View - Kowalewski, M. - Virginia Tech

PALEONTOLOGICAL SOCIETY PAPERS, V. 8, 2002geographic ranges. In addition, the outcome of aninteraction often varies among environments withthe varying context of the other species present. Inmodern habitats predation varies at different scales,from local to regional. Not surprisingly, then, manystudies have shown that predation pressure is alsohighly variable on both large and small spatialscales within time-averaged assemblages in thefossil record (Vermeij, 1980; Geller, 1983;Schmidt, 1989; Hansen and Kelley, 1995;Hoffmeister and Kowalewski, 2001; Nebelsick andKowalewski, 1999; Alexander and Dietl, in press).This sets up the conditions for a possible selectionmosaic of hot and cold spots that have fluctuatedin both space and time.The question that studies on spatial variabilityin predation have neglected to ask is how variabilityin the frequency of hot and cold spots affects therate of evolution of species interactions (or traitsinvolved in the interaction). While it is importantthat we evaluate temporal trends in coevolutionorescalation-related traits among ecologicallysimilar habitats (i.e., because local adaptation iscontext dependent; Vermeij, 1994), in doing so weignore the variation across other populations in theselection mosaic. The degree of connectionbetween populations may change as the frequencyof hot and cold spots (the mosaic of spatiallyvarying selection pressures) changes temporally.If the frequency of hot spots is low relative to coldspots, and the scale of gene flow exceeds the scaleover which selection varies, beneficial adaptivechanges (such as an increase in shell thickness)are unlikely to spread quickly to all populations ofthe species. Further work is needed to determinethe general applicability of the conclusion of Bushet al. (2002) that degree of variation in timeaveragedfossil assemblages mirrors withingenerationvariability of living populations; if theseresults hold, then geographic and temporal mosaicsof change should be observable in the timeaveragedfossil record.The outcomes of interactions also can varyamong environments as changes occur in theeffectiveness of traits involved in an interaction(Thompson, 1988). Thompson has called the rangein effectiveness of a particular trait an interactionnorm—by analogy with reaction norm, which isthe range of phenotypes a given genotype expressesamong environments. In order to understandselection and constraints on the evolution of speciesinteractions, patterns of geographic variation in theoutcomes of such interactions might have to beconsidered (see also Thompson, 1988). Vermeij(1982b) has shown convincingly that predatoryattacks are not always successful. It is this failurethat he argues is driving the evolution of adaptationsto enemies (Vermeij, 1982b, 1987, 2002).Geographic variation in sublethal predation thusprovides an opportunity to understand howcombinations of different traits affect the outcomesof interactions over evolutionary time (Vermeij,1982b), or favor the evolution of new associationsor innovations important to an interaction betweenspecies. How do different environments affectselection on interactions (or the interaction norm)?Thompson has argued that differences in outcomeamong environments could potentially be greatenough to shift the mean outcome along a continuumof antagonism, commensalism, and mutualism.Adaptive phenotypic plasticity.—Heterogeneityin predation pressures among habitats in ecologicaltime also favors the evolution of adaptive phenotypicplasticity. Inducible defenses are phenotypic changesinduced directly by chemical cues associated withbiotic agents. Induced morphological defensesinclude the production of antipredatory structures(bryozoans, cladocerans, rotifers) and protectivevariation in shell thickness (barnacles, gastropods,mussels) in the presence of the predator (Harvell,1984; Lively, 1986; Trussell, 1996, 2000; Leonardet al., 1999). For instance, thicker shells areinduced in the gastropod Littorina obtusata bythe presence of the predatory crab Carcinusmaenus (Trussell, 1996).Induced defenses are likely favored overconstitutive defenses because they account for theunpredictability inherent in most habitats. Inducibledefenses in the prey are thought to evolve via tworoutes: “cost-benefit” (=“inducible” in Fig. 4) and“moving target.” The cost-benefit model (Fig. 4)applies when four conditions are met: 1) Selective366