Remote Sensing is a powerful tool and can be used as a time machine to track biodiversity loss
Primary forests, or old-growth forests as they are sometimes called, are epicenters of rich biodiversity, are more resilient than younger forests, and store significantly more carbon than their younger counterparts, to name just a few of the vital roles of these essential and irreplaceable ecosystems. The preservation of primary forests is the focus of global conservation efforts.
The UConn Global Environmental Remote Sensing (GERS) Lab has developed a new remote sensing method to continuously monitor primary forest loss and determine what factors are driving that loss. Their findings are published in Remote Sensing of Environment.
Lead author and Department of Natural Resources and the Environment Ph.D. student Falu Hong says that they focused on these key habitats on the island of Hispaniola, which includes Haiti and the Dominican Republic, using satellite images from the years 1996-2022.
"We used a satellite time series to track primary forest loss, and we focused on these two countries because they have experienced significant primary forest loss and because they are ignored in previous studies, especially Haiti, which is one of the hotspots of biodiversity loss," says Hong. "We analyzed the forest loss over 27 years of land cover change, which has not been done in previous studies."
The researchers analyzed multiple dimensions of forest loss, including the primary forest inside and outside of protected areas and the drivers of forest loss. They applied a method called the COtinuous monitoring of Land Disturbance algorithm (COLD) and remote sensing data from Landsat to create a map of the primary forest loss.
Ji Won Suh, a postdoctoral researcher in the GERS lab, says this study showcases the power of using Landsat time series data.
"So few studies focus on primary forests because it is very difficult to map them using remote sensing signals. Sometimes it is difficult to differentiate a secondary forest or regenerated forest from a primary forest, but this study successfully classified those primary forests using a random forest machine learning model."
Suh says the accuracy of the map was verified by their collaborator and co-author S. Blair Hedges from Temple University, who is an expert on primary forests on Hispaniola Island.
"Another unique part of this study is we created a primary forest map over time," Suh says. "Usually other studies just focused on a one-time event. We can track the loss of primary forests over many years. Our study is a way where we can map the trajectory of loss as it happens and we can analyze why those losses happen."
They found the main drivers of primary forest loss in Haiti are fire, which caused around 65% of the observed losses, followed by logging which accounted for about 20% of the primary forest loss, and around 10% of the forest loss was attributed to hurricane damage.
"We found that in 2016, Hurricane Matthew destroyed around 12% of the primary forest in Haiti, just in one year," says Hong. "That's a huge amount of loss. With our map we can visualize the primary forest change and analyze the drivers causing that change. We can also analyze forest fragmentation. Usually, primary forests are homogeneous, but activities like construction or logging result in the forest becoming more and more fragmented. We quantified the fragmentation level of the primary forest which could give good insight into biodiversity conservation and preservation."
They also found that primary forest fragmentation is more pronounced in Haiti, where patches of primary forest are smaller and less numerous. Primary forests in both Haiti and the Dominican Republic are located on steep terrain, indicating that primary forests located in flatter and more accessible areas are prone to development and forest destruction.
This paper is the first step in a larger project, says Hong, where the next steps are to begin expanding the mapping across the Caribbean region to evaluate the impact of primary forest loss on biodiversity change.
GERS Lab Director and Associate Professor in the Department of Natural Resources and the Environment Zhe Zhu says that as primary forests have the lion's share of biodiversity, many of the species living there are also endangered, so the preservation of these irreplaceable ecosystems is paramount. Having a reliable method to map primary forests accurately will help in the effort,
"One thing I want to emphasize about this work is that it is very difficult to identify between different forests like primary dry forests, primary wet forests, and secondary forests, for example. A primary forest may look very similar if the secondary forest is old enough. You can have very subtle human disturbances causing it to no longer be a primary forest. You need to know the driver and how severe the drivers are. You also need to know the resilience of the trees."
This work is supported by a $2 million NSF grant with the goal of linking remote sensing to track biodiversity through time.
"We are treating remote sensing as a time machine to backward and forward to forecast future impacts on biodiversity. It is a very fun project that a lot of us in the GERS lab are working on," says Zhu.
Tracking the impacts on biodiversity and the drivers of change is important for conservation and policymaking, and studies like this can yield surprising results and insights into what needs to happen to preserve vital ecosystems like primary forests.
This work was supported by a grant from the NSF Biodiversity on a Changing Planet (BoCP) program (2326013 and 2326014).
