Ecosystem damage by increasing tropical cyclones

The study addresses how increasing tropical cyclone (TC) activity under anthropogenic warming affects coastal marine ecosystems, a key gap compared with well-studied climate stressors such as warming, marine heatwaves, and ocean acidification. TCs cause extensive human and ecological damages, and their global proportion and frequency of very intense events (category 4–5) and lifetime maximum wind speeds are projected to increase at 2 °C warming. Despite evidence of large ecosystem damages to coral reefs, mangroves, salt marshes, and seagrass, quantitative predictions for future TC-driven ecosystem impacts are lacking. The North Atlantic has experienced an accelerated rise in major TC exceedance probability (about 42% per decade since 1979, far above the global average), driven by anthropogenic and natural forcings, offering a window into potential future global impacts. The authors quantify trends in landfalling NATCs (1970–2019) and synthesize immediate ecological responses across five dominant coastal ecosystems to link storm attributes to biological impacts and infer implications under continued warming.

The paper situates TCs among climate-driven extreme events with major economic and ecological consequences, noting projections of increased intensity and frequency of very intense storms. Prior work documents TC damages to coastal habitats and rising economic losses, yet ecosystem-focused, quantitative syntheses tied to storm metrics are scarce. Hypothesized determinants of damage include landfall wind speed and oceanographic drivers (waves, surge, sediment dynamics, salinity), but generalizable relationships across ecosystems remain unclear. The North Atlantic’s recent multidecadal increase in major TCs, potentially influenced by Atlantic Multidecadal Variability, provides an empirical basis to examine ecosystem responses. The study builds on disparate case studies by applying standardized effect sizes to assess species performance and community metrics, addressing a gap in predictive ecology for TC impacts.

Physical storm analysis: Best-track data for the North Atlantic (IBTrACS Version 4, 2020) were analyzed for 1970–2019. Variables included wind intensity (knots), Saffir-Simpson category, and 6-hourly positions with higher-frequency tracking at landfall. Landfalls were defined as storm centers within 60 nautical miles of coastlines (continents and islands >1400 km²). Time series constructed: (1) annual count of landfalling NATCs, (2) annual count of landfalling lifetime maximum intensity (LMI) category 1–5 NATCs, and (3) wind speed at landfall for all LMI categories. Data were aggregated in 3-year bins (final datapoint includes 2018–2019). Monotonic trends were tested using Mann–Kendall; slopes estimated via Theil–Sen regression. Trends were contextualized against high-resolution global model projections at 2 °C warming (after Knutson et al.). Ecological literature synthesis: A systematic Web of Science search (to ~29 July 2021; literature analyzed through 2020) targeted coral reefs, mangrove forests, salt marshes, seagrass meadows, kelp forests, and oyster reefs using TC-related and ecosystem-specific terms. From 248 retrieved articles, 109 relevant NATC impact studies were identified; 80 reported quantitative responses (22 coral reefs, 26 mangroves, 14 seagrass, 13 salt marsh, 7 oysters; plus 2 kelp qualitatively). Each study was reviewed to extract: ecosystem type, impact year, storm name, study location (lat/long), depth (if relevant), co-stressors, spatial extent, taxa affected, pre/post observation dates, response metrics (e.g., growth, abundance, mortality), error and sample sizes (if available), and primary/secondary physical drivers (winds, waves, surge, sediment deposition/erosion, salinity). Impacts of the same TC on different ecosystems or widely separated locations were treated as independent events. Storm attributes (LMI category, landfall wind speed, landfall location) were cross-referenced with IBTrACS. Effect sizes and modeling: Immediate pre- and post-storm values (typically within 90 and 20 months of landfall, median 4 months both) were used to compute unweighted log response ratios (LRR = ln(x_pre/x_post)); unweighted due to missing variance for 77% of cases. Directionality was standardized (e.g., increased mortality treated as negative performance). Zero values were adjusted by a small constant (0.001) when needed. Two response categories were assessed: species performance and community structure/processes. Linear mixed-effects models (LMMs) tested effects of ecosystem type on LRR and relationships between z-scored LRR and landfall maximum wind speed (knots), with storm identity and study-specific effects as random terms; adding distance from landfall to impact site did not improve fit. Qualitative tallies identified primary and secondary physical drivers by ecosystem. Data are available at https://doi.org/10.6084/m9.figshare.27316113.v1.

- NATC landfall trends (1970–2019): - Landfalls of very intense (LMI category 4–5) NATCs increased 68% per decade (Mann–Kendall P = 0.005; Theil–Sen slope = 0.031 yr⁻¹). - Proportion of landfalling very intense NATCs increased 29% per decade (P = 0.035; slope = 0.005 yr⁻¹). - Landfall wind speeds (all LMI cat. 1–5) increased ~4% per decade (P = 0.011; slope = 0.222 yr⁻¹). - Very intense NATC activity correlates positively with ecosystem damage. - Ecological impact synthesis: - 97 NATC landfalls yielded 891 immediate post-storm ecological impacts across coral reefs, mangroves, salt marshes, seagrass meadows, and oyster reefs. - Distribution of unique ecosystem-storm events: coral reefs 30%, mangroves 29%, seagrass 15%, salt marsh 14%, oyster reefs 11%. - 85% of reported impacts involved foundation species; 15% associated species. - 92% of impacts were to species performance; fewer reported for community structure/processes. - Meta-analysis showed strong negative effects (negative LRR) on species performance and community metrics, with the strongest negative impacts in mangrove forests. - LMMs: significant negative relationship between effect size and landfall year for mangroves; no significant temporal trend for other ecosystems. Distance from landfall did not improve models (additional deviance explained ≈ 0.2%; AIC change minimal; χ² = 1.70; p = 0.714). - Most impacting storms were classified as LMI category ≥1; many were very intense. A reporting bias toward very intense storms is likely. - Primary physical drivers varied by ecosystem: winds (mangroves); waves (coral reefs); surge (salt marshes); sediment deposition/erosion (seagrass); salinity changes (oyster reefs). Secondary drivers often co-occurred (e.g., surge with wind in mangroves). - Implication: Increasing NATC intensity and frequency are associated with escalating ecological damages, especially in mangrove forests where damage scales with landfall wind speed.

Findings indicate that immediate ecological impacts from NATCs are broadly negative, with mangroves disproportionately affected and showing damage that scales with landfall wind speed. Ecosystem traits mediate vulnerability: emergent, woody, high-drag mangroves are directly damaged by winds, while lower, flexible or rigid but short habitat-formers (salt marshes, seagrass, oysters) are more impacted by surge, sediment dynamics, and water-quality shifts (e.g., salinity). Coral reefs are strongly influenced by wave action generated by TCs. These mechanistic distinctions suggest that predictive modeling should incorporate multiple storm attributes (wind, waves, surge, rainfall/salinity, sediment transport) and ecosystem-specific functional traits. Rising NATC activity, reflecting both anthropogenic warming and natural variability, serves as an early warning of broader global risks at 2 °C warming, though basin-scale heterogeneity is expected. The study underscores the need to integrate TC extremes into climate-impact frameworks and coastal management, complementing existing focus on warming and heatwaves. Management should target local stressors (sedimentation, hydrology, disease) that amplify TC damages and prioritize resilience, particularly in mangroves, while recognizing that increased TC frequency alone may elevate damages across ecosystems even without stronger intensity–damage relationships outside mangroves.

The paper provides a standardized, cross-ecosystem synthesis linking increasing North Atlantic TC activity to immediate ecological damages, identifying mangrove forests as especially vulnerable with impacts scaling to wind speed at landfall. It demonstrates rising frequencies and proportions of very intense NATC landfalls and associates very intense activity with greater ecosystem damage. Mechanistic insights highlight differing primary drivers across ecosystems (winds vs. waves, surge, sediments, salinity), informing targeted resilience strategies. Future research should: (1) quantify high-resolution physical drivers (winds, waves, surge, sediment, salinity) alongside biological responses; (2) evaluate additional storm metrics (e.g., PDI, ACE); (3) pair detailed ecological sampling with coastal storm models; (4) expand geographic and ecosystem coverage (notably kelp forests) and include reports of limited/no impacts; and (5) improve temporal and spatial resolution of monitoring to enable dynamic, predictive models for TC damages and adaptation planning.

- Geographic and reporting biases: ecological impact studies are concentrated in the U.S. and Caribbean with limited coverage in South/Central America and Mexico; likely bias toward reporting very intense storms and positive (significant) impacts. - Short-term focus: analyses emphasize immediate (days to months) post-storm effects; delayed or chronic impacts and recovery trajectories are underrepresented. - Data constraints: 77% of extracted measures lacked variance, necessitating unweighted effect sizes; inconsistent pre/post sampling intervals (though typically within months) and occasional zero-value adjustments may affect effect size estimates. - Ecosystem coverage gaps: sparse quantitative data for some ecosystems (e.g., kelp forests), limiting generalizability. - Physical driver quantification: primary/secondary drivers largely based on authors’ qualitative attribution; high-resolution, co-located physical measurements (waves, surge, sediments, salinity) were often unavailable. - Model scope: linear mixed-effects models captured wind–damage relationships clearly only for mangroves; other intensity–damage relationships may be nonlinear or mediated by unmeasured factors (e.g., storm size, rainfall totals, antecedent conditions). Basin-scale variability limits extrapolation to global patterns.