Online Resource for:

Predators, environment and host characteristics influence the probability of infection by an invasive castrating parasite

Alyssa-Lois M. Gehman1, Jonathan H. Grabowski2,A. Randall Hughes2, David L. Kimbro2, Michael F. Piehler3, and James E. Byers1.

1Odum School of Ecology, University of Georgia, Athens, GA 30602; 2Northeastern University, Nahant, MA, 01908; 3 University of North Carolina, Institute of Marine Sciences, Morehead City, NC 28557

Corresponding Author Information:

140 E. Green St, Athens, GA 30602

Phone: (206) 251-9115

Fax: (706) 542-4819


Online Resource 1: Additional methods for the selection of predatory fish.

For Spot, Toadfish, Pigfish, and Pinfish, gut content samples were collected between 1997 and 2001 in Middle Marsh, North Carolina on intertidal oyster reefs and mud flats (see (Grabowski et al. 2005) for more details on how fish were collected). For Hardhead catfish, Gaftopsail catfish, Black drum, and Red drum, sampling was conducted with similar methods to Grabowski et al. (2005) but in 2010-2011 on the reefs sampled in this study. For Grey snapper, Franks and Vanderkooy (2000) found thatEurypanopues depressusaccounts for ~10% of the diet of grey snapper caught in eastern Mississippi Sound. We found an average of 7 predatory fish per trap and gill net and a maximum of 36 fish per trap and gill net (Table 1), and we found multiple individuals of each of the species included.

Xanthid crab eaters / Common name / Life stage of xanthid prey / Reference
Arius felis / Hardhead catfish / Adults / Kimbro et al (unpub data)
Bagre marinus / Gaftopsail catfish / Adults / Kimbro et al (unpub data)
Pogonias cromis / Black drum / Adults / Kimbro et al (unpub data)
Sciaenops ocellatus / Red drum / Adults / Kimbro et al (unpub data)
Opsanus tau / Toadfish / Adults / Grabowski (unpub data)
Leiostomus xanthurus / Spot / Juveniles / Grabowski (unpub data)
Orthopristis chrysoptera / Pigfish / Juveniles / Grabowski (unpub data)
Lagodon rhomboides / Pinfish / Juveniles / Grabowski (unpub data)
Lutjanus griseus / Grey snapper / Juveniles / Franks and VanderKooy (2000)

Franks, J.S. and K.E. VanderKooy, 2000. Feeding habits of juvenile lane snapperLutjanus synagrisfrom Mississippi coastal waters, with comments on the diet of gray snapperLutjanus griseus. Gulf Caribb. Res. 12:11-17 10.18785/gcr.1201.02

Grabowski, J. H., A. R. Hughes, D. L. Kimbro, and M. A. Dolan. 2005. How habitat setting influences restored oyster reef communities. Ecology 86:1926-1935 10.1890/04-0690


Online resource 2: Correlation matrix of predicted variables included in the full model analysis.

Pearson correlation matrix of all variables collected in a survey of 50 reefs from 10 estuaries from North Carolina to Florida in August 2010 and included as predictor variables in a model evaluating infection probability of Loxothylacus panopaei infection in Eurypanopeus depressus. The lower left panels illustrate the raw data correlations, and the upper right panels are the Pearson correlation coefficients.


Online Resource 3: Random variance and model fits with different random variance structures.

Comparison of variance attributed to the random variables and the effect of the inclusion of reef as a random variable on model fit. The model fits illustrated include Akaike’s Information Criteria (AICc) and the amount of variance accounted for in the random term (the difference between the Marginal and Conditional R2 in Online Resource 4 and Table 2).

Random variable / Variance (top) / Variance (full) / AICc (top) / AICc (full) / Random variable
R2 (top) / Random variable R2 (full)
Estuary / 0.43 / 0.38 / 1108.7 / 1114.7 / 0.09 / 0.08
Estuary +
Estuary(reef) / 0.43 / 0.38 / 1110.7 / 1116.7 / 0.09 / 0.08
0.00 / 0.01
Reef / 0.45 / 0.44 / 1121.8 / 1127.1 / 0.08 / 0.08

Online Resource 4: Model analysis with reef as a nested random variable.

The candidate model set (ΔAICc<2) for infection status on 50 reefs from 10 estuaries from North Carolina to Florida with both reef and estuary as random variables, with reef nested within estuary. Standardized β-coefficients were reported for predictor variables included in each model. Predictor variables that were significant in the model are bold. Relative variable importance (RVI) was calculated for each predictor variable, scaled from 0-1. A null model with estuary as a random variable was included. To account for model selection uncertainty model averaging was conducted on the candidate set models (A-E). Odds ratios (OR) were reported for the component variables of the best-fitting model (A) according to AICc criteria. All variables are standardized, so the effect of each variable is comparable between predictor variables. Odds ratios are associated with a single standard deviation change in the predictor variable.

Variance / β Coefficients / Pseudo R2
Model / Estuary / Estuary(Reef) / Intercept / Host size / Water depth / Fish Predators' / Host density / C. sapidus / Other Fish / P. herbstii / C. virginica / Vertical relief / AICc / Δ AICc / Weight / Marginal / Conditional
A / 0.43 / 0.00 / -1.3 / 1.08 / 0.72 / 0.35 / 0.21 / 0.29 / 1110.70 / 0.00 / 0.30 / 0.26 / 0.35
B / 0.45 / 0.08 / -1.4 / 1.09 / 0.75 / 0.35 / 0.25 / 1111.43 / 0.72 / 0.21 / 0.26 / 0.37
C / 0.40 / 0.02 / -1.3 / 1.08 / 0.25 / 0.33 / 0.23 / 0.25 / 0.12 / 1111.73 / 1.03 / 0.18 / 0.26 / 0.35
D / 0.05 / 1.5x10-6 / -1.3 / 1.09 / 0.73 / 0.35 / 0.22 / 0.29 / -0.08 / 1112.14 / 1.44 / 0.15 / 0.26 / 0.35
E / 0.04 / 1.4x10-9 / -1.3 / 1.08 / 0.71 / 0.33 / 0.21 / 0.29 / 0.06 / 1112.26 / 1.56 / 0.14 / 0.26 / 0.35
Full / 0.39 / 6.0x10-6 / -1.3 / 1.08 / 0.66 / 0.30 / 0.24 / 0.25 / 0.13 / 0.09 / 0.01 / -0.04 / 1116.71 / 6.01 / 0.02 / 0.27 / 0.35
Null / 1292.47 / 181.77 / 0.00 / 0.00 / 0.12
Average / -1.3 / 1.09 / 0.72 / 0.35 / 0.17 / 0.27 / 0.02 / 0.01 / 0.0003 / -0.01
RVI / 1.00 / 0.83 / 0.69 / 0.55 / 0.58 / 0.46 / 0.41 / 0.31 / 0.30
OR / 2.95 / 2.05 / 1.42 / 1.23 / 1.34