The role of species distribution models and multiple data sources in implementing dynamic ocean management Authors: Emily C. Nazario, Helen Bailey, Rachel Rhodes, Elliott L. Hazen, Steven J. Bograd, Austin Sell, Douglas J. McCauley, Mark F. Baumgartner, Ana Širović, Briana Abrahms, Rachel Seary Published: 2026 ABSTRACT Content of image described in text. Impact statement Dynamic ocean management, management measures that rapidly change in space and time in response to near real-time data, is increasingly being adopted as an approach to balancing marine life protection and human activities in marine environments. This approach requires reliable near real-time biological, oceanographic or social data, which are costly to attain and maintain. Species distribution model (SDM) outputs have been utilized as a complementary source of information to predict species occurrence. Despite previous work indicating that distribution models can be used to help inform dynamic ocean management frameworks, little work has been done to assess the accuracy and reliability of operational modeled outputs compared to in situ surveillance data. In this article, we evaluate the value of SDMs alongside multiple data sources indicating species presence (sightings and passive acoustic monitoring). We conclude that, when available, assessing species presence according to multiple sources of inference will be central to developing robust predictions to support dynamic ocean management. Introduction The demand to balance the sometimes conflicting objectives of marine life protection and human uses has led to the development of dynamic ocean management (DOM) measures that rapidly change in space and time through the integration of near real-time biological, oceanographic or social data (Maxwell et al., 2015). Growing availability and access to near real-time data, particularly from remotely sensed environmental observations, combined with advanced computing power, has made this novel approach possible. Species distribution models (SDMs) based on species–environment relationships, are being increasingly operationalized with near real-time environmental data for the purpose of informing DOM to address a range of conservation threats, including marine megafauna bycatch (Howell et al., 2008; Hazen et al., 2018), entanglements (Samhouri et al., 2021) and vessel collisions (Hazen et al., 2017; Ham et al., 2021; Hausner et al., 2021). There remain some reservations, however, concerning the appropriate use of SDMs in guiding conservation or management decisions designed to mitigate human–wildlife conflict (Guisan et al., 2013; Robinson et al., 2017). In the California Current System (CCS), vessel collisions remain a leading factor limiting the recovery of endangered Northeast Pacific blue whales (Balaenoptera musculus) to pre-whaling-era numbers (NOAA 2023 Stock assessment). An SDM using near real-time environmental data to predict where whales were likely located, WhaleWatch, was developed to mitigate whale-vessel collision risk (Hazen et al., 2017). An updated higher resolution version, WhaleWatch 2.0 (Abrahms et al., 2019), has become one of three data sources, alongside sightings and passive acoustic monitoring (PAM) detections, supporting the Whale Safe initiative (https://whalesafe.com/). This initiative supports the NOAA-led vessel speed reduction program in California by sharing whale data with shipping companies and monitoring their cooperation with speed recommendations during the months of peak endangered whale abundance. It has successfully helped improve voluntary vessel speed reduction cooperation (https://whalesafe.com/). However, few areas and species are as data-rich; thus, understanding the reliability of operational SDM outputs as stand-alone metrics of species presence in relation to in situ data sources is an important step in enabling possible future expansion of the number of species and systems that could support DOM frameworks (e.g., cryptic species where sightings are rare, cost constraints). Here we use Whale Safe, a tool designed to prevent lethal and sublethal ship strikes with whales, and the endangered blue whale as a case study to demonstrate the contribution of an SDM in the context of other available whale presence data. Our objectives were to (1) evaluate the contribution of the SDM relative to concurrent detections of blue whale presence, and (2) to compare the WhaleWatch 2.0 (blue whale SDM, hereafter) predictions with sightings and passive acoustic detections of blue whales. From this analysis, we show the correlation between the model predictions and whale detections. The value added by near real-time model predictions depends on whether species-specific in situ data are consistently available, meaning accessible without interruption (e.g., without technological failures or seasonal limitations). When in situ data are opportunistic or sporadic, model predictions provide essential insights into species presence that may otherwise be unavailable (i.e., in situ data gaps). Additionally, simultaneous contributions from multiple data sources provide inference for different components of whale presence (sightings: where/when they surface, acoustics: where/when they call and SDMs: where/when there is suitable habitat), and also have distinct but complementary limitations. For example, sightings and passive acoustic detections are prone to false negatives (whales present but not detected), while SDM predictions have uncertainties but are optimized to balance false positives and negatives. Such co-occurring sources of insight (hereafter, multiple data sources) allow for a better understanding of whale presence, as well as uninterrupted data delivery even during failure of any single data stream, supporting the long-term success of DOM. Methods We used blue whale presence from each of Whale Safe’s three data sources (sightings, acoustics and SDM predictions), which were aggregated, processed and provided to us by Whale Safe. These data sources were analyzed for two regions in the CCS where Whale Safe operates: (1) the San Francisco Vessel Speed Reduction (VSR) Zone, and (2) the Santa Barbara Channel VSR Zone off Southern California, USA. Sightings data were a combination of opportunistic naturalist observations reported to Whale Alert and Spotter Pro mobile apps and monthly aerial surveys in the Santa Barbara Channel. No metric for effort was available for any of the sightings data. Acoustic detections were collected by near real-time PAM buoy systems, one off San Francisco and one in the Santa Barbara Channel (Baumgartner et al., 2019). The percentage of 15-min summary periods per day with blue whale acoustic detections was calculated by Whale Safe, and represents the number of 15-min periods with blue whale detections (robots4whales.whoi.edu) relative to the total number of monitored 15-min periods for that day. Lastly, the blue whale SDM was developed by Abrahms et al. (2019) using location data from satellite-tagged blue whales (Hazen et al., 2017). The blue whale SDM is a multimodel ensemble (https://coastwatch.pfeg.noaa.gov/projects/whalewatch2/whalewatch2_map.html), and the predictions are generated using both Generalized Additive Mixed Model and Boosted Regression Tree modeling frameworks (Abrahms et al., 2019). Ensemble models can improve model accuracy, as was seen with the approach by Abrahms et al. (2019) for this blue whale SDM. The model predicts blue whale habitat suitability, a score ranging from 0 to 1 where 0 indicates unsuitable habitat and 1 represents regions where the model predicted optimal habitat conditions. Habitat suitability predictions were based on environmental conditions hypothesized to influence blue whale habitat use, including sea surface temperature, sea surface height and isothermal layer depth from Regional Ocean Modeling Systems configured to the CCS (oceanmodeling.ucsc.edu). Habitat suitability was predicted daily (0–100%) across the CCS with a spatial resolution of 10 km × 10 km grid cells. Within the Whale Safe program, detections from each of the data sources are assigned a categorical “high,” “medium” or “low” rating based on the number of detections/sightings per day and the SDM value to qualitatively characterize the probability of whale presence for the purpose of informing voluntary speed reduction (Table 1). To visualize the contribution of the three data sources in each region, we summarized the daily records from each of the continuous data sources into monthly averages, as well as the number of days that each data stream was offline during the study period.Table 1.Summary of methodology followed by Whale Safe to categorically rank blue whale presence for each of its three data sources (sightings, acoustics and a blue whale SDM). More details on each ranking can be found on the Whale Safe website (https://whalesafe.com/methodology/) Table 1. long description.Sightings dataNumber of whales observed (daily)Whale presence rating0Low1Medium>1HighAcoustic dataPercentage of summary periods with confirmed acoustic detections (daily)Whale presence rating0%Low0–10%Medium≥10%HighBlue whale SDMSan FranciscoAverage habitat suitability across vessel speed reduction zoneWhale presence rating<0.70Low≥0.70MediumSanta Barbara ChannelAverage habitat suitability across vessel speed reduction zoneWhale presence rating<0.75Low≥0.75MediumA table summarizing Whale Safe methodology for ranking blue whale presence as Low, Medium, or High based on sightings, acoustic detections, and S D M habitat suitability. See long description. Table 1.Summary of methodology followed by Whale Safe to categorically rank blue whale presence for each of its three data sources (sightings, acoustics and a blue whale SDM). More details on each ranking can be found on the Whale Safe website (https://whalesafe.com/methodology/) Table 1. long description.Sightings dataNumber of whales observed (daily)Whale presence rating0Low1Medium>1HighAcoustic dataPercentage of summary periods with confirmed acoustic detections (daily)Whale presence rating0%Low0–10%Medium≥10%HighBlue whale SDMSan FranciscoAverage habitat suitability across vessel speed reduction zoneWhale presence rating<0.70Low≥0.70MediumSanta Barbara ChannelAverage habitat suitability across vessel speed reduction zoneWhale presence rating<0.75Low≥0.75MediumA table summarizing Whale Safe methodology for ranking blue whale presence as Low, Medium, or High based on sightings, acoustic detections, and S D M habitat suitability. See long description. Summary of methodology followed by Whale Safe to categorically rank blue whale presence for each of its three data sources (sightings, acoustics and a blue whale SDM). More details on each ranking can be found on the Whale Safe website (https://whalesafe.com/methodology/) Table 1. long description. A generalized linear mixed-effect model (GLMM) framework with the glmmTMB package in R (Brooks et al., 2017) was used to evaluate the relationships between the blue whale SDM predictions and the in situ sources of whale presence (sightings and acoustic data) on a daily scale when data were available. Sightings or acoustic data served as the response variable. Seasons form part of a cycle that impacts whale presence (i.e., migratory behaviors), which can be described by the angular equivalent of the state of the cycle using sine and cosine functions (Bailey et al., 2013). Month was converted into two vectors using sine and cosine terms, and both terms were included as fixed effects. An interaction between the SDM predictions and region (i.e., San Francisco and the Santa Barbara Channel) was also included as a fixed effect to allow for regional differences in the relationships between the habitat suitability predictions and in situ data. The Santa Barbara Channel served as the reference level. Year was centered and included as a random effect to account for interannual variation. A Poisson distribution with a log link function was used for the sightings model, where the response variable represented count data, while a beta distribution with a logit link function was used for the acoustics model, where the response variable represented a percentage of summary periods with acoustic detections. We allowed zero inflation to vary by region. Results Blue whale sightings, acoustic detections and SDM predictions were analyzed across 846 days off San Francisco (September 2022–January 2025) and 1,579 days in the Santa Barbara Channel (September 2020–January 2025). The acoustic and SDM data went offline for a prolonged period (n acoustic = 474 days, n model = 170 days off San Francisco; n acoustic = 389 days, n model = 180 days in the Santa Barbara Channel) due to technological failures and damage sustained by the moored PAM systems. Given the opportunistic nature of the sightings data, there was no information available to account for effort or recorded days with missing visual effort. The offline time periods often differed between the acoustic and modeled data sources, meaning there was an increased likelihood that at least one data stream was active for each day of the study period. There were more frequent acoustic detections of blue whales than reported sightings in both regions (Figure 1). Sighting numbers were generally very low and peaked in both regions in summer 2024. There were relatively high blue whale acoustic detections in 2020 in the Santa Barbara Channel, but they were much lower from 2021 onwards, other than briefly in mid-2024. Acoustic detections of blue whales were much higher off San Francisco in 2022–2023 than during the corresponding period in the Santa Barbara Channel (Figure 1).Figure 1.Year-month averages of blue whale sightings records (top panel), acoustic detections (middle panel), and blue whale habitat model predictions (bottom panel). Data are separated by study region, which includes San Francisco (left) and the Santa Barbara Channel (right). Grey rectangles display when the voluntary vessel speed reduction (VSR) zone was active and blue dotted rectangles indicate when a given data type was “offline”. Figure 1. long description.Six line graphs comparing whale sightings, acoustic detections, and habitat suitability across two regions over time. See long description. Year-month averages of blue whale sightings records (top panel), acoustic detections (middle panel), and blue whale habitat model predictions (bottom panel). Data are separated by study region, which includes San Francisco (left) and the Santa Barbara Channel (right). Grey rectangles display when the voluntary vessel speed reduction (VSR) zone was active and blue dotted rectangles indicate when a given data type was “offline”. Figure 1. long description. The blue whale habitat suitability predictions were positively correlated with sightings data in the Santa Barbara Channel (×2.63 as many whales when habitat suitability is 1, P = 0.032, n = 1,402) (Table 2). There was also an insignificant regional difference observed between the WhaleWatch predictions and sightings data (P = 0.079, n = 2080), suggesting a weak, but positive relationship between sightings data and WhaleWatch predictions in San Francisco (×0.90 as many whales when habitat suitability is 1). A positive relationship was also observed between the WhaleWatch habitat suitability predictions and acoustics data in the Santa Barbara Channel (×5.34 as many summary periods with acoustic detections when habitat suitability is 1, P < 0.0001, n = 1,089). The relationship between the WhaleWatch predictions and acoustic detections significantly differed between the two regions (P < 0.0001, n = 1,379). Acoustic detections had a stronger positive relationship with predicted habitat suitability in San Francisco (×25.69 as many summary periods with acoustic detections when habitat suitability is 1, P = 0.0079, n = 290).Table 2.Results from the acoustics and sightings data generalized linear mixed models (GLMMs) Table 2. long description.ParameterCoefficient estimateConfidence interval P Sightings data modelIntercept0.4280.126–1.4530.174WhaleWatch2.6311.086–6.3750.032 Region – SF1.3990.565–3.4650.468Month (sin)0.5360.368–0.7810.001 Month (cos)0.2870.193–0.410<0.0001 WhaleWatch: Region – SF0.3410.103–1.1330.079Acoustic data modelIntercept44.35633.458–55.8250.335WhaleWatch5.3402.081–13.024<0.0001 Region – SF31.74722.294–42.9890.002 Month (sin)21.58014.801–30.358<0.0001 Month (cos)50.54844.932–56.1490.849WhaleWatch: Region – SF85.97475.211–92.528<0.0001 A statistical table showing G L M M results for acoustics and sightings data across parameters like WhaleWatch, Region, and Month. See long description.Variables with P < 0.05 are bolded. The Santa Barbara Channel served as the reference level for the region parameter. Coefficient estimates have been converted back to the response scale, and the 95% confidence interval is provided. The WhaleWatch parameter represents the habitat suitability prediction. Table 2.Results from the acoustics and sightings data generalized linear mixed models (GLMMs) Table 2. long description.ParameterCoefficient estimateConfidence interval P Sightings data modelIntercept0.4280.126–1.4530.174WhaleWatch2.6311.086–6.3750.032 Region – SF1.3990.565–3.4650.468Month (sin)0.5360.368–0.7810.001 Month (cos)0.2870.193–0.410<0.0001 WhaleWatch: Region – SF0.3410.103–1.1330.079Acoustic data modelIntercept44.35633.458–55.8250.335WhaleWatch5.3402.081–13.024<0.0001 Region – SF31.74722.294–42.9890.002 Month (sin)21.58014.801–30.358<0.0001 Month (cos)50.54844.932–56.1490.849WhaleWatch: Region – SF85.97475.211–92.528<0.0001 A statistical table showing G L M M results for acoustics and sightings data across parameters like WhaleWatch, Region, and Month. See long description.Variables with P < 0.05 are bolded. The Santa Barbara Channel served as the reference level for the region parameter. Coefficient estimates have been converted back to the response scale, and the 95% confidence interval is provided. The WhaleWatch parameter represents the habitat suitability prediction. Results from the acoustics and sightings data generalized linear mixed models (GLMMs) Table 2. long description. Variables with P < 0.05 are bolded. The Santa Barbara Channel served as the reference level for the region parameter. Coefficient estimates have been converted back to the response scale, and the 95% confidence interval is provided. The WhaleWatch parameter represents the habitat suitability prediction. Discussion and conclusions The rapid changes taking place in the marine environment due to unprecedented environmental change will require dynamic management approaches that can respond to such change while balancing ecological, social and economic priorities. DOM approaches can help to meet these demands, although they have large data requirements that can serve as a barrier to their success and implementation (Hobday et al., 2013; Lewison et al., 2015; Maxwell et al., 2015; Smith et al., 2021). Thus, identifying possible solutions to such barriers may be key to the broader application of DOM. Our results indicate that SDM outputs can provide valuable contributions alongside in situ monitoring for detecting whale presence (sightings records and PAM). While the multiple data sources available for Whale Safe may serve as the “gold standard” given the benefits of co-occurring sources of insight, the correlations between the independent in situ data sources and the blue whale SDM suggest that SDMs have potential to serve as indicators for the probability of species presence, via habitat suitability predictions, when other data sources are unavailable (e.g., technological failures, seasonal or cost limitations). San Francisco and the area around the Santa Barbara Channel serve as important habitat for blue whales (Irvine et al., 2014), yet additional research should also be completed to investigate the different results observed between these regions that could be attributed to a range of factors (e.g., interannual variability of whale presence in the two regions, differences in sightings effort, changes in vessel traffic and ambient sound levels making it easier/harder to detect whale calls and the number of days the PAM buoys were online/offline). Ideally, stronger relationships between SDM predictions and the sightings data would be observed for both regions before relying on SDMs alone to guide conservation and management decisions. Together, however, these results support the use of SDMs to address some of the data requirements for DOM initiatives, expanding the overall accessibility of this climate-ready management approach. DOM programs may rely on consistent, near real-time access to species presence data to guide the regulation of various ocean resource-user groups (e.g., shipping companies and fisheries (Lewison et al., 2015; Maxwell et al., 2015; Barlow and Torres, 2021). Data sources can include fisheries-dependent logbook records, PAM, SDM habitat suitability predictions and bio-logging or bio-telemetry detections, among others (Hazen et al., 2018; Oestreich et al., 2020; Pons et al., 2022; Schwartz-Belkin and Portman, 2023). Each data source can be used to support DOM frameworks, though each comes with their own unique set of drawbacks (e.g., technological failures, cost and seasonally dependent biases), such that all three sources together in a data fusion framework, such as Whale Safe, may provide a more complete picture than any data source alone (Gilman et al., 2006; Hobday et al., 2013; Maxwell et al., 2015; Thorne et al., 2019). Acoustic recordings rely on whale vocalizations (e.g., demographically and seasonally dependent behaviors) to detect presence, while sightings rely on calm waters, good visibility, sampling effort and surfacing animals. As such, acoustics and sightings data are also prone to false negatives. Additionally, documented effort may be unavailable when using community-based sightings, as is the case in the present study, making records of species absence and relationships between sightings and modeled data difficult to interpret. A notable benefit of in situ data, that is not captured by the SDM, is that in situ data provide important insights into species-specific responses to extreme ocean conditions that may not be captured through models informed by historical data. Model predictions highlight where animals are most likely to occur based on the historical data used to train the model, but anomalous ocean conditions and the ephemeral nature of prey aggregations may result in a mismatch between model predictions and actual whale presence (Santini et al., 2021; Brown and Puschendorf, 2025). In these cases, in situ data provide valuable insight into potentially unexpected patterns of species presence. Simultaneous access to multiple data sources can compensate for each dataset’s respective limitations and provide continuity during periods when some data sources are offline. The multiple data sources used to inform Whale Safe offer a number of strengths, including robust spatiotemporal coverage and multiple forms of inference for whale presence (Hausner et al., 2021). However, when multiple data sources are unavailable, as is the case for many regions and species, our results suggest that SDM outputs can fill an important real-time species presence data gap. SDMs have been presented as a valuable tool for guiding DOM approaches, and our correlation results indicate that these modeled outputs were indeed operationally reliable and could be used to assess the probability of species presence via habitat suitability predictions when other data sources fail or are unavailable (Hazen et al., 2013; Brodie et al., 2015; Scales et al., 2017). Ecologically realistic SDM predictions have been recommended for marine spatial planning (Milles et al., 2025), anticipating marine heatwave impacts on fisheries (Farchadi et al., 2024), and supporting marine conservation and management more broadly (Braun et al., 2023). While the use of SDM outputs for guiding DOM offers many benefits (e.g., broad spatial coverage, multispecies and fine-scale resolution), the validation and maintenance of these models and tools over long periods of time may only be possible with continuous support and near real-time environmental data availability (Leidner and Buchanan, 2018). Academic-sponsored projects dependent on grant cycles may not be able to sustain such programs long term (Welch et al., 2019; Barlow and Torres, 2021). The automated operationalization of WhaleWatch, in addition to NOAA’s continuous support and maintenance after the initial development funding from NASA, has underpinned its longevity. SDMs can require large amounts of species distribution data for their development and validation (Wisz et al., 2008), although approaches for more sparse datasets are being developed (e.g., see Welch et al., 2019). They also need periodic reassessment and potential refitting as the historical relationships identified between species and environmental conditions may alter over time with climate change (Elith and Leathwick, 2009). This may be especially important given potential species redistributions due to anticipated ocean variability. Such factors will be critical to incorporate into the design of future SDM tools to support the broad application of DOM (Welch et al., 2019). DOM may play a core role in climate adaptation strategies, thus identifying the features and mechanisms that contribute to the success of these programs will be key to its efficacy into the future. Summary: This manuscript is directed towards conveying how species distribution models (SDM) can support dynamic ocean management efforts of species of concern through a real-world case study—blue whales in the California Current System off San Francisco and Santa Barbara. To achieve these results, the authors leverage model predictions developed by Abrahms et al. (2019, Diversity and Distributions) in the context of two other data streams, sightings and acoustics data. The authors demonstrate that the SDM predictions were correlated with sightings and acoustics data with two tests, Spearman and Kendall correlation coefficients in both regions although correlation was stronger in San Francisco compared to Santa Barbara. Together, the authors have compiled and presented a manuscript that illuminates the utility of incorporating SDMs into the portfolio data streams supporting dynamic ocean management, such as when in- situ data streams may be unavailable. Combined, findings have clear applications that can assist regional management. While the authors alluded to the shortcomings inherent with in-situ data streams, greater focus on the shortcomings of SDM predictions would assist in demonstrating a more holistic interpretation of observed results. The results are informative and I recommend publication after major revision. Major issues: The greatest weakness of the manuscript is its characterization of the seeming objectivity of the SDMs compared to the sightings and acoustics data streams. There is no doubt that aggregating multiple streams of data is necessary to provide more complete analysis for dynamic ocean management (i rely on SDMs for my work as well!) Nevertheless, while there is an indication that SDMs can be optimized compared to other data streams on lines 80 and 185–186, there is no elaboration on how the model used to drive Whale Watch 2.0 was optimized. Indeed, details of the model can be read more extensively in Abrahms et al. (2019, Diversity and Distributions) but the present manuscript would benefit from a brief overview on what was optimized in the underlying model. It is also worth mentioning that while it is possible to account for underlying biases in SDMs, there will always be some residual level of bias. For example, the model driving Whale Watch 2.0 is a boosted regression tree and is informed by electronic tag data, yet random effects are not possible to include in a boosted regression tree. If actions were taken to reduce the spatial autocorrelation of data driving the SDM referenced here, the authors should mention such. It would also bolster the authors argument that SDMs are robust tools to support dynamic ocean management by giving examples of ecologically realistic predictions from boosted regression tree SDMs in the literature (see Braun et al. 2023, Ecological Applications; Farchadi et al. 2024, Fish and Fisheries; Milles et al. 2025, Marine Ecology Progress Series). Along similar lines, the authors mention that the present SDM was more correlated with the other data streams in San Francisco compared to Santa Barbara. As a reader, I would appreciate further discussion into why this may be; where were the majority of underlying electronic tag data located? While these data again are available in Abrahm et al. (2019, Diversity and Distributions), it would improve the nuance and reduce potential overoptimism of the model performance compared to greater pessimism of the other data streams (this is not to say one data stream is more robust; there needs to be more discussion of the model performance to “level the playing field”). These findings may also allude to the importance of the domain and scale in developing and predicting SDMs. The California Current System is particularly dynamic; therefore, there may be regions in which the model performs better than others. It may be worthwhile to mention the ability of the model to capture (or fail to) the relation between blue whales and predicted suitable habitat compared to the other data streams when analyzing the correlation values stated in the results. Minor Comments: - On line 49, are you referencing the International Union for Conservation of Nature of U.S. Endangered Species Act when referencing “endangered”? - On line 127, there is reference to days where SDM predictions were unavailable due to the model going offline. It is unclear to me if the model is being predicted to environmental data recorded by PAMs. And if the PAM went offline and the model is being predicted to PAM arrays, why would the acoustic stream have more offline days compared to the SDM? I think it is likely a clarity issue. I thank the authors for an interesting read and thank the editor for the opportunity to review. References: Braun, C.D. et al. (2023) “Building use‐inspired species distribution models: Using multiple data types to examine and improve model performance,” Ecological Applications, 33(6), p. e2893. Available at: https://doi.org/10.1002/eap.2893. Farchadi, N. et al. (2024) “Marine heatwaves redistribute pelagic fishing fleets,” Fish and Fisheries, n/a(n/a). Available at: https://doi.org/10.1111/faf.12828. Milles, H.M. et al. (2025) “Characterizing seasonal whale shark habitat in the western North Atlantic,” Marine Ecology Progress Series, 766, pp. 91–106. Available at: https://doi.org/10.3354/meps14906. The work is interesting and could be a necessary contribution to understanding how different sources of information can be useful in modulating species distributions, specifically marine species such as cetaceans. Although the work raises this need and presents an analysis with whale sighting data in different formats, it is actually unclear in several respects, so I suggest that these issues be carefully addressed before it can be accepted for publication. The article is not consistent with the terms most commonly used in SDM, so it would be advisable to review the publication Melo-Merino et al. 2020 Ecol. Model. (https://doi.org/10.1016/j.ecolmodel.2019.108837) for a more comprehensive reference on these issues. In this publication, you will find the need to clarify several aspects that are relevant, on the one hand, to the concepts and, on the other, to the implementation of the tools. For example, the use of the term habitat-based model may be appropriate but at the same time confusing if it is not made clear why this term and not another is used, given the objective of moderation. Another aspect that may be confusing from the title’s approach could be what Dynamic means. Although it is clear that they are talking about dynamic ocean management, this can be very vague and could even be confused (or confuse), for example, with dynamic SDM modeling approaches (e.g., see Dobson et al. 2023 MEE; https://doi.org/10.1111/2041-210X.14101). In other words, this distinction and objective need to be clarified in greater depth. This same idea about the need to be more specific is scattered throughout the text, with other examples of terms including the following: additional/multiple data streams (in the impact statement or on line 187); near real-time habitat; multiple data sources and data redundancy; among the most important. Another aspect that may be important is the relationship between this work/analysis and what is known as time-specific niche modeling. I bring this up because it specifies this idea of near real-time data (gap) and therefore opens up the possibility of thinking about this concept of making time-specific models, which, although it is something that is only just being explored, already has a perhaps more direct meaning with what is being proposed. In general terms, my opinion is that the work offers much more than can be understood after reading it, and it really detracts from leaving a clear notion of how this can be operated (i.e., data + concepts) in a clear and forceful manner. Hence my suggestion that it be carefully reviewed and addressed by pointing out and linking the concepts to existing literature. I am attaching a file for reference to my comments here. Dear authors, We have now received two evaluations of your manuscript. Although we had initially sought an additional review, we believe the two reports obtained provide sufficient and valuable feedback for the manuscript. Both reviewers suggested major revisions so I am therefore returning the manuscript to you with the expectation that you can consider their comments, which I agree will strongly improve the paper. Please pay particular attention to improving the clarity of the overall narrative and the methods section. Both reviewers have provided references that may help guide your revisions. Thank you for your work and for your patience throughout the review process. No accompanying comment. After reviewing the manuscript for a second time, I realized that the presented analysis and text do not adequately account for (1) non-random missing data and (2) unaddressed temporal dependence, both of which may bias the strength and significance of the reported correlations. Moreover, (3) there is no clear explanation of how the SDMs were constructed in terms of environmental predictors. The link provided in the text does not appear to work, and there is insufficient detail regarding the variables or factors included in the models. In addition, there is no clear delineation of the species’ accessibility area (see Barve et al. 2011, Ecological Modelling), which is crucial for many modeling algorithms given its strong influence on model performance and statistical significance. As currently presented, there is no clear justification for what “suitability” represents in the modeling framework. I therefore encourage the authors to carefully consider these comments and strengthen both the analysis and the manuscript accordingly. Overall, these issues should ideally be addressed through additional analyses to ensure the robustness of the results or at least a really robust justification, but preferred the reanalysis of the data. Regarding the first general comment, the analyses only include days when the monitoring system was online. Because offline periods may coincide with environmental conditions that also influence whale presence or detectability, the missing data may not be random. This could introduce selection bias, meaning that the reported correlations may not represent patterns across the full study period. The authors could evaluate whether online/offline status is systematically related to environmental or detection variables and, if necessary, apply statistical adjustments to account for potential bias. Regarding the second point, the correlation tests do not account for temporal autocorrelation or seasonality in the daily time series. Since daily ecological data are typically temporally structured, treating observations as independent may inflate statistical significance. As a result, the reported p-values may overstate the strength of the associations. The authors could reanalyze the data using time-series approaches that account for autocorrelation and seasonal patterns, or implement a more comprehensive model-based framework. I thank the authors for an interesting read and for being responsive to my feedback. I recommend publication after minor revision. The authors have resolved my previous feedback concerning the need for greater context surrounding the models that feed into Whale Safe and have added the details sufficient for a perspective in Cambridge Prisms: Coastal Futures. The authors have also sufficiently revised the manuscript to acknowledge the differences in correlation values among regions and allude to the gap in knowledge that should be explored as DOM efforts are pursued. Minor comment The manuscript sometimes alludes that the SDM predictions are indicative of species presence (80–88; 163–165; 209). While acoustics and visual sightings directly identify whale presence, SDMs do not. I recommend reframing that the SDM predicted values are not directly linked to species presence; rather, the model predicts optimal habitat conditions which may incur greater probable presence. I encourage the authors to review the manuscript for additional examples I may have missed. Dear Authors, As you can see one of the reviewers has further questions regarding the methodology of the SDM employed while the other one is satisfied. I am returning the manuscript to you with the hopes that you can further address the concerns of the reviewer. While I don’t expect you to run all the analysis required by the reviewer, I would like to see strong justifications for those not directly addressed by extra analysis. Many thanks No accompanying comment. Dear Authors, I want to start by personally apologize for the length of the revision process and thank you for your patience. I am pleased with the revisions you have made and delighted to let you know your manuscript has now been accepted for publication. No accompanying comment.