Predicting plant dynamic rates from aerial image derived forest structure

Stephanie Bohlman and Stephen Pacala have a paper in 100:2 in the Journal titled “A forest structure model that determines crown layers and partitions growth and mortality rates for landscape-scale applications of tropical forests“.  Read it here.

Stephanie has provided a short synopsis of the paper and an example photo used in their study.

Photo credit: Richard A. Grotefendt, Tropical Forest Management Foundation,

This aerial photograph of the 50-ha tropical forest dynamics plot on Barro Colorado Island, Panama presents a blanket of subtly varying green, with an occasional splash of color.  This pattern of continuous green is in fact quite surprising if you were to look at the locations of tree trunks or stems, which are highly clumped.  If the trees were modelled as spherical lollipops rigidly placed directly above the stems, this image would not look like a carpet of green, but rather a more discontinuous surface of green spheres interspersed regularly with dark spots representing deep gaps in crown cover, and even brown spots where the soil surface would be viewed in the inter-crown gaps.  Tropical forest canopies present a relatively smooth surface of green because tree crowns do not stay rigidly above tree stems, but instead alter their shapes in order to access the highest amount of direct light.  This crown plasticity, or ability for crown displacement, in reaction to available light has been used to develop a new forest structure model for tropical forests detailed in our paper.

The Perfect Plasticity Approximation (PPA) model used in this paper and applied to tropical forests recognizes this crown plasticity and displacement can generate very simple rules about how forests are structured.  The simple rule used in this paper is that the tallest trees in a local neighborhood of forest tend to all get direct sunlight, even if their stems are directly adjacent to another tall tree.  One or both tall trees will develop their crowns away from each other, where they can access the most direct light.  The top crown layer of a local neighborhood fills up with the tallest trees crowns until it is 100% filled.  The next tallest trees fill the next layer, and so on until all crowns have been assigned to a crown layer.  In this way, the number of crown layers of a patch of forest can be calculated.  But more importantly, if all trees in the crown layer are assumed to have access to the same light level, they can be given the same growth and mortality rates and used as the basis of a dynamic forest model simulator.

The large variation in shades of green in the aerial photo, along with the variation in shapes and textures of the tree crowns, reflect the high species diversity here – 226 species in the canopy and 302 species in total. Some crowns are compact neat round circles while others are irregular, sprawling polygons. The PPA model has so far been applied in temperate systems, with each species having several parameters to determine crown shape.  To do so on Barro Colorado Island or other highly diverse tropical forests would require crown shapes for hundreds of species.  This paper explored the use of one single crown shape, or crown shapes based on only few functional types.

The image not only demonstrates some important properties of the forest, such as high species diversity and crown plasticity, but it was critical for parameterizing and testing the PPA model.  This and other photographs were taken as overlapping stereopairs, so that the forest could be viewed and measured in three dimensions.  Using a stereoplotter, the crown boundaries of each crown were digitized and height measured, producing the sun-exposed crown area and height of each crown that was “sun-exposed”.  In the field, the crowns were linked to individual stems for which species identity and a 25-year diameter history was known.  Thus, diameter-crown area and diameter-height relationships or allometries could be derived from the stereophotos, which was critical for model parameterization.  Second, by process of elimination, all tagged trees that did not appear in the canopy can be assumed to be in the understory.  The appearance of trees in the canopy and understory was used to test how well the model worked.

Overall, the model worked surprisingly well, despite its conceptual simplicity.  Also, it proved robust to using species- versus general parameterization, and to a range of plot sizes.  The model will be used to develop a fully dynamic version of the model.  Despite the well-known clumpiness of tree density in tropical forests, there were consistently three crown layers everywhere.  The variance in crown layers was surprisingly low, even lower than if all tree crowns had been distributed at random across the 50-ha plot, thus creating an even distribution of crowns.  This indicates regulatory mechanisms that maintain the number of crown layers consistently at three.  Crowns in dense areas have smaller crowns, due to competition for light and space.  Inability to survive at extremely low light levels prevents the development of more than three layers.  High growth rates in the tropical forests allow even new gaps to fill up crown layers quickly.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s