Seasonally Intensified Mud Shrimp Bioturbation Hinders Seagrass Restoration

Abstract

Understanding how disturbances affect marine foundation species is critical for enhancing the success of coastal ecosystem restoration. Extreme bioturbation by burrowing animals is increasingly impacting coastal vegetated habitats worldwide, with the potential to undermine the persistence and resilience of key foundation species. However, the role of faunal disturbances in modulating restoration outcomes remains poorly understood. Here, we combine field surveys and manipulative field experiments to examine how mud shrimp (Upogebia major) bioturbation impacts vegetation dynamics and restoration outcomes for intertidal seagrass (Zostera japonica). Field surveys revealed pronounced seasonal variation in shrimp bioturbation intensity, with peak burrow densities occurring in fall (up to 400 burrows m⁻²; 289% higher than in spring). The intensified bioturbation was associated with significant declines in seagrass shoot cover, density, and biomass, with negative associations restricted to fall. To test whether seasonally intensified shrimp bioturbation impairs seagrass restoration, we conducted a 24-day field experiment transplanting seagrass patches of varying initial sizes (5–26 cm diameter) into plots representing three levels of shrimp burrow density observed during the fall peak: control (~9 burrows m⁻²), high (~280 burrows m⁻²), and extremely high (~455 burrows m⁻²). Compared to the control, high and extremely high burrow treatments exhibited accelerated patch losses. By day 24, vegetation was virtually eliminated in all shrimp treatments, but the rate of patch loss was significantly lower in larger patches. These results suggest that seasonal intensification of mud shrimp bioturbation has a potential to compromise intertidal seagrass restoration, while increasing planting scale offers a potential mitigation strategy. Restoration interventions should explicitly consider temporal patterns in faunal bioturbation and integration of positive interactions to improve long-term success of vegetation restoration in bioturbator-dominated coastal systems.

1. Introduction

Coastal vegetated ecosystems (e.g., seagrasses, kelps, salt marshes, and mangroves) provide a wide range of vital ecological functions and services, including shoreline stabilization, carbon sequestration, water quality enhancement, and support for biodiversity and ecosystem productivity [1,2]. However, these ecosystems are experiencing global declines due to intensifying anthropogenic pressures, including climate change, overfishing, pollution, and biological invasion [3,4]. In response, large-scale restoration efforts have intensified in the past decades, particularly under global initiatives such as the UN Decade on Ecosystem Restoration [5]. Recent global-scale meta-analyses support the effectiveness of coastal habitat restoration in reinstating ecosystem structure and essential services [6,7]. Nonetheless, restoration outcomes remain highly variable across sites and are frequently constrained by high implementation costs, limited post-restoration monitoring and management [6,8]. More critically, the lack of integration of abiotic and biotic drivers (e.g., physical stress, habitat quality, and species interactions) into restoration strategies has emerged as a central challenge to long-term restoration success [6,9].

Growing recognition of the complex role community processes play in ecosystem dynamics has prompted increased interest in integrating species interactions into coastal vegetation restoration [10,11,12,13]. This paradigm shift emphasizes not only facilitating positive interactions but also anticipating and managing negative interactions, which may directly or indirectly impede restoration success. Notably, escalating herbivory and bioturbation pressures have emerged as critical challenges for restoring vegetated ecosystems in both terrestrial and coastal environments [13,14,15,16]. A growing body of evidence indicates that excessive disturbance by large vertebrate herbivores and burrowing invertebrates is becoming increasingly widespread across global coastlines [17]. These patterns are likely driven by predator declines from overfishing and habitat loss [14], the expansion of livestock and waterfowl populations [18], and population recoveries of large herbivores through conservation initiatives [10]. Although moderate levels of grazing and bioturbation may enhance biodiversity and support biogeochemical processes [19,20], their drastic increases have been linked to vegetation loss, sediment erosion, and declines in carbon storage capacity. According to a recent global meta-analysis, these effects may be even more pronounced in restored systems, where herbivore or bioturbator populations are often unchecked due to disrupted trophic regulations [13].

Compared to direct suppression of restored vegetation by herbivores, the effects of mechanical disturbances caused by bioturbating animals have received considerably less attention [17,21]. Yet, existing studies suggest that extensive bioturbation, even in the absence of direct consumption, has a potential to undermine vegetation restoration. For example, excavation and burrowing by macroinvertebrates can hinder seedling establishment, physically damage plant tissues, and accelerate lateral erosion by reworking and destabilizing sediments [15,22,23]. The strength and direction of these effects may be density-dependent and modulated by ecological and environmental settings [24,25]. Due to the limited empirical studies and narrow taxonomic and geographic scope of existing data, however, our understanding of the generality and mechanisms by which bioturbation affects coastal plant restoration remains constrained. Addressing this gap is essential for developing robust, transferable guidelines for vegetation restoration in bioturbator-dominated environments.

In this study, we examined the impact of burrowing shrimps, a globally widespread group of coastal bioturbators, on intertidal seagrass restoration. Burrowing shrimps are detritus and particle-feeding decapod crustaceans within the infraorders Upogebiidae and Axiidea, well-known for their ability to modify sediment characteristics, biogeochemical cycling, and benthic communities [26,27,28,29]. Burrowing shrimps are commonly associated with seagrass ecosystems [28,30,31,32], yet their interactions with seagrasses, especially under restoration contexts, remain poorly understood. Previous works have suggested that antagonistic interactions may arise due to contrasting sediment-modifying traits: shrimp burrowing destabilizes sediments, increases turbidity, and may bury or uproot seagrass shoots [32], while seagrass rhizomes stabilize sediments, potentially reducing burrowing activity [33,34]. Nonetheless, how these interactions vary along abiotic and biotic gradients (e.g., season, sediment type, vegetation patch size or density) has received little attention [35].

The main objective of our study was to assess the functional relationship between mud shrimp bioturbation and early-stage stability of restored seagrass, as influenced by bioturbation intensity and transplant scale. Decades of research have provided compelling evidence that positive interactions among neighboring plants critically mediate resilience of restored and natural vegetated habitats [36,37,38], especially under elevated abiotic and biotic stress. For instance, in systems such as salt marsh and kelps, increasing patch size or the clumped configuration of transplants have shown to confer resistance to herbivores via associational defenses [39,40]. Similar mechanisms may also underlie resistance of seagrass to bioturbation, but this has not been explicitly tested in previous works. Higher conspecific density and larger patch size has a potential to provide stronger associational defense against mud shrimp bioturbation by physically interfering with shrimp burrowing activity and complex belowground network mitigating sediment reworking and vegetation losses.

We focused on the interaction between the intertidal dwarf seagrass Zostera japonica and the mud shrimp Upogebia major. U. major is a dominant filter-feeding bioturbator on tidal flats in Korea and Japan, reaching over 10 cm in length [26]. The species constructs deep Y-shaped burrows that may extend beyond 2 m and can attain very high densities (up to 952 burrows m⁻¹) [27,41]. Although U. major has been considered to mainly inhabit unvegetated mudflats, recent observations have reported its encroachment into vegetated habitats, including both Zostera marina and Z. japonica beds (Figure A1) [30,42]. Z. japonica, native to temperate and subtropical East Asia, has experienced rapid declines in recent decades due to intensifying climate change and human disturbances [43]. Accordingly, efforts to develop effective conservation and restoration interventions, spanning seed banking and germination/establishment studies, have expanded in recent years [44,45,46]. Regional interventions for restoring Zostera spp. typically account for abiotic conditions such as wave exposure, sediment type, and water depth, as well as existing human pressures from fisheries, aquaculture, and coastal development [44,47,48]. In contrast, the role of biotic interactions, such as bioturbation stress caused by burrowing shrimps, has been largely overlooked, despite its potential to compromise restoration success. Notably, this seasonal peak in bioturbation coincides with the preferred transplant window for seagrass due to enhanced transplant survival during fall [49]. The study also found that Upogebia bioturbation significantly reduced the persistence of small-scale transplants, but this was tested only with a single transplant size (~10 cm diameter). Building on this prior work, we aimed to address three core objectives: (1) to characterize seasonal variation in the intensity of Upogebia bioturbation across vegetated and unvegetated intertidal areas across multiple sites; (2) to quantify the relationship between bioturbation intensity and Z. japonica vegetation structure; and (3) to experimentally test how transplant patch size mediates the effect of mud shrimp bioturbation on seagrass persistence.

We predicted that fall-intensification of Upogebia bioturbation would be associated with significant declines in seagrass vegetation. Furthermore, we predicted that early-stage transplant persistence would decrease under higher burrow densities, but that larger transplant patches would exhibit greater resistance to bioturbation stress.

2. Materials and Methods

2.1. Study Sites

We conducted field surveys during spring and fall of 2024 across three natural intertidal seagrass habitats in Geoje Bay and Tongyeong Bay, South Korea (Figure 1): Osong (34°47′59.2″ N, 128°35′02.0″ E), Sagok (34°90′00″ N, 128°57′03″ E), and Sunchon (34°51′21.6″ N, 128°27′11.7″ E). All three sites were located within a semi-enclosed bay characterized by low wave energy and sandy–silty sediments. Although long-term disturbance records are lacking, the sites have been consistently affected by mechanical disturbance and vegetation removal from clam harvesting [44]. The three sites have not been subject to restoration intervention previously. Seagrass restoration efforts in Korea remain relatively limited compared to global initiatives but have increased in recent decades in response to widespread declines caused by climate change, coastal development, and destructive fishing practices [44,50,51].

The surveyed areas covered approximately 2.2 ha at Sagok and Osong, and about 0.7 ha at Sunchon, with the areas consisting mostly of unvegetated mudflats. At all sites, Z. japonica occupied the mid-intertidal zone as discrete patches rather than continuous beds. In spring, seawater temperatures ranged from 12.4 to 18.9 °C and salinity from 32.7 to 33.4 PSU, whereas in fall, temperatures varied between 15.2 and 26.2 °C and salinity between 29.1 and 32.5 PSU [52]. Tidal range during the study period spanned from −20 to 213 cm relative to mean sea level [53].

We determined sediment grain size at the three sites by collecting sediments from seagrass beds to a depth of 10 cm using a 5 cm diameter PVC corer. Organic matter was removed with 10 mL of 30% H₂O (overnight), after which samples were dried at 65 °C. Dried sediments were sieved through 63 μm and 2 mm meshes, and the retained fractions were weighed to the nearest 0.0001 g. Sediment composition was expressed as the pro-portion of mud (silt + clay; <63 μm), sand (63 μm–2 mm), and gravel (>2 mm). Sediment analysis revealed that substrates in the three study areas was predominantly sand, with similar sediment compositions observed between Z. japonica (gravel: 0–5.5%, sand: 82.0–93.0%, silt and finer: 6.0–21.9%) and adjacent unvegetated mudflats (gravel: 0–28%, sand: 77.4–95.0%, silt and finer: 2.3–22.3%).

2.2. Seasonal Variations in Upogebia Bioturbation Intensity and Its Relations to Seagrass Vegetation Metrics

We conducted field surveys across the three study sites to (1) compare the intensity of Upogebia bioturbation between Z. japonica patches and adjacent unvegetated mudflats, and (2) assess the relationship between seagrass vegetation characteristics and bioturbation intensity. Surveys were carried out in April (spring), July (summer), and November (fall) 2024 to capture seasonal changes in Upogebia bioturbation activity. At each site, random quadrats (50 cm × 50 cm) were placed during low tide over Z. japonica patches and unvegetated mudflats located at least 2 m from seagrass edge (n = 12 per habitat type per site). To quantify Upogebia bioturbation intensity, we counted the number of visible burrow openings within each quadrat [27,54]. While Upogebia typically construct Y-shaped burrows with two openings, some individuals form more complex burrow networks with multiple openings [26], making precise estimation of shrimp density from burrow counts difficult.

For quadrats placed within Z. japonica beds, we estimated the percent cover of live shoots and collected vegetation samples to measure shoot density and biomass. Seagrass vegetation was subsampled using a PVC corer (10 cm diameter, 15 cm depth) randomly placed within each quadrat. Samples were transported to the laboratory, rinsed with tap water, and cleaned of epiphytes before further processing. Seagrass tissues were separated into aboveground (leaves and leaf sheaths) and belowground (roots and rhizomes) components, oven-dried at 60 °C for 24 h and weighed to the nearest 0.0001 g to determine biomass. The survey data from Sagok were published previously [42] and are included here in full; we merged them with data collected contemporaneously at Osong and Suncheon to enable cross-site modeling.

2.3. Effects of Upogebia Bioturbation and Transplant Patch Size on the Persistence of Restored Seagrass Vegetation

We conducted a field experiment at Sagok Beach in October 2024 to assess the effects of extreme Upogebia bioturbation and seagrass transplant patch size on the persistence of restored Z. japonica. During the study period, the average temperature of the intertidal zone was 22.74 ± 0.05 °C (range = 13.46–34.27 °C; HOBO^® Pendent logger, Onset^®, Lincoln, NE, USA) (Figure A2). Seagrass transplant “plugs” of varying sizes were generated using PVC corers with diameters of 5 cm, 10 cm, 15 cm, and 26 cm, each to a depth of 10 cm (n = 24 per core size) (Figure A3). Transplant plugs were collected to ensure comparable shoot density across sizes (summarized in

Table A1. Vegetation characteristics and sediment grain size for different experimental groups during Zostera japonica transplant experiment.

Treatment	Diam. Core (cm)	Shoot Density (Shoots/Core)	<Silt (%)	Sand (%)	Gravel (%)
Control (9.4 ± 1.9 burrows m−2)	5	11.00 (±0.85)	1.96 (±0.02)	91.33 (±4.00)	6.71 (±3.97)
	10	22.63 (±0.75)
	15	41.63 (±1.28)
	26	72.63 (±1.39)
High (280.2 ± 5.8 burrows m−2)	5	9.63 (±0.91)	7.48 (±2.86)	88.92 (±2.07)	3.60 (±0.80)
	10	22.13 (±0.90)
	15	42.63 (±0.84)
	26	73.25 (±1.32)
Extremely High (455 ± 5.5 burrows m−2)	5	10.75 (±0.73)	4.12 (±0.36)	72.96 (±5.83)	22.92 (±6.19)
	10	22.63 (±0.65)
	15	44.00 (±0.98)
	26	73.63 (±1.45)

). We randomly assigned each plug to one of three burrow density treatments—control, high, and extreme high—yielding eight replicates per plug size per treatment.

Plugs of different sizes were transplanted in blocks marked with 3 cm-diameter PVC pipes, with each block containing one plug of each size. The measured burrow densities (mean ± SE) for each treatment were: control (9.4 ± 1.9 burrows m⁻²), high (280.2 ± 5.8 burrows m⁻²), and extreme high (455.0 ± 5.5 burrows m⁻²). The high burrow density treatment reflected the maximum values observed within Z. japonica during fall, while the extreme high treatment represented peak burrow densities recorded in adjacent unvegetated mudflats in the same season. Because areas completely free of Upogebia burrows were scarce near Z. japonica beds, we established the control plots in locations with the lowest burrow density possible and transplanted seagrass plugs to minimize direct contact with burrow openings. Although all experimental plots were unvegetated at the time of transplantation, their proximity (tens of meters) to existing Z. japonica beds and the pronounced annual fluctuations in bed extent at the site suggest that the area may have historically supported seagrass.

Sediment characteristics were more similar between the control and high burrow plots, with sand constituting 91% and 89% of substrates, respectively. In comparison, extreme high burrow plots showed lower sand (73%) and higher gravel contents (23%) (summarized in Table A1). We monitored the aerial changes in seagrass transplants at 0, 14, and 24 days post-transplantation. At each time point, patch area was estimated by measuring plug radii at 45° intervals from the plug center and calculating the enclosed polygonal area. The experiment was terminated on day 24 due to rapid loss of seagrass in the high and extreme burrow treatments.

2.4. Statistical Analyses

We used a generalized linear model (GLM) to analyze the effect of season, site, habitat type (seagrass vs. unvegetated mudflat), and their interactions on Upogebia burrow density. To account for overdispersion in the data, we opted for negative binomial GLM. To determine the effects of Upogebia burrow density, site, and season on Z. japonica shoot density, we applied a negative binomial GLM. Shoot cover was modeled with a binomial GLM with logit link, applying robust standard errors to accommodate underdispersion in the proportional (0–1) response data. Biomass data were analyzed with linear regression using the same explanatory variables, after applying transformations to meet model assumptions (square-root transformation for aboveground, log transformation for belowground). To examine within-group trends, we estimated regression slopes of burrow density for each site × season combination and tested their significance using false discovery rate (FDR) adjustment within season (package “emmeans”, version 1.6.1).

For seagrass transplant experiment, we used linear mixed-effect model (LME; package “nlme”, version 3.1-168) to test the effects of burrow density, days since restoration, initial patch size of seagrass, and their interactions on % areal changes in seagrass patch. To account for non-independence arising from the use of block design and repeated measurements of individual patches over time, both block ID and days since restoration were included as random intercepts. Finally, to quantify the rate of seagrass patch degradation under bioturbation, we modeled the temporal decline in patch area using an exponential decay model. Analyses were restricted to the high and extreme high burrow density treatments, as patch degradation was minimal in the control group. For each replicate plug, we fitted the exponential decay model with the following formula:

R = A·e^−kt

(1)

where R is the % area of patch remaining at time t, A is the initial % patch area (100% on day 0), and k is the decay constant. Models were fitted using the Levenberg–Marquardt algorithm (package “minpack.lm”, version 1.2-4), allowing up to 1000 iterations. The decay constant k was extracted for each replicate, representing the rate of patch loss. Replicates for which the model failed to converge were excluded from further analysis. All statistical analyses were conducted using R (R Core Team 2024, Vienna, Austria).

3. Results

3.1. Seasonal Variations in Upogebia Bioturbation Intensity and Its Relations to Seagrass Vegetation Metrics

Across all sites, burrow densities were generally higher in mudflats than in seagrass (zero-inflated negative binomial GLM, χ² = 25.10, p < 0.001), and densities increased markedly from spring to fall (χ² = 28.33, p < 0.001). Significant two-way interactions were also observed for habitat × season (χ² = 6.94, p = 0.031), habitat × site (χ² = 10.45, p = 0.005), and season × site (χ² = 37.97, p < 0.001), while the three-way interaction was not significant (χ² = 3.42, p = 0.489) (Figure 2B). The zero-inflation component indicated a small but significant probability of excess zeros (z = –2.16, p = 0.031). At Sagok, burrow density in mudflats increased from 116.3 ± 33.8 m⁻² in spring to 336.0 ± 13.4 m⁻² in fall, while those in seagrass increased from 54.7 ± 10.6 to 183.0 ± 21.4 m⁻² over the same period. A similar seasonal trend was observed at Osong but not at Sunchon (Figure 1).

Seasonal changes in Z. japonica vegetation metrics are summarized in

Table 1. Seasonal variations in Zostera japonica vegetation characteristics and sediment grain size across the three study sites. The values for vegetation characteristics in Sagok were derived from Seo & Lee (2025) [42].

Parameters	Osong			Sagok			Sunchon
	Spring	Summer	Fall	Spring	Summer	Fall	Spring	Summer	Fall
Shoot density (shoots/core)	36.00 (±3.86)	41.92 (±3.52)	26.08 (±3.08)	31.50 (±3.86)	46.08 (±3.76)	37.33 (±4.05)	38.83 (±3.10)	37.75 (±4.53)	30.25 (±3.91)
Percentage cover (%)	99.33 (±0.45)	100.00 (±0.00)	91.00 (±3.34)	97.33 (±1.02)	36.00 (±3.86)	93.67 (±2.63)	95.67 (±2.06)	99.67 (±0.33)	98.00 (±1.67)
Above ground biomass (g/core)	0.31 (±0.06)	0.54 (±0.09)	0.34 (±0.09)	0.26 (±0.33)	1.24 (±0.14)	0.52 (±0.04)	0.67 (±0.07)	1.18 (±0.14)	0.48 (±0.08)
Below ground biomass (g/core)	0.36 (±0.07)	0.60 (±0.13)	0.27 (±0.05)	0.26 (±0.03)	0.72 (±0.10)	0.31 (±0.04)	0.42 (±0.06)	0.45 (±0.05)	0.22 (±0.03)
<Silt (%)	14.36 (±4.5)			5.95 (±1.29)			6.34 (±2.16)
Sand (%)	82.58 (±3.38)			93.13 (±3.38)			80.03 (±4.65)
Gravel (%)	3.05 (±1.24)			0.92 (±0.65)			13.63 (±5.91)

. Burrow density interacted with season and site to influence multiple structural attributes of Z. japonica vegetation. Percentage shoot cover declined significantly with increasing burrow density (binomial GLM, χ² = 6.05, p = 0.014), with marginally significant interactions observed between site and season (χ² = 5.55, p = 0.06) (Figure 3A). Similarly, shoot density showed a negative association with burrow density (negative binomial GLM, χ² = 8.45, p = 0.004); was significantly influenced by burrow density × season (χ² = 8.02, p = 0.005) and site × season interactions (χ² = 11.19, p = 0.004) (Figure 3B). For belowground biomass, a significant main effect of burrow density was detected (gaussian GLM, F_1,60 = 7.57, p = 0.008), along with a strong site × season interaction (F_2,60 = 5.31, p = 0.008) (Figure 3D). In contrast, aboveground biomass showed no significant association with burrow density but varied significantly by site (F_2,60 = 8.45, p < 0.001) (Figure 3C). Examination of within–group regression slope estimates indicated that significant negative relationships between burrow density and vegetation metrics were restricted to fall.

3.2. Effects of Upogebia Bioturbation and Transplant Patch Size on the Persistence of Restored Seagrass Vegetation

Percentage change in patch area of transplanted seagrass was significantly influenced by burrow density (Linear mixed-effect model, F_2,21 = 208, p < 0.0001), day (F_1,180 = 370.8, p < 0.0001), initial patch size (F_3,63 = 12.2, p < 0.0001), and all two-way and three-way interactions among these factors (p < 0.0001 for all) (Figure 4A). In high and extreme burrow treatments, patch area declined rapidly across all patch sizes, with almost all transplants lost by day 24. Patch loss dynamics up to day 14 indicated a size-dependent response, with smaller patches exhibiting disproportionately greater areal losses under bioturbation. In the control plots, this trend was reversed: the smallest patches expanded by 112%, whereas medium and large patches exhibited minimal change or slight contraction (Figure 4A). This pattern was corroborated by the exponential decay models, which showed that decay constants (k) increased progressively with burrow density but decreased with transplant size (Linear mixed-effect model; patch size effect, F_3,42 = 15.01, p < 0.0001; burrow effects, F_1,14 = 6.16, p = 0.026) (Figure 4B). No significant interaction between patch size and burrow density was detected (F_3,42 = 1.52, p = 0.21).

4. Discussion

Despite growing recognition that bioturbators can hinder the restoration of coastal vegetated habitats [17], empirical understanding of their effects and underlying mechanisms remains limited. Through a combination of seasonal field surveys and a manipulative experiment, we demonstrated that U. major can reach extremely high densities within intertidal seagrass beds during the fall, coinciding with the optimal season for seagrass restoration. At such high densities, bioturbation by U. major was associated with declining seagrass shoot cover, shoot density, and belowground biomass. Seagrass patches transplanted into Upogebia-dominated areas experienced rapid vegetation loss in response to increasing bioturbation intensity within just a few weeks. Notably, seagrass patch size influenced the rate of degradation under Upogebia bioturbation, with larger patches exhibiting disproportionately slower rates of areal loss. These results indicate that extreme levels of Upogebia bioturbation can pose a threat to intertidal seagrass restoration, whereas its impact may be partially mitigated by restoration interventions with larger planting scales.

Across the three study sites, the burrow density of U. major inside Z. japonica ranged from 0 to 280 burrows m⁻², with an average of 65.2 burrows m⁻², corresponding to approximately 33 individuals m⁻². Conversely, Upogebia burrow density in unvegetated mudflat averaged 116.5 burrows m⁻², with maximum density exceeding 432 burrows m⁻². In the west coast of Korea, previous studies showed that Upogebia burrow density can be as high as 952 burrows m⁻² [27,54,55]. These observed values are significantly higher than reported densities for U. major and other burrowing shrimp species associated with Z. japonica and unvegetated mudflats. For example, in other East Asian regions, U. major densities have been reported to range between 7 and 36 individuals m⁻² [41,56,57]. Neotrypaea californiensis in North America constructs an average of 43 burrows m⁻² within Z. japonica beds and 117 burrows m⁻² (max. 780 burrows m⁻²) in adjacent mudflats devoid of vegetations [58,59].

Our survey revealed strong seasonality in Upogebia bioturbation intensity. From spring to fall, burrow density increased on average by 190% in Z. japonica and 140% in unvegetated mudflats. This sharp increase likely reflects seasonal recruitment patterns, as Upogebia larvae settle from late spring to early summer and populations typically peak between summer and fall [55,57,60]. Seasonal trajectories, however, differed among sites. In Osong and Sunchon, shrimp abundance in seagrass and mudflat habitats increased in parallel. In contrast, Sagok exhibited a sharper summer increase in mudflats and a delayed fall peak in seagrass. This indicates that post-settlement processes, such as cross-habitat colonization or competition with seagrass or conspecifics, may modulate shrimp abundance across intertidal landscape [31]. Dense seagrass rhizomes may initially restrict settlement by acting as physical barriers for newly settled recruits [34,61,62]. As the season progresses, seagrass belowground biomass naturally declines (Table 1) [42,63]. reducing these barriers and facilitating greater shrimp colonization and persistence [34,61]. That said, these seasonal patterns were not uniform across sites, manifesting only in Osong and Sagok, suggesting that local physical conditions further influence recruitment dynamics.

Season strongly mediated the relationship between Upogebia bioturbation and seagrass vegetation metrics. In spring and summer, when shrimp density was relatively low and seagrass exhibited rapid growth or peak biomass [44,63], vegetation metrics showed no significant relationship with burrow density. In contrast, during fall when Upogebia burrow density peaks with a 41% to 126% increase in maximum density across sites, strong negative relationships with seagrass vegetation metrics emerged. This density−dependent pattern is consistent with previous findings that increasing burrowing shrimp density beyond a threshold drives rapid seagrass decline [32,61]. For instance, Hull & Ruesink (2024) [25] reported that Z. marina shoots declined with increasing callianassid shrimp density and were completely lost above 336 shrimps m⁻². size increases from spring to fall as recruits grow [54,60], amplifying per capita bioturbation impacts [35,54]. Bioturbation stress may also exacerbate natural seasonal declines in seagrass biomass and productivity, resulting in extensive vegetation loss [63,64]. This contrasts with the earlier growing season, when rapid growth and high productivity might allow seagrasses to better compensate for mechanical damages caused by shrimp bioturbation [33].

Our field experiment shows that Upogebia bioturbation strongly suppresses the establishment of seagrass transplant, and this effect is mediated by shrimp burrow density. Relative to the control group, seagrass transplants in the high-and extreme-high burrow treatment groups exhibited significant patch decay, with major vegetation losses occurring rapidly within 14 days. Whereas the extremely high burrow treatment reflected conditions almost exclusively found in unvegetated mudflats, the high burrow treatment was within densities observed in both seagrass and mudflat at our study site during fall (Figure 2B). Thus, the negative impacts of Upogebia bioturbation under our experimental treatments likely represent realistic constraints on seagrass persistence. Several mechanisms likely underlie the negative impacts of Upogebia bioturbation on seagrass transplants: (a) excavation of deep, multi-entrance tunnel matrices can lower sediment shear strength and promotes lateral erosion, thereby hindering transplant anchorage and rhizome extension [31,65]; (b) elevated sediment turnover and burial reduce oxygen availability around below-ground tissues and limit photosynthesis, potentially increasing seagrass mortality [16,32,66]; and (c) direct physical disturbance during excavation and burrow maintenance may damage leaves and rhizomes or remove entire shoots (Figure A4) [15]. Determining the relative contribution of these different mechanisms requires targeted manipulative experiments and observations (e.g., time-lapse/video).

Our results support the hypothesis that patch size mediates the persistence and growth of transplanted seagrass under mud shrimp bioturbation. In the absence of significant bioturbation (control), the percentage change in patch area was unrelated to initial patch size; in fact, the smallest patches (5 cm diam.) performed the best, possibly due to reduced competition [38]. In contrast, under shrimp treatments where patch loss prevailed, percentage change in patch area was inversely related to initial patch size. The positive effect of patch size on bioturbation resistance remained consistent across burrow density levels, as indicated by the non-significant interaction term in the exponential decay model. Thus, larger patch size slowed degradation rates independently of burrow density. Several mechanisms likely contributed to the slower degradation of larger transplants. Larger seagrass patches with structurally complex belowground structures may confer greater associational defense by physically interfering with shrimp burrowing activities [61]. Higher belowground structural complexity may also reduce sediment erosion and associated vegetation loss even under extensive burrowing activities [31,34]. Because high burrow densities caused rapid vegetation loss within weeks, our experiment could not assess the long-term treatment effects on restoration outcomes. Future studies should therefore test a wider range of burrow densities, particularly low to moderate levels, and extend monitoring to capture longer-term vegetation recovery.

Our findings have direct implications for restoring intertidal seagrasses where mud shrimp activity is high. Prior work indicates that fall is often the most effective season for transplantation in temperate Asia because it avoids peak summer heat stress and allows plant establishment before winter [49]. However, this window also overlaps with peak Upogebia activity when bioturbation most strongly threatens transplant persistence. Fall plantings should therefore be sited only in areas with historically low Upogebia densities, verified through pre-planting investigations of recipient sites. We further demonstrate that facilitation may play an important role in the persistence of seagrass transplants under bioturbation stress. Harnessing positive interaction is increasingly recognized as a cornerstone of effective restoration interventions [11,12,67]. Increasing patch size or planting density can promote self-facilitation and enhance resistance to biotic and abiotic stressors [36,37,38,40]. Even so, almost all transplants failed within 24 days, indicating that Z. japonica restoration likely requires larger planting units than those tested and, possibly, additional protective measures. For example, Suykerbuyk et al. (2012) [35] showed that embedding a shell layer beneath Z. noltii transplants suppressed bioturbating lugworms and markedly improved restoration success. By analogy, installing a thin shell or mat barrier to disrupt burrow construction may provide similar benefits for Z. japonica in heavily bioturbated areas [68]. Over longer time scales, recovery of predator communities that consume burrowing shrimp may help reduce bioturbation pressure and increase restoration durability [13], although direct evidence for top-down control of U. major is still extremely limited [26]. Finally, physical factors (e.g., wave exposure, sediment type) and human disturbances (e.g., clam harvesting) should be considered alongside controlling shrimp bioturbation, as they may further amplify bioturbation effects on restoration outcomes. In conclusion, we provide novel evidence that seasonal peaks in Upogebia bioturbation can severely constrain intertidal seagrass restoration, and that leveraging facilitative interactions through increased planting scale can partially offset these impacts. These results underscore the need for strategic restoration siting to avoid areas of extreme bioturbation stress and for restoration designs that build in positive interactions while suppressing antagonistic shrimp effects. Z. japonica remains highly threatened across East Asia and best-practice guidance is limited. Future work should investigate how mud shrimp influence other restoration interventions for the species (e.g., seed-based) and assess their impacts on habitat recovery over multi-year timescales.