Predicting the impact of an invasive seaweed on the fitness of native fauna

Authors

  • Jeffrey T. Wright,

    Corresponding author
    1. Institute for Conservation Biology and School of Biological Sciences, University of Wollongong, Wollongong 2522, Australia; and
    Search for more papers by this author
    • Present address: National Centre for Marine Conservation and Resource Sustainability, Australian Maritime College, University of Tasmania, PO Box 986, Launceston 7250, Australia

  • Paul E. Gribben

    1. Centre for Marine Biofouling and Bio-Innovation and School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney 2052, Australia
    Search for more papers by this author
    • Present address: Department of Environmental Sciences, University of Technology, Sydney, PO Box 123, 2007, Australia


*Correspondence author. E-mail: jeff.wright@amc.edu.au

Summary

  • 1Understanding the impacts of invasive species on natural ecosystems is an important component of developing management strategies. Habitat-forming invasive plants and sessile invertebrates often support a high diversity and abundance of native fauna, suggesting some benefits of invasion. However, the fitness responses of these native fauna, and thus the net benefit from their association with habitat-forming invasive species, are not well understood.
  • 2We determined how fitness-related life-history traits, patterns of resource allocation among life-history traits, and survivorship of an abundant bivalve, Anadara trapezia, responded to invasion by the habitat-forming seaweed, Caulerpa taxifolia, by transplanting A. trapezia into invaded and uninvaded habitats over a 12-month period.
  • 3Although A. trapezia recruits into C. taxifolia in high numbers, adult growth, body condition, shell condition, female reproduction and survivorship were all significantly lower in C. taxifolia compared to unvegetated sediment. Notably, we observed high mortality in C. taxifolia after heavy rainfall events, highlighting a potential link between sublethal effects on condition, stochastic environmental perturbation and survivorship.
  • 4In addition to the overall reduction in fitness, there were changes in scaling relationships between reproduction and body size following invasion. Female A. trapezia in C. taxifolia habitat allocated proportionally more resources to reproduction (including reproductive tissue and number of eggs per follicle) than those in unvegetated sediment despite their poor condition. Maximizing reproduction following invasion may impose a further cost to already stressed A. trapezia and contribute to the higher mortality observed when living in C. taxifolia.
  • 5Synthesis and applications. The full impact of habitat-forming invasive species is complex and understanding it cannot be based solely on descriptions of native species diversity or abundance. Our study has identified how the presence of long-lived species within habitat-forming invasive species may simply indicate an extinction debt. A decline in the fitness of A. trapezia in C. taxifolia appears to increase its probability of mortality in the long-term. We recommend that management approaches for C. taxifolia and other habitat-forming invasive species combine an understanding of impacts on species diversity, abundance and the fitness of associated fauna to provide a more pluralistic understanding of their effects.

Introduction

Invasive species are a major threat to natural ecosystems and biodiversity (Vitousek et al. 1996; Mack et al. 2000). Many studies show that invasive plants have negative effects on native plant community structure, species diversity and abundance, and also on higher trophic levels (Levine et al. 2003). Nonetheless, many native fauna use invasive species as habitat, and recent reviews highlight the high faunal diversity and abundance often associated with ecosystem engineering or habitat-forming invasive plants and sessile invertebrates compared to uninvaded habitats (Crooks 2002; Sax, Kinlan & Smith 2005; Rodriguez 2006). The assumption behind such patterns of community or population-level facilitation is that the native fauna benefit from association with these invasive species. However, a net benefit for native fauna will only occur if their individual fitness is equal or higher in the invaded habitat (Robertson & Hutto 2006). Because habitat-forming invasive species often change ecosystem processes (Ehrenfeld 2003) or physical factors (Levine et al. 2003), invaded habitats may be stressful for native fauna and may reduce their fitness. With few exceptions (e.g. Posey, Wigand & Stevenson 1993; Crooks 2001), it is largely unknown how life-history traits and fitness of native fauna respond to habitat-forming invasive species, and thus, whether fauna gain a net benefit from that association.

In addition to effects on life-history traits, invasion may result in changes in patterns of resource allocation. Such a response in native fauna may be reflected in changes in scaling relationships between life-history traits (particularly reproduction) and body size, or in trade-offs between different life-history traits. Many life-history traits scale with individual size, but in stressful environments, scaling relationships can change (Méndez & Karlsson 2004), and there can be trade-offs between reproduction, growth and survivorship (Horvitz & Schemske 1988), or between reproductive traits (Galen 2000) to maximize fitness. For marine invertebrates, our understanding of how stressful environments affect life-history trade-offs comes largely from laboratory experiments (George, Fenaux & Lawrence 1991). This contrasts with evidence showing rapid changes in life-history associations of invasive species in new regions (Blossey & Notzold 1995; Bøhn et al. 2004).

Evidence for community or population-level facilitation by habitat-forming invasive species is particularly common in marine and estuarine systems. Species of invasive salt marsh, seagrass, seaweed, ascidians, mussels and oysters contain a higher diversity or abundance of native fauna than uninvaded habitats (Posey 1988; Castel et al. 1989; Crooks 1998; Crooks & Khim 1999; Hedge & Kriwoken 2000; Castilla, Lagos & Cerda 2004; Bulleri et al. 2006; Gribben & Wright 2006a). Generally, these positive effects are on epifauna (although see Posey 1988), and have been linked to added habitat complexity for recruitment and the provision of refuges from predation (Crooks 2002; Gribben & Wright 2006a). However, invasion also modifies ecosystem properties (Neira, Levin & Grosholz 2005; Hacker & Dethier 2006; Neira et al. 2006) and life-history traits of marine fauna are sensitive to such changes (Shumway, Scott & Shick 1983; Reusch & Williams 1999; Allen & Williams 2003). Consequently, marine habitat-forming invasive species may, on one hand, have positive short-term effects on recruitment but, on the other hand, have negative long-term effects on adult life-history traits and fitness of associated fauna. Such potentially contradictory impacts make it difficult to clearly identify management goals. Thus, increasing the understanding of changes in life-history traits and fitness of native fauna following the invasion of marine habitat-forming invasive species will increase the understanding of their overall impacts and enhance management decisions.

In this study, we determined the fitness response of the native infaunal bivalve Anadara trapezia (Arcidae, Deshayes 1840) to the habitat-forming invasive alga Caulerpa taxifolia (Vahl) C. Agardh. We hypothesized that the fitness of A. trapezia would decline in C. taxifolia relative to uninvaded habitat, and that patterns of resource allocation among life-history traits would change in response to C. taxifolia invasion. In south-eastern Australia C. taxifolia has invaded numerous sites with established populations of A. trapezia (Wright, McKenzie & Gribben 2007). C. taxifolia has positive short-term effects on A. trapezia recruitment, but surveys indicate that adult densities, dry tissue weight and reproduction are reduced, although adult densities in C. taxifolia still reach 30 m−2 compared to a range of 30–90 m−2 in uninvaded habitat (Gribben & Wright 2006a,b; Wright et al. 2007). To determine the life-history and fitness consequences of C. taxifolia invasion for A. trapezia, we transplanted clams into invaded and uninvaded habitats for 12 months to answer these four main questions:

  • 1Do life-history traits (growth, reproduction, and shell morphology), condition (body and shell dry weight) and survivorship respond positively or negatively to invasion?
  • 2Does the response vary with animal size at the time of invasion or depend on previous exposure to invasion?
  • 3Does scaling of reproductive traits with body size change following invasion?
  • 4Are there trade-offs between reproductive traits following invasion?

We focused primarily on the response of adults because they dominate A. trapezia populations (Wright et al. 2007), are able to tolerate invasion of C. taxifolia for > 5 years (J. T. Wright, personal observations), and are reproductively active.

Materials and methods

species and study location

Caulerpa taxifolia is a coenocytic green alga considered one of the worst 100 invasive species in the world (Lowe et al. 2000). It has invaded several temperate regions worldwide where it covers large areas of shallow soft-sediment habitat (Meinesz et al. 2001; Creese, Davis & Glasby 2004). C. taxifolia forms high-density beds (Meinesz et al. 1995; Wright 2005; Wright & Davis 2006), modifies chemical and physical sediment properties (Chisholm & Moulin 2003), outperforms native seagrasses (Ceccherelli & Cinelli 1997) and alters fish foraging behaviour and species diversity (Longepierre et al. 2005; York et al. 2006).

Anadara trapezia is a large (up to 70 mm shell length), thick-shelled dioecious free-spawning bivalve clam (Gribben & Wright 2006b). A. trapezia are generally estuarine, occurring from the intertidal to shallow subtidal (0–3 m water depth) in unvegetated sediment, native seagrass and now C. taxifolia habitats (Wright et al. 2007). A. trapezia are suspension feeders with short siphons and generally occur at the sediment surface with between 30% and 90% of the shell buried in sediment depending on habitat (T. Wright et al., unpublished data).

Our experiments were conducted in Lake Conjola, a temperate barrier lagoon 590 ha in size located 210 km south of Sydney, Australia. C. taxifolia was first described in Lake Conjola in 2000, but it may have been present there for several years before that (Creese et al. 2004). Several native seaweeds and seagrasses inhabit Lake Conjola but the most abundant macrophyte is now C. taxifolia which covers greater than 150 ha of the lake floor (Creese et al. 2004). At our study location, Sponge Bay (35°15′4″ S, 150°26′47″ E), at the time of the experiment, the benthos was covered by intermingled patches of C. taxifolia and unvegetated sediment. The seagrasses Halophila ovalis and Zostera capricorni only occur as small patches in shallow zones fringing the bay. Sponge Bay also contains large populations of A. trapezia in both unvegetated sediment and C. taxifolia, which are dominated by individuals of > 45 mm shell length with relatively few individuals between 15 and 45 mm (Wright et al. 2007). High densities of A. trapezia recruits occur in C. taxifolia (up to 680 m−2) compared to unvegetated sediment (up to 25 m−2, Gribben & Wright 2006a).

transplant experiment

To examine how C. taxifolia affected A. trapezia life-history traits, condition and survivorship, we transplanted A. trapezia into sites in both C. taxifolia and unvegetated sediment (n = 6 sites per habitat) in Sponge Bay. Sites measured 1·4 × 1·4 m marked at each corner by rope attached to posts and were within 100 m of each other at similar depths (1·5–2 m deep). Once sites were established, they were cleared of A. trapezia by thoroughly searching the C. taxifolia or sediment by hand. A. trapezia used for transplanting into sites were initially collected, brought to the shore and tagged using numbered shellfish tags (8 × 4 mm FPN8O Glue-on shellfish tags, Hallprint Pty Ltd, Victor Harbor, South Australia) attached to shells following the suppliers’ instructions. Once tags were attached to clams, their shell length was measured on the shore using Vernier calipers and they were returned to the water. Clams were only out of the water for between 1 and 2 h. Each site was restocked with A. trapezia from the following four size classes: 15–30 mm (shell length, anterior–posterior axis: ‘small’ size class), 31–45 mm (‘medium’ size class), 46–60 mm (‘large’ size class), > 60 mm (‘very large’ size class). For the large and very large size classes, clams were collected separately from both C. taxifolia and unvegetated sediment in Sponge Bay and then reciprocally transplanted back into both habitats. For the small and medium size classes, clams were collected from a nearby location in Lake Conjola (~400 m from Sponge Bay) where they occurred in high densities. The benthos at this location was a mix of H. ovalis and unvegetated sediment; original habitat was not considered as an important factor for these two size classes. Each experimental site contained six clams from each of the six categories: small, medium, large (C. taxifolia origin), large (unvegetated origin), very large (C. taxifolia origin) and very large (unvegetated origin). The resulting density of 36 A. trapezia per site was realistic for this and nearby estuaries (Wright et al. 2007).

The experiment began in November 2004 and clams were re-measured every 3 months over a 12-month period. Clams were re-collected by visually searching and then by probing the C. taxifolia and/or sediment by hand. The perimeter of each site (up to 1 m) was searched for clams that may have moved outside the designated area. Clam shell length was measured onshore and any dead clams were recorded and removed. After 12 months, sites were searched in the same way as they were every 3 months and then partially excavated to ~5 cm to ensure no buried shells had been overlooked. Only one tagged clam was found during this excavation. At the end of the experiment, shell length, width (dorsal–ventral axis perpendicular to the umbo) and height (maximum thickness) of each clam were measured. For large and very large size classes, approximately half the surviving clams were randomly selected to examine the effects of ‘original’ and ‘transplant’ habitats on reproduction and the other half was used to examine the effects of the same factors on body condition (dry tissue weight). Because of low levels of recovery, all surviving clams ≤ 45 mm shell length were used to examine the effects of transplant habitat on body condition only. For reproduction, histological sections of the visceral mass with associated reproductive tissue were taken from a standard position and prepared as described in Gribben & Wright (2006b). From these sections, we measured the area of follicles and total area of tissue using ImageJ (Rasband 2005) and then calculated the area of somatic tissue (total area minus follicle area) and the ratio of the area occupied by follicles to total area to provide a measure of the ratio of reproductive:total tissue per individual (Gribben & Wright 2006b). Typically, examination of the partitioning of resources between reproductive and somatic tissue is based on biomass but in A. trapezia gonad is incorporated into the viscera and it is not possible to separate the different types of tissue. Instead, we used the ratio based on section area to provide an estimate of the proportion of resources allocated to reproduction (reproductive effort, Thompson & Stewart 1981). We also counted the number of oocytes in six randomly selected follicles for each female (see Gribben & Wright 2006b for more details of these methods). Body condition was determined by removing tissue from shells and drying it at 60 °C for 48 h. Shell condition (dry shell weight) was determined by drying shells at 60 °C for 48 h. To examine whether handling during the experiment had any effect on A. trapezia we set up a control plot in unvegetated sediment. Twenty six clams (52·55 ± 1·92 mm length, mean ± SE) were tagged and measured in the same way as the experimental clams but were left untouched for 12 months and only re-collected and measured at the end of the experiment. Survivorship and growth of these clams were similar to the experimental clams in unvegetated sediment (73·1% survivorship; 1·37 ± 0·23 mm growth, mean ± SE; see Results).

analyses

We analyzed survivorship in two ways: (i) total survivorship over 12 months and, (ii) the percentage of the total mortality over the 12 months that occurred from November 2004–August 2005 (i.e. the first 9 months). We separated the mortality from November 2004–August 2005 as high mortality was recorded in November 2005 (see Results) following heavy rain around Lake Conjola on 31 October 2005 (unpublished data, sourced from Bureau of Meteorology, Australia). Inspection of our sites on 27 October 2005 showed no evidence of high mortality. We observed high turbidity and low salinity on 4 November 2005 at our study location following this rainfall and subsequent flood event. Analysing the data in both ways enabled us to compare mortality over the entire 12 months including mortality linked to this heavy rainfall event and mortality prior to the rainfall event. For both data sets, we used three-factor analyses of variance (anova) with the factors original habitat, transplant habitat (both C. taxifolia vs. unvegetated) and size class (large and very large, all factors fixed) with site as the replicate. Survivorship of small and medium size classes was not determined because of the loss of many clams from these size classes. Tethering experiments revealed that A. trapezia are mobile up to 45 mm shell length (J. T. Wright and P. E. Gribben, unpublished data), and we could not separate loss from movement vs. predation for these small clams. We assumed clams > 45 mm shell length that were missing (15 in unvegetated sediment, 23 in C. taxifolia) had been removed and eaten by predators. Blue swimmer crabs Portunus pelagicus and octopus Octopus tetricus were the most likely predators. O. tetricus were common in Sponge Bay during the experiment and large numbers of A. trapezia shells were observed in their dens (J. T. Wright and P. E. Gribben, personal observations). Haphazard searches of nearby O. tetricus dens located several tagged shells and on one occasion, an individual O. tetricus was observed consuming one of the tagged clams in a C. taxifolia plot.

Differences in growth, reproduction, body condition and shell condition were determined with anova or, where shell length was a significant covariate, with analysis of covariance (ancova), followed by Tukey's tests where required. Effects of size were examined with ancova rather than including size class as a categorical factor as most of these traits scale allometrically with shell size. Multivariate analyses of variance (manova) were used to determine differences in multivariate shell morphology (length, width and height). For clams > 45 mm shell length, analyses examined the effects of original habitat and transplant habitat (C. taxifolia and unvegetated sediment) on traits, while for clams ≤ 45 mm shell length, analyses only examined the effect of transplant habitat. For analysis of the ratio of reproductive:total tissue, we also included sex as a factor. The end of the experiment coincided with the onset of the first spawning period for A. trapezia in late spring, which is synchronous between C. taxifolia and unvegetated sediment (Gribben & Wright 2006b). Spawning was observed at our study location 7 days after harvesting the clams. For all analyses, only clams alive at the end of the experiment were included, and because of low survivorship in C. taxifolia, clams were pooled across sites within habitat.

For females, changes in scaling relationships between body size and life-history traits as a function of C. taxifolia invasion were examined by comparing regression slopes of dry shell weight as a measure of body size (dry shell makes up 95·4 ± 0·1% of total biomass, mean ± SE, N = 91) and three traits: the area of reproductive tissue, the area of somatic tissue, and the ratio of reproductive:total tissue, between habitats. Regression slopes were compared by testing for an interaction between transplant habitat and dry shell weight (covariate) using ancova (Quinn & Keough 2002). Evidence for trade-offs between the ratio of reproductive:total tissue (covariate) and oocytes per six follicles was examined by comparing regression slopes using ancova. The source habitat had little effect on these traits (see Results); hence, analyses focused only on changes in these relationships in response to the habitat to which they were transplanted.

Univariate assumptions were checked by examining distributions of residuals and plots of residuals vs. means (Quinn & Keough 2002) and data were transformed as required. For ancova, homogeneity of slopes between treatments and similarity of covariate values across groups were examined as recommended by Quinn & Keough (2002).

Results

Survivorship over 12 months was significantly lower for A. trapezia transplanted into C. taxifolia than into unvegetated sediment for both the large and very large size classes (F1,43 = 47·933, P < 0·001; Fig. 1). Most mortality, particularly in C. taxifolia, occurred between August and November 2005, which included the flood event of early November 2005. The percentage of total mortality that occurred between November 2004 and August 2005 was higher in C. taxifolia than in unvegetated sediment, although this was marginally non-significant (F1,43 = 3·678, P = 0·062). There was no significant effect of original habitat on 12-month survivorship (F1,43 = 1·094, P = 0·301) or the percentage of total mortality that occurred from November 2004–August 2005 (F1,43 = 0·091, P = 0·764).

Figure 1.

Twelve-month survivorship curves of A. trapezia for large and very large size classes in unvegetated sediment and C. taxifolia. Plots show mean ± 1 SE proportional survivorship at 3-month intervals for A. trapezia transplanted from unvegetated sediment back to unvegetated sediment (unvegetated – unvegetated), C. taxifolia to unvegetated sediment (C. taxifolia– unvegetated), unvegetated sediment to C. taxifolia (unvegetated –C. taxifolia) and C. taxifolia back to C. taxifolia (C. taxifoliaC. taxifolia). N = 36 for each treatment.

Growth was significantly lower for A. trapezia transplanted into C. taxifolia than into unvegetated sediment for clams ≤ 45 mm shell length (F1,26 = 19·499, P < 0·001), but not for clams of > 45 mm shell length (F1,138 = 1·240, P = 0·267; Fig. 2). Body condition was significantly lower for A. trapezia transplanted into C. taxifolia than into unvegetated sediment across all sizes (≤ 45 mm shell length: F1,24 = 10·922, P = 0·003; > 45 mm shell length: F1,62 = 57·225, P < 0·001; Fig. 2). For clams of > 45 mm shell length, A. trapezia transplanted from C. taxifolia also had a significantly lower body condition than A. trapezia transplanted from unvegetated sediment (F1,62 = 6·872, P = 0·011). Shell condition was significantly lower for A. trapezia transplanted into C. taxifolia than into unvegetated sediment across all sizes (≤ 45 mm shell length: F1,25 = 5·1389, P = 0·029; > 45 mm shell length: F1,120 = 17·090, P < 0·001; Fig. 2). manova showed that A. trapezia shell morphology differed between habitats (Table 1), indicating that differences in shell condition were due to differences in the size of shells. For clams of ≤ 45 mm shell length, univariate analyses indicated that all three shell dimensions were lower in C. taxifolia than in unvegetated sediment. For clams of > 45 mm shell length, clams were significantly less wide in C. taxifolia than in unvegetated sediment. There were no effects of the source habitat on shell morphology (Table 1). There were indications that the lower shell weight was due to selective mortality for the very large size class (> 60 mm shell length) in C. taxifolia. Surviving clams in C. taxifolia had a lower shell dry mass than those that died (4·128 ± 0·516 g vs. 4·318 ± 0·276 g, mean ± SE, t48 =–2·565, P = 0·014). By comparison, surviving clams in unvegetated sediment had a similar shell dry mass to those that died (4·404 ± 0·296 g vs. 4·430 ± 0·299 g, mean ± SE, t43 = –0·234, P = 0·820).

Figure 2.

Adjusted mean ± 1 SE growth, body condition (dry tissue weight) and shell condition (dry shell weight) of A. trapezia≤ 45 mm and > 45 mm shell length over 12 months in unvegetated sediment and C. taxifolia. Values are back-transformed least square means from ancova tests. Sample sizes are shown at the top of each bar.

Table 1.  Shell morphometrics of A. trapezia transplanted to unvegetated sediment and C. taxifolia (means ± SE), manova tests on shell morphology (Wilks’λ) and univariate anova tests on each trait examining the effect of transplant habitat (≤ 45 mm shell length) and the effect of original habitat and transplant habitat (> 45 mm shell length). Interactions were all non-significant (P > 0·25) and are not shown : manova tests
N≤ 45 mm> 45 mm
UnvegetatedC. taxifoliaUnvegetatedC. taxifolia
25411132
Dimensions
Length (mm)44·76 ± 0·8138·75 ± 3·9060·05 ± 0·5360·75 ± 0·71
Width (mm)38·64 ± 0·7133·50 ± 2·9051·76 ± 0·4849·72 ± 0·54
Height (mm)30·44 ± 0·4926·75 ± 2·1441·44 ± 0·5340·97 ± 0·72
 d.f.FPd.f.FP
Transplant habitat 
 Multivariate Wilks’λ
 3, 252·5080·0823, 13812·073< 0·001
 Univariate response
  Length1, 275·8860·0221, 140 0·7170·398
  Width1, 276·0470·0211, 140 4·2380·041
  Height1, 276·3100·0181, 140 0·1230·727
Original habitat
 Multivariate Wilks’λ
 3, 138 0·7870·503
 Univariate response
  Length1, 140 1·9940·160
  Width1, 140 1·1280·290
  Height1, 140 0·4820·489

The ratio of reproductive:total tissue of A. trapezia varied with transplant habitat and sex (transplant habitat × sex interaction: F1,61 = 6·224, P = 0·015) and was significantly lower in females in C. taxifolia vs. unvegetated sediment (Tukey's α < 0·05, analysis between sexes within transplant habitat, Fig. 3a). The ratio of reproductive: total tissue of males did not differ between transplant habitats (Tukey's α < 0·05). The number of oocytes per six follicles of females was also significantly lower in C. taxifolia than in unvegetated sediment (F1,35 = 6·603, P = 0·015, Fig. 3b). On average, females transplanted from C. taxifolia produced more oocytes than females from unvegetated sediment, although this was non-significant.

Figure 3.

The mean ± 1 SE (a) ratio of reproductive:total tissue in males and females and, (b) number of oocytes per six follicles in females of A. trapezia reciprocally transplanted between unvegetated sediment and C. taxifolia. Sample sizes are shown at the top of each bar.

There were significant differences in regression slopes between habitats for both the ratio of reproductive to total tissue and the area of reproductive tissue vs. dry shell weight. In C. taxifolia, the direction of these relationships was positive, indicating an increase in reproduction with increasing dry shell weight in C. taxifolia (Fig. 4a,b). There was no difference between habitats for the relationship between the area of somatic tissue and dry shell weight (Fig. 4c). The number of eggs per six follicles increased with the ratio of reproductive: total tissue in C. taxifolia but not in unvegetated sediment; regression slopes were significantly different between habitats (Fig. 4d). However, the range of covariate values differed slightly between groups because smaller A. trapezia in C. taxifolia had low amounts of reproductive:total tissue (Fig. 4d). Note that removing the individual with very low reproduction in C. taxifolia (Fig 4a,b,d) did not change the significance of these tests.

Figure 4.

Regressions between traits of female A. trapezia in unvegetated sediment (solid line) and C. taxifolia (dashed line). (A) Dry shell weight vs. the ratio of reproductive:total tissue, (B) dry shell weight vs. the area of reproductive tissue, (C) dry shell weight vs. the area of somatic tissue, and (D) the ratio of reproductive:total tissue vs. eggs per six follicles. All log(x + 1) – log(x + 1) relationships. Homogeneity of regression slopes between habitats were tested by examining interactions between habitat and dry shell weight (A–C) and habitat and the ratio of reproductive:total tissue (D) using ancova: (A) F1,29 = 7·331, P = 0·011, (B) F1,29 = 7·685, P = 0·010, (C) F1,29 = 0·014, P = 0·907, and (D) F1,34 = 4·506, P = 0·041.

Discussion

fitness consequences

Most previous studies of the response of native fauna to marine habitat-forming invasive species have only described effects on species diversity or abundance. Our results show that despite positive effects on recruitment (Gribben & Wright 2006a), invasion by a habitat-forming seaweed has strong and consistent negative effects on the fitness of a resident bivalve with growth, reproduction, body condition, shell condition and survivorship all lower in C. taxifolia compared to unvegetated sediment. We are aware of only one previous experimental test of the fitness response of native fauna to marine habitat-forming invasive species. Crooks (2001) examined two native bivalves at higher densities following invasion of the mussel Musculista senhousia and showed negative effects on the growth and survivorship of one species but no effects on the second species. Moreover, for A. trapezia that survive in C. taxifolia, there were changes in the pattern of resource allocation among life-history traits that may further reduce fitness. Thus, for A. trapezia, an ontogenetic shift in their habitat requirement between epifaunal at recruitment (Gribben & Wright 2006a) to infaunal as a juvenile and adult suggests it goes from being positively to negatively affected by C. taxifolia. While this finding for A. trapezia is novel for a single species, it is consistent with abundance data showing generally positive effects of marine habitat-forming invasive plants on epifauna but negative effects on infauna (Hedge & Kriwoken 2000; Levin, Neira & Grosholz 2006; Neira et al. 2006).

The observation that A. trapezia appears able to persist in C. taxifolia for at least 5 years (J. T. Wright, personal observations) highlights the importance of understanding the sub-lethal effects of invasive species and suggests at least three important consequences. First, lower adult growth and reproduction results in reduced individual-level fitness. Notably, the negative effects on female reproduction reinforce the idea of greater female vulnerability to habitat-forming invasive species (Gribben & Wright 2006b). These effects on fitness will also have implications for population growth rates (e.g. Williams & Crone 2006). Secondly, the lower survivorship in C. taxifolia suggests that the sublethal effects on body condition accumulate to have lethal effects and contribute to the lower densities of A. trapezia in invaded habitats (Wright et al. 2007). The massive flood-associated mortality in C. taxifolia indicates a potential link between sublethal effects on condition, stochastic environmental perturbation and survivorship which may be an important, and as yet, untested mechanism driving some of the effects of marine habitat-forming invasive species. The rainfall event which preceded this mortality was the highest during our experiment, but 8-year rainfall data suggest an event of this magnitude is likely to occur every 2 to 3 years (J. T. Wright and P. E. Gribben, unpublished data, sourced from Bureau of Meteorology, Australia). Floods are generally stressful events for estuarine bivalves (e.g. Mathews & Fairweather 2004) and in the laboratory A. trapezia have low survivorship at salinities < 15 ppt (Nell & Gibbs 1986). Thirdly, relatively long-lived species such as A. trapezia may initially be resilient to invasion but if their fitness declines thereafter and increases their probability of mortality, there may be an ‘extinction debt’ that delays localized extinction (Tilman et al. 1994). Thus, for long-lived native species, describing changes in sub-lethal fitness may be an important predictor of long-term impacts of habitat-forming invasive species and this information should be incorporated into management approaches.

In addition to the overall high mortality in C. taxifolia, there appeared to be selective mortality with the largest clams most susceptible to C. taxifolia invasion. Loss of the largest clams coupled with lower overall densities in C. taxifolia may have further important consequences for A. trapezia. These include the potential for allee effects resulting from reduced gamete production and fertilization success due to gamete dilution (Levitan 1991), and reduced settlement of larvae and recruitment of post-settlement juveniles as both are gregarious, have strong preferences for adults as habitat, and juveniles are present on adults more than 12 months after settlement (Gribben & Wright 2006a; P. E. Gribben et al., in press). The high densities of recruits of A. trapezia on C. taxifolia (Gribben & Wright 2006a) suggests, however, that factors other than adult density also influence settlement. Additionally, the loss of adults from sites with C. taxifolia may have further implications for downstream recruitment at non-infested sites or estuaries.

We also found that previous exposure to C. taxifolia had little effect on A. trapezia life-history traits, condition and survivorship. The only exception to this general pattern was a lower body condition in A. trapezia transplanted from C. taxifolia (Fig. 2). These weak effects of previous exposure imply that surviving A. trapezia should recover from C. taxifolia's effects once it is removed. Although C. taxifolia decreases or even disappears periodically at some places (Glasby & Creese 2007), by increasing localized control programmes (e.g. Glasby, Creese & Gibson 2005), management authorities should allow A. trapezia populations to recover.

scaling of life-history traits and trade-offs

As well as lowered fitness-related traits (e.g. reproduction, growth and survivorship) in C. taxifolia, changes were found in the scaling relationships between body size and reproductive traits. In unvegetated sediment, there was a weak negative relationship between the ratio of reproductive to total tissue vs. dry shell weight which may reflect a cost to reproduction with increasing size for unstressed clams. By contrast in C. taxifolia, positive relationships between reproduction and body size (including the amount of reproductive tissue and eggs per six follicles) indicated increased allocation to reproduction with increasing size for these stressed clams. The positive relationships in C. taxifolia appeared driven by a lower allocation of resources to reproduction in clams < 4·1 g dry shell weight (maximum size of the large size class in C. taxifolia) compared to the same sized clams in unvegetated sediment. In contrast, in clams of > 4·1 g dry shell weight (very large size class), reproduction in C. taxifolia and unvegetated sediment covered a similar range. Whether these positive relationships reflect trade-offs between allocation to reproduction vs. somatic tissue in C. taxifolia that differ in direction between different sized clams (allocation to reproduction in very large clams but away from reproduction in large clams), or whether shell or somatic tissue would be traded-off under such a scenario remains unclear. Previous studies demonstrate that shell and tissue growth in bivalves are independent (e.g. Lewis & Cerrato 1997). Unlike other marine invertebrates, particularly those with lecithotrophic larvae which maximize egg size in stressful environments (e.g. George et al. 1991), free-spawning invertebrates such as A. trapezia may be more likely to alter egg number than egg size. There is no evidence for a trade-off between egg size and egg number in A. trapezia (Gribben & Wright 2006b), suggesting that in C. taxifolia, very large A. trapezia invest in egg number by maximizing reproductive tissue and eggs per follicle.

Overall, the response of A. trapezia in C. taxifolia is consistent with there being a trade-off between allocation of resources to reproduction at the expense of long-term survivorship in the stressful environment, and this may contribute to the higher mortality observed in C. taxifolia. However, determining whether maintaining reproduction in C. taxifolia had a cost for A. trapezia in terms of survivorship (a survival cost, Reznick 1985) was not possible given the problems in measuring reproduction non-lethally in situ.

potential mechanisms of impact

Our findings suggest that marine habitat-forming invasive plants and seaweed are detrimental to infauna. Invasion by salt marsh plants into mudflats reduces water flow and increases the deposition of organics, mud content, levels of chlorophyll a, anoxia and pore water sulphides in the sediment (Neira et al. 2005; 2006; Hacker & Dethier 2006). Habitats invaded by C. taxifolia also have anoxic and sulphide-rich sediments (Chisholm & Moulin 2003), which can be toxic to bivalves (Shumway et al. 1983; Laudien et al. 2002). To our knowledge, there have been no experimental tests for any marine or estuarine habitat-forming invasive species determining which of these factors impact native fauna, although several have been linked to a lower abundance of surface-feeding fauna (Neira et al. 2006; 2007). In addition to environmental factors, increased crab predation contributes to the effects of invasive Spartina on native fauna (Neira et al. 2006), and high predation rates in C. taxifolia may contribute to the negative effects on A. trapezia. Marine plants and rhizophytic algae such as Caulerpa influence sediment nutrients via uptake of nutrients from the sediment and input of organic matter into the sediment via exudates and detritus (Moriarty, Iversen & Pollard 1986; Chisholm et al. 1996; Williams & Heck 2001). Consequently, habitat-forming invasive plants and seaweed are more likely to impact sediments and native sediment fauna, than habitat-forming invasive invertebrates. As well as affecting nutrient cycles, high organic input can change invaded habitats from algae- to detritus-based food webs (Levin et al. 2006) and have cascading effects on sediment anoxia and sulphate-reducing bacteria (Welsh 2000). Although their faeces can enrich sediments (Reusch & Williams 1999), invasive mussels and oysters affect sediment characteristics primarily through the addition of shell and byssus mats (Castel et al. 1989; Crooks 1998; Crooks & Khim 1999). Overall, the main effects of invasive mussels and oysters may be the addition of complex structure above the sediment which is likely to have largely facilitative effects on epifauna.

conclusions and recommendations

An important component of managing invasive species is understanding their impacts on native species as this should help prioritize management efforts (Hulme 2006). Our study shows that high levels of recruitment of native fauna or their occurrence as adults in habitat-forming invasive species does not necessarily mean they benefit from the invasion. Typically, management frameworks for invasive species focus on impacts on diversity and population abundance. However, these frameworks may be inadequate for managing habitat-forming invasive species. Positive or no effects on diversity and abundance may be interpreted as ‘no apparent negative impact’ leading to erroneous conclusions. This in turn may lead to management inertia regarding their control. The link between poor fitness and low survivorship of A. trapezia within C. taxifolia suggests that the current understanding of the impacts of C. taxifolia, and thus management strategies, would be enhanced by understanding the fitness of other relatively long-lived species that co-occur with it – such as numerous fishes (York et al. 2006). These long-lived species may also have a localized extinction debt. Although eradication of C. taxifolia in south-eastern Australia is unlikely, the current management plan for C. taxifolia focusing on localized control and limiting further spread (Glasby et al. 2005) should be maintained to minimize its impacts on native biodiversity.

Acknowledgements

We are grateful to John Gollan and Adriana Verges for help in the field, Louise MacKenzie for video image analysis and the histological unit, School of Medicine, UNSW for processing samples. Pete Biro, Steve Bonser, Chris Frid, Philip Hulme, David Schiel, Martin Wilkinson and an anonymous reviewer provided valuable comments on early drafts of the manuscript.

Ancillary