I have various past and ongoing projects aimed at understanding how species interactions that form the ecological feedbacks within ecosystems are modified under changing disturbance regimes.
Life after death: material legacies of foundation species shape demographic processes
Foundation species are organisms that create structure within ecosystems and provide habitat for other organisms. Examples include trees that form forests, oysters that form oyster reefs, or stony corals that form coral reefs. Foundation species typically dominate their respective ecosystems in biomass and/or abundance and therefore have a strong influence over ecosystem dynamics, both while they live and after they die. The structural remains of foundation species are a type of material legacy, and these legacies can persist in the environment for long durations and influence processes tied to ecosystem resilience.
Left: Material legacies of foundation species. Clockwise from bottom left: dead grass litter in a grassland; standing burned trees in a forest; dead branching coral skeletons on a tropical reef; dead oyster shells on an oyster reef.
Left: Material legacies of foundation species. Clockwise from bottom left: dead grass litter in a grassland; standing burned trees in a forest; dead branching coral skeletons on a tropical reef; dead oyster shells on an oyster reef.
To understand the breadth of these influences in nature, I am synthesizing time series data from a collection of NSF Long Term Ecological Research (LTER) sites that represent both marine and terrestrial ecosystems spanning the tropics to the Arctic Circle. My goal is to build a more unified and generalized understanding of material legacies across disparate ecosystems, and how global change is modifying their effects.
Right: (clockwise from top left image): BNZ = Bonanza Creek; KNZ = Konza Prairie; HFR = Harvard Forest; VCR = Virginia Coastal Reserve; GCE = Georgia Coastal Ecosystems; LUQ = Luquillo; FCE = Florida Coastal Everglades; MCR = Moorea Coral Reef; SONGS = San Onofre Nuclear Generating Station Mitigation Monitoring Program; AND = H. J. Andrews Experimental Forest.
Right: (clockwise from top left image): BNZ = Bonanza Creek; KNZ = Konza Prairie; HFR = Harvard Forest; VCR = Virginia Coastal Reserve; GCE = Georgia Coastal Ecosystems; LUQ = Luquillo; FCE = Florida Coastal Everglades; MCR = Moorea Coral Reef; SONGS = San Onofre Nuclear Generating Station Mitigation Monitoring Program; AND = H. J. Andrews Experimental Forest.
Coral reef resilience under changing disturbances
A focus of my research has been to compare the impacts of different disturbance types on coral reef resilience. The bottom panel shows a typical, healthy coral reef covered in living coral colonies. Historically, the main form of disturbance on coral reefs has been tropical storms (e.g., hurricanes and cyclones), which create powerful waves that break apart coral colonies and scour them from the reef, leaving behind flattened reefscapes (left panel). In recent decades, there has been a rise of marine heatwaves (that cause coral bleaching), which kill corals but leave their stony skeletons intact on the reef (a type of material legacy) and instead leave behind structurally complex reefscapes (right panel). Through various studies, I have shown that the reefscapes left after marine heatwaves (where dead coral skeletons are left in place) disrupt key ecological processes that support coral resilience, and thereby undermine coral reef recovery. Encouragingly, I have also found that removing dead skeletons from the reef after a heatwave can improve reef recovery.
Underwater remote sensing
While remote sensing technologies have greatly expanded our study of terrestrial systems, environmental and logistical challenges have made using these technologies underwater more challenging. Fortunately, some recent breakthroughs have made remote sensing a more reliable tool for studying marine ecosystems. With an international team of remote sensing specialists, physicists, and engineers, I have developed novel techniques for using underwater remote sensing to detect ecological change on coral reefs.
|
Taking photographs of an experimental reef plot. Photos are later "stitched together" to create a photomosaic and a 3D model of the reef (see video below). The metal pole with a number '6' on it is what we call a fixed reference point. Because GPS doesn't penetrate underwater, we have to create our own reference system in order to create scaled maps of the reef and line up maps from different time points of the same location.
|
|
Video of a 3D model of an experimental reef plot. The model is built from ~300 individual photographs and allows for measuring objects, such as corals, at very high (sub-centimeter) spatial resolution. This enables us to much more accurately quantify changes in coral cover over time, especially in response to disturbance, and more effectively understand the consequences of this change. |
|
Combining AI and remote sensing to quantify ecological change
Remote sensing tools have expanded our ability to make measurements of the environment, but this can be a very labor- and time-intensive process due to the sheer volume of data that remote sensing produces. To reduce this bottleneck, I have collaborated with the Visual Computing Lab (based in Pisa, Italy) to create a pipeline for processing large imagery that uses the AI-assisted image segmentation software, TagLab. This program automates the process of measuring (segmenting) objects within images while maintaining high accuracy and spatial resolution (sub-centimeter), and measures the three-dimensional surfaces of objects. We have used this program to efficiently analyze large photomosaics produced from our underwater remote sensing techniques, allowing us to detect fine-scale changes in reef communities (e.g., from disturbance) at the scale of reefs, but with the resolution of coral colonies.
The TagLab user interface showing a healthy, 5 x 5 meter coral reef plot in 2018 (left) and the same plot after a marine heatwave in 2019. All 910 coral colonies in each photomosaic (pink = live, brown = dead) were measured fully automatically.
Examples of different types of changes in coral colonies (i.e., 'colony fates') that are automatically detectable using our workflow.