Control of Seascapes
In this paper, we consistently use the word ‘control’ to refer to the collection of theories stating that the possibility of surveillance might have been important to past communities. We emphasise that results should be interpreted in light of this diversity of theories rather than exclusively the desire for exercising active control: the need to observe resources (like food or lithic extraction places) to avoid their use by certain people/communities; the ability to observe canoes navigating along the coast and individuals walking through the landscape to be aware of human movements, exchange routes or the possibility of attack. The outcomes of the experiments indicate the locations of land areas particularly suitable to visually control possible sea routes for short- and long-distance exchange. These include the southern coast and parts of the eastern tip of Pointe des Châteaux (TV3; Fig. 3c). These parts may have been particularly important in observing or controlling excursions between La Désirade and Basse-Terre. Only few sites are located in these areas. It is possible that visual control over both sides of Pointe des Châteaux could be exerted from other Grande Terre villages, located outside the study area (cf. De Waal 2006: Figs. 6.1, 6.3 and 6.8). In addition, as the Pointe des Châteaux peninsula only has a width of approximately 1 km, it is expected that communication between people in sites on the south coast with people in settlements in other parts of Pointe des Châteaux was quick and easy. Vantage points around major settlements like Anse à la Gourde could have enabled successful visual control over sea lanes to the north, east and south. The south coast is particularly well suited for visual control over the southern sea leading towards Petite Terre and enabling navigation further east to La Désirade (TV3; Fig. 3c). This is significant given the importance of Petite Terre as a marine resource extraction area (De Waal 2006: 65; De Waal 2009: 8–9) and of the hypothesised contacts between communities on Pointe des Châteaux and La Désirade. As sea currents to the north and north-east of Pointe des Châteaux are particularly strong (De Waal personal observations 1994–2002), we expect navigation from Guadeloupe to La Désirade to have followed the relatively sheltered southern coast before making the crossing to La Désirade. We therefore argue that sites on the southern coast should be considered lookout points as a complementary part of the major agglomeration of Anse à la Gourde on the north coast.
Interestingly, the temporary habitation sites of site 1 and Ouest Résidence Kahouanne, and the strategic outpost at Ouest Anse à Plume, are located in these areas controlling sea routes but permanent habitation sites are found in other locations (Table 3). This suggests that the location of Anse à la Gourde, one of the most prominent settlements in the study area thought to have played an important role in short- and long-distance exchange networks, was not selected for its visual control of sea routes. The Anse à la Gourde area has many other favourable conditions for a settlement (De Waal 2006: Table 5.1), but apparently, visual control of sea routes was not a requirement for a settlement to fulfil this role in exchange networks. Several uninhabited areas on the south coast were more suitable to visually control seascapes. It may also be suggested that other sites, such as site 1 and Ouest Résidence Kahouanne, fulfilled this role for Anse à la Gourde, even though they are at some distance (0.8 and 1.5 km) from this settlement. If this happened, this only occurred during the early phase of the Late Ceramic Age (AD 600/850–1200/1300). During earlier and later phases, Anse à la Gourde is surrounded by less sites, the landscape is more empty and the inhabitants of the settlement had fewer opportunities to control sea routes in their near surroundings.
Land areas particularly suitable to visually control possible sea routes for short- and long-distance exchange at La Désirade include the western tip and the entire southern coastal area of the island (TV3; Fig. 3c). It is noteworthy that the southern edge of the central plateau is not included. In the experiments, the plateau edge has normal visibility, but it has a lower theoretical maximum of visible sea area, being located slightly inland and given our theoretical maximum viewing distance of 3 km. When standing on the edge of the plateau at Morne Cybèle-1 and Morne Souffleur one can oversee the sea to Petite Terre (and identify the Petite Terre lighthouse) and to Pointe des Châteaux. From the plateau, it is also possible to see boats arriving at the island as landing areas for canoes can be observed. Morne Cybèle-2 overlooks Anse Petite Rivière and Morne Souffleur overlooks one of the other few bays where one can easily land a canoe at the south coast.
At La Désirade only six sites are located in areas with high visual control over sea routes (Table 4). Only three of these represent permanent habitation sites: Pointe Colibri, À L’Escalier and Aéroport. Grand Abaque 1, linked in a micro-style area with Pointe Colibri is in an area of low visibility. The natural setting of the sites of Pointe Colibri and Grand Abaque 1 is very different (De Waal 2006: Table 5.1). The regular contacts, suggested on the basis of similarities in ceramics, were apparently not linked to shared preferences for locations overlooking sea routes. They rather seem to complement each other, one site overlooking the west, the other site overlooking the east. Interestingly, habitation sites facing Petite Terre are located in areas of high visual control over the southern sea (Pointe Colibri, Aéroport, À l’Escalier TV3, Fig. 3c). Similarly to Anse à la Gourde, Anse Petite Rivière is located very close to (but not in) such areas (VNC3, Fig. 4c) (Table 4). Uninhabited areas close-by have slightly better visual control over seascapes. But no sites are close-by Anse Petite Rivière that may have taken over this function.
At Petite Terre, possible sea routes for short- and long-distance exchange can be controlled from large coastal areas except for the low-lying northern coast (TV3; Fig. 3c). No sites are located in areas with good visual control over seascapes (Table 5) and uninhabited areas allow visual control over seascapes. However, from Est de Mouton de Bas at the northeastern coast of Terre de Bas, and Pointe Sablé at the southern coast of Terre de Haut, one can control access to the restricted channel separating the islands of Petite Terre, which is reported to be one of the richest marine resource extraction areas for the islands (De Waal 2006: 65; De Waal 2009: 8–9). These sites may have played a role in mobility related to resource extraction at Petite Terre, for example by expeditions from La Désirade or Pointe des Château.
The results further indicate that in the eastern part of Pointe des Châteaux, including the area with the Salinas, smoke columns rising up from villages would be particularly visible from sea (TV1; Fig. 3a). Half of the sites are located in these areas, including half of the habitation sites, including Petites Salines, Est Petite Saline Orientale, Degrat and site 7 (Table 3). The sites are evenly distributed over these areas, which eliminates the question if uninhabited areas were more visible from the sea. Sites belonging to micro-style areas do not share the same characteristics related to the visibility of smoke columns. It is noteworthy that Anse à la Gourde is not in such a location. Regarding its role in regional and micro-regional networks one would expect that the visibility of smoke columns could help to direct interacting groups to the site. However, Anse à la Gourde does have an eye-catching feature that helps people navigating to the site: a small rocky island called ‘Le Diamant’, just north of Anse à la Gourde bay (CV8; Fig. 6b). In addition, even though visibility is low for the settlements of Anse à la Gourde and Grande Saline, high visibilities have been identified for the close-by indistinct site of Pointe a Cabrits 2, west of Anse à la Gourde, and of the strategic outpost at Ouest Anse à Plume, west of Grande Saline. These sites may have complemented this visibility role of both settlements. Smoke columns are best seen from sea stretches at the north-west and south-west of Pointe des Châteaux.
In the western and eastern parts of La Désirade smoke columns rising up from villages would be particularly visible from sea (TV1; Fig. 3a), especially from sea stretches directly south of the island. Only 8 out of 36 sites are located in these areas. Five of these are permanent settlements, including Pointe Colibri and Grand Abaque 1, which together form one of the early micro-style areas (Table 4). The latter site, which seems to be hidden in the eastern hills, must have been well visible when it comes to detecting smoke columns from the sea. However, it may be questioned how important the notion is that both early micro-style area sites have a similar visibility of smoke columns from the sea. Contacts between both sites may well have taken place over land, as they are only separated by approximately 10.5 km as the crow flies and as canoes cannot be landed at the northern and eastern sides of the island close to Grand Abaque 1. Interesting to note are the unremarkable results for the settlements of Morne Cybèle-1 and Morne Souffleur, neither particularly high or low. People establishing sites with presumably defensive locations probably tend to avoid attention to be drawn from a distance, but the results of TV1 (Table 4) do not provide strong support for this. Anse Petite Rivière also has a low visibility of smoke columns from the sea. Interesting to note as well is the high visibility of the south-eastern part of the island, where the (undated) lithic workshops are located. These workshops have played a role in micro-regional interactions as inhabitants from Pointe des Châteaux and from other parts of Guadeloupe collected raw materials for the manufacture of lithic artefacts at these locations (De Waal 2006: 106, 115, 130). The high visibility of the indistinct (special activity) site of Grande Ravine, at some elevation north of the lithic workshops, seems to complement the visibility of the latter.
In the south-eastern part of Terre de Bas in Petite Terre smoke columns rising up from villages would be particularly visible from sea (TV1; Fig. 3a). This seems important as this part of Terre de Bas is furthest removed from presumably important transportation routes between Marie-Galante (or further south) and the north, which may be expected to have passed the islands of Petite Terre in the west. No sites are located in these high visibility areas, whilst two habitation sites are located in areas of very low visibility (Table 5). Sea stretches surrounding the islands, except for the eastern part, would allow good views on smoke columns present on the islands of Petite Terre (TV2; Fig. 3b).
As a concluding remark, we cannot argue that the sites in general are located in places of exceptionally good or exceptionally bad visibility from and to the sea. Considering the presumed importance of the visibility of smoke columns with regards to intersite and interisland movements, more sites were expected to be located in high visibility areas than indicated in the experiments. The experiments do not indicate visibility to have been a factor of importance for the existence and functioning of micro-style areas either.
Signalling Network
The visibility network experiment N1 (Figs. 7 and 8) revealed that the networks’ structures could have enabled smoke signalling networks between sites that could have functioned for communication purposes throughout the research area. The networks per landmass are connected and dense and they incorporate all known sites, allowing for information to be shared through signalling from any site to any other site.
The Pointe des Châteaux visibility network seems to consist of one single component. The settlement of Anse à la Gourde is a visually prominent node in this network. However, it did not play an important role as an intermediary. The Petite Terre visibility network includes one single component where all sites are intervisible (N1; Figs. 7 and 8).
The network on La Désirade consists of two separated components (N1; Fig. 7). Pointe Colibri plays an important role as an intermediary in a hypothetical signalling network, whereas Grand Abaque 1, in the eastern part of the island, does not play an important role. This network structure raises an important question: how could the western and eastern components become connected so that a functional signalling network could have existed for the entire island? There are multiple possible answers (Fig. 9). First, one is tempted to assume the existence of an unidentified coastal site between Pointe à Godard and À L’Escalier, but the coast is too steep here to allow settlement (Fig. 9a). Second, the site Trou Madame, tentatively interpreted as a (yet indistinct) special activity site, is located close to the plateau edge and within 3-km distance of Point Godard in the western component (Fig. 9b). Locations close-by Trou Madame offering lines-of-sight to Pointe à Godard would serve to connect both components. Third, and to us most intriguing, locations along the plateau edge nearby the site of Chemin de M. de l’Orme situated on the central plateau could also connect both components. Chemin de M. de l’Orme has been interpreted as a ceremonial, thus exceptional, site in an ‘empty’ area on the plateau (Fig. 9c). The site itself has low visibility (N1; Fig. 7) but nearby locations offer lines-of-sight to Pointe à Godard in the western component. It consists of a deliberate deposition of a small pelican vessel, which functioned as a container for a small non-used stone axe and adze of St. Martin chert (De Waal 2003, 2006: 99, 304). The site was considered a deposition in an unattractive area with no strategic importance, on a hilly terrain that is difficult to reach and far from the villages and the coast. In the light of the visibility analyses, it can now be suggested to have been close to locations that served as a lookout, connecting two network components on the island and it is tempting to further speculate if the depot was created as reference to this important role as a lookout or anchor point within the islands’ visibility network.
Interestingly, when we only include habitation sites, smoke signalling networks still may have functioned (Fig. 8). On the northern part of Pointe des Châteaux and the southern part of La Désirade, more or less ‘opposing sides’, the visibility networks of habitation sites form chains or paths through which information may have been shared from one side of a landmass to the other. This spatial distribution is thought not to be a result of the similarity in shape of the landmasses, as the pattern did not emerge in other similarly shaped parts of the study area. Due to the maximum viewing distance, artificially set at 3 km for our experiments, interaction between communities by smoke signalling was limited to intra-island contacts.
It is difficult to ascertain whether such hypothesised signalling networks transformed over time, since only a few sites could be assigned in an absolute way to different micro-style areas and phases of the Late Ceramic Age, making networks of securely dated contemporary sites impossible. However, one major change is obvious: a well-functioning smoke signalling network could have existed in the early phase of the Late Ceramic Age as it has a large number of sites all over the landscape, which allows intensive visual signalling networks, but not in the later phase. The decrease to only three later phase sites marks the end of a smoke signalling network if one did exist. Morne Souffleur and Morne Cybèle-1 are intervisible, but a non-trivial signalling network requires a chain of at least three sites where messages can be passed on between mutually intervisible sites A and C via a mutually visible site B. In addition, Morne Souffleur and Morne Cybèle-1 share similar visibility characteristics and do not complement each other’s observations. Interestingly, this later phase is precisely the period in which East-Guadeloupe inhabitants might have wished for relatively dense or tight signalling and communication networks in their close environs, as there would have been only few neighbours to rely on in times of need. Apparently, they depended on different ways to keep contact and they covered larger distances out of necessity (cf. De Waal 2006: 132).
The fact that most of the sites have not been absolutely dated reduces our certainty of these visibility networks’ structures. However, taking one node away from the network (for example because it is not contemporaneous with the other sites) does not cause the network to break up into multiple components, thanks to its density, and thus stop being able to function as a possible signalling network. When only habitation sites are included in the network the removal of a site has a larger impact, simply because less sites can take over its role. It is also interesting to note that when we look at the early phase of the Late Ceramic Age and in turn remove the sites belonging to each of the two micro-style areas, almost nothing changes in terms of the roles of other sites and the ability of the network to function. A problematic exception to this is the removal of Pointe Colibri, which would disrupt the ability of habitation sites on La Désirade to function as a visual signalling network.
Multi-Scalar Networks
We argue the significance of these visibility networks for structuring navigation and communication between communities in the area needs to be understood from a multi-scalar perspective: the chains of intervisible habitation sites created local (short-distance) networks on the different landmasses, and are linked up through long-distance visibility between the landmasses, where similar chains of intervisible sites mirror each other (Fig. 10). We hypothesise that the short-distance visibility networks structured navigation and communication within landmasses, whereas the landmasses themselves served as focal points for regional navigation and interaction.
The patterns in the study area support this hypothesis, but it may also serve as a hypothesis for the wider region. Habitation sites on the Pointe des Châteaux peninsula are regularly spaced and this pattern is mirrored along the western tip and southern coast of La Désirade, enabling a network of sites connected through the intervisibility of smoke columns (N1; Fig. 8). Outside the study area this pattern of regularly spaced habitation sites is continued along the northern and southern coast of Grande Terre (cf. De Waal 2006: Fig. 6.8), suggesting this visibility network could have been an important feature of the Late Ceramic Age settlement pattern of Grande Terre. These local chains are connected through the long distance intervisibility of landmasses from islands. Of particular importance is the visibility of the western side of La Désirade from Anse à la Gourde (CV7; Fig. 6a), and the visibility of Petite Terre and the eastern tip of Pointe des Chateaux from Morne Souffleur (De Waal and Hofman, personal observations 1994–2000). According to this hypothesis, direct visibility between contemporary sites might have been less important than the cumulative effect of a chain of long and short distance visibility: as soon as a new landmass was reached one could link into the local network of intervisible locations. For example, although the contemporary sites of Anse à La Gourde and Anse Petite Rivière are not directly intervisible, from Anse à la Gourde the landmass of La Désirade is visible (CV7; Fig. 6a), whereas the rock of Le Diamant off the coast of Anse à la Gourde is visible from a far bigger sea area south of La Désirade and might have served as an additional visual marker for people approaching this settlement from La Désirade and Anse Petite Rivière specifically (CV8; Fig. 6b).
This interpretation suggests that large parts of the landscape, vastly exceeding the boundaries of the East-Guadeloupe micro-region, were interconnected, even though the individual land masses are separated by long stretches of sea. These sea stretches could therefore be considered to have functioned as passage areas, linking the different land masses into local networks. This hypothesis further underlines the importance of understanding the visual properties of Late Ceramic Age East-Guadeloupe from a multi-scalar landscape perspective.
Micro-Style Areas
The outcomes of the experiments indicate that sites belonging to the Pointe Colibri and Grand Abaque 1 micro-style area have similar visual property values. However, the sites belonging to the other early phase micro-style area, Anse Petite Rivière and Anse à la Gourde, have very different visual properties. With regards to visibility of people navigating the coast in canoes, Grand Abaque 1 and Pointe Colibri are located in areas where smoke columns are highly visible from the sea (TV1; Fig. 3a; VNC1; Fig. 4a), whereas Anse Petite Rivière and Anse à la Gourde, are located in areas with average visibility from the sea (TV1; Fig. 3; VNC1; Fig. 7). Visual control over sea areas does not seem to have been a factor of importance for the early phase micro-style areas. Pointe Colibri is one of only three habitation sites well-positioned to visually control the sea (TV3; Fig. 3c). When we include locations in areas of high visibility then we can also consider Degrat an early site with good visual control over sea (VNC3; Fig. 4c). Grand Abaque 1 is particularly badly positioned for this purpose. Anse Petite Rivière and Anse à la Gourde are located in areas with average visual control over the sea (TV3; Fig. 3c), although regarding locations embedded in local areas of high visibility then we can also consider Anse Petite Rivière a later site with good visual control over sea (VNC3; Fig. 4c). The sites belonging to the early phase micro-style areas also do not share visibility of certain natural features, such as parts of seascapes and landscapes. The views they offer seem rather to be complementary (CV1–4; Fig. 5a–d). Moreover, smoke columns rising up from villages belonging to the same micro-style area are not intervisible. The fact that strong stylistic similarities do not occur between sites that are particularly close to each other, or that are intervisible, indicates that purposeful efforts were made to maintain close contacts with groups at larger distances. It may be hypothesised that the larger the distance, the larger the need for constructing strong ties between communities.
Interesting to note is that sites related to micro-style areas do not occupy the important positions in networks one would imagine, except for Pointe Colibri and to some extent Anse à la Gourde in the network with only habitation sites (N1; Figs. 7 and 8). We may conclude that visibility was not a prime factor in determining the location and role of key villages like Anse à la Gourde and Anse Petite Rivière. They are by far the largest sites in the micro-region, both belong to a micro-style area and display evidence of materials obtained through short and long distance contacts, yet their visual properties are not exceptional.
Later Phase Transformation and Defensive Locations
The results show that the two sites on the La Désirade plateau edge dated to the late phase of the Late Ceramic Age, Morne Souffleur and Morne Cybèle-1, are not particularly hidden from view whilst being close to areas from which access routes to these sites, via the coast and the sea, can be visually controlled. Both sites are in areas with high local variability in visual properties as revealed by the VNC experiments. They offer particularly limited visual control over the sea but are very close to areas offering great vantage points (TV3; Fig. 3c; VNC4; Fig. 4d). In addition, smoke columns at these sites would be visible from much larger sea areas than those in their immediate surroundings but they are also close to areas where they would be far less visible (TV1; Fig. 3a; VNC2; Fig. 4b). This seems contrary to what one would expect for sites in supposedly defendable locations. In hostile situations, village inhabitants are expected to wish for clear look-out facilities at a site, and to be well-hidden at the same time.
To some extent, other uninhabited areas were better suited to serve a defensive purpose. Many areas in the hilly western and eastern tips of La Désirade and on the plateau edge share the same visual properties of Morne Souffleur and Morne Cybèle-1. Parts of these areas were also occupied during the earlier phase of the Late Ceramic Age, so we cannot argue that the visual properties of Morne Souffleur and Morne Cybèle-1 are exclusively a phenomenon belonging to the later phase. However, since these are the only two sites on La Désirade in this later phase and the experiments reveal that their site locations do serve this purpose, we can assume that these visual features were considered particularly important in this phase and that these locations were purposefully selected partly for their visual properties. The eastern areas of La Désirade might have been avoided because they would not offer great views over the southern coast facing Petite Terre. It is not clear why the western areas were avoided. One reason might be a wish to be not in sight of groups travelling north (for example coming from Marie-Galante or further south) through the passage between Pointe des Châteaux and La Désirade.
A number of earlier sites on La Désirade (e.g. Anse Petite Rivière and Pointe Colibri) are better located to visually control seascapes when compared to the later sites of Morne Souffleur and Morne Cybèle-1, from which nevertheless a large area of the sea along the south coast can be visually controlled (CV5–6; Fig. 5e, f; CV10–11; Fig. 6d, e). We can therefore conclude that the ability to visually control the sea from these later sites is not exceptional as compared to earlier sites. However, unlike Anse Petite Rivière and Pointe Colibri, visual properties at Morne Souffleur and Morne Cybèle-1 can change radically within the local area, and the inhabitants of the sites could have profited from precisely this variability. This interpretation again emphasises the importance to understanding visual properties of sites from a landscape perspective and not just from the perspective of the individual sites, as is usually done.
To conclude, it remains questionable whether the locations for Morne Cybèle-1 and Morne Souffleur were selected primarily for their defensive aspects. Villages should preferably be invisible from the sea as smoke columns can easily give away hidden village locations to enemies, whereas good look-outs over the surrounding landscapes and seascapes should be available close-by the village. This is precisely the opposite of what our total viewshed results show (TV1, 3; Fig. 3a, c; Table 4). On the other hand, the visual neighbourhood configuration experiments did reveal that such locations with a defensive nature exist close-by due to the high variability of visual properties in the local area (VNC2, 4; Fig. 4b, d; Table 7), and the cumulative viewshed experiments and the authors’ personal observations during survey work highlight the visibility of the landmasses of Petite Terre and Pointe des Châteaux (CV10–11; Fig. 6d, e). These results reflect a situation that may well have served a defensive purpose, although maybe not exclusively so. If the site locations were not selected for defensive purposes alone, and if they were meant to be seen from a distance, this might also relate to the suggested special, possibly ceremonial, function of the sites, based on the presence of the shell faces or guaizas at both sites (De Waal 2003, 2006; Hofman 1995; Hofman et al. 2004). In addition, we could argue that sites participating in regional networks, which both sites were, are not expected to be hidden.