TY - JOUR
AU - Aurioles-Gamboa, David
AB - Abstract Bottlenose dolphins (Tursiops truncatus) and pantropical spotted dolphins (Stenella attenuata) co-occur in Golfo Dulce, a fjord-like embayment located in the southern Pacific coast of Costa Rica. We evaluated if spatial overlap by these dolphin species is associated with similar environmental requirements. Presence-only models based on niche theory were constructed by contrasting a set of spatial locations with the responses of the target species to abiotic ecogeographical variables (EGVs: depth, slope, distance to rivers, distance to 200 m isobath, mean sea surface temperature, and variation in sea surface temperature). Models were cross-validated with levels of discrimination that ranged from acceptable to excellent based on the area under the curve assessment (T. truncatus, rainy season: 0.76, dry season: 0.83; S. attenuata, rainy season: 0.84, dry season: 0.89). Both dolphin species occur in Golfo Dulce year-round; the lack of seasonality documented previously was supported by the models. Species distribution models showed no spatial overlap, with differences in EGVs affecting their distribution (T. truncatus: distance to river + distance to 200 m isobath, S. attenuata: depth + sea surface temperature). We argue that the coexistence of both predators in Golfo Dulce is linked to habitat heterogeneity, where critical habitats are spatially differentiated. The lack of fine-scale spatial overlap, along with influential abiotic variables, highlights a process of coexistence for dolphins that are sympatric at the scale of Golfo Dulce, but within the Gulf there is fine-scale allopatry. El delfín nariz de botella (Tursiops truncatus) y el delfín manchado pantropical (Stenella attenuata) son delfínidos presentes en el Golfo Dulce, una bahía similar a un fiordo, en el litoral del Pacífico Sur de Costa Rica. Este estudio evalúa si el traslape espacial en estas especies se asocia a requerimientos ambientales similares. Se hicieron modelos de “solo presencia” basados en la teoría de nichos, que ponderan una muestra de localidades geográficas con la respuesta de las especies evaluadas a variables ecogeográficas abióticas (VEGs: Profundidad, Pendiente, Distancia a Ríos, Distancia a la Isobata 200 m, Promedio de la Temperatura Superficial del Mar y Variación de la Temperatura Superficial del Mar). Los modelos se sometieron a validación cruzada con niveles de discriminación en un rango que va desde aceptable a excelente, en base a la medición del Área Bajo la Curva (T. truncatus, temporada de lluvia: 0.76, temporada seca: 0.83; S. attenuata temporada de lluvia: 0.84, temporada seca: 0.89). Ambas especies de delfines están presentes todo el año, los modelos sustentaron la falta de estacionalidad documentada previamente. Los modelos de distribución de las especies no muestran traslape espacial, estos evidencian diferencias en VEGs que afectan la distribución de los delfines (T. truncatus: distancia a ríos + distancia a isobata de 200 m, S. attenuata: profundidad + temperatura superficial del mar). La coexistencia de ambos depredadores en el Golfo Dulce está ligada a la heterogeneidad de hábitat, donde los hábitats críticos están espacialmente diferenciados. La falta de traslape en la dimensión espacial del hábitat, incluyendo la segregación de las variables abióticas, resaltan un proceso de coexistencia para delfínidos que se caracteriza como una simpatría aparente a la escala del Golfo Dulce con alopatría a escala fina dentro del Golfo. coexistence, Costa Rica, ecological niche modeling, Golfo Dulce, habitat partitioning, spatial overlap, Stenella attenuata, sympatry, Tursiops truncatus A crucial part of the survival of a species includes either outcompeting surrounding species or coexisting with them. Sympatric species are those with geographical co-occurrence and the potential for home-range overlaps (Heinrich et al. 2010). Home-range overlap does not necessarily imply direct interaction, due to the potential for species to occupy different microhabitats, have distinct dietary preferences, or show different circadian activity patterns (Davies et al. 2007). Consequently, currently accepted ecological theory predicts that stable communities of coexisting species will diverge to some degree in resource utilization, including but not limited to prey species, habitat selection, and diel patterns (Roughgarden 1976; Day 2000). Coexistence is associated with segregation of occupied niches or “partitioning” of resources. Niche segregation is expected to occur at finer scales, such as the microhabitat level, where environmental factors affect individual behavior (Morris 1996). Furthermore, niche segregation between generalists and specialists also is affected by scale, where there could be underused or unused resources by the specialist at fine spatial scales (Morris 1996; Leimgruber et al. 2014); that is the case for the scale-dependent habitat selection between the pine marten (Martes martes) and the more behaviorally plastic stone marten (M. foina) in the northern Iberian Peninsula (Vergara et al. 2016). Ecological models, such as the Lotka–Volterra model, predict the stable coexistence of 2 competitors in situations where interspecific competition would be, for both species, less significant than intraspecific competition. As a result, niche differentiation will buffer competition between species and tend to concentrate competitive effects among conspecifics. To understand the potential mechanisms facilitating coexistence by species with overlapping ranges, segregation along different niche dimensions should be examined: spatially, by detailing the key environmental factors shaping the spatial distribution; temporally, by describing resource selection and use in time; and, finally, in the trophic dimension, through evaluating differences in diet, prey contribution, and trophic level. Insights on the ecological patterns implicit in coexistence of odontocetes, and particularly delphinids, are scant due to constraints associated with experimental approaches. Empirical data based on field observations support dietary differentiation as the major ecological process to buffer competitive interactions and promote coexistence (Bearzi 2005, 2007), resulting in divergence in habitat use, including its main components: abundance and distribution. Such differentiation can develop in a gradient of spatiotemporal partitioning. For example, in habitats where resources are abundant, there should be a low level of spatial partitioning and a considerable degree of dietary overlap, with species intermingling together, as in the case of white-sided (Lagenorhynchus acutus) and common dolphins (Delphinus delphis) in the Gully, off Nova Scotia, Canada (Gowans and Whitehead 1995), or common dolphins and dusky dolphins (L. obscurus) in Patagonia (Romero et al. 2012; Svendsen et al. 2015). The other side of the spectrum would be complete spatial differentiation; such segregation is generally described by ecogeographical variables (EGVs) influencing dolphin occurrence. For instance, common and white-beaked dolphins (L. albirostris) in waters off the United Kingdom and Ireland (Macleod et al. 2008) and Scotland (Macleod et al. 2007; Weir et al. 2009) showed a switch in dominant species from common to white-beaked dolphins mediated by a temperature threshold (12°C). This segregation provides evidence of potential exclusion of the species with lower tolerance to an increase in temperature. Physical and hydrographical variables have been used to describe cetacean occurrence and habitat use (Doksæter et al. 2008; Praca and Gannier 2008; Gross et al. 2009; Fury and Harrison 2011), but they might only be proxies for prey type and abundance (Benson et al. 2002). We explored the coexistence of inshore bottlenose dolphins (Tursiops truncatus) and pantropical spotted dolphins (Stenella attenuata), occurring year-round in a fjord-like embayment along the Pacific coast of Costa Rica, facilitated by fine-scale habitat partitioning. Species distribution models (SDMs) based on niche theory (Phillips et al. 2006; Friedlaender et al. 2011; Thorne et al. 2012) provided the analytical framework to elucidate spatial segregation, associated with the environmental requirements for each predator. This analysis is based on the premise that fine-scale spatial overlap is supported by the coincidence of at least 1 variable describing the ecological niche of both species of predator. The information presented in this study deals with the spatial dimension of both predators’ foraging habitat. Materials and Methods Study area. Golfo Dulce (GD hereafter) is located at 8°30′N, 83°16′W. This embayment is treated in this assessment as a subsystem of the Osa Peninsula in Costa Rica, which is located in the Nicoya Ecoregion (Spalding et al. 2007; Fig. 1). GD is characterized by a deep inner basin (> 215 m maximum depth) and a shallow sill (70 m), which communicates the inner basin with the open Eastern Tropical Pacific (Wolff et al. 1996; Quesada-Alpizar and Cortes 2006). The total surface area is close to 750 km2, water circulation is restricted as in true fjords, and there is a slow deep-water renewal by occasional intrusion of dense subsurface waters (Svendsen et al. 2006). Productivity in GD’s inner basin is most likely subsidized by riparian discharge to the inner basin, particularly by contributions of the Esquinas, Rincón, Tigre, and Coto Colorado rivers (Fig. 1). Fig. 1. View largeDownload slide Study area: Golfo Dulce, Costa Rica, with details on research effort (survey tracks) and encounters of pantropical spotted dolphins (Stenella attenuata, circles) and inshore bottlenose dolphins (Tursiops truncatus, triangles), 2011–2015. Fig. 1. View largeDownload slide Study area: Golfo Dulce, Costa Rica, with details on research effort (survey tracks) and encounters of pantropical spotted dolphins (Stenella attenuata, circles) and inshore bottlenose dolphins (Tursiops truncatus, triangles), 2011–2015. Dolphin surveys and behavior sampling. We carried out dolphin behavior and photo ID surveys using as research platform a 9-m boat, powered by a 115-HP 4-stroke engine, during 2 seasons, rainy (June–October) and dry (November–May), from March 2005 to March 2015. A representative coverage of the study area was achieved by allocating search effort among sectors (inner basin, sill area, transitional-oceanic) within each season (Oviedo et al. 2015); therefore, a complete area survey required 2–3 days, during which each subarea was navigated along a zigzag course. To obtain a reliable estimate of occurrence patterns of the target species, we recorded presence-absence data based on a point-sample design. We documented the parameters sea surface temperature (SST; using a handheld field thermometer), tide, wind speed on the surface (Beaufort scale), visibility, and boat location (GPS), every half hour at monitoring stations as described by Gowans and Whitehead (1995). We constructed a presence-absence matrix via 5-min scans while monitoring detectability throughout the survey. A total of 2,619 sampling points was used for this study (2011–2015). Data were grouped into the 2 main climatological and oceanographical seasons present in the study area, dry season with 962 points sampled and rainy season with 1,357 points sampled. Additionally, presence-only records (2005–2015, n = 674) for both species were used as a test data set for the model evaluation. All procedures followed American Society of Mammalogists (Sikes et al. 2016) and national Costa Rican guidelines for research on live animals and cetaceans. Ecogeographical variables. Six environmental variables were selected according to the biological information available for the species in the area (Acevedo-Gutierrez and Burkhart 1998; Cubero Pardo 1998a, 1998b, 2007a, 2007b; Oviedo 2007; Oviedo et al. 2009, 2015; Herra-Miranda et al. 2016). We used the GEBCO 2014 Grid bathymetry (The GEBCO_2014 Grid, version 20150318, www.gebco.net) as a base to obtain all the geographical and depth-derived variables. The slope, distance to the 200 m depth-contour line, and the distance to major rivers were obtained using QGIS and the integrated SAGA and GRASS toolboxes (QGIS Development Team 2015). Two oceanographic variables were used, the mean SST and the SST standard deviation. These layers were obtained from the MARSPEC repository (Sbrocco and Barber 2013), using long-term monthly climatological means obtained from remotely sensed and in situ oceanographic observations. The SST standard deviation was calculated to express the variability in each season. The mean value and the standard deviation of each season were estimated using the corresponding months for each case, due to the lack of a dispersion metric in the MARSPEC data set. Four oceanographic layers were built by stratifying these variables into dry and rainy season. All the strata were tested for autocorrelation using a Pearsons test. Correlation was lower than 0.7 for all strata. All the variables were resampled to a 1-km grid resolution. Modeling approach. We used 2 different procedures to obtain reliable estimates of the realized niche. Recent studies (Qiao et al. 2015) highlighted the importance of using more than 1 algorithm for niche modeling, choosing the best model based on evaluation tests. As a first step, we applied a set of generalized additive models (GAMs) as an explanatory analysis. GAMs are widely used statistical modeling tools that facilitate the analysis of the relationships between species distributions and their environmental correlates, resulting in a flexible description of complex species responses to environmental constrains (Leathwick et al. 2006). GAMs are based on the use of nonparametric smoothing functions; they provide a general framework for extending a standard linear model by allowing nonlinear functions of each of the variables while maintaining additivity (James et al. 2013). This analysis was performed using the “mgcv” R package; we used a cubic regression spline shrinkage smoother to penalize the factors less relevant to the final model (Marra and Wood 2011). An offset representing the number of points sampled in each grid cell was included to take into consideration the effort. The second modeling approach was boosted regression trees (BRT) or generalized boosted regression models (GBMs). This technique can be described as a combination of a classical statistical approach (regression trees) and a machine learning (ML) technique (boosting). The inclusion of a ML method adds considerable advantages compared to conventional methods, including the improvement of model selection (Elith et al. 2008). This approach examines a large number of trees and using a boosting approach to select a linear combination of many trees (usually from hundreds to thousands). Fitted values in the final model are computed as the sum of all trees weighted by an estimate of the contribution of each tree to the growing model. A relatively slow learning rate (0.001) with a higher tree complexity (5) was selected following recommendations of Elith et al. (2008), to aim for more than 1,000 trees. Again, the effort was included as an offset term, representing the number of points sampled per grid. Model validation. The model validation was performed using a semi-independent data set, a presence-only data set collected between sampling points. This data set is constructed with the information collected at the beginning of each sighting: date and location through GPS coordinates, best estimates on group size, group composition, and behavior at first encounter. While the data sets were not totally independent, the use of this method allowed us to obtain a test data set in areas not used to build the model, which results in a better performance for the models and provides better estimates of their predictive capability. The area under the curve (AUC) metric of the receiving operator characteristic (ROC) curve (Phillips et al. 2006) was used to evaluate model prediction. The AUC value provides a threshold-independent metric of overall accuracy; a value of 0.5 indicates that the model prediction of presence-absence is no better than random, while greater scores indicate increasing predictive ability. We assessed AUC values of the ROC curve of the models following the scale suggested in Hosmer and Lemeshow (1989): 0.5 indicated no discrimination; 0.5 to 0.7 represented poor discrimination; 0.7 to 0.8 indicated an acceptable discrimination; 0.8 to 0.9 indicated an excellent discrimination; and greater than 0.9 represented outstanding discrimination. All analysis and figures were produced using R 3.2.2 (www.r-project.org) with the raster, reshape2, MASS, dismo, SDMTools, ecodist, pROC, PMCMR, mgcv, and gbm packages. Results A total of 192 and 258 sightings of inshore bottlenose dolphins were used for the dry and rainy season, respectively. A total of 95 and 124 pantropical spotted dolphins sightings were used for the dry and rainy season, respectively. Models performance and validation for inshore bottlenose dolphins and pantropical spotted dolphins in Golfo Dulce. For the 2 sets of models applied, the best predictive power was obtained by the GBM (Table 1). While for the pantropical spotted dolphins the AUC test values obtained by the GAMs were still reasonably good (around 0.8), the results for the bottlenose dolphins were quite low. For this species, the deviance was lower, but moreover the test AUC values were extremely low (AUC ≤ 0.65). Almost all the test AUC values for the GBM approach were in the excellent range (AUC > 0.8), the exception was that for bottlenose dolphins the rainy season values were slightly lower but in the acceptable range (AUC = 0.76). Table 1. Area under the curve (AUC) values for pantropical spotted dolphins (Stenella attenuata, Sa) and inshore bottlenose dolphins (Tursiops truncatus, Tt) during dry and rainy seasons. The range of AUC values in boosted regression trees (GBM), show a better discriminative performance than generalized additive models (GAMs) for inshore bottlenose dolphins, reflecting the more generalist behavior of the species. Standard accuracy value Dry Rainy Sa Tt Sa Tt GAM  Deviance explained (%) 47.6 15.6 46.3 26  AUC test 0.87 0.65 0.83 0.62 GBM  AUC test 0.89 0.83 0.84 0.76 Standard accuracy value Dry Rainy Sa Tt Sa Tt GAM  Deviance explained (%) 47.6 15.6 46.3 26  AUC test 0.87 0.65 0.83 0.62 GBM  AUC test 0.89 0.83 0.84 0.76 View Large Table 1. Area under the curve (AUC) values for pantropical spotted dolphins (Stenella attenuata, Sa) and inshore bottlenose dolphins (Tursiops truncatus, Tt) during dry and rainy seasons. The range of AUC values in boosted regression trees (GBM), show a better discriminative performance than generalized additive models (GAMs) for inshore bottlenose dolphins, reflecting the more generalist behavior of the species. Standard accuracy value Dry Rainy Sa Tt Sa Tt GAM  Deviance explained (%) 47.6 15.6 46.3 26  AUC test 0.87 0.65 0.83 0.62 GBM  AUC test 0.89 0.83 0.84 0.76 Standard accuracy value Dry Rainy Sa Tt Sa Tt GAM  Deviance explained (%) 47.6 15.6 46.3 26  AUC test 0.87 0.65 0.83 0.62 GBM  AUC test 0.89 0.83 0.84 0.76 View Large For inshore bottlenose dolphins, GAMs identified slope, distance to the 200 m isobath, and distance to rivers as significant predictors of occurrence during the dry season, and distance to the 200 m isobath, slope, and SST during the rainy season (Table 2). Encounters of inshore bottlenose dolphins during the dry season were inversely related to distance to the rivers, distance to the 200 m isobath, and a slope gradient between 0° and 5° (Fig. 2). During the rainy season, sightings increased with distance to the 200 m isobaths, corresponding with shallow inshore waters, with slope values between 3° and 6° and more variation in SST (0.6° to 0.8°; Fig. 3). Table 2. Ecogeographical variables (EGVs) and responses for the generalized additive models for inshore bottlenose dolphin (Tursiops truncatus, Tt) and pantropical spotted dolphins (Stenella attenuata, Sa) during dry and rainy seasons. Species Season EGV Response P-value Tt Dry Depth None Distance to river Negative < 0.05 Distance to 200 m Positive < 0.01 Slope Nonlinear—peaks at S = 5° to 10° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) None Tt Rainy Depth None Distance to river None Distance to 200 m Nonlinear—peaks at D = 10–30 km < 0.01 Slope Nonlinear—peaks at S = 3° to 6° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) Nonlinear—peaks at SD = 0.6–0.8 < 0.01 Sa Dry Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.05 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) Nonsignificant > 0.05 Sa Rainy Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.01 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) None Species Season EGV Response P-value Tt Dry Depth None Distance to river Negative < 0.05 Distance to 200 m Positive < 0.01 Slope Nonlinear—peaks at S = 5° to 10° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) None Tt Rainy Depth None Distance to river None Distance to 200 m Nonlinear—peaks at D = 10–30 km < 0.01 Slope Nonlinear—peaks at S = 3° to 6° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) Nonlinear—peaks at SD = 0.6–0.8 < 0.01 Sa Dry Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.05 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) Nonsignificant > 0.05 Sa Rainy Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.01 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) None View Large Table 2. Ecogeographical variables (EGVs) and responses for the generalized additive models for inshore bottlenose dolphin (Tursiops truncatus, Tt) and pantropical spotted dolphins (Stenella attenuata, Sa) during dry and rainy seasons. Species Season EGV Response P-value Tt Dry Depth None Distance to river Negative < 0.05 Distance to 200 m Positive < 0.01 Slope Nonlinear—peaks at S = 5° to 10° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) None Tt Rainy Depth None Distance to river None Distance to 200 m Nonlinear—peaks at D = 10–30 km < 0.01 Slope Nonlinear—peaks at S = 3° to 6° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) Nonlinear—peaks at SD = 0.6–0.8 < 0.01 Sa Dry Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.05 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) Nonsignificant > 0.05 Sa Rainy Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.01 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) None Species Season EGV Response P-value Tt Dry Depth None Distance to river Negative < 0.05 Distance to 200 m Positive < 0.01 Slope Nonlinear—peaks at S = 5° to 10° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) None Tt Rainy Depth None Distance to river None Distance to 200 m Nonlinear—peaks at D = 10–30 km < 0.01 Slope Nonlinear—peaks at S = 3° to 6° < 0.01 Sea surface temperature ( X¯) Nonsignificant > 0.05 Sea surface temperature (SD) Nonlinear—peaks at SD = 0.6–0.8 < 0.01 Sa Dry Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.05 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) Nonsignificant > 0.05 Sa Rainy Depth Nonlinear—peaks at Z < 100 m < 0.01 Distance to river Nonlinear—peaks at D = 5–10 km < 0.01 Distance to 200 m Negative < 0.01 Slope Negative < 0.05 Sea surface temperature ( X¯) Positive < 0.01 Sea surface temperature (SD) None View Large Fig. 2. View largeDownload slide Ecogeographical variables (EGVs) responses for inshore bottlenose dolphins (Tursiops truncatus) derived from a generalized additive model (GAM), during the dry season in Golfo Dulce, Costa Rica (2011–2015). Slope, distance to the 200 m isobath, and distance to rivers are significant predictors of dolphin occurrence during the dry season. SST = sea surface temperature. Fig. 2. View largeDownload slide Ecogeographical variables (EGVs) responses for inshore bottlenose dolphins (Tursiops truncatus) derived from a generalized additive model (GAM), during the dry season in Golfo Dulce, Costa Rica (2011–2015). Slope, distance to the 200 m isobath, and distance to rivers are significant predictors of dolphin occurrence during the dry season. SST = sea surface temperature. Fig. 3. View largeDownload slide Ecogeographical variables (EGVs) responses for inshore bottlenose dolphins (Tursiops truncatus) derived from a generalized additive model (GAM), during the rainy season in Golfo Dulce, Costa Rica (2011–2015). Distance to the 200 m isobath, slope, and the variation of sea surface temperature (SST) are significant predictors of dolphin occurrence during the rainy season. Fig. 3. View largeDownload slide Ecogeographical variables (EGVs) responses for inshore bottlenose dolphins (Tursiops truncatus) derived from a generalized additive model (GAM), during the rainy season in Golfo Dulce, Costa Rica (2011–2015). Distance to the 200 m isobath, slope, and the variation of sea surface temperature (SST) are significant predictors of dolphin occurrence during the rainy season. The final model run for pantropical spotted dolphins highlighted the interplay of more diverse EGVs. Depth, distance to the 200 m isobath, mean SST, distance to rivers, and slope were significant predictor EGVs, describing the habitat that favored the occurrence of pantropical spotted during rainy and dry seasons (Table 2). The occurrence of pantropical spotted dolphins during the dry season was affected primarily by depth greater than 100 m, inversely related to distance to the 200 m isobaths, and positively affected by an increase in SST. Other significant variables that influenced the occurrence of S. attenuata during the dry season were the distance to rivers draining to the inner basin (5–10 km) and low values of slope gradient (Fig. 4). The rainy season GAM for pantropical spotted dolphin was similar to the dry season model (Table 2; Fig. 5). Fig. 4. View largeDownload slide Ecogeographical variables (EGVs) responses for pantropical spotted dolphins (Stenella attenuata) derived from a generalized additive model (GAM), during the dry season in Golfo Dulce, Costa Rica (2011–2015). Depth over 100 m, slope, the distance to the 200 m isobaths, the distance to rivers, and the increase in mean values of sea surface temperature (SST) are significant predictors of dolphin occurrence in the dry season. Fig. 4. View largeDownload slide Ecogeographical variables (EGVs) responses for pantropical spotted dolphins (Stenella attenuata) derived from a generalized additive model (GAM), during the dry season in Golfo Dulce, Costa Rica (2011–2015). Depth over 100 m, slope, the distance to the 200 m isobaths, the distance to rivers, and the increase in mean values of sea surface temperature (SST) are significant predictors of dolphin occurrence in the dry season. Fig. 5. View largeDownload slide Ecogeographical variables (EGVs) responses for pantropical spotted dolphins (Stenella attenuata) derived from a generalized additive model (GAM), during the rainy season in Golfo Dulce, Costa Rica (2011–2015). Depth over 100 m, slope, the distance to the 200 m isobaths, the distance to rivers, and the increase in mean values of sea surface temperature (SST) are significant predictors of dolphin occurrence in the rainy season. Fig. 5. View largeDownload slide Ecogeographical variables (EGVs) responses for pantropical spotted dolphins (Stenella attenuata) derived from a generalized additive model (GAM), during the rainy season in Golfo Dulce, Costa Rica (2011–2015). Depth over 100 m, slope, the distance to the 200 m isobaths, the distance to rivers, and the increase in mean values of sea surface temperature (SST) are significant predictors of dolphin occurrence in the rainy season. The GBM results highlighted the importance of depth (Table 3), particularly for the pantropical spotted dolphin (dry season: 47.71%; rainy season: 44.92%). The distance to the 200 m isobath was important, especially during the rainy season for both species (bottlenose dolphins: 26.10%; pantropical spotted dolphins: 24.43%). Distance to rivers was relatively important for the bottlenose dolphins, with higher values during the dry season (19.67%). The mean SST values were more important for bottlenose dolphins (19.53%) than for pantropical spotted dolphin (18.90%). The best models selected for inshore bottlenose dolphins and pantropical spotted dolphins in GD corresponded with the GBM runs (Fig. 6). Table 3. Relative influence (%) of the ecogeographical variables (EGVs) for the generalized boosted model (boosted regression trees) for inshore bottlenose dolphins (Tursiops truncatus, Tt) and pantropical spotted dolphins (Stenella attenuata, Sa) during dry and rainy seasons. The 3 main variables for each group that contribute the most to explain the variation are marked using an asterisk. EGV Tt-dry Tt-rainy Sa-dry Sa-rainy Depth 18.47* 17.59* 47.71* 44.92* Slope 11.19 14.70 7.41 7.39 Distance to 200 m 14.86 26.10* 8.02 24.43* Distance to river 19.67* 17.92* 13.88* 8.35 Sea surface temperature ( X¯) 19.53* 15.83 18.90* 10.73* Sea surface temperature (SD) 16.28 7.84 4.06 4.14 EGV Tt-dry Tt-rainy Sa-dry Sa-rainy Depth 18.47* 17.59* 47.71* 44.92* Slope 11.19 14.70 7.41 7.39 Distance to 200 m 14.86 26.10* 8.02 24.43* Distance to river 19.67* 17.92* 13.88* 8.35 Sea surface temperature ( X¯) 19.53* 15.83 18.90* 10.73* Sea surface temperature (SD) 16.28 7.84 4.06 4.14 View Large Table 3. Relative influence (%) of the ecogeographical variables (EGVs) for the generalized boosted model (boosted regression trees) for inshore bottlenose dolphins (Tursiops truncatus, Tt) and pantropical spotted dolphins (Stenella attenuata, Sa) during dry and rainy seasons. The 3 main variables for each group that contribute the most to explain the variation are marked using an asterisk. EGV Tt-dry Tt-rainy Sa-dry Sa-rainy Depth 18.47* 17.59* 47.71* 44.92* Slope 11.19 14.70 7.41 7.39 Distance to 200 m 14.86 26.10* 8.02 24.43* Distance to river 19.67* 17.92* 13.88* 8.35 Sea surface temperature ( X¯) 19.53* 15.83 18.90* 10.73* Sea surface temperature (SD) 16.28 7.84 4.06 4.14 EGV Tt-dry Tt-rainy Sa-dry Sa-rainy Depth 18.47* 17.59* 47.71* 44.92* Slope 11.19 14.70 7.41 7.39 Distance to 200 m 14.86 26.10* 8.02 24.43* Distance to river 19.67* 17.92* 13.88* 8.35 Sea surface temperature ( X¯) 19.53* 15.83 18.90* 10.73* Sea surface temperature (SD) 16.28 7.84 4.06 4.14 View Large Fig. 6. View largeDownload slide Habitat suitability from generalized boosted models (boosted regression trees) for inshore bottlenose dolphins (Tursiops truncatus, upper panels) and pantropical spotted dolphins (Stenella attenuata, lower panels) in Golfo Dulce, Costa Rica, during the rainy (right panels) and dry season (left panels). Fig. 6. View largeDownload slide Habitat suitability from generalized boosted models (boosted regression trees) for inshore bottlenose dolphins (Tursiops truncatus, upper panels) and pantropical spotted dolphins (Stenella attenuata, lower panels) in Golfo Dulce, Costa Rica, during the rainy (right panels) and dry season (left panels). Discussion Our analysis showed that there is no evident spatial overlap between areas with the highest habitat suitability for both predators. The major proportion of suitable habitat for inshore bottlenose dolphins in GD during the dry and rainy seasons was along the coastline, specifically associated with areas at or near the mouth of rivers such as Esquinas and Coto Colorado, with particular reference to the subsystem formed by the Tigre and Platanares rivers. For pantropical spotted dolphins, the area with the most suitable habitat was located at the center of the inner basin, within the 100 m isobath. Habitat-model performance. In general, all models had good performance. The use of a semi-independent data set as validation supported the predictive capability of our models. Nevertheless, GBMs outperformed GAMs, especially for the bottlenose dolphin. This result might be due to the more generalist and plastic behavior of this species (Wells and Scott 2008), where the higher flexibility of the GBM approach (James et al. 2013) might be more suitable. Habitat partitioning between sympatric dolphins in Golfo Dulce. Inshore bottlenose and pantropical spotted dolphins did not share the same habitat within the discrete geographic space of GD. The hypothesis of fine-scale spatial differentiation was supported by the characterization of EGVs describing suitable habitat and the contrasting locations of dolphin sightings. The lack of overlap in the spatial dimension was conditioned by the environmental factors bounding habitat suitability. A suite of complex responses characterized habitat suitability for each species. Habitat for pantropical spotted dolphins was consistent with bathymetric features of the inner basin: a rather low slope and deep waters over 100 m. Additionally, the increase in mean SST was associated with locations at considerable distances from fresh water input. Suitable habitat for inshore bottlenose dolphins appeared to be characterized by different EGVs depending on season. During the dry season, suitable habitat was associated with major river drainages at distances of 5 to 15 km away from the 200 m isobath. Suitable habitat for bottlenose dolphins during the rainy season was still affected by rivers, but at distances closer to the 200 m isobaths (between 5 and 10 km). Due to the complex physiography at GD, this distance corresponds to inshore waters, where the variation in mean SST, in magnitudes of 0.6° to 0.8°, would essentially describe areas under the influence of river drainages; this area is substantial during the rainy season due to large amounts of continental water flow. Even though there were differences in the EGVs that affected habitat suitability of these species, the spatial patterns during the rainy and dry seasons showed weak or no seasonality, confirming the year-round occurrence of both delphinids in GD (Oviedo et al. 2015). The relevance of depth as a facilitator of habitat segregation of these species in the Gulf has been previously documented (Acevedo-Gutierrez and Burkhart 1998; Cubero-Pardo 1998a, 2007a; Oviedo 2007, 2008). Oviedo (2007) established the relevance of the 100 m isobath as a limiting factor that separates the realized niche of the inshore bottlenose dolphin from that of the pantropical spotted dolphin, the latter occurring less frequently in shallow water, while the former tolerates a wider breadth of depths. Oviedo (2007) discussed habitat segregation at a fine-scale level as the main facilitator of the coexistence of these species in GD. Habitat segregation described in our analysis was based only on the abiotic factors we measured for GD. We acknowledge the need to incorporate biotic variables in the assessment of ecological niches (Peterson 2011). In addition, hydrographic factors shaping the distribution of cetaceans are often intertwined with one another, including the link to bathymetric features of the habitat (Weir et al. 2012). Nevertheless, areas defined by the highest habitat suitability in our study coincide with those identified as critical foraging habitat, for both inshore bottlenose dolphins (Oviedo 2007; Herra-Miranda et al. 2016) and pantropical spotted dolphins (Oviedo 2007, 2008; Oviedo et al. 2015) based on behavioral sampling. Both dolphin species also conduct key life-history activities within GD; active reproduction and calving have been observed in addition to foraging (Oviedo et al. 2015; Herra-Miranda et al. 2016). Demographic information is required to assess fitness differences, in order to specify how inshore bottlenose and pantropical spotted dolphins differ in their level of adaptation to their common environment (Morris 1996). The effect of Golfo Dulce’s habitat heterogeneity on the coexistence of sympatric dolphins. The definition of sympatry for highly mobile organisms, such as dolphins in GD, could be obscured by the effects of scale. Both inshore bottlenose dolphins and pantropical spotted dolphins move among patches of habitat with different suitability attributes for each species. Those habitats make a mosaic that leads to an apparent sympatry at the scale of the Gulf, but once habitat use is fitted to a well-defined gradient, finer-scale differences in the distribution of each species become evident. Therefore, habitat heterogeneity underlies the spatially close coexistence observed within the GD (Chesson 2000; Parra 2006; Adler et al. 2013). Foraging bouts of bottlenose and spotted dolphins in GD indicate areas of prey availability and productivity (L. Oviedo, D. Herra-Miranda, and J. D. Pacheco-Polanco, pers. obs.), and also provide support for the occurrence of environmental heterogeneity, in contrast to the previous perception of this inshore habitat as a nonheterogeneous, low-productivity environment (Cubero-Pardo 2007a). Adler et al. (2013) considered the role of spatial heterogeneity in the process of coexistence: when different species are favored in different local environments, species show spatial or resource partitioning. Spatial and environmental divergence might reflect selection of the environment where each species’ prey is available. In the case of pantropical spotted dolphins, highly abundant schooling fish in dense aggregations, including flying fish from the family Exocoetidae and ballyhoo (Hemiramphus sp.) among other prey, have been observed where these dolphins were encountered (Oviedo 2008). For inshore bottlenose dolphins, a variety of demersal fish, particularly needle fish (Tylosurus sp.), have been reported as preferred prey (Pacheco-Polanco 2015). The pattern of spatial segregation described in this study contrasts with the spatial overlap documented for other sympatric dolphins populations at higher latitudes; for example, common and dusky dolphins exhibited spatial overlap in San Matias Gulf, Argentina (Svendsen et al. 2015), which was reflected by the consumption of Argentine anchovies as the main prey of both predators (Romero et al. 2012). Similarly, neritic common and striped dolphins (D. delphis and S. coeruleoalba) overlapped in space use and partially overlapped in prey consumption (Pusineri et al. 2008) off the Bay of Biscay. Parra (2006) and Parra et al. (2006, 2011) also documented spatial overlap between tropical snubfin dolphins (Orcaella heinsohni) and Indo-Pacific humpback dolphins (Sousa chinensis) off northeastern Australia, including partial coincidences in their fish prey (Parra and Jedensjö 2014), but with marked teuthophagia only by snubfin dolphins. According to Adler et al. (2013) spatial heterogeneity promotes coexistence when different functional traits are favored in different environments. The interspecific relationship of sympatric bottlenose and spotted dolphins is influenced by this mechanism via 2 main features: 1) morphological differences in their feeding apparatus, specifically in the number and size of their teeth (Perrin 1975a, 1975b; Perrin et al. 1987, 2011); and 2) behavioral differences in their method of prey capture, where both delphinids take advantage of the habitat heterogeneity of GD. Spotted dolphins use the deep inner basin, where anoxic conditions concentrate schooling prey in the upper water column, whereas foraging by inshore bottlenose dolphins is focused on mangrove-lined estuaries (Herra-Miranda et al. 2016), where tidal fronts concentrate prey and facilitate capture (Pacheco-Polanco 2015). Dolphin coexistence and patterns of local and regional diversity. Evidence of species-specific associations of dolphins with the structural features of GD has implications for understanding local (fine-scale) and regional patterns of diversity. Pantropical spotted dolphins and inshore bottlenose dolphins appear to have rather wide ecological niches outside GD with ample spatial overlap, in contrast to the pattern inside the gulf, suggesting that more than one ecotype, population, or subspecies may occur in the eastern tropical Pacific. Pantropical spotted dolphins in the inner basin of GD seem to have phenotypic differences with S. a. graffmani frequently encountered outside this gulf, suggesting differentiation at the ecotype level (Oviedo et al. 2015). The population of inshore bottlenose dolphins at GD might be the most abundant along the entire Pacific coast of Costa Rica. Recent documentation of small populations of inshore bottlenose dolphins at Golfo de Nicoya (300 km north of GD) and Chiriquí Gulf in Panama (less than 150 km south of GD), done by the authors of the current study, suggests the possibility of a metapopulation along the Pacific Coast at the Nicoya Ecoregion. The embayment environment of GD could facilitate partial isolation of dolphins living within the gulf from those occurring along the open coast, which together with ecological associations and behavioral factors documented in this study, could result in genetic differentiation of dolphins from conspecifics outside this embayment (Moller et al. 2007). The spatial separation of inshore bottlenose dolphins and pantropical spotted dolphins that follows from their associations with different abiotic variables in GD may best be described as gulf-scale sympatry with fine-scale allopatry, and suggests a mechanism that may facilitate coexistence of these potential competitors. It remains to be determined whether differences in the trophic ecology of these species underlie the spatial patterns observed, and the extent to which adaptation to local conditions could promote distinct resident populations in semi-closed, coastal-marine habitats such as GD. Acknowledgments We are grateful for the fieldwork support of our Captain M. Loaiziga (Taboga), A. Herra-Miranda, and J. Medina from our research base at El Chontal. We are indebted to all our volunteers for their support in the field. This research has been funded by Earthwatch Institute (Chesoning Underwriting, Hill-Urbina Underwriting), International Student Volunteers, the Society for Marine Mammalogy (SMM), and multiple small grants from the Cetacean Society International (CSI). LO was supported by the BEIFI Scholarship Scheme of the Intituto Politecnico Nacional, a OAS-Conacyt-AMEXID 2014 Scholarship, and the Conacyt-National Scheme of Graduate Scholarship. DA-G received a sabbatical grant from the Secretaria de Investigación y Posgrado of Intituto Politecnico Nacional. Literature Cited Acevedo-Gutierrez , A. S. Burkhart . 1998 . Seasonal distribution of bottlenose (Tursiops truncatus) and pan-tropical spotted (Stenella attenuata) dolphins (Cetacea: Delphinidae) in Golfo Dulce, Costa Rica . Revista de Biología Tropical 46 : 91 – 101 . Adler , P. B., A. Fajardo , A. R. Kleinhesselink N. J. Kraft . 2013 . Trait-based tests of coexistence mechanisms . Ecology Letters 16 : 1294 – 1306 . Google Scholar CrossRef Search ADS Bearzi , M . 2005 . Dolphin sympatric ecology . Marine Biology Research 1 : 165 – 175 . Google Scholar CrossRef Search ADS Bearzi , M . 2007 . Dolphins and African apes: comparisons of sympatric socio-ecology . Contributions to Zoology 76 : 235 – 254 . Benson , S. R. , D. A. Croll , B. B. Marinovic , F. P. Chavez J. T. Harvey . 2002 . Changes in the cetacean assemblage of a coastal upwelling ecosystem during El Niño 1997–98 and La Niña 1999 . Progress in Oceanography 54 : 279 – 291 . Google Scholar CrossRef Search ADS Chesson , P . 2000 . Mechanisms of maintenance of species diversity . Annual Review in. Ecology and Systematic 31 : 343 – 366 . Google Scholar CrossRef Search ADS Cubero Pardo , P . 1998a . Distribución y patrones de actividad del bufeo (Tursiops truncatus) y el delfín manchado (Stenella attenuata) en el Golfo Dulce . M.S. thesis , Universidad de Costa Rica , San Jose, Costa Rica . Cubero-Pardo , P . 1998b . Patrones de comportamiento diurnos y estacionales de Tursiops truncatus y Stenella attenuata (Mammalia: Delphinidae) en el Golfo Dulce, Costa Rica . Revista de Biología Tropical 46 : 103 – 110 . Cubero Pardo , P . 2007a . Environmental factors governing the distribution of the bottlenose dolphin (Tursiops truncatus) and the spotted dolphin (Stenella attenuata) in Golfo Dulce, South Pacific, off Costa Rica . Investigaciones Marinas 35 : 15 – 23 . Google Scholar CrossRef Search ADS Cubero Pardo , P . 2007b . Distribución y condiciones ambientales asociadas al comportamiento del delfín bufeo (Tursiops truncatus) y el delfín manchado (Stenella attenuata) (Cetacea: Delphinidae) en el Golfo Dulce, Costa Rica . Revista de Biología Tropical 55 : 549 – 557 . Google Scholar CrossRef Search ADS Day , T . 2000 . Competition and the effect of spatial resource heterogeneity on evolutionary diversification . The American Naturalist 155 : 790 – 803 . Google Scholar CrossRef Search ADS Doksæter , L. , E. Olsen , L. Nøttestad A. Fernö . 2008 . Distribution and feeding ecology of dolphins along the Mid-Atlantic Ridge between Iceland and the Azores . Deep Sea Research Part II: topical Studies in Oceanography 55 : 243 – 253 . Google Scholar CrossRef Search ADS Elith , J. , M. Kearney S. J. Phillips . 2010 . The art of modelling range-shifting species . Methods in Ecology and Evolution 1 : 330 – 342 . Google Scholar CrossRef Search ADS Elith , J. , J. R. Leathwick T. Hastie . 2008 . A working guide to boosted regression trees . Journal of Animal Ecology 77 : 802 – 813 . Google Scholar CrossRef Search ADS Friedlaender , A. S. , D. W. Johnston , W. R. Fraser , J. Burns , P. N. Halpin and D. P. Costa . 2011 . Ecological niche modeling of sympatric krill predators around Marguerite Bay, Western Antarctic Peninsula . Deep Sea Research II 58 : 1729 – 1740 . Google Scholar CrossRef Search ADS Fury , C. A. P. L. Harrison . 2011 . Seasonal variation and tidal influences on estuarine use by bottlenose dolphins (Tursiops aduncus) . Estuarine, Coastal and Shelf Science 93 : 389 – 395 . Google Scholar CrossRef Search ADS Gowans , S. H. Whitehead . 1995 . Distribution and habitat partitioning by small odontocetes in the Gully a submarine canyon in the Scotian Shelf . Canadian Journal of Zoology 73 : 1599 – 1608 . Google Scholar CrossRef Search ADS Gowans , S. , B. Wursig L. Karczmarski . 2008 . The social structure and strategies of delphinids: predictions based on an ecological framework . Advances in Marine Biology 53 : 195 – 294 . Google Scholar CrossRef Search ADS Gross , A. , J. Kiszka , O. Van Canneyt , P. Richard V. Ridoux . 2009 . A preliminary study of habitat and resource partitioning among co-occurring tropical dolphins around Mayotte, southwest Indian Ocean . Estuarine, Coastal and Shelf Science 84 : 367 – 374 . Google Scholar CrossRef Search ADS Heinrich , S. , S. Elwen S. Bräger . 2010 . Patterns of sympatry in Lagenorhynchus and Cephalorhynchus: dolphins in different habitats . Pp. 313 – 332 in The dusky dolphin: master acrobat off different shores . (B. Würsig and M. Würsig , eds.). Elsevier Academic Press , Amsterdam, The Netherlands . Google Scholar CrossRef Search ADS Herra-Miranda , D. , J. D. Pacheco , L. E. O. Correa M. Iñiguez . 2016 . Análisis espacial de los hábitats críticos del delfín nariz de botella (Tursiops truncatus) y la ballena jorobada (Megaptera novaeangliae) en el Golfo Dulce, Costa Rica: consideraciones acerca de un proyecto de construcción de marina . Revista Ciencias Marinas y Costeras 8 : 9 – 27 . Google Scholar CrossRef Search ADS Hosmer , D. W. S. Lemeshow . 1989 . Applied logistic regression . John Wiley and Sons Inc ., New York . James , G. , D. Witten , T. Hastie R. Tibshirani . 2013 . An introduction to statistical learning . Vol. 112 . Springer , New York . Google Scholar CrossRef Search ADS Leathwick , J. R. , J. Elith , M. P. Francis , T. Hastie P. Taylor . 2006 . Variation in demersal fish species richness in the oceans surrounding New Zealand: an analysis using boosted regression trees . Marine Ecology Progress Series 321 : 267 – 281 . Google Scholar CrossRef Search ADS Leimgruber , P. , W. J. McShea M. Songer . 2014 . Vertical habitat segregation as a mechanism for coexistence in sympatric rodents . Mammalian Biology 75 : 313 – 317 . Google Scholar CrossRef Search ADS MacLeod , C. D. , C. R. Weir , C. Pierpoint E. J. Harland . 2007 . The habitat preferences of marine mammals west of Scotland (UK) . Journal of the Marine Biological Association of the United Kingdom 87 : 154 – 164 . MacLeod , C. D. , C. R. Weir , M. B. Santos T. E. Dunn . 2008 . Temperature-based summer habitat partitioning between white-beaked and common dolphins around the United Kingdom and Republic of Ireland . Journal of the Marine Biological Association of the United Kingdom 88 : 1193 – 1198 . Google Scholar CrossRef Search ADS Marra , G. S. N. Wood . 2011 . Practical variable selection for generalized additive models . Computational Statistics and Data Analysis 55 : 2372 – 2387 . Google Scholar CrossRef Search ADS Morris , D. W . 1996 . Coexistence of specialist and generalist rodents via habitat selection . Ecology 77 : 2352 – 2364 . Google Scholar CrossRef Search ADS Oviedo , L . 2007 . Dolphin sympatric ecology in a tropical fjord: habitat partitioning by bathymetry and topography as a strategy to coexist . Journal of Marine Biological Association of the United Kingdom 87 : 1 – 9 . Google Scholar CrossRef Search ADS Oviedo , L . 2008 . Análisis del uso de hábitat del delfín manchado pantropical Stenella attenuata (Cetacea: Delphinidae) en el Golfo Dulce, Costa Rica . M.S. thesis , Universidad Nacional de Costa Rica (UNA) , Puntarenas, Costa Rica . Oviedo , L. , et al. 2015 . Diversidad de cetáceos en el paisaje marino costeros de Golfo Dulce, Península de Osa, Costa Rica . Revista de Biología Tropical 63 : 395 – 406 . Oviedo , L. , J. D. Pacheco-Polanco D. Herra-Miranda . 2009 . Evaluación de los riesgos de afectación por el establecimiento de granjas atuneras en relación con la distribución espacial de cetáceos en el Golfo Dulce, Costa Rica . Revista de Ciencias Marinas y Costera REVMAR 1 : 159 – 174 . Pacheco-Polanco , J. D . 2015 . La población residente de delfines nariz de botella (Tursiops truncatus) ecotipo costero en el Golfo Dulce, Costa Rica: implicaciones para una gestión integrada del recurso . M.S. thesis , Universidad de Costa Rica (UCR) , San José, Costa Rica . Parra , G . 2006 . Resource partitioning in sympatric delphinids: space use and habitat preferences of Australian snubfin and Indo-Pacific humpback dolphins . Journal of Animal Ecology 75 : 862 – 874 . Google Scholar CrossRef Search ADS Parra , G. , P. Corkeron P. Arnold . 2011 . Grouping and fission-fusion dynamic in Australian snubfin and Indo-Pacific humpback dolphins . Animal Behavior 82 : 1423 – 1433 . Google Scholar CrossRef Search ADS Parra , G. , P. J. Corkeron H. Marsh . 2006 . Population size, site fidelity and residency pattern of Australian snubfin and Indo-Pacific humpback dolphins: implication for conservations . Biological Conservation 129 : 167 – 180 . Google Scholar CrossRef Search ADS Perrin , W. F . 1975a . Variation of spotted and spinner porpoise (genus Stenella) in the eastern Pacific and Hawaii . Bulletin of the Scripps Institution of Oceanography 21 :197 pp. Perrin , W. F . 1975b . Distribution and differentiation of populations of dolphins of the genus Stenella in the eastern tropical Pacific . Journal of the Fisheries Board of Canada 32 : 1059 – 1067 . Google Scholar CrossRef Search ADS Perrin , W. F. , et al. 1987 . Revision of the spotted dolphins, Stenella spp . Marine Mammal Science 3 : 99 – 170 . Google Scholar CrossRef Search ADS Perrin , W. F. , J. L. Thieleking , W. A. Walker , F. I. Archer K. M. Robertson . 2011 . Common bottlenose dolphins (Tursiops truncatus) in California waters: cranial differentiation of coastal and offshore ecotypes . Marine Mammal Science 27 : 769 – 792 . Google Scholar CrossRef Search ADS Peterson , A. T . 2011 . Ecological niches and geographic distributions (MPB-49) (No. 49) . Princeton University Press , Princeton, New Jersey . Phillips , S. J. , R. P. Anderson R. E. Schapire . 2006 . Maximum entropy modeling of species geographic distributions . Ecological Modelling 190 : 231 – 259 . Google Scholar CrossRef Search ADS Phillips , S. J. , et al. 2009 . Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data . Ecological Applications 19 : 181 – 197 Google Scholar CrossRef Search ADS Praca , E. A. Gannier . 2008 . Ecological niche of three teuthophagous odontocetes in the northwestern Mediterranean sea . Ocean Science 4 : 49 – 59 Google Scholar CrossRef Search ADS Pusineri , C. , O. Chancollon , J. Ringelstein V. Ridoux . 2008 . Feeding niche segregation among the North-east Atlantic community of oceanic top predators . Marine Ecology Progress Series 361 : 21 – 34 . Google Scholar CrossRef Search ADS QGIS Development Team . 2015 . QGIS Geographic Information System. Ver. 2.14.22. Open Source Geospatial Foundation Project . http://qgis.osgeo.org. Qiao , H. , J. Soberon A. Townsend-Peterson . 2015 . No silver bullets in correlative ecological niche modelling: insights from testing among many potential algorithms for niche estimation . Methods in Ecology and Evolution 6 : 1126 – 1136 . Google Scholar CrossRef Search ADS Quesada-Alpizar , M. A. J. Cortes . 2006 . Los ecosistemas marinos del Pacifico sur de Costa Rica: estado del conocimiento y perspectivas de manejo . Revista de Biología Tropical 54 ( Suppl. 1 ): 101 – 145 . Romero , M. A. , S. L. Dans , G. M. Svendsen , R. González E. A. Crespo . 2012 . Feeding habits of two sympatric dolphin species off North Patagonia, Argentina . Marine Mammal Science 28 : 364 – 377 . Google Scholar CrossRef Search ADS Roughgarden , J . 1976 . Resource partitioning among competing species: a coevolutionary approach . Theoretical Population Biology 9 : 388 – 424 . Google Scholar CrossRef Search ADS Sbrocco , E. J. P. H. Barber . 2013 . MARSPEC: ocean climate layers for marine spatial ecology . Ecology 94 : 979 . Google Scholar CrossRef Search ADS Sikes , R. S. , and The Animal Care and Use Committee of the American Society of Mammalogists . 2016 . 2016 Guidelines of the American Society of Mammalogists for the use of wild mammals in research and education . Journal of Mammalogy 97 : 663 – 688 . Google Scholar CrossRef Search ADS Spalding , M. D. , et al. 2007 . Marine ecoregions of the world: a bioregionalization of coastal and shelf areas . BioScience 57 : 573 – 583 . Google Scholar CrossRef Search ADS Svendsen , G. M. et al. 2015 . Environmental niche overlap between common and dusky dolphins in North Patagonia, Argentina . PLoS One 10 : e0126182 . Google Scholar CrossRef Search ADS Svendsen , H. , R. Rosland , S. Myking , J. A. Vargas , O. G. Lizano E. C. Alfaro . 2006 . A physical oceanographic study of Golfo Dulce, Costa Rica . Revista de Biología Tropical 54 : 147 – 170 . Thorne , L. H. , et al. 2012 . Predictive modeling of spinner dolphin (Stenella longirostris) resting habitat in the main Hawaiian Islands . PLoS One 7 : e43167 . Google Scholar CrossRef Search ADS Vergara , M. , S. A. Cushman , F. Urra A. Ruiz-González . 2016 . Shaken but not stirred: multiscale habitat suitability modeling of sympatric marten species (Martes martes and Martes foina) in the northern Iberian Peninsula . Landscape Ecology 31 : 1241 – 1260 . Google Scholar CrossRef Search ADS Warren , D. L. , R. E. Glor M. Turelli . 2010 . ENMTools: a toolbox for comparative studies of environmental niche models . Ecography 33 : 607 – 611 . Google Scholar CrossRef Search ADS Weir , C. R. , C. D. MacLeod S. V. Calderan . 2009 . Fine-scale habitat use by white-beaked and common dolphins in the Minch (Scotland, UK): evidence for interspecific competition or coexistence ? Journal of the Marine Biological Association of the United Kingdom 89 : 951 – 960 . Google Scholar CrossRef Search ADS Weir , C. R. , C. D. MacLeod G. J. Pierce . 2012 . Habitat preferences and evidence for niche partitioning amongst cetaceans in the waters between Gabon and Angola, eastern tropical Atlantic . Journal of the Marine Biological Association of the United Kingdom 92 : 1735 – 1749 . Google Scholar CrossRef Search ADS Wolff , M. , H. J. Hartmann V. Koch . 1996 . A pilot trophic model for Golfo Dulce, a fjord-like tropical embayment, Costa Rica Modelo piloto trófico para el Golfo Dulce, Costa Rica, un tipo de fiordo tropical . Revista de Biología Tropical 44 : 215 – 231 . © 2018 American Society of Mammalogists, www.mammalogy.org This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)
TI - Habitat partitioning mediates the coexistence of sympatric dolphins in a tropical fjord-like embayment
JF - Journal of Mammalogy
DO - 10.1093/jmammal/gyy021
DA - 2018-05-18
UR - https://www.deepdyve.com/lp/oxford-university-press/habitat-partitioning-mediates-the-coexistence-of-sympatric-dolphins-in-bN2cy9ajMa
SP - 1
EP - 564
VL - Advance Article
IS - 3
DP - DeepDyve
ER -