Introduction

Grazing by large, mobile herbivores is fundamentally important in regulating algae in both temperate and tropical marine habitats. However, herbivory by small invertebrates, such as amphipods, can also affect algal communities (Brawley and Adey 1981; Hay and Fenical 1988; Hay et al. 1988a; Duffy and Hay 1990, 2000). While the foraging areas of some mesograzers (grazers <25 mm; sensu Brawley 1992) are comparable to larger grazers, others may exhibit more limited movement. Tubicolous marine amphipods demonstrate the range of this behavior. Many species live in fixed tubes formed in the algae they eat; they feed primarily near the openings of their tube (Heller 1968; Hay et al. 1988b) and venture out on short excursions infrequently, abandoning the tube only if conditions deteriorate significantly (Barrett 1966; Heller 1968; Poore and Steinberg 1999). Yet, still others display high rates of spread and colonize new algae rapidly over large distances (e.g., Edgar 1992; Taylor 1998; Poore 2005).

Large, mobile herbivores select algal food based principally on its nutritional value or lack of chemical defenses (Hay et al. 1988b), but the algal preferences of small grazers have been variously attributed to algal morphology (Nicotri 1980; Holmlund et al. 1990), toughness (Littler and Littler 1980; Steneck and Watling 1982), nutritional value (Cruz-Rivera and Hay 2000a, b, 2001), and secondary chemistry (Hay and Fenical 1988). Moreover, since many tube-dwelling amphipods are capable of only limited movement relative to their algal habitat, tube sites must be selected on the basis of both food preference and refuge quality of the algae (Nicotri 1980; Hay et al. 1987, 1988b, 1989, 1990). For instance, Duffy and Hay (1991) suggested that the choice of host algae by Ampithoe longimana was largely determined by predation pressure and was less reflective of feeding preference or diet value.

The herbivorous amphipods A. lacertosa and Peramphithoe humeralis are common species inhabiting protected intertidal and subtidal rocky shores of the San Juan Archipelago, Washington, USA (Heller 1968; Kozloff 1996). Both species are members of the family Ampithoidae, a group that often forms tubes or nests on the fronds of various algae (Skutch 1926; Barnard 1965; Griffiths 1979). With more than 102 species in 12 genera distributed worldwide, ampithoids represent the most species-rich family of herbivorous amphipods (Poore et al. 2008). Their intimate relationship with their host algae has made ampithoids the subject of experiments concerning plant–herbivore interactions (Nicotri 1980; Hay et al. 1987; Duffy 1990; Holmlund et al. 1990; Duffy and Hay 1990; Cronin and Hay 1996). The grazing activities of these amphipods influence the structure and organization of benthic communities (e.g., Duffy and Hay 2000). Population explosions of P. humeralis associated with ENSO events have reduced or eliminated some large kelp beds in the northeastern Pacific (Tegner and Dayton 1987; Dayton and Tegner 1989; Graham 2002).

In the present paper, we evaluated the use of algae as food and habitat by A. lacertosa and P. humeralis, both of which construct tubes from curled portions of algal thalli. Our specific aims were: (1) to describe the relationship between in situ distribution, diet preference, and juvenile survivorship of the species across host algae; (2) to determine whether the species’ preferences are based on characteristics of the host algae, particularly food value or toughness; and (3) to determine how feeding and tube building differed between the species in relation to their ecology and behavior (e.g., mobility and thigmotaxis).

Materials and methods

Organisms and study site

We chose six macroalgae for our experiments: Ulva lactuca, Mazzaella splendens, Alaria marginata, Desmarestia ligulata, Fucus distichus edentatus, and Saccharina latissima. These species vary in taxonomic relatedness, growth form and chemistry (Online Resource 1 in “Supplementary material”). Algae for experiments were collected at low tide from Shannon Point beach (Anacortes, Washington; 48°30.542′ N, 122°41.070′ W) or by SCUBA from nearby Burrows Channel. We minimized chemical and nutritional variability by using only freshly collected algae in good condition and by sampling over a restricted period (July–September). All epiphytes were removed from the algae prior to experiments and specimens were held in tables of running seawater for <1 week before use. Portions of algal tissue used in all experiments and assays were excised from non-reproductive, non-meristematic areas of the blades.

All amphipods used in experiments were collected from Shannon Point beach, between July and September in 1997, 2007, and 2008. Ampithoe lacertosa were captured by snorkeling in shallow water and collecting assorted algae in fine mesh bags; the algae were then sifted in a seawater table to dislodge the amphipods. Peramphithoe humeralis were collected at low tide by carefully inspecting seaweed thalli and removing obvious tubes. The tubes were often fully exposed at low tide and the amphipods remained inside even when the nests were handled. Only A. lacertosa between 20 and 25 mm total length and P. humeralis between 25 and 30 mm total length were used in the experiments because individuals within these size ranges represent the majority of adult specimens present on Shannon Point beach between July and September. For the purposes of the present study, we consider P. humeralis functional mesograzers, despite their large size (see Brawley 1992). The amphipods were maintained in tables with running seawater and fed mixed algae prior to the experiments.

Field distribution of amphipods

To assess the distribution of A. lacertosa and P. humeralis in the field, we sampled the six test algae on Shannon Point Beach during three low tides (July 28, 29, 31) in 1997, with additional observations in July–September 2007 and 2008. We sampled algae by walking a transect along the water’s edge (−0.30 to −0.46 m MLLW tide) and examining exposed blades of S. latissima, A. marginata, D. ligulata, and F. distichus edentatus. To compare amphipod densities on the algae, it was necessary to determine the surface area of the sampled algae within the transect during each low tide sampling event. For the first three species, we measured individual thalli and estimated surface area. This was impractical for F. distichus edentatus since the algal blades were smaller and highly branched. Instead, we collected ten representative thalli, measured their surface area with image analysis software, and multiplied the average surface area by the total number of thalli examined.

The blades of the remaining two species (U. lactuca and M. splendens) were also impossible to measure accurately in the field due to their morphology. In addition, earlier observations suggested that amphipods associated with those species formed less permanent nests and would escape the algae while it was being examined or collected. To accurately measure amphipods on those species, therefore, it was necessary to sample them differently. We accomplished this by quickly scooping up blades of U. lactuca or M. splendens from water at the edge of the transect until we had filled a single 8-l bucket with each. In the laboratory, we vigorously sieved the algae in a seawater table, dislodging any amphipods that were present. We collected and identified the amphipods left after sieving. To estimate the surface area of the collected samples, we took eight single blades, laid them flat on a piece of paper, measured their surface area with image analysis software then dried the sample. We also dried and weighed the total sieved sample. Using the mean mass to surface area relationships of the single thalli, we converted the dry mass measure of each species to a total surface area. This allowed us to directly compare the density of amphipods on algae with different morphologies.

To derive relationships between wet and dry biomass, additional algae and amphipods were collected from Shannon Point beach. We rinsed six thalli of each alga, removed all epiphytes, and excised small sections of non-reproductive, non-meristematic tissue. Individual A. lacertosa and P. humeralis were measured to the nearest mm (total length; TL). All samples were blotted dry, weighed, and dried in an oven at 60°C for 72 h. The dried tissues were re-weighed, ground by hand with a mortar and pestle, and burned in a muffle furnace at 550°C for 3 h to obtain ash-free dry mass (AFDM).

Amphipod feeding behavior

Rates of herbivory were quantified using no-choice and choice assays for A. lacertosa and P. humeralis. Individual amphipods were presented with a single alga in the no-choice assays, while we simultaneously offered groups of amphipods pieces of all six algae in choice assays. In both types of experiments, freshly collected algae were thoroughly rinsed, cleaned of epiphytes, cut into disks (1.43 cm diameter), pre-weighed, and placed in 10-cm-diameter glass bowls with 350 ml of seawater. This was repeated with ten replicate bowls in each assay. In order to differentiate amphipod feeding behavior from habitat preference, it was necessary to prevent the amphipods from building tubes in the algae. We accomplished this by stringing the algal disks individually on monofilament line and attaching them to the dish; individual algal disks were too small for tube building and the monofilament lines kept the amphipods from pulling several disks together (in choice assays). The arrangement did not appear to produce unusual feeding behaviors in the amphipods used in the assays (personal observation).

We wished to keep algae densities similar across assays (~1 amphipod 1.6 cm−2 algae); thus six A. lacertosa or P. humeralis were added to each bowl with six pieces of algae (one of each species) in choice assays. This ensured that the amphipods would consume measurable amounts of the algae over the experiment period. As a result of this design, however, measurements from a single bowl in the choice assay represented the feeding responses of a group of amphipods rather than single individuals, as in the no-choice assay. While it is likely that there were amphipod-to-amphipod interactions in the bowls, the disks were large enough to accommodate feeding by several amphipods simultaneously and intraspecific interactions did not appear to affect feeding (personal observation). Per capita rates of consumption were calculated to make comparisons between experiments.

To control for differences in autogenic change of the test algae, we prepared an equal number of control bowls with seawater and algal disks but no amphipods. We interspersed all bowls (treatment and control) in a sea table of running seawater and allowed the amphipods to graze in the treatment bowls for 54 h; feeding rates on algae in treatment bowls were quantified relative to the change in ungrazed algae in paired control bowls. The sea table received ≈12 h of combined overhead fluorescent lighting and natural sunlight per day. At the end of the experiment, the algae remaining in each bowl were collected, blotted dry, and reweighed. Following Taylor and Brown (2006), we corrected for autogenic change in the test algae by calculating consumption as: (T i  × C f /C i ) − T f , where T i and T f are pre- and post-assay mass of test algae in treatment bowls, and C i and C f are pre- and post-assay mass of algae in paired control bowls with no amphipods. This ensured that differences in algal biomass in the treatment bowls were due to amphipod grazing and not to inherent changes in the algae themselves over the period of the experiment. When necessary, data were log transformed. Differences in feeding rate of A. lacertosa in the no-choice assay were tested with one-way ANOVA followed by Tukey’s HSD tests. Because transformation failed to homogenize variances for the P. humeralis no-choice data, those data were analyzed with a Kruskal–Wallis non-parametric analysis of variance followed by pairwise comparisons of ranks. Choice feeding assays were analyzed with a Friedman non-parametric test of ranks followed by pairwise comparisons (Conover 1980).

We also compared feeding rates of A. lacertosa to those of P. humeralis in the choice assays. Because amphipods were offered multiple algal foods simultaneously, it was necessary to treat the data as a multivariate set with each algal choice as a dependent variable. We graphically compared feeding patterns of A. lacertosa and P. humeralis through MDS (non-metric multidimensional scaling, Clarke and Warwick 2001). We followed the MDS with ANOSIM to statistically test for differences in the feeding choices of the two species. Euclidean distance was used to create the similarity matrix. We also used SIMPER (similarity percentages) to determine which algal choices contributed most to differences in feeding patterns of the two amphipod species.

Amphipod habitat choice

To measure habitat preferences of the amphipods, we offered them pieces of all six algal species and allowed them to build tubes in the alga of their choice. Thalli of each alga were washed thoroughly in seawater and cut into rectangular sections (≈78 cm2). One piece of each of the six test algae was placed in a 20-cm-diameter glass bowl filled with 1,350 ml of seawater. Pieces were folded over to form a loop then clamped at regular intervals around the interior lip of the bowl with the folded loop hanging down into the seawater. A single amphipod was placed in each bowl and allowed 1 h to construct a tube. One hundred and ten amphipods of each species were tested individually and each amphipod and algal sample was used only once. The tests were conducted in a flow-through seawater table covered with black plastic to remove light as a factor influencing amphipod choice. Choice of an alga by the amphipod was indicated by a newly formed tube domicile with cemented fibers; trials in which no obvious tube was formed were eliminated from the analysis. Data were analyzed with a χ2 goodness-of-fit test to determine whether overall algal choice differed from random. To test the significance of individual algal choices, we adjusted standardized residuals to the variance and compared the results to a normal distribution (Whittam and Siegel-Causey 1981; Bingham and Braithwaite 1986). Occupied tubes from each trial were hung in a table of running seawater following completion of the habitat preference test. After 24 h, all tubes were checked to see whether they were still occupied. We used a χ2 test for independence to determine whether A. lacertosa and P. humeralis differed in their fidelity to tubes they had built on different algae.

Survivorship of juvenile amphipods

To determine the food value of each alga to juvenile amphipod performance, we raised juvenile A. lacertosa and P. humeralis on single diets of each of the six algae. Brooding adult females were fed a diet of mixed algae and held until newborn amphipods left the brood pouch. Within 52 h, juveniles were placed individually in separate 5-cm-diameter glass bowls with a single disk (1.43 cm diameter) of one of the six algal species and ~200 ml of 0.02-μm filtered seawater. Ten replicate A. lacertosa and 16 replicate P. humeralis were maintained on each diet; individuals in similar bowls without algae served as starved controls. All juveniles used in the experiment were selected from five mothers of each species and sibling individuals were dispersed among the treatment groups. The bowls were held in an incubator and maintained at 12°C on a 12-h light:12-h dark cycle for the duration of the experiment. Amphipods were examined every 2 days for evidence of feeding and/or mortality and food and water were changed at 4-day intervals until the experiments ended on day 36. If the amphipods consumed >75% of their algal food between water changes, a fresh piece of the alga was added. At the conclusion of the experiment, we calculated a survival time of each juvenile and compared the algal treatments with Kaplan–Meier survival analyses. We used log ranks to test for overall differences in the treatments then compared survival on individual algae by looking for overlap in the 95% confidence intervals on the estimated mean survival times. To determine whether performance differed among broods, survival to day 36 was compared among offspring of the five mothers of each species using a χ2 test of homogeneity.

Relationships between feeding preference, juvenile survivorship, and in situ adult abundance

To determine amphipod responses to macroalgal hosts, the relationships of feeding preference, survival, and in situ abundance were evaluated using the methods of Taylor and Brown (2006). Response variables were ranked such that a rank of one indicates most preferred food in choice trials, highest juvenile mean survival time, or highest adult amphipod abundance; least preferred food, lowest juvenile survival or lowest adult abundance receive a rank of six. Reduced major axis regressions were performed for each relationship (McArdle 1988).

Algal toughness

Because algal toughness could influence amphipod feeding and habitat preferences, we measured toughness of the test alga with a penetrometer designed after Duffy and Hay (1991) and Littler and Littler (1980). The penetrometer consisted of an insect pin attached to a light plastic cone and held point down in a vertical PVC sleeve. A wet algal thallus was clamped horizontally in a bracket below the pin, and dry sand was slowly added to the cone to increase force until the pin completely pierced the alga. The sand, cone, and pin were then weighed to determine the force (N) required to pierce the thallus. Measurements were made on marginal portions of ten thalli of each species and the results were analyzed with a one-way ANOVA and Tukey’s HSD pairwise comparisons. Data were log-transformed prior to analysis to homogenize variances.

Measurement of algal percent nitrogen and C:N ratio

We measured total nitrogen and carbon to nitrogen (C:N) ratios of each alga as an index of nutritional quality. Fresh algal tissue was dried in an oven for 48 h at 60°C and subsequently ground by hand to a powder with a mortar and pestle. Samples (2.0–4.0 mg, n = 5) of each alga were combusted in a Leeman Labs Elemental Analyzer using acetanilide as a standard. Total nitrogen (% dry mass) and C:N ratios were compared among the algal species with one-way ANOVA followed by Tukey’s HSD pairwise tests. Data were log-transformed to homogenize variances.

Results

Field distribution of amphipods

Field data showed a distinct difference in distributions of A. lacertosa and P. humeralis on macrophyte hosts. Ampithoe lacertosa was much more likely to be found on U. lactuca than on any other substrate we examined. In fact, only one individual was found on any other alga (M. splendens; Table 1). Conversely, P. humeralis occupied a broader range of host algae, primarily S. latissima, but they were also abundant on F. distichus edentatus and A. marginata; only two individuals were found on U. lactuca. Relationships between wet mass, dry mass, and ash-free dry mass of test algae and amphipods are included in Online Resource 2 in “Supplementary material”.

Table 1 Field distributions of Ampithoe lacertosa and Peramphithoe humeralis on six algal species on Shannon Point Beach, Washington, USA in July 1997
Full size table

Amphipod feeding behavior

Ampithoe lacertosa feeding differed significantly among the algae in the no choice assay (F = 2.65, P = 0.03; Fig. 1a); A. marginata was consumed at the highest rate and F. distichus edentatus was consumed at the lowest rate. However, all algae were consumed at some level. In the choice assay, A. lacertosa again appeared to be a generalist feeder. While individuals of this species fed most heavily on U. lactuca and D. ligulata, they also ate all other algae present in the bowls, though the amount of F. distichus edentatus consumed was quite low (Friedman’s test, P < 0.001; Fig. 1b). Peramphithoe humeralis feeding differed significantly among the algae in the no choice assay (F = 37.56, P < 0.001; Fig. 1c). Individual amphipods confined to the four kelp diets had much higher feeding rates than their counterparts confined to diets of U. lactuca and M. splendens. In the choice assay, P. humeralis were more selective in their grazing than A. lacertosa; they ate large amounts of S. latissima, D. ligulata, and A. marginata. Like A. lacertosa, P. humeralis in the choice assay ate only small amounts of F. distichus edentatus. However, they strongly avoided U. lactuca (the preferred food of A. lacertosa) and M. splendens (Friedman’s test, P < 0.001; Fig. 1d).

Fig. 1
figure 1

Feeding by a A. lacertosa and c P. humeralis. Amphipods were offered six algal species individually (a, c; n = 10 bowls), and feeding rates were compared. In a separate test, amphipods were presented all choices simultaneously (b, d; n = 10 bowls) and feeding was compared. Treatments with same letter (lower case for A. lacertosa and upper case for P. humeralis) were not significantly different. SE are shown. Inset is a non-metric multidimensional scaling (MDS) plot of square-root transformed data. Open circles represent A. lacertosa samples; filled circles are P. humeralis

Full size image

MDS analysis verified that the feeding behaviors of A. lacertosa and P. humeralis were very different; replicates for the two groups clearly separated on the ordination plot (Fig. 1, inset) and ANOSIM statistically verified the separation (Global R = 0.83, P < 0.001). SIMPER analysis showed that the separation of the two groups was attributable to a large difference in their responses to U. lactuca and S. latissima. Those two algae alone accounted for over 70% of the dissimilarity between the two groups (Table 2).

Table 2 Similarity percentage (SIMPER) results for feeding of Ampithoe lacertosa and Peramphithoe humeralis
Full size table

Amphipod habitat choice

When given a choice of tube-building materials, A. lacertosa showed strong preference among the test algae (G = 48.57, P < 0.01; Fig. 2). Most tubes were built in U. lactuca. The significantly negative z scores for M. splendens, A. marginata, D. ligulata, and S. latissima indicate that A. lacertosa avoided building nests in those algae. In contrast, P. humeralis preferred A. marginata, D. ligulata, and S. latissima while avoiding U. lactuca and M. splendens (G = 33.86, P < 0.01; Fig. 2). The non-significant z score for F. distichus edentatus indicated that P. humeralis neither sought out nor avoided that alga (Fig. 2).

Fig. 2
figure 2

Tube building preferences of A. lacertosa and P. humeralis (n = 110 each species). Data points are standardized residuals (z scores) derived from log-likelihood analysis. Dashed lines indicate z = 1.96, the statistical rejection border for α = 0.05. Positive standardized residuals indicate selection for an alga; negative values indicate selection against an alga. Peramphithoe humeralis showed no significant preference for or against F. distichus edentatus. All other standardized residuals were significant for both amphipod species

Full size image

Ampithoe lacertosa had a low probability of remaining in its tubes after 24 h (occupancy rates were ≤50% in 4 of the 6 algal species; Online Resource 3 in “Supplementary material”). In contrast, P. humeralis usually remained in tubes it had built, except those on U. lactuca. A χ2 test for independence indicated that A. lacertosa and P. humeralis responded differently to the test algae (χ2 = 51.2, P < 0.001). Cell χ2 values showed that most of the significance was due to U. lactuca. Ampithoe lacertosa was unusually persistent on this species; the opposite was true for P. humeralis (Online Resource 3 in “Supplementary material”).

Survivorship of juvenile amphipods

Starved juvenile A. lacertosa did not survive beyond day 8 (Fig. 3a). A Kaplan–Meier survival analysis showed significant differences among the diet treatments (log rank χ2 = 19.5, df = 6, P = 0.003). Comparisons of means and 95% confidence intervals showed that survival was significantly lower only for starved juveniles and the A. marginata diets (Fig. 3a; Online Resource 4 in “Supplementary material”). However, the M. splendens diet was the only one to support >50% survival of juvenile A. lacertosa over the 36-day experiment. Survival to day 36 did not differ among offspring of the five mothers (χ2 = 0.36, P = 0.99).

Fig. 3
figure 3

Survival of juvenile A. lacertosa and P. humeralis starved or fed one of six single algal diets (n = 10 and n = 16, respectively). Experiments were ended after 36 days

Full size image

Survival of P. humeralis differed greatly among the algal diets (Fig 3b; Online Resource 4 in “Supplementary material”). Kaplan–Meier analysis again showed significant differences (log rank χ2 = 150.4, df = 6, P < 0.001). Starved amphipods and those feeding on M. splendens died quickest, followed by those feeding on U. lactuca (no survival beyond day 14). Survival times were significantly longer for amphipods feeding on D. ligulata, F. distichus edentatus and A. marginata; survival of P. humeralis feeding on S. latissima exceeded 90% over the 36-day experiment (Fig. 3b). There was again no difference in survival among the five broods (χ2 = 0.24, P = 0.99).

Relationships between feeding preference, juvenile survivorship, and in situ adult abundance

The ranked feeding preferences of A. lacertosa were not significantly related to ranked juvenile survivorship (Fig. 4a) or adult abundance (Fig. 4b). There was also no statistically significant relationship between ranks of juvenile survivorship and adult abundance (Fig. 4c).

Fig. 4
figure 4

Relationships between in situ adult abundance, feeding preference, and juvenile survivorship for A. lacertosa and P. humeralis feeding on six macroalgae. Amphipod responses were ranked such that a rank of one indicates most preferred food in choice trials (Fig. 1), highest juvenile mean survival time (Online Resource 4 in “Supplementary material”), or highest adult amphipod abundance (Table 1). See Online Resource 1 in “Supplementary material” for descriptions of macroalgae

Full size image

In contrast, P. humeralis feeding preferences were positively related to juvenile survivorship (r 2 = 0.889, P < 0.05; Fig. 4d). Feeding preference was not significantly related to adult abundance in the field, and adult abundance was not significantly related to juvenile survivorship (Fig. 4e, f). However, the non-significant relationships involving adult abundance of P. humeralis were driven by the response of the species to D. ligulata, which received a high rank for feeding preference and juvenile survivorship but was not occupied by adult amphipods in the field.

Algal toughness

The force required to penetrate pieces of algae varied significantly among species tested (F = 125.83, P < 0.0001; Fig. 5), largely reflecting thickness of the particular thalli. The thick-bladed brown alga S. latissima was toughest, followed by F. distichus edentatus, M. splendens, D. ligulata, and A. marginata. Significantly, less force was required to penetrate the thin blades of U. lactuca than any other algae tested.

Fig. 5
figure 5

Toughness of six algal species as measured with a penetrometer (n = 10 each species). Letters over bars show results of Tukey’s HSD pairwise comparisons. Treatments with same letter were not significantly different. SE are shown

Full size image

Measurement of algal percent nitrogen and C:N ratio

Percent nitrogen content differed significantly among the test algae (F = 71.38, P < 0.001). Ulva lactuca had the highest nitrogen concentration followed by M. splendens. Nitrogen content (% dry mass) was lower among the other species, with D. ligulata and F. distichus edentatus having the lowest nitrogen content (Fig. 6). Ratios of carbon to nitrogen also differed significantly among species (F = 72.02, P < 0.001), and the pattern was nearly opposite to that observed for nitrogen content. C:N ratios were highest in F. distichus edentatus, followed by D. ligulata, A. marginata, and S. latissima. The ratios were significantly lower for M. splendens and U. lactuca (Fig. 6).

Fig. 6
figure 6

Nitrogen content (% dry mass) and C:N ratios for six algal species used in amphipod experiments. All samples (n = 5 each species) were collected in July. Letters over bars show results of Tukey’s HSD pairwise comparisons. Treatments with same letter were not significantly different. SE are shown

Full size image

Discussion

The habitat preferences and feeding behaviors of A. lacertosa and Peramphithoe humeralis were fundamentally different in our laboratory experiments. Ampithoe lacertosa built tubes almost exclusively on the green alga U. lactuca, but fed on all algae presented to it (green, red or brown). In contrast, P. humeralis associated nearly exclusively with brown algae (especially S. latissima and A. marginata) for both habitat and food. These data support the strong phylogenetic effect proposed by Poore et al. (2008), who asserted that amphipods in the genus Peramphithoe have a more restricted diet relative to other genera of the family Ampithoidae, tending most often to associate with algae from the orders Fucales and Laminariales (e.g., Poore and Steinberg 2001).

In marine systems, small herbivores tend to be generalists; host-plant specialists are uncommon (Hay 1992; Hay and Steinberg 1992). In addition, most seaweed-associated amphipods are habitat generalists as well and quite mobile (Duffy and Hay 1991). Both of the species we studied could be considered specialists but in very different niches. The low-mobility P. humeralis creates persistent tubes on brown algae and appears unwilling to travel far between algal habitats. Indeed, P. humeralis often remains in desiccated tubes at low tide (personal observation). They appear to feed primarily on their host plant; plants with a large number of tubes often have extensive grazing damage, presumably from the amphipod inhabiting the tube. Ampithoe lacertosa, on the other hand, creates relatively temporary nest tubes that are readily abandoned when disturbed. It is likely that they supplement feeding on their host alga with algal debris that happens to drift into the U. lactuca mat (Heller 1968). This is similar to the amphipod Allochestes compressa, which lives on branching red algae but feeds on pieces of brown algae transported in by currents (Robertson and Lucas 1983). Some amphipod species may also separate host use in space and time by feeding on preferred macroalgae at night and using different species as refuge during the day (e.g., Buschmann 1990).

The difference between A. lacertosa and P. humeralis in their propensity to leave their tubes may be related to the energy they expend to create them. In the laboratory, we recorded one P. humeralis actively building a tube for more than 3 h; A. lacertosa rarely spends more than 30 min on basic tube construction (Heller 1968). The tubes of A. lacertosa may be better characterized as transient hiding places than as permanent homes.

Cruz-Rivera and Hay (2000a, b) found that amphipod mobility can strongly influence feeding behaviors. The sedentary, tube-building species they studied fed primarily on their host plants, while mobile, non-tube building species tended to mix foods to produce high-quality diets. Consistent with this suggestion, P. humeralis, which builds long-term tubes, primarily ate the same kelp on which it built tubes. In contrast, A. lacertosa, with more transient tubes, fed on a more mixed diet. It should be noted that these conclusions are based solely on laboratory studies with limited food options. Bell (1991) has warned against extrapolating such results to the field since the natural diet of the amphipods could include many other things (e.g., epiphytes).

Given the number of potential algal substrates available, there may be a variety of factors, including algal morphology, toughness, and nutritional content, influencing the choices of these two species. Hacker and Steneck (1990) found that algal morphology had an important effect on amphipod habitat choice; in that study, amphipods were much more abundant on branched or filamentous algae than they were on algae with a foliose (flat, leafy) morphology. More complex morphologies may allow the amphipods to escape detection or attack by visual predators (Stoner 1979; Nelson 1979a; Main 1987; reviewed by Hay 1991) or prevent dislodgement by waves (e.g., Sotka 2007). All of the algae we tested have a basic foliose morphology; no filamentous algae were included in the experiment as they are not particularly abundant at the study site. The major morphological difference among the algae was their size. Peramphithoe humeralis used both very large (S. latissima) and very small (F. distichus edentatus) thalli for tube building, suggesting that substrate size had little influence on habitat choice for this species. In the assays, A. lacertosa preferred the small blades of U. lactuca.

Algal toughness and longevity may be more important than morphology and size in determining habitat choice, particularly for P. humeralis. Previous work (Nelson 1979b; Stoner 1979; Holmlund et al. 1990; Duffy and Hay 1991) has shown that habitat stability (as influenced by algal longevity) can affect habitat selection by some amphipods, particularly in habitats where predation risk is high. For P. humeralis, which shows strong tube fidelity, the longevity of its algal home and the strength of the tube may be critical since the tube is a primary defense against predation (e.g., Nelson 1979a; Nicotri 1980) and physical stress (e.g., Sotka 2007). It is not surprising, therefore, that this species preferentially lives on S. latissima, the toughest, thickest alga. The contrasting habitat choice of A. lacertosa (i.e., living in a thin, fragile alga) may represent a fundamentally different lifestyle. The green color of A. lacertosa is very similar to that of U. lactuca. The transient tubes of this species, therefore, may be much more important for camouflage than for structural defense.

Value of an alga as a tube-building substrate is one aspect of habitat selection in these particular amphipod species. However, habitat choice is tightly coupled with feeding preference—particularly in P. humeralis, which creates long-term tubes. Algal characteristics that favor tube building may actually interfere with feeding. For example, toughness, while a desirable characteristic for defense, could prevent effective feeding by the amphipods. We measured the highest and lowest toughness values for S. latissima and U. lactuca, respectively; S. latissima was the preferred food of P. humeralis, while U. lactuca (which is only two cell layers thick; Scagel 1967) was the preferred food of A. lacertosa. This suggests that the mechanics of feeding may be different in these species. Interestingly, although M. splendens was one of the tougher algae tested, it was not consumed by either amphipod in large quantities. Gaines (1985) reported that M. splendens (Iridaea cordata in that study) has an outer multilaminated cuticle that discourages feeding by the isopod Idotea wosnesenskii; perhaps, this cuticle also minimized grazing by P. humeralis and A. lacertosa.

Cruz-Rivera and Hay (2000b) suggested that nutritional requirements of amphipod species can be dramatically different, even when the species are closely related. This appears to be the case with A. lacertosa and P. humeralis. Using nitrogen content as an index of nutritional value, we predicted that the amphipods should feed preferentially on U. lactuca and M. splendens if they were seeking the most nutrient-rich food (but see Cruz-Rivera and Hay 2000a, 2001 for arguments against nitrogen limitation in marine amphipods). Ampithoe lacertosa did, in fact, eat these two algae, but not in amounts that would indicate preference over the nitrogen-poor algae. It is clear from the choice feeding assay that diet mixing is important to A. lacertosa, presumably because the approach allows it to obtain the nutrients it needs (e.g., Cruz-Rivera and Hay 2000b). Peramphithoe humeralis focused on brown algae with relatively low nitrogen content. Adult P. humeralis can apparently satisfy nutritional demands from a more limited diet and there appears to be some evidence for compensatory feeding in this species (e.g., Cruz-Rivera and Hay 2000b, 2001). For instance, the feeding rate on F. distichus edentatus, a low-quality food, was 15-fold higher when amphipods were confined to a single alga diet compared to the choice assay. Thus, amphipods may substitute quantity of food for diet quality. Compensatory feeding would allow P. humeralis to satisfy nutritional demands when high-quality hosts are less available and reduce the need to move among host algae. Cruz-Rivera and Hay (2001) suggest that compensatory feeding may be a common trait among low-mobility herbivores.

The feeding choices of adult P. humeralis were closely related to the survival of juveniles raised on single-alga diets, verifying that our measured feeding preferences reflect nutritional requirements. Juvenile P. humeralis survived well on all four kelp diets but died very quickly when limited to the green (U. lactuca) or red (M. splendens) algae the adults rejected. A similarly strong link between juvenile performance and adult preference has been observed in the congeneric amphipods Peramphithoe tea (Sotka 2007) and Peramphithoe parmerong (Poore and Steinberg 1999). Conversely, juveniles of A. lacertosa did poorly on all single-alga diets and there was actually a weak negative relationship between adult feeding preference and juvenile survivorship. Again, it seems clear from these data that A. lacertosa requires a broader mixed diet.

In addition to differences in nitrogen content, the algal species we used varied widely in the secondary compounds they contain. The habitat choices of A. lacertosa and P. humeralis may be influenced by those compounds. Secondary metabolic compounds discourage grazing by a variety of marine herbivores (Steinberg 1984; Van Alstyne 1988; Hay et al. 1988a, 1994). Hay et al. (1987) hypothesized that mesograzers, such as tubicolous amphipods, should be less affected by algal chemical defenses than larger, mobile herbivores. Yet, tolerance of these compounds may vary dramatically among populations, species, and genera (e.g., Duffy and Hay 1994; Sotka and Hay 2002; Poore et al. 2008). To date, no work has been done on the effects of the chemicals in Online Resource 1 in “Supplementary material” on local amphipod species of the San Juan Archipelago. It will be necessary to gather additional information before it is clear whether algal secondary metabolites play any role in habitat selection or feeding of A. lacertosa or P. humeralis.

Based on our laboratory experiments, we predicted that the four brown algae we tested (S. latissima, A. marginata, F. distichus edentatus, and D. ligulata) are acceptable habitat for P. humeralis. They fed readily on them and accepted them for tube building sites. This held true in field sampling except that P. humeralis was never found on D. ligulata. It is unclear why P. humeralis in the field avoided an alga that was readily accepted, and even preferred, in laboratory trials. It may have been related to a series of low tides during our field sampling that damaged the D. ligulata causing them to leach H2SO4 and rendering the habitat unsuitable for the amphipods. Subsequent observations during July–September of 2007 and 2008 suggest that the abundance of P. humeralis in D. ligulata is highly variable (personal observation); more extensive field sampling across diel cycles or seasons may elucidate the pattern more fully. Consistent with the reports of Heller (1968), A. lacertosa lived almost exclusively on U. lactuca, preferring it to other algae for tube building, but feeding on anything that was available.

The distinct differences in food and habitat preferences of these amphipod species suggest a separation into fundamentally different niches (e.g., Croker 1967). For these small sympatric herbivores, patterns of host use in the laboratory and field reflect differences in their mobility and feeding strategy. Mixing diets requires moving among a variety of hosts, and A. lacertosa appears well adapted to such a strategy. On the other hand, more sedentary species like P. humeralis benefit from greater selectivity and the capacity for compensatory feeding, which may reduce the need to move among hosts thereby limiting exposure to predators. As a result of these differences, the species will exert entirely different impacts on the macrophyte community, and in turn will be differentially affected when preferred foods and habitats become seasonally limited.