The global degradation of coral reefs (Hughes et al., 2003), compromises the large number of ecosystem services that provide (Woodhead et al., 2019). To protect them, Seascape Ecological Units (SEU) cartography is a tool to guide restoration and conservation processes (Mumby, 2001). In Colombia, coral reef mapping began in the 1990s, including the Roncador and Serrana atolls (Díaz et al., 1996). Later, Díaz et al. (2000) presented the Colombia´s coral areas maps, which included almost all the RB Seaflower areas. Taking advantage of this product, Sánchez et al. (2005) presented modified cartographic squemes of Serrana, Roncador y Quitasueño; the first maps of Serranilla, Bajo Nuevo and Bajo Alicia were also produced (Invemar - ANH, 2012). Based on the availability of thematic information for the previously mentioned atolls, during the 2014 Seaflower Expedition, the registration of new biotic information in field was achieved, creating a data set that facilitated cartographic updates.
32 dives were made with Scuba equipment in the lagoon basins, backreefs, and fore reef terraces obtaining information derived from rapid ecological evaluation similar to that carried out by Díaz et al. (1995) and Garzón-Ferreira and Pinzón. (1999), discriminating scleractinian corals at species level in a 20 × 2 m band transect. Based on the types, highest percentages of biotic, substrate coverage, each SEU was determined, and the associated geomorphology. To fill information gaps due to the low number of points the Colombian coral areas atlas were used as reference (Díaz et al., 2000), and modifications from the aforementioned authors with cartographic schemes prepared by Sánchez et al. (2005), thus, the spatial position of each cartographic element through the ArcGIS 10.5 software was identified. General aspects of the Colombian Caribbean oceanic reef zoning were also consulted (Diaz et al., 1996; Díaz - Pulido et al., 2004), allowing a better interpretation of the information.
Before the cartographic edition, the digital image processing included atmospheric corrections through the “dark subtraction” tool on three satellite images of 2012, ALOS AVNIR - 2 sensors (one for each atoll), in where were observed contours up to 30 m deep. Geometric corrections were ruled out due to the absence of reference sites. However, the images show displacements of less than 10 m regarding the information taken in the field. Then each image was divided into three zones (lagoon basin, backreef, and fore reef terraces), as recommends from Andréfouёt and Guzmán (2005). In each zone were made band combinations to create true color compositions and principal component analysis using the blue and green bands, obtaining three components that were visualized with different order and selecting the combination that show the differentiation of predominantly coral cover. This analysis detects and transforms the image variability set of bands or components which explain the maximum of variability but maintaining the significant information (Chuvieco, 1996).
As a reference, the existing cartography mentioned previously was used (Díaz et al., 2000 y Sánchez et al., 2005), confronting it´s correspondence with what was observed in the satellite images, making new assignments when the field information demonstrated it. Besides, the thematic attributes associated using visual interpretation criteria such as texture, tone, and color in true color composition and principal components already described, which allowed assigning levels of uncertainty per polygon on the presence of each unit.
As a result, two cartographic layers were updated, one geomorphological and other biotic, and then rename according to the classification of Díaz et al. (2000). The final step consisted of the cartographic union of these layers, and the concatenation of the respective legends (biotic and geomorphological) generating a layer of SEU. For the assignment of the geomorphological units, the most outstanding features of the landscape were named taking into account the geomorphological classification scheme contained in Díaz et al. (2000) discriminating three zones: 1. Fore reef terraces, 2. Crest and 3. Lagoon basin; two types: 1. Patch reef (located in the lagoon basin) and 2. peripheral patch reef (located at the leeward side of the periphery of lagoon basin). Allowing to differentiate the biotic units by their location with respect to the reef geomorphology. A total 138,967 ha was delimited and 30 SEU based on 7 geomorphological units and 12 biotic units.
The Serrana atoll, despite corresponding to less than a third of the Quitasueño size with 32,908 ha, its reef crest that is part of the peripheral reef with a length of 56 km, and its lagoon basin with 24,883 ha, are the largest in Colombia, However, it represents only 12 SEU, the smallest number of the three atolls (Figure 1). Roncador, despite being the smallest of the three atolls with 4,861 ha, 13 km in length with a northwest - southeast orientation, 6 km in its widest portion and a 12.8 km long reef crest being part of the peripheral reef, represented 15 SEU. This atoll is characterized by having the lagoon basin with the highest relative coverage of coral formations in Colombia (Figure 2). Quitasueño, the largest coral area in Colombia with 101,198 ha, it extends for 63 km in its longest portion with north - south orientation and 24 km in its widest portion. It´s characterized by having a large for reef terrace, with dominance of coral formations towards the center, and large extensions of octocorals and sponges towards the south and north. In Quitasueño the largest number of SEU was identified for a total of 21, furthermore, its reef crest is the second largest in the country with a 43 km length (Figure 3).
Regarding the previous cartography (Díaz et al., 2000), the main differences can be seen in the lagoon basins detail level, which is higher in the aforementioned authors, but the scale is unknown; geometric distortions and displacements were also corrected and new elements of the background were delimited, mainly in Quitasueño.
The most notable change is the restriction of the SEU Arena - Coral debris to the lagoon terraces, since the soft bottoms of the other zones seems to correspond to descriptions of bioturbed sediments - calcareous algae of Diaz et al. (2000), likewise, their representation in the deep zones was improved with respect to the mentioned authors. The most notable difference is appreciated in Quitasueño with the SEU Octocorals - Sponges, at depths between 20 and 30 m in the fore reef terrace, it is similar to the so-called Macroalgae meadows - Octocorals - Sponges in Serranilla bank and bajo Alicia registered in 2012 (Invemar and ANH, 2012), but in this case macroalgae are not mentioned because despite being common in coral reefs (McCook, 1996; Knowlton, 2004), and become increasingly dominant in recent decades including remote coral reefs (Bruno y Valdivia, 2916), its dominance can occur seasonally (Díaz-Pulido and Garzón-Ferreira, 2002). Another notable SEU for not being previously represented in Quitasueño is Octocorals - Mixed corals located in the pre-reef terraces, which coincides with the descriptions of Sánchez et al. (2005). Regarding the SEU that build coral reefs or with hermatypic coral dominance, they cover about 29,000 ha representing about 21 % of the three atolls, where the most relevant are Octocorals - Mixed corals and Orbicella spp. The change in these proportions with respect to Díaz et al. (2000) is almost the same in Roncador, in Serrana the difference is barely 3 % in favor of soft bottoms, but Quitasueño is notably different, since the SEUs that form coral reefs are considerably higher in Díaz et al. (2000) where the mixed corals approach 40,000 ha, but in the present publication they only add up to 1,637; these differences may be due to the low capacity to detect the seabeds at that time, especially those corresponding to the deep areas of the fore reef terraces. The relationship between SEU forming coral reefs was inversely proportional to the atoll size. Quitasueño presents 18 %, Serrana 27 % and Roncador, which is about 21 times smaller than Quitasueño, presents 59 %; it is necessary to know if this behavior is a generality of the complex oceanic reefs of the Caribbean.
It is pertinent to mention that the SEU represented by Díaz et al. (2000) and Sánchez et al. (2005) as Acropora cervicornis was not identified in the field, which, if it persists, is presumed in scattered patches smaller than the scale of the present case. Also, the unit Acropora palmata - Pseudodiploria spp., it is represented in smaller proportions than those reported by the aforementioned authors. However, in several places eroded skeletons of these two species can be seen, vestiges of a past where they were more common, which is consistent with the decrease in the coverage of these species due to the mortality episodes reported throughout the Caribbean in the late 1980s (Aronson and Pretch, 2001).
In conclusion, a good approximation of the extension and distribution of the benthic seascape elements was achieved of an available product for consultation in MEC100K (Ideam et al., 2015) and AACC (Invemar - Minambiente, 2020). However, the need to carry out extensive field work and specifically oriented to the generation of thematic cartography to update the information at a more detailed scale and validate it persists, since the scientific need for updated, spatialized and thematic information that leads to identifying and delimiting SEU to the maximum level of detail is still in force.