1994), and fleeing, defined as one spider moving rapidly away from the other. If both spiders were in or near the web at the end of the 15 min focal observation period, we stopped focal observations but returned one h later to check their final positions. These additional checks were made for 65 of 113 trials.
We analyzed these data with logistic regression using JMP v10.0. Our independent variables were intruder species (L. triangularis or F. communis), intruder size relative to the resident (larger, the same size or smaller), and the interaction of intruder species × size. We coded the dependent variable, interaction outcome, into three categories: takeover, web-sharing, and intruder leaves. Takeovers included trials in which the intruder remained anywhere in the web (on the sheet or in the supports) and the resident had left the web entirely, and trials in which the intruder was on the catching sheet of the web but the resident had retreated to the supports above or below the web. Web sharing included trials in which both spiders were on the sheet, both were in the supports, and those with the intruder in the supports and the resident on the sheet. Intruder leaves were those trials in which the intruder left the web entirely.
Naturally occurring takeovers
During August 2005, we marked 40 adult female F. communis webs and checked them daily for 2 weeks for evidence of invasion by L. triangularis. At this time in the season, L. triangularis is mature or penultimate (4–8 mm body length). The webs of F. communis are the largest linyphiid webs available in the field at the time that L. triangularis is at or near adulthood, making these webs potentially valuable to L. triangularis.
We noted which spiders were present in or near the web, and also whether modifications had been made to the web. In the laboratory, L. triangularis sometimes modifies the structure of bowl-and-doily webs by removing the “doily” and flattening the “bowl” to be more typical of its own web.
Sticky-trap sampling: effect of L. triangularis on prey abundance
In August 2004, we examined the possibility that L. triangularis reduces the availability of flying prey. We conducted similar experiments in two habitats in which L. triangularis is very abundant: short conifers along forest margins and low coastal vegetation. In both habitats, we randomly assigned plots to two groups: in removal plots, L. triangularis and their webs were removed daily, and in control plots, L. triangularis were not removed.
We monitored insect abundance with prefabricated sticky traps (Ben Meadows Company, Janesville WI, USA), consisting of 7.6 × 12.7 cm cards coated with adhesive on one side. While sticky traps are not perfect mimics of webs, they are likely to provide reasonable comparative data for areas with and without L. triangularis. One day after the start of spider removal, we used a metal binder clip to attach one trap to the vegetation in the center of each plot. Thereafter, traps were checked daily to ensure that at least half of the sticky surface was still available for insect capture.
In early August we assessed the ability of L. triangularis to suppress aerial prey around 20 balsam fir saplings (Abiesbalsamea). We clipped the traps to the top shoot of the tree approximately 1–2 m above the ground and oriented them vertically with the sticky surface pointing westward. This position was intended to capture flying insects in a manner analogous to the vertical barrier threads found above the sheet web of L. triangularis. Traps were collected 6 days after being set out in the field.
In late August, we established 12 circular plots of 2 m diameter in short vegetation (<50 cm high) within 10 m of the rocky coastline, and assigned half to the spider removal treatment. Plots were dominated by blueberry (Vacciniumaugustifolium) and juniper (Juniperus spp.) and contained few native web-building spiders and no large native linyphiids. Due to the lack of vertical relief in the vegetation in these plots, L. triangularis webs consisted of little more than the catching sheet, making a horizontal trap orientation more appropriate. Thus, we placed traps horizontally in the center of plots, facing upward. Traps were collected after 2 days.
We measured insect length using a dissecting microscope for small insects and calipers for larger insects. We estimated insect biomass using several published regression equations relating body length to dry mass (Rogers et al. 1976, 1977; Schoener 1980). Most often these equations are specific to order, though in a few cases separate equations are provided for important families that differ substantially in morphology from others in their order (e.g., Formicidae vs. other Hymenoptera). We were able to identify most specimens larger than 2 mm to the necessary level, excluding a few badly damaged specimens. For those, we used the general length–weight relationship published by Rogers et al. (1976).
For each of the two prey suppression experiments, we measured three outcome variables: count of individual insects, mass calculated using Rogers et al. (1976, 1977), and mass calculated using Schoener (1980). We compared these outcome variables for ten categories of prey: Diptera, Coleoptera, Hemiptera, Homoptera, Hymenoptera (excluding the Formicidae), Lepidoptera, Other taxa, Unknown, Total, and Palatable total. “Other taxa” consisted of specimens that we were able to identify (and thus take advantage of the more specific length–mass models) but whose taxa occurred infrequently on the traps: Acarina, Araneae, Blattaria, Collembola, Formicidae, Opiliones, Psocoptera, and Thysanoptera. “Unknown” consisted of arthropods that we could not identify, either due to damage or small size. “Total” includes all specimens collected. “Palatable total” includes only prey that we have observed being captured by L. triangularis and native linyphiids. This excludes Araneae, Blattaria, Coleoptera, Formicidae, Opiliones, and specimens longer than 8.0 mm or shorter than 1.0 mm.
After normalizing the data using ln(x + 1) transformations, we compared each outcome variable between L. triangularis removal (low L. triangularis density) and control (high L. triangularis density) plots using individual two sample t tests. We also conducted a power analysis for each statistical comparison. We used trap captures from the control group (those containing high L. triangularis density) as the baseline, with effect size defined as a doubling of captures on L. triangularis removal plots compared to control plots (Cohen 1988). We used SYSTAT or Minitab for Windows for analyses.
Effect of prey supplementation on web-site tenacity of native spiders
We examined the potential for increased prey availability to increase web-site tenacity of F. communis. Our study area was in the open habitat created by a powerline cut through the middle of the forested interior of Schoodic Peninsula, an area where native species are relatively numerous and L. triangularis is rare (Jakob et al. 2011a). We selected webs on trees isolated from their neighbors and checked to make sure that no L. triangularis webs were on the same tree. As we found webs, we assigned them alternately to control and experimental groups, and marked each with numbered tape affixed to a nearby branch. We surveyed webs daily, weather permitting, noting the presence or absence of the spider and whether a different species was present. We did not mark the spiders themselves, so if spiders are replaced by conspecifics, our data underestimate spider movement. We could not quantify the number of empty sites a spider might have available to it, but the habitat in this study area did not appear to be saturated; there were many small gaps in the vegetation that appeared to be suitable for spider webs but were empty.
Each day, we visited all the webs and added 4–6 D. melanogaster to each experimental web, similar to the number used in other experiments (Jakob 2004; McNett and Rypstra 1997) and similar to the amount we fed spiders that were successfully reared in the laboratory (Houser, unpublished data) and watched to ensure that the spider captured the prey. We conducted daily surveys for 8 days following the experiment, with the exception of days 4 and 7, where surveys were prevented by rain. On the ninth day, a period of heavy rains and wind began. Web-site tenacity was defined as the number of consecutive days after the start of the experiment that a spider was found at the web site.
We compared web-site tenacity between the supplemented group and the control group using a Mann–Whitney–Wilcoxon test. Power values for these comparisons were estimated based on the relative efficiency of the test (95.5 %) as compared to the analogous t test (Lehmann 1975). We defined the effect size of interest as an increase of 50 % in tenacity in the supplemented group above that seen in the control group, e.g., a control group mean of 4 days would result in an effect size of 2 days (Cohen 1988). We used SYSTAT and Minitab for Windows.
During the 15 min period of continuous observation, we observed very little fighting behavior but frequent fleeing. When L. triangularis was the intruder (N = 64), fighting occurred in four (6 %) contests, F. communis fled in ten (16 %), and L. triangularis fled in five (8 %). When F. communis was the intruder (N = 49), fighting occurred in two (4 %) contests, F. communis fled in eleven (22 %), and L. triangularis fled in five (8 %). In the remainder of the trials, we did not discern any interactions, including signaling via web vibrations, between the spiders. We observed one incident of predation, when an F. communis introduced to the web of a larger L. triangularis was preyed upon by the resident.
Naturally occurring takeovers
In 2 weeks of observation, three of the 40 (7.5 %) marked F. communis webs were taken over by L. triangularis. In one case, an L. triangularis was hanging from the underside of the catching sheet while an F. communis of approximately the same size was in the support silk 10 cm above the sheet. The web form was intermediate between that typical of F. communis and L. triangularis. The next day, the web was entirely typical of L. triangularis except for remnants of the secondary sheet below the catching sheet, and F. communis was no longer present. The L. triangularis readily captured a D. melanogaster we dropped into the web. Four days after takeover, a male L. triangularis was found cohabiting with the resident female. In the second case, an adult female L. triangularis was found in the prey-catching position with F. communis located approximately 4 cm above the sheet in the support silk. The L. triangularis was considerably (~1.5 times) larger. The web was still bowl-shaped, more typical of F. communis. Two days later no spiders were in the web. In the third case, L. triangularis was found alone in a web, and remained there for an additional day.
Sticky-trap estimates of prey density
Effect of L. triangularis removal on flying prey near conifers
Hymenoptera (excluding Formicidae)
Effect of L. triangularis removal on flying prey in coastal vegetation
Hymenoptera (excluding Formicidae)
Effect of prey supplementation on web-site tenacity of native spiders
No significant difference in web-site tenacity was found between supplemented and control F. communis (W = 388.5, P = 0.548, power 0.51). Median tenacity of supplemented F. communis was 4.5 versus 6.0 days for control spiders, opposite the direction predicted if spiders are food limited. The experiment was ended on Day 9 by heavy rain and wind that destroyed many webs.
Our results support the hypothesis that L. triangularis negatively affects the native F. communis via web invasion. When introduced directly onto heterospecific webs, L.triangularis was more adept at web takeover than F. communis and was more likely to remain in an occupied web (Fig. 1). Under natural conditions, L. triangularis sometimes usurped F. communis webs, modified them to resemble their own webs, and used them for prey capture. Web usurpation is likely to be costly for linyphiids, which are not known to be able to forage for insect prey without a web (Toft 1990). Large linyphiid webs are metabolically costly to construct (Ford 1977). Pholcid webs similar in size to our species’ webs represent an energetic investment equivalent to approximately 4 days of foraging effort (Jakob 1991); those data, however, are from the dry Central Valley of California where there are fewer insect prey, and web costs are likely to be recouped more quickly in Maine. Beyond the loss of energy invested into the web itself, spiders evicted from their webs also lose time in searching for a new web site and building a web (e.g., Jakob et al. 2001), and may face increased risk of predation during search.
While size is a well-known correlate of competitive advantage in spiders, both between conspecifics and heterospecifics (e.g., Dodson and Beck 1993; Heiling and Herberstein 1999; Jakob 2004; Toft 1990), we found no effect of relative size on the outcome of staged web invasions between L. triangularis and F. communis in the size pairings we tested: L. triangularis was a significantly stronger competitor regardless of relative size. It is possible that effects of size might be revealed if greater size differences were tested. The native species matures and reproduces earlier in the season than does L. triangularis, so if larger differences in body size than those we tested confer an advantage, a seasonal reversal of competitive advantage is possible.
We have observed intraguild predation only rarely between our study species and only once in the present study, when L. triangularis preyed upon a smaller F. communis introduced into its web. We may have found intraguild predation had we manipulated hunger levels: intraguild predation can be dangerous to the predator, and it is possible that only very hungry spiders may attempt it. In addition, L. triangularis may prey upon younger, smaller F. communis than were studied here.
We found no evidence for exploitative competition for food between L. triangularis and F. communis in either of our experimental approaches to this question. First, removal of the invasive species from areas where it is abundant did not result in increased local prey availability. Sticky traps are not perfect mimics of webs (Eberhard 1990): for example, prey may avoid or be attracted to one and not the other, and prey capture by sheet webs, which are not sticky, depends a great deal on the spider’s behavior. However, our traps did capture a broad variety of insect prey that we have seen spiders consume. A more direct test of exploitative competition would be to compare the prey-capture rates of native spiders in areas of high vs. low L. triangularis density. However, given the scarcity of native spiders where L. triangularis are plentiful in our study sites, native populations would have to be established artificially.
Second, supplementation of the diets of native F. communis did not increase their web-site tenacity as would be expected if natives are food limited, contrary to much previous research on other web-building spider species. Individual spiders are generally likely to follow a win-stay/lose-shift strategy, remaining in a given location when food is more abundant (e.g., Bradley 1993; Jakob 2004; McNett and Rypstra 1997; Nakata and Ushimaru 1999; but see Smallwood 1993). Our experiment was as long and our prey supplements as high as in experiments where supplementation increased tenacity (Jakob 2004; McNett and Rypstra 1997). It is possible that our spiders are not food limited, that the impact of food limitation on web-site tenacity is minor relative to other factors such as agonistic encounters (e.g., see Smallwood 1993), or that food is limiting for spiders at other times of their life cycle (Wise 1975).
Many introduced spiders are anthrophilic and largely restricted to areas near human habitations and agricultural settings (e.g., Nentwig and Kobelt 2010), and evidence to date suggests that introduced spiders are more likely to negatively affect native spiders in simpler, less productive landscapes than in natural landscapes (Burger et al. 2001). However, our research provides encouragement to study non-native spiders in natural areas. Some spider families are more prone than others to establish in natural habitats: for example, in Europe, representatives of Gnaphosidae, Salticidae, and Linyphiidae generally occurred in natural habitats, but most non-native families are synanthropic, (Kobelt and Nentwig 2008). With increased numbers of spiders becoming established outside of their ranges, the potential for harm to the native fauna requires more attention.
We greatly appreciate helpful comments from reviewers of earlier versions of the manuscript: R. Brodie, E. Clotfelter, S. Droege, K. Fite, R. Harrington, S. Hoffmann, J. Podos, S. Partan, D. Pope and P. Sievert. Special thanks to those who assisted with data collection: A. Porter, C. Skow, J. Bednarski, M. Gorski, and V. Johnson. We thank the staff at Acadia National Park, especially D. Manski, B. Connery, E. Pontbriand, and B. Wiedner, for assistance throughout this project and for permission to work at Schoodic. M. Bierman, J. McKenna, and the staff of the Schoodic Education and Research Center provided crucial logistic support. We are grateful to D. Jennings for advice and encouragement throughout this project. We thank the following funding sources: an NIH predoctoral training fellowship to J. D. H, a research grant from the American Arachnological Society to J. D. H, and a grant from the National Park Service and U.S. Geological Survey to J. D. H., E. M. J., and Daniel Jennings. Use of trade or product names does not imply endorsement by the U.S. Government.