Introduction

In biodiverse ecosystems, the decomposition of litter mixtures from different plant species is a fundamental ecosystem process that determines nutrient cycling and organic carbon stock1. The decomposition of litter mixture has been a focus of substantial research efforts in recent decades. However, predicting the litter mixture effects on decomposition remains challenging, particularly with the shifting plant community structure along environmental gradients or due to global climate change and anthropogenic disturbances observed in various terrestrial2,3 and aquatic4,5,6,7,8 habitats. Invertebrate faunas have been recognized as critical biotic agents driving decomposition efficiency worldwide9. However, it remains unclear how shifts in the species composition and diversity of plant litter mixtures supplied to the soil food web stimulate changes in invertebrate faunal communities and how these changes, in turn, affect litter decomposition10.

Invertebrate fauna can directly affect litter decomposition through grazing11 and litter fragmenting activities12,13, and indirectly by influencing soil structure and chemical status14 and through interactions with microbial decomposers15. The mechanisms by which invertebrate fauna modulate the decomposition of litter mixtures are likely related to enhanced resource diversity and habitat heterogeneity16,17, complementary resource use13,18, food and substrate preferences19,20, and diffusion of anti-herbivore compounds21. These are potential mechanisms trigging various diversity effects on decomposition, which contrast with what is expected from monospecific treatments22,23. It is suggested that these mechanisms likely differ between marine, terrestrial, and freshwater habitats due to differences in soil biota, litter resources, and litter leaching dynamics under contrasting environmental conditions1. A wealth of experimental evidence from terrestrial forests and streams demonstrates that changes in plant species diversity and composition can significantly alter decomposition processes1,24. However, evidence from marine habitats is extremely scarce, and the mechanism of litter-mixing effects on the decomposition of marine macrophytes is largely unknown25,26.

Seagrass meadows are among the most productive marine ecosystems worldwide, where various macrophyte species (seagrasses and macroalgae) coexist27,28. These meadows produce diverse macrophyte litters that contribute greatly to the food web, nutrient cycling, and organic carbon stock in surrounding areas and even in deep-sea zones29,30. The utilization and decomposition of marine macrophyte litter (or wracks) by benthic faunas are likely influenced by the dietary strategies and habitat preferences of these organisms31,32,33. Generally, invertebrate fauna in marine habitats, including herbivores and detritivores, tend to prefer resources rich in labile compounds and nutrients and low in chemical defenses (e.g. phlorotannins and other phenolics) to meet their trophic needs32,34,35. Additionally, invertebrate fauna often inhabits macrophytes (or their litters) with high structural complexity33,36.

Litter mixtures of seagrasses and macroalgae primarily form during maximum litterfall following the growing season37,38. These litter mixtures vary spatiotemporally in coastal areas as they may compose various species characterized by different nutritional qualities, chemical compounds, and three-dimensional structures36,39,40. Therefore, it is suggested that differences in the species identity and composition of litter mixtures may influence the community structure of invertebrate fauna in marine habitats, with particular litter types promoting specific taxa of fauna, as reported in terrestrial habitats41,42.

Litters of seagrasses and macroalgae decompose at different rates39. Seagrasses are vascular plants characterized by high levels of refractory compounds (e.g. lignocellulose) and are relatively nutrient-poor, with high carbon and low nitrogen and phosphorus content. Consequently, their litters are generally less degradable for microbes and detritivores than those of macroalgae40,43. In contrast, macroalgae lack lignocellulose and are considered more easily degradable by microbes and detritivores40. However, macroalgae are very diverse, and their decomposition rates vary greatly according to taxa/species44,45,46. For example, red algae (Rhodophyta) and brown algae (Phaeophyta) contain taxon-specific polysaccharides that provide recalcitrance and thus delay degradation, while these features are absent in green algae (Chlorophyta)47. It has been reported that seagrasses such as Zostera sp., Cymodocea sp., Halodule sp., and Thalassodendron take about 37–49 days to lose 50% of their initial leaf litter mass48,49,50, while it takes about 150 days for Posidonia sp.51. In contrast, previous studies found that the incubation period needed for macroalgal litters to disappear from litterbags ranged from a few days (e.g. Ulva sp. and Laminaria sp.) to months (e.g. Plocamium sp. and Fucus sp.)44,45,52. Therefore, litter mixtures of seagrasses and macroalgae may exhibit distinct stage-specific characteristics in terms of litter availability, nutritional quality, and species composition. This is due to the rapid decomposition of easily degradable species within a few weeks and the persistence of less degradable species for months to years43,53. In terrestrial and freshwater habitats, it is well-studied that successional changes in the litter mixtures from the initial to later decomposition stages may elicit different responses from various faunal taxa, resulting in different feedback on litter decomposition16,54,55. This process should also be tested in marine ecosystems to understand the mechanisms underlying litter-mixing effects.

In this study, we assessed the possible consequences of macrophyte litter-mixing on benthic faunal communities and litter decomposition by conducting a series of in situ decomposition experiments across gradients of seagrass and macroalgal litter species richness and composition. We hypothesized the following: Hypothesis I: Increases in both initial litter species richness and the degree of macroalgal dominance (indicated by the percentage of macroalgal mass) in the early stage of decomposition (Stage 1) lead to increased faunal abundance through enhanced resource quality/diversity and microhabitat complexity (Fig. 1a, b). However, this trend diminishes in the late stage (Stage 2) when most of the easily degradable macroalgal litters have decomposed (Fig. 1c). Hypothesis II: Seagrass litter decomposition during Stage 1 is negatively influenced by increased both initial litter species richness and macroalgal dominance because benthic faunas selectively utilize macroalgae and neglect seagrasses when given a choice (Fig. 1d). Hypothesis III: Higher initial litter species richness will result in less litter remaining (mainly seagrass) in Stage 2, and the colonization of fauna accelerates the decomposition of the remaining litter during this stage (Fig. 1e).

Fig. 1: Schematic diagrams illustrating the main hypotheses concerning the roles of litter-mixing, decomposition stages, and benthic faunal communities in controlling litter-mixing effects on decomposition.
figure 1

In particular, we hypothesized the following: (1) Increased initial litter species richness enhances faunal abundance in the early stage (Stage 1). This trend diminishes during the late stage (Stage 2) when macroalgae disappears (see a–c). (2) The decomposition rates of seagrass are negatively related to both litter species richness and macroalgal dominance during Stage 1 because of the resource selection of fauna favouring macroalgae (see d). (3) The decomposition rates of seagrass are positively related to both litter species richness and faunal presence in the absence of macroalgae during Stage 2 (see e).

Full size image

Results

Nutrient contents and ratios of seagrass and macroalgae litters

The litter of seagrass Z. marina had the lowest N content and the highest C:N ratio, while the litter of seagrass P. japonicus had the lowest P content and the highest C:P ratio (Table 1). In contrast, the litter of the macroalgae E. prolifera and C. okamurae exhibited relative higher N and P content and lower C:N and C:P ratios than other species. The litter of the macroalgae S. fulvellum and U. lactuca had intermediary values of N and P content, and C:N, and C:P ratios among the six macrophytes (Table 1). C. okamurae, S. fulvellum, and P. japonicus were much richer in C than U. lactuca and E. prolifera (Table 1).

Table 1 Nutrient contents and ratios of seagrass and macroalgae litters
Full size table

Litter-mixing effects on the decomposition of seagrass and macroalgae litters

After 43 days of monospecific incubation, the litter of the four macroalgae species had lost mass more quickly than that of the two seagrass species (Fig. 2a). Specifically, Z. marina and P. japonicus lost 43% and 66% of their initial mass, respectively, while S. fulvellum, E. prolifera, C. okamurae, and U. lactuca lost 80%, 92%, 94%, and 98% of their initial mass (Fig. 2a). After 86 days of incubation, more than 99% of the macroalgal masses was lost. In comparison, Z. marina and P. japonicus had lost 51% and 83% of their mass after 86 days, and 83% and 92% of their mass after183 days of incubation, respectively (Fig. 2b, c). Thus, in the present study, we identified Stage 1 (0–43 days) and Stage 2 (44–183 days) as the early and late stages of decomposition, respectively.

Fig. 2: Net mass loss of monospecific seagrass and macroalgal litters.
figure 2

Net litter mass loss after 43 (a), 86 (b), and 183 (c) days of field incubation. Values are means ± SE (n = 4). Different letters indicate significant (p < 0.05) differences between litter species. Species ID: Zostera marina (Z), Phyllospadix japonicus (P), Sargassum fulvellum (S), Ulva lactuca (U), Enteromorpha prolifera (E), and Caulacanthus okamurae (C).

Full size image

The net mass loss (%) of seagrass litter varied significantly (p < 0.001) over the incubation period, but a significant (p < 0.001) effect of species richness was only observed for P. japonicus after 43 days of incubation (Fig. 3, Supplementary Fig. 1 and Table 2). Stage-specific litter mass loss differed significantly (p < 0.001) between the decomposition stages, and the interactions between species richness and decomposition stage were significant (p = 0.011), indicating contrasting effects of litter species richness on stage-specific litter mass loss between Stages 1 and 2 (Fig. 4a–d). For example, the litter of P. japonicus lost less mass in the four- and five-species treatments compared to the single- to three-species treatments during Stage 1 (Fig. 4c), whereas the litter lost mass faster in higher species richness treatments during Stage 2 when there was no macroalgal material left (Fig. 4d).

Fig. 3: Net mass loss and the corresponding relative mixture effect (RME).
figure 3

Net mass loss and the RME in Zostera marina (a–c) and Phyllospadix japonicus (d–f) under different litter species composition treatments after 43, 86 and 183 days of incubation. Values are means ± SE for seagrass mass loss (%) and means ± 95% confidence intervals for RMEs (n = 4). Significant (p < 0.05) RMEs are indicated by an asterisk (*). Species ID: Zostera marina (Z), Phyllospadix japonicus (P), Sargassum fulvellum (S), Ulva lactuca (U), Enteromorpha prolifera (E), and Caulacanthus okamurae (C).

Full size image
Table 2 Summary of two-way ANOVA testing the effects of litter species richness (SR) and incubation time (T) or decomposition stage (S) on the percentage of seagrass litter mass loss, stage-specific litter mass loss, faunal abundance, and relative faunal mass in each litterbag series
Full size table
Fig. 4: Effects of litter species richness on the stage-specific mass loss.
figure 4

The percentage of mass loss for Zostera marina (a, b) and Phyllospadix japonicas (c, d) during Stages 1 (0–43 days) and 2 (44–183 days) of incubation. Values are means ± SE (n = 4–24). Different letters denote significant (p < 0.05) differences between treatments. ns not significant.

Full size image

The relative mixing effect (RME) of the net litter mass loss (%) varied over time and depended on the species composition of the mixtures (Fig. 3 and Supplementary Fig. 2). Throughout the incubation period, only 14 out of 90 cases (15.6%) of multi-species composition treatments significantly (p < 0.05) altered the net mass loss of seagrass compared to that observed in single-seagrass treatments (Fig. 3). In contrast, 12 out of 30 cases (40%) of litter-mixing treatments exhibited significant (p < 0.05) RMEs on macroalgal litter decomposition at 43 days of incubation, with 10 of them accelerating decomposition compared to the monospecific treatment (Supplementary Fig. 2).

The RME of seagrass litter decomposition was strongly (p = 0.012) related to both litter species richness and the degree of overall macroalgal dominance during Stage 1, i.e., when macroalgae were present (Fig. 5a and Supplementary Table 1). Specifically, lower macroalgal dominance levels of 50% and 67% (in the two- to three-species treatments) triggered positive RMEs for seagrass litter decomposition, whereas higher macroalgal dominance levels of 75% and 80% (in the four- to five-species treatments) triggered negative RMEs (Fig. 5a). In contrast, the estimated RME for macroalgal litter decomposition tended to be more positive under conditions of higher macroalgal dominance (Fig. 5a and Supplementary Table 1). The RME of seagrass litter decomposition was also affected by the macroalgal identity. The presence of each macroalgal species, except C. okamurae, led to significantly (p < 0.05) lower RMEs for seagrass litter decomposition under constant macroalgal dominance (Fig. 5b and Supplementary Table 1).

Fig. 5: Effects of litter species richness and macroalgal identity on the relative mixture effect (RME).
figure 5

Effects of initial litter species richness (a) and the presence/absence of a particular macroalgal species (b) on the RME on seagrass litter decomposition after 43 days of incubation. The bracketed value in (a) represents the relative macroalgal dominance by mass in litter mixtures. Values are means ± SE (n = 4–24). Significant (p < 0.05) effects of macroalgal presence are indicated by an asterisk (*). Species ID: Sargassum fulvellum (S), Ulva lactuca (U), Enteromorpha prolifera (E), and Caulacanthus okamurae (C).

Full size image

Litter-mixing effects on benthic faunal communities

Ascidians and polychaetes were the dominant fauna in the 432 litterbags collected during the decomposition experiment, accounting for 40.3% and 33.8%, respectively, of the total number of individuals collected (Supplementary Table 2). Litter species identity had significant (p < 0.05) effects on the abundance of polychaetes and crustaceans but not on molluscs and ascidians after 43 days of incubation (Supplementary Table 3). Specifically, among monospecific treatments, S. fulvellum hosted the highest number of polychaetes and crustaceans, followed by U. lactuca, C. okamurae, E. prolifera, and Z. marina, while P. japonicus hosted the lowest number (Supplementary Fig. 3a). This trend gradually reversed over incubation time, with the abundance of polychaetes and ascidians highest in P. japonicus and lowest in S. fulvellum monospecific treatments after 183 days of incubation (Supplementary Figs. 4a and 5a). Different litter compositions also triggered dramatic changes in benthic faunal abundance during the incubation period (Supplementary Figs. 3–5).

The abundance of each faunal taxon, except crustaceans in Z. marina litterbags, varied significantly (p < 0.001) over the incubation period (Table 2). Initial litter species richness had significant (p < 0.05) effects on the abundance of polychaetes and crustaceans in all litterbags and ascidians in Z. marina litterbags (Table 2). No effect of species richness on the mollusc abundance was observed (Table 2).

The expected positive effect of litter species richness on faunal abundance was not observed in any taxon of the fauna during Stage 1 (Fig. 6). Instead, the overall macroalgal dominance exhibited a positive effect on the abundance of crustaceans but a negative effect on the abundance of ascidians (Supplementary Fig. 6 and Supplementary Table 4). In addition, litter type had significant (p < 0.05) effects on the abundance of certain taxa of benthic fauna (Supplementary Table 4). The abundance of polychaetes was the lowest in the seagrass litter that decomposed alone (Fig. 6a, m). The presence of macroalgae markedly (p = 0.001) increased the polychaete abundance (Fig. 6a, m and Supplementary Table 4). The abundance of crustaceans was significantly affected by the presence of seagrass (p = 0.004) and macroalgae (P = 0.035), being the lowest in the monospecific seagrass litterbags and the highest in the monospecific macroalgal litterbags (Fig. 6d, p and Supplementary Table 4). The abundance of ascidians was significantly (p = 0.003) affected by the presence of seagrass (Fig. 6g, s and Supplementary Table 4). Litter type had little effect on mollusc abundance (Fig. 6j, v and Supplementary Table 4).

Fig. 6: Dynamics of benthic faunal abundance.
figure 6

The abundance of polychaetes (POL; a–c and m–o), crustaceans (CRU; d–f and p–r), ascidians (ASC; g–i and s–u) and molluscs (MOL; j–l and v–x) under different litter species richness treatments after 43, 86 and 183 days of incubation. Note that the first bar in each panel represents the overall mean value of faunal abundance in monospecific treatment of macroalgal species. Values are means ± SE (n = 4–24). Different letters indicate significant (p < 0.05) differences between treatments. ns not significant.

Full size image

The patterns of abundance of polychaetes, crustaceans, and ascidians along the gradients of the initial litter species richness shifted considerably during Stage 2, when the macroalgal litter had almost been lost, compared to those of Stage 1 (Fig. 6). At 86 and 183 days of incubation, the abundances of polychaetes and ascidians tended to be lower in the macroalgal monospecific and/or five-species treatments, in which the initial proportion of seagrass mass was low (Fig. 6b, c, h, i, n, o, t, u), whereas the abundance of crustaceans was almost independent of the initial litter species richness (Fig. 6e, f, q, r). The overall abundance of fauna exhibited a pattern similar to that observed for polychaetes (Fig. 6 and Supplementary Fig. 7).

The relative faunal mass increased drastically (p < 0.001) over time and changed with species richness (Fig. 7 and Table 2). The values did not differ significantly (p > 0.05) among the species richness treatments after 43 days of incubation (Fig. 7a, d). However, the relative faunal mass increased dramatically after 86 and 183 days of incubation as the species richness increased from 1 to 4. The value in the five-species treatments was slightly lower than or comparable to the value in the four-species treatments (Fig. 7b, c, e, f).

Fig. 7: Relative faunal mass in different litter species richness treatments.
figure 7

Relative faunal mass in litterbag series of Zostera marina (a–c) and Phyllospadix japonicus (d–f) after 43, 86, and 183 days of incubation. Values are means ± SE (n = 4–24). Different letters denote significant (p < 0.05) differences between treatments. ns not significant.

Full size image

Relationships between stage-specific seagrass mass loss and relative faunal mass

No significant (R2 = 0.001, F = 0.064, p = 0.804; R2 = 0.170, F = 2.863, p = 0.113) relationship was observed between the stage-specific seagrass mass loss and the relative faunal mass during Stage 1 (Fig. 8a, c). In contrast, there was a significant linear relationship (R2 = 0.289, F = 5.684, p = 0.032) between Z. marina mass loss and a significant non-linear relationship (R2 = 0.613, F = 22.189, p < 0.001) between P. japonicus mass loss and the relative faunal mass during Stage 2 (Fig. 8b, d). The mass loss of Z. marina increased linearly with an increasing relative faunal mass across the range of 0.2–1.0 (Fig. 8b). In contrast, the mass loss of P. japonicus increased dramatically within a range of 1.0–3.0 (Fig. 8d). The mass loss gradually approached the maximum level with a further increase in the relative faunal mass (Fig. 8d).

Fig. 8: Relationships between the stage-specific mass loss of seagrass litter and the relative faunal mass.
figure 8

Zostera marina litter mass loss (a, b) and Phyllospadix japonicus litter mass loss (c, d) changed with the relative faunal mass at different decomposition stages. In each panel, the data points, colored according to species richness (SR), represent values from the sixteen treatments of litter species composition. Red solid lines represent significantly (p < 0.05) fitted regression lines. Shaded region indicates the 95% confidence interval.

Full size image

Discussion

In the present study, the litter of the two seagrasses exhibited lower nutritional quality, with lower N% and P% and higher C:N and C:P ratios compared to the four macroalgal species. Across various terrestrial and aquatic plant taxa, differences in the carbon, nitrogen, and phosphorus concentrations of plant detritus have been found to account for 89% of the variance in litter decomposition rates39. Our litterbag experiment showed that macroalgal litter decomposed faster than seagrass litter. This may be attributable, at least in part, to differences in nutrient content, as well as the high levels of lignocellulose in seagrass tissues and the lack of lignocellulose in macroalgae40.

However, the differences in nutritional quality alone could not explain the varying decomposition rates among species of either seagrass or macroalgae. For example, S. fulvellum and U. lactuca were quite similar in N% and P% and their ratio with C%, and both were less nutritious than E. prolifera and C. okamurae. However, U. lactuca decomposed at a much faster rate, while S. fulvellum decomposed more slowly than the other macroalgal species. Macroalgae such as Sargassum spp. and Fucus spp. may be rich in refractory compounds or chemical defenses (e.g. polysaccharides, phlorotannins, and phenolics) compared to species like Ulva spp.44,47,56. This may also partially account for the differences in degradation rates among species.

In the present study, litter mixtures of seagrass and macroalgae lost the vast majority of macroalgal mass during Stage 1 (0–43 days), whereas the remaining seagrass litter decayed in the absence of macroalgae during Stage 2 (44–183 days). Previous studies in terrestrial and freshwater habitats have found that successional changes in litter mixtures from the initial to later decomposition stages can influence the abundance and structure of invertebrate fauna16,54,55. A similar result was reported in an algal removal experiment, which measured the effects of algal cover on the structure of an intertidal benthic assemblage57. Thus, in this study, the presence or absence of diverse macroalgal resources in litter mixtures, depending on the incubation time, was expected to modulate benthic faunal assemblages.

Contrary to our Hypothesis I, we did not observe a general trend regarding the effect of litter species richness on the faunal abundance during Stage 1 of decomposition. However, this finding is consistent with the results of previous studies conducted on coastal and terrestrial ecosystems10,58,59. The effects of litter mixing on faunal abundance, such as those observed in the present study, can be largely attributed to variations in litter identity and composition rather than species richness per se. This is because different litter types generally promote different faunal groups41,60. For example, we found that different taxa of fauna responded differently in both magnitude and direction to an increase in macroalgal dominance in litter mixtures. In the present study, the monospecific treatment of macroalgae, particularly S. fulvellum and C. okamurae, hosted a significantly larger number of polychaetes and crustaceans than seagrass litter during the early decomposition stage. Unlike the unbranched and nutrient-poor seagrass, coarsely branched and structurally complex macroalgae, such as S. fulvellum and C. okamurae, may serve as both a habitat and a food source. In contrast, unbranched and nutrient-rich macroalgae, such as Ulva spp., usually serve primarily as a food source for invertebrates, including amphipods, isopods, gastropods, and polychaetes31,32,33,60. These faunas could benefit from an increase in macroalgal dominance, as these macroalgae provide nutrient-rich and palatable food and/or structurally complex habitats32,36,46. However, not all faunal species prefer labile over refractory litter. We found that the presence of less degradable seagrass species significantly promoted the abundance of sessile ascidian species, whereas the presence of macroalgae had a negative effect. Therefore, the varying responses of different faunal taxa to a particular litter species may be responsible for the absence of positive litter species richness effects on faunal communities.

The effects of initial litter mixing on faunal communities diminished during Stage 2. This result supports previous studies that have found short-term (4–6 weeks) effects of macroalgae on faunal assemblages because macroalgal mats are ephemeral and rapidly degrade59,61. After 183 days of incubation, when most of the macroalgal litter was depleted, the abundance of the macrofauna, particularly polychaetes, largely depended on the remaining seagrass mass, suggesting the importance of resource availability57. In contrast, crustacean abundance was almost identical among litter species richness treatments during Stage 2 of decomposition. Crustaceans such as amphipods are typically the first to colonize newly deposited litter. Their high mobility and tendency to forage on freshly senescent materials make them important in the early stages of the degradation process34,62.

Our results showed that only 15.6% of the litter mixture treatments altered the net seagrass litter mass loss (%) during the experiment, suggesting weak effects of litter mixing on the overall breakdown of the recalcitrant seagrass litter. In contrast, positive litter-mixing effects were more pronounced (33.3% of the cases) for the breakdown of the labile macroalgae. This finding is consistent with a previous study conducted in streams16. Moreover, a laboratory study reported that the decomposition of labile carbon fractions (e.g. soluble compounds) is accelerated in mixtures compared to monospecific treatments, while the decomposition of refractory fractions (e.g. lignin) slows down23. The contrasting decomposition patterns of different litter species/fractions in mixtures are likely related to their differing qualities or levels of refractory compounds63,64. However, the mechanisms behind these different decomposition patterns are largely unknown.

In partial agreement with Hypothesis II, we observed a pattern of decreased seagrass mass loss with increasing litter species richness for P. japonicus, but not for Z. marina, during Stage 1 of incubation. Previous studies on the dietary preferences of gastropods and amphipod consuming seagrass and macroalgae found that the presence of nutrient-rich, palatable macroalgae may shift the resource selection of fauna towards macroalgae, which tended to decrease the consumption of nutrient-poor and unpalatable seagrass materials31,32. In the present study, we observed a high abundance of polychaetes and crustaceans and a high probability of increased macroalgal litter decomposition when seagrass was mixed with macroalgae, suggesting a clear faunal preference for macroalgae. However, this did not necessarily lead to a reduction in seagrass litter decomposition. On the contrary, the overall effect of litter mixing on seagrass litter decomposition changed from positive under lower species richness (and lower macroalgal dominance) to negative under higher species richness (and higher macroalgal dominance) (Fig. 5a). The lack of a relationship between relative faunal mass and seagrass mass loss during Stage 1 further suggests that the role of benthic fauna is ambiguous and likely context-dependent (Fig. 8a, c). These results imply that other mechanisms, such as resource complementarity and microbial priming effects65,66, may overrule and reverse negative resource selection effects, particularly when the dominance of labile litter falls below a certain threshold. Further work is required to address this issue.

In agreement with Hypothesis III, we found a positive relationship between seagrass litter decomposition and both litter species richness and the relative faunal mass during the late stage (i.e., Stage 2). Our results agree well with previous findings that an increasing presence of invertebrate fauna accelerates litter decomposition9,67. Therefore, initial differences in litter diversity and composition may have translated into varying levels of faunal presence and, consequently, different rates of litter decomposition, representing a legacy effect on litter quality54. It is noteworthy that global climate change and anthropogenic disturbances are shifting the species composition of macrophyte communities in seagrass meadows worldwide, leading to the replacement of habitat-forming seagrass species with macroalgae6,7,8. The resulting changes in the diversity and composition of macrophyte litters entering the dead organic matter pool may have significant effects on invertebrate assemblages and organic matter decomposition68,69. Since macroalgae-derived organic matter decomposes faster than seagrass-derived matter, a shift in the macrophyte community may reduce the long-term potential for organic carbon stock in meadow systems70.

Conclusion

Our study found that different decomposition rates of seagrass and macroalgal litters were likely related to differences in their nutrient content and structural complexity. These differences triggered changes in benthic faunal communities and the decomposition rates of the litter mixture of the seagrass and macroalgae. The key factors influencing the structure of faunal communities include not only species richness but also the identity and composition of the litter. We found that various taxa of benthic fauna exhibited distinct preferences for litter with different nutritional and physical qualities. Furthermore, litter-mixing effects on decomposition may vary depending on litter characteristics (e.g. identity, quality, and composition) and the decomposition stage, both in terms of magnitude and direction. The rapid transition from the initial presence to the subsequent absence of easily degradable litter, as a result of natural degradation, can trigger varied responses in the decomposition of more resistant litter with the involvement of benthic faunas. Overall, our findings emphasize the importance of recognizing the impact of various benthic faunas on the decomposition of different types of litter to understand the driving mechanisms behind the effects of litter mixing on decomposition.

Materials and methods

Study site

This study was carried out in Koje Bay (34°48′4′′N, 128°35′2′′E), a shallow embayment on the southern coast of Korea (Supplementary Fig. 8). The study site is characterized by semidiurnal tidal regimes with tidal amplitudes ranging from 1.2 to 2.5 m. The sediments are predominantly sandy, consisting of approximately 85% sand and 15% silt. Additionally, shells and rocks are scattered throughout the area, providing a hard substratum that supports the growth of macroalgae. The seagrass Zostera marina dominates the study area and forms extensive perennial meadows in the shallow subtidal zones (0–4 m in depth) of the bay. A variety of macroalgae is distributed sporadically throughout the seagrass habitats, including the brown algae Sargassum fulvellum and S. piluliferum; the green algae Ulva lactuca L, Enteromorpha prolifera, and Chaetomorpha spp.; and the red algae Gelidium amansii and Caulacanthus okamurae. Litter mixtures of seagrass and macroalgae primarily form during maximum litterfall in the summer months, dispersing throughout the seagrass meadow and adjacent bared areas.

Field litterbag experiment: effects of litter mixing on decomposition

We conducted in situ litterbag decomposition experiments with litters from two dominant seagrasses, Z. marina and Phyllospadix japonicus, and four common macroalgal species in the study area: the brown alga S. fulvellum, the green algae U. lactuca L and E. prolifera, and the red alga C. okamurae. Scuba divers collected litter material from May to July 2018, a period of maximum seagrass leaf litterfall and macroalgal detritus formation. Freshly fallen leaves of Z. marina were manually collected from the Z. marina meadow at the Koje Bay study site. Senesced leaves of P. japonicus were collected from parent plants in a mixed Z. marina and P. japonicus seagrass meadow on the southeast coast of Korea (35°10′21′′N, 129°11′54′′E). Litter materials of the four macroalgae were collected by harvesting whole macroalgal thalli from Koje Bay. Efforts were made to exclude young live tissues from the macroalgal litter wherever possible. All litter types were sorted and washed with tap water to remove any adhering particles or litter from other species immediately after collection. The materials were then air-dried at room temperature for one week and stored until the beginning of the experiment.

A series of experimental treatments were established that represented five levels of species richness (one, two, three, four, or five species per bag) and 36 different species compositions (Supplementary Table 5). The litter of one seagrass species (either Z. marina or P. japonicus) was either incubated alone or mixed with the litter of one, two, three, or four macroalgal species in species composition treatments No. 1–32. Thus, there were 1 single-species, 4 two-species, 6 three-species, 4 four-species, and 1 five-species treatments for each Z. marina and P. japonicus. Because the two seagrasses differ in their habitats, i.e., Z. marina colonizes soft sediments while P. japonicus colonizes rocks, using these two seagrass species enabled us to test the litter-mixing effects of different litter mixtures from two distinct habitats. Additionally, single-algal species treatments of 4 macroalgal species (treatments No. 33–36) were performed to investigate the effects of each macroalgal species on faunal communities. Each treatment of 36 different species compositions was sampled at day 43, 86, and 183 with 4 replicates per sampling time, resulting in a total of 432 litterbags.

Air-dried seagrass and macroalgal litter were weighed and placed in litterbags according to the experimental design. Litterbags were 15 × 20 cm in size and made of nylon with a mesh size of 1 mm designed to minimize litter loss through the mesh net but still allowed colonization by benthic fauna71. All litterbags contained 5 g (equivalent dry weight) of air-dried litter material evenly divided among the species. Specifically, the algal biomass accounted for 0, 50, 67, 75, 80, and 100% of the total biomass in the single seagrass species, two species (1 seagrass + 1 macroalgal species), three species (1 seagrass + 2 macroalgal species), four species (1 seagrass + 3 macroalgal species), five species (1 seagrass + 4 macroalgal species), and single macroalgal species treatments, respectively (Supplementary Table 5).

Litterbags were deployed in the Koje Bay study site on 10 August 2018. Four nylon ropes, each 20 m long and 5 mm in diameter, were placed side-by-side at 1 m intervals parallel to the shoreline on the sediment surface of a bare area adjacent to the seagrass stands. This bare area was 3.7–4.0 meters deep, with a bottom substratum similar to that of the seagrass meadow. All 432 litterbags (36 species composition treatments × 3 sampling times × 4 replicates) were evenly divided into four groups, each containing three litterbags for each of the 36 species composition treatments. These four groups of litterbags were randomly assigned to the four nylon ropes. The litterbags were randomly arranged, evenly spaced, and tied along each nylon rope. Floating macroalgal clumps occasionally appearing in the area were removed during each monthly visit. One litterbag for each of the 36 species composition treatments was retrieved from each rope line at each sampling time of 43, 86, and 183 days of incubation. The incubation times were chosen to facilitate the division of the two stages of decomposition based on previous studies on degradation rates of macroalgae and seagrasses39. The litterbags were brought back to the laboratory, rinsed, and sieved through a 0.25 mm mesh to separate the remaining litter and fauna from the fine sediment71. The remaining litter was extracted and sorted into seagrass or macroalgae. The remaining macroalgal litter could not be identified at the species level because of its highly fragmented and decaying state. The residue containing benthic fauna was stored in 95% ethanol for further analysis. The remaining litter was dried at 60 °C for 72 h and weighed, and the percentage of mass loss at each sampling time was calculated. Based on the high degree of decomposition observed in macroalgae (see the result section), the 43rd day was identified as the approximate cut-off point between the two decomposition stages: Stage 1 (0–43 days) and Stage 2 (44–183 days). To enable the separation of litter-mixing effects on decomposition during different decomposition stages, we calculated the stage-specific litter mass loss (%), representing the percentage of initial mass lost during a specific incubation period. The stage-specific mass loss (%) of seagrass litter was calculated as the difference in the net mass loss (%) between the beginning and end of each stage. Specifically, the specific litter mass loss (%) in Stage 1 was determined as the net mass loss (%) after 43 days of incubation. In contrast, the specific litter mass loss (%) in Stage 2 was calculated by subtracting the net mass loss (%) after 43 days from the net mass loss (%) after 183 days.

The fauna in each sample was counted and sorted into four taxa: polychaetes, crustaceans, ascidians, and molluscs. These taxa included almost all the recognized individuals. Faunal abundance of each taxon is expressed as the number of individuals per litterbag. The fauna within each litterbag was pooled and dried at 60 °C to a constant weight. The relative faunal mass in litterbags was calculated by dividing the faunal mass by the litter mass in each sample (g fauna per g litter).

Six additional 4 g litter samples of each macrophyte species were dried at 60 °C for 72 h, and the initial dry weight of the litter used for field incubation was corrected according to the ratio of oven-dried to air-dried weight. The oven-dried samples were homogenized by milling to a fine powder, and the carbon (C) and nitrogen (N) content of each litter sample were determined using a CHN elemental analyzer (Fisons NA1500, Fison Instruments, Milan, Italy). The phosphorus (P) content of the litter tissue was measured using the method described by Fourqurean et al.72.

Statistical analyses

Differences in elemental content among seagrass and macroalgal litters, as well as variations in the percentage of litter mass loss and faunal abundance among monospecific treatments of these species were examined using one-way ANOVAs. The effects of initial litter species richness (SR) and incubation time (T) or decomposition stage (S) on the percentage of litter mass loss, stage-specific litter mass losses, faunal abundance, and the relative faunal mass in each Z. marina and P. japonicus litterbag series were evaluated using two-way ANOVAs. When significant differences were found, a post hoc Tukey’s test for pairwise comparisons was performed to determine the specific locations of these differences. Regression analyses were conducted to analyze the relationships between stage-specific litter mass loss and the relative faunal mass during each stage.

Relative mixture effects (RMEs) of multi-species litter-mixing on the percentage of litter mass loss were calculated separately for seagrass and macroalgae, using the equation:

$${{{\rm{RME}}}}=({{O}}-{{E}})/{{E}}\times 100 \%$$
(1)

where E represents 73 the expected value of litter mass loss (%) estimated as the percentage of litter mass loss of a specific species when decomposed alone (i.e., treatment No. 1 [Z. marina], 17 [P. japonicus], and 33–36 [macroalgae]; Supplementary Table 5), while O represents the observed values of litter mass loss (%) in multi-species treatments (i.e., treatment No. 2–16, 18–32, and 33–36; Supplementary Table 5). One-sample Student’s t tests (two-sided) were used to test whether the RME differed significantly from zero for each treatment at different incubation times (43, 86, and 183 days).

Generalized linear models were used to identify factors linked to litter mixing, including litter species identity, composition, and richness, which could potentially affect the relative mixture effects and faunal community structures after 43 days of incubation. Note that the generalized linear model was not applied to the data collected after 86 and 183 days of incubation because of the complete loss of macroalgal litter (see results for details). Specifically, the first model tested the effect of litter species richness on RMEs in seagrass and macroalgae, and the second model tested the effect of macroalgal identity (i.e., the presence or absence of specific macroalgal species) on RMEs in seagrass. The overall macroalgal dominance (% by mass) was used as a covariate in the second model to control for potential interference. The third model tested the effect of overall macroalgal dominance, and the fourth model tested the effect of litter type (i.e., the presence or absence of overall seagrasses or macroalgae) on macrofaunal abundance. Litter species richness was treated as a covariate in the third and fourth models.

All statistical analyses were performed using SPSS ver. 25.0 software (IBM Corp., Armonk, NY, USA). The data were assessed for normality and homogeneity of variance using box plots and Levene’s test prior to performing statistical tests. If necessary, the data were square-root- or log-transformed. Significance was evaluated at p < 0.05 level in all cases. All p values were two sided.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.