Conserv Genet DOI 10.1007/s10592-014-0646-4
RESEARCH ARTICLE
The relationship between distance and genetic similarity among invasive rat populations in the Falkland Islands Michael A. Tabak • Sally Poncet • Ken Passfield Matthew D. Carling • Carlos Martinez del Rio
•
Received: 1 August 2014 / Accepted: 7 August 2014 Ó Springer Science+Business Media Dordrecht 2014
Abstract Norway rats (Rattus norvegicus) have been either introduced to or have invaded a large number of the world’s islands. Rats have caused population declines and the extinction of island endemics. Their eradication is often a major conservation success as it leads to recovery of affected species and even ecosystem processes. However, eradication efforts can be hampered by the ability of rats to re-colonize eradicated islands. Here we present the results of genetic analyses that inform migration rates and population structure of Norway rats from 14 sampling locations in the Falkland Islands, where rat eradication efforts are taking place. We used 12 microsatellite markers and population genetic tools to estimate rat migration patterns between 12 islands that were separated by distances ranging from 230 m to 112 km. We found evidence of significant migration rates, and hence presumably of rat movements between islands up to 830 m away from each
Electronic supplementary material The online version of this article (doi:10.1007/s10592-014-0646-4) contains supplementary material, which is available to authorized users. M. A. Tabak (&) M. D. Carling C. Martinez del Rio Department of Zoology and Physiology, University of Wyoming, 1000 E. University Ave, Laramie, WY 82071, USA e-mail:
[email protected] M. A. Tabak M. D. Carling C. Martinez del Rio Program in Ecology, University of Wyoming, 1000 E. University Ave, Laramie, WY 82071, USA S. Poncet K. Passfield Beaver Island LandCare, PO Box 756, Stanley FIQQ IZZ, Falkland Islands C. Martinez del Rio Wyoming Biodiversity Institute, University of Wyoming, 1000 E. University Ave, Laramie, WY 82071, USA
other. Norway rats seem capable of swimming this distance even in the cold waters of the Falkland Islands. Our results can inform managers about strategies of rat eradication in the Falklands including minimal distances that reduce recolonization and the choice of island clusters for eradication. Keywords Invasive species Islands Eradication Rattus norvegicus Microsatellites
Introduction Introduced organisms are among the greatest threats to biodiversity (Vitousek et al. 1997), especially on islands (Hilton and Cuthbert 2010). Predatory mammals such as rats, cats, and mongooses (Courchamp et al. 2003) can have especially strong effects on island biotas (Atkinson 1985). The widely documented negative effects of these animals on native island species have led conservation managers to attempt to remove them from islands around the world (Keitt et al. 2011). These eradications are often successful at removing the target species from an island (Genovesi 2011) and can have clear benefits for native species (Towns 2009) and even ecosystem processes (Jones 2010). The removal of invasive species from islands has led to a large number of conservation successes (Towns 2011). Unfortunately, target species can recolonize islands following eradication (Oppel et al. 2011). Thus, eradication efforts on islands must recognize the possibility of recolonization from nearby sources and include design schemes that minimize this possibility (Abdelkrim et al. 2005). Future eradication efforts require a better understanding of the capacity of invasive organisms to move among islands.
123
Conserv Genet
One way to assess the probability that invasives will recolonize islands following eradication is to use genetic data to help quantify the movement of invasive species among islands prior to eradications (Fewster et al. 2011). Genetic data can be used to estimate the probability of migration events between islands at different distances from each other (Wilson and Rannala 2003), and hence be helpful in establishing ‘‘eradication units’’: groups of islands that are geographically close enough that migrants can move between the islands, but are far enough as a group from a source of invaders (Abdelkrim et al. 2010). These eradication units must be eradicated simultaneously to prevent reinvasion from islands within the unit (Robertson and Gemmel 2004). The Falkland Islands (or ‘‘Falklands’’) is a relatively remote archipelago in the South Atlantic Ocean. The archipelago consists of two main islands, East and West Falkland, surrounded by over 500 smaller islands. The Falklands have a unique land bird fauna of nine species, eight of which are considered endemic at the species or sub-specific level (Woods and Woods 2006). They include an endemic species of wren (Cobb’s Wren, Troglodytes cobbi) and a rare ovenbird (Tussacbird, Cinclodes antarcticus) that was historically found in Tierra del Fuego (Pin˜a and Cifuentes 2004), but it is possible that Tussacbird is currently limited in distribution to the Falkland Islands (Strange 1992). Norway rats (Rattus norvegicus, henceforth ‘‘rats’’) are found on many, albeit not all, of the islands in the Falkland archipelago (Woods and Woods 2006). The presence of rats in the Falkland Islands is associated with a reduction in land bird diversity and with the disappearance of Cobb’s Wren and Tussacbird from islands occupied by rats (Hall et al. 2002; Tabak et al. 2014). Cobb’s Wren and Tussacbird nest on the ground and hence are likely especially vulnerable to rats (Woods and Woods 2006). Protecting them from local, and perhaps even global, extinction requires the maintenance of rat-free habitats in the Falkland Islands (Tabak et al. 2014). Rats have been eradicated from 40 islands in the Falklands. In 1–11 years following eradications the number of land bird species on eradicated islands becomes similar to that of rat-free islands, and is significantly higher than on islands with rats (Poncet and Passfield unpubl. data). The presence and abundance of Tussacbird as well as that of many other species increased rapidly after eradication. Eradication of rats seems to be an effective means of restoring land bird diversity in the Falklands. However, rats have recolonized several islands from which they were eradicated, presumably by swimming from nearby islands occupied by rats (Poncet et al. 2011). The possibility of recolonization is a major challenge for the success of eradication efforts in the Falklands (Brown 2008). An estimate of the maximum swimming distance of rats can
123
benefit conservation efforts by helping to determine which islands are good candidates for eradications and allow managers to establish eradication units. Data on swimming endurance of rats can inform the distances that they can traverse between islands. Swimming endurance data are obtained primarily in laboratory experiments (Russell et al. 2008a). The maximum speed at which Norway rats have been recorded swimming in the laboratory is 1.4 km/h (Dagg and Windsor 1972). The length of time that they can survive in water decreases with water temperature. At 9 °C, the time is approximately 9 ± 1 min (Le Blanc 1958). Ocean surface water temperature in the Falklands ranges from 5 to 8 °C (Waluda et al. 1999; Agnew et al. 2000). Hence, assuming that Norway rats can remain in water for 10 min and that they swim at their maximum speed the entire time, their estimated maximum swimming distance is 233 m. This value, however, appears to greatly underestimate the swimming abilities of rats in the Falklands. Using a large data set (158 islands), Tabak et al. (submitted) estimated the relationship between the probability of rat occupancy on an island and the distance to the nearest rat source (i.e., another island with rats). They found that the probability of finding rats on an island decreased with distance from the nearest rat source: 72 % of islands within 500 m of a source of rats had rat populations and 40 % of islands between 500 and 1,000 m from a source had rat populations. Of the 69 islands in their dataset that were farther than 1,000 m from the nearest rat source, only two had rat populations. They concluded that rats in the Falklands are much better swimmers than laboratory data suggests and that 1,000 m was a conservative distance for successful and safe eradication. Here we used the frequencies of genetic markers among rat populations from islands separated by a large range of distances to investigate population structure and migration patterns of rats in the Falklands and hence to add to the body of data that informs eradication strategies.
Methods Sample collection From March to April 2013, we collected rats on 12 offshore islands in the Falkland Islands. We collected rats from three island clusters throughout the archipelago: Arch Islands, River harbour, and Adventure Sound (Fig. 1, Table 1). In each island cluster, we deployed snap traps (VictorPest) along the coast of four offshore islands as well as on the adjacent mainland (‘‘mainland’’ refers to the two large landmasses of East Falkland and West Falkland). Islands ranged in size from 4 to 2,070 hectares. We deployed 100–200 traps at each location for 14 days or
Conserv Genet
Fig. 1 The Falkland Islands, an archipelago 500 km off the coast of South America, is made up of two large landmasses (East and West Falkland) and about 500 smaller islands offshore from these two
larger islands. We sampled rats from four islands in each of three island clusters: Arch Islands, River Harbour, and Adventure Sound. Note that the maps for each island cluster are drawn to different scales
until 30 individuals were collected. On the biggest island in our survey, Bleaker Island (in Adventure Sound, Fig. 1), we sampled the island from three distinct sections; North, Central, and South; and captured rats from each section. The three locations on Bleaker Island were treated as separate locations for all analyses except one analysis of migration rates. We removed the gastrocnemius muscle from all captured rats and preserved it in 70 % ethanol. We sampled a total of 436 rats (Table 1). We did not capture any rats at mainland sites (East and West Falkland) despite substantial effort (815 total trap nights at two mainland locations).
ratgtg3, ratcyp2a3a, ratglut, and ratrjg9 at Ta = 62 °C; and panel three included primer sets ratil6 g, ratthy1 g, and ratiid2 g at Ta = 60 °C. PCR was performed in 10 lL volumes with 4 lL DNA (5 ng/lL), 5 lL Multiplex PCR Mix (Qiagen), and 1 lL panel mix (Supplementary Table 1). All PCR amplifications were carried out on a thermal cycler with an initial denaturation at 95 °C for 15 min; followed by 40 cycles of 95 °C for 30 s, 45 s of panel-specific Ta, and 45 s of 72 °C; and a final extension of 30 min at 72 °C. PCR products were run with Hi-Di Formamide (Life Technologies) and 500 LIZ size standard (GeneScan) on a Prism 3,730 DNA Analyzer (Applied Biosystems). The size of the amplified products were determined using GeneMapper (Applied Biosystems).
Microsatellite genotyping DNA was extracted from the gastrocnemius muscle of each captured individual using DNeasy Blood and Tissue kits (Qiagen). We used 12 microsatellite markers developed previously for Norway rats (Kunieda et al. 1992) to construct multilocus genotypes for all individuals. These markers were separated into three panels based on their PCR annealing temperatures (Ta). Panel one included the primer sets ratelai1, ratmlvi2e, ratapoa02, ratmyc, and ratfibg2 at Ta = 56 °C; panel two included primer sets
Population genetic structure We used Genepop v.4.2 (Raymond and Rousset 1995), to test for Hardy–Weinberg Equilibrium (HWE) at each polymorphic locus and linkage disequilibrium (LD), mean gene diversity, and inbreeding coefficients at each sampling location. For the HWE and LD analyses, we corrected our a value for multiple comparisons using a sequential Bonferroni adjustment (Holm 1979; Rice 1989).
123
Conserv Genet Table 1 Results of sampling from 12 islands and three locations on Bleaker Island Sampling location
Island group
Number of rats sampled
Number of polymorphic loci
Inbreeding coefficient
Mean number of alleles per locus (±SD)
Mean gene diversity over loci (±SD)
Big Arch Island
Arch Islands
35
11
0.41
4.0 (1.7)
0.42 (0.23)
Natural Arch Island
Arch Islands
33
11
0.43
3.6 (1.3)
0.34 (0.19)
Sand Bay Island
Arch Islands
35
11
0.62
3.1 (1.3)
0.35 (0.20)
Tussac Island
Arch Islands
35
11
0.57
3.5 (1.9)
0.37 (0.21)
Kent Island
River Harbour
33
11
0.34
4.5 (1.3)
0.53 (0.28)
Kent Island Knob
River Harbour
22
12
0.35
3.8 (1.3)
0.58 (0.31)
Middle Island Rabbit Island
River Harbour River Harbour
12 32
12 12
0.35 0.41
4.4 (1.4) 4.4 (1.4)
0.60 (0.33) 0.53 (0.28)
Bleaker Island (North)
Adventure Sound
32
10
0.32
4.1 (2.3)
0.29 0.17)
Bleaker Island (Center)
Adventure Sound
35
11
0.35
4.0 (2.1)
0.34 (0.19)
Bleaker Island (South)
Adventure Sound
33
10
0.35
3.7 (2.2)
0.40 (0.22)
Cassie’s
Adventure Sound
28
11
0.52
3.0 (1.0)
0.37 (0.20)
Driftwood Island
Adventure Sound
36
11
0.43
3.4 (1.1)
0.38 (0.21)
Urchin Island
Adventure Sound
35
11
0.67
2.8 (0.7)
0.31 (0.18)
We estimated FST using Weir and Cockerham’s (1984) method. We tested whether each of these pairwise FST values differed from 0 (i.e., whether there was significant differentiation between two populations) by simulating 104 permutations of genotypes among samples using Arlequin v.3.5.1.2 (Excoffier and Lischer 2010). We did not use estimates of FST to infer effective migration (Whitlock and McCauley 1999; Meirmans and Hedrick 2011); instead, we used these statistics as estimates of genetic differentiation. We used analysis of molecular variance (AMOVA) to determine how much genetic variation was explained by variation within sampling locations, among sampling locations within the same island group, and among island groups. This analysis was performed in Arlequin v.3.5.1.2 using 10,100 permutations. We inferred the most likely number of distinct populations (K) using a Bayesian clustering method in STRUCTURE v.2.3.4 (Pritchard et al. 2000). We searched for K by running the program 20 times for each value of K, ranging from 1 to 14. We used an admixture model that did not include a priori information about population membership of individuals. This model allowed for correlation between linked loci, or ‘‘admixture linkage disequilibrium’’ (Falush et al. 2003). Using preliminary tests of the convergence time needed for the Markov Chain Monte-Carlo (MCMC), we settled on a burn-in period of 106 steps followed by 9 9 106 steps. We compared likelihood values for each value of K and calculated DK, a statistic based on the rate of change in the log likelihood of data between successive K values (Evanno et al. 2005). We used DK to detect the optimal number of populations and the uppermost hierarchical level of structure; this statistic was calculated using STRUCTURE HARVESTER v.0.6.93 (Earl and vonHoldt 2012). Once
123
we determined the optimal value for K, we compared the mean posterior proportion of each individual’s ancestry arising from each cluster. We also evaluated K using a Discriminant Analysis of Principal Components (DAPC; Jombart et al. 2010). DAPC was implemented using the R package adegenet v.1.4–2 (Jombart and Ahmed 2011). We selected the optimal value for K using BIC scores and compared this value with K estimated from STRUCTURE. To determine if there was a correlation between geographic Euclidian distance and genetic diversification (i.e., isolation by distance), we used a Mantel test in Arlequin v.3.5.1.2. We ran the Mantel test using data from individuals from all sampling locations. We estimated the significance of the correlation between geographic distance and genetic diversification using randomization techniques with 104 replicates. Migration rates We estimated rates of recent migration (i.e., migration over the last several generations) using a Bayesian method with BAYESASS v.3.0.3 (Wilson and Rannala 2003). Using BAYESASS, we ran two analyses; one in which all islands were treated as separate locations (the three locations on Bleaker Island were treated as one location; 12 total locations) and one in which all sampling locations were treated as separate locations (Bleaker Island was treated as three locations; 14 total locations). For each analysis, we ran 107 MCMC iterations with a burn-in period of 106 iterations and an interval of 1,000 steps between sampling. We experimented with various delta values for mixing parameters: migration rates (m), allele frequencies (a), and inbreeding values (f). We settled on values for the
Conserv Genet
parameters so that the acceptance rate for changes to each of the parameters was between 20 and 60 %. We compared chains for convergence to a stationary posterior distribution by performing multiple runs starting with different initial random seeds. We calculated 95 % credible intervals associated with each migration estimate. We classified migration rates as negligible if the 95 % credible intervals included zero. Migration rates were considered statistically significant if the 95 % credible interval did not include zero. We assume that rat migration among offshore islands is the result of rats swimming on their own accord. It is likely that rat migration is not human-assisted for several reasons: the human population is very small (2,932 inhabitants; Falkland Islands Government 2012), there is no inshore fishing, ‘‘adventure tourism’’ is negligible, and there is no camping or small boat access to most islands. Furthermore, the likelihood of human-caused rat introductions to islands sampled is very low because only three islands in our study are commonly visited by humans (more than two visits in the ten years prior to sampling), two islands (Middle Island and Urchin Island) have two visits a year for livestock movements, and Bleaker Island, although inhabited by humans, has a strict biosecurity policy.
Results Population genetic structure We found that all 12 loci were polymorphic in at least ten sampling locations and each sampling location had at least 10 polymorphic loci (Table 1). At each sampling location where a locus was polymorphic, there were no significant deviations from HWE (Supplementary Table 3) and there was no evidence of LD among loci (Supplementary Table 4). Mean gene diversity over loci at each sampling location ranged from 0.29 to 0.60, with a mean of 0.42 (Table 1). All FST values were highly significantly different from 0 (p \ 0.001) except between Bleaker Island Center and Bleaker Island North. Significant FST values ranged from 0.039 to 0.621. Values were generally lower (\0.4) between islands within the same island cluster than among islands from different clusters. Within island groups, we found two instances with high levels of genetic differentiation (FST [ 0.35): Tussac Island and Sand Bay Island in the Arch Islands (FST = 0.377, p \ 0.00,001) and Cassie’s and Bleaker Island North (FST = 0.351, p \ 0.00001) in Adventure Sound were highly genetically distinct from each other. The AMOVA revealed that a large fraction of the total genetic variation (49.7 %) was found within sampling locations; 15.3 % of the total genetic variation
Fig. 2 Estimates of the number of populations (K) and DK suggest that all rats sampled can be grouped into two populations (top panel). These two populations correspond with the island clusters near East (Arch Islands and River Harbour) and West Falkland (Adventure Sound). For estimates of K, the likelihood of the model given the data increased with K especially for smaller populations. We estimated DK using the Evanno method and found a dramatic decrease in DK after a K value of 2. We also estimated K using a discriminant analysis of principal components (bottom panel), which also predicted two populations
documented was the result of differences among locations within the same island groups; and 35.0 % was determined by differences between island groups. Evaluating model fit using the DK and DAPC methods revealed that the individuals sampled are best described as two genetic clusters (K = 2; Fig. 2). These genetic clusters correspond to islands in western (River Harbour and Arch Islands) and eastern sectors (Adventure Sound; Fig. 1) of the Falkland Islands. We compared the mean posterior proportion of ancestry in population one across all 20 runs for each island. For the eight islands in the western sector (i.e., islands in the Arch Islands and River Harbour), we found that the mean posterior proportion of ancestry in population one ranged from 0.946 to 0.998 with a mean of 0.988. For the six sampling locations in the eastern sector (i.e., locations in Adventure Sound), we found that the mean posterior proportion of ancestry in population one ranged from 0.002 to 0.009 with a mean of 0.004. Mantel Tests revealed a strong correlation between geographic distance and genetic differentiation (FST) across all islands (r = 0.798, Fig. 3 Inset). Migration rates In both Bayesian analyses of migration rates, we found negligible migration rates among most locations (Fig. 3), but high rates of migration between two islands in the
123
Conserv Genet
Fig. 3 The percent of individuals on an island that were migrants from another island decreased with distance to that island. Bayesian estimates of migration rates are plotted along with 95 % credible intervals (CIs). For each pair of islands, there are two estimates of migration rates, and both are plotted. Note that these estimates can be asymmetrical. Migration rates for which the 95 % CI included zero,
were deemed statistically insignificantly different from zero. Migration rates for large distances ([3,000 m) are not included in the figure because it is unlikely that rats swim such distances in the Falkland Islands (see discussion). Inset Genetic differentiation (FST) increased with geographic distance short distances (\3,000 m) between islands, but the relationship appeared to reach an asymptote at larger distances
Table 2 Significant migration rates (% of individuals at location 2 that were migrants from location 1) Locations between which migration occurred (location 1 to location 2)
Migration rate using 14 locations
Migration rate using 12 locations
Straight-line distance between locations (m)
Land distance between locations (km)
Kent Island–Kent Island Knob
22.24 ± 5.0
22.94 ± 4.7
290
Rabbit Island–Middle Island
18.17 ± 6.3
17.09 ± 5.8
830
Bleaker Island Center–Bleaker Island North Bleaker Island Center–Bleaker Island South
24.69 ± 4.0
–
7,300
8.7
24.89 ± 3.9
–
11,000
11.8
* *
Migration rates are reported ±95 % credible interval. Analysis with 12 locations treated all locations on Bleaker Island as one location * Land distance not calculated between locations on different islands
River Harbour group (Table 2, Fig. 3). There was no significant difference between estimates of migration rates using either analysis (i.e., treating all sampling locations as different locations and pooling all samples on Bleaker Island produced the same result; the 95 % credible intervals associated with these estimates overlapped, Table 2). Additionally, in the analysis where we treated locations on Bleaker Island separately, there were high levels of migration among locations on Bleaker Island. The Bleaker Island locations were from the same island, so migration between locations did not necessarily include movement across water.
123
Discussion Genetic differentiation among rat populations increased with the geographic distance between islands in the Falkland archipelago. We found significant rates of rat migration among islands up to distances of 830 m and across land up to distances of 11.8 km. Here we compare our estimate of rat migration with the literature and estimate how far rats are capable of moving among islands in the Falkland archipelago: we find that rats are capable of moving between 830 and 1,000 m among islands without human assistance. We argue that the non-linear relationship
Conserv Genet
between genetic divergence and geographic distances for larger distances suggests that rat populations in the Falkland Islands have not reached equilibrium. Furthermore, we suggest that the asymmetrical migration rates among locations might represent a ‘‘source-sink’’ dynamic. We conclude by using our estimates of rat migration to inform future eradication programs in the Falkland Islands. Population structure We found that all sampled individuals represented two distinct genetic clusters (Fig. 2). When we separated all of the sampling locations into geographic sectors (islands in the Arch Islands and River Harbour were in the western sector and locations in Adventure Sound were in the eastern sector, Fig. 1), rat populations were genetically distinct in the eastern and western sectors. This pattern can be explained by two non-exclusive explanations: continued exchange of rat individuals among populations in each sector and a historical legacy. Rats might be able to migrate occasionally (albeit infrequently—see discussion of migration rates) between the Arch Islands and River Harbour in the western sector. Their close association with the large mass of West Falkland can facilitate migration between these two island groups. Islands in Adventure Sound, in contrast, are less likely to exchange migrants with islands in the western sector because Falkland Sound, the 4.5 km-wide channel that separates the two large landmasses of East and West Falkland, might function as a barrier (Fig. 1). It is also possible that rats introduced to islands in the western sector might have the same (or similar) origin, whereas rats introduced to the eastern sector might have a different one (Miller et al. 2005). The differentiation between rat populations from the eastern and western sectors might be the result of two independent rat colonization events. While we found a significant pattern of isolation by distance using a global Mantel Test, the relationship between geographic Euclidian distance and genetic divergence (FST) does not appear to be linear (Fig. 3 Inset). Instead, with larger geographic distances, genetic divergence appears to asymptote and becomes more variable, as the points are more scattered. Hutchinson and Templeton (1999) argue that if populations within a region are at equilibrium, the relationship between geographic and genetic distances should be positive and monotonic across all distances in the region. The relationship we observed indicates that this group of rat populations is not at equilibrium, which is probably the case for most sets of populations in fragmented landscapes (McCauley 1993). One hypothesis for the disequilibrium among these rat populations is that the relatively recent introduction of rats into the Falkland Islands has not allowed them to reach
equilibrium across the entire archipelago. Rats were introduced to the Falklands in the late eighteenth century (Woods and Woods 2006) and they are not able to travel large distances across the sea. We would expect rats to experience a consistent pattern of localized dispersal and genetic drift between populations that are separated by short distances, resulting in small variation in FST values (Crow and Aoki 1984). However, populations separated by larger geographic distances ([3,000 m) experience lower divergence and more variation in FST values because dispersal and drift have not yet reached equilibrium among these populations (Hutchinson and Templeton 1999). Therefore, genetic and geographic distance among Falkland Islands rats do not increase monotonically, but this relationship might become more monotonic with more time since introduction as these populations approach equilibrium. Rat dispersal between islands Bayesian estimates of migration rates were negligible among most islands, with two notable exceptions in River Harbour (Table 2). Estimating migration rates using 12 locations did not change the estimates of migration compared to estimating migration using 14 locations (the 95 % credible intervals associated with these migration estimates overlapped, Table 2). Therefore, we will discuss the output from the analysis of migration rates using 14 locations. Almost one quarter (22.2 %) of the individuals on Kent Island Knob were estimated to be migrants from Kent Island. Additionally, 18.2 % of the individuals on Middle Island were estimated to be migrants from Rabbit Island. The pairwise Euclidian distances between these islands are 290 and 830 m, respectively. These results indicate that rats are capable of moving distances of at least 830 m between islands in the Falkland Islands. However, we did not find significant reciprocal migration rates in these pairwise comparisons (i.e., migration was not significant from Kent Island Knob to Kent Island or from Middle Island to Rabbit Island). We also did not find significant migration rates between two islands that are separated by only 330 m (Big Arch Island and Sand Bay Island). These results indicate that rats are capable of migration up to distances of at least 830 m, yet there are island pairs where they do not migrate shorter distances of 290 or 330 m. The large residual variation in the patterns of rat migration indicates that distance alone cannot explain the migration frequencies observed. Other factors, including currents, can also determine the exchange of migrants among islands (Quinteiro et al. 2007). Nevertheless, the observation that migration rates were high for some instances up to 830 m, but negligible for other instances indicates that rats might not always move between islands
123
Conserv Genet
that are within their maximum swimming range. It is also possible that rats swim these distances, but they are unsuccessful at breeding with local rats once they reach a new island, so that gene flow does not occur and migration goes undetected (Ostfeld 1990). The unexplained variation in the relationship between migration rates and distance between islands complicates our ability to predict the maximum distance that rats can move between islands.
Island than in the inferior habitat on Middle Island, and migrants move from Rabbit Island to Middle Island. Kent Island (32 ha) is much larger than Kent Island Knob (4 ha; Fig. 1) and it is likely that the small habitat on Kent Island Knob limited the population size of rats, while Kent Island was able to sustain a large population size. Therefore, we hypothesize that the rat population on Kent Island serves as a source for the sink population on Kent Island Knob.
Source-sink relationship
Rat dispersal across land
We hypothesize that the migration patterns in River Harbour represent ‘‘source-sink’’ and ‘‘mainland-island’’ systems. In a source-sink system, ‘‘sources’’ are large populations that exist in patches of good habitat that supply migrants to ‘‘sinks,’’ which are small, unstable populations that exist in inferior habitats (Pulliam 1988; Harrison 1991). Similarly, in a mainland-island system, the ‘‘mainland’’ is a larger patch with a larger population that supplies migrants to an ‘‘island,’’ a smaller patch with a smaller population (MacArthur and Wilson 1967; Morrison 1998). The difference between these two population structures is that in a source-sink system, patches differ in habitat quality; while in a mainland-island system, patches differ in size of suitable habitat (Fronhoffer et al. 2012). Because of the similarities between these two types of spatially structured populations, we will refer to islands with larger populations that supply migrants as ‘‘sources,’’ and islands with smaller population sizes that receive migrants as ‘‘sinks’’ (Heard et al. 2012). We observed low population sizes on both Middle Island and Kent Island Knob, as indicated by less rat sign (field observation) and low capture rates (Table 1) despite greater sampling effort compared to all other locations. We also observed significant levels of migration from Rabbit Island to Middle Island and from Kent Island to Kent Island Knob (Table 2). We hypothesize that Rabbit Island and Kent Island acted as sources to Middle Island and Kent Island Knob. Rabbit Island appeared to have better habitat than Middle Island, as indicated by vegetation cover. On South Atlantic Islands, Norway rats are found at the highest population sizes in dense stands of tussac grass (Poa flabellata; Pye and Bonner 1980), as rats tend to favor areas with dense cover (Whisson et al. 2007). Tussac grass is the most dominant vegetation in the Falklands, and it is the main plant providing shelter for Falkland Island rats (Woods and Woods 2006). A relatively large portion (25 %) of Rabbit Island was covered in tussac grass, while Middle Island did not have any tussac grass (0 % cover) on the island (Poncet and Passfield 2012), likely due to overgrazing by sheep, and the island was devoid of dense vegetation cover (field observations). Therefore, rat populations might be higher in the superior habitat on Rabbit
Rats appear to travel long distances across land on Bleaker Island, as 24.7 % of the individuals at the north end of the island were migrants from the center and 24.9 % of individuals at the south end of the island were migrants from the center. The shortest across-land distances from Bleaker Island Center to Bleaker Island North and South are 8.7 and 11.8 km, respectively (Table 2). These significant migration rates were unidirectional, implying that rats are less likely to travel from the island peripheries to the center. While there are few studies in the literature of Norway rat migration across land (Feng and Himsworth 2014), there is one notable exception where population geneticists estimated dispersal of Norway rats in an urban environment (Gardner-Santana et al. 2009). In their study, GardnerSantana et al. (2009) sampled 277 rats from 11 locations. They found that most ([95 %) individuals were genetically assigned to populations where they were captured, but one individual was assigned to a population 11.2 km away from capture, indicating that long distance dispersal is rare, but possible. We found relatively high levels of recent migration among populations separated by 11.8 km, indicating that rats in the Falklands frequently move even larger distances across landmasses.
123
Developing eradication strategies To identify islands that are good candidates for eradication, a variety of criteria must be considered and population genetic structure is only one. Here we use several criteria to inform future conservation planning. First, we recommend that conservation managers in the Falkland Islands identify a threshold distance for planning future eradication attempts. If an island is farther than this threshold distance from other rat-infested islands and humans infrequently visit it, an eradication attempt may be viable. If islands are closer than this threshold distance, islands in this group are considered to be part of an ‘‘eradication unit,’’ and should be eradicated simultaneously to reduce the risk of reinvasion (Robertson and Gemmel 2004). Our results do not give us the power to provide an exact estimate of this threshold distance. However, when coupled with a previous estimate of the probability of occurrence of rats on islands
Conserv Genet
as a function of distance from a rat source, we can provide a range of distances across which rats move among islands. Tabak et al. (submitted) used an incidence function model to estimate a maximum swim distance of approximately 1,000 m for rats in the Falkland Islands, and in the present study we found rats to be capable of moving among islands (presumably by swimming) 830 m. These two estimates allow us to infer a threshold distance for eradication units of between 830 and 1,000 m between islands. We find the congruence between 830 m and 1,000 m evidence of the robustness of our results and confirmation of the applicability of these results to conservation efforts. It appears that an incidence function analysis and a genetic one give roughly similar results. We further recommend that prior to future eradication attempts, rats on candidate islands and nearby islands should be sampled for genetic analyses. If migration rates from all nearby rat-infested islands to the candidate island are negligible and the island is farther than the threshold distance to the nearest rat-infested island, the island should be considered a good candidate for eradication. If migration rates from all nearby rat-infested islands to the candidate island are negligible and the island is less than the threshold distance to the nearest potential rat source, the case should be investigated further. In such cases, we recommend using genetic data to generate testable hypotheses about why rats tend not to move among the islands (e.g., Is there a barrier to migration that causes migration rates to be low?). These hypotheses should be tested to determine if there is good reason to think that rats will not migrate between the islands. The hypothesis should be well supported, as it is potentially risky and wasteful to eradicate rats from an island in an eradication unit without simultaneously eradicating other islands in that unit. For example, assume that we set the threshold distance at 1,000 m between islands. Natural Arch Island is a potential candidate for eradication because it is separated by greater than the threshold distance from the nearest ratinfested island. We sampled rats on all nearby rat-infested islands and found no significant migration rates from any of these islands to Natural Arch Island. Therefore, we can argue that this island is a good candidate for eradication. Considering Sand Bay Island for eradication is more complicated. This island is only 330 m from the nearest rat-infested island, which is well below the threshold distance. However, migration rates from all nearby rat-infested islands are negligible. This is a case that should be investigated further. Perhaps there is a barrier to dispersal between Sandy Bay Island and Big Arch Island, the nearest rat-infested island. It is possible that a strong tidal current separates these islands such that rats cannot move between these islands. This would be a case where the conservation
strategy could be informed by a coupled seascape genetic analysis (e.g., Galindo et al. 2006). Recolonization of islands by Norway rats following successful eradication has been documented for other archipelagoes as well (Russell et al. 2008b; Harris et al. 2012). Conservation managers in these archipelagoes can benefit from an estimate of the distance that Norway rats can move across open water. Our estimate of 830–1,000 m is from islands surrounded by a relatively cold sea (5–8 °C; Waluda et al. 1999; Agnew et al. 2000). The time that rats can survive in water appears to increase with water temperature (Le Blanc 1958). Consequently, Norway rats can potentially swim greater distances in warmer archipelagoes. Nevertheless, we think that our finding that rats can swim at least 830–1,000 m can be useful as a guide to designing eradication studies in other archipelagoes. Acknowledgments This project was partially funded by the Shackleton Scholarship Fund, Antarctic Research Trust, Wyoming Biodiversity Institute, and NSF Grant #0841298. We wish to thank Nick Rendell, Environmental Planning Officer for the Falkland Islands Government, for his assistance with permits for this research. We also thank Tony and Susan Hirtle, Mike and Phyll Rendell, and Steven Dickson for permission to access sampling locations, and to Derek Brown for his advice and insights. The comments of Liz Mandeville and two anonymous reviewers benefitted previous versions of this manuscript.
References Abdelkrim J, Pascal M, Calmet C, Samadi S (2005) Importance of assessing population genetic structure before eradication of invasive species: examples from insular Norway rat populations. Conserv Biol 19:1509–1518 Abdelkrim J, Byrom AE, Gemmel NJ (2010) Fine-scale genetic structure of mainland invasive Rattus rattus populations: implications for restoration of forested conservation areas in New Zealand. Conserv Genetics 11:1953–1964 Agnew DJ, Hill S, Beddington JR (2000) Predicting the recruitment strength of an annual squid stock: Loligo gahi around the Falkland Islands. Can J Fish Aquat Sci 57:2479–2497 Atkinson IAE (1985) The spread of commensal species Rattus to oceanic islands and their effects on island Avifaunas. In: Moors PJ (ed) Conservation of island birds. ICBP Technical Publication, Cambridge, pp 35–81 Brown D (2008) Criteria for prioritising rat eradications from islands in the Falklands. Rat Eradication Workshop, R Soc for the Prot of Birds (RSPB). Stanley, Falkland Islands, pp 1–22 Courchamp F, Chapuis J-L, Pascal M (2003) Mammal invaders on islands: impact, control, and control impact. Biol Rev 78:347–383 Crow JF, Aoki K (1984) Group selection for a polygenic behavioral trait: estimating the degree of population subdivision. Proc Natl Acad Sci 81:6073–6077 Dagg AI, Windsor DE (1972) Swimming in northern terrestrial mammals. Can J Zool 50:117–130 Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and
123
Conserv Genet implementing the Evanno method. Con Genetics Resour 4:359–361 Evanno G, Regnault S, Goudet J (2005) Detecting the number of clusters of individuals using the software structure: a simulation study. Mol Ecol 51:672–681 Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol Ecol Resour 10:564–567 Falkland Islands Government (2012) The Falkland Islands Census 2012 Falush D, Stephens M, Pritchard JK (2003) Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164:1567–1587 Feng AYT, Himsworth CG (2014) The secret life of the city rat: a review of the ecology of urban Norway and black rats (Rattus norvegicus and Rattus rattus). Urban Ecosyst 17:149–162 Fewster RM, Miller SD, Ritchie J (2011) DNA profiling—a management tool for rat eradication. In: Veitch CR, Clout MN, Towns DR (eds) Island Invasives: eradication and management. IUCN, Gland, pp 426–431 Fronhoffer EA, Kubisch A, Hilker FM, Hovestat T, Poethke HJ (2012) Why are metapopulations so rare? Ecology 93:1967–1978 Galindo HM, Olson DB, Palumbi SR (2006) Seascape genetics: a coupled oceanographic-genetic model predicts population structure of caribbean corals. Curr Biol 16:1622–1626 Gardner-Santana LC, Norris DE, Fornadel CM, Hinson ER, Klein SL, Glass GE (2009) Commensal ecology, urban landscapes, and their influence on the genetic characteristics of city-dwelling Norway rats (Rattus norvegicus). Mol Ecol 18:2766–2778 Genovesi P (2011) Are we turning the tide? Eradications in times of crisis: how the global community is responding to biological invasions. In: Veitch CR, Clout MN, Towns DR (eds) Island invasives: eradication and management. IUCN, Gland, pp 5–8 Hall JR, Woods RW, MdL Brooke, Hilton GM (2002) Factors affecting the distribution of landbirds on the Falkland Islands. Bird Conserv Int 12:151–167 Harris DB, Gregory SD, Bull LS, Courchamp F (2012) Island prioritization for invasive rodent eradications with an emphasis on reinvasion risk. Biol Invasions 14:1251–1263 Harrison S (1991) Local extinction in a metapopulation context: an empirical evaluation. Biol J Linn Soc 42:73–88 Heard GW, Scroggie MP, Malone BS (2012) Classical metapopulation theory as a useful paradigm for the conservation of an endangered amphibian. Biol Conserv 148:156–166 Hilton GM, Cuthbert RJ (2010) The catastrophic impact of invasive mammalian predators on birds of the UK Overseas Territories. Ibis 152:443–458 Holm S (1979) A simple sequentially rejective multiple test procedure. Scand J Stat 6:65–70 Hutchinson DW, Templeton AR (1999) Correlation of pairwise genetic and geographic distance measures: inferring the relative influences of gene flow and drift on the distribution of genetic variability. Evolution 53:1898–1914 Jombart T, Ahmed I (2011) adegenet 1.3-1: new tools for the analysis of genome-wide SNP data. Bioinformatics 27:3070–3071 Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genet 11:1–15 Jones HP (2010) Prognosis for ecosystem recovery following rodent eradication and seabird restoration in an island archipelago. Ecol Appl 20:1204–1216 Keitt B, Campbell K, Saunders A, Clout M, Wang Y, Heinz R, Newton K, Tershy B (2011) The global islands invasive vertebrate eradication database: a tool to improve and facilitate restoration of island ecosystems. In: Veitch CR, Clout MN,
123
Towns DR (eds) Island invasives: eradication and management. Gland, IUNC, pp 74–77 Kunieda T, Kobayashi E, Tachibana M, Ikadai H, Imamichi T (1992) Polymorphic microsatellite loci of the rat (Rattus norvegicus). Mamm Genome 3:564–567 Le Blanc JA (1958) Effect of chlorpromazine on swimming time of rats at different temperatures. Proc Soc Exp Biol Med 98:648–650 MacArthur RH, Wilson EO (1967) The Theory of Island Biogeography. Princeton University Press, Princeton McCauley DE (1993) Genetic consequences of extinction and recolonization in fragmented habitats. In: Kareiva PM, Kingsolver JG, Huey RB (eds) Biotic interactions and global change. Sinauer Associates Inc., Sunderland, pp 217–233 Meirmans PG, Hedrick PW (2011) Assessing population structure: FST and related measures. Mol Ecol Resour 11:5–18 Miller N, Estoup A, Toepfer S, Bourguet D, Lapchin L, Derridj S, Kim KS, Reynaud P, Furlan L, Guillemaud T (2005) Multiple transatlantic introductions of the western corn rootworm. Science 310:992 Morrison LW (1998) The spatiotemporal dynamics of insular ant metapopulations. Ecology 79:1135–1146 Oppel S, Beaven BM, Bolton M, Vickery J, Bodey TW (2011) Eradication of invasive mammals on islands inhabited by humans and domestic animals. Conserv Biol 25:232–240 Ostfeld R (1990) The ecology of territoriality in small mammals. Trends Ecol Evol 5:411–415 Pin˜a DM, Cifuentes GG (2004) Las Aves de Chile: Nueva guia de campo. Ediciones del Naturalista Poncet S, Passfield K (2012) Island visit reports for 11 tussac islands in River Harbour and Jersey Harbour. Beaver Island LandCare report to the Antarctic Research Trust Poncet S, Poncet L, Poncet D, Christie D, Dockrill C, Brown D (2011) Introduced mammal eradications in the Falkland Islands and South Georgia. In: Veitch CR, Clout MN, Towns DR (eds) Island invasives: eradication and management. IUCN, Gland, Switzerland, pp 332–336 Pritchard J, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155:945–959 Pulliam HR (1988) Sources, sinks, and population regulation. Am Nat 132:652–661 Pye T, Bonner WN (1980) Feral brown rats, Rattus norvegicus, in South Georgia (South Atlantic Ocean). J Zool 192:237–255 Quinteiro J, Rodrı´guez-Castro J, Rey-Me´ndez M (2007) Population genetic structure of the stalked barnacle Pollicipes pollicipes (Gmelin, 1789) in the northeastern Atlantic: influence of coastal currents and mesoscale hydrographic structures. Mar Biol 153:47–60 Raymond M, Rousset F (1995) GENEPOP (version 1.2): population genetics software for exact tests and ecumenicism. J Hered 86:248–249 Rice WR (1989) Analyzing tables of statistical tests. Evolution 43:223–225 Robertson BC, Gemmel NJ (2004) Defining eradication units to control invasive pests. J Appl Ecol 41:1042–1048 Russell JC, Towns DR, Clout MN (2008a) Review of rat invasion biology: implications for island biosecurity. Sci Conserv 286:1–54 Russell JC, Beaven BM, MacKay JWB, Towns DR, Clout MN (2008b) Testing island biosecurity systems for invasive rats. Wildl Res 35:215–221 Strange IJ (1992) A field guide to the wildlife of the Falkland Islands and South Georgia. HarperCollins, London Tabak MA, Poncet S, Passfield K, Martinez del Rio C (2014) Invasive species and land bird diversity on remote South Atlantic Islands. Biol Invasions 16:341–352
Conserv Genet Towns DR (2009) Eradications as reverse invasions: lessons from Pacific rat (Rattus exulans) removals on New Zealand islands. Biol Invasions 11:1719–1733 Towns DR (2011) Eradications of vertebrate pests from islands around New Zealand: what have we delivered and what have we learned? In: Veitch CR, Clout MN, Towns DR (eds) Island invasives: eradication and management. IUCN, Gland, pp 364–371 Vitousek PM, D’Antonio CM, Loope LL, Rejmanek M, Westbrooks R (1997) Introduced species: a significant component of humancaused global change. New Zealand J Ecol 21:1–16 Waluda CM, Trathan PN, Rodhouse PG (1999) Influence of oceanographic variability on recruitment in the Illex argentinus (Cephalopoda: Ommastrephidae) fishery in the South Atlantic. Mar Ecol Prog Ser 183:159–167
Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evolution 38:1358–1370 Whisson DA, Quinn JH, Collins KC (2007) Home range and movements of roof rats (Rattus rattus) in an old-growth riparian forest, California. J Mammal 88:589–594 Whitlock MC, McCauley DE (1999) Indirect measures of gene flow and migration: Fst not eual to 1/(4Nm ? 1). Heredity 82:117–125 Wilson GA, Rannala B (2003) Bayesian inference of recent migration rates using multilocus genotypes. Genetics 163:1177–1191 Woods RW, Woods A (2006) Birds and Mammals of the Falkland Islands. WildGuides, Hampshire
123
