Introduction

Invasive species are increasingly spread around the world by human-mediated transport, but what makes a successful invader is not clear (Hulme 2009; van Kleunen et al., 2011; Wardle et al., 2011). A common approach to trying to identify indicators of invasive success is to compare closely related native and invasive species (Dunn et al., 2009), or native and invasive populations of the same species (Lamarque et al., 2015). In some cases, invasive and native populations or species do not differ significantly, perhaps because some species have preadapted traits that predispose them to invasion success, even in their native range (e.g., Elst et al., 2016; Schlaepfer et al., 2010), or because plasticity is key to invasion (Lamarque et al., 2015). However, drift may play a role in differentiation of invasive populations, since initial invasions are expected to involve founder effects (Abdelkrim et al., 2005; Barrett et al., 1996). Selection during the invasion process is also expected (Colautti et al., 2017; Colautti and Lau 2015), and may result in invasive populations that are phenotypically different from native populations (Sakai et al., 2001). The morphological, life history, and behavioural traits that aid invasive establishment and spread past the point of introduction in new environments may differ from those that are adaptive in the native range, leading to phenotypic differences from the native source population (Chapple et al., 2012; Le Gros et al., 2016).

The invasion process can be characterized by a series of stages, including transport, introduction, establishment, and spread (Blackburn et al., 2011). Selective filters at different stages of the invasion process cause phenotypic differences between native and invasive populations of the same species, and different combinations of traits may be adaptive at different stages (Sakai et al., 2001). For example, the transport stage may select for traits that increase survival, such as selection for increased tolerance to heat stress (e.g., mussels, Lenz et al., 2018), or behavioural traits related to higher activity in novel conditions (e.g., songbirds, Mueller et al., 2017). A selective filter at the introduction stage could favour tolerance to new predators, prey, or pathogens and survival in different abiotic conditions than the native habitat. For example, in invasive bunchgrass, selection for drought tolerance occurred after introduction, but before range expansion (Marchini et al., 2018). During invasive range expansion, traits that facilitate movement, such as increased dispersal ability, may be favoured. Such selective filtering through multiple stages of the invasion process can result in invasive populations with a mix of traits adaptive at different stages (Carrete et al., 2012; Chapple et al., 2012). Apart from selection based on the invasion pathway, once a population has established and spread, the new environment is often different and may be unpredictable compared to the native environment. Thus, as invasive populations become more well-established, the signature of selection during transport and spread may become less prominent as selection increases for local adaptation or plasticity. In this study, we focus on morphology, life history, and behavioural traits that may affect invasive success at different stages.

Morphological and life history traits that increase survival in transport, establishment, and spread may differ from those in the native range. For example, compared to native populations of the same species, Asteraceae plants grow larger in the invasive range, which may facilitate movement and dispersal, and also aid in outcompeting native species (Jakobs et al., 2004). However, there is often a tradeoff between advantageous large size and development time (Gadgil and Bossert 1970; Nijhout 2015), which may depend on environmental conditions, both biotic (competitors and predators present) and abiotic (temperature and seasonality). For example, invasive individuals may be smaller if investment in high reproductive output or rapid development decreases offspring size (Thorlacius and Brodin 2018). This trade-off may be adaptive if new conditions are unfavourable, or population density is low (Brownscombe and Fox 2012). Thus, life history traits such as high fecundity and fast development time, which may be important for successful establishment and population growth after initial colonization (fish: Amundsen et al., 2012; reptiles and amphibians: Allen et al., 2017; mammals: Capellini et al., 2015) or at the invasion front (Burton et al., 2010), are linked in complex ways with body size, mediated by environmental factors.

Similarly, different sets of behaviours could be adaptive at different stages of invasion, resulting in opposing selection pressures over time. For example, exploratory behaviour and high levels of dispersal may be favoured during transport to a new habitat (Pettit et al., 2016) and spread. The result can be shifts in allele frequencies at loci governing higher activity levels in response to novelty (Mueller et al., 2017). In addition, natal dispersal distance or propensity may determine the likelihood of colonizing new habitats, and is often positively correlated with activity level, exploratory behaviour (Korsten et al., 2013), or boldness and aggression (Rehage and Sih 2004). These correlated behaviours, called syndromes, often vary within populations, and may show patterns of spatial sorting, with individuals with the highest dispersal propensity (and highest activity levels, boldness, or exploratory behaviours) found closest to the edge of the invasion front (Atwell et al., 2012; Liebl and Martin 2012). This process accelerates invasion (Ochocki and Miller 2017) by supporting population expansion and likelihood of exploiting novel conditions. As a result, more aggressive, bolder, more exploratory individuals may be common at the invasion front, but less common in subsequent waves of individuals establishing in the new habitat, when population densities are higher and local adaptation is important (Duckworth and Badyaev 2007; Sih et al., 2012). Thus, in addition to differences between native populations and those at the early stages of invasion, long-term differentiation between established invasive populations compared to the native source population is expected (Michaelides et al., 2018). Although there may be evolutionary lags (Crooks and Soulé 1999; Pyšek and Jarošík 2005), with sufficient time, introduced populations are expected to adapt to the novel environment (Caut et al., 2008; Colautti et al., 2017), which may lead to new trade-offs relative to the native population or the early stages of invasion (Burton et al., 2010).

We assess traits leading to invasion success in the Australian redback spider, Latrodectus hasselti, native to Australia and now introduced and spread in New Zealand (detected in 1984 Forster 1984) and 11 years later, in Japan (detected in 1995 Ori et al., 1996). We predict that, in comparison to the native Australian population, invasive populations from Japan and New Zealand will be larger, more fecund, more exploratory, bolder, more dispersive, develop more rapidly and have higher rates of sibling cannibalism (a behaviour that may buffer against low food availability). In addition, we expect the more recently-established population (Japan) will score higher on these traits than the longer-established invasive population (New Zealand).

Methods

Latrodectus hasselti has been introduced to Japan and New Zealand, as well as India, Iran, and Papua New Guinea (Chrysanthus 1975; Hippargi et al., 2012; Shahi et al., 2011), likely through cargo ships carrying wood, steel, or produce (Vink et al., 2011). Latrodectus spiders and fertilized egg sacs have been intercepted in international shipments of produce and second-hand cars to Europe (Blick et al., 2004; Van Keer, 2011). Adult females build their webs in refuges in corners of wood and metal structures, so can remain undetected during transport. This iteroparous, high-fecundity species may have 1 to 3 overlapping generations per year, which may facilitate establishment. Latrodectus hasselti develop and hatch inside egg sacs, with clutches of up to 300. Spiderlings emerge at the 2nd instar and stay clustered on their natal web, during which time they may cannibalize siblings (Modanu et al., 2014). Invasive success of these spiders may be facilitated by sibling cannibalism, which may increase survival of cannibals when prey are scarce (Antonialli-Junior and Guimarães 2014). Spiderlings disperse after about one week by ballooning (Forster 1992) and rappelling (Forster 1995), and high dispersal propensity may increase invasive spread. Male and female widow spiders have dramatically different life histories and sizes; males, which are 1–2% of a female’s mass, wander in search of females, while females are sedentary (Andrade 1996). Propensity for sibling cannibalism and dispersal both vary across family lines in Latrodectus spiders (Johnson et al., 2010, 2015; Modanu et al., 2014). These spiders depend on webs for prey capture and defense from predators, so after arriving in a new environment, spiders must rapidly build webs to survive. Web-construction activity may co-vary with boldness in spiders, as found in Cyrtophora citricola, an invasive orbweaver spider (Chuang and Riechert 2021). Like other Theridiidae, widow spiders typically rest in a protected area of the web during the day. Foraging requires movement across the web, but spiders may retreat or ‘huddle’ (pull in their legs and cease movement) if they detect a threat (e.g., predator vibration).

Laboratory conditions and populations

All spiders used in these experiments were reared in the lab under 12:12 L:D and at 25 °C conditions. Juvenile spiders and males were fed flies (Drosophila spp.) and watered twice per week as juveniles, and female spiders were fed crickets (Gryllodes sigillatus or Acheta domesticus) once per week as adults.

Mated L. hasselti females (P, parental generation) were collected from North Epping, NSW, Australia (33.765°S, 151.087°E), Osaka, Japan (34.775°N, 135.490°E) and Fukuoka, Japan (33.666°N, 130.417°E) in October 2015, and from Alexandra, New Zealand (45.235°S, 169.345°E) in February 2019. Additional preserved adult females from New Zealand, used for morphological measurements, were originally collected in Bannockburn (45.085°S, 169.171°E), Cairnmuir (45.079°S, 169.1882°E), and Alexandra (45.254°S, 169.402°E). The Japan population from Osaka was near the site of the initial introduction in 1995, while the population from Fukuoka was likely established after 2000 (Nihei et al., 2003). The habitat where the New Zealand population was collected is low density and not inhabited by humans, whereas the Japanese population is high density and found near ports and cities (Forster 1984; Nihei et al., 2003; Vink et al., 2011). F1 offspring were derived from eggs produced by P females, which were reared to adulthood, then mated in the laboratory using an outbreeding protocol. Experimental individuals for measuring behaviours (boldness, exploration, and dispersal) for the Australian and Japanese populations were from the F2 and F3 (outbred) generations from P females (Supplemental Material 1). Constrained by the timing of field collecting, and the intervals needed for behavioural trials, the New Zealand juveniles tested were from the F1 generation from P females.

Across all trials described below, spiders were only assayed once across different size, life history, or behavioural assays (e.g., no individuals were reused across trials to avoid habituation or stress-related effects).

Size

Spiders were photographed using a Nikon DXM 1200 digital camera mounted on a Zeiss Stemi 2000-C dissecting microscope for later image analysis. The identification of specimens as Latrodectus hasselti was confirmed using morphological characters including abdominal hairs and genitalia (Vink et al., 2009). We measured body size at multiple stages of development to assess how differences varied across sex and lifespan. We photographed a sample of five spiders from each egg sac (with each sac representing one family line) at each developmental stage. Immediately after emerging, second instar spiderlings were photographed (Australia: n = 11 family lines, Japan: n = 10 family lines, New Zealand: n = 10 family lines). We measured cephalothorax width at the second-instar stage (Modanu et al., 2014) as spiderlings were photographed alive, therefore measuring patella-tibia length was not possible. We measured cephalothorax width in fourth instar (juvenile) spiders (Australia n = 21, Japan n = 21), which were also measured alive, thus patella-tibia length measurement was not reliable. We also measured patella-tibia length of penultimate instar (the stage immediately before adulthood) males (n = 34 Australia, n = 26 Japan, n = 21 New Zealand), adult males (n = 23 Australia, n = 17 Japan, n = 17 New Zealand), and adult females (n = 48 Australia, n = 32 Japan, n = 26 New Zealand). The adult females measured from New Zealand were not laboratory reared, but were rather collected as adults, which may present a confound to this part of the analysis.

Life history

Life history traits (reproductive output, generation time) were measured from F3 spiders from Japan and Australia, and P spiders from New Zealand that had been held under laboratory conditions for at least one month. Reproductive output was assessed in two ways: egg sac production rate and fecundity per egg sac, which are positively correlated in other spider species (linyphiid spiders: Thorbek et al., 2004). Mated adult females lay egg sacs throughout their lifetimes. Fecundity (total number of eggs per egg sac) was measured for one egg sac produced by each female (n = 40 Australia, n = 40 Japan, and n = 14 New Zealand females from distinct family lines). For a subset of females, the number of egg sacs produced over a two-month period was recorded, and then averaged over that period to estimate the egg sac production rate (n = 14 Australia, n = 17 Japan, n = 14 New Zealand females). We measured juvenile development time by recording the time from emergence from the egg sac until the fourth instar (Australia n = 95, Japan n = 50). For males, we measured development to the penultimate instar, assessed by the presence of enlarged pedipalps (Bhatnagar and Rempel 1962).

Behaviour

We measured behavioural traits predicted to be relevant to survival and persistence at multiple stages of the invasion process: sibling cannibalism, exploration, boldness, and dispersal Table 1.

Table 1 Morphological, life history, and behavioural traits related to Latrodectus hasselti invasive potential and direction predicted in the difference between invasive and native populations
Full size table

Sibling cannibalism was measured in spiderlings emerged from egg sacs produced by females mated with one male in the laboratory. Cannibalism was measured from when spiderlings emerged from the egg sac to eight days after emergence (n = 15 Australian egg sacs, n = 24 Japanese egg sacs). This eight-day period was chosen as the approximate amount of time on the natal web in the field before juvenile dispersal (MAM, pers. obs.). Immediately after hatching, 30 spiderlings from each egg sac were transferred to a plastic cage (8 × 8 × 12 cm. clear plastic container, Amac Plastics, Ltd.) and observed for eight days. Food and water were withheld during this period, when spiderlings typically do not capture prey in nature. The number of spiderlings cannibalized (wrapped in silk) was counted every other day, as well as death by natural causes (legs curled and not wrapped in silk).

Exploratory behaviour (n = 49 Australian, n = 37 Japanese juveniles) and boldness behaviours (n = 49 Australian, n = 39 Japanese juveniles) were tested in fourth-instar juvenile spiderlings in the Australian and Japanese populations. Each spider was placed in a novel, empty plastic cage (5 × 5 × 7 cm. clear plastic container, Amac Plastics, Ltd.) and tested either for exploratory behaviour or boldness during three three-minute (180 s) trials. For exploratory behaviour, we recorded the time until the spider started to move and release silk to initiate web-building. Movement in a novel environment is a typical way of measuring exploration across taxa (Mead 1960). Spiders that were placed into a novel container and did not move during the trial period were categorized as non-responders. For boldness assays, spiders that had started moving and laying silk were puffed with air from an 88.7 ml air bulb from 5 cm away (Riechert and Hedrick 1993) to simulate a predator threat, after which the spider huddled and stopped moving. Boldness was measured by the time it took for the spider to start moving again after huddling. This response was considered a measure of boldness because it indicated an individual’s propensity for movement under simulated risk of predation. Spiders that were puffed with air and never resumed movement over the trial periods were categorized as non-responders. We did not measure boldness or exploratory behaviour in New Zealand spiderlings at the fourth-instar for comparison because of laboratory rearing constraints.

Dispersal was measured with F3 spiderlings from Australia and Japan (n = 12 family lines per population, five spiderlings per egg sac tested). For the New Zealand population, spiderlings were from egg sacs produced by field-collected female after they were held in constant laboratory conditions for at least one month (n = 14 family lines, five spiderlings per egg sac). Dispersal was tested seven days after hatching, with experimental spiderlings kept communally with their siblings, but food and water withheld. The dispersal arena was a plexiglass box (50 × 34 × 35 cm) laid on its long axis, with one short wall (the endpoint) replaced with a fibreglass screen and the opposite wall (the start point) including a screen-covered hole through which air from a fan blew into the arena Fig. 1. Just inside the start point wall, two wooden dowels (16 cm in length) were joined together with tape and placed upright, 11 cm from the inner surface, with the tip of the dowels centered on the screen hole. The fan created an airspeed of 1 m/s measured at the dowel, and the temperature was held at a constant at 23 °C during trials, as moderate wind speeds and temperatures are optimal for aerial spider dispersal (Simonneau et al., 2016). For each trial, a spiderling was placed inside the arena at the tip of a wooden dowel. The spiderling was allowed to settle for five minutes. The fan was then turned on and the spiderling was observed for 10 min, or until it showed dispersal behaviours (rappelling or ballooning (Forster 1995). Both dispersal behaviours began with the spider attaching a silk dragline to the top of the dowel. A ballooning spider moved into a ‘tiptoe’ posture with posterior end of the abdomen elevated and released additional silk which was caught by the air (Forster 1995). Rappelling occurred when the spider released a dragline of silk and using the released silk, moved from the top of the dowel to the mesh 0.29 m away. During the trial, spiderlings often rappelled for short distances (a few centimeters), but this behaviour was not counted as successful dispersal. Dispersal was measured as a binary variable (disperse or did not disperse) and as time to dispersal for those that did disperse (time to balloon or rappel from the start of the trial).

Fig. 1
figure 1

Arena used for trials to measure dispersal propensity and time to disperse in Latrodectus hasselti spiderlings

Full size image

Statistical analysis

We used the ‘psych’ package in R to calculate summary statistics for all of the variables by population (Revelle 2017). We investigated mean differences and population-level variability in adult size, development time, fecundity, and egg sac production rate of the native and invasive populations using one-way ANOVA tests followed by Tukey post-hoc tests for comparisons among Australia, Japanese, and New Zealand populations, and two-tailed t-tests for comparisons between Australian and Japanese populations. We used Levene’s test to assess the homogeneity of variance. In web building and boldness response times, assumptions of normality were not met. To compare spiderling size among populations, we created a generalized linear mixed model with cephalothorax width as the response variable, egg sac ID as a random factor, and population as a fixed factor.

To compare cannibalism between the populations from Australia and Japan, we used a generalized linear model with the number of spiderlings cannibalized as the response variable (Poisson distribution), population by day as an interaction, and female ID as a random factor using the ‘lme4’ package in R (Bates et al., 2015). To compare boldness and exploratory behaviour between Australian and Japanese populations, we used chi-squared tests for each behaviour to assess differences in response rate (resumed movement or did not resume movement during any of the three-minute trials). To compare differences among those spiders that responded during the trials, we conducted Kruskal–Wallis tests. To compare dispersal of native and invasive spider populations, we first made a generalized linear model with dispersal as the (binary) response variable and population as a predictor and family line as a random factor, with a binomial link function. To test for population differences in dispersal time, we included only those spiders that dispersed within ten minutes and used an ANOVA with dispersal time as a response variable and population as a predictor. All statistical tests were conducted in R version 3.3.2.

Results

Size

Invasive second-instar spiderlings from Japan were similar in size to the native spiderlings from Australia (Fig. 2a, means ± SE, Japan: n = 120, 0.50 ± 0.01 mm, Australia: n = 139, 0.49 ± 0.01 mm; GLMM, t = 1.612, p = 0.116). However, the invasive New Zealand spiderlings were smaller than both the Australian and the Japanese populations (mean ± SE, New Zealand: n = 138, 0.38 ± 0.01 mm, GLMM, t = -11.043, p < 0.001). Fourth-instar juvenile invasive spiders from Japan were larger than native spiders from Australia (Fig. 2b, means ± SE, Australia: n = 21, 0.59 ± 0.01 mm, Japan: n = 21, 0.68 ± 0.02 mm, t-test, t = 4.351, p < 0.001). Among adult females, leg length was the largest for invasive spiders from New Zealand compared to the native Australian and invasive Japanese populations (Fig. 2c, means ± SE, Australia: n = 48, 5.43 ± 0.06 mm, Japan: n = 32, 5.55 ± 0.07 mm, New Zealand: n = 26, 6.23 ± 0.10 mm, ANOVA, F = 33.91, p < 0.001; Tukey HSD, New Zealand and Japan/Australia: p < 0.001, Australia and Japan: p = 0.323). Body size is affected by diet, so field caught New Zealand females could differ from lab-reared Japanese and Australian spiders for this reason. However, we note that adult field-collected (P) and lab-reared (F3) Australian females do not differ in leg length (Supplementary Fig. 1).

Fig. 2
figure 2

Boxplots comparing the size of Australian (native), New Zealand (longer-established invasive), and Japanese (newly-established invasive) populations of Latrodectus hasselti as a second-instar newly-emerged spiderlings (cephalothorax width), b fourth-instar juveniles (cephalothorax width), c adult females (patella-tibia length), and d penultimate stage and adult males (patella-tibia length). Different letters above box plots indicate significant differences between locations based on Tukey HSD post-hoc tests

Full size image

For males, invasive males from Japan at the penultimate instar and as adults were larger than native males from Australia, though males from New Zealand were smaller than the Australian and Japanese populations at both life stages (Fig. 2d, means ± SE, penultimate instar males: Australia: n = 34, 2.32 ± 0.04 mm, Japan: n = 26, 2.45 ± 0.05 mm, New Zealand; n = 21, 2.17 ± 0.03 mm, adult males: Australia: n = 23, 2.89 ± 0.05 mm, Japan: n = 17, 3.07 ± 0.09 mm, New Zealand: n = 17, 2.58 ± 0.05 mm, GLM, population p = 0.006).

Life history

The population from Japan was faster to produce egg sacs than the population from Australia (means ± SE, Australia: n = 14, 12 ± 1 days between sacs, Japan: n = 17, 9 ± 1 days between sacs, two-tailed t-test, t = 3.0764, p = 0.005). Spiders from the New Zealand population (field-caught, but held on a standard diet) produced egg sacs less frequently than spiders from the lab-reared Australian and Japanese populations (Fig. 3a, mean ± SE, New Zealand: n = 14, 18 ± 2 days between egg sacs, ANOVA, F = 14.44, p < 0.001). There were no population-level differences in number of eggs per sac (Fig. 3b, means ± SE, Australia: n = 40, 102 ± 10 eggs, Japan: n = 40, 103 ± 8 eggs, New Zealand: n = 14, 122 ± 15 eggs, ANOVA, F = 0.742, p = 0.479). However, spiders from Japan took longer to develop from emergence to fourth instar than native spiders from Australia. Invasive males from the Japan population took longer to develop from emergence to the penultimate stage of development (means ± SE, Australia: n = 95, mean ± SE, 18 ± 1 days, Japan: n = 50, 30 ± 2 days, two-tailed t-test, t = 4.2253, p < 0.001). We did not measure the development time of females to mature to adulthood.

Fig. 3
figure 3

Boxplots of Latrodectus hasselti a Egg sac production rate (number of egg sacs produced over two months divided by number of days) and b fecundity (number of eggs per sac) produced by females from native Australian, invasive New Zealand, and invasive Japanese populations. Different letters above box plots indicate significant differences between locations based on Tukey HSD post-hoc tests

Full size image

Behaviour

Invasive spiders from Japan were more cannibalistic of siblings than spiders from the native population. Invasive spiders from Japan cannibalized siblings at a higher rate over an eight-day period than native spiders from Australia (Fig. 4, means ± SE, n = 24 groups, Japan: 10 ± 2 individuals cannibalized out of 30 individuals after eight days, n = 15 groups, Australia: 5 ± 1 individuals cannibalized out of 30 individuals after eight days, GLMM, population: z = 2.261, p = 0.024, day: z = 6.388, p < 0.001, population*day: z = -1.106, p = 0.269). There was no significant difference in the propensity to explore (91.9% of Australia spiders [n = 49] and 81.1% of Japan spiders [n = 37] explored during the trials, χ2 = 1.328, df = 1, p = 0.249), nor was there a difference in latency to initiate exploration between native spiders from Australia and invasive spiders from Japan of the spiders that responded during the three-minute trial (means ± SE, Australia: n = 45, 35 ± 7 s, Japan: n = 30, 45 ± 7 s, Kruskal–Wallis test, χ2 = 2.325, p = 0.127). In the boldness assay, spiders from the invasive Japan population (33.3% of 39 spiders) were more likely to stay still after a startle than were those from the Australia population (8.1% of 49 spiders) for the entire testing period of 180 s (χ2 = 7.286, df = 1, p = 0.007). Of those spiders that resumed activity within 180 s, there was no difference in the latency to resume movement (boldness) between native and invasive spiders from Japan (means ± SE, Australia: n = 45, 25 ± 7 s, Japan: n = 26, 22 ± 7 s, Kruskal–Wallis test, χ2 = 0.0058, p = 0.939). When comparing the F3 lab populations from Australia and Japan, spiders from the invasive Japan population were more likely to disperse than spiders from the native Australia population (means ± SE, Japan: n = 60, 0.83 ± 0.06, proportion dispersed out of spiderlings assayed, Australia: n = 60, 0.60 ± 0.07, GLMM, z = 2.611, p = 0.009). When including the offspring of field-collected females from New Zealand (n = 70, mean ± SE, 0.78 ± 0.06 dispersed out of spiderlings assayed), invasive spiders from New Zealand and Japan were more likely to disperse than the native Australian population (Fig. 5a, GLMM, Japan to Australia, z = 2.488, p = 0.0128, New Zealand to Australia, z = 2.135, p = 0.033). Among those spiders that dispersed, when limiting the comparison to the Japan and Australia F3 lab populations, spiders from the Japan population tended to disperse faster than spiders from the Australia population (means ± SE, Japan: n = 50, 200 ± 18 s, Australia: n = 36, 263 ± 32 s, GLMM, t = 1.881, p = 0.060). When comparing the lab populations and offspring of field-collected females from New Zealand, those from the New Zealand population dispersed more quickly than spiders from the native Australian population, and tended to disperse more quickly than spiders from the Japanese population (Fig. 5b, means ± SE, New Zealand: n = 56, 132 ± 20 s, ANOVA, F = 7.866, p < 0.001, Tukey HSD, New Zealand and Australia: p < 0.001, Japan and Australia: p = 0.161, New Zealand and Japan: p = 0.069).

Fig. 4
figure 4

Means and standards errors of sibling cannibalism in invasive Japanese and native Australian Latrodectus hasselti egg sacs over an eight-day period immediately after spiderling emergence

Full size image
Fig. 5
figure 5

Boxplots of Latrodectus hasselti spiderlings seven days after emergence a Proportion dispersed within each family line, b time to dispersal of those that dispersed within ten minutes) in native Australian, invasive New Zealand, and invasive Japanese populations. Different letters above box plots indicate significant differences between locations based on Tukey HSD post-hoc tests

Full size image

Discussion

We found differences in size, life history, and behaviour between a native and an invasive population of Latrodectus hasselti, even after rearing under controlled laboratory conditions for two generations (see Table 1 for a comparison of results to predictions). In general, spiders from the invasive population from Japan were larger at several life stages Fig. 2, took longer to develop, were more cannibalistic Fig. 4, less bold, and more dispersive Fig. 5 than a native population from Australia. Including additional trait comparisons from a New Zealand invasive population, we found that offspring of field-collected females from an invasive New Zealand population were more likely and faster to disperse than the native Australia population Fig. 5. These phenotypic differences may aid survival in transport (e.g., larger size, sibling cannibalism) in colonization of a new environment (e.g., high fecundity) and spread through those novel environments (e.g., dispersal propensity). While many of these differences were consistent with our predictions, some were contrary to expectations Table 1. In particular, we found that Japanese and New Zealand size and fecundity differed from the native population, but these differences varied by life stage and were inconsistent. These unpredictable differences could arise from divergent selection pressures in Japan and New Zealand affecting the evolution of traits in those invasive populations. Neutral processes such as population bottlenecks and genetic drift during the invasion process could also explain these differences in phenotypic traits (Johnson et al., 2018). Some of these effects may arise because the New Zealand spiders used in this experiment were P and F1 rather than having been reared through a generation in the lab. However, we found that field-collected and F3 females from Australia were not significantly different in body size (Supplementary Fig. 1), which suggests that lab rearing for several generations does not lead to shifts in size. Overall our data suggest invasive Latrodectus hasselti spiders may show phenotypic shifts compared to a native population.

We found that spiders from the invasive Japanese populations were larger than native spiders at multiple life stages. Similarly, for New Zealand invasive females, adult females were larger than Australia native females. Consistent with other findings in insects and arachnids, large size may offer selective advantages to colonizing new habitats (Hemptinne et al., 2012). Moreover, for spiders, adult female body size is typically closely correlated with fecundity, which would also aid in invasive success. Interestingly, the adult females were the only life history stage at which New Zealand spiders were larger than those from Australia (the juvenile, penultimate instar males, and adult males were smaller, Fig. 2). The longer-established New Zealand invasive population showed increased sexual size dimorphism (smaller males and larger females than the native population), similar to long-established invasive populations of cane toads (Hudson et al., 2016). In Latrodectus, small male size is selected for in the absence of direct competition, which may occur in low density populations (Kasumovic and Andrade 2006). Various types of environmental effects may thus underlie the results with the New Zealand population, particularly for adult females. Nevertheless, for the Japan compared to Australia populations, we compared 2nd and 3rd generation lab-reared individuals, which suggests that there has been an evolutionary shift in body size among invasive Japanese L. hasselti.

Surprisingly, we found that spiders from the invasive Japanese population took longer to develop at multiple life stages than native spiders. This developmental difference could reflect local adaptation to the lack of specialist predators in Japan. Slow development time results in higher risk of predation in other invertebrates (Häggström and Larsson 1995), so species with high predation pressure tradeoff faster development time for smaller body size (Dahl and Peckarsky 2003). In contrast, in Japan, there may be reduced predation pressure, allowing slower-developing L. hasselti higher fitness because of the overall advantages of larger body size (Honěk 1993). Slower development time could also be caused by adaptation to cooler temperatures in the invasive habitat (Li and Jackson 1996). Though spiders from the invasive population took longer to develop, they may overcome this slower development by producing offspring more frequently as adults. We found higher fecundity among Japanese invasives relative to native spiders, in terms of the higher frequency of egg sac production. This high-fecundity phenotype is similar to that found in other invasive invertebrates (Foucaud et al., 2016; Pechenik et al., 2017), particularly in recently-established populations (Masson et al., 2016). An increase in frequency of reproduction is a bet-hedging strategy, which can increase success in unpredictable environments (Sol et al., 2012).

Spiders from the invasive Japan population were more cannibalistic and dispersive, yet less bold than native spiders. Increased rates of sibling cannibalism could be beneficial particularly in low food environments, such as during the transport stage of invasion, most likely via shipping containers (Vink et al., 2011). In contrast, there was no difference in latency to begin exploring and web building between the invasive Japan and native Australia populations. Increased exploratory behaviour may be selected for during the spread stage, but not during the transport stage, where surviving individuals are those that avoid detection (Chapple et al., 2011). Consequently, this assay in a novel environment may predominantly reflect on mobility/exploratory behaviour. The lack of differences in web building is not surprising because selection pressures on latency to initiate web building may not be different in the novel environment. Being without a web is very costly for spiders regardless of whether they are from an invasive population, since the web allows prey capture, communication, and detection of predators.

We found strong evidence for increased dispersal propensity in invasive populations. Both the Japanese and the New Zealand spiderlings were more dispersive than the native Australian spiderlings. We found similar increased dispersal propensity in the invasive Japanese population, which was reared under standard laboratory conditions for three generations like the native Australian population. This pattern in increased dispersal is consistent with predictions about selection during invasive spread past the point of introduction (Alford et al., 2009; Cote et al., 2010). Increased dispersal ability can be particularly advantageous in changing environments (Kokko and López-Sepulcre 2006). Moreover, founder effects could quickly lead to evolutionary shifts in dispersive tendency since the invasive populations could have been initiated by chance dispersal events, whether to the habitat directly, or dispersal onto ships or other vectors (Wilson et al., 2009).

New Zealand spiderlings were lab-produced offspring of field-collected females held on a constant diet, and their behaviour may have been influenced by maternal effects. In spiders, field and lab-reared individuals can differ in prey capture, activity level, and mating behaviour (Carducci and Jakob 2000; Wiggins et al., 2018). The laboratory can act as a selective environment after several generations in captivity, as was the case of the Australia and Japan populations, and can affect morphology, life history, and behaviour. For example, mosquitos (Culex tarsalis) that were field-collected dispersed further than lab-reared flies (Nelson and Milby 1980), so the finding of lower dispersal in the Australia population compared to the New Zealand population could be influenced by laboratory rearing. In addition, lab conditions of a constant diet and unvaried environment conditions may also affect development time, size, and fecundity.

In contrast to dispersal, we found that invasive spiders from Japan were less bold than native spiders from Australia when presented with a simulated predator threat. Under high predation risk, this decrease in boldness would be adaptive. Although habitats in Japan and New Zealand are less likely to have specialist predators such as parasitoid wasps, spiders in Japan are more likely to live in areas with a high human footprint (Mowery, Luo, & Andrade, unpublished) than in Australia. Behaviourally, invasive redback spiders from Japan are more cannibalistic and dispersive, yet less bold, than native Australia spiders, which suggests that there are no underlying constraints that link boldness, cannibalism, and dispersal in this species, in contrast to other studies that found population-level behavioural correlations in boldness and dispersal (Duckworth and Badyaev 2007). Differences between two invasive populations could also be explained by differences in biotic or abiotic conditions. In addition, adaptation to climatic conditions in Japan and New Zealand could explain the size and fecundity differences. The range of mean minimum and maximum temperatures in Osaka, Japan is 1.4 °C to 32.4 °C, and the range of mean minimum and maximum monthly temperatures in Alexandra, New Zealand is  −0.8 °C to 23.1 °C. Both climates are cooler than Sydney, Australia, where the mean monthly minimum is 7.1 °C and the mean monthly maximum is 26.3 °C (Fick and Hijmans 2017).

Future work will investigate whether these traits are common across invasive arthropods, or whether they are specific to species and transport method. The mechanisms of transport, establishment, and spread of invasive spiders (Hänggi and Straub 2016; Nentwig 2015; Nentwig et al., 2017) could be examined along with species and population-level studies of traits leading to invasion success (Houser et al., 2014; Krehenwinkel and Tautz 2013). By investigating how behavioural shifts influence patterns of establishment and spread in spiders, we can better understand what factors make a successful invader. In this study, we made predictions based on how sequential stages of the invasion process may act as selective filters, resulting in invasive populations with different morphology, life history, and behaviour from the native population of origin. We found substantial phenotypic differences between an invasive and native spider population, even after rearing in common conditions for several generations. These phenotypic differences suggest an integrated invasive phenotype shaped by the invasion process may include traits favoured at different stages of invasion and establishment, which lead to differences in colonization success and spread through the new habitat.