COMADRE: Animal Matrix Database Research Guide

Q: What does MatrixSplit mean in the COMADRE database?

MatrixSplit indicates whether original publications provided sufficient information to separate the projection matrix A into constituent biological submatrices. A value of Divided confirms distinct U , F , and C matrices, whereas Indivisible denotes an unseparated transition matrix.

Q: How are captive and wild populations distinguished in COMADRE?

The MatrixCaptivity metadata field designates whether a study monitored wild populations in natural habitats ( W ), captive groups in zoos or breeding centers ( C ), or laboratory colonies ( L ), preventing captive survival conditions from distorting wild evolutionary analyses.

The COMADRE Animal Matrix Database is an open-access repository of matrix population models (MPMs) and demographic metadata across the animal kingdom. Initiated to resolve data fragmentation in published ecology, COMADRE compiles standardized stage-, age-, and size-classified projection matrices. By structuring vital rates into uniform submatrices, the platform enables comparative analyses of animal life histories, senescence, and extinction risk.

What It Is

COMADRE standardizes empirical animal demographic studies into discrete-time Matrix Population Models. A projection matrix describes how population stage distributions change across discrete census intervals according to the equation:

n(t + 1) = A n(t)

In this formulation, n(t) is the population stage-distribution vector at time t, and A is the population projection matrix. COMADRE explicitly partitions the full projection matrix A into three biological submatrices:

A = U + F + C

These submatrices isolate distinct biological processes across the life cycle:

Matrix U (Survival and Transitions): Encodes stage-specific survival probabilities alongside developmental growth, stasis, and shrinkage. Because entries represent conditional transitions, column sums denote total stage survival and must not exceed 1.0 (Σ_i U_ij ≤ 1.0).
Matrix F (Sexual Reproduction): Contains per-capita sexual fecundity rates representing offspring produced by reproductive stages.
Matrix C (Clonal Reproduction): Captures rates of vegetative reproduction, fission, or budding, occurring in colonial invertebrates and select insects.

Partitioning matrices into U, F, and C allows researchers to automate life-table calculations. Inverting the survival matrix yields the fundamental matrix N = (I – U)^-1, providing mean life expectancy, stage residence times, and age at maturity across species.

Who Runs It and What It Covers

COMADRE was established in 2011 at the Max Planck Institute for Demographic Research (MPIDR) in Rostock, Germany, led by Dr. Roberto Salguero-Gómez and Dr. Owen R. Jones in collaboration with the University of Oxford and an international consortium of ecologists. It serves as the animal counterpart to the COMPADRE Plant Matrix Database.

Taxonomic coverage spans terrestrial, freshwater, and marine animals across major clades: Mammalia, Aves, Reptilia, Amphibia, Actinopterygii, and Invertebrata (including corals, mollusks, and arthropods). The foundational release (version 1.0.0) compiled 1,625 matrix models from 402 studies across 345 taxonomically accepted species validated against the Catalogue of Life.

Geographic coverage is global, with initial sampling centered in North America, Europe, Africa, Asia, and Oceania. Data curation follows rigorous error-checking: approximately 50% of source papers undergo double-blind digitization by two independent researchers, followed by triple review by lead curators to verify stage descriptions and ensure column sums of U do not exceed 1.0.

How to Search or Use It

Researchers access COMADRE through its web portal or programmatically in R. The following six steps demonstrate how to retrieve, filter, and analyze matrix data using data frames in R programming:

Access the Web Portal: Visit compadre-db.org to inspect version notes, digitizing protocols, and species coverage lists. Users can download complete database snapshots as structured .RData files, CSV tables, or JSON.
Install Rcompadre: In an R environment (R ≥ 3.5), install the official CRAN interface:
```
install.packages("Rcompadre")
library(Rcompadre)
```
Fetch the Database: Download the latest standardized animal database into memory as a CompadreDB S4 object containing matrices and standardized metadata:
```
comadre = cdb_fetch("comadre")
```
Filter Quality Records: Subset records using tidyverse verbs to retain wild populations with fully partitioned matrices:
```
library(dplyr)
comadre_clean = comadre |>
  filter(MatrixSplit == "Divided",
         MatrixCaptivity == "W")
```
Extract Submatrices: Access the component projection submatrices and author stage definitions using specialized accessor functions:
```
mat_A = matA(comadre_clean)[[1]]
mat_U = matU(comadre_clean)[[1]]
mat_F = matF(comadre_clean)[[1]]
```
Compute Vital Rates with Rage: Use the companion CRAN package Rage to compute population growth rate (dominant eigenvalue λ), generation time, and life expectancy:
```
library(Rage)
lambda = pop_growth_rate(mat_A)
life_exp = life_expect_mean(mat_U)
```

Strengths and Limits

Evaluating COMADRE requires assessing its comparative modeling capabilities against its empirical scope.

Strengths:

Uniform Matrix Partitioning: Standardizing A into U, F, and C submatrices allows automated comparative demography across diverse life forms.
Structured Contextual Metadata: Every entry records technical metadata including ecoregion, wild versus captive provenance (MatrixCaptivity), and study treatments.
Reproducibility: Distribution via CRAN through Rcompadre ensures reproducible R analysis workflows that adhere to FAIR vs CARE principles.

Limits:

Taxonomic Asymmetry: Mammals and birds in temperate regions are more densely sampled than tropical insects and marine invertebrates.
Deterministic Assumptions: Most entries represent time-invariant matrices under average conditions; density dependence and environmental stochasticity require external parameterization.
Matrix Primitivity Caveats: Users must verify that empirical matrices meet mathematical ergodicity and primitivity criteria before computing asymptotic eigenvalues.

Alternatives

The table below summarizes primary repositories used alongside COMADRE for demographic and ecological modeling:

Repository	Taxonomic Scope	Primary Data Model	Typical Research Application
COMADRE	Animalia	Matrix Population Models (A = U + F + C)	Comparative animal demography and life-history projection
COMPADRE	Plantae, Fungi, Algae	Matrix Population Models (A = U + F + C)	Plant demography, seed bank dynamics, vegetative propagation
DATLife	Animals and Humans	Individual life tables and survival curves	Actuarial senescence, longevity comparisons, mortality hazards
Movebank	Animalia (tracked wildlife)	Bio-logging telemetry and GPS trajectories	Animal movement ecology and migration corridor mapping
Living Planet Database	Vertebrata	Abundance time-series records	Global population trend monitoring and biodiversity indices

Researchers choose COMADRE when investigating stage-structured vital rates and life-cycle dynamics, whereas DATLife serves continuous survival curves and Movebank tracks spatial trajectories.

Frequently Asked Questions

What is the difference between COMADRE and COMPADRE?

Both databases share the same data architecture and S4 R object format, but COMADRE houses matrix population models for animals, while COMPADRE catalogs plants, algae, fungi, bacteria, and viruses. Both decompose projection matrices into survival (U), sexual reproduction (F), and clonal reproduction (C) submatrices.

How should researchers cite COMADRE in academic publications?

Users must follow a dual-citation convention: cite the primary descriptor article (Salguero-Gómez et al., 2016, Journal of Animal Ecology) to acknowledge the data infrastructure, and cite the original empirical studies from which specific matrices were extracted. Source DOIs registered via services evaluated in Crossref vs DataCite are recorded in the metadata.

What does MatrixSplit mean in the COMADRE database?

MatrixSplit indicates whether original publications provided sufficient information to separate the projection matrix A into constituent biological submatrices. A value of Divided confirms distinct U, F, and C matrices, whereas Indivisible denotes an unseparated transition matrix.

How are captive and wild populations distinguished in COMADRE?

The MatrixCaptivity metadata field designates whether a study monitored wild populations in natural habitats (W), captive groups in zoos or breeding centers (C), or laboratory colonies (L), preventing captive survival conditions from distorting wild evolutionary analyses.

Can COMADRE matrices be used for stochastic population viability analysis?

Yes. Studies providing multiple matrices across consecutive years or treatments allow researchers to simulate environmental stochasticity by drawing matrices across time intervals, though continuous environmental covariates require external modeling.

Sources

Salguero-Gómez, R., Jones, O. R., Archer, C. R., et al. (2016). COMADRE: a global data base of animal demography. Journal of Animal Ecology, 85(2), 371–384. DOI: 10.1111/1365-2656.12482. (PMCID: PMC4819704).
COMPADRE and COMADRE Matrix Database Portal. Max Planck Institute for Demographic Research and University of Oxford. URL: https://www.compadre-db.org/
Jones, O. R., et al. Rcompadre: Utilities for using the COM(P)ADRE Matrix Population Databases. Comprehensive R Archive Network (CRAN). URL: https://cran.r-project.org/package=Rcompadre
Jones, O. R., et al. Rage: Life History Metrics from Matrix Population Models. Comprehensive R Archive Network (CRAN). URL: https://cran.r-project.org/package=Rage