Supplementary information
Supplementary methods
All of our experiments were conducted at the Keys Marine Laboratory on Long Key, Florida, in an isolation facility designed for marine disease research. Seawater used in the facility was sand-filtered and UV-sterilized prior to release into the sea. Our designation of lobsters as ‘healthy’ prior to their inclusion in experiments was evaluated by visual evaluation of haemolymph (blood) colour, which changes from a translucent amber colour to a milky-white cast in heavily infected individuals1. However, we could not confirm the absence of a less severe, developing infection before the lobsters’ inclusion in the experiment without killing them for histological examination, which we did at the conclusion of the experiment. Therefore, we anticipated that the prevalence of PaV1 in our controls would be similar to background levels in the field, which is in the range of 5 to 10%.
Den Selection Experiment. We placed two identical artificial dens (25 cm long x 10 cm wide) constructed of PVC and concrete within each of four 2-m diameter x 1-m deep experimental tanks with flow-through seawater, For each experimental replicate, we tethered an untested lobster (either healthy or diseased) to one den by attaching one end of a 20-cm length of monofilament to the abdomen of the lobster and the other end to the den. This permitted free movement of the tethered lobster within the den and within a few centimetres of its opening. The other den in the experimental tank was left unoccupied. Every other morning, we introduced an untethered lobster (either healthy or diseased) to the tank and after 48 hours recorded in which shelter we found the untethered lobster. All lobsters used in this experiment were juveniles (25 to 45 mm carapace length), of a size naturally susceptible to PaV1 infection1. Eighty trials were performed with all possible combinations of tethered, untethered, healthy, and visibly diseased lobsters tested in an orthogonal design (n = 12 to 29 trials/treatment). We used a 2 x 2 contingency table analysis to test the null hypothesis that den choice was random. The results of that analysis demonstrate that the pattern of den choice differed significantly (c2= 15.502, df = 1, P = 0.001) between untethered healthy and diseased lobsters. Healthy lobsters clearly avoided dens containing diseased lobsters.
Infectivity Time-Series Experiment. To determine the time-course of infectivity, we inoculated 12 juvenile lobsters with haemolymph from naturally infected individuals and four with haemolymph from healthy individuals (controls) and held them individually in 11-litre tanks with flow-through seawater. We then exposed 16 naive lobsters (acquired from the field and judged to be healthy by visual inspection) to the inoculated lobsters by placing them in the same tank for two weeks. After the two-week exposure period, the naive lobsters were removed from the exposure tanks and held in isolation to determine if they had been infected with PaV1. We repeated this procedure every two weeks on a new set of 16 naive lobsters, yielding four sets of naive lobsters each independently exposed for two weeks to inoculated lobsters at two, four, six, and eight weeks after inoculation. Four of the lobsters initially inoculated with infected haemolymph died before the final two-week exposure period, reducing the number in that treatment to eight. At the conclusion of the experiment, tissues (epidermis, gill, heart, foregut and hindgut) from all of the lobsters were examined histologically1 to determine whether they had become infected with PaV1. None of the controls was infected. The number of infections among PaV1-inoculated lobsters was analysed using a 2 x 4 contingency table analysis to test the null hypothesis that infection was independent of time (weeks after inoculation). That analysis revealed that infectivity increased significantly eight weeks after inoculation (c2 = 21.33, df = 3, P < 0.0005).
Cohabitation Time-Series Experiment. To test the onset of avoidance of infected lobsters by healthy lobsters relative to infectivity, we tethered the inoculated lobsters from the Infectivity Time-Series Experiment (described above) in den-selection trials similar to those conducted in the initial Den Selection experiment (described above). Every two weeks after inoculation, the 16 inoculated lobsters were removed from the Infectivity Time-Series Experiment and tethered to a shelter within a 2-m diameter x 1-m deep experimental tank (n=4), leaving a second shelter empty. A naive lobster was then added to the tank, and its choice of shelter was recorded after 48 hours, after which the inoculated lobster was returned to its tank in the Infectivity Time-Series Experiment. These trials were repeated until all 16 lobsters were tested during each two-week period. We analyzed these data using a three-way contingency table: factor a = individual lobster (repeated measures effect); factor b = weeks after inoculation; factor c = den choice by naive lobster). Inoculated lobsters began to die during the six-week post-inoculation treatment, reducing the number of lobsters in the six- and eight-week treatments to 10 and 4, respectively. Although 8 inoculated lobsters were alive at the beginning of the final two-week period of the Infectivity Time-Series Experiment, four died prior to testing in the Cohabitation Time-Series Experiment. Analysis of these data showed that there was a significant reduction in the number of healthy lobsters that chose to shelter with inoculated lobsters over time, as indicated by the significant three-way interaction term in the analysis (c2 = 10.854; df = 3; P = 0.013).
1. Shields, J.D. & Behringer, D.C. Dis. Aquat. Org. 59, 109 (2004).