Detecting avian predation on bivalve assemblages using indirect methods.

ABSTRACT Modern ecological studies suggest that avian predation has a significant impact on intertidal mollusc-dominated communities and can potentially bias the fossil record by altering the composition and probability of preservation of shelly fauna in coastal habitats. To assess if bird predation can be detected indirectly using surficial shell assemblages, fourteen samples from an area known for bird predation were compared with an expected model derived from observational studies on bird feeding behavior in the literature. The model predicts: (1) minimal bioerosion and physical decay of mollusc shells; (2) high degree of shell fragmentation; (3) ubiquity of distinct fracture patterns; and (4) shell assemblages dominated by a few species. This model was evaluated using 3 distinct habitats of Argyle Bay (San Juan Island, WA); lagoon, bay, and gravel bar. On the gravel bar, where avian delivery of prey items is observed to be high shell assemblages displayed high fragmentation (>75% of valve missing), dominance of pristine shells, frequent presence (68%) of specimens still retaining ligaments, dominance (72%) of valve fractures that cut straight across growth lines, and nearly monospecific composition. In contrast, shells from the bay and lagoon, where avian export of prey items would be high but processing negligible, have a low proportion (30%) of highly fragmented shells, a dull to chalky appearance, lower frequency (44%) of ligament-preserving specimens, and much more variable fracture patterns. Fidelity analysis, a comparison of species diversity and abundance between the sympatric life and death assemblages, indicates low death fidelity and in all three habitats death and life assemblages differ significantly from one another. However, the gravel bar death assemblage is more similar in taxonomic composition to the life assemblages than to the other two death assemblages, suggesting that the gravel bar approximates the present day composition of the local mollusc fauna ecosystem more closely than either the bay or lagoonal death assemblages. The bar deposit, with a combination of highly fragmented but pristine shells, dominant fracture patterns, and monospecific composition, suggests that supratidal deposits resulting from bird predation can be identified using indirect methods based on damage patterns in shell assemblages. These results offer a promise for applying unique taphonomic and taxonomic signatures of shell accumulations generated by molluscivorous birds to study avian predation from historical, ecological, and paleoecological perspectives.

KEY WORDS: avian; predation; bivalvia; fossil record

INTRODUCTION

Bird predation often has a tremendous effect upon invertebrate community structure despite relatively low avian abundance in intertidal systems (Clegg 1972, Goss-Custard 1980, Baird & Milne 1981, Baird et al. 1985, Marsh 1986, Richardson & Verbeek 1986, Wotton 1997). For example, a single oystercatcher can consume over J00 bivalves in a single day (Cummings 1997), compared with invertebrates, such as a drilling gastropod, which can take hours to days to complete a kill (Boggs et al. 1984, Kabat 1990). In some intertidal environments, birds can consume about 12% of the invertebrate community biomass in a single year (Moreira 1997) and exhibit discriminatory prey selection such as removing specific size classes (Richardson & Verbeek 1987, Cadee 1989) or species in bulk depending on prey availability (e.g., Webster 1941, Heppleston 1971, Sutherland 1982, Richardson & Verbeek 1986, Cummings 1997, Moreira 1997, Rome & Ellis 2004). Bias introduced by avian predation is not limited to modern ecosystems: bird predation on invertebrates have been proposed to explain unusual fossil deposits as far back as the Oligocene (Lindberg & Kellogg 1982, Iribarne et al. 1998). Despite the potential bias introduced into the fossil record by bird predation (Cadee 1989, Cadee 1994, Cadee, 1995), the paleobiological and paleoecological literature has focused on interinvertebrate predation especially for bivalve prey (for overview see Kelley et al. 2003). A significant component of ecological interaction in marine habitats may thus be passed over in paleontological and biological studies if bird predation is disregarded.

Direct observational and experimental studies on the impact of bird predation on modern intertidal communities can nonetheless be problematic because of high avian mobility and metabolism (Heppleston 1971, Maron 1982, Wotton, 1997). In this study the plausibility of bird predation being recorded in the fossil record is evaluated using a modern assemblage by (1) quantifying breakage patterns and degree of overall alteration (taphonomic approach) to determine whether avian predation leaves a unique trace and (2) comparison of life to death bivalve assemblages to evaluate for prey selectivity. If bird predation can be detected through indirect and noninvasive means, especially in intertidal habitats, it would be possible to determine whether the death assemblage is biased, with respect to body size or species, which can then be corrected for accordingly in paleoecological and historical ecological studies. In addition, this study can test whether avian predation can be detected in modern ecosystems using a noninvasive approach, which can be useful for ecological surveys.

Neontological studies can use direct observation to investigate predation and species interaction; paleontological studies have to focus on devising indirect methods to infer interaction. Within the paleontological literature, there are numerous studies dedicated to identifying and creating models to detect predation using a variety of trace fossils such as drill holes (Carriker & Yochelson 1968, Kowalewski et al. 1998, Leighton 2001, Leighton 2002, Kowalewski 2004); repair scars (Vermeij 1987, Ebbestad & Peel 1997, Alexander & Dietl 2001); peeling (Bishop 1975, Schindel et al. 1982); various bitemarks (Alexander 1986); and even coprolites (Hantschel 1968, Kloc 1987). Whereas in most cases it is difficult to identify the predator down to species in the fossil record, it is possible to narrow the list to a few suspects based on unique patterns left on the shell (e.g., circular punctures in Carboniferous goniatites associated with the shark Symmorium) (Mapes & Hansen 1984). Diet preferences of intertidal shorebirds have been well studied through direct observation, revealing different prey-preferences, feeding, and handling methods (e.g., Drinnan 1958, Davidson 1967, Feare 1970, Feare 1971, Heppleston 1971, Baker 1974, Goss-Custard 1977, Goss-Custard 1980, Maron 1982, Richardson & Verbeek 1986, Richardson & Verbeek 1987, Burger 1988, Rome & Ellis 2004).

The patterns of predation marks on bivalve victims of avian predation as a result of different feeding methods recorded by modern observational studies may be sufficiently diagnostic to be identified in subfossil and fossil assemblages. Here an expected model of common characteristics of bird-concentrated shell accumulations is tested against shell assemblages from three distinct modern environments (lagoon, bay, and gravel bar) in a locality known for bird predation on bivalves.

METHODS

The study was conducted at Argyle Bay and Argyle Lagoon, San Juan Island, WA, USA (48[degrees]31' N, 123[degrees]00' W; Fig. 1). Argyle Bay is a small partly protected marine embayment connected to the open ocean to the south. The bay's northern end is connected at high tide to Argyle Lagoon by a 25-m long and several meter wide inlet known as Argyle Creek. Argyle Lagoon is a triangle shaped area surrounded by raised sand bars; a sand and gravel bar separates Argyle Bay and Argyle Lagoon. The gravel bar is an anthropogenic feature built in the 1930s (Willings, Jr. 1957), with the side adjacent to the bay dominated by cobbles and gravel and the side adjacent to the lagoon sand. The entire area is susceptible to diurnal tides and is partially emptied during low tides. As a result, the bay, creek, and lagoon share similar environmental characteristics, such as salinity and temperature (14[degrees]C to 15[degrees]C) (Willings, Jr. 1957, Lazo 2004).

[FIGURE 1 OMITTED]

Seven types of birds were observed in Argyle Bay over the course of the study: oystercatchers (Haematopus bachmani, [Audubon]), gulls (Larus glaucescens [Naumann] and L. occidentalis, [Audubon]), crows (Corvus caurinus [Baird]), Great Blue Herons (Ardea herodias [Linnaeus]), Greater Yellowlegs (Tringa melanoleuca [J. F. Gmelin]), kingfishers (Ceryle alcyon [Linnaeus]), and a variety of ducks. Of those 7, oystercatchers, gulls, and crows are known to rely on bivalves as a substantial part of their diet (Heppleston 1971, Maron 1982. Richardson & Verbeek 1986, Richardson & Verbeek 1987, Rome & Ellis 2004). The hunting and handling strategies of these molluscivores differ, consequently leaving distinct marks on prey shells; from gulls that drop bivalves (Barash et al. 1975, Fine & James 1976, Lewis & Sharpe 1987), to crows that hammer thin-shelled bivalves with their beaks against rocks (Clegg 1972, Richardson & Verbeek 1986) (Table 1).

Using the breakage patterns qualitatively described within the ecological literature, some general characteristics are proposed to identify a bivalve victim of dropping handling methods: (1) Minimal bioerosion and physical decay--previous studies have described the remains of avian prey to have a pristine shell interior, preserved ligament, and articulated valves (Hulscher, 1982, James & Verbeek, 1984, Cadee, 1989). These characteristics in modern environments imply that damage to the shell by physical factors involved with the decay process would be minimized; (2) High degree of fragmentation--highly fragmented shell remains are interpreted often in the literature as the result of bird predation (e.g., Clegg 1972, Meldahl & Flessa 1990). For example, Cadee (1995) estimated that 75% of shells in the Dutch Wadden Sea sediment were fragmented as a result of bird predation; (3) Common fracture patterns--Common patterns that cut across growth lines or are concave into the shell have been described in multiple cases of observed bird predation (e.g., Hulscher 1982, Lomovasky et al. 2005); and (4) Evidence of prey selection--concentration of shells that were victims of bird predation should be dominated by few species and have a limited size range compared with the general bivalve community (e.g., Richardson & Verbeek 1986, Richardson & Verbeek 1987, Rome & Ellis 2004). If bird predation cannot be detected in the Argyle Bay and Lagoon localities, the results should reveal (1) no associated pattern between degree of fragmentation and state of decay; (2) no prominent break patterns along the shell; (3) and random spatial distribution of species.

Data Collection

Sampling localities were selected based on five days of exploratory observations of bird feeding habits during low tide; this restricted the target area to the rim of the bay and the lagoon where birds were active. The majority of initial captures occurred at the intertidal rim of the bay, with numerous birds observed flying to the gravel bar or the lagoon to handle and consume prey. Fourteen localities were sampled: 7 in Argyle Bay, 3 in Argyle Lagoon, and 4 on the gravel bar (Fig. 1). The bar appeared to be the environment where evidence of bird predation would be easiest to identify based on the observations of crows dropping bivalves and the lack of a living bivalve population found on the bar to mask results. The bay is used as a baseline for comparison of taphonomic and fidelity patterns because only prey selection was observed in this area, not handling or feeding. This suggests that degradation and fragmentation patterns in the bay death assemblage would primarily be the result of physical and biological processes other than avian predation. The lagoonal environment may exhibit some products of avian predation, such as high fragmentation and common fracture pattern, because of many birds handling prey items there. However, the presence of reproducing populations in the lagoon may potentially dampen the signal of avian predation by the input of bivalves in the death assemblage that were not predated.

Each sample consisted of a 1-[m.sup.2] biological grid placed randomly within the targeted sampling areas, from which all identifiable surficial bivalve material, whole shells and fragments above a minimum size of 12 mm, were collected. Subsurface samples were taken by penetrating the sediment to a depth of 15 cm and sieving the material through a 12 mm sieve, where bivalve material was exhaustively collected; whole and fragments whether a hinge was present or not. Though the sampling depth biases against deep burrowing species, such as Mya sp, many molluscivores birds prey on bivalve species that are either epifaunal, seminfaunal, or shallow infaunal in habit, and thus the depth of 15 cm is consistent with the depth birds search for bivalve prey (e.g., Baker 1974, Hulscher 1982). Throughout the sampling process, species and abundances of live bivalves in the sampling grids were also recorded. No samples were taken from the creek area because the creek bivalve population is highly disrupted throughout the year by clammers and biological research (Willings, Jr. 1957).

Taphonomic damage for specimens was scored using the semiquantitative taphonomic scheme proposed by Davies et al. (1990). Information recorded for the specimens were: (1) taxonomic identification; (2) valve length and width; (3) valve preservation; (4) presence of fragmentation; (5) location of fragmentation (Fig. 2); (6) degree of fragmentation; (7) manner of fracture (Fig. 2); (8) edge modification; (9) external and internal attributes: surface alteration, bioerosion, encrustation. For characteristics that are judged in a rank order, such as original edge modification and external surface alteration, each rank category was given a number in ascended order that correlated with degree of change (Table 2).

[FIGURE 2 OMITTED]

Data Analysis

Each shell was assigned a degradation score, to summarize the severity of surface alteration and the degree of bioerosion and encrustation to the interior and exterior of the shell. This score refers to the overall damage state of the shell and is assumed to correlate to the amount of time the shell remained at the surface and is separate from degree of fragmentation. A score of 0 indicates a specimen that is pristine on both the interior and exterior of the shell, whereas the highest score possible is 12; a specimen whose surface is chalky and eroded and has more than 20% of its internal and external area covered by encrusters or bioerosion marks. For a bivalve specimen that has a high degradation score caused by surface alteration and degree of bioerosion and encrustation, the interpretation is that the specimen had a long residence time at the surface and exposed to physical, chemical, and biological agents. In addition, the presence/absence of repair scars and drill holes were also recorded to provide a baseline for the impact of nonavian predators. With the exception of some proposed blisters in razor clam shells (Lomovasky et al. 2005), documentation of bivalves repairing their shell after a bird attack is rare, so all repair scars are assumed conservatively to be the result of failed non-avian attacks.

Based on qualitatively described prey-handling methods in the ecological literature, the taphonomic hypothesis is that an assemblage of bivalve victims of bird predation for the Argyle Bay and Lagoon would exhibit either some or all of the following characteristics: (l) minimal bioerosion and physical decay; (2) a high degree of fragmentation; (3) a fracture pattern that is either straight across or concave into the valve; and (4) a concentration of few species in high abundances. If bird predation can be detected on the gravel bar using this model, then the distribution of these characteristics should be significantly different in comparison with the bay area where no bird was observed handling the prey. To investigate fracture patterns, the specimens were grouped into either relatively thick-shelled bivalves (Clinocardium nuttallia [Conrad], Nuttallina californica [Reeve], Protothaea staminea [Conrad], Saxidomus gigantean [Deshayes], and Tapes japonica [Deshayes]) or relatively thin shelled bivalves, (Macoma nasuta [Conrad] and Mya arenia [Linnaeus]). This distinction was made between the relative thickness to account for different prey specific methods that birds use to handle bivalve prey (Table 1) and differences in shell durability because of characteristics such as thickness (Zuschin & Stanton, Jr. 2001, Zuschin et al. 2003).

To assess the potential that bird prey-selectivity biases the taxa composition and relative abundance of bivalve death assemblages the fidelity between the life and death assemblages was assessed within and across the bay, lagoon, and bar localities. The life and death assemblages each contained whole and fragmented specimens that could be identified to the species. Two approaches were used; (1) an exploratory approach using multisample Kolmogorov-Smirnov analysis to explore the differences between life and death assemblage's species abundance distribution (ex. live lagoonal assemblage) in relation to the pooled life and death distribution of species abundance for the study site (e.g., Kowalewski & Demko 1997) and (2) Spearman rank correlation to determine if patterns in the rank of species based on abundance within one assemblage can still be observed in another (ex. comparing the lagoon life versus death assemblage) (e.g., Kidwell 2001).

RESULTS

Twelve-hundred and seven dead bivalve specimens and 251 live specimens from seven bivalve species were observed within the study area. Encrustation, repair scars, and drill holes were rare in the entire death assemblage: 35 specimens with repair scars (2.9% of death assemblage), 17 specimens with either internal or external encrustation (1.2% of death assemblage), and 11 specimens with drill holes (<1.0% of death assemblage).

Taphonoraic Patterns

The biggest contrast in taphonomic patterns occurs between the gravel bar and the bay and lagoon (Table 3, Fig. 3). The median degradation score among specimens from the bay and lagoon is 4, indicating either (a) sparse surface alteration with minimum amounts of bioerosion or (b) high surface alteration with no bioerosion. The distribution of degradation scores observed between the bay and lagoon did not differ statistically, (Fig. 3, Wilcoxon pairwise comparison, Z = 0.41, P = 0.67). In contrast, samples from the gravel bar region are primarily pristine valves with a degradation median score of 0, no surface alteration or bioerosion (Fig. 3). The lack of degradation observed in the gravel bar is significantly different from the amount observed in the other two environments (Wilcoxon pairwise comparisons: bar vs. bay, Z = 17.68, P < 0.0001; bar vs. lagoon, Z = -17.16, P < 0.0001). In the bay and lagoon, 44% of specimens with a recognizable hinge still possessed a ligament; whereas 68% of the gravel bar specimens with recognizable hinges did. A log-likelihood test of the percentages of ligaments present among the three environments is statistically significant ([chi square] = 23.713, P < 0.0001). Because the bay and the lagoon have the same percentage of specimens with ligament, the significance should be attributed to the influence of the gravel bar area (Table 3).

[FIGURE 3 OMITTED]

Fragmentation patterns also vary significantly among the three environments (Kruskal Wallis, H = 140.2, P < 0.0001; Fig. 3), though pairwise comparisons reveal that each environment is significantly different from the other two. In the bay and gravel bar, specimens with greater than 75% fragmentation dominate. However, despite that dominance, patterns differ between the two settings (Wilcoxon pairwise comparison Z = -11.20, P < 0.0001). This difference is mostly likely caused by the higher percentage of complete shells in the bay; 150 specimens (29.2% of bay sample) compared with 13 complete shells from the bar (5.1% of bar sample, Fig. 4). In the lagoon, complete specimens dominate (44.2% of the lagoon sample, Fig. 4). When looking at the degradation score based on class of fragmentation, the bay and the lagoon median degradation score fluctuates between 2 and 5 for all classes of fragmentation. Yet, the median degradation score for gravel bar specimens does not vary from pristine (0) regardless of the class of fragmentation.

[FIGURE 4 OMITTED]

The breakage pattern could be determined on 637 specimens (52.7%) of the entire death assemblage. The remaining specimens were not classified because they were either complete valves or so degraded and fragmented that a breakage pattern could not be reliably determined. Such specimens were more common in the bay than the other settings (Fig. 5). For the bay, 18% of the thin-shelled bivalves and 23% of the thick-shelled bivalves exhibited a breakage pattern straight across growth lines. In the lagoon, 26% of the thick-shelled bivalves and 34% of the thin shelled bivalves had a breakage pattern straight across the growth lines (Fig. 5). On the gravel bar 72% of the specimens collected had a break that went straight across growth lines. Despite high fragmentation, the shells on the bar were pristine enough that growth lines could be determined on very small fragments, contrary to specimens in the bay or the lagoon. The frequency of different types of breakage patterns varies significantly among the three localities (log-likelihood test [chi square] = 324.9, P < 0.0001).

Fidelity Patterns

There was no significant difference in species composition and proportional abundance between the surface and subsurface death assemblages in a locality, so the data was pooled to report death assemblage. Whereas species richness among the three environments are comparable (Table 4), the abundance of species differs notably between the assemblages within and across environments (Table 5) In Argyle Bay, the entire life and death assemblages are both dominated by Macoma (79% and 41% of bay sample respectively), whereas in the lagoon the life assemblage is dominated by Protothaca (95%) and the death assemblage is dominated by Tapes (52%) despite the scarcity of live specimens of Tapes observed during data collection. In the gravel bar, 99% of the death assemblage consisted of bivalve species with thicker-shells, dominated by Tapes and Protothaca (Table 4). Spearman rank correlation between the bay and lagoon life and death agreement with the gravel bar assemblage were weak (r < 0.5) and insignificant (P > 0.05). The correlation between death assemblages of the lagoon and gravel bar, which was higher (r = 0.53), but still statistically insignificant. Within the bay and lagoon correlation was high (r > 0.5), but only in the bay was it significant (Table 5).

Compared with their abundance distribution in pooled data, the bay and lagoon life assemblages deviate positively, whereas the death assemblages deviate negatively from the pooled distribution in the Kolmogorov-Smirnov analysis (Fig. 6). The resulting deviations indicate that the death assemblages are more similar to each other than to their counterpart life assemblages, and vice versa for the life assemblages. This pattern is likely driven by 2 bivalve species (Clinocardium and Saxidomus) that are relatively abundant in all three death assemblages, but completely absent alive in the bay, lagoon, and gravel bar assemblages, which will force the death assemblages to differ from the pooled data. Tapes is much more abundant dead than alive and thus probably also contributes to the deviation of life and death assemblages (Table 4). A life assemblage represents the present-day abundance of bivalve species for an ecosystem and is susceptible to annual fluctuations; in contrast a death assemblage is an average representation of the annual abundance distribution of an ecosystem, sometimes thousands of years of production (Kowalewski et al. 1998). The death assemblage from the bar deviates negatively similar to the bay and lagoon death assemblages. However the deviation from the pooled data is smaller and suggests that the bar assemblage is less time-averaged in respect to the life assemblages compared with either the bar or lagoon death assemblages.

[FIGURE 6 OMITTED]

DISCUSSION

In the study site, the gravel bar death assemblage best fits the features expected from avian predation, suggesting that terrestrial or supratidal deposits are promising settings to detect bird predation. Because no live bivalve populations were found on the gravel bar, all samples collected from this region had to be delivered by either physical or biological transport. Normal high tides in the study site do not cover the gravel bar completely, as suggested by the heavy grass growth and terrestrial insects living on the bar, and no recent marine flooding event has been documented for the area in the past 10 y. Over half of the bivalves from the gravel bar assemblage had greater than 75% fragmentation. As noted by previous studies, high abundance of fragmented shells in a deposit can be interpreted as the product of either (1) high water turbulence or (2) high amounts of predation in low-energy environments (Trewin & Welsh 1976, Cadge 1989, Cadge 1995 Zuschin et al. 2003). The low probability of supratidal deposition by wave or tides makes transport by a biological agent a much more likely explanation. This is further supported by the presence of significant proportion of pristine specimens and dominance of highly fragmented valves with a prevailing breakage pattern of straight fractures across growth lines (Table 3), comparable to avian predation traces described in the literature (Table 1). In addition the gravel bar remains are dominated by a single bivalve species, (Tapes japonica), which is consistent with the strong prey preference behavior that is observed often in shorebirds reported from the study area (e.g., Heppleston 1971, Sutherland 1982, Richardson & Verbeek 1986, Cummings 1997, Moreira 1997, Rome & Ellis 2004). If the gravel bar were to become fossilized it will preserve a biased record of local bivalve fauna.

The high correlation and statistical significance for the fidelity in rank order between the bay life-death assemblages agrees with the results from previous life-death fidelity studies that death assemblage can correctly represent patterns that have occurred in the live assemblages. A strong positive correlation (>0.8) and significant (Table 5), is comparable to what seen in paleontological fidelity studies (e.g., Kidwell & Flessa 1995, Kidwell 2001, Kidwell 2002, Kowalewski et al. 2003). The life-death fidelity within the lagoon is also strongly positive correlated (0.7) and comparable to the correlation in the bay, but not significant at 0.05. The low significance and correlation between the bay, lagoon, and bar localities can be because of a variety of factors: (1) time averaging of variable annual population or ecological noise (e.g., more Prototheca one year over another); (2) a preservational bias (some shells are destroyed more readily than others); or a (3) a predation-induced bias. The distribution of species between the live and death assemblages in the lagoon and gravel bar localities can also be partially explained through predation. The high abundance of Tapes in the gravel bar despite their paucity among the live specimens in the lagoon and bay environments combined with taphonomic characteristics are highly suggestive of bird predation.

If the predation model in the study site is correct, then bird predation plays an important factor in the molluscan ecology and the future fossil record of the area. The lack of repair scars and drill holes, <5% of entire dataset, imply that for the system as a whole, interinvertebrate predation was minimal in comparison number of bivalve remains present on the gravel bar. Marine paleoecology studies would thus miss this important bird-mollusc biological interaction if only repair scars and drill holes were used to evaluate for interinvertebrate predation. Modern studies based solely on observation may also underestimate the intensity of avian predation in the Argyle Bay ecology given that only five captures of predation were witnessed directly over the observation period of the study; the results of this taphonomic study suggest a significantly greater avian predation rate.

A caveat of this model is that no single characteristic is a sure indicator of avian predation, multiple characteristics are necessary. This may pose a challenging problem to paleontological studies, because the initially pristine signature of shells is unlikely to be preserve over the course of time-averaging. However, monospecific shell deposits far above the shoreline with high numbers of specimens provide indirect evidence for avian predation (e.g., >1,000 Prototheca shells were found on hillsides near marine environments (James & Verbeek 1984).

This study suggests that indirect detection of bird predation through taphonomic analysis and life-death is possible in modern settings. The bivalve remains from the anthropogenic gravel bar were not exposed to wave or tidal forces, and exhibit characteristics similar to the model proposed from the literature: (1) high fragmentation rate, <75%; (2) fragmentation dominated by concave or parallel breakage patterns that are a common result of bird handling methods; (3) pristine interior and exterior surfaces of the shell; (4) abundant ligament remains; and (5) high abundance and concentration of specimens in deposits of low species diversity. This is promising for ecological and paleontological studies in that indirect approaches can be used to include bird-mollusc interactions in ecological reconstructions and surveys.

ACKNOWLEDGMENTS

The author thank in alphabetical order, John Huntley, Karen Koy, and Rich Krause for field and mental support. Thanks also to Lindsey Leighton and Mike Kowalewski for their valuable suggestions and meticulous reading that helped in the progression of this manuscript and Susan Kidwell for final suggestions and edits. The initial work for this project was supported by Friday Harbor Labs, University of Washington as part of the 2004 Predator-Prey Interaction field course and later analysis was supported by the Department of Geosciences, Virginia Tech and completed at the Center for Forensic Malacology, Appalachian Division, paper #2.

LITERATURE CITED

Alexander, R. R. 1986. Resistance to and repair of sublethal shell breakage induced by durophages in Late Ordovician brachiopods. J. Paleontol. 60:271-285.

Alexander, R. R. & G. P. Dietl. 2001. Shell repair frequencies in New Jersey bivalves: A recent baseline for tests of escalation with tertiary Mid-Atlantic congeners. Palaios 16:354-371.

Baird, D., P. R. Evans, H. Milne & M. W. Pienkowski. 1985. Utilization by shorebirds of benthic invertebrate production in intertidal areas. Oceanography and Marine Biology: an annual review. 23:573-597.

Baird, D. & H. Milne. 1981. Energy flow in the Ythan estuary, Aberdeenshire, Scotland. Estuarine Coastal and Shelf Science. 13:455-472.

Baker, A. J. 1974. Prey-specific feeding methods of New Zealand oystercatchers. Notornis 21:219-233.

Barash, D. P., P. Donovan & R. Myrick. 1975. Clam Dropping behavior of the Glaucous-winged Gulls (Larus glaucescens). Wilson Bull. 87:60-64.

Bishop, G. A. 1975. Traces of predation. In: R. W. Frey, editor. The study of trace fossils. New York: Springer-Velag. pp. 261-281.

Boggs, C. H., J. A. Rice, J. A. Kitchell & J. F. Kitchell. 1984. Predation at a snail's pace: what's time to a gastropod? Oecologia 62:13-17.

Burger, J. 1988. Foraging Behavior in Gulls: Differences in Method, Prey and Habitat. Colonial Waterbirds. 11:9-23.

Cadee, G. C. 1989. Size-selective transport of shells by birds and its palaeoecological implications. Palaeontology 32:429-437.

Cadee, G. C. 1994. Eider, selduck, and other predators, the main producers of shell fragments in the Wadden Sea: paleoecological implications. Palaeontology 37:181-202.

Cadee, G. C. 1995. 1995. Birds as producers of shell fragments in the Wadden sea, in particular the role of the Herring Gull. Geobios. Ms 18:77-85.

Carriker, M. J. & E. L. Yochelson. 1968. Recent gastropod boreholes and Ordovician cylindrical borings. U.S. Geol. Surv. Prof. Paper. B593:1-26.

Clegg, M. 1972. Carrion Crows Feeding on Marine Molluscs and Taking Fish. Bird Study 19:249-250.

Cummings, V. J. 1997. Multiscale experimental analysis of aggregative responses of mobile predators to infaunal prey. J. Exp. Mar. Bio. Ecol. 216:211-227.

Davidson, P. E. 1967. A study of the Oystercatcher (Haematopus ostralegus L.) in relation to the fishery of cockles (Cardium edule L.) in the Burry inlet, South Wales. Fishery Investigation, Series 2.25. 1-28.

Davies, D. J., G. M. Staff, W. R. Callendar & E. N. Powell. 1990. Description of a Quantitative approach to taphonomy and taphofacies analysis: All Death things are not created equal. In: W. miller III, editor. Paleocommunity temporal dynamics: the long term development of multispecies communities, special publication (Paleontological Society). Knoxville, TN: Dept of Geosciences, University of Tennessee. pp. 328-350.

Drinnan, R. E. 1958. The winter feeding of the Oystercatcher (Haematopus ostralegus) on the edible mussel (Mytilus edulis) on the Conway Estuary, North Wales. Fishery Investigation, Series 2.22. 1-15.

Ebbestad, J. O. R. & J. S. Peel. 1997. Attempted predation and shell repair in Middle and Upper Ordovician gastropods from Sweden. J. Paleontology 7:1007-1019.

Feare, C. J. 1970. Aspects of ecology of an exposed shore population of dogwhelks Nucella lapillus (L.). Oecologia 5:1-18.

Feare, C. J. 1971. Predation of limpets and dogwhelks by oystercatchers. Bird Study 18:121-129.

Fine, K. & M. James. 1976. Gull Predation on Cockles (Clinocardium nuttalli) at Garrison Bay. Unpublished thesis. University of Washington, Friday Harbor Labs. 8 pp.

Goss-Custard, J. D. 1977. The Energetics of Prey Selection by Redshank, Tringa totanus (L.) in Relation to Prey Density. J. Anim. Ecol. 46:1-19.

Goss-Custard, J. D. 1980. Competition of food and interference among waders. Ardea 68:31-52.

Hantschel, W. 1968. Coprolites: an annoted bibliography. Geological Society of America 108:1-132.

Heppleston, P. B. 1971. The Feeding Ecology of Oystercatchers (Haematopus ostralegus L.) in Winter in Northern Scotland. J. Anim. Ecol. 40:651-672.

Hulscher, J. B. 1982. The oystercatcher Haematopus ostralegus as a predator of the bivalve Macoma balthica in the Dutch Wadden Sea. Ardea 70:89-152.

Iribarne, O., J. Valero, M. M. Martinez, L. Lucifora & S. Bachmann. 1998. Shorebird predation may explain the origin of Holocene beds of stout razor clams in live position. Mar. Ecol. Prog. Set. 167:301-306.

James, P. C. & N. A. M. Verbeek. 1984. Temporal and energetic aspects of food storage in Northwestern crows. Ardea 72:207-215. Kabat, A. R. 1990. Predatory ecology of naticid gastrioids with a review of shell boring predation. Malacologia 32:155-193.

Kelley, P., M. Kowalewski & T. A. Hansen. 2003, Predator-Prey interactions in the fossil record. New York: Kluwer Academic/ Plenum Publishers. 464 pp.

Kidwell, S. M. 2001. Preservation of species abundance in marine death assemblages. Science 294:1091-1094.

Kidwell, S. M. 2002. Time-averaged molluscan death assemblages: Pafimpsests of richness, snapshots of abundance. Geology 30:803-806.

Kidwell, S. M. & K. Flessa. 1995. The quality of the fossil record-populations, species and communities. Review of Ecology and Systematics 26:269-299.

Kloc, G. J. 1987. Coprolites containing ammonoids from the Devonian of New York. Geol. Soc. Am. Abstr. Program. 19:23.

Kowalewski, M. & T. Demko. 1997. Trace fossils and populations paleoecology: comparative analysis of size-frequency distributions derived from burrows. Lethaia 29:113-124.

Kowalewski, M., A. Dulai & F. T. Fursich. 1998. A fossil record full of holes: The Phanerozoic history of drilling predation. Geology 26:1091-1094.

Kowalewski, M., G. A. Goodfriend & K. Flessa. 1998. High resolution estimates of temporal mixing within shell beds: the evils and virtues of time-averaging. Paleobiology 24:287-394.

Kowalewski, M., M. Carroll, L. Casazza, N. S. Gupta, B. Hannisdal, A. Hendy, R. A. Krause, Jr., M. LaBarbera, D. G. Lazo, C. Messina, S. Pulchalski, T. A. Rothfus, J. Salgeback, J. Stempien, R. C. Terry & A. Tomasovych. 2003. Quantitative fidelity of brachiopod-mollusk assemblages from modem subtidal environments of San Juan Island, USA. J. of Taphonomy, v.1, 1.56 MB, www.journaltaphonomy. com [checked 01-24-05].

Kowalewski, M. 2004. Drill holes produced by the predatory gastropod Nucella lamellose (Muricidai): Pateobiological and ecological implications. J. Moll. Stud. 70:359-370.

Lazo, D. G. 2004. Bivalve taphonomy: testing the effect of live habit on shell condition of the littleneck clam Protothaca staminea. Palaios 19:451-459.

Leighton, L. R. 2001. New examples of Devonian predatory boreholes and their influences on brachiopods spines on predator success. Palaeogeogr. Palaeoclimatol. Palaeoecol. 265:71-91.

Leighton, L. R. 2002. Inferring predation intensity in the marine fossil record. Paleobiology 28:328 342.

Lindberg, D. R. & M. G. Kellogg. 1982. Bathymetric Anomalies in the Neogene Fossil Record: The Role of Diving Marine Birds. Paleobiology 8:402-407.

Lewis, M. G. & F. A. Sharpe. 1987. Birding in the San Juan Islands. Seattle, WA: The Mountaineers. 210 pp.

Lomovasky, B. J., J. L. Gutierrez & O. O. Iribarne. 2005. Identifying repaired shell damage and abnormal calcification in the stout razor clam Tagelus plebeius as a tool to investigate its ecological interactions. J. Sea Res. 54:163-175.

Mapes, R. H. & M. C. Hansen. 1984. Pennsylvanian shark-cephalopod predation: a case study. Lethaia 17:175-183.

Maron, J. L. 1982. Shell-dropping behavior of Western Gulls. Auk 99:565-569.

Marsh, C. P. 1986. Impact of avian predators on high intertidal limpet populations. J. Exp. Mar. Bio. Ecol. 104:185-201.

Meldahl, K. H. & K. W. Flessa. 1990. Taphonomic pathways and comparative biofacies and taphofacies in a recent intertidal/shallow shelf environment. Lethaia 23:43-60.

Moreira, F. 1997. The importance of shorebirds to energy fluxes in a food web of a South European Estuary. Estuar. Coast. Shelf Sci. 44:67-78.

Richardson, H. & N. A. M. Verbeek. 1986. Diet selection and optimization by northwestern crows feeding on Japanese littleneck clams. Ecology 67:1219-1226.

Richardson, H. & N. A. M. Verbeek. 1987. Diet selection by yearling northwestern crows (Corvis caurinus) feeding on Littleneck clams (Venerupis japonica). Auk 104:263-269.

Rome, M. S. & J. C. Ellis. 2004. Foraging ecology and interactions between herring gulls and great black-beaked gulls in New England. Waterbirds. 27:200-210.

Schindel, D. E., G. J. Vermeij & E. Zipser. 1982. Frequencies of repaired shell, fractures among the Pennsylvanian gastropods of north-central Texas. J. Paleontol. 56:729-740.

Sutherland, W. 1982. Spatial Variation of Cockles by Oystercatchers at Traeth Melynog, Anglesey. I. The Cockle population. J. Anim. Ecol. 51:481-489.

Trewin, N. H. & W. Welsh. 1976. Formation and composition of a graded estuarine shellbed Palaeoecology. Palaeoclimatology, Palaeogeography. 19:219-230.

Webster, J. D. 1941. Feeding habits of the black oyster-catcher. Condor 43:175-180.

Willings, Jr. 1957. A special marine environment created by Argyle Lagoon, San Juan Island. Unpublished thesis. University of Washington, Friday Harbor Labs. 14 pp.

Wotton, J. T. 1997. Estimates and tests of per capita interactions strength: diet, abundance, and impact of intertidally foraging birds. Ecol. Monogr. 67:45 64.

Vermeij, G. J. 1987. Evolution and escalation: an ecological history of life. Princeton, NJ: Princton Univ. Press. 544 pp.

Zuschin, M. & R. J. Stanton, Jr. 2001. Experimental measurement of shell strength and its taphonomic interpretation. Palaios 16:161-170.

Zuschin, M., M. Stachowitsch & R. J. Stanton, Jr. 2003. Patterns and processes of shell fragmentation in modern and ancient marine environments. Earth-Science Reviews 63:33-82.

Figure 5. Fracture patterns observed on thick and thin shells from
death assemblage, plotted separately for the bay, the lagoon and the
gravel bar.

                                   Thin-shell
                                    bivalves

                           Lagoon     Bay      Gravel Bar
                           n=439     n=513       n=255

Complete shell, no
breakage                    35%       33%
Breakage pattern
subparallel to growth
line (Pattern #1)           13%       12%
Breakage pattern
straight across growth
line (Pattern #2)           34%       18%
Concave breakage
pattern into the shell
(Pattern #3)                 9%        5%
Indeterminable breakage
pattern                      9%       32%

                                   Thin-shell
                                    bivalves

                           Lagoon     Bay      Gravel Bar
                           n=439     n=513       n=255

Complete shell, no
breakage                    47%       24%          5%
Breakage pattern
subparallel to growth
line (Pattern #1)            2%        9%          1%
Breakage pattern
straight across growth
line (Pattern #2)           26%       23%         72%
Concave breakage
pattern into the shell
(Pattern #3)                17%        5%         14%
Indeterminable breakage
pattern                      8%       39%          8%

Note: Table made from pie chart.

TABLE 1.

A literature compilation of characteristic traces left on prey shells
by molluscivorous birds.

                                                        Grappling &
           Bird                  Mollusc Prey            Handling

Oystercatchers (Haematopus)
  Black Oyster-catcher        mussels (Mytilus      Rapid blow to the
    (H. bachmani)               edulis and M.         mussel then
                                californianus)        opening mussels
                                                      through leverage.
  South Island                cockles (Chione       Hammering through
    Pied Oystercatcher          stutchburyi)          the umbo or
    (H.ostalegus finschii)                            thrusting into
                                                      the commissure.
  H. ostalegus                Limpets (Patella      Thrust at margin
                                aspera and P.         with closed bill
                                vulgata)              till limpet
                                                      knocked off rock.
  H. ostalegus                Dogwhelks             Dogwhelk turned
                                                      over and pecked
                                                      at the operculum.
Gulls (Larus)
  Herring Gulls               Cerastoderma edule,   Bivalves >3cm in
    (L. argentatus)             Mytilus edulis        length are
                                                      dropped; Bivalves
                                                      <3cm in length
                                                      are ingested
                                                      whole.
Crows (Corvus)
  Northwestern Crows          Littleneck clams      Repeated dropping
    (C. caurinus)               (Venerupis
                                japonica)
  Carrion crow (C. corone)    Mactra corallina,     Insertion of bill
                                Spisula solida,       in gape and shake
                                and Lutraria          animal free.
                                lutraria              Occasional blows
                                                      to thinner
                                                      shells.
  Carrion crow (C. corone)    mussels (M. edulis)   "Pick-axe" blows
                                                      with bill to
                                                      bivalve pressed
                                                      against a rock.

           Bird                Trace on Bivalve          Reference

Oystercatchers (Haematopus)
  Black Oyster-catcher        Fracture marks on     Webster, 1941
    (H. bachmani)               valve.
  South Island                Holes within the      Baker, 1974
    Pied Oystercatcher          umbo or fractures
    (H.ostalegus finschii)      along the
                                commissure
  H. ostalegus                Chips and fracture    Feare, 1971
                                concentrated on
                                the anterior part
                                of the shell.
  H. ostalegus                Chips and fracture    Feare, 1971
                                concentrated on
                                the anterior part
                                of the shell.
Gulls (Larus)
  Herring Gulls               High fragmentation    Cadee, 1995
    (L. argentatus)
Crows (Corvus)
  Northwestern Crows          Prey-selectivity      Richardson &
    (C. caurinus)               for clams with        Verbeek, 1987
                                lengths 24.0-43.0
                                mm
  Carrion crow (C. corone)    No noted marks;       Clegg, 1972
                                valve is smashed
                                resulting in high
                                fragmentation
  Carrion crow (C. corone)    Blows delivered to    Clegg, 1972
                                straight edge of
                                shell

TABLE 2.

Explanation of semiquantitative and rank scales for taphonomic
characteristics, based from Davies et al. 1990.

    Taphonomic
  Characteristic        Min. Max. Values            Explanation

Ligament present on   1/0 (present/absent)
  recognizable
  hinge part
Degree of                     0-4            0 = no fragmentation,
  fragmentation                                complete shell
                                             1 = <25% of shell missing
                                             2 = 25% to 50% of shell
                                               missing
                                             3 = 50% to 75% of shell
                                               missing
                                             4 = >75% of shell missing
Fracture pattern            0-3, (.)         0 = no fracture of the
                                               shell
                                             1 = fracture is parallel
                                               to growth lines
                                             2 = fracture is straight
                                               across growth lines
                                             3 = fracture in concave
                                               into the valve
                                             (.) = fracture pattern
                                               cannot be determined
Surface alteration            0-2            0 = pristine, still
                                               contains original color
                                               and luster
                                             1 = surface is dull, both
                                               color and luster
                                               diminished
                                             2 = chalky and eroded
Bioerosion coverage           0-2            0 = no bioerosion
                                             1 = <20% of valve area
                                               affected
                                             2 = >20% of valve area
                                               affected
Encrustation                  0-2            0 = no encrustation
  coverage                                   1 = <20% of valve area
                                               affected
                                             2 = >20% of valve area
                                               affected

TABLE 3.

Summary of taphonomic characteristics by environment. Percentage values
are for number of specimens within each environment.

                                     Ligaments
                                     Present on
                                    Recognizable
                                      Hinges *
          Total    Degred. Score                        Dominant
            n         Median         n       %       Class of Frag.

Lagoon     439           4          152    44.44%          0%

Bay        513           4          157    44.23%         <75%

Bar        255           0           83    68.03%         <75%

            Complete
             Shells         Frag. <75%
                                                  Most Common
           n       %        n       %            Break Pattern

Lagoon    200    45.56%    110    25.06%    Straight across    26%
                                              growth lines
Bay       150    29.24%    155    30.21%    Straight across    23%
                                              growth lines
Bar        13     5.10%    139    54.51%    Straight across    72%
                                              growth lines

TABLE 4.

Total abundance of life and death bivalves and their rank for each
assemblage for Argyle Bay, Argyle Lagoon, and the gravel bar. Rank
values are in parentheses.

                                                              Argyle
                                    Argyle Bay                Lagoon

                               Live            Dead            Live

    Bivalve Species         n      Rank     n      Rank     n      Rank

Thick-shelled
  Nuttalina californica       0     (5)       5     (7)       1     (3)
  Clinocardium nuttalia       0     (5)      22     (6)       0     (4)
  Prototheca staminea        16     (3)      73     (3)      87     (1)
  Saxidomus gigantea          0     (5)      32     (5)       0     (4)
  Tapes japonica             18     (2)      65     (4)       1     (3)
Thin-shelled
  Macoma nasuta             142     (1)     200     (1)       0     (4)
  Mya arenia                  3     (4)      79     (2)       3     (2)
Unidentifiable                0              37               0
                           Live            Dead            Live

Totals                      179             513              92

                              Argyle
                              Lagoon        Gravel Bar

                               Dead            Dead

    Bivalve Species         n      Rank     n      Rank

Thick-shelled
  Nuttalina californica      11     (5)      32     (3)
  Clinocardium nuttalia       7     (6)      15     (4)
  Prototheca staminea        75     (3)      60     (2)
  Saxidomus gigantea         32     (4)       8     (5)
  Tapes japonica            196     (1)     137     (1)
Thin-shelled
  Macoma nasuta               2     (7)       1     (7)
  Mya arenia                115     (2)       2     (6)
Unidentifiable                1               0
                           Dead            Dead

Totals                      439             255

TABLE 5.

Correlation between assemblages based on rank of bivalve species
abundance using Spearman rank correlation. [alpha] = 0.05.

              Within           Correlation to Gravel Bar
           Environment
            Live-Dead           Live             Dead

           r        p         r       p        r       p

Bay       0.81    0.02 *    -0.03    0.93    -0.46    0.29
Lagoon    0.67    0.09       0.45    0.31     0.53    0.21

           Correlation Between
             Bay and Lagoon

                    r         p

Bay       Live     0.19     0.67
Lagoon    Dead     0.10     0.81

JENNIFER A. STEMPIEN

Department of Geosciences, Virginia Tech, Blacksburg, Virginia 24060; Department of Geological Sciences, University of Colorado at Boulder, Boulder, Colorado 80309

E-mail: [e-mail]