Appendix A. Additional methods and results of percent cover and volume determinations.

METHODS

Sponge volume was estimated from osculum diameter measurements at the start and end of each three-year interval. Two measurements of osculum diameter, with the first diameter chosen as the longest possible diameter and the second perpendicular to the first, were taken from top images of sponges with UTHSCA Image Tool software. If a sponge was not photographed during a census, osculum diameter measurements were taken from pictures from adjacent field seasons and its osculum diameter was estimated by interpolation.

The volume of *X. muta* was found to be reliably predicted from the osculum diameter of the sponge. To determine this relationship, measurements of height, base diameter, and osculum diameter were taken on 104 sponges spanning a broad range of sizes present in the population. Volume was calculated using the equation for the geometric model of a frustum of a cone, which applies to the three most common morphologies of *X. muta*: cylinder, frustum of a cone, and inverse frustum of a cone. Sponge volume was then corrected for spongocoel volume by subtracting the volume of the central cavity of the sponge (McMurray et al. 2008). Because a predictive relationship was sought, ordinary least squares regression was performed on loge-transformed volume and osculum diameter data (Sokal and Rohlf 1995; Warton et al. 2006). Bias in the scaling coefficient estimate resulting from the conversion from log to arithmetic units was corrected with a logarithmic correction factor (Sprugel 1983). The equation used to calculate sponge volume from osculum diameter was determined to be: V_{sponge}=28.514 × OsculumDiameter ^{2.1} (*P* < 0.001, *R*^{2} = 0.901).

To determine percent cover of *X. muta* at the study sites, the base diameter of each sponge at each interval was estimated from osculum diameter measurements using the equation: BaseDiameter=4.834 × OsculumDiameter^{0.624} (*P* < 0.001, *R*^{2} = 0.835). This relationship was determined by the method described for the osculum diameter-volume relationship. The area of substrate covered by each sponge was estimated by solving for the area of a circle. For sponges in the Base stage, surface area was traced from top images using UTHSCA Image Tool. Percent cover data was arcsine-transformed to meet assumptions of ANOVA and repeated measures ANOVA was used to test for differences in sponge cover between sites and among time intervals (2000, 2003, 2006). Repeated measures ANOVA was similarly used to test for spatial and temporal changes in sponge volume. Because the volumes of sponges in the Base stage could not accurately be computed they were omitted from analyses. Examination of the resulting temporal and spatial patterns of volume and the number of sponges in the Base stage revealed that the exclusion of the Base stage had no effect on the outcome.

RESULTS

The mean volume of sponges at the study sites did not change through time (*F*_{2,16} = 2.699, df = 2, *P* = 0.098). Though not significant, volume increased over the 2000–2003 interval and decreased over the 2003–2006 interval for all sites on Conch Reef. At PR15, volume continually increased over the study period. The mean (± SD) volume on Conch Reef for all sites in 2000, 2003, and 2006 was 1420 ± 1242, 1529 ± 1241, and 1488 ± 1175 cm^{3}/m^{2} respectively. Individuals of the largest size class contributed most to the total volume at each site. For all sites at Conch Reef, the total volume of size class V exceeded that of size classes I-IV combined (Fig. A1). In 2000 size class V comprised 77% of the total volume of *X. muta* on Conch Reef and 75% in 2006. There was no significant difference in sponge volume between sites (*F*_{3,8} = 2.357, df = 3, *P* = 0.148). Spatial patterns of volume were similar to patterns of density and percent cover. The mean (± SD) volume of sponges at CR15, CR20, CR30, and PR15 for spring 2000 were 1266.42 ± 402.05, 2418.76 ± 1750.92, 575.97 ± 584.87, and 1304.59 ± 344.25 cm^{3}/m^{2}, respectively, compared to fall 2006 volume of 1305.90 ± 326.43, 2577.01 ± 1440.66, 581.74 ± 530.20, and 1737.11 ± 674.01 cm^{3}/m^{2}.

FIG. A1. Mean (±SD) volume of Xestospongia muta in each size class at all sites on Conch Reef. Volume of size classes I and II are < 100 cm^{3}/m^{2} |

The percent cover of sponges increased throughout the study at all sites, however this change was not significant (*F*_{1.271,10.165} = 2.390, df = 1.271, *P* = 0.150) (Fig. A2). Additionally, percent cover did not differ between sites (*F*_{3,8} = 3.684, df = 3, *P* = 0.062). Spatial patterns of sponge cover paralleled patterns of sponge density, with cover greatest at CR20 and PR15 and lowest at CR30 (Fig. A2). The mean (± SD) sponge cover at CR15, CR20, CR30, and PR15 for spring 2006 were 0.64 ± 0.18, 1.04 ± 0.37, 0.36 ± 0.30, and 0.84 ± 0.20 % respectively. The area covered by each stage on Conch Reef increased with the exception of the Base stage and size class III.

FIG. A2. Mean (±SD) percent substratum covered by Xestospongia muta at each site in 2000, 2003, and 2006. |

LITERATURE CITED

McMurray, S. E., J. E. Blum, and J. R. Pawlik. 2008. Redwood of the reef: growth and age of the giant barrel sponge *Xestospongia muta* in the Florida Keys. Marine Biology 155:159–171.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and practice of statistics in biological research. 3rd edition. W. H. Freeman and Co. New York, New York, USA.

Sprugel, D. G. 1983. Correcting for bias in log-transformed allometric equations. Ecology 64:209–210.

Warton, D. I., I. J. Wright, D. S. Falster, and M. Westoby. 2006. Bivariate line-fitting methods for allometry. Biological Reviews 81:259–291.