MOSAIC - A Unified Trait Database to Complement Structured Population Models

Abstract

Despite exponential growth in ecological data availability, broader interoperability amongst datasets is needed to unlock the potential of open access. Our understanding of the interface of demography and functional traits is well-positioned to benefit from such interoperability. Here, we introduce MOSAIC, an open-access trait database that unlocks the demographic potential stored in the COMADRE, COMPADRE, and PADRINO open-access databases. MOSAIC data were digitised and curated through a combination of existing datasets and new trait records sourced from primary literature. In its first release, MOSAIC (v. 1.0.0) includes 14 trait fields for 300 animal and plant species: biomass, height, growth determination, regeneration, sexual dimorphism, mating system, hermaphrodism, sequential hermaphrodism, dispersal capacity, type of dispersal, mode of dispersal, dispersal classes, volancy, and aquatic habitat dependency. MOSAIC includes species-level phylogenies for 1,359 species and population-specific climate data. We identify how database integration can improve our understanding of traits well-quantified in existing repositories and those that are poorly quantified (e.g., growth determination, modularity). MOSAIC highlights emerging challenges associated with standardising databases and demographic measures.

Fig. 1

Structure of the MOSAIC database (v1.0.0). The MOSAIC database comprises a combination of existing trait records centralised from data servers, databases, and datasets and new records collected from the primary literature by the MOSAIC team. Existing records are labelled, MOSAIC-A, in the mosaic metadata attribute field and include provenance in the metadata. New data records that adhere to existing data standards and that fit into the remit of an existing database (i.e., a data gap filled by the MOSAIC team) are labelled MOSAIC-B. New records that do not have an existing data standard or for which there is not an existing database are labelled MOSAIC-C.

Fig. 2

Trait fields in the MOSAIC database (v1.0.0). The MOSAIC database contains a combination of continuous and discrete trait fields. Six of the 14 trait fields are illustrated here for animals, and organised by trait level (all discrete fields) and taxonomic classification. Trait distributions vary strongly by trait and taxonomy. Observed variation in trait values by taxonomic group is a prerequisite condition to their potential value toward explaining reported variation in vital rates and resulting demographic properties across the Tree of Life. Species included in the MOSAIC database are limited to those for which stage-structured population data exist in the COMADRE, COMPADRE, and PADRINO databases.

Fig. 3

The spatial distribution of MOSAIC (v1.0.0) records relative to the (a) COMADRE and (b) COMPADRE databases. The COMADRE and COMPADRE dataset include animal and plant demographic data, respectively, on all continents except Antarctica, with a substantial bias to North America and Central Europe. MOSAIC introduces trait records that include models in all major geographies of COMADRE and COMPADRE. The map shows the density of matrix population models globally (color graded by density per area in ca. 150 km² hexagonal cells). Cells bordered by white represent localities of population models for which there are trait records included in the first release of the MOSAIC trait database. Species prioritised in v1.0.0 of MOSAIC did not include the PADRINO dataset, so no distribution figure is shown for integral projection models.

Fig. 4

Phylogenetic mapping of traits included in MOSAIC (v1.0.0). (a) Phylogenetic distribution of seven major traits for animals: (1) growth determination; (2) sexual dimorphism; (3) monogamy; (4) gonochory; (5) volancy; (6) dispersal; and (7) biomass. (b) Phylogenetic distribution of four traits in plants: (1) dispersal capability; (2) dispersal type; (3) dispersal mode; and (4) dispersal class. Figures developed in R using the ggtree package.

Fig. 5

Trait covariance in the MOSAIC database (v1.0.0). Some traits exhibit high correlation in the MOSAIC database, such as between growth determination and between growth regeneration. Trait associations are expected to occur in the MOSAIC dataset and may reflect widespread constraints or statistical anomalies, particularly when dealing with small samples or specific taxonomic subgroups. Trait covariation can be symptomatic of biomechanical constraints (e.g., flight and biomass), major growth characteristics (e.g., modularity and growth determination), or other past or presently compelled associations (e.g., height and vessel density).