INTRODUCTION

Mangrove wetlands are blue carbon ecosystems and are considered among the most carbon-rich and productive ecosystems in the world (Alongi, 2012;Donato et al., 2011). They provide essential ecosystem services to coastal communities and industries, including coastal protection and storm attenuation, water-quality improvement, wildlife habitat, support of fisheries, and nutrient cycling (Barbier et al., 2011;Gedan et al., 2010;Lee et al., 2014;Nagelkerken et al., 2008). Despite their small area (0.7%) compared to tropical forests (Giri et al., 2011), these forested wetlands are highly efficient at sequestering and storing large amounts of carbon (C) in vegetation (above-and below ground) and in soil and sediments (Donato et al., 2011;Lovelock et al., 2017;Rovai et al., 2018Rovai et al., , 2021)), and thus play an essential functional role in the global C cycle (Bouillon et al., 2008;Breithaupt et al., 2012;Breithaupt & Steinmuller, 2022;Twilley et al., 1992). Such high rates of C storage and sequestration underscore their potential contribution to global climate change adaptation and mitigation (Lovelock et al., 2017;Murdiyarso et al., 2015;Siikamaki et al., 2012).

For decades, mangrove forests around the world have experienced rapid declines as a result of natural and anthropogenic disturbances (Polidoro et al., 2010;Sasmito et al., 2020). Even though human impacts account for most of global mangrove loss and degradation (Friess & Webb, 2014;Sasmito et al., 2020), natural factors such as climate change, saltwater intrusion, shoreline erosion, and extreme weather events account for more than 11% of mangrove loss worldwide (Goldberg et al., 2020). Climate extreme events such as hurricanes and droughts can cause severe forest damage and often lead to localized mangrove diebacks (Beckett et al., 2023;Lagomasino et al., 2021;Mafi-Gholami et al., 2020), with long-term consequences for regional economies, inland infrastructure, and the provision of ecosystem services (Alongi, 2008;Menendez et al., 2020). Hurricanes are one of the most destructive natural disturbances affecting mangrove forests, and present an even greater risk in regions with high recurrence frequency such as south Florida, the Gulf of Mexico, and the Caribbean region, changing the extent, distribution, species composition, and trajectories of ecosystem structure and function of mangrove vegetation (Danielson et al., 2017;Imbert, 2018;Lagomasino et al., 2021;Rivera-Monroy et al., 2019).

Large-scale episodic disturbances such as hurricanes create biological legacies that interact with local and regional environmental conditions, thereby regulating ecological processes and defining trajectories of ecosystem recovery post-disturbance (Johnstone et al., 2016;White & Jentsch, 2001). For example, south Florida is regularly affected by hurricanes, and mangrove forests in the region have been impacted by four major hurricanes (Andrew in 1992, Katrina and Wilma in 2005, and Irma in 2017) over the last three decades changing community structure and ecological processes (Baldwin et al., 2001;Castañeda-Moya et al., 2010;Rivera-Monroy et al., 2019;Smith et al., 2009). The most recent hurricanes, Wilma (2005) and Irma (2017), caused major forest structural damage including defoliation and tree mortality, as well as erosion of surface sediments across mangroves in the Florida Coastal Everglades (FCE) (Chavez et al., 2023;Danielson et al., 2017;Feher et al., 2019;Lagomasino et al., 2021;Rivera-Monroy et al., 2019;Smith et al., 2009;Xiong et al., 2022). However, despite these disturbances, mangroves in this region have shown a high degree of ecological stability and resilience (Chavez et al., 2025;Danielson et al., 2017;Rivera-Monroy et al., 2019;Xiong et al., 2022). Mangrove species are well adapted to quickly recover from disturbances due to their resilient traits such as resprouting from epicormic shoots and high rates of waterand nutrient-use efficiency (Alongi, 2008;Lugo, 1980Lugo, , 2008)). As a result, most of the affected forests in the Everglades not only recovered to pre-disturbance conditions but also benefited from the input of hurricane-induced phosphorus (P)-rich mineral sediments associated with the storm surge. These sediments contribute to building elevation, stimulating peat soil development, enhancing soil nutrient availability for plant growth, and promoting forest recovery and resilience to sea-level rise (SLR) (Castañeda-Moya et al., 2010, 2020;Danielson et al., 2017;Davis et al., 2018;Feher et al., 2019;Giri et al., 2023). Although many studies have documented hurricane impacts on spatiotemporal patterns of vegetation structure, aboveground biomass, and productivity across Everglades mangroves (Baldwin et al., 2001;Danielson et al., 2017;Rivera-Monroy et al., 2019;Zhang et al., 2008), few have quantified changes in belowground root processes (i.e., biomass, production, and decomposition) following a major hurricanes (Kuhn et al., 2021;Radabaugh et al., 2019).

Root processes play a critical role in soil formation and vertical accretion in mangrove forests by storing C below ground and promoting root volume expansion enabling them to maintain surface elevation relative to SLR (Chen & Twilley, 1999a;Krauss et al., 2017;McKee et al., 2007). Roots are also essential for water and nutrient uptake, transport, and storage (Eissenstat et al., 2000), and nutrient cycling (Nadelhoffer et al., 1985). Furthermore, recent studies have underscored the significant contribution of root dynamics in estimating global C budgets of mangrove ecosystems, with root production accounting for 38% of the total net production (Bouillon et al., 2008).

Soil nutrient availability is considered one of the major factors controlling mangrove biomass allocation and C partitioning between above-and belowground components (Castañeda-Moya et al., 2011, 2013;Lovelock et al., 2006). The allochthonous input of P-rich sediments during hurricanes into FCE mangroves acts as a natural fertilization mechanism that could have long-term legacies on soil fertility gradients and vegetation dynamics, including root biomass accumulation and C sequestration influencing forest recovery (Castañeda-Moya et al., 2020). It is well established in the literature that plants in nutrient-poor soils tend to allocate more biomass to roots to capture the most limiting nutrient to maintain a favorable carbon:nutrient balance (Castañeda-Moya et al., 2011;Chapin et al., 1986). Therefore, resource limitation in mangroves, particularly in karstic oligotrophic P-limited settings such as south Florida (Noe et al., 2001), may be significant in determining patterns of C storage between above-and belowground compartments. Mangroves respond differently to short-and long-term changes in soil nutrients and disturbance legacies (Castañeda-Moya et al., 2020) based on species-specific life history traits and adaptations (Chapin et al., 1987). Indeed, previous pre-hurricane (2000)(2001)(2002)(2003)(2004) studies in FCE mangroves reported strong links between belowground root dynamics (root biomass, productivity, turnover, and decomposition) and environmental gradients-notably P availability, sulfide concentrations, and hydroperiod (e.g., frequency of inundation)-suggesting high phenotypic plasticity of mangrove root systems (Poret et al., 2007;Castañeda-Moya et al., 2011). Given the significance of root dynamics to ecosystem production, C budgets, and ecological processes in mangrove forests (Arnaud et al., 2023;Bouillon et al., 2008;Cormier, 2021), there is an increasing need to quantitatively assess what changes, if any, occur within the belowground component of FCE mangroves, especially given the projected increase in the frequency of major hurricanes (categories 4/5) due to climate change (Kossin et al., 2020;Walsh et al., 2016;Webster et al., 2005) and the rapid increase in SLR in this coastal region over the last decade (Wdowinski et al., 2016).

Here, building on previous pre-hurricane baseline studies in the FCE (Castañeda-Moya et al., 2011;Poret et al., 2007), we quantified and compared long-term changes in root processes before (pre-hurricane period: 2000-2004) and after post-hurricane periods, including post-Wilma (May 2012), immediate-post-Irma (March 2018), and post-Irma (March 2023). Because our long-term (25 years) experimental plots in the FCE mangroves were established prior to both Hurricanes Wilma's and Irma's impacts, the recurrent occurrence of these storms allowed us to perform a comprehensive evaluation of mangrove root process responses and long-term vegetation trajectories post-disturbance. To our knowledge, this research represents the first comparison of pre-and post-hurricane influences on long-term root dynamics of neotropical mangroves in carbonate settings. We focused on the Shark River estuary (SRE) located in the southwestern region of the FCE (Figure 1). SRE is considered one of the largest mangrove-dominated estuaries in the southwest coast of Florida and is characterized by strong soil P fertility and hydroperiod gradients, resulting in distinct forest structure and productivity patterns along the estuarine gradient (Castañeda-Moya et al., 2011, 2013;Chen & Twilley, 1999b).

Hurricanes Wilma and Irma, Category 3 storms, made landfall on the southwest Florida coast on October 24, 2005, and on September 10, 2017, respectively. Maximum sustained hurricane-force winds over the FCE were estimated at 105 km h -1 (65 mph, Wilma) and 180 to 193 km h -1 (112-120 mph, Irma) during landfall (Castañeda-Moya et al., 2010, 2020). The storm surges produced water levels up to 2-3 m above ground level along the southwest coast of Florida (Pasch et al., 2006;NHC, 2018), and up to 1-4 m above the mangrove forest soil surface across the FCE (Castañeda-Moya et al., 2010, 2020). Wilma's and Irma's winds caused a large-scale physical damage to forest structure (i.e., defoliation, tree snapping, and uprooting), and deposition of P-rich allochthonous mineral sediments associated with the storm surge across the FCE mangroves (Castañeda-Moya et al., 2010, 2020;Chavez et al., 2023;Lagomasino et al., 2021). In contrast to these two storms, Hurricane Katrina, a Category 1 storm, also made landfall in south Florida on August 25, 2005, but only caused defoliation across the FCE mangroves (Danielson et al., 2017). Since both Wilma and Irma caused substantial damage to mangrove forest structure, we expected to observe more pronounced changes in root responses similar to aboveground compartments. Therefore, the patterns in root responses we document in this study are interpreted specifically in relation to the impacts of Wilma and Irma.

These two hurricanes provided a unique opportunity to further quantify the long-term positive and negative effects of hurricanes and their legacies on mangrove root processes in the FCE. We addressed the following questions: (1) What are the relative changes in mangrove root biomass C stocks, production, turnover, and decomposition along the P fertility gradient in SRE post-hurricane periods? (2) How does root size class distribution in biomass and production pools change post-hurricanes? (3) What are the relative changes in root nutrient content post-hurricanes? (4) Are the post-hurricane spatiotemporal patterns of root processes consistent with pre-hurricane patterns along the P fertility gradient of SRE? We hypothesized that (1) root biomass C stocks would be higher in post-hurricane periods than in pre-hurricane periods, with higher fine root biomass allocation across all sites to maximize plant nutrient uptake and facilitate canopy re-foliation and recovery; (2) fine root turnover would be higher post-hurricane periods, due to increased soil fertility resulting from P-rich sediment inputs into mangrove soils; (3) root nutrient (N, P) content would increase post-hurricane at all sites, especially at downstream sites because of greater nutrient uptake as a result of P inputs; and (4) root decomposition rates would be higher after Hurricane Irma, particularly at downstream sites compared to the upstream site where P deposition was not observed.

MATERIALS AND METHODS

Study site

This study was conducted in the SRE located in the southwestern region of Everglades National Park (ENP; Figure 1). Riverine mangroves along SRE are dominated by mixed species of Rhizophora mangle, Avicennia germinans, Laguncularia racemosa, and Conocarpus erectus; the latter species is limited to upstream locations of the estuary in the mangrove-marsh ecotone boundary where salinity does not exceed 10 ppt (Castañeda-Moya et al., 2013;Chen & Twilley, 1999b;Zhao et al., 2020). Three mangrove sites (SRS-4, SRS-5, and SRS-6) were established along SRE, as part of the Florida Coastal Everglades Long Term Ecological Research (FCE-LTER) program since December 2000 (http://fcelter.fiu.edu/; Childers, 2006). SRS-6 is located downstream approximately 4.1 km from the mouth of the estuary, while SRS-5 (9.9 km) and SRS-4 (18.2 km) are located in the midstream and upstream regions, respectively (Figure 1). A new permanent site (SRS-7) was established at the mouth of the estuary after Hurricane Irma to assess mangrove recovery since this area experienced severe forest damage from both hurricanes (Castañeda-Moya et al., 2020). Hereafter, mangrove sites are referred to as upstream (SRS-4), midstream (SRS-5), downstream (SRS-6), and mouth (SRS-7).

Two permanent plots (20 × 20 m) were established at each site (in December 2000) approximately 30-50 m from the mangrove edge to monitor forest structural and functional attributes, soil biogeochemical properties, and hydroperiod (i.e., frequency and duration of inundation and water depth). There are distinct environmental gradients among Shark River sites, where soil total P concentrations (0.05-0.22 mg cm -3 ), porewater salinity (4-27 ppt), and hydroperiod (3965-5592 h year -1 ; 217-395 tides year -1 ) increase from upstream to downstream sites (Castañeda-Moya et al., 2013, 2020;Chen & Twilley, 1999b). These well-defined gradients influence vegetation patterns along the estuary, with higher forest development and productivity at near-coast mangroves (i.e., downstream) compared to mid-and upstream regions. Mangrove tree height (6-18 m) and basal area (20-40 m 2 ha -1 ) decrease from downstream to upstream sites, whereas tree density (2838-7746 trees ha -1 ) increases with distance inland from the mouth of the estuary (Castañeda-Moya et al., 2013;Chen & Twilley, 1999a). Because of strong differences in environmental gradients along SRE, L. racemosa is the dominant species downstream, while R. mangle dominates at up-and midstream sites (Castañeda-Moya et al., 2013;Danielson et al., 2017;Rivera-Monroy et al., 2019). Mangroves in SRE are tide-dominated, although the upstream site is also influenced by runoff and groundwater, particularly during the wet season (Castañeda-Moya et al., 2013;Chen & Twilley, 1999b;Saha et al., 2011). Tides in ENP are semidiurnal with a mean tidal amplitude of 1.1 m in the southwestern region and from negligible to 0.5 m in the southeastern region and Florida Bay (Wanless et al., 1994). South Florida's climate is subtropical savanna with distinct dry (December-May) and wet (June-November) seasons. The region is influenced by the passage of winter cold fronts (early dry season) and tropical cyclones that usually occur between June and October (Duever et al., 1994).

Root biomass carbon stocks

Previously published root biomass estimates (Castañeda-Moya et al., 2011) for the pre-hurricane period (root core collection in December 2000 and May 2003) in the shallow (0-45 cm depth) root zone were used as a baseline to assess changes in root biomass after Hurricanes Wilma (October 2005) and Irma (September 2017) impacted all four Shark River mangrove sites. These published estimates represented the overall estimate (mean ± 1 SE) of both root collection periods (data were pooled for statistical analysis; no significant differences between periods: F 5,132 = 1.95, p = 0.1) (Castañeda-Moya et al., 2011), and thus captured the spatiotemporal variability in root biomass across mangrove sites during the pre-hurricane period. The post-hurricane sampling periods included post-Wilma (May 2012), immediate-post-Irma (March 2018), and post-Irma (March 2023). During each post-hurricane period, root cores were collected at all sites to estimate root biomass in the shallow root zone using the same sampling protocol as in the pre-hurricane period (Castañeda-Moya et al., 2011). The estuary's mouth site was not sampled during the pre-hurricane period or the post-Wilma period. Briefly, two sampling points (replicates) were established in opposite corners of the two permanent plots at each site and duplicate root cores (0-45 cm depth) were collected from each sampling point using a PVC coring device (10.2 cm diameter × 45 cm length). Extracted root cores were stored separately in Ziploc bags at 4 C and brought to the laboratory for further analyses. Root core samples were rinsed with water through a 1-mm synthetic mesh screen to remove soil particles. Live roots were separated manually based on their buoyancy, turgor, and color (Castañeda-Moya et al., 2011;Cormier et al., 2015;Medina-Calder on et al., 2021). Live roots were further sorted into three size diameter classes: fine (<2 mm), small (2-5 mm), and coarse (>5-20 mm).

Roots greater than 20 mm in diameter were not included in this study due to sampling limitations (i.e., core area). All root samples were oven-dried at 60 C to a constant mass and weighed. Root biomass estimates are expressed in grams per square meter by scaling up the core area. We applied a root tissue carbon content of 44% (derived from analysis of subsamples) to convert root dry biomass (in grams per square meter) to C stocks (in megagrams of carbon per hectare) in the shallow root zone.

Root biomass samples were ground by size classes (all post-hurricane periods) using a Wiley Mill to pass through a 40-μm-mesh screen and analyzed for total carbon (TC), nitrogen (TN), and phosphorus (TP) contents. It was not possible to determine C and nutrient content of root samples by size classes during the pre-hurricane period due to small sample volumes at all sites (Castañeda-Moya et al., 2011). Root TC and TN contents were determined with an ECS 4010 elemental analyzer (pre-hurricane and post-Wilma samples; Costech Analytical Technologies, Valencia, CA) and a Carlo-Erba NA-1500 elemental analyzer (immediate-post-Irma and post-Irma samples; Fisons Instruments, Danvers, MA, USA). Total P was extracted using an acid-digest (HCl) extraction after combustion in a furnace at 550 C (Aspila et al., 1976), and concentrations of soluble reactive P were determined by colorimetric analysis (Methods 365.4 and 365.3, US EPA 1983). Root total C and nutrient contents were expressed in milligrams per gram.

Root production

Root production was estimated using the ingrowth core technique (Vogt et al., 1998) during pre-hurricane and post-hurricane periods. Similar to root biomass, previously published root production estimates (Castañeda-Moya et al., 2011) for the pre-hurricane period in the shallow (0-45 cm depth) root zone were used as a baseline to assess changes in annual root production during post-hurricane periods, including post-Wilma (May 2013), immediate-post-Irma (March 2019), and post-Irma (March 2024). During the pre-hurricane period, ingrowth cores were retrieved at 1-year (May 2004) and ~3-year (February 2006) intervals from all sites to estimate root production. These published estimates represented the average of both root production cycles (i.e., 1 year and ~3 years; no significant differences between them) and thus captured the spatiotemporal variability in annual root production across mangrove sites during the pre-hurricane period. Ingrowth cores (10.2 cm diameter × 45 cm length) were made of synthetic material (3-mm mesh) and filled with commercial sphagnum peat moss. This material has similar soil properties (i.e., bulk density, organic matter content, total C and N) as mangrove peat in our study sites as previously reported by Castañeda-Moya et al. (2011). At each site, eight ingrowth cores were deployed vertically into the same holes from which root biomass cores were previously collected during all sampling periods. Ingrowth cores were retrieved 1 year later during each post-hurricane period in all sites. Root growth within each ingrowth core following 1 year of incubation was used to estimate annual root productivity. After collection, ingrowth cores were processed individually following the same protocol as described for root biomass estimation. Root production is expressed in units of C (in megagrams of carbon per hectare per year) using the same 44% C conversion factor. Annual fine root turnover ( -per year) was calculated for each period as the ratio of root production to biomass (Eissenstat & Yanai, 1997). To maintain scientific consistency and integrity in our pre-and post-hurricane comparisons, we rely on the baseline root biomass and production estimates as originally defined and published by Castañeda-Moya et al. (2011).

Root decomposition rates post-Irma

Root decomposition experiments were initiated in January 2018 (i.e., immediate-post-Irma) following Hurricane Irma at each of the four mangrove sites (no experiments were conducted after Hurricane Wilma). Root decomposition rates were compared among sites and with soil depth using the same mesh bag (1-mm 2 ; 10 × 40 cm) technique used in a previous study at the same sites (except for the mouth site) during the pre-hurricane period (Poret et al., 2007). At each Shark River site, live belowground roots were initially extracted by coring, rinsed with water to remove soil particles, air-dried for 24 h, and then placed inside bags. Mesh bags were divided into two 20-cm-depth sections, each one containing ~5-6 g of fresh root material from each site with an equal mixture of three root size classes: 1-4, >4-8, and >8-12 mm. At each site, a total of 18 root decomposition bags were randomly buried in the soil within two sampling points (nine bags per point) that were established adjacent to the permanent plots approximately 40-50 m from the water edge; sampling points were treated as experimental units.

Triplicate bags were retrieved as sub-replicates from each sampling point at all sites after 190, 358, and 541 days of incubation, stored separately in plastic bags at 4 C, and brought to the laboratory for further analyses. Root material inside each bag (i.e., 0-20 and >20-40 cm depths) was removed, gently washed to remove sediment particles, oven-dried at 60 C to a constant mass, and weighed. Root samples in the >20-40-cm-depth section from the last incubation period (541 days) at the upstream site were not included in the analysis due to logistic problems. At the beginning of the experiment, additional triplicate root bags from each site were saved and oven-dried at 60 C to a constant mass to determine the air-dried to oven-dried mass conversion (i.e., dry:wet ratio) of the initial root dry masses for all deployed decomposition bags. For each site, the remaining root material from each incubation period was combined by depth at each sampling point, considered as replicates, and ground with a Wiley Mill to pass through a 40-μm-mesh screen. Samples were then analyzed for TC, TN, and TP contents. The initial root C and nutrient contents were determined on the triplicate samples saved at the beginning of the experiment for all sites.

Decay rates of mangrove roots were calculated for all sites and depths during the immediate post-Irma period using the negative exponential model, M t = M 0 e -kt , where M t = the percentage of dry root mass remaining at time t, M 0 = original dry root mass, and k = decay constant (Nye, 1961;Olson, 1963). It was not possible to statistically assess differences in root decay rates within sites between both pre-hurricane and immediate post-Irma sampling periods due to the unavailability of the raw data from the pre-hurricane study (Poret et al., 2007).

Statistical analyses

All statistical analyses were performed with JMP (Version 18.2, SAS Institute., Cary, NC, USA). Variation in total (sum of all size classes) root biomass C stocks, total root production, fine root biomass C, fine root production, and fine root turnover rates was tested separately for differences among sites (upstream, midstream, and downstream) and sampling periods (pre-hurricane, post-Wilma, immediate-post-Irma, and post-Irma) using a two-way ANOVA. Data from the mouth site (SRS-7) were excluded from these analyses due to missing pre-hurricane and post-Wilma measurements. Root N and P contents of biomass samples were analyzed independently with a two-way ANOVA to determine differences among sites and sampling periods. Differences in root nutrients were also assessed among sites and root size classes using a separate two-way ANOVA. We used a split-plot repeated measures ANOVA to test for differences in decay rates (k d ) of decomposing root material among sites (upstream, midstream, downstream, and mouth), depths (0-20 and >20-40 cm), and incubation periods (190, 358, and 541 days), with site considered as the main plot and incubation period and depth as the subplot. We used a two-way ANOVA to test for differences in C and nutrient content of decomposing roots among sites and incubation periods.

All main effects and their interactions were considered fixed for all analyses. Sampling points were nested within each site, considered as random effects, and treated as experimental units. Sampling units (i.e., root cores, ingrowth cores, and root decomposition bags) were also nested within each site and considered random effects. The Kenward-Roger procedure was used to adjust the degrees of freedom of the F test statistics in all analyses (Kenward & Roger, 1997). Pairwise comparisons were performed using Fisher's least significant differences (LSD) test when significant differences (p < 0.05) were observed within a main effect or interaction. Variables were transformed (i.e., ln(x + 1)) when required prior to analysis to meet ANOVA assumptions (i.e., normality and homoscedasticity). Unless otherwise stated, data presented are means (±1 SE) of untransformed data.

RESULTS

Root biomass carbon stocks

Total (all root classes; 0-45 cm depth) root biomass C stocks varied among mangrove sites (upstream, midstream, downstream) and sampling periods (pre-hurricane, post-Wilma, immediate-post-Irma, post-Irma), with a notable site × period interaction indicating that the variation in root biomass C among periods was dependent on site differences (Table 1; Figure 2). Root biomass C stocks were higher during the immediate-post-Irma (15.4 ± 2.0 Mg C ha -1 ) and post-Irma (19.7 ± 2.7 Mg C ha -1 ) periods compared to pre-hurricane (10.3 ± 1.1 Mg C ha -1 ) and post-Wilma (8.7 ± 0.8 Mg C ha -1 ). Overall, root biomass C estimates were highest midstream (17.8 ± 1.9 Mg C ha -1 ) and lowest downstream (7.9 ± 0.8 Mg C ha -1 ) across all periods. At the upstream and midstream sites, root C stocks were consistently higher during both the immediate-post-Irma and post-Irma periods. In contrast, root estimates at the downstream site were similar across all periods. Root biomass C stocks at the estuary mouth remained low during both immediate-post-Irma and post-Irma periods (range: 8.3-8.7 Mg C ha -1 ) and were comparable to the downstream site.

The distribution of biomass C stocks among root size classes also differed among periods at all sites (Figure 3). On average, during the pre-hurricane period, fine and small root size classes contributed 12%-25% of the total root C across sites, whereas coarse roots accounted for 52%-69% of the total C stocks. In contrast, the contribution of root size classes shifted considerably during the post-hurricane periods, with fine roots accounting for ~32% (post-Wilma) to 66% (immediate-post-Irma and post-Irma) of the total root C stocks across sites. Small and coarse roots only contributed on average 12% and 33% of the total C across all sites and post-hurricane periods, respectively. Similarly, at the mouth site, the contribution of fine roots to the total root C stocks was considerably higher during both the immediate-post-Irma (75%) and post-Irma (35%) periods compared to that of small and coarse roots.

Fine root C stocks also varied among mangrove sites and among periods and followed the same trend as total root biomass C stocks (Table 1; Figure 4). Fine root C stocks were 1.2-1.9 times higher in the post-Wilma period relative to the pre-hurricane period at all sites, and 6-9 times higher during immediate-post-Irma and post-Irma periods. Overall, fine root C stocks ranged from 1.1 ± 0.2 Mg C ha -1 (downstream, pre-hurricane) to 25.7 ± 3.5 Mg C ha -1 (midstream, post-Irma) across all sites and periods. On average across periods, fine root C stocks were higher at upstream (6.6 ± 1.0 Mg C ha -1 ) and midstream (7.7 ± 1.5 Mg C ha -1 ) sites than at the downstream site (1.9 ± 0.2 Mg C ha -1 ). At the mouth site, fine root C stocks ranged from 3.1 ± 0.6 Mg C ha -1 (post-Irma) to 6.2 ± 0.6 Mg C ha -1 (immediate-post-Irma) across periods, and estimates were comparable to the downstream site.

Root production and turnover rates

Total root production varied among sites, with higher estimates at the midstream site (2.06 ± 0.23 Mg C ha -1 year -1 ) compared to the upstream (1.07 ± 0.10 Mg C ha -1 year -1 ) and downstream (1.31 ± 0.17 Mg C ha -1 year -1 ) sites. In contrast, root production estimates did not vary among periods (Table 1; Figure 5). Overall, root production estimates ranged from 0.79 ± 0.10 Mg C ha -1 year -1 Source of variation df F p Root biomass C stocks Site 2, 66.1 29.3 *** Period 3, 59.7 7.9 *** Site × Period 6, 60.0 4.5 *** Fine root biomass C stocks Site 2, 48.5 90.7 *** Period 3, 41.2 161.2 *** Site × Period 6, 41.4 16.2 *** Root productivity Site 2, 42.7 7.6 ** Period 3, 41.3 1.0 ns Site × Period 6, 41.4 0.6 ns Fine root productivity Site 2, 51.2 3.9 * Period 3, 48.7 6.9 *** Site × Period 6, 48.8 1.2 ns Fine root turnover Site 2, 20.0 14.2 *** Period 3, 25.3 45.9 *** Site × Period 6, 26.4 2.2 ns Root total N Site 2, 36.0 16.9 *** Period 3, 36.0 47.3 *** Site × Period 6, 36.0 9.3 ns Root total P Site 2, 36.0 216.4 *** Period 3, 36.0 9.8 ns Site × Period 6, 36.0 11.8 *** Root decay rates Site 3, 3.9 20.2 * Depth 1, 57.3 1.0 ns Site × Depth 3, 57.1 1.4 ns Incubation period 2, 8.4 22.3 *** Site × Incubation period 6, 8.4 2.3 ns Depth × Incubation period 2, 57.3 0.8 ns Site × Depth × Incubation period 6, 56.9 1.6 ns Abbreviation: ns, not significant. *p < 0.05; **p < 0.01; ***p < 0.001.

(upstream, post-Wilma) to 2.60 ± 0.61 Mg C ha -1 year -1 (midstream, pre-hurricane) across all sites and periods. At the estuary mouth site, total root production in the immediate-post-Irma and post-Irma periods was similar to that of the downstream site. Unlike total root production, fine root production varied among sampling periods (Table 1; Figure 4), with higher estimates during the pre-hurricane period (0.63 ± 0.10 Mg C ha -1 year -1 ) compared to lower estimates in the post-Wilma (0.44 ± 0.05 Mg C ha -1 year -1 ), immediate-post-Irma (0.44 ± 0.04 Mg C ha -1 year -1 ), and post-Irma (0.34 ± 0.03 Mg C ha -1 year -1 ) periods. Overall, fine root productivity ranged from 0.15 ± 0.03 Mg C ha -1 year -1 (mouth, post-Irma) to 0.76 ± 0.07 Mg C ha -1 year -1 (upstream, pre-hurricane) across all sites and periods. There was no site × period interaction, nor an overall site effect on fine root production. The distribution of root size classes between pre-hurricane and post-hurricane periods was rather similar at each mangrove site, with fine roots accounting for 20%-50% of the total root production across all sites and periods (Figure 6). Coarse roots had the largest contribution (22%-75%) to the total root production, whereas the small root size class accounted for 5%-38% of the total production across all sites and periods. Noteworthy, at the estuary mouth site, the contribution of small roots increased during the post-Wilma period (32%) relative to the immediate-post-Irma (Figure 6).

Fine root turnover rates declined after hurricane disturbances. The highest root turnover was observed in the pre-hurricane period (0.43 ± 0.05 year -1 ) at all sites, whereas rates were lowest during the immediate-post-Irma (0.05 ± 0.01 year -1 ) and post-Irma (0.07 ± 0.03 year -1 ) periods (Table 1; Figure 7). Overall, fine root turnover rates were higher at the downstream site (0.28 ± 0.06 year -1 ) compared to the upstream (0.11 ± 0.03 year -1 ) and midstream (0.17 ± 0.04 year -1 ) sites and ranged from 0.016 ± 0.002 year -1 (midstream, post-Irma) to 0.599 ± 0.073 year -1 (downstream, pre-hurricane) across all sites and periods. At the mouth site, fine root turnover in the immediatepost-Irma and post-Irma periods was 2-3 times lower than that at the downstream Shark River site.

Root nutrient content

Root total N differed among sites (Table 1), with higher values downstream (5.8 ± 0.3 mg g -1 ) compared to upstream and midstream sites (range: 4.9 ± 0.3 to 5.0 ± 0.3 mg g -1 ) across all periods (Figure 8). Overall, root total N was higher in the post-Irma period (6.1 ± 0.2 mg g -1 ) and lowest in the immediate-post-Irma period (4.2 ± 0.2 mg g -1 ). There was no interaction between sites and periods. At the mouth site, root total N during both the immediate-post-Irma (5.1 ± 0.7 mg g -1 ) and post-Irma period (5.7 ± 0.5 mg g -1 ) was similar to values at the downstream site. Among root size classes, fine roots consistently had the highest N content (8.4 ± 0.8 mg g -1 ) relative to small (4.2 ± 0.3 mg g -1 ) and coarse roots (3.2 ± 0.3 mg g -1 ) across all sites (F 2,80 = 143.2, p < 0.001).

Root total P varied among sites and among periods (Table 1) and consistently increased from the upstream site (0.17 ± 0.01 mg g -1 ) to the downstream (0.38 ± 0.01 mg g -1 ), and mouth (0.42 ± 0.03 mg g -1 ) sites (Figure 8), following the observed soil P fertility gradient along the estuary. Overall, root total P was highest in the post-Irma period (0.35 ± 0.03 mg g -1 ), and consistently lower in the other three periods (range: 0.25 ± 0.03 to 0.27 ± 0.03 mg g -1 ). At the mouth site, root total P had similar values and trends during both the immediate-post-Irma and post-Irma periods compared to the downstream site. Among root size classes, fine roots consistently had the highest P content (0.36 ± 0.03 mg g -1 ) compared to small (0.31 ± 0.02 mg g -1 ) and coarse roots (0.24 ± 0.02 mg g -1 ) across all sites (F 2,79 = 48.7, p < 0.001).

Root decomposition rates immediate-post-Irma

Root mass loss over ~18 months (541 days) of incubation followed a generally similar pattern at both depths and across sites (Figure 9). In the top soil section (0-20 cm depth), root mass loss averaged 28% at the upstream site, 40% at downstream, and 63%-67% at midstream and mouth sites. In the bottom soil section (>20-40 cm depth), mass loss was similar: 31% upstream, 39% downstream, and 62%-65% at midstream and mouth sites. Overall, between 25% (upstream) and 63% (midstream) of root mass loss occurred during the first ~6 months of the experiment (i.e., 190 days) across sites and depths. After 6 months of incubation, root mass loss remained relatively similar between both depths at each site, ranging from ~30% (midstream) to ~75% (upstream) across sites.

Mean decay rates of root material did not differ between soil depths (0-20 cm: 0.0024 ± 0.0001 day -1 ; >20-40 cm: 0.0023 ± 0.0001 day -1 ) but varied overall across sites (Table 1; Figure 10). The highest root decay rates were observed at midstream and mouth (0.0034 ± 0.0002 and 0.0033 ± 0.0002 day -1 , respectively) sites and the lowest rates at upstream (0.0011 ± 0.0002 day -1 ) and downstream (0.0016 ± 0.0002 day -1 ) sites. There was no interaction between sites and depths indicating that the variation in root decay rates between depths is independent of site differences.

Root decay rates also varied among incubation days (Table 1; Figure 10), with faster decomposition rates at 190 days (0.0032 ± 0.0002 day -1 ) compared with 358 days (0.0022 ± 0.0002 day -1 ) and 541 days of incubation (0.0017 ± 0.0002 day -1 ). After 190 days, root decay rates ranged from 0.0018 ± 0.0003 day -1 (upstream) to 0.0046 ± 0.0003 day -1 (midstream), from 0.0010 ± 0.0001 day -1 (upstream) to 0.0033 ± 0.0002 day -1 after 358 days, and from 0.0004 ± 0.0001 day -1 (upstream) to 0.0027 ± 0.0002 day -1 (mouth) after 541 days of incubation across all sites. There was no interaction between sites and incubation days suggesting that the variation in root decay rates among incubation days is independent of site differences.

Nutrient content of decomposing roots

Changes in C and nutrient content and ratios of decomposing root material among mangrove sites and incubation days (0, 190, 358, 541 days) were evident throughout the study (Tables 2 and 3). Root C content decreased along the estuary gradient, from 461.5 ± 6.5 mg g -1 upstream to 414.6 ± 6.8 mg g -1 at the mouth. Root C content was higher at the beginning of the experiment (0 days, initial) and decreased after 190 and 541 days of incubation.

Root N varied among sites, with higher values at downstream and mouth sites (range: 6.9-7.0 mg g -1 ) compared to midstream and upstream sites (range: 5.6-6.1 mg g -1 ; Tables 2 and 3). N content in decomposing roots increased over time and ranged from 6.4 ± 0.4 mg g -1 by 190 days to 7.6 ± 0.3 mg g -1 by 541 days of incubation, compared to 4.1 ± 0.2 mg g -1 at the beginning of the experiment.

Root P content showed a strong gradient along the estuary, consistently increasing from the upstream site (0.19 ± 0.01 mg g -1 ) to the midstream (0.37 ± 0.02 mg g -1 ), downstream (0.46 ± 0.03 mg g -1 ), and mouth (0.52 ± 0.05 mg g -1 ) sites (Tables 2 and 3). Root P content at the beginning of the experiment (0.27 ± 0.02 mg g -1 ) was 1.6 times lower compared to values at 190, 358, and 541 (range: 0.40-0.44 mg g -1 ) days of incubation.

Root C:N ratios were 1.6-2 times lower after 190-541 incubation days compared with initial values (130.8 ± 5.6), indicating N enrichment of decomposing root material. Root C:N ratios were consistently lower at the downstream and mouth sites compared to midstream and upstream sites (Tables 2 and 3). N:P ratios of decomposing roots were higher upstream (69.3 ± 3.9 mg g -1 ) relative to the other three sites (range: 30.5--34.0 mg g -1 ), indicating strong P limitation. Overall, N:P ratios of decomposing roots were lower at the beginning of the experiment (34.8) and increased (range: 41.3-47.5) after 190-541 days of incubation (Tables 2 and 3). Collectively, these changes in root tissue nutrients indicate that decomposing roots became enriched in N and P relative to C over time (C content declined, and N and P increased), leading to much lower C:N and higher N:P ratios by the end of the incubation period.

DISCUSSION

Spatial variation in root dynamics before and after Hurricanes Wilma and Irma

Understanding the spatiotemporal patterns of belowground processes in mangrove forests is particularly important given the increasing frequency and/or intensity of hurricanes in coastal regions worldwide (Emanuel, 2021; Kossin et al., 2020;Kuhn et al., 2021). Hurricanes Wilma and Irma impacted mangrove forests in the FCE altering forest structure and function due to extensive canopy defoliation and tree mortality (Castañeda-Moya et al., 2020;Chavez et al., 2023;Danielson et al., 2017;Lagomasino et al., 2021;Rivera-Monroy et al., 2019;Smith et al., 2009). Subsequent studies have also documented mangrove canopy regrowth and recovery post-Wilma and post-Irma (Danielson et al., 2017;Rivera-Monroy et al., 2019;Xiong et al., 2022) and underscored the positive effects of hurricane-induced P-rich sediment deposition on soil fertility and plant-soil feedback interactions post-disturbance (Castañeda-Moya et al., 2010, 2020). These post-Wilma and post-Irma mineral deposits particularly in near-coast mangroves within the FCE, with P concentrations twice the average surface (top 10 cm) soil nutrient P density (0.19 ± 0.02 mg cm -3 ), create hurricane legacies that influence the long-term ecological properties of mangroves and facilitate rapid forest recovery, stimulation of peat development, and resilience to SLR (Castañeda-Moya et al., 2010, 2020;Feher et al., 2019).

There is a notable lack of studies quantifying how hurricanes affect mangrove belowground processes (i.e., F I G U R E 1 0 Variation in root decay rates with (A) soil depth and (B) after three (190, 358, and 541 days) incubation periods in mangrove sites along Shark River estuary during the immediate-post-Irma period (means ± 1 SE). In panel (A), different lowercase letters indicate significant differences ( p < 0.05) among sites and depths. Different uppercase letters indicate significant differences in decay rates among sites. The blue solid line represents root decay rates at all sites during the pre-hurricane period as reported by Poret et al. (2007). In panel (B), different lowercase letters indicate significant differences in root decay rates among incubation days within each site. biomass, production, and decomposition). Belowground root responses following hurricane impacts are less evident and understood compared to the visible aboveground effects on mangrove forests (Lugo, 2008). Without robust, long-term root datasets covering pre-and post-hurricane periods, it remains challenging to assess the direct effects of hurricanes and their legacies on root dynamics. The scarcity of such comparative studies limits our ability to identify thresholds of recovery and trajectories of response, and to predict the resilience of mangrove forests post-disturbance.

We found that root biomass allocation was higher in all post-Irma periods compared to pre-hurricane conditions at all sites, except at the downstream site, where root biomass C stocks did not vary between pre-and post-hurricane periods. Overall, in our study, total root C stocks increased along the freshwater-estuarine gradient, with higher estimates at up-and midstream sites compared to downstream and mouth sites (Figure 2). Fine root biomass increased at all sites in every post-hurricane period, particularly at upstream and midstream sites where root C estimates were on average 7-13 times higher during the immediate-post-Irma and post-Irma periods compared to the pre-hurricane period (Figure 4). Root biomass and production are critical processes in mangrove wetlands that contribute to soil C storage and sequestration, and represent a significant adjustment to soil elevation relative to rising sea levels (Arnaud et al., 2023;Castañeda-Moya et al., 2011;Krauss et al., 2014;McKee et al., 2007). Furthermore, fine roots play an important role in the acquisition of water and nutrients in forest ecosystems (McCormack et al., 2015), whereas coarse roots function as storage and structural support (Eissenstat & Yanai, 1997). In our study, fine root biomass accounted for up to 32%-66% of the total biomass allocation in the post-hurricane periods (Figures 3 and 4), a significant increase in root C allocation relative to pre-hurricane conditions. These findings support the hypothesis that root biomass allocation is a significant T A B L E 2 Two-way analysis of variance (ANOVA) for carbon and nutrient content, and atomic C:N and N:P ratios of decomposing roots in mangrove sites along Shark River estuary.

Source

of variation df F p Root total C Site 3, 44 29.4 *** Days incubated 3, 44 12.1 *** Site × Days incubated 9, 44 8.6 *** Root total N Site 3, 44 5.2 ** Days incubated 3, 44 26.8 *** Site × Days incubated 9, 44 1.6 ns Root total P Site 3, 44 27.9 *** Days incubated 3, 44 7.1 *** Site × Days incubated 9, 44 1.0 ns Root C:N Site 3, 44 14.0 *** Days incubated 3, 44 67.4 *** Site × Days incubated 9, 44 1.6 ns Root N:P Site 3, 44 161.5 *** Days incubated 3, 44 12.9 *** Site × Days incubated 9, 44 8.0 *** Abbreviation: ns, not significant. *p < 0.05; **p < 0.01; ***p < 0.001. T A B L E 3 Variation in carbon, nutrients, and atomic C:N and N:P ratios of decomposing roots in mangrove sites along Shark River estuary in the Florida Everglades. Variable Site Days incubated Upstream Midstream Downstream Mouth 0 (initial) 190 358 541 (SRS-4) (SRS-5) (SRS-6) (SRS-7) Root total C (mg g -1 ) 461.5 a (6.5) 433.6 b (4.0) 425.9 b (5.2) 414.6 c (6.8) 447.7 a (2.2) 429.7 b (9.9) 441.3 a (5.9) 420.5 b (5.1) Root total N (mg g -1 ) 6.1 b (0.5) 5.6 b (0.3) 7.0 a (0.5) 6.9 a (0.5) 4.1 c (0.2) 7.6 a (0.3) 6.9 ab (0.3) 6.4 b (0.4) Root total P (mg g -1 ) 0.19 c (0.01) 0.37 b (0.02) 0.46 a (0.03) 0.52 a (0.05) 0.27 b (0.02) 0.40 a (0.03) 0.44 a (0.04) 0.41 a (0.05) Root C:N 97.6 a (8.4) 95.5 a (6.7) 77.1 b (6.6) 75.4 b (5.9) 130.8 a (5.6) 66.9 c (2.1) 76.4 b (3.3) 82.6 b (5.6) Root N:P 69.3 a (3.9) 34.0 b (1.6) 33.3 b (1.2) 30.5 b (1.5) 34.8 c (1.8) 47.5 a (5.4) 41.3 b (4.9) 41.6 b (4.8) Note: Means (±1 SE) followed by different letters across each row are significantly different (Fisher's least significant difference [LSD] post hoc test, p < 0.05).

contribution to soil C storage in mangrove forests (Arnaud et al., 2023;Chmura et al., 2003;Khan et al., 2007). This higher fine root biomass allocation, particularly after Hurricane Irma (Figure 3), likely reflects an adaptive response of mangrove species to cope with disturbances. It promotes recovery by increasing root:shoot ratios, enhancing nutrient uptake to support rapid canopy regrowth (Figure 11). Mangrove areas in Shark River with strong soil P limitation (i.e., N: P > 40; up-and midstream sites) (Castañeda-Moya et al., 2013) had the highest fine root biomass allocation post-hurricanes, suggesting an increase in root area to capture the limiting nutrient and support canopy recovery.

In our study, nutrient inputs from both hurricanes (Castañeda-Moya et al., 2010, 2020) appeared to facilitate rapid forest recovery across all sites. For example, fine root biomass allocation increased at the higher soil P fertility downstream and mouth sites (Castañeda-Moya et al., 2020), particularly following Irma. Indeed, following both Wilma and Irma, we observed rapid (~5 years) canopy re-foliation and return of litterfall productivity (Castañeda-Moya et al., 2025b;Danielson et al., 2017), along with renewed structural growth in Shark River mangroves (Rivera-Monroy et al., 2019, Xiong et al., 2022). Our results underscore the unique effect of these recurrent storms on mangrove ecological properties and highlight the strong phenotypic plasticity and resilience capacity of mangroves in response to hurricane impacts (Figure 11) (Castañeda-Moya et al., 2020;Krauss & Osland, 2020).

Sediment deposition, nutrient inputs, and erosion are important drivers of post-hurricane recovery and species responses in coastal wetlands. For instance, in Louisiana coastal marshes, Hurricane Katrina sediment deposits increased marsh root biomass accumulation up to tenfold from 2005 to 2007, indicating a storm-induced stimulation of root productivity (Day et al., 2013;McKee & Cherry, 2009). In the FCE, similar plant tissue responses have been described in downstream mangroves of Shark River post-Irma, where litter foliar total P concentrations increased by 1.5-2.5 times across all mangrove species in early 2018, 3 months post-Irma (Castañeda-Moya et al., 2020). The increase in foliar P in all species was explained by increases in soil porewater soluble reactive P (SRP) concentrations, particularly at downstream mangroves, where SRP in December 2017 (14.5 ± 0.5 μM) was seven times higher than the long-term average wet season value (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016). Similar responses in SRP concentrations, but with lower magnitude, were observed at the downstream mangrove site in Shark River following Hurricane Wilma's impact (Castañeda-Moya et al., 2025c). In contrast, studies in Texas following Hurricane Harvey reported an ~80% decrease in root biomass in mangroves/saltmarshes sites, where surface sediment P concentrations were reduced by 50% post-hurricane compared to pre-hurricane conditions. This reduction in soil nutrients was attributed to erosion of surficial sediments and soils followed by storm surge deposition of organicand nutrient-poor sediments, which increased nutrient limitation and stress in these ecosystems (Kuhn et al., 2021). Thus, the observed post-hurricane increases in root biomass and higher fine root C allocation at our mangrove sites likely represent a recovery response, paralleling the foliar regrowth and nutrient uptake patterns observed after these storms (Figure 11). These findings underscore the significant effects of P-rich sediment inputs on plant-soil feedbacks and interspecies differences in nutrient uptake in response to hurricane disturbances (Castañeda-Moya et al., 2020).

Mangrove root responses varied spatially in relation to hurricane-induced nutrient deposition patterns. Following Hurricanes Wilma and Irma, P deposition in the SRE (and other estuaries in southwestern Everglades) increased dramatically, particularly in near-coast mangroves, contributing up to 54% (Wilma) and 49% (Irma) to the soil nutrient pools (Castañeda-Moya et al., 2010, 2020). Notably, upstream mangrove areas that did not receive P-rich sediments post-hurricanes still showed a significant increase in root biomass. The likely drivers were the combination of severe defoliation (~40% canopy loss), chronic P limitation (N:P = 105), and increases in water levels and flooding duration at this site over the last decade (Castañeda-Moya et al., 2013;Castañeda-Moya, Solohin, & Kominoski, 2025;Danielson et al., 2017;Zhao et al., 2020), which together triggered compensatory root growth even without a direct nutrient subsidy. Thus, landscape variability in sediment deposition (both inland and laterally) coupled with the nature and chemical composition of mineral deposits (e.g., P-rich vs. nutrient-poor) across coastal wetlands results from the interaction between coastal geomorphology, local microtopography, hydrology, and storm characteristics. These interactions influence mangrove ecological and biogeochemical (e.g., nutrient cycling) processes across spatiotemporal scales, as observed in the FCE (Castañeda-Moya et al., 2020).

Mangrove root biomass and production are also influenced by the interplay of abiotic (i.e., soil pH, salinity, nutrient availability, and tidal regime) and biotic factors (species composition and forest age) that vary with local topography and hydrology (Adame et al., 2014(Adame et al., , 2017;;Krauss et al., 2014). For instance, Zhang et al. (2021) showed that elevated soil salinity can stimulate fine root production in mangroves, suggesting that this response is an adaptation to salinity stress, which adversely affects water uptake by fine roots (Adame et al., 2014;Ball, 1988;Zhang et al., 2021). Other studies have shown that mangrove sites with higher soil nutrients and lower salinity favor root production and turnover, leading to enhanced soil C accumulation (Xiong et al., 2016). Notably, in the FCE, previous studies have documented that the interaction of environmental gradients including soil P fertility, hydroperiod, and sulfide concentrations is the main factor controlling mangrove belowground C allocation patterns during the pre-hurricane period (Castañeda-Moya et al., 2011). Indeed, during the pre-hurricane period, fine root biomass was negatively correlated with soil P density and frequency of inundation, whereas fine root turnover decreased with increasing soil N:P ratios across FCE mangroves. These results suggest the significant effect of nutrient availability and hydroperiod on root dynamics, and the potential interaction with other soil processes such as decomposition (Castañeda-Moya et al., 2011). Moreover, fine root turnover increased along the soil P fertility gradient in SRE and was higher at the downstream site. Contrary to our hypothesis that hurricane P inputs would enhance fine root turnover, we found that turnover rates decreased post-hurricane at all sites (Figure 7). This indicates that fine root longevity increased after the storms (more biomass was retained below ground while root production remained the same or lower than before). These results contrast with the post-hurricane periods, where fine root turnover rates were similar within each site, mainly due to increased fine root biomass relative to production compared to pre-hurricane conditions. Environmental conditions coupled with mangrove species-specific adaptations and functional traits determine the degree of plasticity in mangrove belowground allocation (Krauss et al., 2014). Moreover, the unique effect of these recurrent storms in the FCE creates legacies of nutrient distribution across mangrove soils that interact with pre-existing environmental gradients-such as soil fertility, hydroperiod, and salinity (Castañeda-Moya et al., 2013;Twilley & Rivera-Monroy, 2009)-thus defining trajectories of ecosystem recovery post-disturbance (Figure 11) (Jentsch & White, 2019;Johnstone et al., 2016). Although salinity is typically considered a major stressor shaping mangrove structure and root dynamics (Ball, 1988;Castañeda-Moya et al., 2025a), salinity in FCE mangroves usually remains below 35 ppt due to groundwater mixing (Zhao et al., 2022). Therefore, salinity is not considered a significant driver of root C allocation in Everglades mangroves (Castañeda-Moya et al., 2011), in contrast to other mangrove ecotypes on karstic oceanic islands, where increased salinity promotes greater root biomass allocation to mitigate physiological stress (Medina-Calder on et al., 2021). Thus, our results in conjunction with other studies provide insights and are further evidence of the strong effect of hurricane disturbances on root processes in the FCE mangroves. Mangrove recovery patterns and species-specific responses in root functional processes following a hurricane define resilience thresholds. Forest recovery is contingent not only on the initial forest structure and plant evolutionary adaptations but also on storm properties (e.g., intensity and direction) and their legacies, all increasingly influenced by climate change (Patrick et al., 2022;Twilley et al., 1998).

Trends in root decomposition following hurricane Irma

Our findings are consistent with previous studies that showed overall low rates of decomposition in mangrove roots compared to other mangrove tissues (e.g., leaves) and roots of other wetland species (Arnaud et al., 2024;Mackey & Smail, 1996;Middleton, 2020;Middleton & McKee, 2001). We found no difference in root mass loss with soil depth, except for the midstream site, where roots in the top 20 cm of soils decomposed faster than those in the >20-40 cm depth (Figure 10). This pattern is consistent with other studies of mangrove root decomposition that reported no differences in mass loss with depth (Middleton & McKee, 2001;Poret et al., 2007). Interestingly, mean decay rates of roots sampled after Hurricane Irma (i.e., immediate-post-Irma period) at upand downstream sites were 1.5 times lower than pre-hurricane estimates reported by Poret et al. (2007) at the same sites. Conversely, at the midstream site, roots decayed at a greater rate, with the average decay constant (0.0035 day -1 ) exceeding pre-hurricane estimates (0.0028 day -1 ) reported by Poret et al. (2007) (Figure 10). Despite these site-specific changes, overall root decay rates remained low and showed a similar pattern along the estuary in both periods, suggesting interactions between nutrient availability and other localized factors, such as redox conditions, porewater chemistry, and hydroperiod.

Initial root chemistry can strongly affect decomposition rates, and fine and coarse roots differ in their decay responses because of variations in root nutrient content (Zhang & Wang, 2015). For example, initial root N and P influence coarse, but not fine root decomposition rates in terrestrial forests (Zhang & Wang, 2015). A combination of low tissue N and P content and a high C:N ratio is often associated with slower decomposition rates in mangroves (Ola & Lovelock, 2021), which likely contributed to the generally slow decay rates observed in our study. Moreover, increased water levels and flooding duration observed at the upstream site over the last decade (Anderson, Kominoski, Osburn, & Smith, 2024a;Castañeda-Moya, Solohin, & Kominoski, 2025) likely enhanced anaerobic conditions in mangrove soils, potentially slowing root decomposition rates, as observed during the post-Irma period. Our findings align with a recent study reporting faster decay rates of organic matter (e.g., roots, leaves) under reduced inundation and greater soil aeration (Arnaud et al., 2024). Similarly, Krauss et al. (2014) reported that prolonged flooding inhibits mangrove root decomposition. Furthermore, in our study, root C content decreased from upstream to downstream locations along the estuary. Root C:N ratios of decaying roots were also lower and within similar ranges at downstream and mouth sites compared to up-and midstream sites. C:N ratios of decaying roots were ~2 times lower after 190-541 days of incubation compared to initially higher C:N values, indicating N enrichment of decomposing root material (Table 2), and serving as a good predictor for decomposition rates as has been described in other studies (Huxham et al., 2010;Ola & Lovelock, 2021). The differences in root tissue nutrient (N, P) content observed in our study sites may contribute to variation in decomposition along Shark River mangroves.

The interactive effect of other abiotic and biotic factors can also influence mangrove root decay rates. Different mangrove species can have distinct rates of root decomposition due to root morphological properties and adaptations to environmental conditions. For example, root tissues of Avicennia sp. decompose faster than those of Rhizophora sp. due to higher N content (Middleton & McKee, 2001;Ola & Lovelock, 2021). Also, in a study of fine root decomposition in tropical Australian mangroves, Robertson and Alongi (2016) found that P limitation at sites dominated by Rhizophora likely resulted in greater fine root production and buildup of dead fine roots, which showed slow rates of decomposition. Moreover, they noted that high groundwater salinity and infrequent inundation can slow root decay rates. Litter carbon lability, based on cellulose and lignin content (not measured in our study; see Anderson, Kominoski, & Sah, 2024b;Kominoski et al., 2022), is another important driver of decomposition rates, in addition to nutrient availability. Indices of nonstructural carbohydrates and lignin:N ratios correlate more accurately with patterns of root decomposition in mangroves, with smaller ratios indicating greater decomposition rates (Huxham et al., 2010;Middleton & McKee, 2001). The typically slower decomposition of mangrove roots compared to leaves is often attributed to their higher lignin content, which makes roots more refractory and difficult to decompose under anaerobic conditions (Chen & Twilley, 1999a), similar to other wetland species (Solohin et al., 2020). For instance, in the FCE, mangrove roots show high levels of lignin content (19%-26% dry mass), yet root lignin:N ratios indicated no relationship with decomposition rates in mangrove sites (Poret et al., 2007). These results suggest that mangrove root decomposition in the Everglades is more strongly controlled by environmental conditions than by substrate quality (Poret et al., 2007). Indeed, higher decomposition rates were observed in tidally dominated riverine mangroves along Shark River (although no significant differences were detected among sites) compared to lower rates in permanently flooded scrub mangroves in Taylor River (Poret et al., 2007). In addition, they did not observe a clear relationship between soil P fertility and root decomposition rates. Further, the strong tidal regime in Shark River decreases the buildup of soil toxins such as sulfides and tannins (Castañeda-Moya et al., 2013) that may slow down decomposition rates, as observed in Taylor River (Poret et al., 2007;Robertson, 1988).

Variations in soil redox conditions can also interact with carbon lability and litter nutrient concentrations to enhance decomposition (Anderson, Kominoski, & Sah, 2024b;Kominoski et al., 2022). For example, higher water depths and sulfide accumulation under reduced conditions can decrease litter decomposition in wetlands (Kominoski et al., 2022). In contrast, the presence of labile dissolved organic matter and increased water nutrient availability, associated with saltwater intrusion, can enhance litter decomposition (Anderson, Kominoski, & Sah, 2024b;Stagg et al., 2017). Some studies have shown that crab burrows can create oxic soil conditions at the root surface, enhancing aerobic decomposition. This bioturbation increases the volume of oxidized Fe (III) in the sediment, neutralizing toxic sulfides in the soil (Gribsholt et al., 2003;Kristensen, 2008). Such conditions are particularly prevalent in SRE, where alternating wet-dry hydroperiod regimes and higher soil redox conditions due to tidal flushing (Castañeda-Moya et al., 2013) primarily control root decomposition rates rather than the chemical composition of the roots (i.e., substrate quality). Additionally, site-specific conditions such as species composition and soil bulk density can strongly influence mangrove root decay rates (Ola & Lovelock, 2021), which may explain the spatial trends observed along Shark River.

Effects of hurricane disturbances on ecosystem carbon storage and development

Hurricanes can affect long-term C storage and wetland development in different ways depending on geomorphic settings and plant-soil feedbacks. In mangrove forests, like other coastal wetlands, soil organic matter development is mainly driven by rates of primary productivity, especially root production, decomposition rates, and organic and inorganic matter retention, all of which control elevation gains relative to sea level (Cahoon et al., 2020;Charles et al., 2020). Wetlands with higher productivity generally support greater soil organic matter stocks and carbon storage (Castañeda-Moya et al., 2020;McKee et al., 2007;Rovai et al., 2018). However, the impact of storms on soil and root processes can vary significantly due to local differences in elevation, soil properties, and plant responses to stress conditions. For example, we found that although root biomass increased significantly post-hurricane, root production did not vary between pre-and post-hurricane periods nor along the soil P fertility gradient of the estuary. In addition, fine roots consistently had the highest N and P content at all sites, and root P was higher during the post-Irma period compared to pre-and other post-hurricane periods (Figure 8), suggesting enhanced nutrient pools of root tissue post-disturbance. The accumulation of thicker sediment layers, coupled with prolonged inundation, tends to reduce oxygen and slow down root decomposition in the soil, potentially leading to delayed mangrove mortality (Lagomasino et al., 2021;Radabaugh et al., 2019). Differences in soil bulk density and organic matter content can also influence net carbon gains or losses in mangrove soils post-hurricane (Castañeda-Moya et al., 2013;Ola & Lovelock, 2021;Ola et al., 2018). Although soil nutrients in mangrove soils increased in the FCE following Hurricanes Wilma and Irma (Castañeda-Moya et al., 2010, 2020), this region has also experienced rapid SLR over the past decade (Wdowinski et al., 2016). Consequently, higher water levels and increased flooding conditions can stress mangroves, potentially decreasing root turnover rates and root nutrient concentrations despite overall increases in root biomass stocks.

How disturbances interact with changing environmental conditions to influence post-disturbance recovery is a long-standing question in ecosystem ecology (Jentsch & White, 2019;Kominoski et al., 2020). Our study showed that mangrove root biomass and decomposition were generally elevated post-hurricane, while root turnover was lower. These findings suggest that while mangroves in the FCE are adapting to increases in storm-driven nutrient availability, higher root decomposition rates may not necessarily enhance net belowground root carbon storage. The balance between root production and decomposition is thus critical for the long-term stability of mangroves under climate change and rising sea levels. This is particularly significant to the karstic region of south Florida with little allochthonous inputs, where fine root production and the accumulation of refractory root fractions (i.e., lignin, cellulose) that resist decay, coupled with root necromass, are the primary processes controlling soil formation and C sequestration in mangrove ecotypes (Biswas et al., 2025;Chen & Twilley, 1999a). Our results underscore the complex interactions among environmental gradients, hurricane legacies, species responses, and mangrove aboveand belowground processes that influence recovery trajectories, C storage and sequestration, and soil formation (Figure 11). Further assessment of these complex mechanisms is needed to quantify the long-term interactions between natural positive (e.g., P fertilization) and negative (e.g., forest mortality, structural damage) effects of hurricanes on the long-term ecological properties of neotropical mangrove wetlands (Castañeda-Moya et al., 2020).

CONCLUSIONS

Our study provides important insights into how mangrove forests in the Florida Everglades adapt and respond to the impacts of hurricanes. We further offer a detailed assessment of changes in root biomass C stocks, production, size class distribution, nutrient content, and decomposition rates across mangrove sites in the SRE. By comparing data from the pre-hurricane period (2000)(2001)(2002)(2003)(2004) to the aftermath of Hurricane Wilma (2005) and Hurricane Irma (2017), our study characterized the long-term root dynamic responses of FCE mangrove forests to environmental stress. Increases in total root biomass C stocks post-hurricane periods were most notable at up-and midstream sites of the estuary characterized by strong soil P limitation. Moreover, the contribution of root size classes shifted considerably during post-hurricane periods, with fine roots accounting for 32%-66% of the total root C stocks across all sites, suggesting an increase in root area and nutrient uptake to help restore the lost canopy.

Although near-coast mangrove areas (downstream) had the highest impacts during both storm (e.g., >90% defoliation, mortality) (Castañeda-Moya et al., 2020;Danielson et al., 2017;Rivera-Monroy et al., 2019), total root biomass C stocks did not change between pre-and post-hurricane periods. This trend may be associated with the high soil fertility condition at this site coupled with the additional input of P-rich sediments from hurricanes that enhance the soil nutrient pools, thus decreasing root stimulation post-disturbance (Castañeda-Moya et al., 2011).

Further, anticipated increases in nutrient content and root turnover post-hurricanes were not uniformly observed across all sites, suggesting that local environmental factors-such as soil fertility and hydroperiodplay a crucial role in influencing root dynamics. Our findings also showed low rates of decomposition in mangrove roots, consistent with other studies. Although root decay rates immediate post-Irma were slightly higher at one site (midstream) compared to pre-hurricane values, overall decomposition rates remained slow in Shark River mangroves.

Collectively, these belowground responses (i.e., greater total root biomass, higher fine root C allocation and nutrient uptake, and prolonged root lifespan) indicate that mangrove forests have a strong capacity to mitigate hurricane disturbances by investing C allocation in their root systems. These root traits can enhance soil carbon sequestration and stability of mangrove forests in the face of future disturbances and SLR.

Such long-term studies are pivotal for improving our mechanistic models predicting mangrove wetland responses and trajectories of recovery from hurricane disturbances. Better understanding of these belowground processes will guide effective conservation and management strategies of coastal natural resources that not only protect mangrove forests but also leverage their natural adaptive capacity and resilience to mitigate the broader impacts of climate change and SLR.