Biol Invasions (2010) 12:677–687 DOI 10.1007/s10530-009-9473-9
ORIGINAL PAPER
Supercolony mosaics: two different invasions by the yellow crazy ant, Anoplolepis gracilipes, on Christmas Island, Indian Ocean Melissa L. Thomas Æ Katrin Becker Æ Kirsti Abbott Æ Heike Feldhaar
Received: 11 September 2008 / Accepted: 8 April 2009 / Published online: 23 April 2009 Ó Springer Science+Business Media B.V. 2009
Abstract Invasive species are one of the main reasons for the ongoing global loss of biodiversity. Anoplolepis gracilipes is an invasive ant that has recently received significant attention due to its negative effect on the native fauna and flora of Christmas Island, Indian Ocean. This species has contributed to a drastic change in the structure of the Christmas Island rainforest through its negative impact on the island’s endemic red land crab, the dominant consumer on the islands forest floor. In this study, we investigate the population structure of M. L. Thomas Parks Australia North, P.O. Box 867, Christmas Island, Indian Ocean, ACT 6798, Australia Present Address: M. L. Thomas Centre for Evolutionary Biology, School of Animal Biology, University of Western Australia, Crawley, WA 6009, Australia K. Becker H. Feldhaar Department of Behavioural Physiology and Sociobiology (Zoology II), Biocenter, University of Wu¨rzburg, Am Hubland, 97074 Wu¨rzburg, Germany K. Abbott Science Faculty, Monash University, Melbourne, VIC 3800, Australia H. Feldhaar (&) Behavioural Biology, University of Osnabru¨ck, Barbarastr. 11, 49076 Osnabru¨ck, Germany e-mail:
[email protected]
A. gracilipes on Christmas Island in order to determine whether multiple introductions occurred on the island and how they correspond to known infestations. We genotyped 578 individuals collected from 50 nests across the Island. We identify two distinct subgroups in the population that represent two different supercolonies. These supercolonies are interspersed across the island, however both nuclear (microsatellites) and mitochondrial markers strongly suggest that there is no gene flow between the two colonies. Significant heterozygote excess within the entire sampling area, with all but one worker examined being heterozygous for all seven microsatellite loci, suggests an unusual reproductive system in these ants. Our results are consistent with recent sociogenetic findings in a population of A. gracilipes in Northern Borneo. Keywords Anoplolepis gracilipes Yellow crazy ant Microsatellites Invasive species Supercolony Christmas Island
Introduction Biological invasions are a leading threat to biodiversity worldwide, and invasive ants are among the most damaging of invasive species (Moller 1996). Invasions by non-native ants can cause considerable economic damage (Lard et al. 2002; Gutrich et al. 2007), disrupt local arthropod communities (Sanders et al. 2003; Holway and Suarez 2006), and once
123
678
established, are extremely difficult to control (Causton et al. 2005; Silverman and Brightwell 2008). One of the most devastating outcomes of an invasion by non-native ants has been reported from Christmas Island, an uplifted oceanic island that lies in the Indian Ocean, 360 km south of Java. Here, the invasive yellow crazy ant, Anoplolepis gracilipes (Smith), is responsible for dramatic changes in the composition and structure of the rainforest (O’Dowd et al. 2003). By causing dramatic declines in local populations of a keystone species, the endemic red land crab (Gecarcoidea natalis), the yellow crazy ant has indirectly caused seedling recruitment in plant species whose recruitment is normally suppressed by the crabs (O’Dowd et al. 2003). Moreover, the yellow crazy ant has been implicated in promoting damaging outbreaks of scale insects on Christmas Island (Abbott and Green 2007; O’Dowd et al. 2003), which promote the growth of sooty moulds, and probably cause canopy dieback and tree death (O’Dowd et al. 2003). One of the factors thought to contribute to the ecological success of invasive ants is their colony structure. In their introduced ranges, invasive ants often form unicolonial populations or supercolonies (Holway et al. 2002), a trait which has also been assigned to populations of A. gracilipes that are thought to be introduced (Abbott 2005, 2006; Drescher et al. 2007; Holway et al. 2002; O’Dowd et al. 2003). Characteristics of supercolonies include low relatedness amongst nestmates, the presence of multiple queens, and a lack of clear behavioural boundaries or aggression between workers from distinct nests. One hypothesis for the success of invasive ants is that changes in social organization lead to a lack of aggression between workers from distinct nests within a supercolony, which may reduce costs associated with territoriality, and enable high worker densities and therefore numerical dominance within the invaded habitat (Holway et al. 1998; Tsutsui and Suarez 2003; Errard et al. 2005). However, this hypothesis is often debated in the literature, and a number of alternative hypotheses for the success of invasive ants have also been proposed, e.g. a lack of biotic and abiotic constraints present in the native range or higher competitiveness (Human and Gordon 1999; Holway et al. 2002) With the exception of nestmates showing high rather than low relatedness
123
M. L. Thomas et al.
amongst nestmates (Drescher et al. 2007), colonies of A. gracilipes display all the characteristics of supercolonies. On Christmas Island, workers of A. gracilipes display little to no aggression against conspecifics from spatially distinct nests (Abbott 2005). However, Abbott (2005) found consistent aggressive interactions between workers from two nests, suggesting that more than one supercolony may in fact exist on Christmas Island. Anoplolepis gracilipes was most probably introduced accidentally to Christmas Island sometime between 1915 and 1934 (Crawley 1915; Donisthorpe 1935), it was not until the early 1990’s that populations exploded, resulting in the significant ecological damage present today. Interestingly, the yellow crazy ant was not reported on Christmas Island between 1935 and 1989, despite a number of arthropod surveys, suggesting particularly low numbers during this period (Framenau and Thomas 2008). A second invasion by genetically distinct individuals of A. gracilipes could provide an explanation for the temporal variation in abundance and impacts observed on Christmas island since 1915, given that a higher genetic diversity may enhance adaptive evolution of the invasive species (Sakai et al. 2001). A similar hypothesis has been put forward to explain the invasion success of a number of other species (Sakai et al. 2001; Simberloff and Gibbons 2004), including the population of A. gracilipes on Tokelau (Abbott et al. 2007). Here we examine the population structure of A. gracilipes on Christmas Island. We use microsatellite markers and mitochondrial DNA sequences to identify distinct populations, and determine whether multiple introductions might have occurred on the island. The number of effective introductions and how they spread has important implications for quarantine and control measures on oceanic islands, and invasion epicentres elsewhere.
Materials and methods Sampling methods for genetic analysis We collected workers of A. gracilipes from 50 nests distributed across Christmas Island (105°400 E, 10°300 S) from June to August 2005 (Fig. 1). Christmas island covers approximately 135 km2, with a
Population structure of Anoplolepis gracilipes on Christmas Island
679
Fig. 1 Sampling sites of Anoplolepis gracilipes on Christmas Island. The symbols correspond to clusters inferred with STRUCTURE based on microsatellite data (circle supercolony A, triangle supercolony B). From sampling sites marked with open symbols a total of 904 bp of mitochondrial DNA was sequenced from one individual per nest in addition to the nuclear microsatellite loci. Workers collected at sites marked
with a closed symbol were only genotyped with microsatellites. Supercolony B (triangles) contains a single mitochondrial DNA haplotype only, whereas in supercolony A (circles) contains a second frequent haplotype as well as one unique haplotype (circle marked with asterisk) (see Fig. 6) that was one mutational step apart from all other individuals sequenced from this supercolony
coastline consisting predominantly of sheer rocky cliffs from 10 to 20 m high interspersed with a few small beaches. The interior is a slightly undulating plateau, from 160 to 360 m above sea level and predominantly covered by tall evergreen closed forest (Claussen 2005). A series of steep slopes or cliffs with intervening narrow terraces separate the central plateau from the shore. Although A. gracilipes is present on the central plateau of Christmas Island, this species is predominantly found on the Island’s terraces. Our collection sites reflect this somewhat patchy distribution. For more details on the distribution of A. gracilipes on Christmas Island please refer to Framenau and Thomas (2008). Workers were collected by hand on foraging trails near the nest entrance to ensure colony affiliation, and were immediately transferred into 100% EtOH. Ants were transferred to propyleneglycol for international transport, but immediately returned to 100% EtOH upon arrival and before analysis.
Microsatellite analysis In total, 578 individuals from 50 nests (range 7–17, and mean 11.3 individuals per nest) were analyzed at seven polymorphic microsatellite loci: Ano1, Ano3, Ano4, Ano5, Ano6, Ano8, and Ano10 (Drescher et al. 2007; Feldhaar et al. 2006). DNA was extracted using the PuregeneÒ DNA Purification Kit (Gentra Systems) according to the manufacturer’s recommendations. After isolation, the DNA pellet was resuspended in 50 ll sterile double distilled water and stored at -20°C. PCR amplification was performed in an Eppendorf or Biometra thermocycler in a total reaction volume of 12.5 ll containing approximately 10 ng of template DNA, 19 PCR-buffer, 2 mM MgCl2, 160 lM dNTPs, 2.5 lM of each primer (forward primer labelled with fluorescent IR-700 or IR-800 dye by Licor) and 0.5 U of Taq DNA polymerase (MolTaqÒ by Molzym GmbH). Cycle parameters were as
123
680
follows: 3 min at 94°C, followed by 30 cycles of 94°C for 40 s, annealing step 40 s, 72°C for 40 s, and a final extension of 3 min at 72°C (annealing temperatures: Ano1, 57°C; Ano3, 55°C; Ano4, 60°C; Ano5, 60°C; Ano6, 51°C; Ano8, 62°C and Ano10, 52°C). PCR-products were diluted between 1:10 and 1:40 and analyzed on a LICORÒ 4300 DNA Analyzer. Mitochondrial DNA analysis Mitochondrial DNA of one individual each from 22 nests was sequenced. Samples were chosen from both putative supercolonies of A. gracilipes identified using microsatellite data. We picked nests from all areas where both supercolonies occurred adjacent to each other from different regions of the island. For amplification we used primer pairs Horst [50 -AC(TC) ATACTTTTAACTGATCG-30 ] designed by D. Kronauer (unpublished data)/Ben [50 -GC(AT)AC(AT) AC(AG)TAATA(GT)GTATCATG-30 ] (Dejean et al. 2005) or partial cytochromeoxidase I (COI) (GenBank accession numbers: EU931670–EU931672) corresponding to positions 2407–2427 (Horst) and 2891– 2914 (BEN) relative to the mitochondrial genome of Apis mellifera (Crozier and Crozier 1993) and CBI/ CBII for partial cytochrome B (Cytb) (GenBank accession numbers:EU931667–EU931669) (Crozier et al. 1991). PCR was performed in a total reaction volume of 25 ll containing approximately 10 ng of template DNA, 19 PCR buffer, 2 mM MgCl2, 240 lM dNTP’s, 800 lM of both forward and reverse primer and 1.2 U of Taq DNA polymerase (MolTaq by Molzym GmbH). Amplification via PCR of both fragments of mtDNA was performed either in an Eppendorf or Biometra thermocycler at the following conditions: 3 min at 94°C, followed by 30 cycles of 94°C for 1 min, 1 min for the annealing step at 45°C, 1.5 min at 72°C and a final extension of 3 min at 72°C. The purified mtDNA fragments were sequenced by SEQLAB sequence laboratories (Go¨ttingen, Germany), analyzed and aligned using BioEdit version 7.0.5.3 (Hall 1999) (http://www.mbio.ncsu.edu/ BioEdit/BioEdit.zip). A haplotype network was built based on the concatenated sequence of 904 bp (consisting of a 460 bp fragment of the COI gene and a 444 bp fragment of the Cytb) using the
123
M. L. Thomas et al.
95% parsimony algorithm implemented in TCS 1.21 (Clement et al. 2000). Data analysis: genetic diversity, Hardy– Weinberg-equilibrium and linkage disequilibrium Characteristics of the microsatellite markers were calculated over loci using the programmes Genepop 3.4 (http://wbiomed.curtin.edu.au/Genepop/Genepop_ op1.html) (Raymond and Rousset 1995) and Fstat 2.9.3 (http://www.unil.ch/izea/softwares/fstat.html) (Goudet 2001), update from (Goudet 1995). The calculated parameters include the number of alleles A, allelic richness An, the expected heterozygosity HE and the observed heterozygosity HO. Hardy–Weinberg-equilibrium (HWE) and linkage disequilibrium (LD) were tested using Genepop 3.4 (Raymond and Rousset 1995), taking a single randomly chosen worker from each nest sampled in order to avoid pseudoreplication due to high relatedness of individuals within colonies. For linkage disequilibrium (LD), we conducted a total of ten calculations by using a different individual from each nest in each replicate. P-values were obtained after a sequential Bonferroni adjustment for multiple tests (Rice 1989). Data analysis: test of population structure and assignment to supercolonies To assess whether nests could be grouped into genetically distinct supercolonies, we estimated the number of populations present on the island using STRUCTURE vers. 2.2 (Falush et al. 2003; Pritchard et al. 2000). The program is based on the minimization of Hardy–Weinberg and linkage disequilibrium, which would be the result of population substructure within the population studied. We derived the most probable number of populations K by calculating the posterior probability distribution for different fixed values of K. The program was run with the whole dataset (all individuals per nest genotyped as well as with one worker per nest only (first worker genotyped per nest). We performed 10 runs each for K fixed from 1 to 10 without prior population information with 50,000 replicates of the MCMC after 10,000 replicates that were discarded as burnin. We used the default admixture model, with uniform prior on the degree of admixture (default settings) and correlated allele frequencies (default settings).
Population structure of Anoplolepis gracilipes on Christmas Island
681
Due to high heterozygosity levels of A. gracilipes workers, and the possibility of an unusual mating system in this ant species (Drescher et al. 2007), we also tested the most probable population structure found with STRUCTURE with a principle component analysis of the multilocus genotypes using PCAGen (www2.unil.ch/popgen/softwares/pcagen. htm). This program compares similarity of genotypes without any assumption concerning the mating system or Hardy–Weinberg-equilibrium. 10,000 permutations were run, with each of the 50 nests representing a population. Following the identification of putative supercolonies we estimated geneflow among colonies with the private allele method (Slatkin 1985) implemented in Genepop.
Table 1 Genetic diversity of Anoplolepis gracilipes on Christmas Island
Data analysis: relatedness
worker that was homozygous at six of the loci. Thus, all seven microsatellite markers showed highly significant (P \ 0.0001) departure from Hardy– Weinberg-equilibrium (Table 1).
Relatedness was calculated within all 50 nests sampled for genetic analysis using Relatedness 5.0.8 (Queller and Goodnight 1989) (http://es.rice.edu/ projects/Bios321/relatedness.html), using all genotyped workers as the reference population. The index of relatedness R by Queller and Goodnight (1989) weights each allele inversely by its frequency in the population, so that rare alleles are given a relatively higher weight. Following population clustering with STRUCTURE and PCAGen we also calculated relatedness within and between the two putative supercolonies.
Results Genetic diversity at microsatellite loci We genotyped a total of 578 individuals at 7 microsatellite loci. All individuals collected were genotyped successfully for all loci, except for 10 individuals from different nests that were genotyped at only six loci. The number of alleles per locus (A) ranged from 3 to 20 in the sampled area (Table 1). Across loci, the expected heterozygosity HE ranges from 0.565 to 0.811 (overall HE = 0.697), while the observed heterozygosity HO was 1.0 (overall HO = 1.0), strongly exceeding the expected heterozygosity HE in all seven loci (Table 1). All workers were heterozygous at all loci, except for a single
Range (bp)
NI/NC
H
HO
PHWE
Ano 1
99–105
578/50
Ano 3
140–166
577/50
3
0.565
1
