Category Archives: Tropical dry forests

What’s new in tropical ecology I: Mapping tropical dry forests

Blogs began as virtual notebooks, places people kept track of the things that caught their interest. And the flood of information makes it hard to keep track of anything on a regular basis. So why not challenge yourself?
Pick up an atlas and you’ll find vegetation maps of some sort which delineate different vegetation types (biomes, life zones, or something else of the sort). At broad scales you’ll see things like savannas and rainforests, at finer scales you’ll see things like oak-hickory forests. Everywhere you look, you’ll see vegetation mapped with what appears to be a high degree of confidence. But ever stop to wonder what those colourful areas on the map come from?

Traditionally there were two ways to map vegetation – ground surveys and aerial surveys. In the 19th century people like Charles Flauhert at Montpellier began mapping and classifying vegetation. People like Robert Smith brought this work back to Britain; after his death, his brother William Gardner Smith continued his work. W.G. Smith was one of the organising members of the British Vegetation Committee (together with Arthur Tansley and others), which later gave rise to the British Ecological Society. They used field surveys to map different vegetation types, and standardised sampling procedures to characterise the different forms of vegetation. In America, E. Lucy Braun, Edgar Transeau and William S. Cooper played a similar pioneering role in vegetation mapping.

The availability of aerial surveys, starting in the 1920s and 30s, added another dimension to vegetation surveys. Now, it was possible to collect data over a large area, but the quality of the data was low – you could probably distinguish forest from grassland, farmland from urban. As aerial photography developed it was possible to extract more information about vegetation – coniferous forest from deciduous forest, open forest from closed forest, even distinguish younger successional forest by the roughness of the canopy (younger forests tend to be dominated by even-aged stands of the same species, so the canopy is made up of relatively similar trees. Older forests are more uneven in age and species composition, leading to a canopy that varied more in height and colour.) Things began to take off as additional sensors, many of them space-based, became available. By measuring additional wavelengths, outside the visible spectrum, and by actually collecting quantitative data about the absolute reflectivity in each band, it was possible to develop much more structured data from remotely sensed data. Some of these sensors can estimate the amount of chlorophyll in vegetation, while others can measure vegetation structure.

In the March 2013 issue of Biotropica, Sebastián Martinuzzi, William Gould and others used LIDAR data to classify dry forest1 in Guánica, Puerto Rico, into “forest type” (semi-deciduous forest, semi-evergreen forest, scrub forest, dwarf forest and mesquite forest, a “relatively homogeneous stand of Prosopis pallida with a dense herbaceous understory”) and beyond that, into successional stage: mid-secondary forest with a logging2 past, late secondary forest with a logging past, late-secondary forest with an agricultural past, and ‘primary’ forest (undisturbed for more than 90 years). This part of their analysis they restricted to semi-deciduous forest areas. Since LIDAR penetrates the forest canopy, it can identity the underlying land surface (and thus, elevation and topography), and the height not only of the forest canopy, but also of various layers in the canopy.

It’s hard to explain how appealing the idea of being able to map and age these forests using remote sensing can be. I’ve mapped forest cover using aerial photographs – first you map current forest cover, then you map historical forest cover, and use that to estimate areas of “older growth” forest. Not only is it tedious, there’s also a risk of ending up with subjective results. If an area was cut-over and regrew in the interval between two images, you may never noticed it. And if an area lost all its understorey without losing its canopy, you’d probably be unaware.

While avoiding subjectivity in that regard, this approach added a different sort of subjectivity.

We used GPS surveys and visual interpretation of 1-m spatial resolution color aerial photos supported by expert knowledge, for a total of 83 sample locations. All samples represented an area of at least 30 m by 30 m of the same forest type (to coincide with the geospatial grain size used by this study), and were separated by >60 m (consistent with Agosto Diaz 2008).

As far as I know, the definitions of the different forest types in Guánica Forest are subjective. They are somewhat self-evident, as long as you stay in the park (and avoid certain grey areas), but I’m not aware of any formal delineations of these vegetation types. Granted, there’s nothing unique about that, but having poked around in Guánica Forest, uncertain (at times) whether I was in one forest type or another…I’m a tad bothered.

Perhaps the thing that interested me most was this

The most important predictors in the LiDAR canopy model included the median absolute deviation of vegetation heights (HMAD), the 90th percentile of vegetation heights (H90th), and the percent of returns >1.0 m (CDENSITY2; Table 3).

Canopy closure is a useful, but incomplete predictor of forest type. The most closed canopies tend to be in ravines and arroyos (semi-evergreen forest), but certain young secondary forests could also have a closed canopy (like the Prosopis forest they mentioned). But when you combine that with with the variation in vegetation height, you can probably separate out the species-poor secondary forests dominated by Prosopis or Leucaena leucocephala, since these forests lack structural diversity – most of the trees are the same age and belong to just a single species. This has me wondering…

  1. Martinuzzi, Sebastián, William A. Gould, Lee A. Vierling, Andrew T. Hudak, Ross F. Nelson, and Jeffrey S. Evans. 2013. Quantifying Tropical Dry Forest Type and Succession: Substantial Improvement with LiDARBiotropica 45(2): 135-146
  2.  Logging in this case means harvest for fence posts and charcoal production; this process left the rootstocks intact and allowed rapid coppice regeneration.
  3. Lugo, Ariel E., Jose A. Gonzalez-Liboy, Barbara Cintron, and Ken Dugger . 1978. Structure, productivity, and transpiration of a subtropical dry forest in Puerto Rico. Biotropica10: 278–291.

Disturbance and recovery in tropical dry forests

[Repost from my old blog]
When people think about the destruction and degradation of tropical forests, they tend to focus on rainforests. Tropical dry forests tend to get overlooked. They aren’t as striking – no cathedral-like understorey, no mind-boggling biodiversity. But more importantly, they often just aren’t there. Over much of their potential range they have simply been erased from the landscape. They may have covered as much as 42% of the land area in the tropics1, but have been reduced to less than 27% of their former range in Mexico2, and as little as 2% in Central America3 and New Caledonia4.

Despite the fact this, tropical dry forests are often seen as being quite well-adapted to human disturbance. Being less species-rich than wetter forests, they tend to support fewer rare species, and may be less extinction-prone. In addition, dry forests are dominated by trees that sprout after being cut. This means that if you cut down a patch of dry forest, most of the stumps will re-sprout. This type of recovery is much quicker than you would get if the trees had to germinate from seeds – not only does it take much longer for seedlings to grow large (stump sprouts can draw on resources stored in the roots of the tree), but there’s likely to be a time lag as seeds disperse into the area from surviving trees (tropical forests tend to lack long-lived seedbanks).

Much of our understanding of succession in tropical dry forests comes from Jack Ewel’s dissertation work. Ewel looked at the effect of cutting and herbicide application on succession in a series of plots across the Neotropics. One of his important findings was the dry forests were quicker to recover their stature that wetter forests. Since most of the recovery comes from stump sprouts, the recovering forest is also close to the original forest in terms of species composition.

While lightly used dry forest sites recover rapidly, recovery is slower in more intensively used sites. Seedling survival rates are very low in dry forests – while seedlings establish in the wet season, most (often all) of them die in the subsequence dry season. So while intensively used sites in Guánica Forest recovered well in terms of structure, biomass and leaf fall in 50 years after abandonment, the recovery of species composition was very slow6.

Resilience is the rate of recovery of disturbed sites to their pre-disturbed state. Ewel’s work helped to establish the idea that dry forests are more resilient than wetter forests. But there is no single rate – or pathway – of recovery. Measures of “recovery” depend on the parameter measured – canopy height, biomass, species richness, nutrient cycling… It also depends on the baseline against which recovery is measured: if the same site is measured before and after disturbance, you need to know if the site represented “mature” forest before disturbance. If another site is used, you need to wonder if it is really representative of initial conditions in your experimental plot.

In an article7 published in the journal Biotropica, Edwin Lebrija-Trejos and coauthors looked at what it really means to say that tropical dry forests are more resilient than wetter forests. They looked at a sequence of 15 sites in Oaxaca, Mexico, which had been cultivated and then abandoned for 0-40 years, and compared them with nearby mature forest. All of the sites had been cultivated for a short period (1-2 years) and then abandoned without being converted to pasture8. They considered a variety of different ways to measure resilience – they looked at forest height, plant density, basal area (the area occupied by tree stems), crown cover, species richness, species density (number of species per 100 m2), Shannon evenness and Shannon diversity. Not surprisingly, they found that certain features (canopy height, plant density, crown cover) recovered rapidly (in less than 20 years) while others (including basal area and species richness) had not recovered after 40 years.

When compared their sites with other comparable studies, they found that their sites were among the quickest to recover canopy cover and height. On the other hand, they found that their sites were among the slowest to recover species diversity and average in terms of the recovery of species richness. Overall, in terms of the structural measures that Ewel focussed on, it’s reasonable to conclude that dry forests are more resilient that wetter forests. On the other hand, with regards to things like basal area and species richness, the assertion of resilience for dry forests isn’t well supported.

  1. Brown, S., and A. E. Lugo. 1982. The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14:161-187.
  2. Trejo, I., and R. Dirzo. 2002. Floristic diversity of Mexican seasonally dry tropical forests. Biodiversity and Conservation 11:2063–2084
  3. Janzen, D. H. 1988. Tropical dry forests: The most endangered major ecosystem. In E. O. Wilson (Ed.). Biodiversity, pp. 130–137. National Academy Press, Washington, DC
  4. Gillespie, T. W., and T. Jaffré. 2003. Tropical dry forests in New Caledonia. Biodiversity and Conservation 12:1687–1697.
  5. Ewel, J. J. 1971. Experiments in arresting succession with cutting and herbicide in five tropical environments. Ph.D. University of North Carolina, Chapel Hill.
  6. Molina Colón, S., and A. E. Lugo. 2006. Recovery of a subtropical dry forest after abandonment of different land uses. Biotropica 38:354–364.
  7. Lebrija-Trejos, E., Bongers, F., Pérez-García, E.A., Meave, J.A. (2008). Successional Change and Resilience of a Very Dry Tropical Deciduous Forest Following Shifting Agriculture. Biotropica 40(4):422-431 DOI: 10.1111/j.1744-7429.2008.00398.x
  8. Conversion to pasture tends to slow recovery significantly; not only does the prolonged period eliminate almost all root stocks, it also establishes a grassy layer that makes it more difficult for tree seedlings to establish.

Survival and rebound of Antillean dry forests: Role of forest fragments


Antillean dry forests have experienced high levels of human impact for almost five centuries. Economic changes in the second half of the 20th century have facilitated forest recovery in Puerto Rico. We quantified the extent of forest cover and the community composition of representative forest fragments in the subtropical dry forest life zone (sensu Holdridge, 1967) in southwestern Puerto Rico. Forest cover, which was largely eliminated by the 1940s, stood at 48% in the western dry forest life zone in 1993. Fragments varied in land-use history and supported from 1% to 86% of the reference species sampled in Guánica Forest, a 4000-ha protected area. Reference species were well represented in forest fragments, even those smaller than 1 ha, if they had never been completely cleared, but were uncommon in forests regenerating on previously cleared sites. The studied fragments are novel ecosystems which combine native and introduced elements; Leucaena leucocephala (Lam.) De Wit, an introduced legume, was the most common species, regardless of land-use history.

I.A., Murphy, P.G., Burton, T.M., Lugo, A.E.  2012. Survival and rebound of Antillean dry forests: role of forest fragments. Forest Ecology and Management 284:124-132.

Full article: ScienceDirect (subscription) or International Institute of Tropical Forestry.