Spatiotemporal Variation in Antlion (Neuroptera: Myrmeleontidae) Density and Impacts on Ant (Hymenoptera: Formicidae) and Generalized Arthropod Foraging

Home , Antlion, Myrmeleon, Myrmeleontini, Neuroptera, Solenopsis texana

ECOLOGY AND POPULATION BIOLOGY Spatiotemporal Variation in Antlion (Neuroptera: Myrmeleontidae) Density and Impacts on Ant (Hymenoptera: Formicidae) and Generalized Arthropod Foraging

LLOYD W. MORRISON1

School of Biological Sciences, Section of Integrative Biology andBrackenridgeFieldLaboratory, University of Texas, Austin, TX 78712

Ann. Entomol. Soc. Am. 97(5): 913Ð922 (2004) ABSTRACT Antlion larvae that construct conical pits to capture prey may strongly affect foraging of ants and other arthropods, yet are usually abundant only in sheltered microhabitats. Larval antlion (Myrmeleon crudelis Walker) densitiesincreasedin exposedareas in central Texas in late summer and early autumn of 1998, presumably because of extended dry conditions. I conducted a study to quantify larval antlion pit densitiesin shelteredandexposedareas over time, andto examine the effect of variation in pit density on the foraging activity of ants and other arthropods. Isolated rainfall events decreased pit densities in exposed areas, sometimes to zero, but pit densities returned to high levels as the soil driedout.Pitfall traps at shelteredsites caught signiÞcantly fewer ants andother arthropods inside antlion zones (i.e., areas of high antlion density) than in adjacent areas without antlions. At exposedsites, pitfall traps caught signiÞcantly fewer ants in antlion zones when pits were present (dry conditions) than when they were absent (wet conditions); there was no signiÞcant difference in foraging outside the antlion zones in wet compared with dry conditions. SigniÞcantly fewer ants were caught insideantlion zones at shelteredsites (that were permanent) comparedwith exposedsites (that were transient), although pit densities were similar at both types of sites. Attraction of ants to baits revealedsimilar patterns. Spatiotemporal variation in antlion pit densitiesandthe associated predation risk to ants and other arthropods may result in behavioral modiÞcations of foraging patterns, higher mortality rates, or both.

KEY WORDS antlion zone, Formicidae, Myrmeleon crudelis, Myrmeleontidae, predation risk

ANTS ARE IMPORTANT COMPONENTS of many terrestrial 1981, MacKay 1982, Waller andMoser 1990, Briano et ecosystems. The Formicidae is both taxonomically di- al. 1995, Orr et al. 1995, Porter et al. 1995, Gotelli 1996, verse (an estimated15,000 living species) andgeo- Morrison 1999), yet such organisms have not received graphically widespread (present everywhere but the as much attention as competitors. One reason may be polar regions) (Bolton 1994). Accordingly, a great that the effects of some members of the former group variety of life histories, morphologies, andforaging are more restrictedin space or time. For example, strategies exists. Despite the diversity that character- antlions (Neuroptera: Myrmeleontidae) are ant pred- izes this subfamily, the single most important factor ators that are common in many regions, although most limiting many ant populations is thought to be intra- pit-buildingspecies are foundonly in shelteredloca- or interspeciÞc competition with other ants. Indeed, tions where dry, loose soil makes pit construction many studies have conclusively demonstrated the im- possible (Topoff 1977). Antlion pits may become very portance of competition in structuring ant communi- dense in such areas, and their presence may greatly ties (Wilson 1971, Cole 1983, Fellers 1987, Ward1987, reduce ant foraging activity (Gotelli 1993, 1996). Savolainen andVepsa¨la¨inen 1988, Ho¨lldobler and Wil- These shelteredareas represent a very small propor- son 1990, Porter andSavignano 1990, Morrison 1996, tion of the overall landscape, however. Holway 1999). During extended dry conditions, however, soil in Predators, pathogens, and parasites may also be ca- exposedareas may become suitable for pit construc- pable of limiting ant populations under certain con- tion, andantlion larvae may become more abundant ditions (Ho¨lldobler 1970, Edwards et al. 1974, Feener and widespread, resulting in an increased predation risk for foraging worker ants andother arthropods. Predation risk is known to be an important factor 1 Current address: Center for Medical, Agricultural and Veterinary Entomology, UnitedStates Department of Agriculture/Agricultural inßuencing foraging decisions in animals in general Research Service, P.O. Box 14565, Gainesville, FL 32604 (e-mail ad- (Lima andDill 1990; Sih 1992, 1994; Kats andDill 1998; dress: [email protected]ß.edu). Lima 1998; Lima andBednekoff1999) andants in

0013-8746/04/0913Ð0922$04.00/0 ᭧ 2004 Entomological Society of America 914 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 97, no. 5 particular (e.g., Nonacs 1990, Nonacs andDill 1990). days per week in the autumn of 1998, when pit den- A greater predation risk associated with higher larval sities were highest andmost variable, andat least once antlion densities may result in decreased foraging ac- every 2 wk thereafter. Transects were locatedalong tivity, greater mortality of foragers, or both. wildlife (primarily white-tail deer, Odocoileus virgin- In the summer of 1998, larval antlion densities in ianus [Boddaert]) trails, and the disturbance of the some areas of central Texas reachedunusually high soil by passing animals preventedovergrowth of veg- levels, anda studywas initiatedto elucidatetheir etation throughout the year. effects on ants andother arthropods.The following Six permanent rectangular study plots were estab- questions were addressed: 1) How did antlion pit lishedon 8 August 1998 to compare pit densitiesin densitiesin exposedareas vary over a periodof 1 yr, exposedandshelteredareas. Three plots were located andhow didthesedensitiescompare with those in in sheltered areas, under abandoned structures that shelteredareas? 2) What environmental factor(s) was hadsolidroofs andwere open on one, two, or all sides. correlatedwith variation in antlion pit densitiesin At these plots, antlion pits were abundant in the area exposedareas? 3) Didthe presence of antlion pits protected from falling rain (i.e., immediately under- affect the foraging of ants andother arthropodsin the neath the roof), but rare outside this area. Three plots same way in shelteredandexposedareas? were in exposedareas with mostly bare soil. The ex- posedplots consistedof a concentration of antlion pits surrounded by an area with few antlions. Materials and Methods At each plot, the area of dense antlion pits (i.e., the Antlion larvae (Neuroptera: Myrmeleontidae) are antlion zone) was circumscribedandmeasured.Ant- predators of small arthropods. The construction of lion zones variedfrom 6.5 to 15.0 m 2. A subplot of each funnel-shapedpits to capture prey occurs within the antlion zone was delimited, and all pits in this subplot tribe Myrmeleontini (New 1986), andis characteristic were countedandmeasured.Subplots were estab- of the genus Myrmeleon (Lucas andStange 1981). The lishedin the central portion of the antlion zone, in a nocturnally active, wingedadultsoviposit in substrates region representative of the overall plot. Subplots composedof small, dry,loose particles; anddevelop- rangedfrom 3.2 to 4.3 m 2. A wooden scaffold was ing larvae construct conical pits to capture prey. While placedover the plots to accurately measure pit diam- antlion larvae feed on a diversity of arthropods, ants eters from directly above, without disturbing any of usually comprise the majority of the prey items the pits. All antlion pits in the subplots at all six sites (Topoff 1977). were measuredat Þve intervals: August andDecember This study was conducted from 15 August 1998 to 30 1998; andMarch, June, andAugust 1999. August 1999 on the grounds of the University of TexasÕ Pitfall trapping was conducted on 4 dates (8, 27, 31 Brackenridge Field Laboratory (BFL; Austin, TX). August and8 October 1998) at each of the 6 study BFL contains numerous overhanging limestone cliffs plots. On each date, 10 traps were set out at each plot, andseveral abandonedstructures(without ßoors) 5 inside and 5 outside the antlion zone. Pitfall traps that create ideal antlion habitat, and antlion larvae were plastic vials (3 cm diameter ϫ 5.5 cm deep), and were common in these areas for at least several years containedpropylene glycol as a preservative. The before the initiation of this project (unpublished 3-cm-diameter mouth of the pitfall traps approximated data). Two major habitat types were deÞned for the an average size antlion pit. The traps were placedout purpose of this study: 1) “Sheltered” areas were re- between 16:00 and18:00 andwere left out for 24 h. gions immediately beneath abandoned structures that Baits were set out on 9 and28 August, 1 September, providedprotectionfrom rain anddirectoverhead and9 October 1998. Six baits were usedat each of the sunlight (but not from running water during ßooding, six study plots, three inside and three outside the or sunlight at low angles). The soil was dry, Þne, mostly antlion zone. The baits were Ϸ3-g sections of Oscar free of vegetation, andrepresentedidealantlion hab- Mayer hot dogs(pork andturkey), placedon 5 ϫ 5-cm itat. 2) “Exposed” areas were deÞned as having no solid white plastic cards. Baits in direct sun were shaded. protection (i.e., rock, building materials) within a few Baits were set out between 16:00 and18:00. After 1 h, meters above groundlevel, but most were beneath the number andspecies identitiesof all ants present on tree canopies of varying coverage, which allowedrain the cards were recorded. and some degree of direct, overhead sunlight to pen- Pitfall traps andbaits were always locatedinsidethe etrate. Exposedareas also containedopen patches of subplots where pits were measuredwhen placedin- loose soil, primarily from disturbances (see Discus- side the antlion zone, and away from any visible pits sion). All antlion larvae examinedfrom both sheltered when placedoutsidethe antlion zone. Those placed andexposedsites were Myrmeleon crudelis Walker. At outside were never located Ͼ1 m away from the least 49 species of ants were present at BFL (Morrison boundary of the antlion zone. Baiting trials were al- 2002). ways conductedafterpitfall traps hadbeen collected, Three permanent strip transects were establishedin to prevent recruitment to baits biasing pitfall trap exposedareas of BFL on 15 August 1998 to document catches. More pitfall traps than baits were usedbe- variation in pit density in exposed areas over time. cause some pitfalls were disturbed by nocturnally ac- Each transect was 8 m long and0.5 m wide.Antlion pits tive mammals. Pitfall traps that were disturbed were were countedalong each transect from 15 August 1998 not included in the analyses. Baits, which were left out to 30 August 1999. Counts were conducted several for only 1 h, were never disturbed. September 2004 MORRISON:SPATIOTEMPORAL VARIATION IN ANTLION DENSITY 915

During two of the sample periods (27Ð28 August and31 August-1 September), the soil was dry,and many pits were present inside the exposed antlion zones. During the other two periods (8Ð9 August and 8Ð9 October), the soil in open areas was moist from recent rainfall, andno pits were visible insidethe exposedsites when baits andpitfall traps were placed out. (Antlion larvae may have been present, but simply unable to construct pits in the moist soil.) Pit densities inside the sheltered sites were not obviously affectedby these rainfall events. Thus, two time periods were sampled: with and without recent rainfall. Specimens caught in pitfall traps were sortedto order except for Oligochaeta, Diplopoda, and Gas- tropoda, which were sorted to class. Collembolans were not counted, but recorded as present or absent. The Formicidaewere further sortedto species, andall other taxa to morphospecies. Ants that recruitedto baits were determined to species. Reference specimens of ants have been deposited at BFL. Variation in the distribution of pit sizes was ana- lyzedby ␹2 tests of homogeneity (Daniel 1990). All pits in the three plots for each type of site (sheltered or exposed) were combined. Comparisons of pit size distributions were made between the sheltered and exposedsites in August 1998, andamong the Þve different sampling dates for the sheltered sites. Average Fig. 1. (A) Abundance of antlion pits over 1 yr along the pit size was also evaluatedin the same comparisons, by strip transects. Points represent averages of the three a two-tailed t-test anda one-way analysis of variance transects; error bars indicate 1 SE. (B) Abundance of antlion (ANOVA), respectively. Data were log transformedto pits along each of the three strip transects andrainfall events normalize the distributions. Because antlion larvae of in the late summer andearly fall of 1998. the same size constructedrelatively larger pits in the shelteredareas (unpublisheddata),comparisons of six treatments that hadforaging ants. StatView 5.0.1 pit size and distribution were made to elucidate the (SAS Institute, Cary, NC) was usedfor all statistical relative predation risk to potential prey, rather than to analyses. indicate differences in larval antlion size or age. Three-way ANOVAs were usedto analyze the pit- Results fall trap data. The three factor levels were site (sheltered vs exposed), location (inside vs outside the ant- Antlion pit densities varied greatly over time in lion zone), andmoisture (wet vs dry;i.e., with vs exposedareas. Along the three strip transects, pit den- without recent rainfall). All factors in the model were sities were highest from August to early October 1998 Þxed. Four separate analyses were conducted for: 1) (Fig. 1A). In mid-October 1998, pit densities declined total number of ant species, 2) total number of ant to zero; pits were present at very low densities in the individuals, 3) total number of nonant morphospecies, spring andsummer of 1999. Pits were not censused and 4) total number of nonant individuals. Log trans- before 15 August 1998, but, qualitatively, it was ob- formations (ln[n ϩ 1]) were performedto normalize servedthat in the preceding3 yr, in exposedareas, pit the data before analyses. After the ANOVAs, four densities did not reach the levels observed in the contrasts of treatment means were evaluatedby the summer of 1998 (unpublisheddata). Bonferroni methodof multiple comparisons (Neter et The relatively high pit densities recorded in AugustÐ al. 1985). October 1998 followeda periodof low rainfall that Ant recruitment to baits was measuredas the num- allowedsoil in exposedareas to become dryenough ber of individual workers on the bait cards summed for pit construction. The 30-yr average rainfall in Aus- over all species after 1 h. Ant recruitment was com- tin, Texas, for the 4-mo periodAprilÐJuly is 32.7 cm paredfor the six treatments (outsideshelteredsites, (National Climatic Data Center, Asheville, NC). Ac- dry; outside sheltered sites, wet; outside exposed sites, tual rainfall for AprilÐJuly 1998 was only 10.1 cm (31% dry; outside exposed sites, wet; inside exposed sites, of 30-yr average). Conversely, the relatively low pit dry; inside exposed sites, wet; in which wet and dry densitiesobservedin the summer of 1999 followeda refer to the soil) by a one-way ANOVA. Because no relatively wet AprilÐJuly, in which rainfall was 39.8 cm ants were foundinsidethe shelteredsites in wet or dry (122% of 30-yr average). conditions, these two treatments were not included. A Although densities of antlion pits along the three one-way ANOVA was usedin this analysis because strip transects reachedhigh levels duringAugustÐ interest was on whether recruitment variedamong the October 1998, pit numbers variedgreatly duringthis 916 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 97, no. 5

ANOVAs revealedno three-way interactions for any analysis. For both ant species and ant individuals, the interaction between site andlocation was highly sig- niÞcant. For both nonant morphospecies andnonant individuals, the interaction between location and moisture was highly signiÞcant. The importance of the interaction between site andmoisture variedamong the dependent variables (Table 1). Four contrasts of interest revealedthe following patterns (Table 2; Fig. 3): 1) At shelteredsites, traps caught less where antlion pits were present (inside, wet and dry conditions) than where they were absent (outside, wet and dry conditions). This was true for ant species and individuals as well as nonant mor- Fig. 2. Mean density of antlion pits at the sheltered and phospecies and individuals. 2) Inside antlion zones at exposedsites over 1 yr. Error bars represent 1 SE. exposed sites, more ant species and individuals were caught during moist conditions when antlion pits were absent, than during dry conditions when they were time period. Pit densities always decreased after mea- present. This contrast revealedno signiÞcant differ- surable precipitation, andincreasedback towardtheir ences for nonants. 3) Outside antlion zones at exposed original levels as the soil dried out again (Fig. 1B). sites, there were no signiÞcant differences between There was a strong negative correlation between pit wet and dry conditions for ant species, ant individuals, density(averagedover the three transects) andrain- or nonant morphospecies. More nonant individuals fall (cumulatedover the previous 3 d)( r ϭϪ0.80; F ϭ were caught during dry than wet conditions. 4) Traps 28.29; df ϭ 1, 16; P ϭ 0.0001). After heavy rains in the caught less inside sheltered sites where antlion pits month of October 1998 (31.47 cm, 360% of normal), no were present (wet and dry conditions) than inside pits were foundagain until January 1999 (Fig. 1A). exposedsites where antlion pits were present (dry The absence of pits in the late autumn andearly winter conditions only). This was true for ant species and may have been because of numerous factors, including individuals, and nonant individuals, but not nonant increasedrainfall, lower temperatures, andaccumu- morphospecies. lation of leaf litter. In other words, at sheltered sites there were lower The average density of antlion pits in the sheltered catches of all potential antlion prey, on the basis of plots remainedrelatively high throughout the year, both species richness andnumerical abundance, ranging from 30 to 50 pits per m2 (Fig. 2). The average where antlion pits were present than where they were density of antlion pits in the exposed sites was similar absent. In comparisons of exposedsites, more ants to that for the shelteredsites in August 1998, but were caught inside the antlion zone during moist con- declinedtozero by the December census andre- ditions, when pits were absent, than during dry con- mainednear zero until the following August census, ditions, when pits were abundant. No signiÞcant dif- when Ϸ5 pits per m2 were recorded. This is the same ference was foundfor ants between wet anddry general pattern as foundin the strip transects (com- conditions outside the antlion zones. There were no pare with Fig. 1A). signiÞcant differences for any of the comparisons of In August 1998, an average of 40.4 pits/m2 was re- nonants at the exposedsites except that more nonant cordedfromthe three shelteredsites, whereas an individuals were caught outside the antlion zone when average of 43.0 pits/m2 was foundin the three exposed it was dry. Finally, more ants and nonants (on an sites. The overall distributions of pit sizes were dif- abundance basis only) were caught inside the exposed ferent for the exposedandshelteredsites ( ␹2 ϭ 56.2, sites when pits were present than inside the sheltered df ϭ 24, P ϭ 0.0002, ␹2 test of homogeneity). The sites (which always containedpits). average pit diameter was not signiÞcantly different, Fifteen ant species were capturedin pitfall traps however (27.0 Ϯ 13.5 vs 26.7 Ϯ 10.2 mm [mean Ϯ SD], over the course of this study (Table 3). At sheltered shelteredvs exposed,respectively; P ϭ 0.26, two-tailed sites, Solenopsis invicta Buren was the most common t-test). ant species, being very abundant outside, but rare Over the year, the distribution of pit sizes in all three inside, the antlion zones (20.36 vs 0.13 individuals/ shelteredsites consideredtogethervariedsigniÞ- trap, respectively). At the exposedsites, 3 Pheidole cantly (␹2 ϭ 479.8, df ϭ 116, P Ͻ 0.0001, ␹2 test of species, P. ﬂoridana Emery, P. bicarinata Mayr, and P. homogeneity). Average pit diameter also varied sig- dentata Mayr, were more common than other species, niÞcantly (F ϭ 65.2; df ϭ 4, 2190; P Ͻ 0.0001; one-way both inside and outside the antlion zones. The shel- ANOVA), ranging from a low of 16.2 Ϯ 8.3 mm in teredsites tendedtobe surroundedbyopen, sunny December 1998 to a high of 26.9 Ϯ 13.5 mm in August areas that representedfavorable habitat for S. invicta, 1998. while the exposed areas were more shaded. Overall, A total of 2,283 specimens was recorded from pitfall similar numbers of species were caught outside the traps (not including Collembolans). Overall, ants ac- shelteredareas comparedwith insideandoutsidethe counted for 68% of the individual specimens. The exposedareas (12, 10, and11 species, respectively). In September 2004 MORRISON:SPATIOTEMPORAL VARIATION IN ANTLION DENSITY 917

Table 1. Summary of three-way ANOVA results for the pitfall trap data; four different dependent variables were evaluated

Source df. MS FP Ant species Site 1 11.034 97.40 Ͻ0.0001 Location 1 10.814 95.46 Ͻ0.0001 Moisture 1 0.236 2.08 0.1521 Site ϫ location 1 2.827 24.95 Ͻ0.0001 Site ϫ moisture 1 0.502 4.43 0.0375 Location ϫ moisture 1 0.029 0.25 0.6160 Site ϫ location ϫ moisture 1 0.087 0.77 0.3821 Error 110 12.461 Ant individuals Site 1 20.195 43.58 Ͻ0.0001 Location 1 63.183 136.36 Ͻ0.0001 Moisture 1 0.461 1.00 0.3208 Site ϫ location 1 23.256 50.19 Ͻ0.0001 Site ϫ moisture 1 11.228 24.23 Ͻ0.0001 Location ϫ moisture 1 1.055 2.28 0.1342 Site ϫ location ϫ moisture 1 0.021 0.05 0.8327 Error 110 0.463 Nonant morphospecies Site 1 1.850 10.23 0.0018 Location 1 5.735 31.71 Ͻ0.0001 Moisture 1 1.184 6.55 0.0118 Site ϫ location 1 0.143 0.79 0.3750 Site ϫ moisture 1 0.774 4.28 0.0409 Location ϫ moisture 1 1.582 8.75 0.0038 Site ϫ location ϫ moisture 1 0.172 0.95 0.3316 Error 110 0.181 Nonant individuals Site 1 3.33 8.22 0.0050 Location 1 22.61 55.77 Ͻ0.0001 Moisture 1 11.73 28.93 Ͻ0.0001 Site ϫ location 1 0.580 1.43 0.2343 Site ϫ moisture 1 0.156 0.38 0.5365 Location ϫ moisture 1 7.627 18.81 Ͻ0.0001 Site ϫ location ϫ moisture 1 0.047 0.12 0.7342 Error 110 0.405 contrast, pitfalls inside the sheltered areas captured discoveredbyants andto which recruitment occurred only 4 ant species (Table 3). was similar for all treatments except inside the shel- Twenty-one invertebrate groups (primarily orders; teredantlion zones, where no ants were found(Fig. see Materials and Methods) were caught in pitfall traps 4a). Although pitfall trapping revealedthe species over the course of the study (Table 4). As observed richness of ants to be similar at the shelteredand with ants, a similar taxonomic richness was captured exposedsites, the number of ant species recruiting to outsidethe shelteredareas andinsideandoutsidethe baits was higher at the exposedsites (Fig. 4b). Eight exposedareas, whereas Ϸ30% fewer groups were cap- species of ants were attractedto baits at the exposed turedinsidethe shelteredareas. sites: Crematogaster laeviuscula Mayr, Monomorium The baits conÞrmedthe same general patterns as minimum (Buckley), P. bicarinata, P. ﬂoridana, Phei- the pitfall traps. The total number of baits that were dole dentata, Pachycondyla harpax (F.), S. invicta, and

Table 2. Contrasts of treatment means evaluated by the Bonferroni method of multiple comparisons (Neter et al. 1985)

Ant Ant Nonant Nonant Contrast species individuals morphospecies individuals 1) Antlion pits present vs absent at shelteredsites ** ** ** ** 2) Antlion pits present (dry) vs absent (wet) *** NS NS inside exposed sites 3) Antlion pits absent outside exposed sites: wet NS NS NS ** vs dry 4) Antlion pits present at shelteredsites vs ** ** NS * present at exposedsites

For each signiÞcant contrast, pitfall trap catches of the respective groups were less for the Þrst combination of factor level means speciÞed in the contrast. * , SigniÞcant at family conÞdence coefÞcient of 0.90. ** , SigniÞcant at family conÞdence coefÞcient of 0.99. NS, not signiÞcant. 918 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 97, no. 5

ference in recruitment to baits at any of the six treatments with ants (F ϭ 1.47, df ϭ 107, P ϭ 0.21, one-way ANOVA).

Discussion The larvae of most antlion species usually construct pits in shelteredareas that are protectedfrom direct sun, rain, wind, and disturbances from large animals (Haub 1942, Topoff 1977). The importance of shading from the sun, which results in loweredsoil temperatures, has been emphasizedas a primary mechanism for the observeddistributions(Haub 1942, Klein 1982, Lucas 1989). In the only Þeldmanipulation of the factors underlying antlion zones, Gotelli (1993) con- cludedthatan interaction of temperature andrainfall producedtheobserveddistributionsof M. immacula- tus DeGeer and M. crudelis in central Oklahoma, and that neither factor acting alone wouldbe likely to limit larval distribution. Some species of antlion larvae do construct pits in areas with high soil temperatures (Marsh 1987, Lucas 1989). The temperature at the base of an antlion pit, however, is often much lower than the surface soil temperature (Green 1955, Gotelli 1993). Moreover, antlion larvae are able to regulate their thermal exposure by changing position in the pit, and, at ex- tremely high temperatures, may enter a state of torpor (Green 1955). Additionally, antlion larvae are more likely to construct pits during cooler times of the day (YouthedandMoran 1969, Klein 1982). Thus, although high temperatures affect larval antlion activity, high temperatures per se may not be the most important factor limiting larval antlion distribution in many situations. In fact, Gotelli (1993) foundno effect of temperature on mortality; moreover, larvae trans- plantedto exposed,higher temperature areas gained more biomass than those in shaded areas, presumably because of greater foodavailability. Rainfall is an obvious variable limiting antlion distribution, as larvae can only construct pits in dry sub- strate. In fact, the phenology of some antlion species appears to be an adaptation to regular dry and rainy seasons (GrifÞths 1985). Gotelli (1993) foundsignif- icant effects of moisture on antlion abundance, and hypothesizedthat rainfall events prevent pit construction, after which larvae perish from high temperatures on the soil surface. This scenario may very well char- acterize some antlion zones, depending upon soil composition andsun exposure. The patterns of antlion abundance in the exposed areas documented in this work, however, can be entirely attributedto rainfall events without any interactions with high tempera- Fig. 3. Summary of pitfall trap captures by site, location, tures. Antlion pit abundances decreased dramatically and moisture for (A) ant species, (B) ant individuals, (C) after rainfall events andincreasedto previous levels as nonant morphospecies, and (D) nonant individuals. In ϭ inside the soil driedout(Fig. 2). Moreover, all exposedareas the antlion zone. Out ϭ outside the antlion zone. Dry ϭ dry soil. ϭ were mostly shaded, and antlions were abundant in Wet moist soil from recent rain. Error bars represent 1 SE. the exposedareas in the late summer andearly autumn, when soil temperatures wouldhave been high- Solenopsis texana Emery. Only three species of ants est. were attractedto baits at the shelteredsites: M. min- Although a strong negative correlation between re- imum, P. ﬂoridana, and S. invicta. There was no dif- cent rainfall andantlion pit densitywas foundfor September 2004 MORRISON:SPATIOTEMPORAL VARIATION IN ANTLION DENSITY 919

Table 3. Ant species caught in pitfall traps inside and outside the antlion zones at sheltered and exposed sites

ShelteredExposed Species Inside Outside Inside Outside Aphaenogaster texana Wheeler 0.12 Ϯ 0.55 Brachymyrmex depilis Emery 0.06 Ϯ 0.35 0.30 Ϯ 0.77 1.34 Ϯ 4.78 Cyphomyrmex rimosus (Spinola) 0.04 Ϯ 0.21 Dorymyrmex ﬂavus McCook 0.12 Ϯ 0.55 0.07 Ϯ 0.32 Forelius mccooki (McCook) 0.03 Ϯ 0.17 0.62 Ϯ 1.12 Monomorium minimum (Buckley) 0.88 Ϯ 1.45 0.88 Ϯ 1.88 0.90 Ϯ 1.21 Pachycondyla harpax (F.) 0.06 Ϯ 0.24 0.03 Ϯ 0.17 0.21 Ϯ 0.37 Paratrechina terricola (Buckley) 1.03 Ϯ 2.26 0.07 Ϯ 0.23 Pheidole ﬂoridana Emery 0.04 Ϯ 0.21 1.06 Ϯ 2.25 2.42 Ϯ 4.71 5.00 Ϯ 4.88 Pheidole bicarinata Mayr 0.06 Ϯ 0.24 1.55 Ϯ 4.05 2.21 Ϯ 3.46 Pheidole dentata Mayr 0.64 Ϯ 0.81 2.15 Ϯ 2.80 4.28 Ϯ 4.02 Solenopsis invicta Buren 0.13 Ϯ 0.34 20.36 Ϯ 24.99 0.30 Ϯ 1.05 1.14 Ϯ 1.42 Solenopsis texana Emery 0.43 Ϯ 1.67 0.52 Ϯ 0.68 1.09 Ϯ 2.14 1.21 Ϯ 1.54 Strumigenys louisianae Roger 0.03 Ϯ 0.17 Trachymyrmex turrifex (Wheeler) 0.03 Ϯ 0.17 Number of individuals/trap 0.65 Ϯ 1.70 24.85 Ϯ 22.82 8.88 Ϯ 7.32 17.03 Ϯ 10.04 Total number of ant species/site 4 12 10 11

Numbers are mean Ϯ 1 SD, pooledover wet anddryconditions. autumn of 1998, this study was conducted over a single the exposedareas of this study(the three strip year, andannual patterns of pit densitymay vary, transects as well as the three study plots) were located particularly in wetter years (i.e., 1999). Soil moisture in areas that receivedrelatively frequent disturbance was not quantiÞed during this study, although this from large animals. The bare areas of loose soil inhab- variable may be a goodpredictorof overall antlion pit itedby the antlions were apparently maintainedby density (i.e., in the transect censuses). Soil moisture frequent disturbances from large animals, as areas ad- was not measuredduringthe pitfall trapping because jacent to the animal trails were vegetatedandcon- interest was not on the degree of moisture, but simply tainedlittle suitable habitat for antlion pits. Observa- whether the soil was too wet for pit construction. tions of the strip transects after deer were seen on the Manipulation of soil moisture was not done, but rep- trail revealed that many pits were disturbed, but resents a way of isolating this effect from variables relatedto rainfall events (e.g., temperature, cloud cover). Disturbances that damage or destroy antlion pits (or kill antlions) are another obvious limiting factor. All

Table 4. Invertebrate groups caught in pitfall traps inside and outside the antlion zones at sheltered and exposed sites

ShelteredExposed Taxa Inside Outside Inside Outside Acari XXXX Araneida XXXX Coleoptera XXXX Collembola XXXX Dermaptera X X X Diptera XXXX Hemiptera X X Homoptera X X Hymenoptera (non-Formicidae) XXXX Isopoda X X X Isoptera X X Lepidoptera X X Diplopoda X X Neuroptera X Oligochaeta X Orthoptera X X X Pseudoscorpionida X Psocoptera X X Gastropoda XXXX Thysanoptera X X X Trichoptera X X Fig. 4. Summary of baiting trials by site, location, and Total number of groups 10 15 15 17 moisture for (A) number of baits occupiedand(B) number of ant species at baits. Categories are the same as in Fig. 3. An X represents the presence of the indicated taxa. Error bars represent 1 SE. 920 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 97, no. 5 within several hours were reconstructed(unpub- tion, than when pits were present. The Þnding of no lished data). Thus, while such disturbances result in signiÞcant difference in successful ant foraging out- pit destruction and probably some mortality, such side the antlion zone in wet vs dry conditions dem- frequent disturbances appear necessary to maintain onstrates that the differences documented inside the the appropriate microhabitat for antlion pit construc- zone were not simply because of a variation in abiotic tion, at least in some exposedareas. factors (i.e., relative humidity or soil moisture). At the shelteredsites, pit densitiesremainedrela- Ants andother groups of arthropodssuccessfully tively high throughout the year, although signiÞcant foragedless in, or simply avoidedtravelingacross, differences in the distribution of pit sizes and average permanent antlion zones. Foraging in the transient pit size were documented. These differences may be antlion zones at the exposed sites in dry conditions was attributable to the phenology of M. crudelis, as larger greater than that in the antlion zones at the sheltered instars are relatively more abundant at certain times of sites, however, even though pit densities were similar. the year (Lucas 1989). Another possibility is that al- This suggests that, over the time scale considered, ants though the shelteredsites were protectedfrom falling do not respond completely to temporal changes in rain, several heavy rainfall events occurred, resulting predation risk in transient antlion zones. in runoff that ßoodedpartsof the shelteredsites. This Other factors in addition to predation risk may af- may have resultedin mortality of some larvae or a fect the activity of ant foragers in nature. For example, change in soil moisture that affectedpit construction. most permanent antlion zones are locatedin dry,loose Relatively lower catches of ants andother arthro- soil with little or no vegetation present. This type of pods in pitfall traps inside concentrations of antlion microhabitat probably represents a very poor food pits couldresult from either a decreasein foraging source for ants (or other arthropods) and, thus, is not activity in these areas, an increase in mortality if ants an attractive foraging area. This may be especially true or arthropods are preyed upon by antlions, or both. for antlion zones at the base of cliffs or other objects Fewer foraging ants at baits couldbe similarly inter- that may represent a terminal habitat edge for ant preted.Examination of the groundsurface aroundpits foragers (Gotelli 1993). An advantage of studying ant- in both exposedandshelteredareas resultedin the lion zones under structures open on the sides is that discovery of few exoskeletons. It is likely that a be- areas adjacent to the antlion zone are accessible havioral impact on foraging is the primary mechanism through the zone. Even if there is little foodwithin the at work, although this couldnot be determinedwith zone, ants may be expectedto travel through the zone certainty. Thus, throughout the discussion, I will use to forage on the other side. the term “successful foraging” in interpreting pitfall It is possible that the observedvariation in foraging catches, to indicate that ants or other arthropods were activity between shelteredandexposedsites was af- present in an area, apparently engagedin foraging, and fected to some degree by differences in the ant species that they hadnot fallen prey to antlions. diversity at the 2 types of sites, if some species reacted Both the pitfall and bait data indicate that very little differently than others to this type of predation risk. successful foraging by ants occurredwithin the antlion Nine of the 15 ant species trappedin pitfalls were zones at the shelteredsites. Only 31% (4 of 13) of the present at both types of sites, however, andthe 6 ant species and59% (10 of 17) of the total invertebrate species that were present at only 1 type of site were groups recordedfromthe shelteredsites were found relatively rare (Յ 0.12 individuals per trap for all but inside the antlion zones. In contrast, 91% (10 of 11) of Forelius mccooki [McCook], which reachedan abun- the ant species and79% (15 of 19) of the total inver- dance of 0.62 individuals per trap outside exposed tebrate groups known from the exposedsites were sites). Yet some species (e.g., S. invicta, P. bicarinata) foundinsidethe antlion zones (Tables 3 and4). The were more abundant at 1 type of site than another. antlion zone is a permanent feature of the sheltered Although elucidation of such variation was beyond the sites, anddensitiesof antlion pits were observedto scope of this study, a species-by-species comparison of exhibit relatively little seasonal variation. ant foraging behavior in a common habitat containing The cumulative numbers of both ant species and high densities of antlion pits would be an interesting other taxonomic groups capturedoutsidethe shel- future study. teredantlion zones were similar to those captured Basedon a studyof antlions inhabiting cliff bases, outsidethe exposedantlion zones (Tables 3 and4). Gotelli (1996) suggestedthat generalist ant predators Thus, the lower number of ant species andother in- such as antlions may have community-wide impacts vertebrates caught inside the sheltered antlion zones on the distribution, abundance, and behavior of ants. is not because of a lower species richness in the vi- This conclusion was supportedby a Þndingof strong cinity of the shelteredsites. effects on ant foraging behavior, although the distri- The pitfall data indicate that signiÞcantly more suc- bution of antlions was so spatially restricted(Gotelli cessful foraging by ants occurredwithin the antlion 1993) that the magnitude of the overall effects at the zones of the exposedsites than within those of the community level must have been very small. However, shelteredsites, even though antlion pit densitieswere if antlions (or other predators) are able to increase slightly higher at the exposedsites. Successful foraging their distribution or abundance at certain times, as by ants was signiÞcantly greater within the antlion shown in this study, the overall degree of predation zones of the exposedsites when antlion pits were pressure on the community of ants andother arthro- absent because the soil was too moist for pit construc- pods may be much greater. September 2004 MORRISON:SPATIOTEMPORAL VARIATION IN ANTLION DENSITY 921

A large literature suggests that animals possess the Grifﬁths, D. 1985. Phenology andlarval-adultsize relations ability to assess the relative risk from predators and in the ant-lion Macroleon quinquemaculatus. J. Anim. Ecol. alter their behavior accordingly (Lima and Dill 1990, 54: 573Ð581. Kats andDill 1998, Lima 1998). Until recently, how- Haub, J. G. 1942. The pit-building activities of Ohio ant- ever, the effects of temporal variation in risk on prey lions. Ohio J. Sci. 42: 113Ð116. behavior have been largely ignored. Yet, such tempo- Ho¨lldobler, B. 1970. Steatoda fulva (Theridiidae), a spider that feeds on harvester ants. Psyche 77: 202Ð208. ral variation may be a fundamental driving force be- Ho¨lldobler, B., and E. O. Wilson. 1990. The ants. Harvard hindmuch of the antipredatorbehavior observedin University Press, Cambridge, MA. nature (Lima andBednekoff1999, Sih et al. 2000). If Holway, D. A. 1999. Competitive mechanisms underlying periods characterized by a high risk of predation are the displacement of native ants by the invasive Argentine relatively brief andinfrequent, as in the exposedant- ant. Ecology 80: 238Ð251. lion areas of this study, theory predicts that the ma- Kats, L. B., and L. M. Dill. 1998. The scent of death: che- jority of feeding by prey animals should be allocated mosensory assessment of predation risk by prey animals. to low-risk situations (Lima andBednekoff1999). In Ecoscience 5: 361Ð394. the context of this study, relatively more foraging Klein, B. G. 1982. Pit construction by antlion larvae: inßu- wouldbe expectedto occur at the transient antlion ences of soil illumination andsoil temperature. Bull. NY Entomol. Soc. 90: 26Ð30. zones during wet conditions, when predation risk Lima, S. L. 1998. Stress and decision-making under the risk wouldbe lower. Foraging of ants didappearto be less of predation: recent developments from behavioral, re- during periods of higher predation risk at the exposed productive, and ecological perspectives. Adv. Stud. Be- sites, although foraging still occurred. hav. 27: 215Ð290. Few experiments have addressed temporal varia- Lima, S. L., and L. M. Dill. 1990. Behavioral decisions made tion in predation risk, and more data are needed to under the risk of predation: a review and prospectus. Can. fully understand the effects on prey behavior (Sih et J. Zool. 68: 619Ð640. al. 2000). This antlion/prey arthropodcommunity Lima, S. L., and P. A. Bednekoff. 1999. Temporal variation represents an ideal system to test these ideas at the in danger drives antipredator behavior: the predation risk behavioral level. allocation hypothesis. Am. Nat. 153: 649Ð659. Lucas, J. R. 1989. Differences in habitat use between two pit-building antlion species: causes and consequences. Am. Midl. Nat. 121: 84Ð98. Acknowledgments Lucas, J. R., and L. A. Stange. 1981. Key anddescriptionsto the Myrmeleon larvae of Florida (Neuroptera: Myrme- E. A. Kawazoe provided valuable assistance with this leontidae). Fla. Entomol. 64: 207Ð216. project. L. A. Stange identiÞed the antlion larvae. S. D. Porter, MacKay, W. P. 1982. The effect of predation of western L. A. Stange, andL. E. Gilbert madevaluable comments on widow spiders (Araneae: Theridiidae) on harvester ants the manuscript. This research was funded by a State of Texas (Hymenoptera: Formicidae). Oecologia 53: 406Ð411. Fire Ant Research andManagement AdvisoryAccount Com- Marsh, A. C. 1987. Thermal responses andtemperature tol- mittee grant to L. E. Gilbert. erance of a desert ant-lion larva. J. Therm. Biol. 12: 295Ð 300. Morrison, L. W. 1996. Community organization in a recently assembledfauna: the case of Polynesian ants. Oe- References Cited cologia 107: 243Ð256. Bolton, B. 1994. IdentiÞcation guide to the ant genera of the Morrison, L. W. 1999. Indirect effects of phorid ßy parasi- world. Harvard University Press, Cambridge, MA. toids on the mechanisms of interspeciÞc competition Briano, J., R. Patterson, and H. Cordo. 1995. Long term among ants. Oecologia 121: 113Ð122. studies of the black imported Þre ant (Hymenoptera: Morrison, L. W. 2002. Long-term impacts of an arthropod- Formicidae) infected with a microsporidium. Environ. community invasion by the importedÞre ant, Solenopsis Entomol. 24: 1328Ð1332. invicta. Ecology 83: 2337Ð2345. Cole, B. J. 1983. Assembly of mangrove ant communities: Neter, J., W. Wasserman, and M. H. Kutner. 1985. Applied colonization abilities. J. Anim. Ecol. 52: 349Ð355. linear statistical models. Irwin, Homewood, IL. Daniel, W. W. 1990. Appliednonparametric statistics, 2nd New, T. R. 1986. A review of the biology of Neuroptera ed. PWS-Kent Publishing, Boston, MA. Planipennia. Neurop. Int. Suppl. 1: 1Ð57. Edwards, G. B., J. F. Carroll, and W. H. Whitcomb. 1974. Nonacs, P. 1990. Death in the distance: mortality risk as Stoidis aurata (Araneae: Salticidae), a spider predator of information for foraging ants. Behaviour 112: 23Ð35. ants. Fla. Entomol. 57: 337Ð346. Nonacs, P., and L. M. Dill. 1990. Mortality risk vs. food Feener, D. H., Jr. 1981. Competition between ant species: quality trade-offs in a common currency: ant patch pref- outcome controlledby parasitic ßies. Science 214: 815Ð erences. Ecology 71: 1886Ð1892. 817. Orr, M. R., S. H. Seike, W. W. Benson, and L. E. Gilbert. Fellers, J. H. 1987. Interference andexploitation in a guild 1995. Flies suppress Þre ants. Nature 373: 292Ð293. of woodland ants. Ecology 68: 1466Ð1478. Porter, S. D., and D. A. Savignano. 1990. Invasion of poly- Gotelli, N. J. 1993. Ant lion zones: causes of high-density gene Þre ants decimates native ants and disrupts arthro- predator aggregations. Ecology 74: 226Ð237. podcommunity. Ecology 71: 2095Ð2106. Gotelli, N. J. 1996. Ant community structure: effects of Porter, S. D., R. K. Vander Meer, M. A. Pesquero, S. Cam- predatory ant lions. Ecology 77: 630Ð638. piolo, and H. G. Fowler. 1995. Solenopsis (Hymenop- Green, G. W. 1955. Temperature relations of ant-lion larvae tera: Formicidae) Þre ant reactions to attacks of Pseudac- (Neuroptera: Myrmeleontidae). Can. Entomol. 87: 441Ð teon ßies (Diptera: Phoridae) in Southeastern Brazil. 459. Ann. Entomol. Soc. Am. 88: 570Ð575. 922 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 97, no. 5

Savolainen, R., and K. Vepsa¨la¨inen. 1988. A competition hi- Vander Meer, K. Jaffe, and C. Cedeno (eds.), Applied erarchy among boreal ants: impact on resource partition- myrmecology, a worldperspective. Westview Press, ing andcommunity structure. Oikos 51: 135Ð155. Boulder, CO. Sih, A. 1992. Prey uncertainty andthe balancing of anti- Ward, P. S. 1987. Distribution of the introduced Argentine predator and feeding needs. Am. Nat. 139: 1052Ð1069. ant (Iridomyrmex humilis) in natural habitats of the lower Sih, A. 1994. Predation risk and the evolutionary ecology of Sacramento Valley andits effects on the indigenousant reproductive behavior. J. Fish Biol. 45: 111Ð130. fauna. Hilgardia 55(2): 1Ð16. Sih, A., R. Ziemba, and K. C. Harding. 2000. New insights on Wilson, E. O. 1971. The insect societies. HarvardUniversity how temporal variation in predation risk shapes prey Press, Cambridge, MA. behavior. Trends Ecol. Evol. 15: 3Ð4. Youthed, G. J., and V. C. Moran. 1969. The solar-day activity Topoff, H. 1977. The pit andthe antlion. Nat. Hist. 86: 65Ð71. rhythm of Myrmeleontidlarvae. J. Insect Physiol. 15: Waller, D. A., and J. C. Moser. 1990. Invertebrate enemies 1103Ð1116. andnest associates of the leaf-cutting ant Atta texana (Buckley) (Formicidae, Attini), pp. 255Ð273. In R. K. Received 20 May 2002; accepted 23 February 2004.