Report

A Loiczview Application to Map Enviromental Proxies for Coral Reef Distribution in Mexico, Central America and the Caribbean

By: Jose D. Carriquiry

(May 1, 2001)

DESCRIPTION OF CLUSTERING PROCEDURES:

The Loiczview porgram was initially used for zones 9 + 10, using many different atmospheric, geomorphic and coastal variables during the first clustering attempts.

Variables were systematically eliminated until a minimun set was established that was capable of creating clusters related to coral reefs distribution. The critical variables needed to generate coral reef distribution clusters were: (1) Average Elevation Bathimetry; (2) Average Minimum SST, (3) SST Range (max-min), (4) Minimum Monthly Salinity, (5) Maximum Monthly Runoff.

Variables had to be weighted: 3xSST_Range and 2xMin_Monthly_Salinity.

The number of clusters was also critical. It was found that 5 clusters using these variables always yielded more realistic cluster separation.

RESULTS:

1. Two clusters were needed to explain the distribution of most of the Pacific and Caribeean coral reefs (East Pacific coral reefs did not fall in the same cluster as those from the Mexican Caribbean, although many other Caribbean coral reefs did).

2. The distribution of most Coral Reefs in the Eastern Tropical Pacific was strongly related to a steep bethimetric slope (i.e., average elevation bathimetry). Apparently, coastal areas off the southern tip of Baja California, off cabo Corrientes (Puerto Vallarta), off the western Mexican coast, off Central America (Honduras, Costa Rica and Panama), as well as off all the oceanic islands (Revillagigedo, The Galapagos and Hawaii) present a steep slope. This can be thought as an opposite variable to gentle sloping offshore areas that usually are associated to mangroves, coastal lagoons, coastal plains and beaches. All these latter coastal types would prohibit coral reef development. That is why, coral reefs are more commonly associated to steeper bathimetric relief. Most of the coral reefs in the Caribbean seem to be related to this type of coastal submarine geomorphology.

3. The second coral reef cluster explain additional (but not all) coral reefs off the Yucatan penninsula.

4. Of the five clusters formed, cluster 3 and 4 (the two that clustered areas populated by coral reefs) showed the greatest similarity between them. Cluster 3 (Caribbean + Pacific) had a gentler slope than cluster 4 (Pacific). However, both clusters showed in common: (1) intermediate values in mean monthly SST (26.3 to 27.3 C); (2) showed lowest SST variability with a range of 5.1 to 5.4 C (more stable, thermally); (3) Minimum Salinity was 34.9 to 35.0; (4) intermediate runoff. Runoff, however, did not seem to contribute much to the clustering (eliminating it did not change too much the distribution, but keeping it in presented a slightly better distribuion)

5. The variables used for establishing clusters with a distribution resembling those of coral reefs are shown below.

STATISTICS:

Variables Used:	Cluster 3	Cluster4	Cluster's Mean
Cluster Symbol Color	PINK	Light BROWN	Mean	StdDev
Region	Pacific	Pacif+Caribbean
AVG_ELEV_BATH:	-1380.95	-3010.12	-2195.54	1151.997
SST_MEAN_MONTHLY:	27.2563	26.3279	26.7921	0.656478
SST_MIN_MAX_RANGE:	5.16679	5.48952	5.328155	0.228205
SALINITY_MIN_MONTH:	35.0171	34.9731	34.9951	0.031113
RUNOFF_MAX_MONTH:	107.89	83.5653	95.72765	17.20016

6. The cluster map showing the distribution is:

Note: In order to clearly see the colors for the resulting clusters, please increase the zoom (upper right corner of Word menu) to 200%. The resulting clusters that coincide with areas known containing coral reefs are: BROWN (Pacific Islands: Hawaii, Revillagigedo, Galapagos) and PINK (Fringing Pacific and Caribbean coral reefs).

7. In order to see how well the set of variables obtained to cluster coral reef areas in the Pacific and Caribbean worked in other reefs of the world, a new attempt was made using the same set of variables (with the same weighting as before) to map the coral reefs off Australia and Oceania. As a refresher, the variables used (and their weight) is given next: Avg_Elev_Bath (1x), SST_mean_monthly (1x), SST_Min_Max_Range (3x), Sal_Min_Month (2x) and Runoff_Max_Month (1x). Also, because the Great Barrer Reef in Australia is not a fringing reef system that could be found in the coastal cell, this reef system is located offshore in the first oceanic cell. For this reason, in addition to the coastal cell, the new clustering attempt for Australia and Oceania included also the first oceanic cell. The results obtained are somehow encouraging. One cluster (cluster # 7, Light BROWN color symbols) was obtained that followed the distribution of coral reefs in the Australia-Oceania region. This cluster maps some of the most important coral reefs in: the Great Barrier Reef (Australia), New Guinea, New Caledonia and Vanuatu, Fiji, and Indonesia in lesser extent.

Note: see the distribution of cluster #7 (Light BROWN) and cluster #6 (light BLUE) as a proxy of the coral reefs in this region. Increase the zoom to 200% in order to clearly see this cluster off NE Australia (GBR), New Guinea and Indonesia.

8. The average value for each of the variables in this cluster

Variables Used	Cluster 6:	Cluster 7:	Clusters-mean
	mean	Mean	Mean	StdDev
AVG_ELEV_BATH:	-4269.66	-2150.56	-3210.11	1498.43
SST_MEAN_MONTHLY:	28.7171	26.9835	27.8503	1.22584
SST_MIN_MAX_RANGE:	3.99973	5.30375	4.65174	0.922081
SALINITY_MIN_MONTH:	34.1381	34.7829	34.4605	0.455942
RUNOFF_MAX_MONTH:	405.5	321.774	363.637	59.20322

9. The average value for each of the variables in this cluster are somewhat similar considering the overall variability in the settings of the regions under comparison (SE Asia, Australia and Oceania versus Eastern Tropical Pacific and Caribbean). The main differences are in the existing runoff (larger in SE Asia, NE Australia and Oceania) as well as in the bathimetric relief (larger also in the latter region because of geomorphology of tectonic basins upon which reefs develop).

Variables Used	Pacific+ Caribbean	Australia+ Oceania
AVG_ELEV_BATH:	-2195.54	-3210.11
SST_MEAN_MONTHLY:	26.7921	27.8503
SST_MIN_MAX_RANGE:	5.328155	4.65174
SALINITY_MIN_MONTH:	34.9951	34.4605
RUNOFF_MAX_MONTH:	95.72765	363.637

10. Limitations.

a) Nota all Coral Reefs could be map effectivelly. The coral reefs off the Eastern and Western coasts of Yucatan, Belize and the Florida Keys could not be mapped with the other coral reefs obtained in the coral reef clusters.

b) A proxy for water clarity/turbidity would help in improving the mapping of coastal waters suitable for coral reef development. Probably chlorophyl data (from CZCS or Seawifs) would help to discriminate more productive waters (which are less clear than coral reef waters).