Spatial distribution and habitat selection of common bottlenose dolphins (Tursiops truncatus) within two estuaries in southern coastal USA Authors: Thibaut Bouveroux, Carl S. Cloyed, Kevin Barry, Keith D. Mullin, Carrie Sinclair, Ruth H. Carmichael Published: 2026 ABSTRACT Coastal waters are increasingly exposed to anthropogenic pressures, resulting in the degradation of critical habitats that shape the distribution of marine predators. This study used photo-identification data to define drivers of common bottlenose dolphin habitat selection in Mobile Bay and Perdido Bay, two urbanized estuaries in southern USA. Dolphins exhibited distinct seasonal habitat selection between estuaries. In Mobile Bay, dolphin occurrence decreased with water temperature and increase with dissolved oxygen during winter, and dolphins were more frequently observed closer to shore, near a key confluence point (Mobile Bay Pass) and in deeper water near ship channels in summer. In Perdido Bay, dolphin occurrence increased with pH and dissolved oxygen in winter, while in summer, dolphin occurrence increased in deeper waters, near ship channels and closer to gillnet exclusion zones. Across both systems, dolphins occurred in moderate-to-high salinity waters but also used very low salinity areas. Our results indicate a seasonal transition from physiochemical drivers in winter to structural and habitat-related factors in summer and suggest that ship channels may function as salinity refuges and prey hotspots, due to water stratification and higher prey availability. This work highlights dolphins’ adjustment to local conditions within dynamic estuarine environments under natural and anthropogenic influences.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-52366-1. Introduction Given the current challenges caused by habitat degradation and environmental changes, understanding how animals use their habitats and identifying the ecological variables that influence their distributions are essential for developing effective conservation strategies and population management1–5. Defining the relationships between ecological variables and species distribution, however, can be challenging, especially in coastal marine ecosystems that are highly dynamic and include many highly mobile predator species. These mobile species, such as marine mammals, can have large home ranges and small sample sizes based on observational data, making accurate spatial prediction difficult. Recently, resource selection functions have become a useful tool to explore the link between environmental predictors and species occurrence, even with limited sample sizes4,6–8. As a result, these predictive statistical models are increasingly used to study distribution and habitat selection by generating relative habitat suitability with high accuracy2,9,10. The model outcomes support conservation and management efforts, for example, by guiding design of protected areas and assessing impacts of long-term environmental variation on marine predator species3,11–14. Common bottlenose dolphins (Tursiops truncatus; hereafter referred to as bottlenose dolphins) exhibit remarkable habitat plasticity, enabling a cosmopolitan distribution15. Dolphins utilize a wide range of habitats that frequently overlap with human activities, exposing them to various anthropogenic pressurese.g.16–19. For example, heavy shipping traffic may force dolphins to change their space use by avoiding high-disturbance areas, increasing their dive-times, modify their movement or adapting their behavioral patterns20,21. In contrast, fishing activities, such as trawling or aquaculture farms, often attract dolphins by enhancing access to prey through discards or fish aggregation associated with fishing gears and depredation22,23, while increasing the risk of entanglement or bycatch24. This adaptability and environmental responsiveness make dolphins an ideal but challenging top marine predator to investigate how both anthropogenic and natural environmental factors affect habitat selection. With the ocean warming, some cetacean species, including bottlenose dolphins, may shift their distribution towards cooler water25. In the southeastern USA, dolphins are becoming vulnerable to environmental change, because factors like water temperature, pH or dissolved oxygen (DO) alter habitat conditions and prey availability18. These changes are often additive, and accumulation of human-induced and natural changes can significantly affect dolphin movements, behaviors, and overall distribution, with potentially profound ecological consequences26,27. As a result, small dolphin populations (i.e., 100–1,000 animals), such as coastal and inshore populations that are characterized by both higher residency patterns and small home ranges18,28, are particularly vulnerable to environmental change. In southern coastal USA waters from Texas to the Florida Keys (hereafter referred to as the Gulf), dolphins are common residents in bays, sounds and estuaries (BSEs). The National Oceanic and Atmospheric Administration (NOAA) has determined that these dolphins are distributed across 32 distinct populations (or stocks)29 within this region. Strandings related to human interactions, including entanglement from fisheries interactions, boat collisions, and direct human-associated feeding that can enhanced the risk of entanglement, boat collisions and ingestion of inappropriate food items, are common in nearshore waters, particularly in areas experiencing significant increases in human activity17,30–33. Dolphin habitats are often environmentally heterogeneous and subject to dynamic conditions. For example, large volumes of freshwater are seasonally discharged into BSE systems in the Gulf region during winter and spring months, and these areas are also regularly affected by extreme weather events, including tropical and non-tropical systems34–36. Such episodic events can substantially alter the habitat suitability of resident bottlenose dolphins37–39. Sudden or prolonged exposure to cold temperature and/or freshwater can negatively affect dolphin health, and in some cases, lead to mortalitye.g.38,40, although dolphins can at least temporarily tolerate a certain level of environmental changes associated with these events41–45. In addition to short-term and seasonal variability, many environmental factors that affect dolphin distribution and habitat selection also vary over longer time periods46. Rising temperatures, variation in salinity associated with freshwater runoff, declining oxygen concentrations, and changes in pH have long-term implications for coastal habitats, concurrently with direct anthropogenic alterations such as ship channel dredging, vessel activity, and point sources of pollutione.g.47,48. These changes are occurring at both local and global scales, with potential for drastic impacts on marine populations and communities, including bottlenose dolphins and other cetaceans49–52. Data to assess these changes at the regional or population levels remain limited for most cetaceans, but are vital for informing effective conservation52. Therefore, to better understand how adaptative responses in dolphin populations are affected by combinations of environmental changes, resource selection studies are needed for habitats where anthropogenic and natural factors vary in known or predictable ways. The objectives of this study are to identify habitat selection of common bottlenose dolphins in adjacent embayments that differ in natural environmental attributes and human activity and presence. We identified key dolphin habitats in both bays using kernel density estimates (KDEs) and used resource selection functions (RSFs) to characterize relationships between dolphin distribution and environmental parameters. In the absence of data on dolphin prey abundance and distribution, we used environmental parameters known to influence prey availability to infer relationships between dolphin distribution and space use related to potential food availability. Comparing across years and seasons between these two embayments allowed us to establish an important baseline for dolphin habitat selection responses to local conditions. The findings offer insights into how natural environmental variables and human-altered habitats affect dolphin habitat use, with local and global implications for conservation and management strategies. Materials and methods Study sites Mobile Bay (MOB) (30°27’30” N, 87°59’44” W; Fig. 1) is a large, semi-open embayment (1,070 km2) that is the sixth largest estuary in the USA, draining the third largest volume of freshwater discharge from watersheds contained in the USA (mean annual discharge = 1,840 m3 s−1), usually peaking in winter and spring37,53,54. MOB is a shallow, micro-tidal estuary with an average water depth of ~ 3 m that is bisected north to south by a ~ 13 m deep and ~ 120 m wide ship channel that connects the Port of Mobile and adjacent industrial canals to the mouth of MOB and Gulf waters55,56. This shipping channel, which currently supports vessels with a capacity of up to 16,000 containers, also connects MOB to the Gulf Intracoastal Waterway (GIWW; 3.7 m deep; Fig. 1) that runs west-east through the southern portion of the bay53,56,57 and supports smaller recreational boats and commercial vessels, such as freight vessels and barges. Seasonal sea surface temperatures range on average between 13 °C in winter and 30 °C in summer, the pH ranges from 5 to 9, and there is a north-south salinity gradient that often range from 0 PSU to 32 PSU58,59. MOB has typically low bottom dissolved oxygen (DO) concentration (~ 4.5 mg/L) and can experience hypoxic conditions, with oxygen concentrations below 0.5 mg/L, especially in summer, deep channels, and low-wind conditions that can enhance water stratification55,56,60. Cargo transport, ship building, and commercial fishing are among the major industries in MOB. Local commercial and recreational fisheries target a variety of seafood year-round, including brown (Farfantepenaeus aztecus) and white shrimps (Litopenaeus setiferus), oysters (Crassostrea virginica), red snapper (Lutjanus campechanus), Spanish mackerel (Scomberomorus maculatus), fringed flounder (Etropus crossotus), menhaden (Brevoortia patronus) and mullet (Mugil cephalus), with many fisheries including the commercial shrimp fishery most active in summer61–63. Perdido Bay (PER) (30°18’37.1” N, 87°30’25.0” W; Fig. 1) is a small-enclosed bay (130 km2) located on the Alabama-Florida border with an average water depth of ~ 2.6 m64,65. PER primarily receives freshwater inflow from the Perdido River in the north, at a much lower flow than MOB (mean daily discharge = 62 m3 s−1), resulting in a fundamentally different type of embayment. A series of several small lagoons and bayous connect PER to the Gulf in the south through the narrow channel of Perdido Pass, and to Pensacola Bay, FL on the east via Big Lagoon. The only major boating channel in PER is the GIWW (Fig. 1), which runs west-east in the southernmost portion of the bay and connects PER to MOB through a 15 km long and narrow (< 100 m wide) channel. Seasonal sea surface temperatures range on average between 15 °C in winter and 30 °C in summer59. There is a north-south salinity gradient, usually ranging from 8 PSU to > 30 PSU, but can drop < 5 PSU near Perdido River. DO concentration is usually higher at the surface in northern PER compared to the southern PER Pass, due to the oxygenated water from the freshwater discharge, but can nevertheless experience hypoxic conditions at the bottom65. PER is characterized by an urbanization gradient, where the northern areas near the Perdido River are largely vegetated and agricultural, while residential development increases down bay, with the southern portion heavily developed, primarily for tourism66. Human activities are primarily residential, recreational, and tourism-related, with smaller-scale commercial and industrial activities compared to MOB. Military activities, particularly at Pensacola Naval Air Station, also take place in the southern part of the PER watershed along Big Lagoon. In this embayment, fisheries are primarily recreational, mainly targeting brown and white shrimp, with tighter harvest regulations in Florida waters (e.g., daily catch limit per vessels and specific gear restrictions under the Florida Fish and Wildlife Conservation Commission) compared to adjacent Alabama waters. Fig. 1Transect (solid) and contour (dashed) lines surveyed for bottlenose dolphins (Tursiops truncatus) between January 2020 and June 2022 in Mobile Bay and Perdido Bay in Alabama, USA. TX = Texas, LA = Louisiana, MS = Mississippi, AL = Alabama, GA = Georgia, FL = Florida. Transect (solid) and contour (dashed) lines surveyed for bottlenose dolphins (Tursiops truncatus) between January 2020 and June 2022 in Mobile Bay and Perdido Bay in Alabama, USA. TX = Texas, LA = Louisiana, MS = Mississippi, AL = Alabama, GA = Georgia, FL = Florida. Field sampling Dedicated boat-based surveys following systematic transect and contour lines were conducted in winter (December - February) and summer (June - July) from January 2020 to June 2022 in the MOB and PER estuaries67 (Fig. 1). To maximize the probability of dolphin detection, surveys were restricted to favorable weather conditions that included wind speeds ≤ Beaufort Sea State 4, clear visibility (no fog or haze), and a swell height of ≤ 1 m. All transect and contour lines were surveyed three times every season at a consistent speed of ~ 14–16 knots, using two outboard engine boats (6 m and 7 m length) in MOB and one 6 m boat in PER. Transects and contour lines took about three to six days of survey effort in MOB and two to three days in PER to be completed. Each vessel typically had three crew members onboard (with a maximum of five for training purposes), including a captain, a photographer, and a data recorder. All crew members visually searched for dolphins and scanned the water 180° ahead of the research vessel. Perimeter contour lines in each embayment ranged approximately 0.2 to 2.0 km from shore, following the 2 m water depth bathymetry contour (Fig. 1). Data collection A sighting event began when one or more dolphins were observed67. The surveyed transect or contour line was paused, and animals were slowly approached to minimize potential disturbance from the research vessel. A group of dolphins was defined as several dolphins swimming within 10 m of any other individuals (i.e., 10 m “chain” rule), moving in the same direction, and engaged in similar behavioral activities68,69. Photos of dorsal fins of every dolphin in the group were taken for individual identification70, using a Canon EOS 7D digital camera equipped with a 100–400 mm telephoto zoom lens (Canon USA, Inc.). For each sighting, the date and time, the geographical position using a handheld GPS Garmin GPSMAP 76, and the group size (minimum, maximum and best) and composition (number of adults, calves and neonates) were recorded. The minimum, maximum and best number of animals in groups were estimated as a consensus among the crew members aboard the vessel. Estimated group sizes were revised after photo analysis if field-based estimates were lower than found after the identification of individuals from all dorsal fins collected during sightings67. At the end of each sighting, the sea surface salinity, water temperature (℃) and DO (mg/L) were collected using a handheld data sonde YSI® Professional Series Pro30, and water depth (m) was recorded using a depth sounder. After all data were collected, the survey was resumed at the location where the vessel initially departed from the transect or contour line. Additionally, environmental data were continuously collected onboard from summer 2020 through summer 2022 to characterize the sea surface profile of the bay. At least one boat was equipped with a Dataflow data logging system, equipped with a Manta multiprobe sonde (©Eureka Water Probes, Austin Texas) to collect salinity, water temperature (℃), DO (mg/l) and pH at the water surface (~ 50 cm depth) while underway during surveys and sightings. To obtain environmental data of homogenous resolution throughout the study areas, the Dataflow system was programmed to collect data every 5 s while cruising on the transect at higher speeds, and every 30 s during focal observations of dolphins at lower speeds. During the first season of sampling (winter 2019-20), prior to the acquisition of the Dataflow system, salinity and water temperature were recorded manually using a handheld data sonde (YSI® Professional Series Pro30) at gridded waypoints located at 9 km intervals in MOB and 7 km in PER along the transect and contour lines. Data analyses Before using explanatory variables in subsequent analyses, collinearity was tested using Pearson correlation tests. Depth and distance to shore were not found to be collinear, as both bays had relatively homogeneous depths overall, while the deep ship channels were situated relatively close to shore. Therefore, both variables were retained in the tested models. Differences in salinity between embayments were assessed using Welch’s t-tests to account for unequal variances, as indicated by Levene’s test. We finally tested the potential effect of the crew size onboard survey vessels on the probability of detecting dolphins using a linear regression model. Subsequent steps were calculated in the Universal Transverse Mercator (UTM) projection and WGS 1984 datum. The closest distance between each dolphin sighting and the distances to key features: shore, ship channels (including the MOB ship channel and secondary navigable waterways), reefs, fishing areas, shrimping restriction zones and gillnet exclusion zones, as well as the distance to Mobile Bay Pass (for MOB only) were calculated using the Simple Features71 and gdistance72 R packages. (Fig. 2). All ship channels were kept in the tested models because distance to the MOB ship channel alone (removing secondary navigable waterways from the shapefile) made no difference in the results generated by RSF models. Sighting locations of dolphins were also used to investigate the presence of dolphin hotspots using kernel density estimate (KDE) analyses in the ArcMap 10.8 (©ArcGis, Esri 2020) spatial analyst toolbox and with group sizes as a weighted factor. To allow for sufficient resolution in respective embayments, the output cell size was set to 4000 × 4000 m in MOB and 2000 × 2000 m in PER. Finally, to characterize the dolphin habitat from sampled environmental parameters (salinity, water temperature, DO, pH) collected by the Dataflow, we used the inverse distance weighted (IDW) interpolation analyses for in ArcMap to generate rasters of predicting values. Fig. 2Seasonal sighting distribution of common bottlenose dolphins in Mobile Bay (A-B) and Perdido Bay (C-D), related to key habitat features. Panels (A-C) are for the winter and (B-D) are for the summer. Seasonal sighting distribution of common bottlenose dolphins in Mobile Bay (A-B) and Perdido Bay (C-D), related to key habitat features. Panels (A-C) are for the winter and (B-D) are for the summer. To quantify habitat selection of dolphins, we used RSFs derived by a presence-availability framework73. To do so, we first generated binary dependent data using “presence” locations (areas where dolphins were encountered) and random “available” locations (areas that dolphins could have used but were never documented during our surveys). For each season, “available” locations were randomly generated at a ratio of 10 to 1 relative to “presence” locations74. To investigate the relationships between dolphin distribution and environmental parameters, we extracted environmental data from generated rasters, including depth, on both “presence” and “available” locations, and we extracted key distances to produce RSF predictive maps. We used several different models in our RSF analyses, that included general linear models (GLMs), general additive models (GAMs), random forest models (RFs), boosted regression tree models (BRTs) and maximum entropy models (MaxEnt). These models represent traditional (GLM and GAM) and machine learning (RF, BRT, MaxEnt) techniques that associate the above environmental variables with “presence” and “available” locations of dolphins. GLMs provide linear relationships among variables that can be helpful to understand how much one variable changes in response to another75. GAMs and machine learning models, however, can better explain complex, often non-linear ecological relationships, but they may be difficult to interpret76. Partial dependence plots can be used to help understand these complex relationships, but they do not quantify change77. Therefore, comparing multiple models provides a more robust way to understand and define habitat selection. The predictive performance of each model was first assessed using the areas under the curve (AUC) for the receiver operating characteristic (ROC) and precision-recall (PR) for best fitting models selection8,78. Then, the supporting model was chosen through seasonal cross-validation. AUC ROC provides a threshold-independent metric that calculates the overall accuracy, ranging between 0 and 1. Models with AUC ROC ranging between 0.85 and 0.9 were considered to have useful predictive capacity, while models with AUC values > 0.9 were considered to have high predictive capacity78,79. All statistical analyses were conducted using the statistical software program R (version 4.5.0)80. Results From January 2020 to June 2022, a total of 85 survey days were conducted in MOB (40 surveys in winter and 45 surveys in summer) and 38 in PER (19 surveys in each season), for a total boat survey effort of 6,599 km and 2,412 km, respectively (detailed survey results were previously reported67). During surveys, between 3 and 5 persons were present onboard the survey vessels, and the size of the crew had no effect on the probability of detecting dolphins (F2, 116 = 0.50, p = 0.61). In MOB, except in winter 2019/20 (n = 26), the number of sightings was higher in winter (2020/21 n = 70; 2021/22 n = 74) compared to summer (2020 n = 37; 2021 n = 41; 2022 n = 61). In contrast, the number of dolphins encountered was higher in summer, with highest numbers recorded in summer 2022 (n = 618) and lowest in winter 2019/20 (n = 148), compared to other sampling periods (winter 2020/21 n = 489; winter 2021/22 n = 523; summer 2020 n = 346; summer 2021 n = 369). In PER, no clear seasonal differences were observed in the number of sightings and dolphins encountered, with ~ 150 sightings recorded overall (47–56 sightings per season) and ~ 760 dolphins encountered in total across seasons (211 and 267 dolphins per season), except in summer 2020 and 2022, which had respectively the lowest and highest number of dolphins encountered (n = 185 and n = 296). During the study period, we continuously collected 127,126 environmental data points in MOB and 88,760 in PER, which included salinity, pH, DO and water temperature values (Figs. 3a & b). In both study areas, salinity was generally low throughout the study period, with an average of 10.3 PSU ± 6.0 in MOB and 15.9 PSU ± 7.4 in PER. Overall, the salinity profile in MOB was lower than in PER (Welch’s t-tests: p < 0.001 for all seasons) and showed variability among study periods (Fig. 3a; Table S1). On average, the salinity was lower during summer (9.3 PSU ± 5.3, median 8.5, range 0.06–31.4) compared to winter (13.5 PSU ± 5.1, median 13.8, range 0.1–31.0; One-way ANOVA: df = 1, F = 12421, p < 0.001). In contrast, salinity in PER was more stable among study periods compared to MOB (Table S1), and was higher in winter compared to summer (winter: 17.5 PSU ± 6.0, median 17.6, range 0.1–30.3; summer: 15.2 PSU ± 7.9, median 15.7, range 0.05–34.1; One-way ANOVA: df = 1, F = 1948, p < 0.001), except in the summer of 2022, when salinity across almost the entire bay dropped below 11 PSU (Fig. 3b; Table S1). Spatial distribution of dolphins and kernel density estimates In both embayments, dolphins were more frequently observed at the higher salinities (> 14 PSU; MOB: X2 = 9.09, df = 1, p = 0.003; PER: X2 = 134.22, df = 1, p < 0.001). However, dolphins were also frequently found at lower salinities (< 10 PSU; 41% of sightings, n = 128 in MOB; 17% of sightings, n = 50 in PER), mostly in the northern part of MOB and the northeast of PER, near riverine sources of freshwater into the bays (Figs. 3a & b). Fig. 3Locations of common bottlenose dolphins in (A) Mobile Bay and (B) Perdido Bay (black points) and the salinity profiles collected by the Dataflow data logging system during surveys. Maps were generated using the interpolation tool in QGIS processing toolbox. Locations of common bottlenose dolphins in (A) Mobile Bay and (B) Perdido Bay (black points) and the salinity profiles collected by the Dataflow data logging system during surveys. Maps were generated using the interpolation tool in QGIS processing toolbox. KDEs were higher in summer than in winter in both embayments (Figs. 4a-d). In MOB, the highest KDEs of dolphins ranged between 7 and 11 dolphins/km² in summer and between 5 and 8 dolphins/km² in winter and were observed in the southern part of the bay, where it connects to open oceanic water and salinity levels were the highest (Figs. 2a and 4a and b). High KDEs of dolphins were also found in the northwestern part of the bay, along ship channels. In PER, dolphin KDEs were twice as high in summer compared to winter, with maximum KDEs ranging from 24 to 36 dolphins/km² in summer and from 12 to 19 dolphins/km² in winter (Figs. 4c & d). In this embayment, the highest KDEs were recorded in two areas of high salinity levels; the northwestern part of PER and the eastern part of Big Lagoon, where it also connects to open oceanic water (Fig. 3b). Fig. 4Kernel density estimates (expected number of individuals per square km) for common bottlenose dolphins in (A-B) Mobile Bay (radius = 4 km) and (C-D) Perdido Bay (radius = 2 km) from January 2020 to June 2022. A-C = winter, B-D = summer. Kernel density estimates (expected number of individuals per square km) for common bottlenose dolphins in (A-B) Mobile Bay (radius = 4 km) and (C-D) Perdido Bay (radius = 2 km) from January 2020 to June 2022. A-C = winter, B-D = summer. Habitat selection MaxEnt was found to be the best fitting and more robust model to the data collected (highest cross-validated AUC ROC and PR values compared to other RSF models, Table S2 and Table S3) and was, therefore, used to infer dolphin habitat selection and estimate likelihood of dolphin presence for each survey season and each embayment. Model selection was based not only on overall performance of the model but also consistency across seasonal cross-validations. In both embayments, the AUC ROC values exceeded 0.85 for both seasons, indicating robust models for inferring dolphin habitat selection. MOB had AUC ROC values of 0.980 in winter and 0.912 in summer, and AUC ROC values of 0.922 in winter and 0.857 in summer in PER (Table S2). Although, AUC ROC values for the random forest models were > 0.999 for both seasons in both embayments (Table S2), it is likely that these models overfitted the data, which is a common problem when random forest models are applied to small datasets. In our case, the sample size (∼ 50 sightings per season) was too small to split into 80% training data and 20% test data to accurately and independently cross-validate the results among seasons, which makes random forest results less reliable81. In contrast, MaxEnt provided strong predictive performance, while maintaining greater stability across seasonal validations, supporting its selection as the final modelling framework (Table S3). In MOB, the strongest MaxEnt predictors of the resource selection function (proportional variable importance ≥ 0.1) were, in order: water temperature, salinity, and DO in winter; and distances to shore, ship channels, Mobile Bay Pass (MOB Pass), and water depth in summer (Table 1; Fig. 5). In PER, the most important predictors were pH, salinity, and DO in winter; and distances to reefs, ship channels, salinity, depth, and distance to gillnet exclusion zones in summer (Table 1; Fig. 6). In winter, dolphin sightings in MOB were more likely to occur in water temperatures ranging 12–14 °C, with occurrence decreasing above 10 °C and increasing again at around 20 °C. Sightings also increased in waters with higher DO concentrations, beginning around 9 mg/L and peaking at approximately 11 mg/L (Fig. 5a). In summer, dolphin had higher occurrences closer to shore in MOB (< 5,000 m), near ship channels, around Mobile Bay Pass, and at greater water depth (Fig. 5b). In PER during winter, the likelihood of encountering dolphins increased in waters with pH > 6.5, salinity < 10 PSU or > 20 PSU, and DO > 9.0 mg/L (Fig. 6a). In summer, dolphins were more likely to be observed at distances of approximately 10–20 km from reefs, near ship channels, in either very low salinity < 5 PSU or high salinity > 30 PSU waters, at depths > 5 m, and near gillnet exclusion zones (Fig. 6b). Table 1Absolute and proportional variable importance weights derived from MaxEnt models. The most important variables predicting dolphin habitat selection are highlighted in bold.Variable weightsAbsoluteProportionalWinterSummerWinterSummer MOB Distance to shore0.016 0.186 0.036 0.286 Distance to ship channel0.019 0.068 0.042 0.105 Distance to MOB Pass0.008 0.066 0.018 0.101 Depth0.002 0.065 0.004 0.100 Salinity 0.095 0.049 0.217 0.075Distance to fishing areas0.0050.0400.0110.061DO 0.047 0.037 0.109 0.057Distance to gillnetexclusion zones0.0020.0350.0040.054Distance to reef0.0010.0340.0020.052Distance to shrimping exclusion zones0.0200.0320.0450.049Temperature 0.211 0.027 0.483 0.041pH0.0140.0110.0310.017 PER Distance to reef0.009 0.135 0.011 0.182 Distance to ship channel0.060 0.096 0.071 0.129 Salinity 0.196 0.094 0.230 0.127 Depth0.016 0.088 0.019 0.118 Distance to gillnetexclusion zones0.016 0.080 0.019 0.108 pH 0.213 0.071 0.251 0.095Distance to shrimping exclusion zones0.0620.0700.0620.094Temperature0.0800.0520.0940.071Distance to shore0.0250.0330.0300.044DO 0.173 0.023 0.203 0.031 Table 1Absolute and proportional variable importance weights derived from MaxEnt models. The most important variables predicting dolphin habitat selection are highlighted in bold.Variable weightsAbsoluteProportionalWinterSummerWinterSummer MOB Distance to shore0.016 0.186 0.036 0.286 Distance to ship channel0.019 0.068 0.042 0.105 Distance to MOB Pass0.008 0.066 0.018 0.101 Depth0.002 0.065 0.004 0.100 Salinity 0.095 0.049 0.217 0.075Distance to fishing areas0.0050.0400.0110.061DO 0.047 0.037 0.109 0.057Distance to gillnetexclusion zones0.0020.0350.0040.054Distance to reef0.0010.0340.0020.052Distance to shrimping exclusion zones0.0200.0320.0450.049Temperature 0.211 0.027 0.483 0.041pH0.0140.0110.0310.017 PER Distance to reef0.009 0.135 0.011 0.182 Distance to ship channel0.060 0.096 0.071 0.129 Salinity 0.196 0.094 0.230 0.127 Depth0.016 0.088 0.019 0.118 Distance to gillnetexclusion zones0.016 0.080 0.019 0.108 pH 0.213 0.071 0.251 0.095Distance to shrimping exclusion zones0.0620.0700.0620.094Temperature0.0800.0520.0940.071Distance to shore0.0250.0330.0300.044DO 0.173 0.023 0.203 0.031 Absolute and proportional variable importance weights derived from MaxEnt models. The most important variables predicting dolphin habitat selection are highlighted in bold. Distance to gillnet exclusion zones Distance to gillnet exclusion zones Fig. 5Partial dependence plots for MaxEnt modeling of common bottlenose dolphin habitat use in Mobile Bay during (A) winter and (B) summer. DO = dissolved oxygen, Temp. = water temperature, D. shore = distance to shore, D. Pass = distance to Mobile Bay Pass, D. shrimp = distance to shrimping exclusion zones, D. gillnet = distance to gillnet exclusion zones, D. ship channel = distance to ship channel, D. reef = distance to reef, D. fishing = distance to fishing areas. Partial dependence plots for MaxEnt modeling of common bottlenose dolphin habitat use in Mobile Bay during (A) winter and (B) summer. DO = dissolved oxygen, Temp. = water temperature, D. shore = distance to shore, D. Pass = distance to Mobile Bay Pass, D. shrimp = distance to shrimping exclusion zones, D. gillnet = distance to gillnet exclusion zones, D. ship channel = distance to ship channel, D. reef = distance to reef, D. fishing = distance to fishing areas. Fig. 6Partial dependence plots for MaxEnt modeling of common bottlenose dolphin habitat use in Perdido Bay during (A) winter and (B) summer. DO = dissolved oxygen, Temp. = water temperature, D. shore = distance to shore, D. shrimp = distance to shrimping exclusion zones, D. gillnet = distance to gillnet exclusion zones, D. ship channel = distance to ship channel, D. reef = distance to reef. Partial dependence plots for MaxEnt modeling of common bottlenose dolphin habitat use in Perdido Bay during (A) winter and (B) summer. DO = dissolved oxygen, Temp. = water temperature, D. shore = distance to shore, D. shrimp = distance to shrimping exclusion zones, D. gillnet = distance to gillnet exclusion zones, D. ship channel = distance to ship channel, D. reef = distance to reef. Discussion We identified important habitats for the common bottlenose dolphin in Alabama waters and determined the ecological drivers influencing habitat selection. Our findings revealed that dolphins in each embayment exhibited distinct habitat selection, with a seasonal shift from primarily natural abiotic environmental drivers in winter to a combination of structural and anthropogenic factors in summer. In both systems, oceanographic parameters such as DO, salinity, and either water temperature or pH strongly influenced habitat selection during winter, while distance to a ship channel emerged as a key predictor in summer. In the larger, open, and freshwater-influenced MOB, dolphin distribution in summer was additionally driven by proximity to the shore and the Mobile Bay Pass. In contrast, in the narrower and more semi-enclosed PER, summer habitat selection was strongly associated with distance to a reef and salinity, as well as depth and distance to gillnet exclusion zones. Several mechanisms could explain the seasonal patterns observed in dolphin habitat selection in MOB and PER. In this region, freshwater discharge from the watersheds typically peaks in winter37,53,54, resulting in greater fluctuations in salinity, DO, and pH as colder water temperatures affect water-column mixing and oxygen solubility. Water temperature and distance to shore, in particular, are frequently identified as key predictors of dolphin distribution and occurrence in coastal systems82. In addition, reduced human activity during winter, such as lower levels of recreational and fishing boats, may relatively accentuate the role of natural environmental drivers. In contrast, in summer, abiotic variation (e.g., salinity, pH, DO) is generally less pronounced as warmer water leads to stronger stratification and higher overall ecosystem stability, except during extreme weather events such as tropical storms55,83. Seasonal oxygen depletion in deep channels may also create favorable foraging conditions for dolphins by aggregating their prey in upper layers of higher DO concentration waters84 or impairing prey escape under these hypoxic conditions85. In MOB, the shrimp fishery is one of the most important commercial fisheries and is most active in summer61,62. This increased shrimping activity and boating traffic during summer, particularly in ship channels and nearshore zones, may further influence dolphin distribution through attraction to fishery discards, depredation in nets and/or avoidance of disturbance86,87. This pattern suggests a seasonal relationship with increased prey availability and greater movement of dolphins into MOB from adjacent coastal waters that highlight the importance of the seasonal interaction of natural and anthropogenic factors in shaping dolphin habitat usee.g.38,88,89. Although salinity affected dolphin distribution in our study, it was less important than expected. Prior studies conducted in shallow estuaries and in the absence of deep shipping channels, such as in Barataria Bay (Louisiana), Mississippi Sound (Mississippi) and Indian River Lagoon (Florida), have found highest dolphin densities in areas of consistently high salinity82,90–92. Additionally, numerous studies have shown negative health effects of prolonged low-salinity exposure in dolphins40–42,93, which could lead to avoidance of low salinity waters. Despite the notable salinity gradients and general preference of dolphins for higher salinity waters in both areas in this study, salinity did not fully explain dolphin distribution and habitat selection. The deep ship channel and proximity to the estuary mouth were key factors influencing dolphin distribution within the most freshwater-influenced estuary in summer (MOB). Of note, high densities of dolphins were found during summer and in northern MOB despite lower salinities. Salinity had a more pronounced influence on dolphin habitat selection in the shallower estuary of PER across seasons. However, dolphins were still regularly observed in the northern portion of PER in close proximity to a freshwater source, despite lower salinity levels. Broad-scale habitat suitability studies have shown predators may tolerate less suitable habitats to access areas of higher food resource availability, resulting in a disconnect between predator distribution and abiotic environmental factors alone94. Because salinity data in this study were collected near the surface (~ 50 cm depth), however, they may not represent habitat conditions below the surface where dolphins spend time, raising questions about potential links among salinity, food availability, and physical features, particularly depth variation, of each embayment. Deep ship channels and proximity to the mouth of each estuary have relationships to salinity and movement that have potential to affect dolphin behavior and distributions. The Mobile Bay ship channel is characterized by pronounced vertical salinity stratification83,95,96. Rather than avoiding areas with low-salinity surface waters, dolphins may select deeper, stratified waters as a higher-salinity refuge while foraging or traveling39,97,98. For example, following the landfall of Hurricane Harvey in Texas during August 2017, Galveston Bay was flushed with freshwater34, and dolphins remained in the bay but occurred more frequently near deeper channels39. In MOB, there is a general trend toward stronger water stratification during the summer months, with as much as a 30 PSU difference between surface and bottom salinity, depending on local discharge and wind conditions83,96. This stratification could support the use of deeper areas of the bay, and particularly the ship channel, as a refuge for local dolphins during low salinity periods. This notion requires additional targeted study for verification. Additionally, dolphins can utilize ship channels, deeper passes, and embayment mouths as energetically advantageous travel corridors, facilitating movement among habitats. In Texas, dredged channels connecting bays to the Gulf may serve as pathways for dolphins87. Dolphins in ship channels often engage in bow-riding, using the pressure waves of large vessels to conserve energy while traveling at higher speeds99–101. In estuaries with low-salinity surface waters, this behavior may also reduce the time dolphins spend in less favorable habitat conditions. These observations suggest that dolphins in MOB and PER have potential to adjust their space-use patterns to mitigate physiological stress and take advantage of movement benefits associated with location-specific use of deeper waters. Deep channels, passes, and reef structures can further affect dolphin distributions by aggregating prey and providing areas of dense and diverse dietary resources. For example, fish often select deep channels, where at the bottom salinity and temperature conditions are more stable, whereas jetties or artificial reefs can offer fish food supplies and shelter areas from predators102. Habitat use and distribution of marine predators, including dolphins, are often influenced by food availability103–105. Deep ship channels with steep sides may aggregate prey fish by providing refuge and attracting dolphins while they forage106. Passes between inshore bays and open waters can act as natural bottlenecks for migrating fish and can concentrate prey for dolphinse.g.107,108. Accordingly, several studies conducted in bays and estuaries have identified areas within or near dredged ship channels or passes as foraging hotspots, including in and around MOB28,87,109–111. Reef structures are known for increasing overall species richness and enhancing biomass of prey fish112. Prior study investigating dolphin diet in Alabama waters, revealed that dolphins predominately fed on Perciform fish, cephalopods, and decapods28, and many of these prey; including brown shrimp, blue crabs (Callinectes sapidus), and sand seatrout (Cynoscion arenarius), have been associated with low salinity habitats63. Dolphin hotspots identified in our study overlapped with these key structural features, highlighting their importance in shaping dolphin distribution and the potential links to prey availability. This study addresses the gaps in understanding dolphin distribution and habitat selection at a local scale, offering valuable insights into how both natural and human-influenced environmental factors affect dolphin space use. We delineated hotspots of dolphin presence that may serve as critical habitats within two historically understudied, urbanized estuaries in southern coastal USA waters. Our findings underscore the significant roles of multiple environmental factors in driving dolphin distribution, particularly their proximity to freshwater sources, food resources, and salinity refuge in deeper water areas, whether natural or the result of anthropogenic alterations. Our results suggest that surface environmental data alone may not fully capture the factors influencing dolphin distribution, and studies that include environmental dynamics at different depths would provide additional insights into dolphin distribution and habitat selection. In such dynamic estuaries, local environmental conditions can also change rapidly on a scale of days to weeks60,96, making it difficult to confidently link environmental attributes to dolphin distribution. For this study, surveys typically took 1–3 weeks to complete during two discrete periods annually, providing a seasonal snapshot of habitat selection. Year-round and longer-term boat-based monitoring, combined with targeted surveys conducted at finer temporal scales, will help to more accurately capture how environmental and anthropological parameters concurrently influence dolphin occurrence and space use. Further monitoring is essential to assess how natural and anthropogenic factors interact to affect the occurrence and space use of dolphins and other marine biota. Increasing urbanization, ongoing fishing pressures, rapid growth of commercial shipping activities, and plans to widen and deepen shipping channels globally (including in MOB), demand that the planning, expansion and maintenance of these anthropogenic activities be carefully evaluated to determine the potential impacts on dolphin and their prey populations and habitats. In addition to local anthropogenic pressures, ongoing and long-term environmental changes that alter habitat suitability will further influence dolphin distribution. Characterizing these factors that shape dolphin space use patterns is essential for understanding how species will respond to the net effects of rapidly changing environmental conditions. Further integrating distribution and habitat-use analyses with improved vertical water-column measurements and assessments of prey abundance and quality will be particularly important for distinguishing direct abiotic drivers (e.g., salinity) from indirect biotic influences (e.g., prey availability), particularly in areas that do not exhibit optimal oceanographic conditions. Such integrative approaches will support more effective, ecosystem-based management of altered estuarine systems and improve the capacity of resource managers to anticipate the combined effects of coastal development, freshwater input, and climate change on dolphin populations. Our study, conducted in a region affected by a range of natural and anthropogenic factors common to coastal estuaries, provides an important case study for understanding how such alterations may affect marine mammals globally and factors for consideration in developing effective conservation plans and management strategies for bottlenose dolphinse.g.113. Electronic Supplementary Material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1