Skip to main content

Modeling Tropical Forests: An Interview with Dr. Maria Uriarte

Sporadic weather events can alter the structure of forests. When subject to intense climate conditions, trees face new survival obstacles. The recent tsunami activity in Southeast Asia is testimony to nature’s ability to alter the landscape. In many affected regions, such as India, vegetation was stripped from coastal areas. In the tsunami’s wake, a range of trees, from mangroves to cultivated bananas, was damaged.

Post-tsunami, will different species become dominant? Will land use patterns influence how forests recover? These are the types of questions explored by IES Postdoctoral Associate Dr. Maria Uriarte. Working with IES forest ecologist Dr. Charles D. Canham, she is developing statistical tools to help understand how forests respond to hurricanes. This summer, Dr. Uriarte will begin teaching at Columbia University. Recently, we discussed her research in Puerto Rico’s Luquillo Forest.  

Why are tropical forests important?

Tropical forests are being cut down at a rapid rate; understanding their dynamics is crucial to sound management. They contain an estimated 50% of the World’s biodiversity and they help regulate the carbon cycle. Fossil fuels release carbon dioxide when they are burned. Increased carbon dioxide levels are linked to climate change. Tropical forests, particularly young ones, have the potential to sequester and store large amounts of carbon.

Why are models useful for understanding forest dynamics?

Forest change is a slow process. Trees often outlive ecologists, making forests difficult to study in real time. Models allow ecologists to simulate how environmental changes, such as invasive species or severe weather events, might impact forest dynamics.

How did the project in Puerto Rico originate?

Charlie Canham developed a model for interpreting how trees grow in response to their neighbors. His neighborhood dynamics model is being used in British Columbia to help develop sustainable forestry. There was an interest in applying it in Puerto Rico’s Luquillo Forest, in an effort to understand how hurricanes shape forests. I began working on the project in 2002.

Describe neighborhood dynamics.

I would define it as how trees respond to their neighbors based on their individual characteristics and who is growing near them. Trees are sedentary. If they find themselves growing in an inhospitable environment, they can’t relocate. In a forest, the trees growing within a certain radius around a tree often determine its survival success. By looking at the trees growing in this “neighborhood” you can gain insight into dynamic processes, such as growth or survival. For instance, a shade intolerant tree is less likely to survive if it is surrounded by tall trees with dense canopies.

How does the model work?

Forest modeling requires inventorying the trees on a plot and accounting for the variables that regulate their survival. Data are recorded on things like tree size, light availability, species composition, the spatial position of trees within the plot, and so forth. By entering this information into a computer simulator, we can ask scenario-based questions that quantify how different variables impact tree growth and survival. In the case of Puerto Rico, we could ask— what will the forest look like if we have a severe hurricane twice in the next decade? 

Tell me about the site you are studying.

We are researching Caribbean forest dynamics on a 16-hectare plot in Puerto Rico’s Luquillo Forest. The plot is part of the Smithsonian Institution’s Center for Tropical Forest Science, a network of 18 research forests throughout the tropics. By standardizing data collection— trees are mapped, measured and identified in similar ways— the network is generating the research needed to understand the processes driving tropical forest dynamics. In 1934, the U.S. Forest Service purchased the land that houses the Luquillo plot. Aerial photos taken at that time provide a window into the forest’s past. The southern two-thirds of the plot, which is relatively flat, had been extensively logged and farmed. Due to steeper terrain, the northern end of the plot was left relatively intact.

What makes the plot well suited to studying forest dynamics?

Globally, human land use is arguably the most pressing environmental issue facing tropical forests. Decades after human disturbance has ceased, land modification can leave a lasting footprint. We are interested in how human land use impacts the way forests recover from hurricanes, and how hurricanes affect human land use legacies. The Luquillo plot has a rich history of human-induced land use legacies and hurricane exposure.Good models require extensive survey data. The Luquillo plot has been inventoried three times: following Hurricane Hugo in 1990, in 1995, and in 2000. A second hurricane, Hurricane Georges, struck the forest in 1998. There are 140,000 trees that have been mapped, measured, and identified. Damage sustained from the two hurricanes has also been documented. Very few tropical forests have been surveyed to this extent.

What's the connection between land use and hurricanes?

When humans modify the land, and slow-growing shade-tolerant trees are removed, forest structure is altered. After logging or farming activities cease, shade-intolerant species tend to dominate forests. Their wood is less dense than shade-adapted trees, making them vulnerable to snapping during hurricanes. In the event of a severe hurricane, areas with a history of human land- use will suffer greater damage than areas of undisturbed forests.

How do hurricanes shape tropical forests?

The most obvious effect of hurricanes is broken trees— trees are uprooted, their crowns are damaged, or they are crushed when a neighboring tree falls on them. Immediately after a hurricane there is a lot of damage, but few trees actually die. In the short term, a tremendous amount of debris  falls on the forest floor, burying seedlings. In the longer term, gaps in the forest canopy change regeneration patterns. In the absence of disturbance, canopy gaps are rare in tropical forests. If you go to Panama, for instance, gaps are infrequent events that occur when a tree dies, through age or disease. In Puerto Rico, hurricanes knock over large numbers of trees. Post-hurricane, a lot of light reaches the forest floor. Young saplings, many of them shade-intolerant, vie for a position in the canopy. Only a few are successful; our plot has a high rate of sapling mortality. 

What makes gap species more prone to hurricane damage?

Trees that exploit canopy gaps need to grow fast to prevent being shaded out by their neighbors. They grow quickly by producing low-density wood; high-density wood takes longer to build. The problem is, as mentioned earlier, tall low-density trees are prone to breaking. Slow growing, shade tolerant species are much more hurricane-resistant.

What challenges have you faced?

Most forest models have been developed for temperate zone forests, where species diversity is low and forests are well inventoried. Tropical forests have much higher diversity, but we know less about them. In Puerto Rico, 90 of the 140 species on our site are rare, with less than one tree per hectare. You need to survey much more land to come up with a model that provides insight into how rare species respond to environmental changes. 

Is modeling limited by technology? 

We are working with huge databases. When we model how 140,000 trees respond to a given change, each time the computer simulates the forest it has to account for every tree. These models were not possible 15 years ago; we are able to build them now because computing power has become fairly cheap.

Can models really be predictive?

There are so many factors regulating forest dynamics— it is almost impossible to predict exactly how a single change, such as a rise in temperature, will impact a forest. That said, models help us explore scenarios and organize our thoughts. They force you to be explicit about what you are thinking; you decide on the factors that go into a model and the relative importance they have. Models also provide us with probabilities that scenarios will occur. Many of the things we do in life, from investing to medical decisions, are based on probability.

How did you come to be an ecological modeler?

When working on my PhD, I did a little modeling and quickly became intrigued. Each dataset presents a new challenge, as does generating the equations that drive the models. There are three components to research— collecting data, analyzing data, and writing papers. For me, the most interesting part is analyzing data and writing papers. Spending six months in the forest is not my passion, but if you give me a dataset I can find interesting things.

What advice would you give to a student interested in pursuing this type of work?

Don’t be intimidated by math, there is nothing magical about it. Take a lot of statistics and computer classes and be prepared to spend a lot of time in front of your computer. It is a rewarding field for people who like puzzle- oriented problem solving.