The Native Ant Lasius niger Can Limit the Access to Resources of the Invasive Argentine Ant

Simple Summary Invasive ants are often highly dominant competitors, having strong impacts on native species. Such invaders often exploit resources better than native species, finding them first or collecting them faster. They are also often more efficient when interfering with other species, suffering fewer losses or preventing access to resources. We assessed the competitive behavior of the invasive Argentine ant when facing another invasive species or a native dominant species. The exploratory behavior of the Argentine ant was strongly inhibited by the native dominant species. The Argentine ant brought very few prey resources to its nest and killed few opponents. Conversely, the other invasive species had low impact on the Argentine ant. Contrary to expectations, the invasive species lacked the ability to hinder resource exploitation by the Argentine ant, whereas the native dominant species did. These results suggest that a native species could impact invasive populations of the Argentine ant by interference competition, perhaps better so than some invasive species. In the northern half of Europe, it could prevent further expansion of this highly invasive species. Abstract Within ant communities, the biotic resistance of native species against invasive ones is expected to be rare, because invasive species are often highly dominant competitors. The invasive Argentine ant (Linepithema humile (Mayr)) often demonstrated numerical dominance against its opponents, increased aggressiveness, and ability to quickly recruit to food. The present study aimed to assess the behavioral mechanisms involved in the interspecific competition between L. humile, facing either an invasive species (Lasius neglectus Van Loon, Boomsma and Andrásfalvy) or a native dominant species (Lasius niger (Linnaeus)). The resource exploitation by the Argentine ant was investigated during one-hour competitive interactions using 10 dead Drosophila flies as prey. When facing La. niger, L. humile exploratory behavior was strongly inhibited, it brought very few prey resources, and killed few opponents. Conversely, La. neglectus had a low impact on L. humile. Contrarily to expectations, the invasive La. neglectus lacked the ability to hinder L. humile resource exploitation, whereas the native La. niger did. These results suggest that La. niger could impact invasive populations of L. humile by interference competition, perhaps better so than some invasive species. While L. humile has become invasive in Southern Europe, the invasion process could be slowed down in the northern latitudes by such native dominant species.


Introduction
Competition between species occurs in two ways. Exploitation competition involves the ability of species to find and exploit rapidly a resource before others, thereby making it unavailable to We measured 33 behavioral descriptors of the interactions covering the space occupation, the ability to exploit resources and the aggressiveness of L. humile, as well as the strength of the competition by the other species. We used controlled laboratory experiments where the behavioral characteristics that discriminate against the competitive ability of the three species can be assessed in the absence or presence of the competitor. As an invasive species, we expected to find evidence that La. neglectus has greater competitive abilities than La. niger, and therefore may hinder L. humile resource exploitation. The species established before the other, and therefore more familiarized with the nesting and foraging site, was also expected to have an advantage in the competitive interactions.

Studied Species
Three distinct species have been studied: the Argentine ant L. humile, the invasive garden ant La. neglectus and the black garden ant La. niger. Linepithema humile is currently a widespread and abundant invasive species, forming supercolonies (arising from the low levels of intraspecific aggression among colonies), with polydomic, highly polygynous nests and totally sterile workers [7]. Lasius neglectus also forms polygynous colonies with the presence of several functional queens within the nest [24]. This invasive species has ecological impacts on the biodiversity of Formicidae and other invertebrates (e.g., reducing the spatial and temporal foraging of native ants; [25,26]). The range of La. neglectus has increased rapidly and steadily in non-native locations over the last 30 years [27]. These two invasive species are present across almost all Europe and their distribution partly overlaps, especially in the easternmost part of Europe (Supplementary Material, Figure S1). Lasius niger is a widespread monogynous species, inhabiting all of Europe and parts of Asia and North America and colonizing a broad diversity of environments, but particularly abundant in arable land, as well as in cities, parks and gardens [23,28]. This species is aggressive towards other ants, including conspecifics [29].

Biological Material
The biological material used for the experiments came from colonies sampled between 23  In L. humile, the letter A indicates that it corresponds to the main supercolony and the letter B indicates that it corresponds to the Corsican supercolony [30,31]; in La. neglectus, the workers from colonies A and B demonstrate strong aggressiveness between each other, suggesting different supercolonies [32]. Each collected colony corresponds to three or more pooled nests on the same site. The colonies collected comprised several thousand workers, brood in substantial quantities, and at least five queens for nests belonging to polygynous species (only workers have been collected in La. niger monogynous colonies, preventing the colonies' destruction).
Colony fragments were kept in the laboratory at 25 ± 3 • C with a mean hygrometry at 47%, and maintained in their original nesting substrate in boxes of 370 × 255 × H160 mm 3 . Colonies were supplied with a water-honey dilution (50%), proteinate food (insects such as mealworms, fruit flies or crickets) and water ad libitum. The interaction experiment was conducted in the two months after the installation and acclimation of the colonies.

Interaction Experiments
Interaction experiments were conducted between 20 June and 15 August 2019. The experimental set-up was as follows (Supplementary Material, Figure S2): two small peripheral plastic boxes (79 × 79 × H34 mm 3 ) were connected by small plastic tubes (diameter 0.8 cm, length 2 cm) to a central arena (165 × 100 × H85 mm 3 ). Each box was closed with stainless steel mesh to prevent leakage. These boxes were characterized by a resting site opposite the arena, with a source of humidity (small tube with wet cotton), with a 15 × 15 mm 2 area covered with a red filter (favorable brightness). For each interaction, one hundred (±2) "naive" workers (i.e., never tested before) were taken from the donor colony, as closely as possible from the food resources to select preferentially foraging workers, and placed in a peripheral box with five brood elements. Water was supplied ad libitum. The taken ants were starved for 72 h prior to each trial [18]. During this period, one side of the device was open, allowing exploration of the central arena by one of the groups (resident), whereas the other did not have access to the arena (colonizer). The entrance of the resident was re-closed one hour prior to the prey being added.
Prior to the experiment, 10 dead Drosophila melanogaster (Meigen) individuals (adequate prey in field experiments with L. humile and native species; [33]) were placed in the center of the arena, equidistant from the two groups. At the start of the trial, the two entrances to the arena were simultaneously opened, and recruitment and interactions were filmed for one hour with two consumer-electronics RGB cameras BRIO 4K Ultra HD (Logitech, Lausanne, Switzerland). On each movie recorded, the species, the source colony, the date, and the conditions (temperature, hygrometry, status) were indicated on the video. All combinations of interactions with L. humile facing (i) no opponent, (ii) La. neglectus opponent and (iii) La. niger opponent were tested for both colonies of each species and successively for resident and colonizer statuses. Each individual test was replicated three times, resulting in 60 one-hour interactions recorded, covering a wide panel of situations encountered by a new colonizer, e.g., from no competitor (n = 12 videos), to an invasive or native competitor (n = 48 videos).

Video Analyses
Each video was analyzed by one of three observers familiar with the species and the ant's behaviors. To ensure the consistency of results, ten percent of the videos were monitored by two observers simultaneously and ten percent of the videos were examined by two observers separately. All the videos have been analyzed twice (one time per interacting species). For each analysis, several global metrics were measured separately for each species: the time before the first worker leaves the peripheral box, the time between the entrance in the arena and the discovery of the resource and the time before the first interspecific interaction. We systematically recorded the temperature and hygrometry of the testing environment. Kinetic metrics were measured every five minutes for each species: the number of ongoing fights in the arena, the number of dead workers, the number of preys brought into the nest, moved in the arena or still available on the bait, the number of workers in the whole arena and the number of workers on the bait.

Statistical Analyses
All statistics were carried out using R v. 3.3 (RC Team, Vienna, Austria) software. Based on the kinetic and global variables resulting from the video analyses, 33 ethological descriptors of the interactions have been summarized in four categories, reflecting various aspects of the interactions: the space occupation by L. humile (5 descriptors), L. humile's ability to exploit resources (8 descriptors), the aggressiveness of L. humile (8 descriptors) and the strength of the competition by the other species (12 descriptors; for interactions with opponents only; Appendix A, Table A1). We reduced these variables following three consecutive procedures: (i) remove correlated variables, (ii) run two principal component analyses (PCAs) for the different sets of descriptors, and (iii) run linear discriminant analyses (LDAs), based on the opponent species (coupled in previous PCAs), as explained below.
First, for each category of descriptors, we ran a collinearity analysis, eliminating the descriptors having a Spearman correlation value >0.7 with others, to establish a set of uncorrelated variables [34]. In each pair of correlated variables, the variable with the highest absolute correlation (i.e., the maximum average correlation with other variables) was identified using the find correlation function of the package caret [35] and removed (in italics in Appendix A, Table A1).
Second, the 22 remaining descriptors from the four categories (Appendix A, Table A1) have been used to perform two PCAs. The first PCA (Appendix A, Figure A1) was performed on 60 trials, including all variables regarding space occupation by L. humile and its ability to exploit resources, to reveal contrasts between tests with L. humile (i) alone, (ii) facing La. neglectus and (iii) facing La. niger. The second PCA (Appendix A, Figure A2) was performed on 48 trials, including all variables regarding the aggressiveness of L. humile and strength of the competition by the other species, to reveal contrasts between tests with L. humile facing (i) La. neglectus and (ii) La. niger.
Third, to identify variables varying according to the Argentine ant opponent (La. neglectus, La. niger or no opponent), we performed two discriminant analyses (LDAs) based on the two previous PCAs (package ade4; [36]). The significance of the eigenvalues was evaluated using a non-parametric version of Pillai's test. Variables having a high proxy of the contribution to the discriminant function, i.e., whose cosines between the variables and the first axe of the linear discriminant analysis were >0.5 (as an absolute value) were considered the best drivers of the differences between species. The discriminant analyses suggested that eight variables were mainly responsible for the differences between the situations with different status and opponent species: the mean number of L. humile workers simultaneously present in the whole arena (mean_arena_Lh, M1), the number of preys brought by L. humile (n_totprey_Lh, M2), the standard deviation of the numbers of L. humile workers on the bait over time (ETbait_Lh, M3), the mean number of L. humile workers simultaneously present on the bait (mean_bait_Lh, M4), the number of dead L. humile individuals (n_deadtot_Lh, M5), the number of dead opponent individuals (n_deadtot_opp, M6), the mean number of simultaneous fights during the contest (mean_fights, M7) and the time when the maximal number of simultaneous fights occurs (t_maxfights, M8) (Appendix A, Table A2).
After reducing the 33 variables to eight, using the described three-step procedures, these eight variables were used to investigate the importance of (i) the opponent (La. neglectus, La. niger or no opponent), (ii) the status (resident vs. colonizer) and their interaction. In order to do that, we performed eight mixed models using the package glmmTMB [37]. The identity of the 20 combinations of source colonies and status was introduced in the model as a random effect. M1, M3, M7 and M8 were fitted with linear models, M2 and M4 were fitted with linear models performed following a square root transformation of the dependent variables, M5 and M6 were fitted with generalized linear models using Poisson family. The significance of each explanatory term was tested using Chi-squared tests. For every significant variable, post hoc, pairwise contrasts among treatments were performed by calculating the least square means (package lsmeans; [38]), and t-tests were adjusted to the number of tests using Tukey corrections. The raw data and scripts are available on the Zenodo repository (https://doi.org/10.5281/zenodo.4327129).

Results
Results of the PCA and LDA indicated likely differences in L. humile space occupation and ability to exploit resources according to the opponent species, but not regarding the colonizer/resident status, with more individuals in the arena and on the bait when alone in the arena than when in the presence of La. niger. The response seemed to be intermediate in the presence of La. neglectus (Appendix A, Figure A1, Table A2). The aggressive and competitive behaviors of L. humile seemed to differ against La. neglectus or La. Niger, and also differed according to status when facing La. niger (Appendix A, Figure A2, Table A2).
The models confirmed that six of these eight discriminant variables significantly varied among the opponent species (see below), whereas only one has been detected as significantly impacted by the status of L. humile (resident/colonizer) and its interaction with the opponent species (mean_arena_Lh, the mean number of L. humile workers simultaneously present in the whole arena) ( Table 1).

Discussion
Among the behavioral descriptors of the interactions between the competing ant species, eight significantly discriminated against the opponent species (invasive or native opponents) or status of the opponent (resident or colonizer). When facing La. niger, the exploratory behavior of L. humile workers was inhibited, especially when L. humile was colonizer. Workers of L. humile brought very few resources, were almost absent from the resources, engaged in fewer fights and killed fewer opponents when facing La. niger. In contrast, La. neglectus was not so impactful, not modifying the number of L. humile in the arena or the quantity of resources brought by L. humile, although L. humile engaged in more fights with La. neglectus and killed more opponents. Contrary to what we expected, La. niger showed greater competitive ability to hinder L. humile resource exploitation than La. neglectus (Figure 4). workers was inhibited, especially when L. humile was colonizer. Workers of L. humile brought very few resources, were almost absent from the resources, engaged in fewer fights and killed fewer opponents when facing La. niger. In contrast, La. neglectus was not so impactful, not modifying the number of L. humile in the arena or the quantity of resources brought by L. humile, although L. humile engaged in more fights with La. neglectus and killed more opponents. Contrary to what we expected, La. niger showed greater competitive ability to hinder L. humile resource exploitation than La. neglectus (Figure 4). Our main outcomes suggested that the native, and not the invasive Lasius species, may reduce the propensity of the Argentine ant to collect resources through interference competition. This finding is in contrast with the general established knowledge that aggressiveness and fight strategies of L. humile may explain its superiority over native ant species [8,12]. For instance, Carpintero and Reyes-López [9] showed that the Argentine ant is a competitively dominant species, because of its aggressive behavior and relative abundance, compared to native species such as Cataglyphis floricola Tinaut, Camponotus pilicornis (Roger) or Pheidole pallidula (Nylander) or Aphaenogaster senilis Mayr. When confronted with the Argentine ant, eight native species tended to retreat more frequently than Argentine ants (which also had initiated most of the encounters), which could help them to displace native species [16]. However, some cases in the literature have also found that specific local ants can Our main outcomes suggested that the native, and not the invasive Lasius species, may reduce the propensity of the Argentine ant to collect resources through interference competition. This finding is in contrast with the general established knowledge that aggressiveness and fight strategies of L. humile may explain its superiority over native ant species [8,12]. For instance, Carpintero and Reyes-López [9] showed that the Argentine ant is a competitively dominant species, because of its aggressive behavior and relative abundance, compared to native species such as Cataglyphis floricola Tinaut, Camponotus pilicornis (Roger) or Pheidole pallidula (Nylander) or Aphaenogaster senilis Mayr. When confronted with the Argentine ant, eight native species tended to retreat more frequently than Argentine ants (which also had initiated most of the encounters), which could help them to displace native species [16]. However, some cases in the literature have also found that specific local ants can offer strong resistance and delay or prevent the spread of Argentine ants, especially when encountering ecologically dominant or functionally similar native species (e.g., T. group nigerrimum [18], I. rufoniger [17], La. grandis Forel [14] or Prenolepis imparis (Say) [39]).
When facing La. niger, the risk perceived by L. humile appeared to decrease strongly its propensity to explore the arena, both for fighting and foraging, especially when exploring novel areas already colonized by La. niger. It has been suggested that chemicals laid in the environment could be used by L. humile as cues for the presence of local species, as they play an important role in the interactions [39]. A behavioral response to allocolonial or allospecific footprint cues could prevent encounters of potential competitors and thus be beneficial by reducing costs from competition [40]. Subordinate ant species avoided cuticular hydrocarbons of dominant species [41]. Lasius niger strongly differentiated between different cue types and avoided cues of allospecifics and allocolonial conspecifics [40]. In the present study, L. humile could thus have used chemical signals as well as direct allospecific interactions to evaluate the risk, and limited more its exploration activity when resident La. niger were present than when La. neglectus was present. By decreasing the ability of L. humile to collect resources, La. niger could also limit its expansion. This result is even more important as this dominant species is widely established in many environments, and the Argentine ant could therefore be challenged by habitats already occupied by the native ant La. niger when spreading north along Europe.
Conversely, some invasive species could be ineffective when facing L. humile, such as La. neglectus in the present study. For instance, T. magnum Mayr has shown considerable potential as an invasive species in Northern Europe, where it thrives in many urban areas and is considered a pest [42]. A recent study showed that T. magnum was systematically excluded by L. humile from resources and underwent a visible reduction in activity [43]. The foraging activity of L. humile was nevertheless reduced in the presence of T. magnum due to an increased worker commitment, both in the arena fights and directly in the colony of the competitor. In the present study, the native species La. niger was not excluded from the arena by L. humile, and its presence also resulted in a reduced foraging activity of L. humile due to an increased worker commitment in the arena fights. Moreover, Leonetti et al. [43] suggested that the nest of the opponent could have been perceived as a new potential threat or resource and, therefore, a more valid target for recruitment than the trophic resource. However, this result could differ for other food contents, such as carbohydrate food that could be more attractive than protein sources [44]. For instance, La. niger showed clear avoiding behavior when encountering the dominant species Formica fuscocinerea (Forel) at a carbohydrate-rich food source [45].
Our results showed that La. niger suffered very few losses during interactions with colonizer Argentine ants (mortality <5%), especially compared to La. neglectus (mortality~10%). This finding contrasts with Bertelsmeier et al. [21], where dyadic interactions between La. neglectus and L. humile led to 90% mortality for L. humile and 35% for La. neglectus. The latter ranked second in the dominance hierarchy established between seven highly invasive ant species, and ranked before L. humile. The results observed here when L. humile interacted with La. niger or La. neglectus were therefore unexpected. The fact that the study of Bertelsmeier et al. [21] was based only on dyadic or ten versus ten interactions could explain the differences observed with our results. Although interference competition involves aggressive encounters between workers, including physical and chemical aggressions, it does not imply an important effect at the colony level [46,47]. In our experiments, we used colony fragments constituted by a hundred of workers with brood and queens for both L. humile and La. neglectus. In these situations, the possibility of many recruits implicated in the fights could thus be the cause of reduced differences in the outcomes of interactions with respect to the results of Bertelsmeier et al. [21]. This is especially relevant for L. humile, which often relies on group-level processes to successfully displace other species based on cooperative fighting [7,12,16]. Moreover, irrespective of their position in the dominance hierarchy, L. humile can adopt "the bourgeois strategy" during agonistic encounters with other species, changing its behavior based on numerical dominance: lone workers tend to be submissive in encounters, whereas, when numerically dominant, workers are aggressive. Linepithema humile has been shown to adopt this strategy against native species such as P. pallidula, T. group nigerrimum, or Monomorium antarcticum (Smith) [9,18,48]. Moreover, the bourgeois strategy could be a mechanism used by L. humile to co-occur with La. neglectus [49]. In this sense, because of the recent expansion of La. neglectus in Europe, where L. humile already invades, it was unexpected that the "resident effect" was not a primary determinant of the behavioral syndrome involved in competitive interactions for these species. Most of the parameters studied suggested that the species' identity was more important than the status as resident or colonizer. However, we evaluated the resident effect as the access to the foraging arena, without accounting for the advantage of an earlier establishment and the consequent increase in density and numerical dominance [50], which could partly limit our results.

Conclusions
Our results suggest that native La. niger species could impact invasive ant populations of L. humile by means of interference competition. Although rare, this report is not the first instance of biotic resistance of native species against Argentine ants through physical aggression or chemicals to successfully defend themselves and their territory (e.g., [39]). Whereas L. humile has become invasive in Southern Europe, the invasion process could be slowed down in Northern Europe by such native dominant species rather than by other invasive ones, even under the scenario of co-occurrence between several highly invasive species [21]. However, the antagonistic behavior observed in this study does not necessarily imply that interspecific competition would take place in real conditions, because competition is a process of populations, not of individuals [5]. Further studies on the natural invasion progress and species competitive interactions in the field, especially including the study of long-term interactions, would be needed to eventually extend our conclusions.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
Appendix A Table A1. Description of the sets of 33 variables used to describe the competitive interactions. Descriptors in italics were removed from the study following the collinearity analysis (Spearman correlation value >0.7 with others).  Figure A1. PCA1 biplot along the two principal components; associating opponent species (LhC: Linepithema humile colonizer, LhR: L. humile resident; Ln: facing Lasius neglectus, Lni: facing Lasius niger, alone: without opponent) with eight variables reflecting space occupation by L. humile and its ability to exploit resources (n = 60; ellipses confidence intervals of 0.9; large points = mean points of groups (barycenters)).    Table A2. Cosines between the variables and the first (DS1) and second (DS2) axes of the linear discriminant analyses 1 and 2 (LDA1 and LDA2). Cosines >0.5 (as an absolute value) were indicated in bold.