Tropical cyclone-induced coastal acidification in Galveston Bay, Texas

The study investigates how tropical cyclones, via extreme rainfall and runoff, drive coastal acidification in estuarine systems, focusing on Galveston Bay, Texas, around Hurricane Harvey (2017). Coastal and ocean acidification, driven globally by rising atmospheric CO2, reduce seawater pH and carbonate mineral saturation states, threatening calcifying organisms such as oysters and corals that provide critical ecosystem services. In coastal zones, additional local processes (river discharge, storms, nutrient inputs) modulate acidification. While tropical cyclones are known to physically damage reefs, their chemical impacts are less well constrained. Hurricane Harvey produced unprecedented rainfall and extended reservoir releases into Galveston Bay, creating conditions likely to alter carbonate chemistry. Given the ecological and economic importance of Galveston Bay’s oyster reefs and existing long-term acidification trends in Texas estuaries, the study aims to characterize post-storm changes, mechanisms, and recovery timescales of the carbonate system and assess implications for calcifying ecosystems.

Prior work documents global ocean acidification from anthropogenic CO2 and its impacts on calcifying organisms and human communities. Coastal acidification is further influenced by local processes including freshwater inputs and eutrophication. Tropical cyclones have been predicted to induce short-term aragonite undersaturation in coastal waters by 2100, and case studies (e.g., Tropical Storms Kate and Isaac) reported post-storm acidification. Projections indicate increasing tropical cyclone intensity and rainfall under continued emissions, heightening risks to coastal ecosystems. In Texas estuaries, long-term alkalinity decreases and acidification have been observed. Post-Harvey studies in adjacent estuaries noted elevated pCO2 attributed to enhanced respiration, and work in Galveston Bay reported microbial and biogeochemical changes and unprecedented freshwater inflow following Harvey, as well as massive oyster mortality linked primarily to low salinity.

Sampling design and timeline: Discrete carbonate chemistry sampling was conducted in Galveston Bay at multiple stations from June 2017 through September 2018, including pre-storm (June 2017) and post-Harvey dates (September 9 and 16, 2017; November 4, 2017; March 24, 2018; June 16, 2018; September 22, 2018). Surface (~0.5 m) and bottom (~1.2 m above bottom) samples were collected with Niskin bottles; CTD profiles (SonTek CastAway) recorded temperature and salinity. Parameters and analyses: Total alkalinity (TA) and dissolved inorganic carbon (DIC) were measured on a VINDTA 3C (coulometric DIC; open-cell potentiometric TA) calibrated with Dickson CRMs; mean precision ±2.5 µmol kg−1 (DIC) and ±2.0 µmol kg−1 (TA). Samples were filtered (0.45 µm inline polypropylene; some 0.2 µm Polycap 36 AS), poisoned with HgCl2, and sealed. Duplicate low-salinity samples analyzed at Scripps (TA/pH) supported internal results; carbonate calculations from those duplicates suggested our reported Ω values may be slightly high (average differences: Ωar +0.57, Ωca +0.34) using Millero constants. Carbonate system calculations: pH, pCO2, aragonite and calcite saturation states (Ωar, Ωca) were computed with CO2SYS using estuarine constants (Millero 2010); seawater constants (Lueker et al., S≥15) were also compared. pCO2 was temperature-normalized to 25.9 °C (npCO2) following Takahashi et al. Statistical analyses: Linear regressions were fit for TA–Salinity, DIC–Salinity, and TA–DIC for each cruise; one-way ANOVA with multiple comparisons tested differences among timepoints (significance p<0.05). Freshwater endmember estimates (TAFW, DICFW) were derived from regression zero-salinity intercepts. Endmember-mixing and fractional composition: A Deffeyes TA–DIC diagram assessed whether carbonate variability fell within the three-endmember mixing envelope defined by Gulf of Mexico (GOM) seawater, Galveston-watershed river water, and rainwater. Mixing fractions (seawater, river, rain) were computed assuming ternary mixing using TA (conservative tracer) and, for comparison, DIC; results were similar, indicating dominance of physical mixing. GOM endmembers were derived from regional cruise regressions at salinities representative of TABS Buoy B (median S=31.7; quartiles S=33.5, 29.3). River endmembers used zero-salinity intercepts of non-storm TA–S and DIC–S regressions; June 2019 Trinity and San Jacinto river samples (weighted by typical contributions ~80% and ~20%, respectively) corroborated estimates. Rainwater was assumed S=0, TA=0; DIC estimated from atmospheric equilibrium at pH 5.4 and pCO2 ~405 ppm (14–16 µmol kg−1) via formula and WEB-PHREEQ. Quality considerations and caveats: Immediate post-storm conditions precluded sampling before September 9. Potential influences of submarine groundwater discharge could not be quantified but appear minor under non-storm conditions based on endmember agreement.

• Pre-Harvey (June 2017) Galveston Bay conditions: average pCO2 470 ± 107 µatm; pH 8.0 ± 0.1; Ωca 3.7 ± 1.0; Ωar 2.3 ± 0.6; no undersaturation observed. • Two weeks after Harvey (September 9, 2017): widespread, severe acidification with average pCO2 985 ± 359 µatm; average pH 7.6 ± 0.2 (>200% increase in acidity relative to June); all stations undersaturated for aragonite (Ωar<1); all stations undersaturated for calcite (Ωca<1) except one bottom-water sample (station 1, Ωca=1.47). Temperature did not change significantly, and npCO2 rose to 981 ± 372 µatm, indicating non-thermal drivers. All carbonate parameters and salinity differed significantly from pre-storm values (p<0.05). • Three weeks after Harvey (September 16, 2017): undersaturation persisted in parts of the bay. Local measurements showed aragonite Ωar=0.04–0.93 and calcite Ωca=0.07–0.99 at multiple stations/samples. Baywide averages trended toward recovery: pCO2 435 ± 288 µatm; pH 8.18 ± 0.24; Ωar 0.6 ± 0.4; Ωca 1.1 ± 0.7, though Ω values remained significantly lower than pre-Harvey (p<0.001) and spatial variability remained high. • By November 2017: average salinity, pH (8.10 ± 0.10), pCO2 (382 ± 83 µatm), Ωar (2.0 ± 0.4), and Ωca (3.3 ± 0.6) returned to pre-storm ranges. From November 2017 to September 2018, near-normal carbonate chemistry prevailed, with minimal undersaturation except at one site in March 2018 during elevated river discharge. • Freshwater endmember shifts: Estimated combined freshwater TA (TAFW) and DIC (DICFW) plummeted 2 weeks post-storm (TAFW from 1853 to 847 µmol kg−1; DICFW from 1843 to 905 µmol kg−1), partially rebounded by week 3 (TAFW 1260; DICFW 1239), remained depressed in early November (TAFW 1504; DICFW 1417), and returned to pre-storm levels by March 2018. • Source-water composition: Post-Harvey freshwater fractions were ≥90% at most sites (≤10% GOM seawater), with rainwater fractions increasing from 1–21% (June 2017) to 23–48% (September 2017), while river fractions were similar to pre-storm. This indicates the acidification was primarily driven by increased rainwater inputs characterized by very low TA, DIC, pH, O2, buffering capacity, and negative [TA−DIC]. • Process attribution: TA–DIC data fell within the three-endmember mixing envelope across all months, showing dominant control by physical mixing; any biological signals (e.g., respiration, CaCO3 dissolution) were not resolvable beyond mixing uncertainty. Temperature changes were not responsible for pCO2 shifts. • Ecological context: The extreme, prolonged freshwater pulse also coincided with high oyster mortality in Galveston Bay; storm-induced acidification likely impeded reef recovery.

Findings demonstrate that Hurricane Harvey caused an ecosystem-scale coastal acidification event in Galveston Bay, inducing bay-wide undersaturation with respect to both aragonite and calcite for two weeks, with localized persistence beyond three weeks. The event coincided with unprecedented freshwater inflow dominated by rainwater, which displaced seawater within the estuary and lowered TA and DIC endmembers, overwhelming the system’s buffering capacity. Deffeyes mixing analysis and consistent TA- vs DIC-derived fractions show that physical mixing of three endmembers (GOM seawater, river water, rainwater) controlled the carbonate system during and after the storm; temperature effects were negligible, and potential biological processes (enhanced respiration, CaCO3 dissolution) could not be distinguished within the mixing uncertainty. Recovery of bay-average carbonate chemistry occurred by ~10 weeks post-landfall, but freshwater endmember properties remained depressed into early November, possibly reflecting residual stormwater and/or seasonal variability. Given the documented massive oyster mortality from low salinity and the sensitivity of larval and adult oysters to depressed Ω, storm-driven acidification likely compounded salinity stress and impeded reef recovery. With projections of increased tropical cyclone rainfall, similar or more severe acidification events may become more frequent, posing significant risks to calcifying coastal ecosystems and resource management.

This study provides direct evidence that an extreme tropical cyclone under current climate conditions can drive prolonged, ecosystem-scale coastal acidification, including bay-wide aragonite and calcite undersaturation, primarily via rainwater-dominated freshwater inputs and prolonged reservoir releases. Physical mixing among rainwater, river water, and seawater governed carbonate chemistry changes; system-wide conditions largely recovered within ~2–3 months. The results imply that increasing tropical cyclone rainfall with climate change represents a significant threat to coastal calcifying ecosystems, such as oyster reefs, potentially hindering recovery following storm damage. Future research should expand post-storm carbonate monitoring across estuaries, resolve the roles of respiration and mineral dissolution during undersaturation, quantify groundwater contributions, and integrate hydrodynamic–biogeochemical models to predict acidification severity and duration under changing storm regimes.

• Sampling could not occur immediately post-landfall; earliest post-storm data are from September 9, potentially missing peak changes. • Attribution of biological processes (respiration, CaCO3 dissolution) is limited; TA–DIC variability was dominated by physical mixing within the endmember uncertainty envelope. • Only one November sampling constrains interpretation of depressed freshwater endmembers (TAFW, DICFW) as storm-related versus seasonal. • Submarine groundwater discharge volumes and chemistry were unknown and not quantified. • Choice of carbonate constants and inter-lab comparisons indicate Ω values may be slightly overestimated; results should be viewed as conservative for saturation state. • Spatial coverage and station availability varied by date; some sites closer to the bay mouth were sampled earlier post-storm.