Introduction
Ants (Hymenoptera: Formicidae) and various Sternorrhyncha species (Hemiptera) have a long history of close association (Johnson et al. Reference Johnson, Agosti, Delabie, Dumpert, Williams, Von Tschirnhaus and Maschwitz2001). Their relationship is mostly mutualistic whereby the ants benefit from “honeydew” (carbohydrate-rich secretions) produced by the sternorrhynchans and the sternorrhynchans are protected from predators. Four families of Sternorrhyncha commonly exhibit this “trophobiotic” form of mutualism with ants: Aphididae (Aphidoidea) and Coccidae, Pseudococcidae, and Stictococcidae (Coccoidea) (Delabie Reference Johnson, Agosti, Delabie, Dumpert, Williams, Von Tschirnhaus and Maschwitz2001). Most research has focused on this interaction above ground (reviewed by Way Reference Way1963; Buckley Reference Buckley1987; Stadler and Dixon Reference Stadler and Dixon2005; Styrsky and Eubanks Reference Styrsky and Eubanks2007) and has shown that the association of the tending ants and their symbionts ranges from facultative to obligate (Hölldobler and Wilson Reference Hölldobler and Wilson1990; Lapolla et al. Reference LaPolla, Kjer, Schultz and Bischoff2006). Interactions of tending ant species and root-feeding aphids and mealybugs are poorly understood, even though this phenomenon has long been recognized (e.g., Lubbock Reference Lubbock1882).
In a study conducted near Kinsella, Alberta, Canada, Coupe (Reference Coupe2003) suggested that the ants Tapinoma sessile (Say), two species of Lasius F., and a species of Myrmica Latreille were possibly tending unidentified aphids. In this study we followed up on Coupe's observations to determine plant-insect associations, the specificity of interactions, and whether or not particular soil and landscape features provide clues to the presence of ants tending root-feeding sternorrhynchans.
Methods
The study site (centered around 53°01.1′N, 111°32.1′W) is located at the University of Alberta Research Ranch in Kinsella. The ranch is situated in native fescue grassland within aspen parkland, geographically positioned between grasslands to the southeast and boreal forest to the northwest. The landscape consists of knob and kettle topography with short slopes interspersed with intermittent wetlands. Soils are characterized as orthic black and dark brown chernozems (Howitt Reference Howitt1988). The vegetation is dominated by graminoids (Festuca hallii (Vasey) Piper and species of Bromus L., Elymus L., Hesperostipa (Elias) Barkworth, and Poa L. (Poaceae) and species of Carex L. (Cyperaceae)), low-growing forbs and shrubs (Amelanchier alnifolia (Nutt.) Nutt. ex M. Roem. and Rosa arkansana Porter (Rosaceae), Artemisia frigida Willd. (Asteraceae), Elaeagnus commutata Bernh. ex Rydb. (Elaeagnaceae)), and patches of trembling aspen (Populus tremuloides Michx. (Salicaceae)) (S.R. White, University of Alberta, personal communication).
In mid-June and late July 2009, 230 soil cores were taken from a variety of locations at the ranch over an area of approximately 50ha. Nineteen transects of 50or 100m length were selected within this area to maximize plant diversity and variation in aspect and slope, but avoiding aspen stands. A 1 m2 quadrat was sampled every 10m along each transect, resulting in a total of 115 quadrats. To avoid visual bias, each quadrat was blindly placed approximately 1–3 m from the transect, alternating between left and right.
Percent cover of “grasses”, “forbs”, “shrubs” V, and “bare ground” in each quadrat was determined by rough visual estimation. Other measurements taken were soil moisture content (Theta probe type ML2x, Delta-T Devices, Cambridge, England), slope category (none, shallow, steep), aspect, and number of visible entrances to ant nests. Two soil cores approximately 15 cm × 20 cm wide and 15cm deep were collected per quadrat using a spade and a hand shovel, placed in plastic bags, and returned to the laboratory for examination. We attempted to maximize the number of plant species included within each coring site.
Each core was manually broken apart in a white tray and carefully inspected for ants, aphids, and mealybugs. Cores that had gone through ant nests were recognized by the presence of chambers, ant eggs and pupae, and a relatively high abundance of ants. Plants with sternorrhynchans on their roots or root crowns were identified to the lowest level possible with the aid of local botanists. Ants were identified using Wheeler and Wheeler (Reference Wheeler and Wheeler1963), Bolton (Reference Bolton1995), and unpublished keys to Albertan species created by J.G. Sternorrhynchans were identified by H.E.L.M. using reference collections at the Canadian National Collection of Insects, Arachnids and Nematodes, Agriculture and Agri-Food Canada, Ottawa. Ant-sternorrhychan associations were categorized as “direct” (sternorrhynchans found in an ant nest or observed being carried by ants) or “indirect” (sternorrhynchans and ants simply found in the same quadrat or soil core). We looked for evidence of associations among insects, plants, and environmental variables using bivariate correlation tests in SPSS (SPSS Inc. 2008).
Results
A total of 23 of the 87 species of ants reported for Alberta (Glasier Reference Glasier2011) were collected. Eleven of these were not observed in association with sternorrhynchans: Camponotus modoc Wheeler, Formica hewitti Wheeler, F. lasioides Emery, F. limata Wheeler, F. neogagates Viereck, F. neorufibarbis Emery, F. obscuriventris Mayr, F. oreas Wheeler, F. podzolica Francoeur, Leptothorax muscorum (Nylander), and Myrmica brevispinosa Wheeler. The remaining 12 species (Table 1) were considered to be directly or indirectly associated with subterranean Sternorrhyncha. Most of the ants listed in Table 1 were found in association with more than one root-feeding sternorrhynchan species, sometimes in the same ant nest.
Associations of ant species (Formicidae) and Sternorrhyncha.
Note: Letter codes (a–z, aa–ii) represent individual quadrats. Ant-sternorrhynchan pairs with the same letter code came from the same quadrat; boldfaced capital letters denote an observed direct association (either sternorrhynchans found in an ant nest, or ants observed transporting sternorrhynchans); lower-case letters denote co-occurrence in the same quadrat, but no observed direct association, e.g., Anoecia sp. found in direct association with L. pallitarsus (V), but also in indirect association with L. pallitarsus and T. rugalutus in one quadrat (cc), along with F. marginata and G. utricularia (cc).
Eight species of Aphididae were collected, the most common being Forda marginata Koch and a member of the Geoica utricularia (Passerini) species complex. Specimens of Anoecia Koch collected from Elymus roots differ from all described species in this genus but match specimens previously known only from roots of Hordeum jubatum L. (Poaceae) in Winnipeg (see Blackman and Eastop Reference Blackman and Eastop2006). However, intraspecific variation in many species of Anoecia is poorly documented, and we are unsure that this is indeed an undescribed species. Also, a single specimen of Geoica Hart found on Hesperostipa curtiseta (Hitchc.) Barkworth fell outside the documented variation within species of Geoica known to occur in North America. Finally, we collected Pleotrichophorus pseudoglandulosus (Palmer) and Pseudoepameibaphis tridentatae (Wilson). Both are known as aboveground feeders on sage (Artemisia L. (Asteraceae)) (Foottit and Richards Reference Foottit and Richards1993), but we collected them from unidentified graminoid roots and root crowns, respectively.
Four species of Pseudococcidae were collected. One specimen of Heliococcus osborni (Sanders), found on the soil surface and not associated with ants, represents a considerable northern extension of the distribution of this species and the first record for Canada (Kosztarab Reference Kosztarab1996). The other three species were ant-associated. Chnaurococcus trifolii (Forbes) is widespread in North America and has been previously reported from Alberta (Ben-Dov Reference Ben-Dov2010a, citing Ben-Dov Reference Ben-Dov1994 (but no such record exists in Ben-Dov Reference Ben-Dov1994)). Tridiscus sporoboli (Cockerell) was previously known only from New Mexico and Nebraska (Ben-Dov Reference Ben-Dov2010b). Phenacoccus solenopsis Tinsley is a widespread pest of a variety of economically important plants (CAB International 2011); our collection is the first record of a population in Canada under field conditions (as opposed to greenhouse collections; H.E.L.M., personal observation). Phenacoccus solenopsis was the only sternorrhynchan in this study found on non-graminoids (Artemisia frigida (Asteraceae) and Rosa arkansana (Rosaceae)). Graminoid hosts plants for the various pseudococcids were species of Elymus, Festuca, Hesperostipa, Poa, and Carex. There was no evidence that graminoid-feeding sternorrhynchan species were restricted to one host species.
None of the environmental data collected were significantly correlated with ants or subterranean sternorrhynchans (P > 0.10). Although we frequently found ants without sternorrhynchans (Fig. 1), only four samples had root-feeding aphids (Geoica utricularia and Anoecia sp.) without tending ants, and all subterranean pseudococcids (Table 1) were found with ants.
Frequencies of occurrence and co-occurrence of ants and subterranean sternorrhynchans in 115 quadrats at Kinsella Ranch, Alberta, Canada.
Discussion
Although sampling intensity and study-area size were limited, a large number of target species were collected, including new records for Alberta or Canada. This demonstrates the current gap in knowledge of the diversity of subterranean arthropods in Canada. This may be due in part to a lack of taxonomists specializing in these taxa, but the inherent difficulty of locating, sampling, and observing the behaviour of cryptic soil fauna also plays a role. For example, not all ant species collected with sternorrhynchans (notably Myrmica latifrons Starcke, Tapinoma sessile, both species of Temnothorax Mayr, and more surprisingly Lasius coloradensis Wheeler and L. fallax Wilson) were directly observed interacting with their associated sternorrhynchans (Table 1). To our knowledge, no published record exists of a species of Temnothorax tending sternorrhynchans above or below ground, but species of Lasius are generally known to tend sternorrhynchans (Fisher and Cover Reference Fisher and Cover2007). Lasius coloradenis is particularly noted as being entirely subterranean and dependent on symbiotic relationships with root-feeding sternorrhynchans (Hölldobler and Wilson Reference Hölldobler and Wilson1990; Fisher and Cover Reference Fisher and Cover2007).
The lack of observed direct interaction may be a function of sample size or miscategorization of some of the indirect associations. Breaking up of soil cores occasionally made it difficult to identify ant tunnels or nest chambers, and the disturbance caused by sorting may have caused ants to abandon their sternorrhynchan associates. Almost all subterranean sternorrhynchans were found in the presence of ants and, for those species that were occasionally found alone, there were other samples in which they were tended by ants. Perhaps the pattern shown in Figure 1 is caused by a high density of ants in the area, resulting in almost all Sternorrhyncha co-occurring with ants. However, this seems unlikely, as only 60% of all sampled quadrats contained ants above or below ground. We suggest that the very small number of Sternorrhyncha found alone, limited to G. utricularia and Anoecia sp. in 3.4% of all quadrats, were either temporarily not tended by ants or may be the only two species in our study area to not always have an association with ants. This implies that although the association may be facultative for ants, it might be obligatory for the other belowground sternorrhynchans.
Although we had hoped to find environmental cues that would increase the likelihood of our locating ants together with sternorrhynchans, there were no significant correlations with plant cover, soil moisture content, slope, aspect, or presence of visible entrances to ant nests. This implies that either there are none, we did not measure the relevant environmental factors, or our sample size was not adequate. Lastly, as previously noted for some taxa by Vogel and Kindler (Reference Vogel and Kindler1980), none of the sternorrhynchans in our study were found to be specific to a host-plant species.
Biological surveys of belowground ant and sternorrhynchan symbioses are rare and behavioural studies often limited to a few taxa (e.g., Lapolla et al. Reference Lapolla, Cover and Mueller2002; Stuart and Polavarapu Reference Stuart and Polavarapu2002; Kishimoto-Yamada et al. Reference Kishimoto-Yamada, Itioka and Kawai2005), suggesting that such associations may be uncommon. However, our data suggest that they may be more widespread than was previously thought. We feel that this is an overlooked relationship that merits more attention from taxonomists and ecologists.
Acknowledgements
We are grateful to Barry Irving (Department of Renewable Resources, University of Alberta) for generously sharing his knowledge of the natural history of the study site. Dorothy Fabijan, Gord McNickle, Shannon R. White, and Jonathan Bennett (Department of Biological Sciences, University of Alberta) helped with plant identifications. We thank Sonya Widen for field and laboratory assistance. Finally, we gratefully acknowledge the Alberta Conservation Association Grants in Biodiversity for funding.
