Introduction

The Argentine ant, Linepithema humile (Mayr), is an invasive species that displaces ecologically important native ant species, causing devastating effects on entire ecosystems as the ecological function of the native inhabitants goes unfulfilled (Christian, 2001; Erickson, 1971; Human and Gordon, 1997; McGlynn, 1999; Sanders et al., 2001; Sanders et al., 2003). This invasive species is highly polydomous and unicolonial, with colonies consisting of expansive and fluid networks of nests and trails (Heller and Gordon, 2006). L. humile changes its nest preferences seasonally according to food availability and proximity (Brightwell and Silverman, 2011; Heller and Gordon, 2006; Heller et al., 2006), as well as to microclimatic conditions of humidity and temperature in the nest location, looking for cool and wet sites during summer and warm and dry places in winter (Heller and Gordon, 2006; Heller et al., 2006).

Up to now, efforts to eradicate L. humile have involved mainly the use of chemical products (e.g., Choe et al., 2010; Hara and Hata, 1992; Harris et al., 2002; Klotz et al., 2002; Krushelnycky et al., 2004; Phillips et al., 1987; Rust et al., 2004; Silverman et al., 2006; Vega and Rust, 2003) but have had low success, and should not be used in natural environments since they are non-selective, i.e., other ant species can have access to them and get poisoned, diminishing native ant competition. Therefore, new studies are necessary to develop other kinds of control approaches that have not been tested. Those strategies might be focused on slowing the rate of spread of this invasive species to limit its establishment into yet non-invaded areas, and subsequently, its negative impact on the whole ecosystem. For this, the elimination of queens has been recommended (Abril et al., 2012; Abril et al., 2008a), principally in the front (edge) of the invasion, where the density of queens is higher (Abril et al., 2012). Moreover, the right time to perform the queen control should be in winter, because it is when the highest density of queens per nest is found, those nests being essential for the species dispersion in spring and thus for its expansion into non-invaded areas (Abril et al., 2012). The implementation of any queen control method implicates, therefore, the necessity to identify rapidly and correctly suitable places for the Argentine ant to nest, especially in winter.

The distribution of the Argentine ant has been linked to its physiological limitations, which can predict its potential distribution worldwide (Abril et al., 2009; Hartley and Lester, 2003; Krushelnycky et al., 2005), together with other factors such as vegetation type, topography (Roura-Pascual et al., 2004; Roura-Pascual et al., 2006) and soil type (Paiva et al., 1998; Way et al., 1997). Depending on the habitat, the Argentine ant can nest in a variety of places, like the topsoil (Heller and Gordon, 2006), the base of trees (Brightwell and Silverman, 2011), leaf litter piles (Newell and Barber, 1913) and decaying logs (Fellers and Fellers, 1982). At small scale, few studies have assessed the Argentine ant nesting behaviour focussing on nests under rocks (Cole et al., 1992; Fellers and Fellers, 1982; Ingram, 2002b), and in the topsoil (Heller and Gordon, 2006). Although Heller and Gordon (2006) studied the seasonal and spatial dynamics of L. humile nests, none of the studies evaluated if the Argentine ant exhibited nest selection, and if this selection had any relation with its seasonal movements.

Therefore, this study aimed to investigate if the Argentine ant shows nest site selection and if this selection has any relation with seasonal changes. We evaluated the type of nests and their longevity in relation to physical factors such as orientation (solar exposition), topography (slope of the terrain where the nest is located) and canopy cover (%). Moreover, we took into account the availability of suitable nesting sites, such as rocks and logs, since they could be another factor driving nest site selection. In addition, we compared nest site selection and density of nests between the edge of the invasion and the full invaded area to assess if these characteristics varied between the different areas of invasion. The knowledge of nest site preferences by the Argentine ant may allow a faster and accurate identification of nests, saving time and efforts and thereby making control measures more cost-effective.

Finally, we conducted an experimental test in situ to evaluate if the Argentine ant could be attracted by artificial nests (i.e., bricks and tiles), instead of nesting under natural structures. If that happened, artificial nests could be acting as a “nest-trap” and could constitute a possible tool for control.

Methods

Study area

To assess nest site selection, nine plots of 12 × 12 m (144 m2) were sampled in three invaded secondary cork oak forests situated in the Northeast of the Iberian Peninsula. Two of them were located in the Cadiretes Massif: one alongside the Puntabrava housing development in the Municipality of Tossa de Mar (41°46′30″N 2°59′54″E), and the other in “Pedralta”, Municipality of Sant Feliu de Guíxols (41°47′31″N 2°58′52″E). The third one was located in the Gavarres Massif in the Municipality of Santa Cristina d’Aro, around the golf camp “El Masnou” (41°49′24″N 3°00′18″E) (Fig. 1a).

Fig. 1
figure 1

a Location of the three study sites in Catalonia, Spain. CA Santa Cristina d’Aro front, PB Puntabrava front, PD Pedralta front; b sampling design used at each one of the study sites. F front area, I1 invaded area at 250 m, and I2 invaded area at 450 m from the edge of the invasion

Full size image

Sampling design

First, to identify the invasion edge or front, bait sampling was carried out in each forest, placing tuna and marmalade baits every four metres on 100-m long random transects, and the invasion limit was identified by the last bait visited by Argentine ants. After identifying the invasion edge, one plot (called F) was placed in the front of the invasion, at 20 m on average from the localized edge, and the two others further inside the invaded area: the first one (called I1) at 250 m, and the second one (called I2) at 450 m from the edge on average, setting in total nine plots (three within each sampled area) (Fig. 1b). In December 2008, we searched for Argentine ant nests in the topsoil, and under all the available structures such as rocks, twigs, logs and in the base of trees. All the individual nests (with queens and brood present) found inside the plots were marked with fluorescent spray paint and mapped on a grid subdivided into 1 × 1 m2.

Nests were detected by upturning and immediately replacing rocks, twigs or logs. In the case of nests located in tree bases, ant trails were followed down the trunk, until a nest entrance was located, and at this place, a gentle movement of the leaf litter was made with the help of a garden shovel to verify the presence of the nest. Previous observations in the study area confirmed that the Argentine ant nests were not greatly disturbed by this sampling method, and that nests remained in place after several weeks. All the nests found were placed underneath rocks, logs or at tree bases and none of them was located in the topsoil. At each plot, all active nests were counted, marked, mapped and monitored every 2 months from December 2008 to February 2010, and in October 2010, December 2010 and February 2011 to determine nest site preferences throughout one complete year and in three winters.

To determine whether ants chose rocks with particular attributes, all rocks available and rocks used for nesting were sampled. For this, all the rocks with a surface area greater than 5 cm2 hosting a nest or surrounding one (in a radius of up to 1 m) were measured. Their surface (S) was calculated as: S = maximum dimension × minimum dimension, and their volume (V) as: V = total surface area × thickness. Also, sun exposure (orientation) of each rock was recorded. The orientation of a given rock/log was considered to be the direction (measured with a compass) towards which the main surface of the rock/log was facing. In the case a nest was found in the base of a tree, the tree species name and its diameter at breast height (DBH, in cm) were recorded and nest orientation was considered to be the direction towards which the nest entrance was facing.

Additionally, the slope and the canopy cover of each plot were measured in five points, the first one at the centre of the plot and the four others 7 m from the centre, on the diagonals of the plot. The canopy cover was measured by means of digital photographs analysed with the software Gap Light Analyzer version 2.0, which estimates the canopy openness percentage. The canopy cover (CC) percentage was then calculated as CC (%) = 100 − Canopy openness (%). The slope was measured with a clinometer (Abney level), letting the instrument to lay parallel to the ground to have a direct angle measurement at each sampling point. Although this method is not completely accurate, and can overestimate the angle measured, it is the only one that gives information on the micro-relief variations of the terrain (Gilg, 1973), which was necessary to make comparisons of the slight differences in slope between nest locations.

Set up of artificial nests as traps

In order to test the suitability of artificial nests as a nesting place for the Argentine ant, three varieties of artificial nests were set in March 2010, in a plot of 208 m2 (52 × 4 m), located inside the invaded area of the Pedralta front: six hollow clay bricks (B), six partition clay tiles (P) and six concrete tiles (T) (with the rough side in contact with the ground) (Fig. 2). They were arranged randomly in six different sets (i.e., a total of 18 nest-traps), which were separated by 10 m from each other. Within sets, a distance of 1 m was let between the three types of nest-traps (Fig. 3). Hollow clay bricks measured 28.6 × 13 × 9.5 cm, partition clay tiles 49.5 × 19.5 × 3.5 cm and concrete tiles 30 × 30 × 3.5 cm. Bricks and tiles were chosen for this experiment because they can be easily found on the market and could be recycled for this purpose when they have been accidently damaged and discarded for construction. All nest-traps were placed oriented south to homogenize this factor which is known to influence nest choice, as was observed by Markin (1970b).

Fig. 2
figure 2

Artificial nests used as nest-traps: a Concrete tiles, b partition clay tiles, and c hollow clay bricks

Full size image
Fig. 3
figure 3

Sampling design used in the nest-trap plot. Circles represent hollow clay bricks, squares represent partition clay tiles, and triangles represent concrete tiles

Full size image

Nest-traps were monitored monthly, from April 2010 to March 2011. In early March 2011, the entire plot was sampled for the presence of Argentine ant nests under nest-traps, rocks, twigs, logs and in tree bases, and the same methodology as for measuring nest availability and selection was followed, taking the same measurements from rocks and trees and evaluating the canopy cover and the slope of the terrain at each nest-trap set.

Data analysis

Generalized linear mixed models (GLMMs) with Poisson error distribution and log link were performed to test nest type preferences, and differences in the overall number of nests and the number of nests of each type (trees, logs, and rocks) between the areas of invasion F, I1 and I2, at 20, 250 and 450 m from the invasion edge, respectively. Type of nest and area of invasion were included as fixed factors, and site (forest fronts) as a random factor with the purpose of controlling possible site-based differences.

GLMMs with Gaussian error distribution and identity link were performed to test differences in rock dimensions (surface, thickness and volume) for the nests located under rocks depending on their time of occupancy (number of seasons throughout 2009 and number of winters) which was considered as the fixed factor. Additionally, differences in rock dimensions between rocks with and without nests were tested using Binomial error distribution and the logit link, including the nest presence as the fixed factor. For both analyses, sites were considered as a random factor. Rocks that were partially buried were not included in the analyses.

Differences in the number of nests between seasons, in accordance with physical variables, including canopy cover (%), slope (°) and orientation (°) were tested through GLMMs with Poisson error distribution and log link. Moreover, differences in the presence/absence of nests underneath rocks in relation to these physical variables within seasons were tested by means of GLMMs with binomial distribution and the logit link. In both analyses, physical variables were included as fixed factors and site as a random factor.

For the experimental plot, generalized linear models (GLMs) with binomial error distribution and logit link were run to test if the presence of Argentine ant nests was due to the type of nest (traps, trees or rocks) together with the influence of canopy cover (%), and slope (°). The same kinds of GLMs were run to test if there was a preference for any type of artificial nest (hollow clay bricks, concrete tiles or partition clay tiles) and if this preference was influenced by canopy cover and/or the slope of the nest-trap. These analyses were made with data from March 2011, when all types of nests and physical factors were surveyed. An additional GLM with Poisson error distribution and log link was performed to test if there was a nest-trap preference throughout the sampled year (April 2010–March 2011). All statistical analyses were run with R version 2.15.2 (R Development Core Team, 2011). P values lower than 0.05 were considered as significant in all analyses.

Results

Nest site selection

We found a total of 310 Argentine ant nests in the sampled plots, located under rocks (236 nests), in tree bases (53 nests) and under logs (21 nests) in order of preference. This preference was maintained through all the sampled plots and the number of nests from the different types differed significantly (GLMM rock vs. log: t = 2.78, rock vs. tree: t = −2.55, log vs. tree: t = 3.13; P < 0.02 in all cases). Additionally, the number of nests found under rocks, logs, or in tree bases varied according to the time of the year. This species preferred to nest in the base of trees from June to August, and under rocks and logs in winter and spring (although nests under logs were very scarce) (Fig. 4). The preference to nest under rocks was maintained as well during the three consecutive winters.

Fig. 4
figure 4

Cumulative abundance of Argentine ant nest types for the nine plots sampled bimonthly from December 2008 to December 2009. Black line denotes nests under logs, grey line nests under rocks and dotted line nests in tree bases

Full size image

Regarding the distance to the edge of the invasion, we found significant differences in the total number of nests between plots F vs. I2 (GLMM: t = 8.68, P < 0.02), finding higher number of nests as the plot was furthest from the invasion edge. The number of nests under rocks was higher than nests on tree bases and nests under logs at all distances from the edge of the invasion, this difference being statistically significant within F (GLMM log vs. rock: t = 5.98, P = 0.03) and I2 plots (GLMM log vs. rock: t = 8.56, rock vs. tree: t = −17.03; P < 0.01).

Overall, the nest density per plot increased as the plot was further into the invaded area for all types of nest. The mean density for nests under logs was 0.01 nest/m2 for F plots, 0.02 nest/m2 for I1 plots and 0.02 nest/m2 for I2 plots; for nests at tree bases was 0.04 nest/m2 for F plots, 0.05 nest/m2 for I1 plots and 0.03 nest/m2 for I2 plots; and for nests under rocks was 0.07 nest/m2 for F plots, 0.08 nest/m2 for I1 plots and 0.39 nest/m2 for I2 plots. When we compared the number of nests per type and area of invasion, we detected only significant differences in the number of nests under rocks between plots F vs. I2 (GLMM: t = 7.38, P < 0.02), those being more abundant in the later.

The Argentine ant nested throughout the year at the tree bases of Quercus suber L., Pinus pinaster Aiton and Arbutus unedo L., in order of preference. The mean DBH from trees sheltering nests was 103.8 cm ± 17.9 for P. pinaster, and 52.4 cm ± 20.3 for Q. suber. In winter, Q. suber was the preferred tree species to nest (26/34 nests). Nests located under logs were unusual and were found only in some of the plots sampled. Logs used to nest came mainly from fallen branches of Q. suber (90 %) and in a fewer proportion from P. pinaster (10 %). They had a mean length of 42.2 cm ± 33.2 SE and a mean volume of 3034.5 cm3 ± 6340.2 SE.

Rock selection for nesting

We measured a total of 230 rocks with nests and 188 rocks without nests. The surface (cm2), thickness (cm) and volume (cm3) between rocks with and without nests were not significantly different (GLMM: P > 0.05 for all dimensions). Most of the nests were found under rocks with a surface smaller than 50–200 cm2, a thickness smaller than 2–5 cm and a volume smaller than 200–400 cm3.

The time of rock occupancy was measured as the number of seasons a rock was used to nest by the Argentine ant from December 2008 to October 2009. Overall, larger rocks sheltered nests during a longer period of time. Rocks that were occupied during one or two seasons showed a higher variability in their dimensions than those that were occupied during three seasons, and only two of the biggest rocks sheltered nests during the whole year (Fig. 5). The dimensions of those biggest rocks differed significantly from rocks that were occupied for one season (GLMM surface: t = 4.42, P < 0.001; thickness: t = 1.95, P ≤ 0.05; volume: t = 4.62, P < 0.001), two seasons (GLMM surface: t = 3.05, P < 0.01; thickness: t = 2.13, P < 0.05; volume: t = 3.44, P < 0.01) and three seasons (GLMM surface: t = 5.42, P < 0.01; thickness: t = 3.97, P < 0.01; volume: t = 6.06, P < 0.001). Nest longevity during successive winters was higher as well for bigger rocks. Seventy per cent of the rocks used during the three winters sampled, had a surface between 104 and 255 cm2, a thickness between 3.5 and 8 cm and a volume between 364 and 2040 cm3.

Fig. 5
figure 5

Variation in a surface, b thickness, and c volume of the rocks in relation to their time of occupancy (No. seasons). Boxes delimitate the lower and upper quartiles, horizontal bars denote the median (solid line) and the mean (dashed line), whiskers the minima and maxima, and circles the outliers

Full size image

Influence of physical factors on nest site selection

Our results indicate that the Argentine ant nested preferentially throughout the year in terrains with a slope around 14°, a reduced canopy cover (56 % of the nests were found with a cover between 0 and 10 %) and oriented mostly south (54.2 % of the nests). Regarding orientation (°), we found that nests were significantly more abundant and present at almost all cardinal directions during summer in comparison to spring (GLMM: t = −2.28, P = 0.03) and winter (GLMM: t = −2.28, P = 0.02), when nests were orientated mostly towards the South (Fig. 6a). Furthermore, nest longevity (considered as the number of seasons a nest was active) was consistently higher in south-facing nests, this orientation being together with the east-facing orientation, the preferred throughout the whole year (from December 2008 to October 2009).

Fig. 6
figure 6

Number of Argentine ant nests by season and area of invasion in accordance to their a orientation (°), b canopy cover (%), and c slope (°). Open bars denote edge of the invasion area (F), grey bars denote invaded area at 250 m from the edge (I1), and black bars denote invaded area at 450 m from the edge (I2). N for canopy cover and orientation: winter = 52, spring = 71, summer = 48, autumn = 62. N for slope: winter = 42, spring = 56, summer = 11, autumn = 32

Full size image

At different canopy cover percentages, we found a significantly lower number of nests in winter compared with spring (GLMM: t = 2.34, P = 0.02) and autumn (GLMM: t = 2.94, P = 0.004). Nests were concentrated mostly under low canopy cover values during winter. Moreover, the canopy cover was significantly different between rocks with and without nests, finding a higher presence of nests under lower canopy cover values in winter and higher canopy values in summer (GLMM t winter = −2.32, t summer = 1.98; P ≤ 0.05) (Fig. 6b). When we evaluated the slope (°) from nests beneath rocks, we found that the majority of nests were located in terrains with a slope of around 14°. This preference was continuous throughout the year, and no differences were detected between seasons (Fig. 6c).

Artificial nest selection

Within the 84 available places to nest in the experimental plot, a total of 17 Argentine ant nests were found, from which six (35 %) were located under artificial nests, eight (47 %) in tree bases and three (18 %) under rocks. Nest site selection was explained partially by the added effect of the type of nest and slope of the terrain (according to the explained deviance of the GLM (Zuur et al., 2009): D 2 = 13.51 %, P ≤ 0.05).

Regarding the types of artificial nests, all of them were used throughout the year (from April 2010 to March 2011), without any preference (GLM: P > 0.05), although concrete tiles (T) were the nest-traps most used, followed by hollow clay bricks (B) and partition clay tiles (P). At the time of the plot survey in early March 2011, the Argentine ant did not show preferences for any type of artificial nest (GLM: z T = −0.80, z P = 0.58, z B = −0.66; P > 0.05 for all), and the only factor that had an influence on the presence of nests was the slope of the terrain (GLM: z = 2.02, P < 0.05).

Between July and August 2010, some of the artificial nests were moved during forest cleaning activities for the extraction of fallen wood. Nests were returned to their place, but this movement affected two of the Argentine ant nests present before the perturbation: one of them under a concrete tile, which disappeared in July but returned in August, and the other one under a hollow clay brick, which disappeared in August and never returned. Artificial nests were used as well by the native ant Plagiolepis pygmaea Latreille, especially in May 2010, but in all cases, at the next month of the survey, nests had been re-occupied by L. humile.

Discussion

The Argentine ant is a highly mobile species that build usually superficial and poorly elaborated nests, possibly to benefit from the higher temperatures near the soil surface to escape adverse conditions in winter (Brightwell et al., 2010), and to diminish excavations costs (Halley et al., 2005). It is known that L. humile changes seasonally nest location to maintain appropriate microclimatic conditions inside the nest (Benois, 1973; Heller and Gordon, 2006; Heller et al., 2006; Markin, 1970c; Newell and Barber, 1913). In California, for example, Markin (1970c) found that entire populations of L. humile switched from a shaded and irrigated zone of a citrus grove in summer, to a steeped south–east-facing hillside in winter. Our results agree with previous literature, underlining that the Argentine ant moved seasonally to keep regular microclimatic conditions (humidity and temperature) inside its nests. The fact that the Argentine ant chose trees to nest mainly in the hottest months of the surveyed year suggests that this species uses the tree structure as isolation against unfavourable weather.

Recently, Brightwell and Silverman (2011) found that L. humile was able to survive in natural habitats under cold air temperatures by nesting around the loblolly pine Pinus taeda L. and to forage under its sun-warmed bark at temperatures well below its minimum foraging threshold. Herein, the results suggest that in our study area the Argentine ant uses pines and cork oaks for the same purpose in warmer seasons, but this time to protect the nest against high temperatures and low humidity, helped by the canopy cover and the isolation effect from the tree bark.

The preference of tree nesting at this time of the year could be linked as well to the foraging activity which is strongly constrained by temperature (Markin, 1970a) and humidity (Abril et al., 2007). This happens because L. humile has a high cuticular permeability and thus suffers an important water-loss rate when exposed to high temperatures (Schilman et al., 2007). Unlike the daily foraging activity on the ground that is strongly inhibited by high daily temperatures (Markin, 1970b; Oliveras et al., 2005), the foraging activity of L. humile on cork oak trees is constant all day long (Abril et al., 2007), allowed most probably by the protection that the tree offers to the foragers.

The highest increment in the number of nests on tree bases happened during June and August (when we observed a higher quantity of ant trails on trees as well), coinciding with the increment in the number of workers and males which begin to hatch into the nest in June (Benois, 1973; Markin, 1970c), having, therefore, a higher demand for food, especially sugary liquids from hemipteran exudates, which are the main type of food consumed by these castes (Markin, 1970a). Hence, the tree nesting behaviour during the warmer months has a dual purpose: first, to gain protection against unfavourable hot weather conditions, and second, to be closer to the food resources needed to meet the trophic requirements during this reproductive period, improving at the same time its foraging efficiency and its competitive ability (Holway and Case, 2000).

In contrast, during colder and rainy seasons, the Argentine ant strategy is to nest underneath rocks which may allow to increase nest temperature (Heller et al., 2006) helped by a southern orientation. To keep an appropriate temperature inside the nest is crucial for this species, since low annual soil temperatures may limit the successful development of workers (Hartley et al., 2010), and a prolonged exposure to subfreezing temperatures can be lethal (Jumbam et al., 2008). Nests under rocks can also keep appropriate moisture levels with the help of a moderate slope which may allow draining and evaporating the excess of soil water. The terrain features at micro-scale (orientation and slope) are thus important factors influencing directly the distribution of Argentine ant nests, especially during cold and rainy seasons.

Argentine ant nests and inter-nest trails suffer also a seasonal change in their size, with bigger nests in winter when the colony is contracted, and smaller nests during the expanding period from spring to autumn (Heller and Gordon, 2006). Changes in nest size also happen in relation to the time a place has been invaded: L. humile nests are bigger in highly invaded areas (Díaz et al., 2013), where there is was as well a higher number of workers (Enríquez et al., unpubl. data). Large nests may be advantageous in highly competitive environments because numerous workers are expected to increase the ability of a colony to monopolize resources (Human and Gordon, 1996). Ingram (2002a) argued that a large nest size is correlated to a high queen number in Argentine ant populations since a large number of workers increases the productivity of a nest (Herbers, 1984) and ensures colony longevity by offering protection against predators and competitors (Herbers, 1986). This could explain why the nests beneath bigger rocks were those that lasted longer throughout the year, and were the most used during successive winters.

Similar results were found by Tinaut et al. (1999) for Proformica longiseta Collingwood, which occupied preferentially larger rocks and showed a higher presence of brood beneath thicker rocks. This was explained by the protection that thicker rocks gave to this ant against extremely high temperatures and by the better heat retention compared with thinner rocks at the sunset. However, this species was found to nest under a variety of rock dimensions, which suggested that there was not a priori rock selection and that workers were not capable of recognizing suitable rocks from external physical characteristics (dimensions), but were capable of such recognition a posteriori on the basis of their thermal properties. Other mobile ant species such as Temnothorax albipennis Curtis and T. curvispinosus Mayr seem to be able to recognize when a nest site has a higher quality than theirs, and move nests regularly when they find better places (Dornhaus et al., 2004; Pratt, 2005). In the case of the Argentine ant, Scholes and Suarez (2009) found that this species was able to recognize a better place where to nest when it was exposed to flooding conditions. In our study sites, L. humile nested under a wide range of rock dimensions, although it showed higher nest longevity for nests underneath thicker rocks, which make us assume that those rocks where the best places to nest during the year.

Higher longevity of Argentine ant nests has also been associated with the size of the nest. Díaz et al. (2013) found that bigger nests (measuring 0.94 m2 ± 0.29 SE) were those that prevailed during the whole year (especially in winter), similar to what was found by Heller and Gordon (2006) in California. Moreover, the former authors and Heller et al. (2008) concur with the idea that those bigger nests could constitute the most important source of queens for the colonization of new areas in the expanding period of this species, acting, therefore, as mother/parent nests. Besides, and given that the size of the colony can be constrained by the size of the nest cavity (Thomas, 2002), those mother nests are expected to be found under bigger rocks as well. Therefore, control methodologies should pay particular attention to nests under larger rocks to have a stronger impact on L. humile populations and provide a more efficient management of the invasion.

Concerning nest density, it was lower as the nests were closer to the edge of invasion, maybe due to inter-specific competition with native ants (Ingram, 2002a) which experience a decrease in their diversity with the increment in abundance of the Argentine ant (Enríquez et al., unpubl. data). Nevertheless, in the Haleakala National Park, Ingram (2002b) found that this enlargement in nest density occurred only within 80 m from the edge and that nest density remained constant until a distance of 500 m. The differences with the present study could be explained by three reasons: first, Ingram sampled only rocks between 10 and 15 cm in diameter which could exclude a certain quantity of nests and give biased results; second, the differences in altitude throughout the transects were not taken into account, but seem to be affecting the abundance of L. humile in this Hawaiian park, as was found by Fellers and Fellers (1982); and third, Ingram sampled the nests just once (not specifying when) whilst our sampling took place throughout one complete year.

Regarding artificial nests, the fact that concrete tiles were the most used throughout the year could be related to their nature, since concrete tiles had a more similar composition to the stones from the study area and were more efficient at keeping soil humidity underneath than the other two artificial nests (pers. obs.). Although in February 2011 the partition clay tiles were preferred to nest, maybe because they were able to raise faster the temperature of the underlying soil (since they were thinner), they were little used when air temperatures increased, and were less efficient at keeping soil moisture, especially in the warmer months. Additionally, the fact that the concrete tiles were thicker and compact could have favoured the retention of heat for a longer time overnight, as was found by Robinson (2008) when using pavers to attract ants to nest in an Australian grassland.

Control methods, such as the removal of queens in winter, could be helped by the use of artificial nests. As nesting sites were not found to be a limiting factor either in this or in previous studies (Ingram, 2002b), artificial nests should be set in places that provide to the underneath soil the most suitable microclimatic conditions for the Argentine ant nesting, to have a higher probability of colonization and nest longevity. In addition, to have a greater impact on Argentine ant populations, artificial nests should be set in autumn, before the drop in temperature to attract nests that start to migrate at this time of the year looking for better conditions in winter. In this manner, the possibility for the foundation of winter colonies under artificial nests should be higher, and any kind of control exerted to the nest should have a stronger impact on L. humile populations. Our results suggest that the best places should have a moderate slope that allows drainage of excessive soil water and under low canopy cover. Moreover, concrete tiles or similar objects should be used in preference. The use of artificial nests as traps could help improve the performance of control methodologies by reducing the searching time of L. humile winter nests.