View - Kowalewski, M. - Virginia Tech

PALEONTOLOGICAL SOCIETY PAPERS, V. 8, 2002The picture is not entirely one of gloom anddoom. By examining the number of repairedinjuries in successive size classes of prey, whichin a rough way correspond to age classes, we caninfer something of the shape of the function relatingthe number of unsuccessful attacks and time. If,for instance, the frequency rises in successive sizeclasses, we can reasonably infer that the ratio ofunsuccessful to total number of attacks also does.What can we conclude about samples with nomarks of unsuccessful predation? There are twopossible interpretations. The first is that all attackson those individuals that happened to beencountered by predators were successful, makingthe skeleton an ineffective device against attack.Such an interpretation would be supported weaklyby the presence of co-occurring species with moreeffective defenses that do sustain repaired damage,and more strongly by the presence of fragmentsand the remains of potential predators. In theabsence of such evidence, the second interpretationis supported: shell-breakers were absent. It is alsopossible that some species capable of very rapidmovement or characterized by venom, aggressivebehavior, or distastefulness could lack repairedinjuries simply because their intact skeletons reflectantipredatory adaptations other than passive armor.These same remarks apply to the interpretation ofdrilling.Many paleontologists have assessedfrequencies of predation and frequencies ofunsuccessful predation in entire assemblages(Kelley and Hansen, 1993, 1996; Kelley et al.,2001; Hoffmeister and Kowalewski, 2001). Suchfrequencies are, however, difficult to interpretbiologically, because assemblages contain adiversity of adaptive types. Moreover, assemblagesdiffer in the relative abundances of species as wellas in the proportions of adaptive syndromes, so thatthey are not comparable in any meaningful way.In essence, there is no biological signal in suchassemblages until they are divided into species. Theproportions of different species or of adaptive typesmay be informative in themselves, but inferencesof successful and unsuccessful predation shouldbe made at the level of the species or the population,and comparisons among places or among timeintervals should be restricted to the same speciesor the same adaptive syndrome.There is a large literature showing that highincidences of unsuccessful predation on shellbearingmolluscs are associated with variousdefenses such as a thick-walled shell, strongexternal sculpture (or relief), microstructurespreventing the propagation of cracks perpendicularto the shell’s growing edge, and deep withdrawalof edible tissues into the shell (for a review seeVermeij, 1993). Demonstrated effectivenessagainst drilling predation is provided by a thickshell and by the presence of organic layers outsideand between mineralized shell layers (see, e.g.,Harper, 1994; Kardon, 1998; Kelley, 1989, 1992;Dietl and Alexander, 2000; Kirby, 2001). Forexample, Kardon (1998) showed that manyboreholes in corbulid bivalves stop at conchiolinlayers. Similarly, breaks made by crabs in snailshells characterized by the presences of varices(periodic thickenings) stop at the varices.Partial predation requires a slightly modifiedinterpretation. Part of the victim is eaten, but thevictim lives to repair the damage. This occurs whenthe extended siphons of bivalves are nipped by fish,when the arms of brittle stars or crinoids arenibbled, and when some of the modular units—polyps, zooids, etc.—of colonial animals areconsumed. The observed incidence of regeneration(number of parts per individual or number ofregenerating individuals compared to total numberof individuals examined) is a very conservativeestimate of both the intensity of predation and theselective, or evolutionary, importance of predation.Because regeneration is often rapid, we can recastthe measurement of regeneration as a source-sinkproblem. The source is predatory attacks that causenonlethal injury; the sink is the “removal” ofregenerating structures through completerestoration. Sublethal damage that is notregenerating should not usually be included in thecalculation of the incidence of regeneration inanimals that are no longer living, because this couldhave resulted from post-mortem damage. Theincidence of regeneration, expressed at the380