Matching model food-web matrices, and other matrices for comparison

[The data files, together with an older version of this page, can also be downloaded as a zipped archive.]

The table below contains links to data files with food web connection matrices.

The format of the files is very simple: A matrix is represented by (1) a line with the number of rows (=number of columns) in the following matrix, and then (2) the matrix, with each row corresponding to a resource, each column corresponding to a consumer, and diagonal elements corresponding to cannibalism. Present trophic links are indicated by the character ‘X’, absent links by ‘.’. The files with model data contain 1000 consecutive matrices, the files with empirical data only one.

The empirical data was provided by Jennifer Dunne and Neo Martinez, please see  [1] for references. The data for the matching model  [1], the niche model  [2], and the nested hierarchy model  [3], was generated as described in detail in  [1]: Model parameters were adjusted to match the standardized (S, see below and  [1]) webs with the standardized empirical data. Only webs with the number of species deviating less than 30% from the empirical data are retained. The raw model food webs (R) are included since the standardization might not be reasonable for all kinds of analysis. For the same reason, the table contains webs with all disconnected components removed (C) and with trophic species lumped (L).

For each model, the data in the C, L, and S files are derived directly from the data in the R files.

Axel G. Rossberg              

Data:

Degrees of Standardization:

• R: The raw model of empirical food webs, without any data standardization applied.
• C: Only the largest connected subgraph.
• L: Like C, but with trophically identical species lumped.
• S: Full standardization: Select largest connected subgraph, lump the lowest level, and then lump trophically identical species.

References: