Ecological trait data promise to usher in an era of predictive ecology by bridging the gap between species identity and ecosystem function. However, globally predictive models—for example, plant community response to global environmental change—require the discovery and integration of multiple disparate trait datasets. Currently, locating these data is time-consuming so a central registry of trait databases would greatly facilitate data discovery. The TraitNet Semantic Data Registry leverages the Plant Trait Ontology (PLATON, developed in collaboration with colleagues at NCEAS and CEFE, France), which implements the Extensible Observation Ontology (OBOE, developed by colleagues at NCEAS). Our search model leverages PLATON by allowing the client to search along three facets of the ontology. The first facet describes a partonomy of plant morphological features, which represent the entities upon which observations are made. For example, a Leaf is part of a Branch, which is part of the Shoot of a particular Plant. The second facet includes the characteristics of the plant that might be measured. For example, Physical Characteristics like Mass, Height, and Area, and Physiological Characteristics like Respiration and Photosynthesis. The third facet includes the traits themselves, which are composed of entities and characteristics. For example an entity Leaf and the characteristic Mass are combined to form a measurement type that unambiguously describes the trait LeafMass. In contrast, a more general measurement type of Leaf and PhysicalCharacteristic would return all traits involving physical characters of leaves, such as LeafArea and LeafMassPerArea, as well as LeafMass. The registry lists traits based on the selection of entity types, characteristic types, or the combination of both. Based on the selection of a trait, the user is returned the metadata of datasets—located with NCEAS’ Knowledge Network for Biological Complexity (KNB) data repository—that contain the trait data of interest.

            The key advancement of the TraitNet Semantic Data Registry is that its interface enables a user to search and locate data driven by controlled vocabularies specific to trait types as opposed to traditional keyword based search strategies. These ontology-based searches are not only more intuitive for users, but offer users new pathways to additional related data based on the ontological relationships in the underlying ontology (which is not possible using traditional keyword searches). Further, searching through the TraitNet Semantic Data Registry can not only save and optimize a user’s time, but also increases the quality of search results. Our approach is extensible both within and across multiple scientific domains as long as the ontology conforms to a basic model of observations and measurements, which is gaining traction in several earth and life sciences (e.g. OBOE` model, EQ model, and OGC’s O&M model). Our search model as implemented in the TraitNet Semantic Data Registry represents an attempt to incorporate OWL reasoning for enhancing data discovery by the biodiversity research community. While it is initially focused on discovering plant functional trait data, it has the potential to be extended to enable semantic search across any types of functional trait data.