Model predicts catastrophic decline of common bottlenose dolphin ( <i>Tursiops truncatus</i> ) population under proposed land restoration project in Barataria Bay, Louisiana, <scp>USA</scp>

The Barataria Bay Estuarine System (BBES) Stock is one of 31 identified stocks common bottlenose dolphin (Tursiops truncatus) occupying the bays, sounds, and estuaries northern Gulf Mexico (NOAA, 2019). stock (hereafter “dolphin population”) centered on Bay, a shallow (largely <2 m depth) estuarine system in Louisiana, close to mouth Mississippi River (Figure 1). This area was heavily impacted by Deepwater Horizon (DWH) oil spill April 2010, population model developed quantify impact abundance (Schwacke et al., 2017, 2022). As result Natural Resources Damage Assessment arising from DWH spill, many environmental restoration projects have been proposed, which Mid-Barataria Sediment Diversion (MBSD) project. MBSD project proposes build operate channel upper Basin 1) that would deliver seasonally varying quantities fresh water, sediment, nutrients order rebuild marsh reduce land loss (U.S. Army Corps Engineers, 2021). water will decreased levels salinity basin. Prolonged exposure low has linked skin lesions other pathologies dolphins can mortality (Booth & Thomas, 2021; Duignan 2020). potential extent this examined recent report National Oceanographic Atmospheric Administration (NOAA; Garrison That gave predictions annual survival rates four geographic regions (“strata”) within (Island, Southeast, Central West; see Figure under two scenarios presented Draft Environmental Impact Statement for proposed project: “Applicants Preferred Alternative” “Diversion scenario,” DS), assumes constructed begins operation 2027, “No Action NDS), not constructed. Here, we integrate each strata (DS NDS) al. (2020) into Schwacke (2022) predict consequences population. an age- sex-structured discrete-time deterministic includes regulation through density-dependent fecundity. allows cohort animals exposed lower fecundity, proportion these recover over time. Animals born after baseline accounts scientific uncertainty its input parameters sampling distribution parameter simulating trajectory sample, thereby giving predictions. Input distributions are given table 1 (2022). We used 1,000 samples as basis possible effect (DS) For partitioned BBES (N = 3,045 95% CI [2,720, 3,611]) same (2020), using estimates with home range centers strata. These come spatial capture recapture analysis (Glennie 2021) forms part inputs model. purposes analysis, assumed stratum demographically independent, i.e., do move another. stratum, ran 66 years (2010–2076) scenarios. In first scenario, representing NDS, without modification. second DS, year 2027 until last 50-year planning horizon 2076, adjusted values multiplying them random draw percentage difference between DS DNS, defined next paragraph. equivalent assuming factors driving vary (see later discussion). multiplier applied all sex-classes. calculated following metrics summarize outcomes runs. (1) (i.e., 2027–2028): (a) excess mortality—the total number expected die minus NDS; (b) change size (c) 2028 NDS. (2) After 10 2037): probability functional extinction, where extinction ≤1 animal; (3) final operations 2076): extinction; case, median value model, together lowest 2.5th highest 97.5th percentile year—these latter represent pointwise confidence interval (CI) prediction. Predicted trajectories NDS up when start (Figures 3 4). During period (2010–2027) populations predicted experience negative and, starting around 2020, begin recover. continue reach steady state long before end modeled time period. Under prediction Island decline, while average rapid declines 3). entire population, summing across strata, decline precipitously at then more gradually, reaching very relative simulation Individual realizations show erratic 5) than averages simulations described previously. occasional large decreases associated there DS; most years, however, be little or no decline. By 50th operation, experienced overall none level corresponding Southeast frequent decreases, but also increases increase DS. Nevertheless, zero. parts West towards zero realizations. 537 [112, 1,400] (Table 2). represents 26% (95% [6, 63]) killed operation. MBSD, 2,056 animals, [1,408, 2,759]. 1,529 [714, 2,270], 24%, [3, 60] contrast, 3, [1, 4] because still recovery spill. Therefore, operations, smaller, 61] compared MSBD, 2038, substantial occurred Dolphins functionally extinct (≤1 animal remaining) 1, 0.37 5). 78% [57, 94] planned 50 Central, West, 0.85 respectively 91% lower, [47, 99] 6). Overall, 85 dolphins, [7, 516] 3,216, [2,740, 3,080] words, 97% [84, 100] analyses here case study use models make quantitative management actions, including expressions Outputs alongside information weigh alternative actions. considered relatively simplistic so briefly consider how complex alternatives could affect results. set limit “functional extinction” animal. To our knowledge agreed threshold; higher thresholds indicate point few they longer form functioning ecosystem. With thresholds, higher. example, if threshold raised 30 MSBD rises 1. results were generated combining separate analyses: some overlapping information—the photo-ID surveys undertaken 2010 2019. Hence it possible, modeling effort, closely building components However, qualitative noted earlier, randomly replicate effects annually. While assumption correct sources (e.g., field hydrography, movement, hence exposure), fully others response exposure). Ideally, different separated sample appropriate just once per projection. reduction variability produce somewhat positive projection, particularly stratum. important source variability, neglected simulations. based single hydrograph, 1970 (cycle0; 2020), reality hydrography substantially among years. mean survival. interannual trajectory, size, good only small amount constrained birth rate. 25% bad year, cannot year. Given this, anticipate addressing issues related discussed paragraph lead overall. Another issue historical hydrograph unlikely accurately reflect future especially climate change. salinity, such prey base, term. much greater anticipated freshwater released positively flow rate river, likelihood high-flow (Garrison A further factor makes projections optimistic dynamics (Lande 2002). First, does account nature births deaths, noninteger counts. Incorporating demographic stochasticity restricting sizes whole numbers, stronger smaller populations, Second, apart effects, include external cause fluctuate time, generally referred “environmental stochasticity.” tends hasten again independent. If away three affected population-level may lower. On hand, disperse regard changes, tend strongly less-affected greater. Genetic supported structure genetically distinct groups Western, East/Central, portions basin (Rosel 2017). Tracking movement patterns via satellite-linked tags shown multiyear site fidelity ranges (Wells 2017), changes coincident fluctuating (Takeshita vertically integrated hydrographic did possibility patches bottom deeper refugia. deep-water channels evidence during periods Further, bay, sound, estuary typically take dives must visit surface breathe. addition, conservative counted longest stretch rather attempting determine multiple bouts exclude adapt behaviorally behavioral adaptation documented previously pace combined genetic unlikely. summary, assumptions catastrophic (97%, 100]) caused scenario. likely become individual) out greatly reduced (91% 95%CI 99]) fourth. those place No happen research written request Marine Mammal Commission; note letter submitted Commission (https://www.mmc.gov/wp-content/uploads/21-05-13-BB-dolphin-popn-trajectory-MMC-response.pdf). Consortium Advanced Research Health (CARMMHA) funding Initiative. thank Lance NOAA providing estimated Applicant Alternative editor, associate reviewers their thoughtful comments suggestions. T.A.M. acknowledges partial support CEAUL (funded FCT - Fundação para Ciência e Tecnologia, Portugal, UIDB/00006/2020). Len Thomas: Conceptualization; data curation; formal analysis; investigation; methodology; software; visualization; writing – original draft; review editing. Tiago A. Marques: Cormac Booth: Ryan Takeshita: Lori H. Schwacke: All R code available repository https://github.com/TiagoAMarques/CARMMHApapersSI shortcut summary page Note https://htmlpreview.github.io/?https://github.com/TiagoAMarques/CARMMHApapersSI/blob/master/FolderArchitecture2runCode/Diversion_ElectronicSupplements.html