This study used a combination of field survey and remote sensing techniques to produce a vegetation community map. A preliminary classification of the atoll was produced before visiting it using the remotely-sensed data. This classification was tested and revised in the field. Upon return to the UK ordination and statistical clustering techniques allowed us to further check the validity of these field classes. With the vegetation classification thus confirmed, the information was added to a GIS (Geographical Information System). Combining these observations with digitised maps of Turneffe and the overlays produced from the aerial photographs allowed us to produce vegetation and land-use maps for the atoll.
Two sources of remotely-sensed data were used - satellite images and aerial photographs. Before the visit, we examined a Landsat TM image of Turneffe. This established general vegetation zones, but because of its relatively coarse resolution (30m x 30m pixels) it could not be used for identifying individual trees. The Lands and Survey Department of Belize allowed us access to their aerial photograph collection. At a scale of approximately 1:40,000 and 1:10,000 these black and white photographs provide the necessary resolution to identify the vegetation to species level. They also clearly show sand suspended in the creeks, useful for establishing sediment transportation patterns. These images were supplemented by a series of colour oblique photographs of the islands, taken by the authors during previous research visits.
|The top left image shows a black and white aerial photograph. The two dark areas are small mangrove covered cays. The white area around the cays is sand showing through the shallow sea. The edge of the barrier reef is picked out by the line of breaking waves to the right.
The bottom left image shows an oblique close-up of two small cays, taken by hand from a light aircraft. In this colour picture, small features such as individual trees, larger coral heads and structures such as huts and piers can be picked out.
The image to the right hand side shows a satellite image of the entire atoll. The display shows the atoll in "false colour" - the colour of an area does not match that seen by the human eye. Instead it is generated by computer to represent a particular combination of different wavelengths of light detected by the satellite sensors. Surface features absorb and emit light at different wavelengths, giving them a "spectral signature". By using different combinations of wavelengths it is possible to begin classifying the vegetation.
Published maps of Turneffe are relatively rare, most such as the Admiralty chart are designed for marine navigation and give little information about the vegetation. Three sources that we found useful however, are:
The maps produced by David Stoddart (who examined the impact of Hurricane Hattie upon Turneffe in the mid-Sixties)
A provisional copy of a new geo-referenced map of Turneffe being developed at the time by the Land Information Center, Lands and Survey Department
The second edition Belizean National Mangrove Maps were used to give an impression of the distribution of mangrove forest on the atoll.
|The National Mangrove Map is an example of a vegetation classification produced using a combination of satellite image processing, aerial photograph interpretation and field survey. It was created by members of the Department of Geography at Edinburgh University between 1990 and 1992.
The image to the right shows the classification for Turneffe Atoll. The mangrove cover is classified by height and density: Tall dense mangrove areas are shown in black; Medium dense mangrove in red; Areas of dwarf mangrove are shown in blue.
Limited fieldwork time available during the production of the original National Mangrove Map meant that there had been relatively little field checking carried out on Turneffe. Thus a subsidiary aim of this project was to check and if necessary correct these maps.
Sampling locations were selected to ensure representativeness - we wanted to be sure that the sites sampled included all the vegetation zones identified from the satellite images and aerial photographs. Also, we had to consider the effect of differences in the geomorphic environment - leeward and windward sides of the atoll, coastal and lagoon locations.
In total fifty-five samples were obtained from over thirty-five locations. The low, flat shape of the islands means that for large parts of the atoll there are very few landmarks suitable for accurate position finding. Instead, sample locations were plotted using hand-held GPS receivers.
The vegetation at each site was recorded, noting the main species which form the canopy, shrub and herb layers. Identification was aided using herbarium collections held by the Forest Department in Belmopan and at the Royal Botanic Garden, Edinburgh. As well as red and black mangrove species (Rhizophora mangle and Avicennia germinans), trees typical of drier, truly terrestrial conditions such as tea-box (Myrica cerifera) and red gumbolimbo (Bursera simaruba) were recorded.
|During the survey, sites were visited by boat.
A skiff with a shallow draft was chosen, to allow access to the winding creeks.
|Red mangrove (Rhizophora mangle) is the most common plant species found on Turneffe Atoll.
This picture shows a single specimen growing 10 metres from the shore in water approximately 50 cm deep.
Red mangrove is easily recognised in the field by its characteristic prop roots, which help to anchor the tree in the ground.
The soil at each site was carefully described, the depth of any standing water recorded and notes made about the surrounding geomorphic environment. The soils were primarily derived from coral rubble and sand, sites covered by mangrove forest had a characteristic surface layer of peat. At many sites the "soils" were only very poorly developed - little more than a shallow mineral layer. Others showed a greater organic content, with several distinct horizons. The soils were named using the Belizean Soil Classification System developed by Bruce King et al. Soil and water samples were taken measuring pH, conductivity and the level of sulphates and sulphides. Water samples were also taken from wells where available. These data provide information about the different physical and chemical environment available to plants across the atoll.
|In some areas the soils were so poorly developed that a simple visual inspection was sufficient to describe them.
In areas where vegetation input results in a more mature soil, pits were dug to expose a profile for description. These were filled in later, to prevent them accidentally trapping animals.
In well drained sites, water samples had to be extracted with a syringe.
The soil and vegetation data were used for ordination analysis. This arranges the plant species in a mathematical space produced using the soils data. Thus it is possible to establish the preferred soil conditions for each plant species. These can be combined into "environmental gradients" which describe the general patterns. At Turneffe, vegetation groups could be differentiated according to two environmental gradients. The first separated sites into well-drained and water-logged. The secondly differentiated between sheltered, low energy sites with acidic soils, exposed, high energy sites with alkaline soils.
The vegetation and soil classes produced in the field were tested using numerical clustering techniques. These take all the sample data and arrange them into groups according to how similar the sites are to each other. The grouping of the sample sites can be shown graphically on a dendrogram.
|This dendrogram shows the sites grouped by soil type.
The sample sites are arranged along the right hand side of the graph and similar sites are connected by the coloured lines drawn to the left.
Branches to the left hand side of the figure represent divisions into broad groups, branches at the right hand side split up the sites into smaller and smaller sub-groups.
The four pink horizontal lines indicate the groupings defined by a cut-off of level of 0.8 - this splits the samples into five groups.
The numerical classifications were very similar to those produced in the field. This provides extra confidence in the final output - the vegetation and land-use maps.
All the data from the survey was brought together using Arc View GIS. A digital version of the Land Information Center map was used as a base. On this, an overlay was added showing the vegetation communities. Observations about the soil conditions, geomorphology and land-use at individual sites were added as further "layers" of information.