Introduction
The sperm whale Physeter macrocephalus is a deep diving, top marine predator distributed worldwide, commonly in offshore and deep waters (usually > 1,000 m), concentrated in areas known as ‘grounds’ (Whitehead, Reference Whitehead2003). The distribution of sperm whales is associated with areas of upwelling, temperature gradients, seafloor relief and with processes supporting food webs that include mesopelagic or demersal cephalopods, on which sperm whales feed (Baumgartner et al., Reference Baumgartner, Mullin, May and Leming2001; Evans & Hindell, Reference Evans and Hindell2004). However, females with their young are usually restricted to temperate and tropical waters at low latitudes, where sea surface temperatures are > 15 °C. Males leave their mothers at c. 10 years of age, moving to colder waters at higher latitudes, returning in their late 20s to the tropical and subtropical habitat of females, to mate (Whitehead, Reference Whitehead2003; Whitehead et al., Reference Whitehead, Antunes, Gero, Wong, Engelhaupt and Rendell2012).
Sperm whales are categorized as Vulnerable on the IUCN Red List (Taylor et al., Reference Taylor, Baird, Barlow, Dawson, Ford and Mead2019) and nationally in Colombia (Rodríguez-Mahecha et al., Reference Rodríguez-Mahecha, Alberico, Trujillo and Jorgenson2006), having been hunted for 2 centuries, until the 1990s, across all oceans (Whitehead, Reference Whitehead2002). Although commercial hunting has ceased, sperm whales face threats from incidental catch, interactions with fishing gear, collisions with boats, and pollution (Avila et al., Reference Avila, Kaschner and Dormann2018). The global population trend of the species is unknown (Taylor et al., Reference Taylor, Baird, Barlow, Dawson, Ford and Mead2019), but the sperm whale population in the eastern Caribbean declined during 2005–2015 (Gero & Whitehead, Reference Gero and Whitehead2016). The global offshore distribution of sperm whales and their habitat use are poorly known (Whitehead, Reference Whitehead2003), and there have been few studies of the species in Colombia despite its occurrence in both Pacific and Caribbean waters (Trujillo et al., Reference Trujillo, Gärtner, Caicedo and Diazgranados2013).
The Colombian Caribbean region is important for cruise tourism (Aguilera et al., Reference Aguilera, Bernal and Quintero2006) and fishing activities (Suárez & Rehder, Reference Suárez and Rehder2009), but there is little information on the ecology of this region to support management decisions. Studies of the occurrence of marine mammals in the Colombian Caribbean Sea have been limited to coastal areas, with few studies offshore (e.g. Pardo et al., Reference Pardo, Mejía-Fajardo, Beltrán-Pedreros, Trujillo, Kerr and Palacios2009; Farías-Curtidor et al., Reference Farías-Curtidor, Barragán-Barrera, Chávez-Carreño, Jiménez-Pinedo, Palacios and Caicedo2017). Information about sperm whales in Colombian waters is scarce, and the few studies on sperm whales in this region have been in The Bahamas (Ward et al., Reference Ward, Thomas, Jarvis, DiMarzio, Moretti and Marques2012) and close to Dominica and surrounding islands (e.g. Gordon et al., Reference Gordon, Moscrop, Caroson, Ingram, Leaper, Matthews and Young1998; Gero et al., Reference Gero, Gordon, Carlson, Evans and Whitehead2007, 2014; Gero & Whitehead, Reference Gero and Whitehead2016), where small aggregations of up to 14 adult females and subadults of unknown sex have been reported (Ward et al., Reference Ward, Thomas, Jarvis, DiMarzio, Moretti and Marques2012; Gero et al., Reference Gero, Milligan, Rinaldi, Francis, Gordon and Carlson2014). It has been suggested that the eastern Caribbean is an ecological trap for the species (Whitehead & Gero, Reference Whitehead and Gero2015). Given the importance of the eastern Caribbean region for sperm whales and the documented population decline there, which potentially suggests migration to surrounding areas with better conditions (Whitehead & Gero, Reference Whitehead and Gero2015; Gero & Whitehead, Reference Gero and Whitehead2016), such as The Bahamas, it is important to determine the environmental conditions that affect sperm whale distribution in other areas in the Caribbean, such as Colombian waters.
Here we report our research on the behaviour, encounter rate, group structure and density of sperm whales in the Colombian Caribbean through the compilation and analysis of data from offshore surveys. To describe the potential distribution of the species in this region, we investigate how environmental conditions at various depths influences its distribution. We identify areas of the Colombian Caribbean where sperm whales are present, and demonstrate the importance of offshore monitoring of marine mammals to generate data for management plans in this region.
Study area
The study area is the Colombian Caribbean, which comprises a total area of 132,288 km2 (Fig. 1). This region is characterized by a mean depth of 2,700 m, with a maximum of 4,500 m, and a wide continental shelf (70–150 km) that extends to 200 m depth (Tabares et al., Reference Tabares, Soltau and Díaz1996). Temperature at the surface is 27–29 °C, and varies from 18 °C at 200 m to 6 °C at 800 m; salinity is 35–37 ppt (Andrade et al., Reference Andrade, Rangel and Herrera2015).
Fig. 1 The Colombian Caribbean Sea, showing the bathymetry of the study area and the locations of the Archipelago of San Andrés, Old Providence and Saint Catherine (the latter two islands labelled Providencia), the Gulf of Urabá, the Gulf of Darién, and the rivers Atrato, Sinú, Magdalena and Ranchería.
Methods
Surveys
During February–November in 2011 and 2013–2016 we recorded occurrences of sperm whales and collected environmental data in the Colombian Exclusive Economic Zone in the Caribbean Sea from aboard seismic vessels in an area being explored for oil and gas (Supplementary Fig. 1). Visual surveys were during 06.00–18.30 by two biologists trained in observation of marine fauna. Observations, using 10 × 50 binoculars, were made from the highest platform of five survey vessels (Veritas Viking, Osprey Explorer, Polar Duke, Oceanic Sirius and Oceanic Vega), with a mean observation height of 18 m and at a mean speed of 4.2 knots. When possible, observed sperm whales were photographed. Sperm whales were located via their lateral blowing out and by their dorsal fins, tails or body (Farías-Curtidor et al., Reference Farías-Curtidor, López, Alarcón, Duque, Barragán-Barrera, Avila, Puentes and León2020). For each sighting, date, time, location (with a GPS), number of individuals, presence of any juveniles, follow-up time, depth of bottom, sea conditions (Beaufort), cloudiness and visibility were recorded. Behaviour was recorded using ad libitum sampling. Vessels halted any seismic activity when a marine mammal was close to the vessel (< 500 m), to avoid or mitigate any potential negative impact of this activity on them (JNCC, 2017), and therefore we did not evaluate the behaviour of whales in relation to seismic activity.
For seismic surveys during 2013–2016, acoustic data were collected by a passive acoustic monitoring operator during 18.31–05.59). Sperm whales were detected acoustically from their wideband clicks, which can be distinguished from other marine sounds (Mellinger et al., Reference Mellinger, Thode, Martinez, McKay and Nides2003).
Sperm whale occurrence and density data analyses
Encounter rate and group size statistics were estimated for sighting surveys, the former as the number of individuals and groups sighted per 100 h of observation effort and the latter as the number of individuals and groups observed per 1,000 km2.
Modelling sperm whale distribution
We used the maximum entropy algorithm, in Maxent (Phillips et al., Reference Phillips, Anderson, Dudík, Schapire and Blair2017), to model the potential distribution of sperm whales in the Colombian Caribbean Exclusive Economic Zone using our observations combined with published data. Maxent estimates the geographical range of a species by finding the distribution that has the maximum entropy constrained by the environmental conditions recorded at occurrence locations. Models were performed using the maxnet function in Maxent with a complementary log–log transformation, which appears to be most appropriate for estimating probability of presence (Phillips & Dudík, Reference Phillips and Dudík2008; Phillips et al., Reference Phillips, Anderson, Dudík, Schapire and Blair2017).
As our data were not collected along survey track lines, real absences were not available, and therefore to represent pseudo-absences we randomly selected locations lacking presence data (Phillips & Dudík, Reference Phillips and Dudík2008; Merow et al., Reference Merow, Smith and Silander2013). We created one random point per 4.5 × 4.5 km grid cell for the surface model and one random point per 9.2 × 9.2 km grid cell for the models at various depth levels, using the randomPoints function in the Dismo package (Hijmans, Reference Hijmans2017) in R 3.6.3 (R Core Team, 2020) over the study area defined for each model (Supplementary Material 1).
Environmental variables used in the modelling were selected based on information in Baumgartner et al. (Reference Baumgartner, Mullin, May and Leming2001), Tobeña et al. (Reference Tobeña, Prieto, Machete and Silva2016) and Barragán-Barrera et al. (Reference Barragán-Barrera, do Amaral, Chávez-Carreño, Farías-Curtidor, Lancheros-Neva and Botero-Acosta2019). As the sperm whale is a deep diving species, we included environmental variables from different levels of the water column: on the surface, at c. 0.5 m depth (Level 1), c. 500 m (Level 2), c. 1,000 m (Level 3), c. 1,500 m (Level 4) and c. 2,000 m (Level 5) (Supplementary Fig. 2). Source, types of environmental data available, spatial resolution and time span of sea surface data differ from those of data for various depths, and therefore we chose data that we considered to be equivalent. One exception was the inclusion of the ocean mixed layer thickness data in the models at different depth levels, which was not included as a surface layer. We analysed variable importance and selected uncorrelated environmental variables following the method proposed by Dormann et al. (Reference Dormann, Elith, Bacher, Buchmann, Carl, Carre and Garcia-Marquez2013), implemented by Zurell et al. (Reference Zurell, Zimmermann, Gross, Baltensweiler, Sattler and Wüest2020). Firstly, we examined the importance of variables for the surface and different depths using a simple generalized linear model for each potential predictor, and ranked variable importance with Akaike's information criterion. We then inspected correlations between environmental layers to identify all pairs of variables that had a Spearman correlation coefficient > 0.7, removing the less important variable from further analyses (Supplementary Material 1, Supplementary Tables 1–3). From a total of 111 environmental layer candidates, we selected 22, including dynamic (ocean mixed layer thickness, salinity, temperature, total chlorophyll a and phytoplankton) and static (bathymetry, distance to shore, seafloor aspect and slope) variables. We extracted predictor values for each occurrence, and background points, using the function Fun_Extract (Derville et al., Reference Derville, Torres, Iovan and Garrigue2018), which returns the closest values for empty cells. To summarize the environmental conditions at the surface and depth levels, and to describe the environmental heterogeneity of the water column, we conducted a principal component analysis (PCA) for the 22 selected layers, using the function rasterPCA in the Rstoolbox package (Leutner & Horning, Reference Leutner and Horning2016) in R (Supplementary Table 4).
Maxent model settings were defined through the ENMevaluate function of the ENMeval 0.3.0 package (Muscarella et al., Reference Muscarella, Galante, Soley-Guardia, Boria, Kass, Uriarte and Anderson2014) in R, which provides species-specific settings such as feature classes and regularization multipliers to generate models (see Supplementary Material 2 for details). Model performance and cross-validation predictions were estimated using a series of adapted functions (Zurell et al., Reference Zurell, Zimmermann, Gross, Baltensweiler, Sattler and Wüest2020). The functions partition the data into k folds (k = 15), determine the model algorithm, update the model for the new training data, and make predictions for the hold-out data. The values of the area under the curve (AUC) and the true skill statistic were used as indicators of the predictive ability of the models. The best model is that with an AUC value closest to 1 (Phillips et al., Reference Phillips, Anderson and Schapire2006). For the true skill statistic, a value > 0.5 indicates a good prediction (Tobeña et al., Reference Tobeña, Prieto, Machete and Silva2016). Two types of model output are commonly used to describe the potential distribution of a species: continuous results in which sites are assigned a probability of being part of a species’ distribution, and binary results in which sites are classified as either part of the distribution of the species or outside their distribution (Liu et al., Reference Liu, White, Newell, Anderssen, Braddock and Newham2009). For the former, the final output maps derived from the cross-validation predictions for each type of model were exported in raster format with values in the range 0–1, and were interpreted as an estimate of occurrence probability. For the latter, binary maps were constructed to indicate where sperm whales could be present. To do this, the maps of occurrence probability were transformed by calculating an optimal threshold (a value that maximizes the sum of model sensitivity plus specificity) using the PresenceAbsence package (Freeman & Moisen, Reference Freeman and Moisen2008) in R, implemented in the script of Zurell et al. (Reference Zurell, Zimmermann, Gross, Baltensweiler, Sattler and Wüest2020). As we used several record types of sperm whale occurrence (published and opportunistic acoustic and sighting data), data were first cleaned, removing duplicate records and retaining only one occurrence record per grid cell.
Results
Monitoring
During 2011 and 2013–2016 we surveyed for a total of 13,069.8 h in 703 days over an area of 68,904.7 km2 (Table 1). We covered 52.1% of the offshore areas of the Colombian Caribbean Sea (Supplementary Fig. 1). We recorded a total of 98 individual sperm whales in 50 groups, of which 77 individuals (73 adults and four juveniles) in 35 groups were during daytime in 7,950.3 h of surveys over 50,024 km2, and 21 individuals in 15 groups were during night-time in 5,119.5 h of acoustic surveys over 18,880.7 km2. Group sizes were 1–10 (mean 2.0 ± SD 1.5) and each group was recorded for a mean of 18.9 ± SD 22.1 minutes. All observations of sperm whales were during seismic surveys and were outside the mitigation zone (a radius of 500 m around the sound source; JNCC, 2017); 62 and 38% of these observations were when the seismic airgun was active and inactive, respectively. Sperm whales exhibited slow and fast swimming, exposure of pectoral and caudal fins, resting, spyhopping (putting head out of water and looking around) and breaching behaviours (Plate 1).
Table 1 Number of groups and individuals of sperm whales Physeter macrocephalus recorded during the daytime and night-time during 2011 and 2013–2016 in the Colombian Caribbean (Fig. 1), with hours of survey effort and area surveyed. Number of individuals by age was only recorded during the daytime.
Plate 1 Some of the behaviours of sperm whales Physeter macrocephalus that we recorded in the Colombian Caribbean: (a) two adults swimming slowly (photo: Nohelia Farías-Curtidor); (b) an adult resting (photo: Javier Alarcón); (c–d) a juvenile breaching (photo: Nohelia Farías-Curtidor).
Mean encounter rate was 0.8 individuals and 0.4 groups per 100 h, with 0.7 individuals and 0.5 groups per 100 hours during daytime, and 0.8 individuals and 0.3 groups per 100 h during night-time. The estimated density of sperm whales was 1.42 individuals and 0.7 groups per 1,000 km2, with 1.6 individuals and 0.7 groups per 1,000 km2 during daytime and 1.1 individuals and 0.8 groups per 1,000 km2 during night-time.
Distribution models
Data for a total of 66 groups of sperm whales and at least 124 individuals were recorded in our surveys and opportunistic sightings, and published data, combined, during 1988–2020, of which eight records were juveniles (0.8 per year). Sperm whales were recorded at a mean distance of 72.5 km from the coast (6–237 km), in a mean water depth of 1,700 m (244–4,191 m; Table 2, Fig. 2).
Table 2 Records of sperm whales Physeter macrocephalus in the Colombian Caribbean during 1988–2020, with corresponding depth range and distances to the coast.
1 S, south (≤ 12°N); N, north (> 12°N); W, west (< 75°W); E, east (≥ 75°W); C, central.
2 San Andrés indicates Archipelago of San Andrés, Old Providence and Saint Catherine.
* Minimum number of individuals (total number not confirmed).
Fig. 2 Records of sperm whales Physeter macrocephalus in the Colombian Caribbean during 1988–2020.
The PCA indicated environmental heterogeneity across the water column, with the surface the most differentiated compared to the five water depth levels (Supplementary Fig. 3). The generalized linear models, ranked by Akaike's information criterion, indicated the most important variables that explain sperm whale distribution were distance to shore, and range and standard deviation of ocean mixed layer thickness (Supplementary Table 4).
After data cleaning, 61 occurrences of sperm whale groups were available for Maxent modelling, but with differing numbers of records at different depth levels (Table 2). All models generated had an AUC > 0.77 and true skill statistic > 0.47, indicating a good performance in general. However, when we considered the optimal threshold, models of the surface and Level 1 failed to assign presence to 59.3% and 17.2% of real sperm whales occurrences used as training data, respectively, whereas the other models failed to assign < 12%. The resulting maps indicated that the probability of occurrence of sperm whales differed between the surface model and the models for Levels 1–5 (Supplementary Figs 4 & 5). The Level 3 model, at c. 1,000 m depth, best represented the potential distribution of sperm whales in the Colombian Caribbean, with the highest AUC (0.84) and true skill statistic (0.55), and assigned presence to > 88% of real sperm whale occurrences (Table 3). The map resulting from this model indicated there is high probability of occurrence of sperm whales in the south and north-east Colombian Caribbean over the shelf break to waters up to c. 3,000 m deep, and near the Archipelago of San Andrés, Old Providence and Saint Catherine in the north-west (Fig. 3). The area of high probability of sperm whale occurrence is characterized by a distance to shore of 9.8–220.5 km (median 106.9 km) and an ocean mixed layer thickness of 5.2–50.3 m (median 29.7 m). The area of occurrence resulting from the model for Level 3 (Fig. 3) had similar features to the real occurrence data (Supplementary Table 5).
Fig. 3 Potential distribution of the sperm whale in the Colombian Caribbean Sea. Occurrence probability is based on the Level 3 model for environmental conditions at c. 1,000 m depth (Table 3).
Table 3 Definition of the six models, from the surface to a depth of c. 2,000 m, generated using Maxent, with the number of presences used as training data, number of presences with missing data, number of background points, modelling settings, and metrics of cross-validation model performance (area under the curve, true skill statistic, and optimal threshold). The best model is in bold. For additional details, see Supplementary Table 4 and Supplementary Material 2.
1 L, linear; Q, quadratic; P, product; T, threshold; H, hinge.
2 Area under the curve.
3 True skill statistic.
Discussion
This study provides the first assessment of the occurrence of sperm whales in the Colombian Caribbean, which appears to be an important habitat for this species. The mean encounter rates of 0.8 individuals and 0.4 groups per 100 h are similar to rates reported in the Gulf of Mexico (1.3 individuals and 0.6 groups per 100 h), where the sperm whale is considered the most abundant large cetacean (Barkaszi et al., Reference Barkaszi, Butler, Compton, Unietis and Bennet2012). The sperm whale density we recorded in the Colombian Caribbean (1.42 individuals per 1,000 km2) is similar to that reported worldwide (1.4 individuals per 1,000 km2; Whitehead, Reference Whitehead2002) and to that of other American tropical regions of the Pacific Ocean, but lower than reported for the Colombian Pacific (3.8 individuals per 1,000 km2; Gerrodette & Palacios, Reference Gerrodette and Palacios1996; Supplementary Table 6). Previous studies in the Caribbean had reported lower encounter rates for sperm whales (e.g. 0.35 individuals per 1,000 km2; Mullin & Fulling, Reference Mullin and Fulling2004). However, most of these data were from coastal research platforms. Our findings highlight the value of marine mammal occurrence data obtained during seismic surveys, which cover offshore areas that researchers may not usually be able to survey.
Whitehead et al. (Reference Whitehead, Antunes, Gero, Wong, Engelhaupt and Rendell2012) found that sperm whale social units in the North Atlantic are based around 6–12 often matrilineally related individuals that move together, raise their calves communally, and probably share important knowledge among themselves. Our findings for the Colombian Caribbean are similar, with groups of up to 10 individuals, and similar to the Gulf of Mexico where groups have up to 16 individuals (Barkaszi et al., Reference Barkaszi, Butler, Compton, Unietis and Bennet2012). Given our documentation of juveniles and the stranding of a juvenile in 2009 in the Urabá Gulf on the Colombian Caribbean coast (Trujillo et al., Reference Trujillo, Gärtner, Caicedo and Diazgranados2013), it is possible that the waters of the Colombian Caribbean are a breeding area for Atlantic sperm whales. Further studies are required to examine this possibility. Genetic and photo-identification studies are also required, to assess whether sperm whales sighted in the Colombian Caribbean belong to either of the better-known populations of the eastern Caribbean (Gordon et al., Reference Gordon, Moscrop, Caroson, Ingram, Leaper, Matthews and Young1998; Gero et al., Reference Gero, Gordon, Carlson, Evans and Whitehead2007) or the Gulf of Mexico (Weller et al., Reference Weller, Würsig, Lynn and Schiro2000), or whether the Colombian Caribbean is an area of connectivity between the eastern and western Atlantic populations. It has been proposed that the eastern Caribbean is a sink with favourable conditions for sperm whales (Whitehead & Gero, Reference Whitehead and Gero2015), but current threats related to human activities in this area (e.g. around the Lesser Antilles) are probably affecting the species (Whitehead & Gero, Reference Whitehead and Gero2015; Gero & Whitehead, Reference Gero and Whitehead2016). Although the sperm whale does not appear to conduct long migrations in equatorial waters, when feeding and survival conditions are poor sperm whales tend to roam widely (Whitehead, Reference Whitehead, Würsig, Thewissen and Kovacs2018). Therefore, considering the Caribbean is a relatively small basin, if Colombian waters offer suitable conditions, it is not surprising to find sperm whales from the eastern Caribbean there.
This is the first study to describe the potential distribution of sperm whales in the Colombian Caribbean, using high resolution spatio-temporal variables that are likely to influence sperm whale distribution. Some studies of cetacean distribution in Colombian waters have used line transect surveys (e.g. Palacios et al., Reference Palacios, Herrera, Gerrodette, García, Soler and Avila2012), which is the most widely used method to estimate cetacean occurrence. However, this method has high costs and logistical challenges, and low detectability for many cetacean species (Kaschner et al., Reference Kaschner, Quick, Jewell, Williams and Harris2012). Species distribution models, as used here, are useful to estimate the potential distribution of species, particularly in areas where there have not been any line transect surveys, such as in the offshore Colombian Caribbean. In our study, the model built with conditions at 1,000 m depth had the best performance; the surface model had a relatively poor performance, failing to predict probability of occurrence in areas where the species was recorded over the shelf break. This suggests that the analysis of sea surface conditions alone is insufficient to describe the distribution of sperm whales. This is not unexpected, as the sperm whale dives deeply to feed (Whitehead, Reference Whitehead2003; Evans & Hindell, Reference Evans and Hindell2004). Our results indicate that of the environmental variables tested, the most important were distance to shore and ocean mixed layer thickness. The model for 1,000 m depth identified that the area with a high probability of sperm whale occurrence is close to the shore (median = 106.9 km), with an average range in ocean mixed layer thickness of 29.7 m. These variables may be related to the presence of sperm whale prey. Areas close to the continental shoreline are influenced by rivers and their nutrients, which favour the presence of prey. The ocean mixed layer, which has homogeneous density, temperature and salinity, varies greatly in time and space (e.g. in subpolar latitudes it can be < 20 m in summer but > 200 m in winter; de Boyer Montégut et al., Reference de Boyer Montégut, Madec, Fischer, Lazar and Iudicone2004), and plays an important role in phytoplankton and food chain dynamics (Carvalho et al., Reference Carvalho, Kohut, Oliver and Schofield2017).
Our modelling indicates that the potential distribution area of sperm whales includes the south and north-east Colombian Caribbean over the shelf break to waters up to c. 3,000 m deep, and near the Archipelago of San Andrés, Old Providence and Saint Catherine in the north-west (Fig. 3). The south and north-east Colombian Caribbean are influenced by freshwater discharge from the Atrato, Sinú, Magdalena and Riohacha Rivers and their nutrients, and by upwelling off the southern Caribbean coast, which is probably the main nutrient source supporting biological productivity in this sea (Correa-Ramírez et al., Reference Correa-Ramírez, Rodriguez-Santana, Ricaurte-Villota and Páramo2020). Rivers provide a nutrient source for the oceans, and their plumes can be zones of high biological productivity, supporting phytoplankton and zooplankton (Montoya et al., Reference Montoya, Toro-Botero and Gómez-Giraldo2016). Additionally, such plumes can generate an alluvial fan and an offshore canyon (a geological process that occurs at the mouth of rivers), which could provide habitat suitable for deep-sea species such as the sperm whale. The Archipelago of San Andrés, Old Providence and Saint Catherine, designated a Seaflower Biosphere Reserve in 2000, has extensive coral reefs, seagrass beds and mangroves (CORALINA–INVEMAR, 2012), productive ecosystems that could provide prey for sperm whales. In addition, the Archipelago is close to the 2,350 m Roncador seamount (Idárraga-García & León, Reference Idárraga-García and León2019). This could explain the high probability of occurrence of sperm whales in this area, as in the Azores where seamount complexes are prefered habitat for the sperm whale (Tobeña et al., Reference Tobeña, Prieto, Machete and Silva2016). Sperm whales are deep divers and their main diet is cephalopods, usually subtropical and muscular cephalopod species of the Onychoteuthidae and Histioteuthidae families (Evans & Hindell, Reference Evans and Hindell2004), which range throughout the water column to at least 3,000 m depth. In the Colombian Caribbean, 48 cephalopod species have been documented, including two species of Onychoteuthidae (Guerrero-Kommritz, Reference Guerrero-Kommritz2021).
Our findings indicate that the Caribbean waters of Colombia may be an important tropical feeding and breeding habitat for sperm whales. In the Caribbean Region there are increasing pressures on marine mammals from coastal development, fishing, boat traffic, river sediment loading, alien species and climate change (Miloslavich et al., Reference Miloslavich, Díaz, Klein, Alvarado, Díaz and Gobin2010; SPAW-RAC, 2020; Avila & Giraldo, Reference Avila and Giraldo2022). Sperm whales in the region are affected by incidental catch in fishing nets and collision with boats (SPAW-RAC, 2020). In the Colombian Caribbean Sea commercial fisheries are concentrated in the south and north-east (Kroodsma et al., Reference Kroodsma, Mayorga, Hochberg, Miller, Boerder and Ferretti2018), areas where there is a high probability of sperm whale presence. The Colombian Caribbean region is also an international and national tourist destination (during 2000–2019 an average of 145 tourist cruise ships > 150 m long arrived annually in San Andrés, Cartagena and Santa Marta; CITUR, Reference Carvalho, Kohut, Oliver and Schofield2021). The transit of such large ships puts whales at risk of collisions (Laist et al., Reference Laist, Knowlton, Mead, Collet and Podesta2001). In 2009 the tourist cruise ship Summer Flower (169 m long) on the route from Santa Marta (Colombia) to Antwerp (Belgium) collided with a fin whale Balaenoptera physalus, and arrived in Belgium with the dead whale on its bow (Haelters et al., Reference Haelters, Kerckhof and Jauniaux2018).
Our analyses suggest that the Vulnerable sperm whale may be both feeding and breeding in the Colombian Caribbean Sea, and provides information that will be useful for management of this cetacean species. Efforts for the conservation and sustainable use of the distribution area of this species, identified here, need to be implemented in this region.
Acknowledgements
We thank Anadarko Colombia Company (since 2019, Occidental Petroleum Corporation), Repsol Exploración Colombia and Shell for allowing the use of data. OSC Limited, EPI Group, CGG & Vision Project Services provided the services of marine fauna observers and passive acoustic monitoring operators and we acknowledge all observers and operators who supported data collection (Salomé Dussán, Juan Manuel Salazar, Alexandra Gärtner, Chad Leedy, Israel Ribeiro, James Doom, Marilia Olio, Leif Halvors and Nicholas Engelmann). We thank fishers from San Andrés and Luis F. Batista from Urabá Gulf for providing information on whales, Javier Alarcón for provision of a photograph, Hal Whitehead for providing access to his data, Ana Marroquim, Paula Gómez, Andrés Palacios and Octavio Ortega for comments on the text, Greg Jenssen for editing help, and two anonymous reviewers and the Editor for their critiques. ICA and DCB-B were supported in 2020–2021 by the Universidad del Valle and Centro de Investigaciones Oceanográficas e Hidrográficas del Caribe, respectively, and the Colombian Sciences Ministry (Call no. 848 of 2019).
Author contributions
Study design: ICA; data collection: ICA, NF-C, LC-M; data analysis: ICA, NF-C, KBdA; writing: ICA, NF-C; revision, editing: all authors.
Conflicts of interest
None.
Ethical standards
This research abided by the Oryx guidelines on ethical standards.
Open access
