1
2
3 SUSPENSION FEEDING AND KLEPTOPARASITISM WITHIN THE
4 GENUS TRICHOTROPIS (GASTROPODA, CAPULIDAE)
5
6 ERIKA V. IYENGAR
7 Department of Biology, Muhlenberg College, 2400 W. Chew St., Allentown, PA 18104, USA.
8
9
10 RUNNING HEAD: SUSPENSION FEEDING & KLEPTOPARASITISM IN TRICHOTROPIS
11
12 Telephone, fax, email:
13 Correspondence: E.V. Iyengar; e-mail: [email protected]
14
15 phone: (484) 664-3731; FAX: (484) 664-3002
16 17 Suspension feeding and kleptoparasitism in Trichotropis
1 ABSTRACT
2 The marine gastropod Trichotropis cancellata is a facultative kleptoparasite, either
3 suspension feeding or parasitically stealing food from tube-dwelling polychaete worms. To
4 determine whether conclusions drawn from long-term studies in the San Juan Islands,
5 Washington, about the relative importance of suspension feeding and kleptoparasitism can be
6 applied generally to T. cancellata across its biogeographic range, I expanded earlier studies to
7 various locales in Alaska and British Columbia. Kleptoparasitism is pervasive throughout T.
8 cancellata 's range, occurring with equal frequency throughout the areas studied. The average
9 density and size of worm hosts are relatively constant across this range. Snail and worm densities
10 are not significantly correlated at the larger scale of site (averaged over nearby sampling
11 locations clustered around a city), but are correlated at the smaller local scale (within a sampling
12 location). Larger worms do not support more snails. The abundance of uninfected worm hosts is
13 usually not limiting, except potentially in some sampling locations in southwest Alaska where
14 the use of a novel host (a sea cucumber) may be the result of low densities of uninfected worms.
15 Additionally, I documented the feeding behaviors of other trichotropid species in these regions.
16 Trichotropis conica is the second confirmed kleptoparasite within the genus Trichotropis, with
17 kleptoparasitism as frequent in this species as in T. cancellata. Like T. cancellata, all sizes
18 observed of T. conica are kleptoparasites. On the other hand, Trichotropis insignis is an obligate
19 suspension feeder. Further studies are needed to determine exactly how many times this
20 behavior has arisen and how many times species have evolutionarily reverted and lost the
21 behavior, but kleptoparasitism evolved multiple times within this clade.
22
23 Keywords: feeding strategies, biogeographic studies, marine snails, symbiosis
p. 2 Suspension feeding and kleptoparasitism in Trichotropis
1 INTRODUCTION
2 The pressure to maximize resources, including food, can drive evolutionary innovations
3 (Stearns, 1992; Halanych, 1993). Determining the relative importance of various trophic
4 strategies, such as kleptoparasitism compared with suspension feeding, can reveal some of the
5 selection pressures driving the evolution of a species or larger taxon. Suspension feeders
6 dominate sublittoral, rocky, benthic communities and are responsible for a large portion of the
7 energy flow from the pelagic to the benthic system (Gili & Coma, 1998). Some suspension
8 feeding snails are evolutionarily successful, both in terms of abundance and species longevity
9 (e.g., Turritellidae; Allmon, 1988). However, suspension feeding is largely confined to a few
10 families within the gastropods, despite evolving multiple times (Owen, 1966; Hickman, 1983)
11 and despite the dominance of this feeding mode in the closely related bivalves. Suspension
12 feeding gastropods that use their gill to capture food possess a distinctive set of ctendial and
13 mantle cavity traits (Declerck, 1995). Little is known about the factors promoting the evolution
14 of suspension feeding in snails, but investigations of facultative suspension feeders may shed
15 light on this topic.
16 The marine snail Trichotropis cancellata HINDS, 1843 can either suspension feed
17 independently or kleptoparasitically steal food from a host (Pernet & Kohn, 1998; Iyengar,
18 2002). Previous long-term studies in the San Juan Islands, Washington, USA, demonstrated that,
19 in the summer, the majority of snails in all size classes of T. cancellata are positioned at the
20 opening of worm tubes (Iyengar, 2005) and grow significantly more quickly as kleptoparasites
21 than suspension feeding snails (Pernet & Kohn, 1998; Iyengar, 2002, 2004). Therefore, this
22 feeding mode is likely important to the evolutionary ecology of this snail. However, the
23 frequency of feeding modes utilized by this species outside of the San Juan Islands is unknown,
p. 3 Suspension feeding and kleptoparasitism in Trichotropis
1 as are the feeding mechanisms utilized by other species within the genus. Whether
2 kleptoparasitism is an alternative feeding strategy that is utilized only by a few populations of T.
3 cancellata and not used by congeneric species is unknown. Such a restriction of kleptoparasitism
4 would suggest that environmental, morphological or host factors affect the relative benefit of
5 each feeding mode. The studies reported herein examined the frequency of these two feeding
6 strategies in T. cancellata throughout a significant part of its biogeographic range and compared
7 population characteristics of this snail and its hosts in different areas with those measured in the
8 San Juan Islands. I also documented the prevalence of feeding strategies used by other species in
9 the same genus and compared the importance of kleptoparasitism in these species with its
10 importance in T. cancellata.
11 As a suspension feeder, T. cancellata, and likely other suspension feeding species within
12 Trichotropis, beats cilia to draw seawater into the mantle cavity where food particles are trapped
13 in the mucus that covers the gill. The snail moves the food-ladened mucus to grooves in the base
14 of the mantle cavity, translocates the food to within close proximity of its mouth, and uses its
15 pseudoproboscis (a flexible, extensible lower lip) to collect and ingest the mucus (Yonge, 1962).
16 As a kleptoparasite, T. cancellata extends its pseudoproboscis between the feeding appendages
17 of its host, usually a tube-dwelling polychaete worm (Iyengar, 2005), and into the host’s mouth.
18 Cilia on the pseudoproboscis create a counter-current and divert food from the host's mouth to
19 the snail's mouth (Pernet & Kohn, 1998). The snail does not injure the host in any way other
20 than through nutrition deprivation.
21 Thompson (1998) emphasized that a population-level, rather than a species-level, approach
22 is needed to understand fully the dynamics of the coevolutionary process due to the potential for
23 the existence of selection mosaics among populations. Because few traits will be favorable in all
p. 4 Suspension feeding and kleptoparasitism in Trichotropis
1 populations, few will be fixed for a species (Thompson, 1998). Trophic shifts and the process of
2 the evolution of feeding behaviors necessarily involve interactions with the surrounding biotic
3 and abiotic environments, even if the animals are not necessarily involved in strict co-evolution
4 (sensu Janzen, 1980). New feeding modes are not instantaneously pervasive across the species
5 range, nor will they necessarily ever become fixed, as the benefits of the new trophic strategy
6 may vary in different environments. Suspension feeding may be profitable for snails only in
7 certain restricted microhabitats, with alternative feeding modes selectively advantageous in other
8 situations. Thus, generalizations about a species based on studies performed in only one region
9 are fraught with uncertainty, as the amount of variation within the species across its entire
10 biogeographic range is unknown.
11 I intended to determine whether conclusions drawn from long-term studies in the San Juan
12 Islands, Washington, can be applied to T. cancellata in general, or whether the importance of
13 suspension feeding versus kleptoparasitism varies across the biogeographic range of this species:
14 Bering Sea to Oregon (Abbott & Dance, 1983). Therefore, I expanded earlier studies to various
15 locales in southeast and southwest Alaska and Vancouver Island, British Columbia.
16 Additionally, I searched for other trichotropid species (and found two: Trichotropis conica
17 MOLLER, 1842 and Trichotropis insignis MIDDENDORFF, 1849) and observed the feeding
18 behaviors utilized by these species. I wanted to consider these behaviors within a phylogenetic
19 context in an attempt to investigate the number of times that kleptoparasitism has arisen within
20 this clade, ostensibly driven by inefficient suspension feeding. Unfortunately, there is
21 uncertainty concerning the phylogeny of groups closely related to Trichotropis, and only a few
22 studies beyond the present report have explicitly examined the feeding modes of trichotropids.
23 These issues, for now, preclude an accurate assessment of the number of times that
p. 5 Suspension feeding and kleptoparasitism in Trichotropis
1 kleptoparasitism has evolved and been lost within the trichotropids and their close relatives. The
2 trichotropids and capulids (the latter restricted to the genera Capulus and Hyalorisia) have
3 recently been assigned to the same family (Bouchet & Warén, 1993), within the superfamily
4 Capuloidea (Bouchet & Rocroi, 2005). Because other kleptoparasites have been found within
5 the clade Littorinimorpha (sensu Bouchet & Rocroi, 2005), a preliminary assessment of the
6 minimum number of times that kleptoparasitism has evolved within this clade is possible.
7
8 MATERIALS AND METHODS
9 Biogeographic sampling
10 My methodology was largely the same as that used in long-term studies in the San Juan
11 Islands, WA (Iyengar, 2005). However, here I used haphazardly placed quadrats that were
12 surveyed only once, rather than using regularly-spaced permanent quadrats that were sampled
13 multiple times over years. Trichotropis cancellata snails remain on the same host for extended
14 time periods (Iyengar, 2002), and snails located in close proximity to a host steal food from it
15 (Pernet & Kohn, 1998; Iyengar, personal observation). Consequently, I can identify
16 kleptoparasitic snails in one-time surveys, even if the act of kleptoparasitism is not observed. I
17 chose sample sites based on general descriptions of local topography (some rock
18 boulders/cobbles were necessary). In Alaska and British Columbia, I sampled locations a single
19 time by swimming (using SCUBA) long enough in an area to verify that Trichotropis sp.
20 occurred there. I placed quadrats (0.5m × 0.5m) haphazardly in areas that contained Trichotropis
21 sp. Within each quadrat, all tubiferous worms were identified to species if possible (to family if
22 not) and each tube diameter was recorded. I recorded the number and lengths (± 0.5 mm,
23 measured from the shell apex to the posterior tip of the siphonal canal) of all snails present on
p. 6 Suspension feeding and kleptoparasitism in Trichotropis
1 worms and other substrata. In T. cancellata, shell length is a good surrogate for body size
2 (Iyengar, 2005). Locations within the Kasitsna Bay site (in the southwest Alaska region) were
3 sampled between 13 July and 17 July 2000. Locations within the Juneau, Sitka, Ketchikan and
4 Craig sites (all in the southeast Alaska region) were sampled between 24 June 1999 and 14 July
5 1999, and locations within the Vancouver Island site (on the eastern side of the island, in the
6 British Columbia region) were sampled between 18 July 1999 and 21 July 1999 (Table 1 lists all
7 locations sampled). In the laboratory, I observed representative individuals of each snail species
8 to confirm whether they stole food from hosts. Among all sites, I found three trichotropid
9 species: T. cancellata, T. conica, and T. insignis. Trichotropis insignis (collected only at the
10 Kasitsna Bay site) was usually strongly aggregated, clustered in boulder cracks. Therefore,
11 rather than using haphazard quadrats, I actively searched for T. insignis and recorded the
12 substratum and nearby microhabitat whenever this species was found. As T. insignis showed
13 complete consistency across all individuals observed (see results), this method of sampling
14 should not affect my conclusions.
15
16 Data analysis
17 Previous work (Iyengar, 2005) investigated kleptoparasitism and suspension feeding in T.
18 cancellata at six subtidal locations around the San Juan Islands, Washington, between June 1998
19 and June 2000. I repetitively sampled the locations using SCUBA and 0.5m × 0.5m quadrats
20 along permanent transect lines (18 replicate quadrats per location) to monitor inter- and intra-
21 annual variation (see Iyengar, 2005 for a more comprehensive description of transect
22 parameters). The Alaska and British Columbia data for T. cancellata described above were
23 compared with only summer data from the San Juan Island site (June and August 1999) to
p. 7 Suspension feeding and kleptoparasitism in Trichotropis
1 control for seasonal variation in behavior. I used single-factor analysis of variances (ANOVAs)
2 to investigate latitudinal trends in the frequency of kleptoparasitism (defined as the percentage of
3 T. cancellata engaged in kleptoparasitism), snail size, snail density, prevalence of infection
4 (defined as the percentage of worms that are infected by at least one snail), Serpula columbiana
5 (the most common host) size, host worm density and intensity of infection (defined as the
6 average number of kleptoparasites per worm host) using site as the independent factor (Kasitsna
7 Bay, Juneau, Sitka, Ketchikan, Craig, Vancouver Island, or San Juan Island) with sampling
8 location (various dive areas around each site) nested within site. Pairwise comparisons using
9 Fisher's LSD tests with Bonferroni adjustments (for 6 simultaneous tests) compared each site to
10 San Juan Island (family-level alpha < 0.05). All statistics were performed using the program
11 PROC MIXED within SAS (SAS Institute Inc., Cary, NC, USA).
12 I tested for the frequency of kleptoparasitism by T. cancellata on size of this species and
13 the dependence of T. cancellata density on worm host density using model I regression on the
14 values averaged over sampling locations at a site. The relationship between T. cancellata density
15 and worm host density also was analyzed using the average densities of all quadrats at a
16 sampling location, considering each location separately. I predicted that snail density would
17 increase proportionally with host density. I also predicted that the frequency of kleptoparasitism
18 would be inversely related to snail size, as suspension feeding is potentially more difficult for
19 smaller snails due to an increase in drag within the mantle cavity (Declerck, 1995). Model I
20 regression with values averaged over sampling locations at a site was used to investigate whether
21 infection intensity increased with worm diameter. I predicted that infection intensity would be
22 positively correlated with worm diameter, as larger tube-dwelling polychaete worms have higher
23 feeding rates (Henderson & Strathmann, 2000) and so can potentially support more competing
p. 8 Suspension feeding and kleptoparasitism in Trichotropis
1 kleptoparasites as there is more food to share. Quadrats with fewer than 4 snails or fewer than
2 ten worms were excluded from statistical analyses to avoid potential impacts of low sample
3 sizes. Data from the John Brown's Beach (Sitka) and Steep Island (Vancouver Island) locations
4 were excluded from statistical analyses because the former was intertidal (all other sampling
5 locations were subtidal) and the latter was dominated by a unique host species (the sabellid
6 polychaete Eudistylia vancouveri KINBERG, 1867 rather than the serpulid polychaete Serpula
7 columbiana JOHNSON, 1901 elsewhere). All regressions were performed using P < 0.05
8 (Snedecor & Cochran, 1989).
9 I compared the frequency of kleptoparasitism and snail size between T. cancellata and T.
10 conica using one-way ANOVAs with species as the independent variable. Sampling locations
11 were intially nested within site and site used as a blocking variable, but these terms were dropped
12 from both analyses because they were not significant. Only sampling locations that contained
13 both T. cancellata and T. conica and had at least four snails of one of these species were included
14 in the analyses. These statistics were performed using the program PROC MIXED within SAS.
15 The discussion of kleptoparasitism in T. insignis is restricted to qualitative observations because
16 individuals of this species were located using directed searching rather than haphazardly placed
17 quadrats.
18
19 RESULTS
20 Trichotropis cancellata was abundant on rocky substrata in all regions studied.
21 Trichotropis conica was also present in all areas except San Juan Island, and was less common
22 on Vancouver Island. I found T. insignis only in Kasitsna Bay (southwest Alaska), but still on
23 rocky substrata.
p. 9 Suspension feeding and kleptoparasitism in Trichotropis
1
2 Characteristics of T. cancellata populations
3 Trichotropis cancellata engages in kleptoparasitism throughout the range studied (Fig.
4 1A), with the majority of snails on hosts, although there was variation among individual
5 sampling locations. Trichotropis cancellata did not differ significantly among sites in the
6 frequency of kleptoparasitism, defined as the percentage of T. cancellata engaged in
7 kleptoparasitism (Fig. 1A; df = 6, F = 2.41, P > 0.05), or in size (Fig. 1B; df = 6, F = 0.69, P >
8 0.5), or in density (Fig 1C; df = 6, F = 2.231, P > 0.05). Although not considered in the statistical
9 analyses, the frequency of kleptoparasitism was also high at the intertidal sampling location
10 (John Brown's Beach, Sitka: 91.7%) and the subtidal sampling location dominated by the
11 sabellid host Eudistylia vancouveri (Steep Island, Vancouver Island: 96.6%). Average snail size
12 at the intertidal location (12.7 mm in length) was within the range reported at other sampling
13 locations, but was larger (18.2 mm) at the sampling location dominated by E. vancouveri, a host
14 that is at least four times larger than any other known host of T. cancellata, than at the locations
15 where S. columbiana was the dominant host (range = 6.1 to 11.9 mm). Although host size does
16 not promote faster growth in kleptoparasitic snails (Iyengar, 2004), E. vancouveri or this
17 particular sampling location (with strong currents) must either elevate snail growth rates or
18 encourage longevity in snails. Trichotropis cancellata twice was observed stealing food from a
19 sea cucumber (Eupentacta quinquesemita (SELENKA, 1867)) at locations in Kasitsna Bay, which
20 is the first time this species (or any member of the phylum Echinodermata) has been reported as
21 a host of T. cancellata.
22 Due to the increased proportional amount of drag, suspension feeding should be more
23 difficult for smaller T. cancellata, and so I predicted that kleptoparasitism should be more
p. 10 Suspension feeding and kleptoparasitism in Trichotropis
1 common in these size classes. However, although the frequency of kleptoparasitism tended to
2 decrease with increasing snail size, this trend was not significant (Fig. 2; P > 0.1).
3
4 Characteristics of worm populations
5 Prevalence of infection, defined as the percentage of potential host worms that had at
6 least one kleptoparasite, varied significantly among sites (Fig. 3A; df = 6, F = 3.26, P < 0.05),
7 but only Kasitsna Bay had a significantly higher prevalence of infection than San Juan Island (P
8 < 0.05). Prevalence of infection was less than 65% at all sites except Kasitsna Bay. Intensity of
9 infection, defined as the average number of kleptoparasites per worm host, varied significantly
10 between sites (Fig 3B; df = 6, F = 4.61, P < 0.01), with Sitka and Kasitsna Bay having
11 significantly more snails per host than San Juan Island (P < 0.05), but even these sites had
12 average infection intensities of less than 2 snails per host. The higher intensities of infection at
13 Sitka and Kasitsna Bay were probably the direct consequence of a higher prevalence of infection
14 (although only the latter site, represented by a single quadrat, was significantly different from
15 San Juan Island).
16 Host worm density (Fig. 4A) and size (Fig. 4B) did not differ significantly among sites
17 (df = 6, F = 0.60, P > 0.5; df = 6, F = 1.54, P > 0.1, respectively). Average worm density and
18 diameter ranged from 11 to 21 worms per 0.25m2 and 4.09 to 4.96 mm, respectively, for all sites
19 but Kasitsna Bay, which had an average host density of 4 worms per 0.25m2 and worm diameter
20 of 3.06 mm.
21 Contrary to my predictions, infection intensity was inversely related to worm diameter (y
22 = -0.66x + 3.72, df = 6, r2 = 0.563, P < 0.05), but this relationship was highly influenced by
23 Kasitsna Bay, represented by only one quadrat. After removing these data from the analysis,
p. 11 Suspension feeding and kleptoparasitism in Trichotropis
1 infection intensity no longer was significantly dependent on worm diameter (y = -0.12x + 1.26,
2 df = 5, r2 = 0.01, P > 0.1). I conclude that in general there typically was no relationship between
3 intensity of infection and worm diameter.
4 There was no significant relationship between snail density and host worm density when
5 analyzed at the level of site (y = -0.11x + 14.43, df = 7, r2 = 0.008, P > 0.1). However, when I
6 performed the analysis at the smaller level of sampling location, snail density increased with
7 worm host density (Fig. 5; P < 0.05).
8
9 Other trichotropids
10 Like T. cancellata, T. conica is also a kleptoparasite that steals food from tube-dwelling
11 polychaetes. Trichotropis conica is probably a facultative kleptoparasite that uses suspension
12 feeding as an alternate feeding mode (as T. cancellata does), as individuals of T. conica were
13 occasionally found distant from any potential host. Additionally, the frequency of
14 kleptoparasitism in T. conica, which averaged 62.08% (SE=10.99%, n=8 sites) and ranged
15 between 20% and 100% at each site, was not significantly different (df = 1, F = 1.17, P > 0.1)
16 from that in T. cancellata, which averaged 77.12% (SE=6.93%, n=14 sites). However, growth
17 studies need to be conducted to verify that T. conica restricted to suspension feeding can obtain
18 enough food to survive before the facultative feeding mode status of this species is confirmed.
19 As with T. cancellata, I observed T. conica on both serpulid and sabellid worms. Laboratory
20 observations confirmed kleptoparasitism by Alaskan snails of both of these species. Trichotropis
21 conica and T. cancellata were not significantly different in size (df = 1, F = 0.07, P > 0.5); the
22 former was 9.56±0.87 mm (mean±1SE; n=8 sites) in length, while the latter was 9.68±1.12 mm
23 (mean±1SE; n=14 sites) in length, measured from the shell apex to the posterior tip of the
p. 12 Suspension feeding and kleptoparasitism in Trichotropis
1 siphonal canal. Although I again predicted a negative relationship would exist between snail size
2 and the frequency of kleptoparasitism in T. conica (due to the negative correlation between the
3 importance of drag and the cross-sectional area of the mantle cavity), sampling locations with
4 larger snails on average had higher rates of kleptoparasitism (df = 7, y = 9.364x - 27.43, r2 =
5 0.548, P < 0.05).
6 In contrast with T. cancellata and T. conica, I never observed T. insignis on, or even near
7 (within a pseudoproboscis length), a potential host species. Over 70 individuals of a wide range
8 of sizes (4 to 21 mm in length) were observed at five sampling locations (Table 1). Despite
9 abundant potential worm hosts in close proximity (indeed, these hosts often were parasitized
10 heavily by T. cancellata and T. conica), T. insignis usually clustered in boulder crevices,
11 sometimes stacking on top of one another. Apparently this species is sedentary for long periods.
12 Coralline algae sometimes grew around a snail, leaving a patch of clear substratum when I
13 removed the animal. At least once, a worm tube (Pseudochitinopoma sp.) grew along the rock
14 substratum and across the shell of a T. insignis. I interpreted this sedentary habit to indicate that
15 this species is a suspension feeder (rather than a mobile grazer or predator). When placed in the
16 laboratory with potential hosts, T. insignis rarely climbed onto the worm tubes and never
17 remained on the same host more than a day (indeed, usually less than a few hours). I never
18 observed a T. insignis using its pseudoproboscis to probe the tentacular feeding crown of a
19 worm, while T. cancellata and T. conica engaged in this behavior multiple times. However,
20 worm hosts apparently are oblivious to the presence of T. insignis, as they are to T. cancellata:
21 members of either of these snail species can sweep their shells through the extended feeding
22 crown of a serpulid worm and the worm will continue feeding undisturbed rather than retreating
23 into its tube.
p. 13 Suspension feeding and kleptoparasitism in Trichotropis
1
2 DISCUSSION
3 Snail demographics
4 Trichotropis cancellata and T. conica predominantly exploit tube worms as hosts and are
5 kleptoparasitic throughout their entire biogeographic ranges, regardless of water depth or which
6 tube worm species was most common. Although the percentage of the T. cancellata and T.
7 conica populations engaged in kleptoparasitism varied between sampling locations, these
8 differences disappeared at the slightly larger level of site, and thus there was no significant
9 latitudinal trend. Therefore, although further studies need to verify the occurrence of
10 kleptoparasitism in the Bering Sea, it is likely T. cancellata steals food even in the northernmost
11 part of its biogeographic range. Characteristics of T. cancellata (size, density, frequency of
12 kleptoparasitism) were consistent throughout its range, so conclusions based on studies in the
13 San Juan Islands likely can be generalized to the species as a whole. Trichotropis insignis, on
14 the other hand, is an obligate suspension feeder. Individuals of this species remained on hosts
15 less than 24 hours in the laboratory and I never found this species on worms in the field.
16 Because suspension feeding may be disproportionately difficult for smaller snails
17 (Declerck, 1995), I predicted that smaller trichotropids are more likely to feed as kleptoparasites,
18 with a shift to obligate suspension feeding later in ontogeny. However, the frequency of
19 kleptoparasitism was not inversely related to average snail size for either T. cancellata (for
20 which the relationship was in the predicted direction but not significant) or T. conica (for which
21 the relationship was significant but opposite the direction predicted). All sizes of both T.
22 cancellata and T. conica engage in kleptoparasitism. Therefore, an ontogenetic threshold, after
23 which snails shift to obligate suspension feeding, does not exist. In retrospect, the lack of
p. 14 Suspension feeding and kleptoparasitism in Trichotropis
1 correlation between snail size and likelihood of kleptoparasitism is not surprising. While
2 previous experiments (Iyengar, 2002) demonstrated that small snails gain a proportionally
3 greater advantage when they shifted from suspension feeding to kleptoparasitism, large snails
4 still gained a significant benefit from the same shift. Therefore, snails of all sizes should
5 associate with worm hosts and participate in kleptoparasitism when possible.
6 Whether there is a set size limit below which trichotropids as a clade cannot effectively
7 suspension feed, and so are obligate kleptoparasites, remains to be determined. Parries and Page
8 (2003) found that recently metamorphosed T. cancellata, ~1 mm in shell length, were obligate
9 kleptoparasites. Other work (Iyengar 2002) has shown that T. cancellata as small as 2.75 mm in
10 length can grow when restricted to suspension feeding, although much more slowly than if the
11 snails are allowed to steal food. In the present study, I found Trichotropis insignis as small as 4
12 mm in length without nearby hosts, suggesting these individuals can suspension feed effectively.
13 Future studies examining the smallest size at which each trichotropid species can gain enough
14 nutrition to promote growth through exclusively suspension feeding and studies determining
15 whether this minimum size is affected by traits of the ctenidium and mantle cavity would be of
16 interest in this clade.
17
18 Characteristics of worm populations
19 Overall, there was no latitudinal gradient in the density or size of worm hosts throughout
20 T. cancellata’s range. The use of a novel host, the sea cucumber Eupentacta quinquesemita, in
21 Kasitsna Bay may have been driven by a local paucity of worm hosts. Prevalence of
22 kleptoparasitism was 100% in all quadrats at this site; uninfected worms occurred here but were
23 rare (Iyengar, personal observation). Although host density did not differ statistically between
p. 15 Suspension feeding and kleptoparasitism in Trichotropis
1 sites, Kasitsna Bay had, at most, 36% of the host density of any other site and also tended to have
2 higher snail densities. Whether kleptoparasitism of sea cucumbers is truly patchily distributed
3 based on the density of alternative uninfected hosts, or merely the result of limited observation
4 time, is not yet clear. However, use of this novel host provides important insight as to the cues
5 snails use to identify hosts. If the cue is chemical, it must be a compound that is shared by
6 members of the Annelida and the Echinodermata.
7 The planktotrophic larvae of T. cancellata have likely evolved to use worm-derived
8 settlement cues because the adult snails are sedentary and suffer a selective disadvantage (slower
9 growth rates) if they are kept from hosts and forced to suspension feed (Iyengar, 2002). The
10 movement-limited adult snails of T. cancellata likely cannot move over a distance as far as was
11 covered by the “site” level in this study (averaged over the cluster of nearby sampling locations),
12 while the planktonic larvae develop long enough in the water column (more than five weeks;
13 Parries & Page, 2003) that water currents likely carry them across multiple sites. Parries & Page
14 (2003) demonstrated that competent T. cancellata larvae settle in water containing serpulids or
15 spirorbids. Host density and snail density were correlated at the level of individual sampling
16 locations, as predicted, but not at the larger scale of site. Thus the snails are responding only to
17 local cues, not the potential of available hosts a kilometer away. This restricted response is
18 likely adaptive, given the variability in host density among sampling locations at a site. Due to
19 microhabitat differences among sampling locations that have not yet been determined, worms
20 either settle or survive differentially, and the T. cancellata snails seem well attuned to these
21 subtle differences, implying that a tight symbiotic interaction has evolved. Perhaps it is an
22 important component of this symbiosis that the snail larvae travel large distances and potentially
23 have the opportunity to assess host availability at various sampling locations.
p. 16 Suspension feeding and kleptoparasitism in Trichotropis
1 Despite previous studies demonstrating that worm tentacle area can be positively
2 correlated with maximum clearance rates (Riisgård & Ivarsson, 1990; Mayer, 1994; Henderson
3 & Strathmann, 2000), larger hosts did not support more parasites than did smaller hosts. This
4 result mirrors studies from the San Juan Islands that indicated there was no correlation between
5 host size and intensity of infection (Iyengar, 2005), nor between host size and parasite growth
6 rate (Iyengar, 2004). Thus, T. cancellata throughout its range does not choose among hosts
7 based on size, likely because any host worm in the summer, whether large or small, can support
8 the maximum growth rate in a snail. Further, the average intensity of infection was low at all
9 sites (<2 snails per host on average). Therefore, although snails grow more slowly when sharing
10 a host with competitors (Iyengar, 2002), this is likely not a severe agent of selection, as usually
11 there will be no more than one competitor (and usually none) on any given host.
12 Overall, at the level of site, there were many uninfected hosts (see above). However,
13 some individual sampling locations in Alaska had many suspension feeding snails, usually as a
14 result of sparse hosts (personal observation). When serpulid worms were present at these
15 locations (sometimes below a sampled quadrat), they had many snails clustered on them.
16 Around San Juan Island, no sampling location had many snails but few hosts; there were always
17 more worms than snails. Further investigations should reveal whether the snails in these
18 suspension-feeding dominated populations still grow at the same slow rate as suspension feeders
19 in populations with many worm hosts or grow more quickly (perhaps due to locally higher food
20 concentrations or morphological adaptations that render suspension feeding more efficient).
21
p. 17 Suspension feeding and kleptoparasitism in Trichotropis
1 Evolutionary implications of the distribution of feeding modes
2 The fact that kleptoparasitism is prevalent throughout the geographic range of both T.
3 cancellata and T. conica suggests that kleptoparasitism is not an extremely recent evolutionary
4 innovation within this clade. If it were a new innovation, I would expect to find more variation
5 in its prevalence and some populations that were still obligate suspension feeders. Additionally,
6 the widespread distribution of kleptoparasitism within these species indicates that this feeding
7 mode provides a beneficial advantage over obligate suspension feeding across the range of
8 environments and selection pressures encountered in these studies. It is unclear whether the
9 similar traits of the populations of host worms across the areas studied are due to coevolution
10 between the snails and their hosts, evidence of certain host qualities required by the
11 kleptoparasitic trichotropids, or just the normal demographics of these tube worms, irrespective
12 of the presence of the snails. Future studies examining the population parameters of these worm
13 species in areas where kleptoparasitic trichotropids are absent are needed to distinguish among
14 these possibilities.
15 From the limited data available to date, the genus Trichotropis, within the family
16 Capulidae and the clade Littorinimorpha (sensu Bouchet & Rocroi, 2005), contains facultative
17 kleptoparasites (T. cancellata and T. conica) that steal concentrated food directly from hosts and,
18 at least in the case of T. cancellata, intercept host currents, which I term indirect
19 kleptoparasitism. Trichotropis also includes an obligate suspension feeder (T. insignis). The
20 genus Capulus includes facultative kleptoparasites that steal food directly, indirectly, or both
21 (Table 2). This distribution of feeding modes implies that kleptoparasitism may have evolved
22 once within the Capulidae (sensu Bouchet & Warén, 1993), before Trichotropis sp. and Capulus
23 sp. diverged, and T. insignis subsequently reverted and lost the kleptoparasitic habit.
p. 18 Suspension feeding and kleptoparasitism in Trichotropis
1 Alternatively, kleptoparasitism has evolved independently at least twice within the Capulidae:
2 once in an ancestral Capulus sp. and once in a common ancestor shared by T. cancellata and T.
3 conica but not by T. insignis.
4 The clade Littorinimorpha includes the superfamilies Calyptraeoidea and Vanikoroidea
5 (Bouchet & Rocroi, 2005), which contain species that participate in kleptoparasitism. All the
6 Calyptraeidae species studied to date are obligate suspension feeders, except Crepidula onyx,
7 which is believed to engage in indirect kleptoparasitism at least occasionally (Peterson 1983).
8 Members of the Hipponicidae, within the Vanikoroidea, are also believed to engage in indirect
9 kleptoparasitism (Table 2). Apparently strong selection pressures are driving multiple
10 evolutionary transitions in feeding mode within this family, but future studies of additional
11 species are needed to allow a definitive determination of the number of independent evolutionary
12 events.
13 The fact that indirect kleptoparasitism is practiced by one species of the Calyptraeidae,
14 but the rest of the species studied to date within this family are obligate suspension feeders, is
15 especially important. The hippponicids and a number of capulids also engage in indirect
16 kleptoparasitism (Table 2). The distribution of this behavior among species suggests possible
17 steps that may be involved within the clade Littorinimorpha in the evolution of direct
18 kleptoparasitism. First, a species may evolve the ability to recognize a potential host and
19 opportunistically intercept food before it is under the host’s control. Later, some species that
20 already engaged in indirect kleptoparasitism may have evolved the more intimate interaction of
21 reaching into a host to remove food already completely procured.
22 Future studies examining the feeding modes of additional trichotropids, creating a
23 detailed phylogeny of the superfamily Capuloidea and the clade Littorinimorpha (Bouchet &
p. 19 Suspension feeding and kleptoparasitism in Trichotropis
1 Rocroi, 2005), and comparing the mantle cavity and the process of suspension feeding in both
2 obligately suspension feeding and facultatively kleptoparasitic trichotropids, will be immensely
3 informative in determining the factors promoting and limiting the evolution of suspension
4 feeding and kleptoparasitism in marine snails.
5
6 ACKNOWLEDGEMENTS
7 I would like to extend tremendous thanks to Andy Mahon and Darin Waller for dive
8 assistance and to the numerous kind Alaskans, especially Aaron and Laura Baldwin, Brian
9 Mortensen, and Clay Culbert in Sitka; Rita O'Clair, Chuck O'Clair and Larry Musarra in Juneau;
10 Scott Walker in Ketchikan; and Craig Sempert in Craig. I would also like to thank the Friday
11 Harbor Laboratories and Kasitsna Bay Laboratories for research space, and Francoise Vermeylen
12 for statistical advice. Warren Allmon, Nelson Hairston, Jr., C. Drew Harvell, David Reid, and
13 two anonymous reviewers commented on the manuscript. Financial support for this project was
14 received from the PADI Foundation, Mellon Foundation, The Malacological Society of London,
15 Conchologists of America and a National Science Foundation Graduate Fellowship. This
16 research was performed in partial fulfillment of the Ph.D. requirements at Cornell University.
p. 20 Suspension feeding and kleptoparasitism in Trichotropis
1 REFERENCES
2 ABBOTT, R.T. & DANCE, S.P. 1983. Compendium of seashells. E.P. Dutton, Inc. New York.
3 ALLMON, W.D. 1988. Ecology of Recent turritelline gastropods (Prosobranchia, Turritellidae):
4 Current knowledge and paleontological implications. Palaios, 3: 259-284.
5 BOUCHET, P. & WARÉN, A. 1993. Revisions of the northeast Atlantic bathyal and abyssal
6 mesogastropoda. Bollettino Malacologico, Suppl. 3. 840 pp.
7 BOUCHET, P. & ROCROI, J.-P. 2005. Classification and nomenclator of gastropod families.
8 Malacologia, 47: 1-397.
9 DECLERCK, C. H. 1995. The evolution of suspension feeding in gastropods. Biological
10 Reviews of the Cambridge Philosophical Society, 70:549-569.
11 GILI, J.M. & COMA, R. 1998. Benthic suspension feeders: their paramount role in littoral
12 marine food webs. Trends in Ecology and Evolution, 13: 316-321.
13 HABE, T. 1964. A new capulid snail, Capulus spondylicola, from Japan. Venus, Japanese
14 Journal of Malacology, 26: 37-38.
15 HALANYCH, K.M. 1993. Suspension feeding by the lophophore-like apparatus of the
16 pterobranch hemichordate Rhabdopleura normani. Biological Bulletin, 185: 417-427.
17 HAYAMI, I. & KANIE, Y. 1980. Mode of life of a giant capulid gastropod from the Upper
18 Cretaceous of Saghalien and Japan. Palaeontology, 23: 689-698.
19 HENDERSON, S.Y. & STRATHMANN, R.R. 2000. Contrasting scaling of ciliary filter in
20 swimming larvae and sessile adults of fan worms (Annelida: Polychaeta). Invertebrate
21 Biology, 119: 58-66.
22 HICKMAN, C. 1983. Comparative morphology and ecology of free-living suspension-feeding
23 gastropods from Hong Kong. In: The malacofauna of Hong Kong and Southern China.
p. 21 Suspension feeding and kleptoparasitism in Trichotropis
1 (B. Morton & D. Dudgeon, eds). 2: 217-234. Hong Kong University Press, Hong Kong.
2 IYENGAR, E.V. 2002. Sneaky snails and wasted worms: Kleptoparasitism by Trichotropis
3 cancellata (Mollusca, Gastropoda) on Serpula columbiana (Annelida, Polychaeta).
4 Marine Ecology Progress Series, 244: 153-162.
5 IYENGAR, E.V. 2004. Host-specific performance and host use in the kleptoparasitic
6 marine snail Trichotropis cancellata. Oecologia, 138: 628-639.
7 IYENGAR, E.V. 2005. Seasonal feeding-mode changes in the marine facultative kleptoparasite
8 Trichotropis cancellata (Gastropoda, Capulidae): trade-offs between trophic strategy and
9 reproduction. Canadian Journal of Zoology, 83: 1097-1111.
10 JANZEN, D.H. 1980. When is it coevolution? Evolution, 34: 611-612.
11 MATSUKUMA, A. 1978. Fossil boreholes made by shell-boring predators or commensals. I.
12 Boreholes of capulid gastropods. Venus, Japanese Journal of Malacology, 37: 29-45.
13 MAYER, S. 1994. Particle capture in the crown of the ciliary suspension feeding polychaete
14 Sabella penicillus: videotape recordings and interpretations. Marine Biology, 119: 571-
15 582.
16 MORTON, B. & JONES, D.S. 2001. The biology of Hipponix australis (Gastropoda:
17 Hipponicidae) on Nassarius pauperatus (Nassariidae) in Princess Royal Harbour,
18 Western Australia. Journal of Molluscan Studies, 67: 247-255.
19 OKUTANI, T. 1997. Marine shell-bearing Mollusks. Tokai University Press, second edition,
20 238pp. (in Japanese.)
21 ORR, V. 1962. The drilling habit of Capulus danieli (Crosse) (Mollusca: Gastropoda). The
22 Veliger, 5: 63-67.
p. 22 Suspension feeding and kleptoparasitism in Trichotropis
1 ORTON, J.H. 1949. Notes on the feeding habit of Capulus ungaricus. Report of the Marine
2 Biological Station, Pt Erin, 61:29-30.
3 OWEN, G. 1966. Feeding. In: Physiology of mollusca. (K.M. Wilbur & C.M. Yonge, eds). 1-51.
4 Academic Press, New York.
5 PARRIES, S.C. & PAGE, L.R. 2003. Larval development and metamorphic transformation of
6 the feeding system in the kleptoparasitic marine snail Trichotropis cancellata (Mollusca,
7 Caenogastropoda). Canadian Journal of Zoology, 81: 1650-1661.
8 PERNET, B. & KOHN, A.J. 1998. Size-related obligate and facultative parasitism in the marine
9 gastropod Trichotropis cancellata. Biological Bulletin, 195: 349-356.
10 PETERSON, C.H. 1983. Interactions between two infaunal bivalves, Chione undatella
11 (Sowerby) and Protothaca staminea (Conrad), and two potential enemies, Crepidula
12 onyx Sowerby and Cancer anthonyi (Rathbun). Journal of Experimental Marine Biology
13 and Ecology, 68: 145-158.
14 RIISGÅRD, H.U. & IVARSSON, N.M. 1990. The crown-filament pump of the suspension-
15 feeding polychaete Sabella penicillus: filtration, effects of temperature, and energy cost.
16 Marine Ecology Progress Series, 62: 249-257.
17 SCHIAPARELLI, S., CATTANEO, V.R. & CHIANTORE, M. 2000. Adaptive morphology of
18 Capulus subcompressus Pelseneer, 1903(Gastropoda: Capulidae) from Terra Nova Bay,
19 Ross Sea (Antarctica). Polar Biology, 23: 11-16
20 SHARMAN, M. 1956. Note on Capulus ungaricus (L.) Journal of the Marine Biological
21 Association of the U.K., 35: 445-450.
22 SNEDECOR, G.W. & COCHRAN, W.G. 1989. Statistical methods. Iowa State University Press,
23 Ames.
p. 23 Suspension feeding and kleptoparasitism in Trichotropis
1 STEARNS, S.C. 1992. The evolution of life histories. Oxford University Press, Oxford.
2 THOMPSON, J.N. 1998. The population biology of coevolution. Researches on Population
3 Ecology, 40: 159-166.
4 THORSON, G. 1965. A neotenous dwarf-form of Capulus ungaricus (L.) (Gastropoda,
5 Prosobranchia) commensalistic on Turritella communis Risso. Ophelia, 2: 175-210.
6 YAMAHIRA, K. & YANO, F. 2000. Distribution of the bonnet limpet, Hipponix conicus
7 (Gastropoda: Hipponicidae), among host species in western Kyushu, Japan. The Veliger,
8 43: 72-77.
9 YONGE, C.M. 1962. On the biology of the mesogastropod Trichotropis cancellata Hinds, a
10 benthic indicator species. Biological Bulletin, 122: 160-181.
11
p. 24 Suspension feeding and kleptoparasitism in Trichotropis
1 Table 1. Sites sampled during the biogeographic survey. Species are listed as occurring at a 2 sampling location if even one individual of that species was found there. 3 Region Site Sampling location Species found can = T. cancellata, con = T. conica, insig = T. insignis Washington San Juan Island Shady Cove can Pt. George, Upper can Pt. George, Lower can South of Pt. George can O'Neal Island can Bell Island can British Columbia Vancouver Island 10 Mile Point (Victoria) can, con Deep Cove (Victoria) can Madrona Point (Nanaimo) can, con Row & Be Damned can (Campbell River) Steep Island** can, con (Campbell River) Southeast Alaska Craig Rancheria Island can, con (Prince of Wales) Little San Juan Baptista can, con Pt. Iphigenea can, con Ketchikan Mountain Pt Lighthouse can, con Mountain Pt Boat ramp can, con Sitka John Brown's Beach** can Magic Island can, con Mary's House can, con Juneau Sunshine Cove can Auke Bay Marine can Laboratories Bird Island can Aron Island can The Shrine of St.Theresa can, con Amalga Harbor can, con False Cove can, con Southwest Alaska Kasitsna Bay Hesketh Island can, con, insig Yukon Island can, con, insig Kasitsna Bay in front of can, con Univ. of AK labs Cronin Island can, con, insig MacDonald's Spit can, con 4 5 ** indicates that these sampling locations were excluded from statistical analyses because of unique properties 6 (interidal site or a unique dominant host) discussed further in the text 7 8
p. 25 Suspension feeding and kleptoparasitism in Trichotropis
1 Table 2. Known kleptoparasitic members of the Capulidae and other families within the clade 2 Littorinimorpha (Bouchet & Rocroi, 2005). Family Species Mode of Hosts Confirmed? References kleptoparasitism1 Capulidae Trichotropis D tube worms, Yes Pernet & Kohn cancellata holothuroids (1998); Iyengar I brachiopods Inferred (2002, 2004) Capulus D, I brachiopods, Yes Orton (1949); ungaricus bivalves, Sharman (1956); tube worms, Thorson (1965) gastropods Capulus D tube worms Inferred Schiaparelli, subcompressus Cattaneo & Chiantore (2000) Capulus danieli D bivalves Yes Orr (1962) (drill shell) Capulus D bivalves Yes Orr (1962); dilatatus (drill shell) Thorson (1965) Capulus D bivalves Yes Orr (1962); sycophanta (drill shell) Matsukuma (1978); Hayami & Kanie (1980) Capulus D bivalves Yes Habe (1964) spondylicola (notch shell) Capulus I bivalve Inferred Orr (1962) californicus Capulus I bivalve Inferred Orr (1962) sericeus Gigantocapulus D?, I? bivalve Inferred Hayami & Kanie giganteus2 (1980) Separatista D tube worm Inferred Okutani (1997) helicoides Calyptraeidae Crepidula onyx I bivalve Inferred Peterson (1983) Hipponicidae I gastropods Inferred Thorson (1965); Yamahira & Yano (2000); Morton & Jones (2001) 3 4 1D= directly steals concentrated food, I = Indirectly steals food by intercepting the feeding currents of a host 5 6 2 This is a fossil species, from the Upper Cretaceous. Its position on its host, similar to that of C. ungaricus on its 7 hosts, implies that this species was a kleptoparasite. This may be the only marine prosobranch species so far 8 identified as an obligate, rather than facultative, kleptoparasite (Hayami & Kanie, 1980). 9 10 11
p. 26 Suspension feeding and kleptoparasitism in Trichotropis
1 Figure 1. A. Mean (+ SE) frequency of kleptoparasitism (defined as the percentage of T.
2 cancellata snails on worm hosts). B. Mean length (+ SE) of T. cancellata snails. C. Mean (+ SE)
3 snail density, defined as the number of snails per 0.5 × 0.5 m quadrat. Dependent variables were
4 averaged across all quadrats within all sampling locations at a site. The numbers within the bars
5 indicate the number of quadrats included in the analysis. Over 75 snails were measured within
6 each site. For A and B, only quadrats with at least four snails were included. All quadrats were
7 included in C. Kbay = Kasitsna Bay, VI = Vancouver Island, SJI = San Juan Island. There were
8 no significant differences between any site and SJI for any of these variables (Fisher's LSD
9 comparisons, P > 0.05).
10
11 Figure 2. The relationship between T. cancellata size (length from shell apex to posterior tip of
12 siphonal canal) and the frequency of kleptoparasitism by T. cancellata was not significant (P >
13 0.1). Each plotted point is a mean value for a site (average of all the quadrats at all sampling
14 locations within a site). When each sampling location was considered separately, the
15 relationship between these two variables was still not significant (df = 126, y = -0.710x + 91.13,
16 r2 = 0.026, P > 0.05).
17
18 Figure 3. A. Mean prevalence (+ SE) of kleptoparasitism (defined as the percentage of host
19 worms parasitized) at a site. B. Mean (+ SE) intensity of infection (defined as the number of
20 snails per host). Each dependent variable was averaged over all quadrats at all sampling
21 locations. The numbers within the bars indicate the number of quadrats included in the analysis.
22 Only quadrats with at least ten worms were included. Kbay = Kasitsna Bay, VI = Vancouver
23 Island, SJI = San Juan Island. Note that there are no error bars on Kbay, as only one quadrat at
p. 27 Suspension feeding and kleptoparasitism in Trichotropis
1 that sampling location had at least ten worms. Any significant difference (P < 0.05) between a
2 site and SJI is indicated by a * over the non-SJI site.
3
4 Figure 4. A. Mean (+ SE) worm density, defined as the number of worm hosts per 0.5 × 0.5 m
5 quadrat. B. Mean (+ SE) host worm size. Dependent variables were averaged across all
6 quadrats of all sampling locations at a site. The numbers within the bars indicate the number of
7 quadrats included in the analysis. All quadrats were included in A. For B, only quadrats with at
8 least ten worms were included. Kbay = Kasitsna Bay, VI = Vancouver Island, SJI = San Juan
9 Island. Note that there are no error bars on Kbay in B, as only one quadrat at that sampling
10 location had at least ten worms. In both A and B, no site was significantly different from SJI (P >
11 0.05 for all).
12
13 Figure 5. Trichotropis cancellata density increased directly with worm host density at each
14 sampling location (P < 0.05). Each plotted point is the average across all quadrats at a sampling
15 location, and sampling locations are plotted separately.
p. 28 Figure 1
A 100
75
50
25 Frequency of
kleptoparasitism 7 12 5 4 10 7 82 0 KBay Juneau Sitka Ketchikan Craig VI SJI 15
B 10
5 (in mm)
Snail length 7 12 5 4 10 7 82 0 KBay Juneau Sitka Ketchikan Craig VI SJI
C 40
30
20
10 Snail Density (# per quadrat) 7 28 13 6 12 11 206 0 KBay Juneau Sitka Ketchikan Craig VI SJI
Site
Suspension feeding and kleptoparasitism in Trichotropis
Figure 2
y = -5.527x + 126.877 r 2 = 0.329, n = 7 100
90
80
70
60
50 Frequency of kleptoparasitism 6 7 8 9 10 11 12
Snail size (mm)
p. 30 Suspension feeding and kleptoparasitism in Trichotropis
Figure 3
A
* 100
75
50
25
1 92310494
Prevalence of kleptoparasitism 0 KBay Juneau Sitka Ketchikan Craig VI SJI Site
B 2 * * 1.5
1
0.5
Infection intensity 1 9 3 2 10 4 94 0 KBay Juneau Sitka Ketchikan Craig VI SJI
Site
p. 31 Suspension feeding and kleptoparasitism in Trichotropis
Figure 4
A 25 20 15 10 5 Worm density 7 28 13 6 12 11 206 0 KBay Juneau Sitka Ketchikan Craig VI SJI
Site
B 6 5 4 3 2
Worm size 1 1 9 3 2 10 4 94
(diameter in mm) 0 KBay Juneau Sitka Ketchikan Craig VI SJI Site
p. 32 Suspension feeding and kleptoparasitism in Trichotropis
Figure 5
y = 0.388x + 4.548 r2 = 0.169, n = 29 30
20
Snail density 10
0 0 1020304050
Worm density
p. 33