| Abstract | Scientific study of animals in situ requires vigilant observation of detailed animal behavior over weeks or months. When animals live in remote and/or inhospitable locations, observation can be an arduous, expensive, dangerous, and lonely experience for scientists. Emerging advances in robot cameras, long-range wireless networking, and distributed sensors make feasible a new class of portable robotic observatories that can allow groups of scientists, via the internet, to remotely observe, record, and index detailed animal activity. As a shorthand for such an instrument, we propose the acronym CONE: Collaborative Observatory for Natural Environments. One challenge is to develop a mathematical framework for collaborative observation. Collaborative observation includes (1) collaboration between humans of different backgrounds, skill sets, and authority/permission levels, (2) collaboration between humans and automated agents whose behavior arises from sensor inputs and/or computation, and (3) automatic detection of species and activities. In this talk, I will summarize our four-year development of algorithms, CONE systems, lessons learned, and results of field experiments. |
| Abstract | The ECOCEAN Whale Shark Photo-identification Library is an Internet-based software application for cooperative whale shark research. The primary purpose of the Library is to increase our understanding of whale sharks on a global and local level and to promote related conservation efforts through high quality research and scholarship. On a purely functional level, the Library is used to collect, protect, store, and share whale shark mark-recapture data gathered from a variety of individuals and institutions worldwide. It also has the capability of storing a variety of other specific data for each individual animal, including DNA profiles, satellite tracks, behavioural profiles etc. Certain data from the ECOCEAN Library are shared with other biology and ecology portals, such as the Global Biodiversity Information Facility and Fishbase. This sharing occurs at regular intervals via an installed TapirLink provider. The Library utilises several unique software applications to maximise the gain from data and identification images submitted from all stakeholders participating in this ‘citizen science’ program. Interconnect is the spot mapping tool; we employ a pattern-matching algorithm (modified from NASA-employed Groth algorithm used to map stars); Spot! is a program that employs perspective correction to enable the use of highly skewed images; sharkGrid is a distributed computing system that enable the public to participate in the monitoring program – even when they have not had the opportunity to swim with and collect data from these animals in the field. The Library currently houses 17000+ photos; 6300+ whale shark reports; 1400+ whale sharks collaboratively tagged from 1500+ individual data contributors. The system has potential to be applied to a wide range of naturally marked species, with a Polar Bear Library recently completed. Plans are being developed for the ECOCEAN Biodiversity Framework which will provide an open-source software platform that enables researchers of species other than whale sharks to use the online software to track population health in the wild. |
| Abstract | In this presentation I will give an overview of the first-ever community science project on observable evolution, using the common European garden snail Cepaea. The Evolution MegaLab is the brainchild of evolutionary ecologist Jonathan Silvertown of the UK’s Open University. It will revolve around a website, www.evolutionmegalab.org, which will be launched in early 2009 (the Darwin Year), and which will allow the general public in all European countries to enter frequencies of colour morphs of Cepaea nemoralis and Cepaea hortensis. These snails have been used for population genetics since the beginning of the 20th century. The genetics of their colours and banding patterns are known, as well as some of the more important selection pressures. Also, many old data on frequency distributions are available from all over Europe. Volunteers who enter data from their local area, will therefore receive immediate and automatic response from the website whether or not allele frequencies are significantly different since the previous record, and will also contribute to a large combined data set, which will allow the elucidation of evolutionary trends across Europe. Such trends are expected because three ecological factors involved in natural selection may have changed over time: (a) the snails’ chief predator, the song thrush, has declined in many parts of Europe, which may have lessened selection on camouflage colouration; (b) temperatures have gone up, which may have reduced selection on dark colour in higher latitudes; (c) distributions of open and closed habitat have changed, which will have changed the selection on different colour patterns that afford camouflage in different kinds of habitats. |
| Abstract | The Internet has connected resources and people around world. However, producing reliable systems that take advantage of the connectivity has often required heroic efforts. The biggest success story so far, the World Wide Web has made the organisation and sharing of documents simple enough to consider it a utility, usable without expert assistance. Distributed computing, the organisation and sharing of more general IT resources, requires a similar breakthrough. The e-Science and cyber-infrastructure projects of recent years have attempted to create a Web for processing capacity, data repositories and on-line instruments. We examine the place of grids, cloud computing, workflows, semantic metadata and ontologies in creating a simple ready-to-use environment for biodiversity research. |
| Abstract | In this paper we describe a proposed system for automatic visual monitoring of seabird populations. Image sequences of cliff face nesting sites are captured using time-lapse digital photography. We are developing image processing software which is designed to automatically interpret these images, determine the number of birds present, and monitor activity. We focus primarily on the the development of low-level image processing techniques to support this goal. We first describe our existing work in video processing, and show how it is suitable for this problem domain. Image samples from a particular nest site are presented, and used to describe the associated challenges. We conclude by showing how we intend to develop our work to construct a distributed system capable of simultaneously monitoring a number of sites in the same locality. |
| Abstract | We developed a new system for continuous radio tracking and tagging of animals, respectively, based on alternative technical processes that allows one to monitor automatically and continuously several specimens (up to 256) in same and real-time, with high accuracy and transmitting biomedical information, as well. |
| Abstract | This position paper discusses the potential of harnessing human computation via social networking technologies for citizen science and biodiversity monitoring. In particular, the potential usefulness of exploiting small applications installed on social networking websites, such as Facebook, for delivering remotely captured imagery of birds is discussed. A short review of previous relevant work in the areas of technology-assisted citizen science for bird monitoring applications, autonomous systems for bird detection and classification, human computation and games with a purpose (GWAPs), and social networking sites and their support of social games is given. It is then hypothesized that social networking sites like Facebook can be used to recruit, motivate, and mobilize large numbers of volunteers for citizen science applications and thereby facilitate a meaningful impact in a scientific discipline such as conservation biology and a list of research questions is presented. Finally some preliminary work is described and the route to further investigation is discussed. |
| Abstract | One of the major barriers to understanding the effects on biodiversity of extensive grazing in complex semi-natural habitats is the standardization of data collection between observers and data processing limitations. The advent of cheaper and more sophisticated digital camera technology has lead to a sudden increase in the number of habitat monitoring images and information that is being collected. We report the use of automated trail cameras (designed for the game hunting market) to continuously capture images of grazer activity in a variety of habitats. These techniques are being used in conjunction with GPS collars and more well established survey methods (dung counting and vegetation surveys). We are now developing Artificial intelligence based methods to assist in the analysis of the large number of images collected. This paper describes the data collection techniques, outlines the quantitative and qualitative data collected and proposes online and offline systems that can reduce the manpower overheads and increase focus on important subsets in the collected data. |
| Abstract | We present first results and ongoing work in the MITCH project on automatic ontology construction and a faceted browsing interface for the natural history domain. Specimen databases provide researchers in the field with invaluable knowledge about collected specimens throughout the world and throughout time. Accessing these databases to answer complex research questions requires either specialized knowledge of database query languages such as SQL, or interfaces that help the user narrow down a complex search with visual aids and query expansion or reformulation suggestions. The particular search aid we are developing, utilises ontological relations between the different concepts in the data, that assumes the presence of a domain ontology, mapping all concepts and their relations. Yet, ontologies are not always available, and manual ontology construction is a time-consuming task. We present a method that leverages the construction of an ontology by a database-driven approach to ontology construction, utilising the implicit domain knowledge already present in specimen databases, and combining this knowledge with an external semantic resource representing general world knowledge, viz. Wikipedia. We present a prototype of a search and browse interface to the data in which this ontology is integrated. |
| Abstract | In this paper we present an overview of various computer vision and image processing methods applied in the context of photo-identification of naturally marked marine mammals. These include segmentation, saliency detection and matching of salient regions and points. We have proposed new generic saliency-based recognition methods which are comparable to the state of the art-methods on repeatability and matching scores and outperform them in perceptual saliency. These methods are shown to work as part of computer-assisted photo-identification system for humpback whales and Rissos’ dolphins and can be applied to aid the recognition of any marked species. |
| Abstract | The behaviour and population ecology of seabirds can be indicators of the changing conditions in wide scale oceanic ecosystems. Understanding the spatial distribution, breeding ecology and at-sea behaviour of such species allows a fuller picture of the health of such ecosystems and the fragility of these species to environmental change. Recent advances in miniature GPS telemetry and wireless sensing provide a powerful set of tools to address such questions. Here, we report on our recent experiences of designing, deploying and maintaining a sensor network to detect individual visitations of Manx Shearwaters (Puffinus puffinus) to their breeding grounds on islands in the Irish Sea. Over two field seasons, the system has combined radio frequency identification (RFID) technology, environmental sensing and remote data transmission via GSM with an existing programme of GPS tracking on the Island. This work represents the beginning of a wider programme to automate the monitoring of animal behaviour and environmental variation and we discuss future plans for these systems and the new kinds of questions we believe they allow us to address. |
| Abstract | A semi-automated framework for analysis/mapping species' distrubution is presented and its functionality demonstrated using a species Buteo buteo L. (records from the NDFF and GBIF). A number of environmental predictors (gridded maps) was used to generate the maps of Habitat Suitability Index and to map the actual distribution for this species (for the Netherlands at resolution of 500 m; for Europe at resolution of 5 km). The preliminary results show that automated mapping of potential and actual distribution of species is possible, but also heavily dependent on the quality of inputs, especially on the quality of the field records (spatiotemporal reference, sampling coverage, taxonomic accuracy). With a further sophistication of statistical techniques that can extract spatio-temporal patterns, and with a further increase of spatial detail of environmental maps, we can anticipate that many data providers (especially GBIF) will be interested to include such mapping tools as a part of their regular service. |
| Abstract | Our ability to understand, conserve, and manage the planet’s marine biodiversity is fundamentally limited by the availability of relevant taxonomic, distribution, and abundance data. The Spatial Ecological Analysis of Marine Megavertebrate Animal Populations (SEAMAP) initiative is a taxon-specific geo-informatics facility of the Ocean Biogeographic Information System (OBIS) network. OBIS-SEAMAP has developed an expanding geo-database of marine mammal, seabird, and sea turtle distribution and abundance data globally. The OBIS-SEAMAP information system is intended to support research into the ecology and management of these important marine megavertebrates and augment public understanding of the ecology of marine megavertebrates by: (1) facilitating studies of impacts on threatened species, (2) testing hypotheses about biogeographic and biodiversity models, and (3) supporting modeling efforts to predict distributional changes in response to environmental change. To enhance the research and educational applications of this database, OBIS-SEAMAP provides a broad array of web-based products and services, including rich species profiles, compliant metadata, and interactive mapping services. This system takes advantage of recent technological advances in Geographic Information Systems (GIS), Internet data standards, and content management systems to stimulate a novel community-based approach to the development of a data commons for biogeographic and conservation research. New developments include (1) animal photo-id information; (2) advanced telemetry data, (3) passive acoustics data; and (4) statistical models.To date, the global OBIS-SEAMAP2.0 database includes 2.5 million observation records from more than 200 datasets, spanning 73 yr (1935 to 2008) provided by a growing international network of data providers. |
| Abstract | New developments in computer science, such as Grids, ontologies, workflows, and virtual labs are constantly being advertised as THE future, very promising, and probably causing a paradigm shift in ecological research. Although we acknowledge these new opportunities, we stress that very large gaps still have to be bridged. With interdisciplinary teams we are working towards large scale research infrastructures. EcoGRID is an information system and research environment for the management, integration and analyses of field observations and model results of the spatial and temporal distributions of flora and fauna in the Netherlands. It is built upon the heterogeneous databases of many organizations, growing to 50 million records in 2009. It facilitates both scientific ecological research and municipal spatial planning in relation to conservation. Second the Virtual Lab for Bird Migration Modelling (VLBMM), an international information system that was primarily developed to enhance military flight safety but that is currently used for scientific research as well. The VLBMM combines radar observations in NW Europe, GPS tracks, environmental data of weather and landscape, models and visualization. Having a few years of experience in building this kind of infrastructure, we will present the pitfalls and the lessons we learned, especially concerning the non-technological issues that are often ignored. |
| Abstract | In order to monitor bird movements we need sensors which can record or calculate location, time and potentially other relevant attributes such as speed, direction, altitude, behavior and environmental conditions. Due to the mobility of birds and physiological constraints, no one sensor is optimally designed to study bird movement at different scales in space, time. Furthermore, movement can be studied at the level of the individual or population, posing different technical requirements. The suitability of the information from a particular sensor is dependent on the particular process being studied and modeling framework used. In our research we are interested in monitoring and modeling bird movement at different scales, particularly in relation to environmental dynamics. Through the European Space Agency FlySafe initiative we are studying bird movement using a suite of complementary sensors including military air surveillance radar, meteorological radar and the tracking of individual birds using GPS technology. The aim of this initiative is to improve military flight safety in Northwest Europe. In this presentation, we will describe the characteristics of each of these sensors, how data is processed to provide useful information and how this information can contribute to our understanding of bird movement. Globally there are many communities interested in monitoring bird movement (see Movebank http://www.movebank.org/) and struggling with similar problems; we will briefly describe current efforts to bring these communities and information sources together. |
| Abstract |
| Abstract | Although Radio-frequency Identification (RFID) tagging technology is now commonly used in many familiar applications (e.g. tracking parcels and ticket-less transport systems), it has only recently been adapted to study insect behaviour. Insect's small size places strict constraints on the size and mass of tags which can be used without affecting behaviour. These constraints are particularly acute for flying insects. The size of passive tags used is a trade-off between absolute size/mass and the distance over which the tag can be detected (i.e. the tag's read-range). Working within these constraints it is now possible to reliably monitor the movement of individual bees, wasps and ants every time they enter or leave their nest, and to record when individuals visit specific sites in their environment (such as feeding stations). Crucially this can be done for long periods, potentially the animal's entire lifespan, with minimal disturbance to their behaviour. This approach has been used to monitor the foraging behaviour of bumblebees (to assess the effectiveness of artificial foraging recruitment pheromones) and the extent to which individual wasps move (drift) among colony nests. It is also being used to monitor the dynamics of ants during 'house-hunting', the activity of foraging bumblebees under 24 hour arctic daylight and to establish if bumblebees can help us to find possible solutions to complex routing problems. |
| Abstract | The LOFAR research project develops an ICT infrastructure to facilitate astronomical observations by means of thousands of small antenna’s. This infrastructure exists of a high capacity glass fiber connection from the antenna fields to a central location with a large data processing and storage capacity. The existence of such an infrastructure in rural areas challenged the agricultural community, apart from geologists and meteorologists, to develop agricultural production systems that are based on a large communication and processing capacity. The general idea in the LOFAR project, to monitor actual astronomical events in real time and to adjust and improve model based representations of these events by means of this observations is also applied in agriculture. The status of the micro climate in potato crops is monitored to improve models for disease control, the status of soil moisture and groundwater in addition to remote sensing observations is used to improve crop growth models which are used for fertilization and irrigation scenarios, and sensors on cows are used to monitor health status. Wireless sensor networks play a key role in the real time observations, due to the nature of the agricultural environment. Also the communication of field data to the LOFAR infrastructure requires the development of a so called Last Mile solution, based on Wi-Fi based mesh networks. |
| Abstract |
Data loggers have traditionally been used for various environmental
monitoring applications. However, their high per-unit cost implies that
they can only be deployed in very small numbers. Effectively, they only
allow point measurements. Wireless sensor networks (WSNs), however, are
usually deployed in much larger numbers and can therefore provide high
resolution spatial and temporal data. Their wireless capability
also allows easier deployment especially in harsh environments.
But WSNs are generally made up of sensor nodes that are battery
powered. Thus in order to have a network that is able to run for
long periods without carrying out battery replacements, the various
protocols running on the sensor nodes need to be designed to
maximize energy efficiency. One of the main techniques to improve
network lifetime is to reduce the duty cycle of the transceiver on a
sensor node as it is one of the primary sources of energy consumption.
The tradeoff is that the amount of data that can be transmitted by every
node in the network is greatly diminished. This can be a major problem especially
when the end-user requires every sensor node in the network to report its readings
periodically as it leads to excessive energy consumption and also to reduced data
quality caused by high rates of data loss due to the limited bandwidth.
In this presentation, we provide an overview of two separate algorithms we have developed, that minimize energy consumption by reducing the amount of data that needs to be transmitted and by reducing the number of sensor sampling operations. Both the solutions have been developed as part of our efforts together with the Australian Institute of Marine Science, to deploy a large scale WSN on the Great Barrier Reef (GBR). This network will be used to study the effects of global warming and agriculture on the coral reefs. Details of our deployment of sensor nodes on the GBR using buoys are also described. The first algorithm takes advantage of the spatial correlations of sensor readings that may exist between adjacent nodes. The algorithm uses a few as representative nodes that perform in-network aggregation. This reduces the total number of transmissions. We present theoretical performance estimates and upper bounds of our algorithm and evaluate it by implementing the algorithm on actual sensor nodes, demonstrating an energy-saving of up to 80% compared to raw data collection. Data quality is also greatly improved by minimizing data loss rates. The second algorithm exploits temporal correlations that may exist between consecutive sensor readings. We describe an adaptive sensor sampling scheme where nodes predict readings instead of sampling sensors when the readings follow a predictable trend. Our results based on real and synthetic data sets, indicate a reduction in sensor sampling by up to 93%, reduction in message transmissions by up to 99% and overall energy savings of up to 87%. We also show that generally more than 90% of the collected readings fall within the user-defined error threshold. |
| Abstract | We are still playing catch-up to the possibilities of the information age. While technological leaps have led to the ubiquitous nature of the Internet, we are only beginning to realize the scientific leaps that it may enable. Within the realm of biodiversity monitoring, we are already seeing efforts to advance science by coordinating distributed information gathering to aid in furthering our understanding of ecology. The challenge with these projects is to understand how to better facilitate the flow of information in ways useful for both scientists and those collecting data. What will be presented is a vision of how aspects of such a structure could look given developments in similar crowdsourcing communities. Additional insights will be drawn from the characteristics of the open source movement, along with a discussion of how the latest web technologies are changing our relationship with information. |
| Abstract | The European Strategy Forum on Research Infrastructures (cordis.europa.eu/esfri/) selected LifeWatch (www.lifewatch.nl) in 2006 as an essential new infrastructure to promote breakthroughs in science. The LifeWatch infrastructure for biodiversity research, presently in its preparatory phase, addresses the huge gaps we face in our understanding of life on Earth. Its design offers a new methodological approach to study biodiversity systems with large-scale data resources, advanced algorithms and computational capability to support knowledge development on biodiversity from the genetic level up to ecosystems and landscapes. Life Watch is not only meant to serve scientific research, but also the understanding and the rational management of our natural environment. Generating the crucial data resources is a main focus of LifeWatch, and it is planned to allocate quite some funds for new enabling technologies. This includes sensor development for monitor sites and digitization techniques for biological collections. The development of lightweight instruments, sensor networks, data repositories and knowledge development in ubiquitous environments in the context of LifeWatch will be discussed. The current LifeWatch preparatory project runs from early 2008 for three years with the objectives to develop a construction blue-print and to convince interested governments to invest for the actual construction and successive operations. |