Deducing the selective pressures which shaped a trait in an organism is a difficult if not impossible task (Williams 1996). It is often possible to determine what function or functions a trait currently serves. Current function does not however equate with evolutionary "purpose" (Williams 1996).
Rolling into a ball may serve multiple functions for individuals of the terrestrial isopod genus Armadillidium. Proposed primary roles for this ability include limitation of water loss (Sutton 1980) and predator deterrence (Schmalfuss 1984, Sutton 1980), especially deterrence against arthropod predators (Schmalfuss 1984). Armadillidium do roll into a ball when conditions are dry or when disturbed by any potential predator (personal observations).
We found no published experimental results that clarify whether rolling into a ball actually deters predators or limits loss of water. The arguments put forth in print are based on anecdotal evidence (Schmalfuss 1984, Sutton 1980). We therefore designed a number of experiments that will allow us to test both the predator deterrence hypothesis and the water loss hypothesis. This paper deals only with the proposed predator deterrence role of rolling-up for Armadillidium.
Experiments were designed and conducted during the fall of 1998 and spring of 1999. Experiment I was a prey choice experiment using both vertebrate (Ia) and invertebrate (Ib) predators. Experiment II was a palatability test using only invertebrate predators. Animals were only used once.
Two species of vertebrate predators, Deer mice (Peromyscus maniculatus) and Short-tail shrews (Blarina brevicauda), were trapped from locations in Chautauqua County New York. Each predator (P. maniculatus: n = 5; B. brevicauda: n = 4) was kept in a plastic cage (16.5 cm x 27 cm) without food for twenty-four hours prior to testing. Vertebrate predators were given one Armadillidium and one Oniscus (a genus of non-rolling terrestrial isopods) and then were observed until they attacked and consumed one of the isopods. Prey species and time needed to attack and consume the prey were recorded.
Invertebrate predators (Centipedes: Order Chilopoda) were obtained from a live supply company. Centipedes have been shown to consume terrestrial isopods (including Armadillidium) in previous studies (Sunderland and Sutton 1980). Each centipede (n = 36) was kept in a cage (diameter of 9 cm) without food for two weeks prior to testing and then given one Armadillidium and one Porcellio (another genus of non-rolling terrestrial isopods) which were matched for size. Each centipede and isopod was weighed prior to testing. Centipede cages were inspected at least twice per day until either an isopod was consumed or five days had elapsed. When an isopod was consumed, the centipede and remaining isopod were weighed to determine which ingested the missing isopod. Prey species and predator were recorded for each case of predation.
Experiment II allowed centipedes (n = 12) to choose between freshly sacrificed Armadillidium and Porcellio. Centipedes were kept without food for two weeks and then transferred to larger cages (16.5 cm x 27 cm) in which dead isopods had been placed at opposite ends, seven centimeters from the end walls. Each centipede was observed until it choose an isopod to consume.
Chi-square tests were performed for both the invertebrate prey choice experiment and the palatability experiment. At least ten predators were necessary for each experiment in order to ensure no bias in the Chi-square test (Zar 1999); therefore, no statistical analyses were performed on the vertebrate prey choice experiments (Experiment Ia).
Vertebrate predators (n = 9) consumed the first isopod encountered regardless of species. Time required to fully consume each isopod was below our ability to accurately record and therefore was not used for comparisons between species or prey types.
Only eleven of thirty-six centipedes in Experiment I actually consumed an isopod within the five day test period. Of those eleven, only one consumed an isopod within the first six hours of the experiment. Centipedes that consumed an isopod consistently chose Porcellio (X2 = 5.82, P = 0.0177). No isopods were consumed by other isopods in this experiment; however, five centipedes were consumed by isopods.
Centipedes (n = 12) in Experiment II chose isopods to consume within thirty minutes of the experiment's beginning. Prior to approaching an isopod carcass, centipedes exhibited escape behavior (followed the walls of the cage and attempted to scale the walls), and only occasionally crossed the middle areas of the cage. When a centipede approached an isopod carcass, it either veered away prior to contact or directly approached and began eating. Once a centipede began eating an isopod, it continued to do so until disturbed by the observer (two to three minutes later). Centipedes did not show a statistically significant preference for either genus of isopod (X2 = 0.75, P = 0.446) although they consumed Armadillidium (n = 8) twice as often as Porcellio (n = 4).
Armadillidium roll into a ball when disturbed, and this ability to roll-up could serve as a predator deterrent. Rolling into a ball did not appear to deter the small mammal predators which quickly captured, manipulated and consumed both rolling isopods (Armadillidium) and non-rolling isopods (Oniscus). Other vertebrate predators such as birds (eg. Starlings, Sturnus vulgaris) readily consume Armadillidium by placing an entire individual in their mouths (Moore 1981). Salamanders and lizards are also known to consume Armadillidium (Paris 1963). Capture and consumption (of isopods by small mammals) were only dependent upon the predator encountering an isopod.
Rolling into a ball does appear to deter arthropod predators such as centipedes. Centipedes did not avoid eating Armadillidium when presented with a dead unrolled individual, even though a dead Porcellio was also present. We believe this indicates that Armadillidium are not less palatable than Porcellio and that another explanation is needed for the disparity of predation in Experiment I. A predator deterrence mechanism is a likely explanation. We believe the difference is the ability of Armadillidium to roll into a hard-shelled ball which other isopods are incapable of forming. This finding agrees with the hypothesis put forth by Schmalfuss (1984).
Further studies are needed to ascertain whether the ability to roll into a ball deters other arthropod predators. Studies are also needed to determine whether rolling into a ball prevents water loss and whether that loss prevention is sufficient to increase survivorship. While we cannot conclude that the ability to roll-up was selected as a predator deterrent, we can conclude that it currently serves that purpose in at least a limited capacity.
Literature Cited
Paris, OH. 1963. The ecology of Armadillidium vulgare (isopoda: oniscoidea) in California grassland: Food, enemies, and weather. Ecological Monographs 33:1-22.
Schmalfuss, H. 1984. Eco-morphological strategies in terrestrial isopods. Symp. Zool. Soc. Lond. 53:49-63.
Sunderland, KD and SL Sutton. 1980. A serological study of arthropod predation on woodlice in a dune grassland ecosystem. Journal of Animal Ecology 49:987-1004.
Sutton, SL. 1980. Woodlice. Oxford: Pergamon Press.
Williams, GC. 1996. Adaptation and natural selection. Princeton, NJ: Princeton University Press.
Zar, JH. 1999. Biostatistical analysis, 4th ed. Upper Saddle River, NJ: Prentice Hall.
Moore, JK. 1981. The ecology of the acanthocephalan (Plagiorhynchus cylindraceus) in the isopod (Armadillidium vulgare) and the starling (Sturnus vulgaris). Thesis (Ph.D.) - University of New Mexico.