Transcript Diapositivas

Miguel Porto, Pedro Beja, Rui Figueira
CIBIO/InBIO
IICT

Ecological systems are highly intricate networks:
every species may be related, in different ways,
to every other species
changes in the abundance of any one species may
affect, directly or indirectly, N other species

Much of the knowledge on ecological networks
is at the conceptual level, not at the factual level

Biodiversity data has traditionally been based on
species occurrences

Biodiversity databases, as a norm, fail to
document relations

Ecological relations, like species, have a spatial
and temporal dimension, i.e. they occur
Why not store relationship occurrences rather
than species occurrences?
(the former includes the latter, anyway)
Cytinus ruber parasitizing rockrose
at 38.7654 ºN 7.9866 ºW
it might look like a detail but it makes a huge difference in the amount
of fundamental ecological information that is recorded

An infrastructure to store and manage
occurrences of ecological relations that:
 is connected bidirectionally to existing species
occurrence databases
 strictly follows the data standards for existing types
of data (e.g. DarwinCore for species occurrence data)
 proposes new standards for describing ecological/
biological relationship data
 provides an array of relationship-based web services
to allow interoperability with existing platforms

Raw data: published relationship data comes in
an immense array of formats and with varying
levels of detail and aggregation
▪ Orobanche gracilis parasitizing Retama sphaerocarpa
▪ Blackbird feeding on the fruits of ivy in Portugal
▪ Fish of genus Barbus feeding on filamentous algae in Tagus
river, in Spring
▪ Beetle pollinating an unidentified red flower at 38.4532 ºN
8.2456 ºW in May-2007
data model able to accommodate all kinds of raw
data without loss of information or generalization

Computational resources: for example, a small country like
Portugal may have ca. 40000 species which can all potentially
interact
 relations are directional and may be of several types and subtypes:
▪ Feeding on
▪ Parasitizing
▪ Dispersing
▪ Pollinating
▪ Co-occurring
▪ ...
 … and relations may have different weights/strengths (e.g. species A
is more frequently found feeding on species B than on C)
 … and occurr at different places and dates
 … and pertain to different body parts

Computational resources
 The network easily attains great complexity because
▪ different data facets are covered – taxonomy, morphology,
ontology, ...
▪ data is highly structured in each facet
▪ relationship occurrences are stored – not “conceptual”
relationships – which leads to large amounts of data
accumulating over time
 but it needs to be efficiently traversed and
summarized in an intelligible and meaningful way
aim
To build a virtual lab infrastructure for
 storing ecological relationship data
 conducting network-based analyses
 testing ecological network-based hypotheses

Data is either compiled from published studies
or from direct observation (citizen science
platform)

Provides services and interfaces for querying,
visualizing, summarizing and analyzing the
network

Highly flexible as to the nature of underlying data:
 relations may be solely “conceptual” without further




details, as obtained from classical bibliography (e.g.
species A parasitizes species B), but this is far from ideal
relations may have precise geographical coordinates and
timestamp
relations may connect any two entities of any
taxonomical rank (e.g. species, genus, family, order...)
relations may connect entities which are not necessarily
taxonomic (e.g. arbitrary trait-based entities)
relations may refer to precise organs, structures or life
stages (e.g. caterpillar feeding on the leaves of species A)

A virtual lab infrastructure for conducting
ecological network-based analyses
 Analyze the spatial patterns of relations and their
relationship with environmental drivers, and predict
network-level changes upon environmental change
 Infer functional relationship patterns from
documented relationship occurrences
 Predict the n-th order impacts of removing nodes
(e.g. species) in the integrity of ecological networks
 Test the ecological significance of observed
relationship patterns using simulated random
networks