Overview

This application area focuses on creating and applying a new genre of sensor networks for making observations in aquatic ecosystems that will enable the generation (and testing) of novel hypotheses regarding the processes that control the distribution, growth and demise of planktonic microbial populations. Our primary long-term scientific goal is to understand and ultimately predict the conditions under which specific populations of marine microorganisms develop in nature. A fundamental requirement for attaining this objective is the correlation of environmental conditions and driving forces with microorganismal abundances at the small spatial and temporal scales that are relevant to the microorganisms.

Our work combines laboratory development and field deployment of technology for in-situ, real-time studies of microbial assemblages and concomitant environmental parameters. Laboratory studies provide a testbed for the development and testing of hardware and software for novel sensing/sampling approaches, and for experimental studies of plankton behavior (growth, vertical migration, trophic interactions). Networks for Aquatic Microbial Observing and Sensing (NAMOS) employ coordinated measurements between stationary sensing nodes and robotic vehicles to provide in-situ, real-time presence for observing plankton dynamics (e.g. chlorophyll concentration), linking them to environmental variables (e.g. temperature, light, nutrients, etc.), and enact adaptive sampling protocols for the collection and analysis of samples by autonomous vehicles.

Approach

Our goal for the field component of the Aquatic Microbial Observing Systems is to develop robust, decentralized algorithms and supporting hardware that enable a wireless sensor network consisting of sensor-equipped buoys and a sensor/sampler-equipped autonomous vehicle (robotic boat) that can perform adaptive hydrographic and biological sampling using the information provided by the network. Field studies to date have focused on Lake Fulmor at the James Reserve in the San Jacinto Mountains. Lake Fulmor provides an interesting natural testbed for our work and an excellent opportunity for interacting with other components of the CENS community.

The laboratory work involves the development and testing of novel sensors/samplers, the development of supporting software and hardware, and the testing of these novel approaches in ‘artificial water columns’ used to simulate natural plaktonic environments. For this work we have focused on studies of the ecology of harmful algal blooms (Aureococcus anophagefferens, the cause of brown tides of the Middle Atlantic States); Lingulodinium polyedrum, the cause of intense red tides of the Southern California Bight). Studies of these organisms not only provide an excellent means of testing new sensing/sampling approaches but also have yielded useful insights into the factors controlling the growth and aggregation of harmful algal bloom-forming species of phytoplankton.

Systems/Experiments

Our field work during the past year has centered on the deployment and refinement of NAMOS (Networks for Aquatic Microbial Observing and Sensing) within Lake Fulmor. The system consisted initially of 5 moored buoys (Fig. 1) equipped with strings of thermistors to provide information on the vertical temperature structure within the lake, and sumersible fluorometers (for measurement of chlorophyll concentrations) at 0.5 m depth in the lake. Thermistor strings provide vertical and horizontal patterns of temperature, and temporal patterns temperature on time scales from seconds to days). Fluorometers provide near-surface spatial (horizontal) patterns of chlorphyll, and temporal patterns scales of seconds to days). For details of this system see Actuation section.

Figure 1. Moored sensor-buoy within the NAMOS wireless sensor network.

Three studies have been conducted in Lake Fulmor during the last year (May, 2005; July, 2005; October, 2005). Experiments during all three deployments employed at least five buoys along the length of the lake in order to examine spatial heterogeneity in temperature and chlorophyll throughout the lake (Fig. 2). In addition, the deployment in October, 2005 involved the deployment of four additional buoys in the lower lake in order to begin to address questions of multiscale actuated sensing and sampling.

Figure 2. Areal photograph of Lake Fulmor. Arrows show approximate locations of the five main sensor buoys.

Measurements made by static (i.e. moored) sensor buoys in NAMOS are augmented by the activities of an autonomous surface vehicle (a robotic boat) that can take carry out sensing/sampling missions based on a set of preprogrammed commands or can be controlled wirelessly and manually (Fig. 3). The boat is equipped with a fluorometer to enable the acquisition of high-resolution spatial patterns of chlorophyll concentrations. In essence, the robotic boat

Figure 3. Robotic boat equipped with GPS, compass, fluorometer and water sampler.

The primary purpose of the field studies has been to test and debug the buoy-based sensor network and robotic boat (for a report on these aspects of this application, see Multiscale Actuated Sampling). However, these studies have also enabled unique and insightful observations of the planktonic assemblages in this environment (see Accomplishments below). Moreover, these initial observations serve as the basis for new hypotheses on ecosystem structure and function in this freshwater environment, which in turn stimulates new sensor network approaches. This iterative process drives the development of embedded networked sensing, and the scientific findings of these experiments drive the application domain. New hypotheses for more in-depth measurements planned for this coming year (see Future Directions below).

Experiments in the laboratory component of the Aquatic Microbial Observing Systems are carried out primarily in a 2-meter water column. This ‘artificial water column’ is illuminated from above and cooled at the bottom to produce a realistic, stratified water column. This laboratory testbed is simplistic in its composition but it is an effective means of simulating a natural water column (Fig. 4). The use of such experimental microcosms are a mainstay of aquatic microbial ecology. The development of our testbed enables the design and testing of new approaches for sensing and sampling vertically in the water.

Figure 4. Experimental testbed for examining plankton dynamics, and also for the development and testing of hardware and software for sensing and sampling in aquatic ecosystems.

To date, water column experiments in the laboratory have focused on two species of harmful bloom-forming algae (Fig 5).

Figure 5. Red tide off Orange County, CA formed by the dinoflagellate Lingulodinium polyedrum(left), and a brown tide formed by the pelagophyte alga Aureococcus anophagefferens in an experimental water column on Long Island, NY.

These lab setups also provide a mechanism for testing new sensing technologies based in novel genetic, immunological and instrumentational approaches for microorganismal detection and enumeration (see Sensors section). The long-term goal of this work is the integration of these different technologies and approaches (embedded networked sensors, novel microorganismal identification) to provide the next generation of tools for studying these organisms in natural settings.

Accomplishments

Field deployments: A major effort has been exerted during the past year by the Aquatic Microbial Observing Systems application to get a functioning network working in a natural freshwater ecosystem. Towards this end, a multi-node (5-9), wireless network of sensor-equipped buoys has been deployed repeatedly in Lake Fulmor of the James Reserve. Initial coordinated measurements between stationary sensing nodes and a robotic boat have been demonstrated to provide high-resolution temporal and spatial information on temperature and phytoplankton (chlorophyll) distribution within the lake. Preliminary observations have resulted in novel hypotheses focused on the role of vertical migratory behavior in nutrient acquisition and community dominance by phytoplankton species. Increasingly sophisticated deployments and experiments will drive the development of robust adaptive sensing and sample collection, enabling previously unachievable observations.

Previous years’ work to build an aquatic sensor network has incrementally provided the hardware and software advances necessary to enable deployment. Among other accomplishments, these advances have included: (1) Design and construction of moored sensor buoys (presently capable of continuous depth-resolved temperature measurements and fluorometric measurements of chlorophyll at 0.5 m). (2) Design and construction of an autonomous sensing/sampling vehicle (robotic boat) with directed navigation that can integrate with the static node network, make continuous underway fluorometric measurements of chlorophyll, and collect surface water samples at designated locations. (3) Development of software for network communication, within-network data synthesis, and transmission of network information to a land-based station for visualization and data storage. (4) Development of efficient, distributed navigation algorithms to enable autonomous operation of the robotic boat (including a bacterium-inspired algorithm for robot navigation and homing with minimal computational requirements), integration into the existing network of static nodes and adaptive sampling by the vehicle.

Three field deployments of the NAMOS sensor network were conducted in 2005 in Lake Fulmor of James Reserve past year. Deployments during May, July and October represented different goals of development, testing, and improvement of the sensor network and the robotic boat (see Actuation and Multiscale Actuated Sensing). However, data collected during each deployment provided an interesting and unique insight into the structure of the planktonic community in the lake. All three deployment involved measurements made from a static, moored sensor array of at least five nodes (the deployment in October also involved a deployment of 4 additional nodes to begin to examine multiscale sensing).

Placement of the five main buoys was approximately the same for all three deployments (Fig. 6). This arrangement provided information from various locations along the length of the lake. Retention of these same locations facilitated comparisons among the three deployments.

Figure 6. Location of static sensor nodes (buoys) within Lake Fulmor during network deployments in May, July and October 2005.

Sensor data for chlorophyll fluorescence (at 0.5 m depth) and temperature (depth-resolved) is provided in Figures 7-10. Information was collected over the course of two to four days for each deployment. Figures 7-9 provide contours of the spatial patterns of the two parameters collected by the buoy network at four specific times of day (approximately 6am, noon, 6pm and midnight).

The May deployment (Figure 7) revealed minor thermal stratification of the water column, with considerable warming of surface waters during the daylight period, with cooling during the night. Nighttime cooling yielded a nearly isothermal water column by morning. Chlorophyll concentrations were modest during the May deployment and did not indicate major spatial heterogeneity throughout the lake or temporal heterogeneity over a 24 hour period. The lower lake area (southwest) had slightly higher chlorophyll concentrations than the upper (inlet) end, but overall chlorophyll remained below approximately 20 µg l-¹.

Temperature and chlorophyll patterns observed using the sensor network during the July deployment were in sharp contrast to patterns observed during May (Figure 8). Water temperature was higher than May (up to 25°C) and relatively isothermal throughout the lake, with the exception of one area of the lake that indicated significant turnover of the lake (presumably due to late afternoon wind). Chlorophyll concentrations during most of the day were relatively homogeneous throughout the lake, but showed a dramatic increase during the night (note midnight pattern; Fig. 8, bottom right). We interpret this pattern as indicating a major diel vertical migration of phytoplankton (see Figures 11-13, below).

Water temperature throughout the lake during the October deployment was again isothermal and moderate (approximately 10-13°C), and a highly heterogeneous pattern of chlorophyll was observed spatially throughout the lake and temporally over a diel cycle.

Figure 7. Chlorophyll at 0.5 m depth (top graph in each panel) and depth-resolved temperature patterns during May 2005 in Lake Fulmor. Measurements were made at approximately 6am (top left), noon (top right), 6pm (bottom left) and midnight (bottom right).

Figure 8. Chlorophyll at 0.5 m depth (top graph in each panel) and depth-resolved temperature patterns during July 2005 in Lake Fulmor. Measurements were made at approximately 6am (top left), noon (top right), 6pm (bottom left) and midnight (bottom right).

Figure 9. Chlorophyll at 0.5 m depth (top graph in each panel) and depth-resolved temperature patterns during October 2005 in Lake Fulmor. Measurements were made at approximately 6am (top left), noon (top right), 6pm (bottom left) and midnight (bottom right).

Figure 10. Chlorophyll at 0.5 m depth (top graph in each panel) and depth-resolved temperature patterns during October 2005 in Lake Fulmor. Measurements were made at approximately 6am (top left), noon (top right), 6pm (bottom left) and midnight (bottom right). Data from 4 additional buoys have been added to provide higher resolution in the lower portion of the lake.

The sensor network during October was supplemented with additional buoys in the lower (southwest) portion of the lake in order to examine how high-density spatial sampling could improve resolution of the temperature and chlorophyll patterns (Figure 10). This constitutes our first attempt to directly examine multi-scale sensing. Increased sensor density revealed heterogeneity in both chlorophyll and temperature that were not detected by the less-dense sensor network (compare panels Figs. 9 and 10).

The static buoy sensor network has also enabled the investigation of temporal patterns of temperature and chlorophyll on scales ranging from seconds to days for the three deployments (Figure 11-13). As observed for the spatial patterns at four time points,

Figure 11. Variations in chlorophyll at 0.5 m depth (green) and depth-resolved temperature at four static sensor nodes over approximately three days during May 2005 in Lake Fulmor.

continuous chlorophyll measurements indicated peaks of chlorophyll in the 20-30 µg l-¹ range. Also evident were distinct peaks occurring near midnight. This pattern is highly unusual, and raises interesting questions about the biology giving rise to this pattern. A similar diel pattern was observed for chlorophyll during July of 2005 (Figure 12) and October of 2005 (Figure 13), although chlorophyll concentrations were dramatically higher at the time of these subsequent deployments, and the period of time over which chlorophyll remained high was much longer during October (several hours) relative to the May or July deployments. Differences in the phytoplankton assemblages of the lake were observed between the July and October deployments (Figures 14, 15).

Figure 12. Variations in chlorophyll at 0.5 m depth (green) and depth-resolved temperature at four static sensor nodes over approximately three days during July 2005 in Lake Fulmor.

Figure 13. Variations in chlorophyll at 0.5 m depth (green) and depth-resolved temperature at four static sensor nodes over approximately three days during October 2005 in Lake Fulmor.

Figure 14. Phytoplankton present during the July 2005 plankton of Lake Fulmor. Dominant taxa included several filamentous cyanobacteria Spirulina (left top), Anabaena (top center), Microcystis (left bottom and right) and the dinoflagellate Ceratium, (center bottom). Spirulina is known to undergo vertical migration and aggregation.

Figure 15. Phytoplankton present during the October 2005 plankton of Lake Fulmor. Dominant taxa included several filamentous cyanobacteria Anabaena (left top and bottom), Aphanizomenom (center top and bottom), and diatoms (left top and bottom). Aphanizomenom was a major contributor to phytoplankton biomass during the October experiment.

Considerable progress has also been realized this year with respect to the application of our robotic boat in Lake Fulmor. We have advanced the control and navigation of the boat considerably (see Multiscale Actuated Sensing section). This has allowed fully autonomous activity of the boat. In addition to adding directed sampling capability to the embedded network, the robotic boat also provides a means of obtaining spatial patterns of surface chlorophyll concentrations at very high resolution. An example is shown in Figure 16. Continuous underway fluorometry was carried out during a single, autonomous transect of the robotic boat along the length of the lake. Based on this very simple experiment, a high degree of spatially-resolved chlorophyll was obtained, with a major accumulation noted at the lower end of the lake, and smaller peaks in chlorophyll noted at various locations along the transect.

Figure 16. The robotic boat in Lake Fulmor, November 2005 (top left). Path of the boat running a long transect through the middle of the lake (top right). Contour plot of chlorophyll (arbitrary fluorometric units) data obtained from the transect.

Laboratory experiments: The laboratory component of the Aquatic Microbial Observing Systems application has focused on methodological approaches for examining the vertical distribution or microorganisms in a water column (with the intent of eventual application in the field. Our work has focused on the dinoflagellate Lingulodinium polyedrum. This alga is a major contributor to red tides within the Southern California Bight off our west coast (Fig. 5, left). Massive aggregations of the alga have implicated swimming behavior of the dinoflagellate (specifically, daily vertical migration) as a fundamental factor affecting these aggregations. We have examined the importance of this behavior in the formation of red tides of the dinoflagellate L. polyedrum.

Figure 17. Spatial (vertical) distributions of Lingulodinium polyedrum in our laboratory testbed water column. Note distinct diel pattern to the vertical position of the alga. Brown discoloration at the surface of the water column (bottom left) indicates massive accumulation of algae near the surface during the day. Light microscopy (bottom right) of the red tide dinoflagellate.

Experiments were conducted using our water column, and applying a newly developed genetic detection/enumeration approach based on the application of quantitative real-time PCR. High-resolution vertical profiling of the dinoflagellate population throughout the course of a 24-hour period have indicated that diel (daily) vertical migration plays an important role in the development of massive aggregations of this red tide alga. This work has now been accepted for publication in a leading microbial ecological journal (Moorthi, S.D. P.D. Countway, B.A. Stauffer and D.A. Caron. In press. Use of quantitative real-time PCR to investigate the dynamics of the red tide dinoflagellate Lingulodinium polyedrum, Microbial Ecology).

Future Directions

Our goals for the coming year include an extensive field program to deploy NAMOS in Lake Fulmor at the James Reserve.  Specific objectives aimed at the improvement of embedded networked sensors include (1) full coordination between the static sensor nodes (buoys) and the autonomous vehicle (robotic boat) to enable within-network decisions on boat navigation, sensing and sampling, (2) multiscale sensing/sampling using the buoys, and in coordination with the robotic boat, (3) build-out and integration into the network of a new robotic boat (and/or buoy) that possesses the ability to profile vertically within the water column (including enabling sample collection at specified depths, (4) collection, comparison and integration of multi-scale sensed data, and (5) actuated sampling (adaptive sampling) aimed at identifying discrete biological features in time and/or space.

Other objectives of this field season in Lake Fulmor include the following:
A primary goal will be the exploitation of the sensor network to make our first ‘systems-level’ measurements of the lake.  We will conduct a series of four NAMOS lake deployments spaced throughout the year.  These measurements work towards developing a model that will capture the important chemical, physical and biological structure of the lake, and thereby enable characterization of the major biogeochemical processes within the lake.  The NAMOS sensor network of static buoys has begun to produce information about Lake Fulmor that has already revealed previously unknown and unexpected observations.  The network is now sufficiently developed and robust to support novel experimental and observational studies of the structure and function of the planktonic microbial community of Lake Fulmor.  Seasonal studies in Lake Fulmor are aimed at obtaining system-level (i.e. whole lake) reconstructions of the spatial (vertical and horizontal) and temporal (scales of minutes to seasons) characterization of features pertinent to aquatic microbial ecology.  Observational measurements made by the static network and robotic boat will be complemented with measurements of nutrient chemistry, microbial biomass, water flow (to establish residence times for water in the epilimnion and hypolimnion of the lake), and rates of important biological processes.  These deployments, aided by hypotheses generated as a consequences of this past year’s deployments, will be the first true application of the network to address question of import to limnological research.

A fundamental improvement to the robotic boat will be an extension of the within-network data synthesis and visualization to include within-network decision making and using that information to define boat missions.  This feature will be designed as both an autonomous activity that will facilitate ‘smart sampling’.  We will also improve and exploit the user interface for improved manual control of the boat.  We believe that human-assisted embedded networked sensing and sampling will be essential for decisions that cannot yet be adequately made by the network.

The four NAMOS lake deployments will be conducted in conjunction with the Networked InfoMechanical Systems (NIMS) group.  This work will allow a first-ever integration and comparison between these sensing networks.  This work will be part of a thrust into developing multi-scale sensing as a component of embedded networks.  We are already well along in this endeavor.  We wish to examine the value-added of sampling at high spatial/temporal resolution (compared to the relatively sparse sampling that characterizes most biological studies in aquatic ecosystems).  The vast number of scales over which environmental processes operate make multi-scale sensing an essential component of environmental networks.  In addition, network flexibility will also constitute an important aspect of our efforts to improve network functionality for environmental sensing.  The use of network output from low-density sensors to trigger reconfiguration of the network or to trigger high resolution measurements/sampling when it is needed will be an important part of future network designs and applications.