Skip to main content

Multiple dimensions of dietary diversity in large mammalian herbivores

Cite this dataset

Kartzinel, Tyler; Pringle, Robert (2020). Multiple dimensions of dietary diversity in large mammalian herbivores [Dataset]. Dryad.


  1. Theory predicts that trophic specialization (i.e., low dietary diversity) should make consumer populations sensitive to environmental disturbances, yet diagnosing specialists is complicated both by the difficulty of precisely quantifying diet composition and by definitional ambiguity—what makes a diet ‘diverse’?
  2. We sought to characterize the relationship between taxonomic dietary diversity (TDD) and phylogenetic dietary diversity (PDD) in a species-rich community of large mammalian herbivores in semi-arid East African savanna. We hypothesized that TDD and PDD would be positively correlated within and among species, because taxonomically diverse diets are likely to include plants from many lineages.
  3. By using DNA metabarcoding to analyze 1,281 fecal samples collected across multiple seasons, we compiled high-resolution diet profiles for 25 sympatric large-herbivore species. For each of these populations, we calculated TDD and PDD with reference to a DNA reference library for local plants.
  4. Contrary to our hypothesis, measures of TDD and PDD were either uncorrelated or negatively correlated with each other. Thus, these metrics reflect distinct dimensions of dietary specialization both within and among species. In general, grazers and ruminants exhibited greater TDD, but lower PDD, than did browsers and non-ruminants. We found significant seasonal variation in TDD and/or PDD for all but four species (Grevy’s zebra, buffalo, elephant, Grant’s gazelle), but the relationship between TDD and PDD was consistent across seasons for all but one of the 12 best-sampled species (plains zebra).
  5. In this large-herbivore assemblage, taxonomic generalists can be phylogenetic specialists, and vice versa. These two dimensions of dietary diversity suggest contrasting implications for efforts to predict how species will respond to climate change and other environmental perturbations. For example, network models predict that populations with high TDD but low PDD should be insensitive to phylogenetically ‘random’ losses of food species, yet highly sensitive to environmental changes that locally disadvantage entire plant lineages—such as changing rainfall patterns, altered fire regimes, and woody encroachment, all of which influence the relative abundance of the grass lineage (Poaceae) relative to trees and shrubs in ecosystems worldwide.


The data and scripts contained within the zip file are organized into three subdirectories. 


The subdirectory titled 'MammalInfoAndDietProfiles' contains 5 input data used for diet analyses: 

1. The file SpListForTree.csv contains information on mammalian species identifiers, the names used in the megaphylogeny cited in the main text, the body mass (g) extracted from the sources listed (Bmsource), digestive morphophysiology (ruminant or non-ruminant) and the domestication status (domesticated or wild). 

2. otutable_compositeprofile.csv contains composite diet profiles generated from data within the file otutable_rarefiedsamples.csv in order to enable species-level analyses in this paper.


The subdirectory titled 'Phylogeny' contains 5 input files used to construct and analyze the dietary plant phylogeny:

1. RaxmlScript.txt is the code used to run RAXML with all input files described below.

2. metabarcodeP6alignment.phy represents the alignment of the trnL-P6 sequences.

3. metabarcodeP6alignment_INDEL.phy is a modification of the trnL-P6 alignment that codes for INDELs as a 5th state for analysis in RAXML.

4. constraint_tree_bladj represents the constraint tree used to define family-level relationships among taxa in the phylogenetic analysis.

5. 20190918_RAxMLbladj is the output of the RAXML analysis, with branchlengths adjusted to represent time before present using BladJ and fossil calibration dates as described in the main text.


The subdirectory titled 'VegetationAvailability' contains 1 input file used to analyze vegetation availability:

1. VertVeg_Master_Vv326_20171009.csv represents 


National Science Foundation, Award: DEB-1930820

National Science Foundation

National Science Foundation, Award: DEB-1457697

National Science Foundation, Award: IOS-1656527