Intercontinental union of Argentine ants: behavioral relationships among introduced populations in Europe, North America, and Asia

Many invasive ants, including the Argentine ant Linepithema humile, form expansive supercolonies, within which intraspecific aggression is absent. The behavioral relationships among introduced Argentine ant populations at within-country or within-continent scales have been studied previously, but the behavioral relationships among intercontinental populations have not been examined. The present study investigated the levels of aggression among intercontinental Argentine ant populations by transporting live ants from Europe and California to Japan and conducting aggression tests against Japanese populations. Workers from the dominant supercolonies of Europe and California did not show aggressive behavior toward workers from the dominant supercolony of Japan, whereas they fought vigorously against workers from minor supercolonies. The three massive supercolonies, together with Argentine ants from Macaronesia, may be the largest non-aggressive unit formed by a social insect species in which intraspecific aggression exists. Absence or low levels of aggression at transcontinental scale, which may have derived from low genetic variation, may help introduced Argentine ants maintain expansive supercolonies. The lack of aggression implies possible frequent exchanges of individuals among the intercontinental populations mediated by human activities.     

Impacts of the invasive Argentine ant on native ants and other invertebrates in coastal scrub in south‐eastern Australia

Invasive ants threaten native biodiversity and ecosystem function worldwide. Although their principal direct impact is usually the displacement of native ants, they may also affect other invertebrates. The Argentine ant, Linepithema humile (Dolichoderinae), one of the most widespread invasive ant species, has invaded native habitat where it abuts peri‐urban development in coastal Victoria in south‐eastern Australia. Here we infer impacts of the Argentine ant on native ants and other litter and ground‐dwelling invertebrates by comparing their abundance and taxonomic composition in coastal scrub forest either invaded or uninvaded by the Argentine ant. Species composition of native ants at bait stations and extracted from litter differed significantly between Argentine ant‐invaded and uninvaded sites and this was consistent across years. Argentine ants had a strong effect on epigeic ants, which were either displaced or reduced in abundance. The native ant Rhytidoponera victoriae (Ponerinae), numerically dominant at uninvaded sites, was completely absent from sites invaded by the Argentine ant. However, small hypogeic ants, including Solenopsis sp. (Myrmicinae) and Heteroponera imbellis (Heteroponerinae), were little affected. Linepithema humile had no detectable effect upon the abundance and richness of other litter invertebrates. However, invertebrate group composition differed significantly between invaded and uninvaded sites, owing to the varied response of several influential groups (e.g. Collembola and Acarina). Floristics, habitat structure and measured environmental factors did not differ significantly between sites either invaded or uninvaded by Argentine ants, supporting the contention that differences in native ant abundance and species composition are related to invasion. Changes in the native ant community wrought by Argentine ant invasion have important implications for invertebrate communities in southern Australia and may affect key processes, including seed dispersal.     

Exploitation and interference competition between the invasive Argentine ant,Linepithema humile,and native ant species

Interactions between the invasive Argentine ant, Linepithema humile, and native ant species were studied in a 450-ha biological reserve in northern California. Along the edges of the invasion, the presence of Argentine ants significantly reduced the foraging success of native ant species, and vice versa. Argentine ants were consistently better than native ants at exploiting food sources: Argentine ants found and recruited to bait more consistently and in higher numbers than native ant species, and they foraged for longer periods throughout the day. Native ants and Argentine ants frequently fought when they recruited to the same bait, and native ant species were displaced from bait during 60% of these encounters. In introduction experiments, Argentine ants interfered with the foraging of native ant species, and prevented the establishment of new colonies of native ant species by preying upon winged native ant queens. The Argentine ants' range within the preserve expanded by 12 ha between May 1993 and May 1994, and 13 between September 1993 and September 1994, with a corresponding reduction of the range of native ant species. Although some native ants persist locally at the edges of the invasion of Argentine ants, most eventually disappear from invaded areas. Both interference and exploitation competition appear to be important in the displacement of native ant species from areas invaded by Argentine ants.     

Nest connectivity and colony structure in unicolonial Argentine ants

Unicolonial ant colonies occupy many nests and individuals rarely show aggression across large geographic distances. These traits make it difficult to detect colony structure. Here we identify colony structure at scales of hundreds of square-meters, within an invasive population of unicolonial Argentine ants. In experiments using labeled food, and in a 3-year census of nests and trails, we found that food was shared and nests were linked by trails at distances up to 50 meters. Food was not distributed to all nearby Argentine ant nests, showing that ants tend to share resources within a spatially bounded group of nests. The spatial extent of food sharing increased from winter to summer. Across different habitats and nest densities, nests were consistently aggregated at spatial scales of 3- 4 meters in radius. This suggests that new nests bud from old nests at short distances regardless of local conditions. We suggest that a ‘colony’ of Argentine ants could be defined as a group of nests among which ants travel and share food. In our study population, colonies occupy up to 650 m² and contain as many as 5 million ants. In combination with previous work showing that there is genetic differentiation among nests at similar spatial scales, the results suggest that Argentine ant populations do not function ecologically as single, large supercolonies, but instead as mosaics of smaller, distinct colonies consisting of groups of interacting nests. Received 6 June 2008; revised 30 June 2008; accepted 2 July 2008.     

Effects of interspecific competition between two urban ant species, Linepithema humile and Monomorium minimum, on toxic bait performance

We evaluated the effects of interspecific competition on ant bait performance with two urban pest ants, the Argentine ant, Linepithema humile (Mayr), and the little black ant, Monomorium minimum (Buckley). In a laboratory study, the impact of a solid sulfluramid bait on M. minimum was diminished when L. humile were present, whereas the presence of M. minimum reduced the performance of a liquid fipronil bait against L. humile. Argentine ants were not adversely affected by sulfluramid bait at any time, whereas M. minimum was unaffected by fipronil bait until 14 d of exposure. In field studies measuring diel foraging activity, M. minimum seemed to delay L. humile foraging to food stations by approximately 30 min during summer 2001. However, L. humile subsequently recruited to food stations in very high numbers, thereby displacing M. minimum. L. humile visited food stations over an entire 24-h period, whereas M. minimum was only observed visiting food stations during daylight hours. Adjusting the timing of bait placement in the field may minimize any negative effects of interspecific competition between these two species on toxic bait performance.     

Four mutually incompatible Argentine ant supercolonies in Japan: inferring invasion history of introduced Argentine ants from their social structure

In recent years, highly invasive ant species successively invaded warm regions of Asia. In Japan, the Argentine ant, Linepithema humile, has become established in several coastal regions. This species forms unusual social organizations called supercolonies consisting of numerous mutually non-aggressive nests. We studied the behavioral relationships, similarity of cuticular hydrocarbon profiles (nestmate recognition cue), and genetic relationships among the introduced Argentine ant populations of Japan. The Japanese populations were divided into four behaviorally, chemically, and genetically distinct supercolonies, which may have derived from independent source populations. The result represents the recent trend of increasing invasions of invasive ants to Asia. The discontinuous distribution of one supercolony throughout most of the Japanese range suggests rapid expansion of the supercolony via human-mediated jump dispersal. Meanwhile, localization of the other three supercolonies in Kobe Port provides the first evidence for multiple invasions of distinct supercolonies into a base for international trade.     

Factors governing rate of invasion: a natural experiment using Argentine ants

Predicting the success of biological invasions is a major goal of invasion biology. Determining the causes of invasions, however, can be difficult, owing to the complexity and spatio-temporal heterogeneity of the invasion process. The purpose of this study was to assess factors influencing rate of invasion for the Argentine ant (Linepithema humile), a widespread invasive species. The rate of invasion for 20 independent Argentine ant populations was measured over 4 years in riparian woodlands in the lower Sacramento River Valley of northern California. A priori predictors of rate of invasion included stream flow (a measure of abiotic suitability), disturbance, and native ant richness. In addition, baits were used to estimate the abundance of Argentine ants and native ants at the 20 sites. A multiple regression model accounted for nearly half of the variation in mean rate of invasion (R ² = 0.46), but stream flow was the only significant factor in this analysis. Argentine ants spread, on average, 16 m year⁻¹ at sites with permanent stream flow and retreated, on average, −6 m year⁻¹ at sites with intermittent stream flow. Rate of invasion was independent of both disturbance and native ant richness. Argentine ants recruited to more baits in higher numbers in invaded areas than did native ants in uninvaded areas. In addition, rate of invasion was positively correlated with the proportion of baits recruited to by native ants in uninvaded areas. Together, these findings suggest that abiotic suitability is of paramount importance in determining rate of invasion for the Argentine ant. Received: 16 September 1997 / Accepted: 8 February 1998     

Characterization of microsatellite loci for the Argentine ant,Linepithema humile,and their potential for analysis of colony structure in invading Hawaiian populations

We developed primers for five polymorphic microsatellite loci to analyse the genetic structure of colonies in an invading Argentine ant population located in Haleakala National Park on the island of Maui, Hawaii. Microsatellite loci were isolated using both a polymerase chain reaction‐based and a cloning‐based method. With a range of 3–18 alleles and expected levels of heterozygosity of 0.46–0.77, these loci provide useful markers for the detection of colony and population structure in new or expanding populations of this species.     

The role of abiotic conditions in shaping the long-term patterns of a high-elevation Argentine ant invasion

Analysis of long‐term patterns of invasion can reveal the importance of abiotic factors in influencing invasion dynamics, and can help predict future patterns of spread. In the case of the invasive Argentine ant (Linepithema humile), most prior studies have investigated this species’ limitations in hot and dry climates. However, spatial and temporal patterns of spread involving two ant populations over the course of 30 years at a high elevation site in Hawaii suggest that cold and wet conditions have influenced both the ant's distribution and its rate of invasion. In Haleakala National Park on Maui, we found that a population invading at lower elevation is limited by increasing rainfall and presumably by associated decreasing temperatures. A second, higher elevation population has spread outward in all directions, but rates of spread in different directions appear to have been strongly influenced by differences in elevation and temperature. Patterns of foraging activity were strongly tied to soil temperatures, supporting the hypothesis that variation in temperature can influence rates of spread. Based on past patterns of spread, we predicted a total potential range that covers nearly 50% of the park and 75% of the park's subalpine habitats. We compared this rough estimate with point predictions derived from a degree‐day model for Argentine ant colony reproduction, and found that the two independent predictions match closely when soil temperatures are used in the model. The cold, wet conditions that have influenced Argentine ant invasion at this site are likely to be influential at other locations in this species’ current and future worldwide distribution.     

The worldwide expansion of the Argentine ant

Aim The aim of this study was to determine the number of successful establishments of the invasive Argentine ant outside native range and to see whether introduced supercolonies have resulted from single or multiple introductions. We also compared the genetic diversity of native versus introduced supercolonies to assess the size of the propagules (i.e. the number of founding individuals) at the origin of the introduced supercolonies. Location Global. Methods We used mitochondrial DNA (mtDNA) markers and microsatellite loci to study 39 supercolonies of the Argentine ant Linepithema humile covering both the native (n = 25) and introduced range (n = 14). Results Data from three mitochondrial genes and 13 nuclear microsatellites suggest that the introduced supercolonies studied originated from at least seven founding events out of the native area in Argentina (primary introductions). The distribution of mtDNA haplotypes also suggests that supercolonies in the introduced range each derive from a single source supercolony and that one of these source supercolonies has been particularly successful, being the basis of many introduced populations spread across the world. Comparison of the genetic diversity of supercolonies based on the five most diverse loci also revealed that native and introduced supercolonies have greatly overlapping ranges of diversity, although the genetic diversity is on average less in introduced than in native supercolonies. Main conclusions Both primary introductions (from the native range) and secondary introductions (from sites with established invasive supercolonies) were important in the global expansion of the Argentine ant. In combination with the similar social organization of colonies in the native and introduced range, this indicates that invasiveness did not evolve recently as a unique and historically contingent event (e.g. reduction of genetic diversity) in this species. Rather, native L. humile supercolonies have characteristics that make them pre‐adapted to invade new – and in particular disturbed – habitats when given the opportunity. These results have important implications with regard to possible strategies to be used to control invasive ants.     

Improving liquid bait programs for Argentine ant control: bait station density

Argentine ants, Linepithema humile (Mayr), have a positive effect on populations of mealybugs (Pseudococcus spp.) in California vineyards. Previous studies have shown reductions in both ant activity and mealybug numbers after liquid ant baits were deployed in vineyards at densities of 85-620 bait stations/ha. However, bait station densities may need to be <85 bait stations/ha before bait-based strategies for ant control are economically comparable to spray-based insecticide treatments-a condition that, if met, will encourage the commercial adoption of liquid baits for ant control. This research assessed the effectiveness of baits deployed at lower densities. Two field experiments were conducted in commercial vineyards. In experiment 1, baits were deployed at 54-225 bait stations/ha in 2005 and 2006. In experiment 2, baits were deployed at 34-205 bait stations/ha in 2006 only. In both experiments, ant activity and the density of mealybugs in grape fruit clusters at harvest time declined with increasing bait station density. In 2005 only, European fruit lecanium scale [Parthenolecanium corni (Bouché)] were also present in fruit clusters, and scale densities were negatively related to bait station density. The results indicate that the amount of ant and mealybug control achieved by an incremental increase in the number of bait stations per hectare is constant across a broad range of bait station densities. The results are discussed in the context of commercializing liquid ant baits to provide a more sustainable Argentine ant control strategy.     

Integrating physiology, population dynamics and climate to make multi-scale predictions for the spread of an invasive insect: the Argentine ant at Haleakala National Park, Hawaii

Mechanistic models for predicting species’ distribution patterns present particular advantages and challenges relative to models developed from statistical correlations between distribution and climate. They can be especially useful for predicting the range of invasive species whose distribution has not yet reached equilibrium. Here, we illustrate how a physiological model of development for the invasive Argentine ant can be connected to differences in micro‐site suitability, population dynamics and climatic gradients; processes operating at quite different spatial scales. Our study is located in the subalpine shrubland of Haleakala National Park, Hawaii, where the spread of Argentine ants Linepithema humile has been documented for the past twenty‐five years. We report four main results. First, at a microsite level, the accumulation of degree‐days recorded in potential ant nest sites under bare ground or rocks was significantly greater than under a groundcover of grassy vegetation. Second, annual degree‐days measured where population boundaries have not expanded (456–521 degree‐days), were just above the developmental requirements identified from earlier laboratory studies (445 degree‐days above 15.9°C). Third, rates of population expansion showed a strong linear relationship with annual degree‐days. Finally, an empirical relationship between soil degree‐days and climate variables mapped at a broader scale predicts the potential for future range expansion of Argentine ants at Haleakala, particularly to the west of the lower colony and the east of the upper colony. Variation in the availability of suitable microsites, driven by changes in vegetation cover and ultimately climate, provide a hierarchical understanding of the distribution of Argentine ants close to their cold‐wet limit of climatic tolerances. We conclude that the integration of physiology, population dynamics and climate mapping holds much promise for making more robust predictions about the potential spread of invasive species.     

Rainfall facilitates the spread, and time alters the impact, of the invasive Argentine ant

Climate change may exacerbate invasions by making conditions more favorable to introduced species relative to native species. Here we used data obtained during a long-term biannual survey of the distribution of ant species in a 481-ha preserve in northern California to assess the influence of interannual variation in rainfall on the spread of invasive Argentine ants, Linepithema humile, and the displacement of native ant species. Since the survey began in 1993, Argentine ants have expanded their range into 74 new hectares. Many invaded hectares were later abandoned, so the range of Argentine ants increased in some years and decreased in others. Rainfall predicted both range expansion and interannual changes in the distribution of Argentine ants: high rainfall, particularly in summer months, promoted their spread in the summer. This suggests that an increase in rainfall will promote a wider distribution of Argentine ants and increase their spread into new areas in California. Surprisingly, the distribution of two native ant species also increased following high rainfall, but only in areas of the preserve that were invaded by L. humile. Rainfall did not have a negative impact on total native ant species richness in invaded areas. Instead, native ant species richness in invaded areas increased significantly over the 13 years of observation. This suggests that the impact of Argentine ants on naïve ant communities may be most severe early in the invasion process.     

A Disjunct Argentine Ant Metacolony in Macaronesia and Southwestern Europe

The Argentine ant, Linepithema humile, originally from South America, is now a major pest in many parts of the world with Mediterranean-like climates. Earlier research indicated that southwestern European L. humile populations are segregated into two distinct supercolonies, the ‘Main’ supercolony extending through Portugal, Spain, southern France, and Italy, and the ‘Catalonian’ supercolony in eastern Spain. Both supercolonies are unicolonial, with workers showing no aggression towards members of the same supercolony, but severe aggression towards members of the other supercolony. Here we evaluated the behavioral relationships among non-native L. humile populations on the Macaronesian islands of Madeira and the Azores and non-native populations in southwestern Europe. We conducted aggression assays among L. humile workers from Madeira, the Azores, and the two southwestern European supercolonies. We found no aggressive interactions among any combination of workers from Madeira, the Azores, and the Main supercolony. However, workers from Madeira and the Azores always fought aggressively with workers from the Catalonian supercolony. Thus, the populations of L. humile in Madeira and the Azores appear to be unicolonial and act as if they belong to the Main supercolony of southwestern Europe. This set of geographically separated but mutually compatible supercolonies, which we term a ‘metacolony,’ appears to descend from one supercolony. Historical evidence suggests that the first L. humile population in the greater Mediterranean region was established in Madeira and that propagules from the dominant supercolony in Madeira gave rise to the dominant supercolonies in to other parts of the region.     

Carbohydrate supply limits invasion of natural communities by Argentine ants

The ability of species to invade new habitats is often limited by various biotic and physical factors or interactions between the two. Invasive ants, frequently associated with human activities, flourish in disturbed urban and agricultural environments. However, their ability to invade and establish in natural habitats is more variable. This is particularly so for the invasive Argentine ant (Linepithema humile). While biotic resistance and low soil moisture limits their invasion of natural habitats in some instances, the effect of food availability has been poorly explored. We conducted field experiments to determine if resource availability limits the spread and persistence of Argentine ants in remnant natural forest in North Carolina. Replicated transects paired with and without sucrose solution feeding stations were run from invaded urban edges into forest remnants and compared over time using baits and direct counts at feeding stations. Repeated under different timing regimes in 2006 and 2007, access to sucrose increased local Argentine ant abundances (1.6–2.5 fold) and facilitated their progression into the forest up to 73 ± 21% of 50-m transects. Resource removal caused an expected decrease in Argentine ant densities in 2006, in conjunction with their retreat to the urban/forest boundary. However, in 2007, Argentine ant numbers unexpectedly continued to increase in the absence of sugar stations, possibly through access to alternative resources or conditions not available the previous year such as honeydew-excreting Hemiptera. Our results showed that supplementing carbohydrate supply facilitates invasion of natural habitat by Argentine ants. This is particularly evident where Argentine ants continued to thrive following sugar station removal.     

Interspecific Aggression and Resource Monopolization of the Invasive Ant Anoplolepis gracilipes in Malaysian Borneo

Invasions by introduced ant species can be ecologically destructive and affect a wide range of taxa, particularly native ants. Invasive ant species often numerically dominate ant communities and outperform native ant species in effective resource acquisition. Here, we describe interactions between the invasive ant Anoplolepis gracilipes (Smith) and resident ant species in disturbed habitats in NE Borneo. We measured interference competition abilities of A. gracilipes by performing arena bioassays between two A. gracilipes colonies and seven local ant species, and measured its effective resource competition at baits within supercolonies and at supercolony boundaries. Furthermore, we compared ant species diversity and composition at baits among (A) core areas of A. gracilipes supercolonies, (B) supercolony boundaries and (C) outside supercolonies. Anoplolepis gracilipes was behaviorally dominant over most ant species except Oecophylla smaragdina. Within supercolonies, A. gracilipes discovered all food baits first, and monopolized the vast majority throughout the course of the experiment. At supercolony boundaries, A. gracilipes discovered baits later than resident ant species, but subsequently monopolized half of the baits. Furthermore, the activity and diversity of the ant community within A. gracilipes supercolonies was lower than at its boundaries and outside supercolonies, and the ant communities differed significantly between infested and noninfested areas. Our study supports the hypothesis that successful establishment of A. gracilipes in anthropogenically disturbed habitats may negatively affect resident ant communities through high levels of direct interspecific aggression and almost complete monopolization of resources within high‐density supercolonies.     

DYNAMICS AND GENETIC STRUCTURE OF ARGENTINE ANT SUPERCOLONIES IN THEIR NATIVE RANGE

Some introduced ant populations have an extraordinary social organization, called unicoloniality, whereby individuals mix freely within large supercolonies. We investigated whether this mode of social organization also exists in native populations of the Argentine ant Linepithema humile . Behavioral analyses revealed the presence of 11 supercolonies (width 1 to 515 m) over a 3-km transect. As in the introduced range, there was always strong aggression between but never within supercolonies. The genetic data were in perfect agreement with the behavioral tests, all nests being assigned to identical supercolonies with the different methods. There was strong genetic differentiation between supercolonies but no genetic differentiation among nests within supercolonies. We never found more than a single mitochondrial haplotype per supercolony, further supporting the view that supercolonies are closed breeding units. Genetic and chemical distances between supercolonies were positively correlated, but there were no other significant associations between geographic, genetic, chemical, and behavioral distances. A comparison of supercolonies sampled in 1999 and 2005 revealed a very high turnover, with about one-third of the supercolonies being replaced yearly. This dynamic is likely to involve strong competition between supercolonies and thus act as a potent selective force maintaining unicoloniality over evolutionary time.     

The role of competitive dominance in the invasive ability of the Argentine ant (Linepithema humile)

To assess the importance of competition in the advance of invasive species, bait stations have been used to determine the dominance hierarchy of a community of native ants in Doñana National Park, southern Spain, and the status of the introduced species Linepithema humile (Argentine ant). Some native species, e.g. Cataglyphis floricola or Camponotus pilicornis, seem to be subordinate, i.e. occupy a low position in the competitive hierarchy; some are dominant (e.g. Pheidole pallidula), and others (e.g. Aphaenogaster senilis) occupy an intermediate position in the hierarchy. The Argentine ant is a competitively dominant species, because of its aggressive behavior and relative abundance. Irrespective of their position in the dominance hierarchy, L. humile and some native species adopt what games theory terms “the bourgeois strategy” during agonistic encounters with other species. Lone workers tend to be submissive in encounters whereas workers in the presence of other colony members are aggressive. L. humile was the only species which aggressively displaced large numbers of ants of other species from the bait. L. humile also expanded its range in the course of the experiment, displacing native species from parts of the study area.     

The role of cuticular hydrocarbons as chemical cues for nestmate recognition in the invasive Argentine ant (<Emphasis Type="Italic">Linepithema humile</Emphasis>)

Argentine ants (Linepithema humile) in their native South American range, like most other ant species, form spatially restricted colonies that display high levels of aggression toward other such colonies. In their introduced range, Argentine ants are unicolonial and form massive supercolonies composed of numerous nests among which territorial boundaries are absent. Here we examine the role of cuticular hydrocarbons (CHCs) in nestmate recognition of this highly damaging invasive ant using three supercolonies from its introduced range. We conducted behavioral assays to test the response of Argentine ants to workers treated with colonymate or non-colonymate CHCs. Additionally, we quantified the amount of hydrocarbons transferred to individual ants and performed gas chromatography-mass spectrometry (GC/MS) to qualitatively characterize our manipulation of CHC profiles. The GC/MS data revealed marked differences in the hydrocarbon profiles across supercolonies and indicated that our treatment effectively masked the original chemical profile of the treated ants with the profile belonging to the foreign individuals. We found that individual workers treated with foreign CHCs were aggressively rejected by their colonymates and this behavior appears to be concentration-dependent: larger quantities of foreign CHCs triggered higher levels of aggression. Moreover, this response was not simply due to an increase in the amount of CHCs applied to the cuticle since treatment with high concentrations of nestmate CHCs did not trigger aggression.The results of this study bolster the findings of previous studies on social insects that have implicated CHCs as nestmate recognition cues and provide insight into the mechanisms of nestmate recognition in the invasive Argentine ant. Received 6 February 2007; revised 31 May and 27 July 2007; accepted 16 August 2007.     

New mutualism for old: indirect disruption and direct facilitation of seed dispersal following Argentine ant invasion

The indirect effects of biological invasions on native communities are poorly understood. Disruption of native ant communities following invasion by the Argentine ant (Linepithema humile) is widely reported to lead indirectly to the near complete collapse of seed dispersal services. In coastal scrub in southeastern Australia, we examined seed dispersal and handling of two native and two invasive alien plant species at Argentine ant-invaded or -uninvaded sites. The Argentine ant virtually eliminates the native keystone disperser Rhytidoponera victoriae, but seed dispersal did not collapse following invasion. Indeed, Argentine ants directly accounted for 92% of all ant-seed interactions and sustained overall seed dispersal rates. Nevertheless, dispersal quantity and quality among seed species differed between Argentine ant-invaded and -uninvaded sites. Argentine ants removed significantly fewer native Acacia retinodes seeds, but significantly more small seeds of invasive Polygala myrtifolia than did native ants at uninvaded sites. They also handled significantly more large seeds of A. sophorae, but rarely moved them >5 cm, instead recruiting en masse, consuming elaiosomes piecemeal and burying seeds in situ. In contrast, Argentine ants transported and interred P. myrtifolia seeds in their shallow nests. Experiments with artificial diaspores that varied in diaspore and elaiosome masses, but kept seed morphology and elaiosome quality constant, showed that removal by L. humile depended on the interaction of seed size and percentage elaiosome reward. Small diaspores were frequently taken, independent of high or low elaiosome reward, but large artificial diaspores with high reward instead elicited mass recruitment by Argentine ants and were rarely moved. Thus, Argentine ants appear to favour some diaspore types and reject others based largely on diaspore size and percentage reward. Such variability in response indirectly reduces native seed dispersal and can directly facilitate the spread of an invasive alien shrub.