Nonsynchronous recovery of community characteristics in island spiders after a catastrophic hurricane

Abstract

We monitored spiders on 41 Bahamian islands for 4 years before and then 4 years after the catastrophic Hurricane Floyd passed directly over the site, inundating the study islands with its storm surge. The respective recoveries of major community properties after this annihilation were far from synchronous. Before the hurricane, the species-area relation was generally strong and the slope showed no temporal trend. After the hurricane, the slope increased from near zero (7 months later) to a value about equal to its prehurricane state. The lizard effect (difference in spider abundance or species richness between islands with and without the lizard Anolis sagrei) was generally strong before the hurricane; 7 months after, the lizard effect on abundance was weak and the effect on richness had vanished. In subsequent years, the lizard effect on abundance became strong again, but the effect on species richness remained weak. The strength of the lizard effect on both abundance and richness over the 8 years was strongly positively related to the density of lizards measured on a subset of the study islands. Twelve months after the hurricane, species richness averaged over all islands rebounded to the last prehurricane value, but abundance attained only about half that value; this finding was remarkably similar to results found in an earlier study of spiders impacted by Hurricane Lili (1996) in a different Bahamian region. Nonetheless, in the next 3 years, species richness failed to increase further, part of its long-term decline at the study site.

Fig. 1.

Temporal trends in spider abundance before and after Hurricane Floyd. Error bars are ± 1 standard error. Lizard effect (one-tailed test from ANCOVA): 1996, F_1,38 = 11.97, P = 0.0007; 1997, F_1,38 = 17.50, P = 0.0001; 1998, F_1,38 = 9.54, P = 0.0019; 1999, F_1,38 = 5.04, P = 0.0154; 2000, F_1,38 = 0.79, P = 0.1891; 2001, F_1,38 = 12.25, P = 0.0006; 2002, F_1,38 = 6.42, P = 0.0078; 2003, F_1,36 = 14.66, P = 0.0003.

Fig. 2.

Log number of spider individuals (Left) and number of spider species (Right) versus log island area for no-lizard and lizard islands in 1998, before Hurricane Floyd, when the effects of both area and lizards were strong.

Fig. 3.

Temporal trends in spider species richness before and after Hurricane Floyd. Error bars are ± 1 standard error. Lizard effect (one-tailed test from ANCOVA): 1996, F_1,38 = 3.74, P = 0.0303; 1997, F_1,38 = 6.08, P = 0.0092; 1998, F_1,38 = 3.60, P = 0.0327; 1999, F_1,38 = 0.52, P = 0.2385; 2000, F_1,38 = 0.07, P > 0.5; 2001, F_1,38 = 2.26, P = 0.0705; 2002, F_1,38 = 2.03, P = 0.0813; 2003, F_1,36 = 1.38, P = 0.1240.

Fig. 4.

Temporal trends in the spider species vs. log island area slope before and after Hurricane Floyd. Lines are fitted regressions. Error bars are ±1 standard error.

Fig. 5.

Mean numbers of individuals and species on study islands before and after Hurricane Floyd in Abaco (Upper) and Hurricane Lili in Exuma (Lower) (exposed islands). Error bars are ±1 standard error.

Fig. 6.

Relation of the magnitude of the lizard effect on spiders (difference between adjusted means on lizard and no-lizard islands, as in Fig. 3) to lizard density. (Upper) Number of individuals. (Lower) Number of species. Lizard density was measured on six islands (control islands in ref. 18) by using a multiple-capture-recapture procedure with 3 censuses (methods in ref. 19). 1999 = census after Tropical Storm Mitch; 2000 = census after Hurricane Floyd. Years 1996–1999 were before Floyd, years 2000–2003 were after.