Salt spray limits the inland penetration of a coastally restricted invertebrate: a field experiment using landhoppers (Crustacea: Amphipoda: Talitridae)



  • 1 Invertebrates from at least three major groups (crustaceans, gastropods and insects) have distributions that are restricted to within a few hundred metres of the coast, but unlike coastal plants there has been little discussion and no tests of the mechanisms that might control such distributions.
  • 2 Coastally restricted landhoppers, Austrotroides maritimus Friend, were transported inland of their natural distribution and established in enclosures at a site in far southern Tasmania. Salt (dry and in solution) was added to these enclosures to test the hypothesis that this species is confined to the coast by a requirement for salt.
  • 3 Over 5 months, A. maritimus persisted and reproduced in the dry salt treatment, but numbers remained low or declined in the other treatments.
  • 4 The addition of salt in solution did not produce the increase in numbers seen in the dry salt treatment. There was no evidence that non-coastal species declined under the salt treatments.
  • 5 The restriction of A. maritimus to the coast is explained in terms of its dependence on a supply of ions from salt spray, rather than a resistance to conditions which other more competitive species cannot tolerate.
  • 6 This conclusion is qualified by the possibility that there may be occasional salt-concentrating events in the coastal zone which raise the salt concentration above the tolerance levels of non-coastal species.
  • 7 Coastal landhoppers, in Tasmania at least, are not plesiomorphic members of their genera, so their distributions appear to be secondary, rather than representing an early stage in land colonization.


On all but the most sheltered coasts, wave and wind action combine to deposit aerosol droplets of sea salts on adjacent coastal land. This may extend far inland – affecting, for example, the ion dominance in lakes (Tyler 1974) over 80 km from the coast – but it is most intense in the first kilometre inland. The effect of salt on the distribution and growth of plants in this zone is well recognized (e.g. Oosting & Billings 1942; Boyce 1954; Dickinson 1977; Barbour 1978; Avis & Lubke 1985), although there has been some discussion about the relative importance of salt and moving sand in controlling the vegetation of coastal sand dunes (Maun et al. 1999). The distribution of coastal vegetation is generally explained in terms of the superior tolerance of some species to salt deposition, but Maun (1994) also lists nutrient deficiency, lack of moisture, sand accretion and predation as the factors that coastal plants must tolerate.

Coastally restricted distributions of land animals are less well known, but nonetheless they exist in several groups. At least two families of terrestrial pulmonate snails have members with coastal distributions. The ellobiids are typical of the upper regions of saltmarshes, but also occur in coastal forest in the tropics (Little 1990). The Orthalicidae are found in Australia, New Zealand and the Pacific Islands, and some species, such as Bothriembryon tasmanicus (Pfeiffer), have distributions confined to a narrow coastal range (Dartnall 1972; Smith & Kershaw 1979). Bristle tails (Insecta: Thysanura), e.g. Petrobius spp., are characteristic of the upper splash zone of rocky shores in temperate regions. Davies & Richardson (1970) suggested that their distribution in Britain is limited by the availability of salt, as evidenced by their deeper inland penetration in places where the prevailing winds carried salt spray further. Several species of woodlice (Crustacea: Isopoda) from the relatively well-mapped fauna of the British Isles are confined to coastal habitats: saltmarshes, the strandline of sandy beaches, pebble or boulder shores and sea cliffs (Harding & Sutton 1985). Some of these species are also found inland at synanthropic sites (Hopkin 1991) but their natural ranges seem to be correlated with elevated salt concentrations. Although rarely explicit, it often seems to be assumed that these distributions are the result of the animals’ requirement for salts, rather than the tolerance of saline conditions seen in the coastal flora.

Fully terrestrial talitrid amphipods, or landhoppers, have radiated most extensively in the Southern Hemisphere (Friend & Richardson 1986), where they can be found hundreds of kilometres from the coast in habitats ranging from grasslands to subalpine moorland and rainforest. Amongst this terrestrial fauna, at least in Australia (Friend 1987) and New Zealand (Duncan 1994), are a few species, which although fully terrestrial in all other respects, are confined to the coastal zone. Their distributions extend around the coast, but only a few tens of metres inland from the shore (Richardson et al. 1991). Circumstantial evidence suggests that these species penetrate further inland where exposure to onshore winds is greater (Richardson et al. 1991; Richardson 1993).

These observations lead to two principal hypotheses to explain the restricted distribution of coastal landhoppers:

  • 1Coastal landhoppers are limited by a physiological requirement for salt spray.
  • 2Coastal species have a superior tolerance of elevated salt concentrations in the coastal zone, but are excluded from inland habitats by competition or some other biological interaction.

In both the above scenarios, two subsidiary hypotheses are relevant:

  • 3Inland species are excluded from coastal habitats by intolerance of elevated salt concentrations, or
  • 4Inland species are excluded from coastal habitats by inferior competitive ability under conditions of elevated salt concentrations.

This study was designed to test hypothesis (1), but the nature of our experimental design will shed some light on (3) and (4).

In Tasmania, at least four species of landhoppers show coastally restricted distributions (Friend 1987, unpublished observations). We selected Austrotroides maritimus Friend for our study since it is one of the most widespread, with a range extending around much of the western and southern coast of the island. At all sites where it has been recorded it is confined to a zone less than 100 m from the high-water mark.

We examined the survival and reproduction of A. maritimus over 5 months in field enclosures to which they had been transferred at sites inland of their normal range, under various treatments of added salt and water. Evidence of improved survival and/or reproduction in A. maritimus at inland sites when supplied with extra salt would support hypothesis (1). The experiment was not designed explicitly to test the other hypotheses, but a decline in the number of non-coastal species under a regime of salt addition would support hypotheses (2) and (3), and negative interactions between A. maritimus and non-coastal species would support hypotheses (2) and (4).


South Cape Bay, in the extreme south of Tasmania (43°36′ S, 146°47′ E), provides an undisturbed transition from a sandy beach to native forest, and has been the site of previous studies of landhopper distributions (Richardson et al. 1991; A. M. M. Richardson & R. Swain, unpublished data). The bay faces almost due south, but is sheltered from the prevailing southwesterly and westerly winds by promontories to the west. Low coastal scrub and forest appear within a few metres of high water; closest to the shore the shrubs Correa backhousiana Hook., Cyathodes abietina R. Br., Westringia brevifolia Benth. and Olearia phlogopappa (Labill.) DC. form a band of low dense cover, 5–10 m deep, which gives way to low forest with emergent trees of Eucalyptus nitida Hook. F., Sassafras (Atherosperma moschatum Labill.), Celery Top Pine (Phyllocladus aspleniifolius (Labill.) Hook. F.), and Southern Beech (Nothofagus cunninghamii (Hook.) Oerst.). This coastal forest has an understorey of native laurel (Anopterus glandulosus Labill.), and often large clumps of tussock-forming grasses, Gahnia grandis (Labill.) S.T. Blake and Lomandra longifolia Labill. The ground layer is mostly absent, apart from a few mosses and herbs. The patterns of coastal vegetation in Tasmania are described by Kirkpatrick & Harris (1999).

The topography immediately inland from the beach consists of a series of two or three low ridges, apparently stabilized dunes, separated by damper slacks. The soils are sandy closer to the sea, but become more clayey inland; the litter layer also becomes better-developed inland.

A site was chosen in the centre of the bay, at a point where A. maritimus extends about 40 m inland from the high-water mark/coastal edge of the terrestrial vegetation. Although A. maritimus is by far the most abundant species in the coastal zone, it shares the zone with low numbers of some non-coastal species, notably Mysticotalitrus tasmaniae (Ruffo). An experimental area was established 30 m inland from the most inland record of A. maritimus; extensive hand collecting at this site confirmed that A. maritimus was absent. Sixteen circular enclosures (2 m diameter) were established using 40 cm deep galvanized iron sheet let into the ground about 10 cm with a knife and held in place with wooden pegs. Like all the landhoppers studied here, A. maritimus lives in surface leaf litter and does not penetrate the soil profile. The enclosures were placed more or less linearly along a shallow gully running parallel to the shore (Fig. 1).

Figure 1.

Diagrammatic plan of the study site to show the location of the enclosures and distribution of the treatments.

Four more enclosures were set up within the range of A. maritimus, in the first slack, 5 m from the seaward edge of the terrestrial vegetation. No treatments were applied to these enclosures, but they were sampled at the same frequency and over the same period as the experimental enclosures to monitor changes in the density and reproductive status of the natural A. maritimus population. The same soil factors were recorded from these enclosures as from the experimental ones (see below).

Austrotroides maritimus was collected in bulk from the seaward end of its range. Approximately 0·5 m2 of litter was quickly scraped onto a plastic sheet and then carried to the experimental area where it was added to one of the enclosures. Given the density of A. maritimus at the coastal end of its range, this meant that approximately 400 animals were added to each enclosure. The numbers of animals required and their susceptibility to damage when handled individually meant that it was not feasible to add a known number of animals to each enclosure. Five pitfall traps were set up at randomly chosen positions in each enclosure. The traps consisted of a plastic cup, 50 mm in diameter and 100 mm deep, let into the ground, and containing 20 ml of ethylene glycol. The traps were protected from rain by a Petri dish lid supported on twigs; this lid was used to close the traps between sampling periods. There were no significant differences in the catch of A. maritimus between enclosures over the three days immediately after the transfer (anova, F15,64 = 0·847, P = 0·624).

The following treatments were allocated at random to the enclosures.

(A) DS (salt): 66 g of dry sea salts (Coral Reef™ Red Sea Salt; Red Sea Fish pHarm, Eilat, Israel) was scattered evenly over the surface of the litter. This was equivalent to the salt content of 2 l of sea water.

(B) WS (water + salt): 2 l of sea water, collected at the site, was applied evenly to the enclosure with a watering can.

(C) WT (water): 2 l of fresh water was applied evenly to the enclosure with a watering can. The water was collected from the nearby South Cape Rivulet.

(D) NT (no treatment): enclosures were untouched.

Salt was applied both dry and in solution because it was impossible to apply it in the way that it arrives naturally (i.e. more or less continuously during rainfall). The water only treatment provided some control for the addition of water in the water + salt treatment, but no control for the dry salt addition was possible. Given the high rainfall at the site (c. 1200 mm per annum), the salt is likely to have dissolved quickly. These treatments were applied immediately after the translocation of A. maritimus and the first density samples, in October 1996, and thereafter at monthly intervals until March 1997. On each of these occasions, the pitfall traps in each enclosure were left open for 3 days to estimate the density of landhoppers. The catch from each trap was identified to species, using Friend’s key (1987). The animals were not measured, but very small (<2 mm body length) individuals of A. maritimus were assumed to have been expelled from the females’ marsupia during preservation, since they were smaller than any animals seen in other collections. Because of the very open brood pouch in all Austrotroides species (Friend 1987), almost all the eggs or young being brooded are ejected from the pouch when the animals are preserved. These small animals were counted separately to give an index of the reproductive status of the population.

On each sampling occasion, five replicate soil samples (approximately 200 g wet mass) were collected from each enclosure using a 5-cm-diameter corer, placed in sealable polythene bags (Polythene Co., Glenorchy, Tas., Australia) and returned to the laboratory where the following soil variables were measured.

  • 1Soil moisture: percentage mass loss on drying for 48 h at 60 °C.
  • 2Soil organic content: percentage mass loss from dry soil after combustion for at least 8 h at 400 °C.
  • 3Soil sodium: 2 g of dry soil was added to 100 ml distilled water and allowed to stand for 24 h. The sodium content of the solution was determined by flame photometry (Corning Model 410; Ciba Corning Diagnostics, Halstead, UK).

Statistical analyses were performed using systat for the Macintosh Ver 5·2 (Wilkinson 1992).


The mean densities of A. maritimus in each treatment at each sampling occasion are shown in Fig. 2. Table 1 summarizes the results of a factorial analysis of variance of these data, in which salt and water were the two factors, and the six sampling occasions were treated as repeated measures.

Table 1.  Results of a repeated measures analysis of variance of pitfall trap catches of the coastal landhopper Austrotroides maritimus within four soil treatments: DS (salt alone), WS (salt + water), WT (water alone) and NT, an untreated control. Since there was a positive relationship between the variance and the mean, the density data were transformed to log10(x + 1)
Between subjects
Salt 9·034 19·0349·467 0·010
Water 1·016 11·0161·064 0·323
Salt × Water 4·241 14·2414·445 0·057
Within subjects
Time 4·707 50·9418·213<0·001
Time × Salt 3·053 50·6115·327<0·001
Time × Water 0·607 50·1211·059 0·392
Time × Water × Salt 2·628 50·5264·586 0·001
Error 6·877600·115  
Figure 2.

The catch of Austrotroides maritimus (mean catch per trap over three days in each treatment; four enclosures per treatment ± SE) over the 5 month study period. Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus.

Austrotroides maritimus persisted in the treatment enclosures to which salt had been added, whereas in the non-salt treatments its numbers declined over time (Fig. 2). The species had effectively disappeared from the no-treatment enclosure after 5 months. Numbers of A. maritimus in the salt + water treatment did not show any significant change over the experimental period, while in the dry salt treatment they increased sharply, in the same way as in the enclosures within the species’ natural range, until February when they began to decline.

Figures 3, 4 and 5 show the changes in soil parameters over time in the experimental and natural range enclosures, and Tables 2, 3 and 4 summarize the results of one-way repeated measures anovas for each factor. These analyses were carried out on the four inland treatments only, since the soil within the natural range of A. maritimus was clearly distinct from that in the inland experimental sites (see Figs 3–5). The soils within the normal ranges were more saline, moister and had a greater organic content than those within the experimental sites.

Figure 3.

Average soil sodium in each of the treatments (four enclosures per treatment) over the experimental period. Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus. Error bars have been omitted for clarity.

Figure 4.

Average soil moisture in each of the treatments (four enclosures per treatment) over the experimental period. Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus. Error bars have been omitted for clarity.

Figure 5.

Average soil organic matter in each of the treatments (four enclosure per treatment) over the experimental period. Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus. Error bars have been omitted for clarity.

Table 2.  Results of a repeated measures analysis of variance on soil sodium (g Na kg soil−1) in the experimental treatments at monthly intervals from October to February
Between subjects
Salt 107·197  335·732 2·993 0·036
Error 907·470 7611·940  
Within subjects
Time 194·423  448·60613·472<0·001
Time × Salt 160·824 1213·402 3·715<0·001
Error1093·805304 3·608  
Table 3.  Results of a repeated measures analysis of variance on soil moisture (%, arcsine transformed) in the experimental treatments at monthly intervals from October to February
Between subjects
Moisture0·238  30·0795·5340·002
Error1·091 760·014  
Within subjects
Time0·095  40·0243·6890·006
Time × Moisture0·061 120·0050·7870·664
Table 4.  Results of a repeated measures analysis of variance on soil organic content (%, arcsine transformed) in the experimental treatments at monthly intervals from October to February
Between subjects
Organic0·169  30·0564·1070·009
Error1·045 760·014  
Within subjects
Time0·112  40·0283·2130·013
Time × Organic0·132 120·0111·2660·238

Salinity increased in both the dry and wet salt treatments over time, the dry salt treatment approaching the sodium content of the natural range enclosures close to the shore by the end of the experiment. Salinity also increased in the water-only treatment for the first 3 months, but it declined in the no-treatment enclosures overall. Soil moisture showed some tendency to increase over the experimental period. The dry salt treatments showed consistently lower soil moisture than the other enclosures, but since the no-treatment enclosures were consistently moister than the others, these differences may reflect pre-existing local soil conditions, rather than the cumulative effects of water addition. Soil organic matter also varied significantly between the treatments, and showed a small but significant increase in the final two months.

As a means of roughly estimating the number of brooding females in each sample, the numbers of animals smaller than 2 mm was divided by 7, an average clutch size for this species (Friend 1987). The population of A. maritimus was apparently non-reproductive in October, since no brooding females were recorded in any traps. No brooding females were recorded in any month in the water-only treatment, and numbers in all the experimental treatments were lower than the natural range enclosures (Fig. 6). Of the experimental treatments, the dry salt treatment showed the most brooding females.

Figure 6.

Estimated number of brooding females per month in each treatment (four enclosures per treatment). Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus. Error bars have been omitted for clarity.

Eight other species of landhopper were collected in the traps. In order of abundance they were Mysticotalitrus tasmaniae , Keratroides angulosus Friend, an undescribed species of Keratroides, Austrotroides longicornis Friend, Neorchestia plicibrancha Friend, Tasmanorchestia annulata Friend, A. leptomerus Friend and K. rex Friend. Of these, T. annulata and K. rex are described as coastal species by Friend (1987) and were probably transported into the experimental enclosures along with the litter containing A. maritimus. Neither of these species was sufficiently abundant to warrant a quantitative analysis. Repeated measures analysis of variance of the catches of all the non-coastal species, using the same approach as above, showed that time was the only factor explaining a significant proportion of the variance (Table 5). However, Fig. 7 shows that there was no tendency for the abundance of non-coastal species to decline over the experimental period in any of the treatments. It also shows that the numbers of non-coastal species were lower than those of A. maritimus in the natural range enclosures (Fig. 7: an average catch of about 2 per trap, compared with 5 per trap in October, rising to over 25 per trap in March for A. maritimus).

Table 5.  Results of a repeated measures analysis of variance of pitfall trap catches of non-coastal landhopper species within four soil treatments: salt alone, salt + water, water alone and an untreated control
Between subjects
Salt  1·010 110·0100·7250·411
Water 27·094 127·0941·9630·187
Salt × Water 52·510 152·5103·8050·075
Within subjects
Time 95·469 519·0942·9980·018
Time × Salt 22·552 5 4·5100·7080·628
Time × Water 44·719 5 8·9441·4040·236
Time × Salt × Water 74·302 514·8602·3330·053
Error382·12560 6·369  
Figure 7.

The catch of non-coastal landhoppers (mean catch per trap over 3 days in each treatment; four enclosures per treatment) over the 5-month study period. Treatments: WS salt solution; DS dry salt; WT fresh water; NT no treatment; NR untreated enclosures within the natural range of A. maritimus. Error bars have been omitted for clarity.

There was no evidence of either positive or negative association between the numbers of A. maritimus and those of all other non-coastal species caught in each treatment enclosure on each occasion (Fig. 8).

Figure 8.

Scatter plot of the total catch of A. maritimus and all non-coastal species in each pitfall trap over the experimental period. (n = 96, Pearson’s r = 0·390).


On the basis of these results we suggest that the persistence and reproduction of the coastal landhopper Austrotroides maritimus at sites inland of its normal range are enhanced by the addition of sea salt to the soil. This effect is seen most clearly when salt is added in a dry form. However, there was some decline in the abundance of A. maritimus in the inland enclosures in the later months of the trial and it is not clear whether the animals would have continued to survive there in the long term. Reproductive rates in the salt treatments were also below those in the natural range of the species.

At least two factors confound the experimental design. There were distinct differences in soil conditions at the inland experimental site compared with the normal range of A. maritimus. The experimental soils were drier and contained less organic matter than the coastal sites; although a full soil analysis was not carried out, it was obvious that the coastal soil was largely made up of sand, while at the experimental site the soils were clayey. Thus it is impossible to separate the effects of these differences in soil conditions from the experimental manipulation of moving A. maritimus inland. Further, at the experimental site there was a pre-existing community of terrestrial landhoppers to which the experimental populations of A. maritimus were added. Although the numbers of these resident animals were relatively small, and some of the species co-occur with A. maritimus in its natural range, this also represents a confounding factor alongside the transplant treatment.

Neither of these confounding factors could be easily removed from the design. It would not have been possible to move soil from the coastal site to the inland site without causing massive disruption in the soil profile, and even if that were possible it would not have altered the local site effects of canopy cover, etc. Similarly there was no obvious way to remove the pre-existing community of landhoppers before adding A. maritimus.

The results support hypothesis (1), that A. maritimus is restricted to the coastal zone by its requirement for a supply of salt. Since there is no evidence of a decline in non-coastal species under the salt addition treatments, hypothesis (2), that the distribution of A. maritimus can be explained by a tolerance/competition mechanism, can be provisionally rejected. It may be unwise to reject this mechanism fully at this stage since the conditions to which A. maritimus is tolerant may only appear periodically, for example, as a result of unusually dry weather and evaporative concentration of salt.

The reduced survival of A. maritimus in the wet salt treatment may be evidence that salt leached away more quickly in the wet treatments. Alternatively, it may suggest that excessive wetness of the litter and soil is inimical to A. maritimus. The latter explanation is more likely since the sodium concentration in the dry salt enclosures did not exceed those in the wet salt enclosures until the final two sampling occasions, while the soil moisture in the dry salt enclosures was always lower than in any of the wet treatments. The inland penetration of A. maritimus is severely restricted at another nearby site where water-logged clay soils come close to the beach (A. M. M. Richardson & R. Swain, unpublished data). Since soil moisture levels were always higher within the natural range of A. maritimus, it may be that the interaction between high soil moisture and clay soil is more important than soil moisture alone.

Morritt & Richardson (2000) found that the osmoregulatory patterns of three coastal landhoppers (Tasmanorchestia annulata , Keratroides rex and A. maritimus) were broadly similar to those of two inland species; all but one species showed a fundamentally hyper–iso-osmotic osmoregulatory pattern. The exception, the coastal species Tasmanorchestia annulata, showed a tendency towards a hyper–hypo-osmotic pattern, and was able to maintain its haemolymph concentrations significantly below those of the other coastal species at external concentrations equivalent to sea water. At the lowest concentrations (40 mOsm kg−1), A. maritimus had significantly lower haemolymph osmolality than the other coastal species. The inland species showed a significantly higher haemolymph osmolality at low (<200 mOsm kg−1) external osmolalities (inland species: 500–600 mOsm kg−1; coastal species c. 400 mOsm kg−1). Austrotroides maritimus survived much more poorly at lower external concentrations than the other two coastal species (Morritt & Richardson 2000). Tasmanorchestia annulata survived well at all external salinities (tap water to full sea water), while K. rex survived poorly at the highest and lowest external salinities.

These results restrict the generalizations about coastal landhoppers that can be drawn from the present study. While they support the hypothesis that A. maritimus has a requirement for salt spray, it is not clear whether the same mechanism is responsible for the coastally restricted distributions of other landhoppers, especially T. annulata.Tasmanorchestia annulata is one of the cuspidactylate landhoppers, which are thought to be quite phylogenetically distant from the simplidactylate group in which Austrotroides and Keratroides lie (Bousfield 1984; Friend 1987), although this is not yet recognized in the taxonomy of the group.

The apparent dependence of A. maritimus on elevated salt concentrations in the soil shown here, and its poorer osmoregulatory performance and survival at low external concentrations (Morritt & Richardson 2000), might be taken as evidence that it is a phylogenetically primitive species that has not developed the physiological mechanisms to deal with the reduced availability of ions in terrestrial habitats. However, this does not correspond with morphological evidence. From a preliminary phylogenetic analysis (A. M. M. Richardson, unpublished data) A. maritimus, the only coastally restricted member of the genus, is not the most plesiomorphic species among its six described congeners. Similarly, K. rex is not the most plesiomorphic within the five described members of its genus. Neither do the distributions of coastal landhoppers support the idea that they are remnants of the initial colonization of land by the talitrids. Sandy beaches are unlikely to be the habitat where talitrids first crossed the sea–land interface (Bousfield 1984; Richardson & Mulcahy 1996; Richardson & Swain 2000). The harsh environmental conditions there have led to specializations for life in sand, and the sandhoppers as a group are phylogenetically remote from the landhoppers and their probable ancestors (Bousfield 1984). While other supra-littoral talitrids may colonize terrestrial habitats in suitable conditions (Bagenal 1957; Richardson & Mulcahy 1996; Morritt & Spicer 1998), sandhoppers rarely extend landward of the extreme high tide mark and rarely overlap with the fully terrestrial talitrid fauna (Richardson et al. 1991). Thus, if coastal landhoppers have always been confined to the coast, it is difficult to understand how they have colonized coastal sites far from the saltmarshes and estuaries, which are the most likely sites of initial land colonization, separated as they may be by large tracts of unsuitable habitat. This leads to the conclusion that A. maritimus, and perhaps all coastal landhoppers, have acquired their dependence on salt spray secondarily, though why this should have occurred is unclear.

Further work to unravel some of these problems is clearly required. It would be useful to know whether the distributions of the other coastal landhoppers in Tasmania are controlled in the same way as A. maritimus. The New Zealand landhopper fauna also provides a further test of our hypotheses since it includes several coastally restricted species (Duncan 1994), although they are all in the cuspidactylate group. Information on the dynamics of salt input and loss in the coastal zone is also required, especially the occurrence of events that might concentrate sufficient salt to eliminate non-coastal species from the zone. Finally, better taxonomic and phylogenetic information would allow evolutionary hypotheses about the mechanisms of land colonization by talitrid amphipods to be formulated with more confidence. This should include elucidation of the status of various undescribed forms of coastal landhopper, and well-supported phylogenies, both within genera that include coastal species, and at the subfamily level to test the validity of the cuspidactylate and simplidactylate groups.


Many people helped with the fieldwork; thanks to Alistair Hobday, Richard Holmes, David Morritt, Colin Shepherd, Sam Thalman and all others. Leon Barmuta is thanked for statistical advice, and David Morritt for many helpful discussions on talitrid physiology. We are grateful to the Tasmanian Parks & Wildlifez Service for permission to work within the Western Tasmania World Heritage Area. This study was partly funded by an Australian Research Council Small Grants award.

Received 28 June 2000; revised 6 December 2000; accepted 8 December 2000