SEARCH

SEARCH BY CITATION

FilenameFormatSizeDescription
mee312192-sup-0001-DataS1.pdfapplication/PDF478K

Data S1. Supporting Information document containing additional methods and results.

Table S1. Properties of the six marine food webs used in this analysis.

Table S2. Group size resulting in the best score for 100 matrices generated using a known model and partition according to log-likelihood (LL), AIC, BIC, Bayes factors with hyper-priors matching the generating distribution (BF), and NML. Each matrix was generated using a group model with a different but known partition into five groups (True), with interaction probabilities drawn from a beta distribution B (a = 0·5, β = 0·5).

Table S3. Group size resulting in the best score for 100 matrices generated using a known model and partition according to log-likelihood (LL), AIC, BIC, Bayes factors with hyper-priors matching the generating distribution (BF), and NML. Each matrix was generated using a group model with a different but known partition into five groups (True), with interaction probabilities drawn from a beta distribution B (a = 0·7, β = 0·7).

Table S4. Model selection ranking (1: best to 5: worst) of empirically determined models within each model family according to AIC, BIC, Bayes factors (BF) and total length based on NML. Cascade and MPN hierarchies specified by body mass (BM) and trophic level (TL). Modular and group partitions specified by habitat (H), kingdom (K), Phylum (P), class (C) and order (O).

Figure S1. Total length for cascade, MPN, modular and group model families.

Figure S2. Total length for cascade and MPN model families when empirical data on body mass (BM) and trophic level (TL) were used to define model hierarchies.

Figure S3. Total length for modular and group model families when empirical data on habitat (H) and taxonomic information-Kingdom (K), Phylum (P), Class (C) and Order (O)-were used to define model partitions.

Figure S4. Total length for MPN model family and N2 and N3 variants formed by successively relaxing model feeding constraints.

Figure S5. Total length for Hybrid model family, which is a combination of cascade and modular model families.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.