Identifying Keystone Species in the Mangrove Benthic Food Web of Yanpu Bay: Integrating Stable Isotope and Network Analysis Approaches

Abstract

Keystone species play a critical role in sustaining ecosystem structure and function. Thus, accurately identifying keystone species is essential for effective biodiversity conservation. This study investigates the benthic ecosystem of Yanpu Bay’s mangroves, utilizing stable isotope analysis in combination with Bayesian mixture models and ecological network analysis to characterize trophic relationships and topological network structures, with the aim of identifying keystone species within the community. The benthic food web in this study comprised 96 connections and 27 nodes. Among them, Scartelaos histophorus preyed on eight benthos species, constituting 18.51% of the total prey sources in food web. Sedimentary organic matter (SOM) was identified as a critical food source, sustaining 17 consumer species, 62.96% of the total species recorded in the community. Quantitative analysis using criticality indices and key player problem indices identified Cerithidea cingulate, Littorinopsis scabra, Periophthalmus magnuspinnatus, S. histophorus, Bostrychus sinensis, and Metaplax longipes as keystone species. The identification of these keystone species provides valuable insights for developing targeted biodiversity conservation strategies and offers a robust scientific foundation for the restoration and sustainable management of the mangrove benthic food web.

1. Introduction

The term “keystone species” refers to species that are of primary importance in determining the structure of a community and maintaining the stability of an ecosystem [1]. These species linchpins sustain homeostatic equilibrium through dense networks of symbiotic relationships and trophic interactions. Their extirpation typically precipitates a cascading decline in biodiversity through both direct and indirect pathways, potentially triggering irreversible degradation or systemic collapse of the ecosystem [2,3]. Consequently, accurate identification and protection of keystone species are imperative for maintaining the ecological equilibrium and enhancing biodiversity.

Mangrove ecosystems currently face unprecedented threats from the combined pressures of anthropogenic activities and climate change, manifesting in biodiversity declines and diminished ecosystem service [4,5]. Consequently, a significant challenge in the realm of mangrove ecological restoration and biodiversity conservation pertains to the effective maintenance of ecosystem stability and mitigation of biodiversity loss. Keystone species have become a major research focus in the field of mangrove protection and restoration [6,7,8]. The accurate identification of potential keystone species within a community and exploration of their regulatory mechanisms within the ecosystem will facilitate the formulation of management strategies for priority conservation groups. Such identification and exploration will assist in assessing the relationship between species and community structures, thereby providing a scientific basis for the restoration and management of mangrove ecosystems.

The identification and selection of keystone species represent a complex and challenging undertaking [9]. In preliminary studies, alterations in species diversity and community structure preceding and following the artificial deletion or introduction of a species were typically compared to ascertain whether the species constituted a keystone species within the community. For instance, Paine [10] used controlled simulation experiments to ascertain whether the ochre starfish constitutes a keystone species within the intertidal zone community. However, most of these experimental methods are limited and cumbersome. Moreover, the direct removal or introduction of a species can have an immeasurable impact on the ecosystem. In this regard, some ecologists used field survey data to calculate the effect of a change in the abundance of one species in a community on another species to assess the strength of the interaction between that species and other species in the community, thereby determining whether that species is a keystone species in the community [11,12,13]. However, it should be noted that such experiments require years of data accumulation and can only focus on a few species in the community. Therefore, prior assumptions must be made to select the species required for the experiment, and artificial selection of some species will lead to a bias in the perception of keystone species [13]. In light of the aforementioned points, several ecologists have proposed a methodology for identifying keystone species within communities. This methodology involves calculating the strength of the effect of a species on the characteristics of the community or ecosystem in question. It should be noted that the identification method had gradually evolved from a qualitative to a quantitative method. For instance, Power et al. [14] and Hurlbert [15] proposed the use of a community importance index and a functional importance index, respectively, to calculate the importance of species in an ecosystem. These methods offer a more intuitive approach to demonstrating the degree of species importance through quantitative calculation; however, it should be noted that they are not universally applicable but only to the calculation of specific communities. For instance, the community importance index method encounters certain difficulties in measuring subtle changes in plant community diversity.

The identification of keystone species is currently a focal point in community ecology research and a fundamental challenge in ensuring targeted protection of biodiversity [16]. Numerous researchers have conducted extensive studies to identify keystone species in communities. Paine [17] was the first to identify keystone species using a controlled analog removal method. This method is theoretically feasible; however, directly removing or introducing species into natural ecosystems is challenging. Subsequently, Yeaton [18] and Wootton [12] proposed the competitive advantage hindrance and species interaction strength method, respectively. Through field experimental simulations, these methods have been shown to better compensate for the shortcomings of controlled simulation removal methods. However, both methods require long-term experimental observations to determine the key species in a community, which is not a practical approach. In the field of community ecology, scholars such as Power et al. [14], Hurlbert [15] and Khanina [19] proposed a series of methods, including the community importance index, functional importance index, and equivalent dominant species methods. Since then, community structure analysis has been employed to identify key species in communities, and it has gained considerable popularity in the study of plant communities. However, this method has been criticized for its limited scope of application in simple biological communities with limited mobility [15,20]. In recent years, topological ecological network analysis has emerged as a powerful framework, integrating graph theory and ecology to assess species interactions. By quantifying degree strength, centrality indices, keystone indices, and network stability metrics (e.g., key player problem), this approach evaluates species’ structural and functional roles within food webs [21]. It subsequently determines the key species in the topological network that are instrumental in establishing and maintaining the network structure and stability [22]. Consequently, ecological network analysis has emerged as a potent tool to identify key species within communities. For example, Xing et al. [23] used this method to confirm that Exopalaemon modestus and Bellamya aeruginosa were keystone species in Xingkai Lake, China. Similarly, Wang et al. [24] explored the impact of the removal of key species, such as Muraenesox cinereus and Leptochela gracilis on the stability of the food web in the East China Sea. In summary, ecological network analysis quantitatively describes the interactions between species in a food web through topological network indices, thereby providing new perspectives for research on the stability of food web structures, simplifying food webs, and protecting species diversity. Thus, this is an effective method for screening keystone species.

Despite the well-established application of topological network analysis in aquatic food web studies, study on mangrove benthic ecosystems in intertidal zones remains limited. This gap arises from two main challenges: (1) the complex three-dimensional structure of mangrove root systems (prop roots and buttress roots) creates physical barriers that, coupled with tidal dynamics, render benthic sampling more difficult than open-water fish collection [25]; and (2) the high dietary diversity of benthic organisms, along with the limitations of traditional stomach content analysis in resolving feeding relationships, complicates the accurate reconstruction of trophic interactions [26]. To address these challenges, this study integrates stable isotope analysis with Bayesian mixing models to elucidate benthic feeding relationships. Additionally, ecological network analysis was applied to construct the community’s topological structure and identify keystone species. The study had two primary objectives: (1) to clarify the trophic linkages among mangrove benthos, and (2) to determine key species within the community. The findings of this study contribute to the understanding of the community-level effects of keystone species within mangrove benthic food web and provide a scientific basis for the restoration and conservation of mangrove biodiversity.

2. Materials and Methods

2.1. Study Area

Yanpu Bay (120°27′44.046″, 27°12′56.171″) is located at the border between Zhejiang and Fujian provinces and is the only geographical unit in Zhejiang province that experiences a subtropical climate. The bay has extensive beaches and mudflats, and the sea covers the area for extended periods of time, providing an environment that is frequently flooded and suitable for the growth of mangrove plants. Following the successful introduction of the Kandelia candel in 2015, a substantial mangrove area has been established, constituting the largest bay mangrove wetland in the northern periphery of China [27,28]. Currently, the mangrove ecosystem in this region is in a mature phase of development and is vulnerable to external pressures such as environmental changes and anthropogenic disturbances. The surrounding sea areas and tidal flats have long been affected by factors, such as sea hunting and extensive aquaculture, resulting in habitat fragmentation and an unstable community structure.

2.2. Sample Collection

Fifteen stations were established in the Yanpu Bay mangrove area in May and August 2021 to collect sediment organic matter (SOM), particulate organic matter (POM), zooplankton, phytoplankton, mangrove leaf litter, and benthos (Figure 1). At each station, a 50 mL disposable syringe with the needle cut off was used to collect SOM; 1 L plastic bottles were used to collect surface seawater at stations Y3, Y4, Y8, Y12, and Y13; zooplankton and phytoplankton were collected at high tide using shallow water type III and type II zooplankton nets; mangrove leaf litter was collected by selecting four trees per station; and a cage net was placed at each station to collect benthos for 2 days (four high and low tides). All samples were frozen, preserved in a vehicle-mounted freezer, and transported to the laboratory for subsequent analysis. The benthic organisms identified to species level in the laboratory through examination under a stereomicroscope, with reference to the Fauna Sinica.

2.3. Stable Isotope Analysis

In the laboratory, three to five individuals of varying sizes from the same species were selected. The experimental samples were dissected from the following tissues: dorsal muscle in fish, cheliceral muscle in crustaceans, and foot muscle in shellfish. The samples were then thoroughly acidified with 1 mol/L hydrochloric acid to remove inorganic carbon. POM samples were pumped and filtered through Whatman GF/F membranes that had been pre-scorched in a muffle furnace at 450 °C. Apoplastic zooplankton samples were washed using a small amount of distilled water to remove impurities and then placed under a dissecting microscope to select larger zooplankton individuals. Phytoplankton samples were filtered through a 160 µm sieve after removing zooplankton and debris. The samples were subsequently lyophilized and ground.

The samples were analyzed for isotopic composition of C (δ¹³C) and N (δ¹⁵N), using an elemental analyzer coupled with a stable isotope ratio mass spectrometer (EA-IRMS; Flash EA 112; Thermo Finnigan MAT 253, Thermo, Waltham, MA, USA), with a Conflo III interface (Thermo Finnigan, San Jose, CA, USA). Stable isotope ratios (δ) were expressed as parts per thousand (‰), relative to the C and N reference materials (Vienna Pee Dee Belemnite (V-PDB) for ¹³C, and atmospheric air nitrogen (N2-atm) for ¹⁵N). The following equations were used:

$$\mathsf{\delta }\mathrm{X} = \left({\mathrm{R}}_{\mathrm{sample}}/{\mathrm{R}}_{\mathrm{standard}}-1\right){\mathrm{R}}_{\mathrm{sample}} \times {10}^3$$

In this context, X represents the measured stable isotope (either¹³C or ¹⁵N), Rsample denotes the sample isotope ratio, Rstandard signifies the standard substance isotope ratio.

2.4. Diet Analysis

A Simmr mixing model was employed to estimate the dietary composition of each species. The model inputs included the carbon and nitrogen stable isotope ratios of the consumer tissues, along with the means and standard deviations of these isotopes for all potential food sources. A default Dirichlet prior (with concentration parameters set to 1) was applied, which assumes equal prior contributions from all sources. For Markov chain Monte Carlo (MCMC) sampling, a burn-in phase of 5000 iterations was conducted, followed by 10,000 productive iterations. Convergence of the model was confirmed, as all Gelman–Rubin diagnostic values were below 1.05. The trophic enrichment factor (TEF) was calculated for each retained species with an average contribution greater than 5% of the model output. This was performed to identify potential food sources for consumers. The TEF in this study was 1.0 ± 0.5‰ for carbon and 3.4 ± 1.0‰ for nitrogen [29,30]. Potential food sources were identified through a combination of literature review and Monte Carlo simulations [31], and foods within the 95% confidence interval of the calculated results were selected as potential food sources for the target consumers [30].

2.5. Ecological Network Analysis

Based on the food web topology, the topological index of a species was calculated using ecological network analysis, following the methodology of Wang et al. [24]. The degree, including the in-degree and out-degree, indicates the numbers of predators and prey, respectively [32]. The centrality indices include information centrality, closeness centrality, and betweenness centrality. Information centrality measures a species’ ability to mediate information transfer. Higher values indicate greater capacity to facilitate interspecific communication, which consequently enhances their potential to influence information exchange dynamics within ecological communities [32,33]; Closeness centrality is indicative of the efficiency with which information is transmitted within a community. Higher values indicate stronger species-level transmission capabilities [34,35], and centrality is a measure of the control species in a community exerted over information exchange. Higher values signify greater control, which is critical for community stability. The topological importance index (TI) is a measure of the capacity of a species to disseminate information within a community. A high TI signifies a strong propensity for information propagation [36]. Keystone indices (K) encompass the bottom-up keystone index (K_b) and top-down keystone index (K_t), which denote the upward and downward effects of species, respectively [20]. The key player problem index (KPP) comprises two components: KPP-1 and KPP-2. KPP-1 calculates two indices: fragmentation (F) and distance-weighted fragmentation (^DF). KPP-2, conversely, calculates the distance-weighted reach (^DR), which indicates the maximum range within which the screened species can transmit information to the community. These indices represent the impact of removing the screened species from the community-on-community fragmentation. The greater the values of F and ^DF, the greater the fragmentation of the community [37,38]; the greater the ^DR, the greater the maximum range of information transmission by the screened species in the community. Both indices indicated the effect of the removal of the screened species from the community-on-community fragmentation. The greater the values of F and ^DF, the greater the fragmentation of the community [39].

The analysis was conducted utilizing the packages “SIMMR” and “MixPolySim” in R software 4.5.1, employing Bayesian mixed models and Monte Carlo simulation analysis, respectively. The calculation of topological indicators was facilitated using Ucinet6, Keyplayer1.44, and CoSBiLab Graph1.0; Cytoscape3.9.1 was employed to plot the food web topology and its topological indicators.

3. Results

3.1. Food Web Structure

A total of twenty benthos species were recorded in this study, with most benthic organisms primarily dependent on plankton (phytoplankton, zooplankton), organic debris (POM, SOM), and leaf litter as their main food sources. Among these, S. histophorus had eight food sources, accounting for 29.62% of the total, followed by Metopograpsus quadridentatus and B. sinensis, which had five food sources, accounting for 18.51%. SOM was identified as the food source for 17 species, accounting for 62.96% of the total, followed by leaf litter and phytoplankton, which were consumed by 14 and 13 species, accounting for 50.85% and 48.14% of the species, respectively. Further analysis of the community feeding relationships revealed that the food web comprised 96 connections, with a maximum of four links in the food chain and 27 nodes (Figure 2). Food webs are characterized by three trophic levels. Mollusks such as Littoraria melanostoma and Cerithidea rhizaphorarum occupy the base of the food web, whereas fish such as P. magnuspinnatus, Periophthalmus modestus and B. sinensis occupy the pinnacle of the food chain.

3.2. Centrality and Topological Importance Indices

The difference between the centrality and topological importance indices of benthos was significant (Figure 3). The degree, out-degree, and information centrality of C. cingulata were the highest at 10.00, 8.00, and 5.11, respectively. This indicates that C. cingulata produces the most complex feeding relationships and that most benthic animals (e.g., B. sinensis) feed on C. cingulata, which thus has the greatest impact on interspecific information transfer. Gastropod (L. scabra) exhibited the highest betweenness centrality of 5.99, confirming its predominant capacity to regulate interspecific information exchange and its pivotal role in preserving community structural stability. Fish (P. magnuspinnatus) is close to the maximum centrality, with a value of 9.66, indicating that it has the strongest ability to transmit information in the community and the fastest speed of information transmission. Similarly, S. histophorus exhibits the highest values for both in-degree and TI, at 8.00 and 2.67, respectively. These findings suggest that S. histophorus is the most abundant food source and has the strongest capacity for information propagation.

3.3. Keystone Indices

The keystone indices of S. histophorus and the top-down keystone index were the largest, at 3.98 and 3.8, respectively. This suggests that S. histophorus plays a pivotal role in the transfer of information and energy flow between communities, primarily influencing the interspecific relationships of the community by feeding on other species to affect the interspecific relationships in the community. The bottom-up keystone index of C. cingulata was the largest, at 1.98, indicating that mainly affected the community through being eaten by other species (Figure 4).

3.4. Key Player Problem Indices

During the KPP-1 analysis, benthos manifested when k = 4, with the four nodes representing S. histophorus, B. sinensis, SOM, and phytoplankton. This suggests that community fragmentation was most pronounced when these four nodes were removed from the food web, with a value of 0.55 and a ^DF of 0.67. In the subsequent KPP-2 analysis, when k = 3, the presence of benthos was observed, including M. longipes, SOM, and zooplankton (

Table 1. KPP-sets of the mangrove benthic food web of Yanpu Bay (at k = 1, 2, 3, and 4).

	k	Filtered Out Nodes
Key Player Problem-1	1	Soil organic matter				F = 0.07
	2	Soil organic matter	Phytoplankton			F = 0.08
	3	Particulate organic matter	Phytoplankton	Zooplankton		F = 0.16
	4	Phytoplankton	Soil organic matter	S. histophorus	B. sinensis	F = 0.55
	1	Soil organic matter				DF = 0.03
	2	Soil organic matter	Phytoplankton			DF = 0.49
	3	Particulate organic matter	Phytoplankton	Zooplankton		DF = 0.52
	4	Phytoplankton	Soil organic matter	S. histophorus	B. sinensis	DF = 0.67
Key Player Problem-2	1	Soil organic matter				DR = 77.8%
	2	Soil organic matter	Phytoplankton			DR = 92.6%
	3	Soil organic matter	Zooplankton	M. longipes		DR = 100%
	4	Soil organic matter	Zooplankton	Meretrix meretrix	B. sinensis	DR = 100%

). This finding suggests that the elimination of these three nodes from the food web optimized the transfer of information, achieving 100% coverage.

4. Discussion

4.1. Diet Determination

Dietary analysis is a pivotal component of elucidating the feeding relationships of marine organisms and identifying keystone species. In the early stages of research, stomach content analysis was frequently employed in conjunction with data from the literature to ascertain the feeding habits of organisms. This method has been extensively used for larger organisms such as large-sized fish. However, the application of stomach content analysis for benthos has considerable limitations. The difficulties inherent to the process of obtaining stomach contents from small organisms, the predominance of debris in stomach contents, and the inability to identify stomach contents with certainty are well documented [40]. Consequently, stomach content analysis is ill-suited to provide reliable insights into the long-term dietary patterns of benthic organisms. Recent years have seen notable advancements in the field of stable isotope analysis, driven by the deepening of theoretical research on thermodynamics and the enhancement of mass spectrometer instruments. This development has led to the widespread use of stable isotope analysis to determine interspecific feeding relationships and the construction of food webs [41]. The unique enrichment process of stable carbon and nitrogen isotopes in the food web has been shown to accurately reflect feeding over long periods, thereby clarifying the trophic position of food web organisms and energy flow. For example, Akbari et al. [42] utilized the Stable Isotope Mixing Model (SIMMR) to assess macrobenthic dietary composition in Persian Gulf mangroves, revealing that benthic microalgae accounted for as much as 40% of their food sources. Despite their utility, Bayesian mixing models present certain methodological constraints. The unrestricted inclusion of potential food sources risks extending consumer isotope signatures beyond the source-derived mixing space, which obscures result interpretation [43]. To address this, researchers must first conduct Monte Carlo simulations to verify that consumer isotope values fall within the distribution range of potential food sources [31]. Since Bayesian models utilize discrete isotope values rather than means for target consumers, the “MixPolySim” package has emerged as the preferred tool for calculating model priors [44]. This package effectively handles discrete isotope values while accounting for trophic enrichment factors, making it the standard approach for prior calculation in most contemporary studies [31].

This study utilized a Bayesian mixture model to precisely quantify the contributions of potential benthic food sources. The approach effectively accounts for uncertainties in isotopic variation and fractionation coefficients, establishing a solid foundation for subsequent identifying keystone species. Analysis revealed soil organic matter (SOM) as a significant food source for mangrove benthos, consistent with findings by Yang et al. [45] in Lianzhou Bay mangroves where SOM contributed substantially (11.9%) to macrobenthic diets. Among the species examined, the diets of the green mudskipper and Chinese black sleeper comprised eight distinct species, a finding that may be associated with the growth and development of mangroves in Yanpu Bay. This phenomenon has the potential to diversify the organic matter sources present in sediments, as previously documented [46]. Consequently, detritivorous and high-trophic-level carnivorous benthic animals have access to a wider range of organic carbon sources. Litter constitutes a significant food source for various crabs, including Chiromantes haematocheir and Uca arcuata. Research has demonstrated that certain plant-consuming crabs of the Sesarmidae and Uce exhibit a high rate of mangrove litter removal [47,48]. For instance, up to 84.3% of the fallen leaves in Brazilian mangroves are consumed by Ucides cordatus [49]. Furthermore, the “litter export theory” of mangroves posits that fallen leaves constitute the primary source of organic carbon for large benthic animals in the forest [49,50], thereby contributing significantly to the high primary productivity of mangrove ecosystems.

4.2. Keystone Species Identification Method Selection

The identification of keystone species is currently a focal point in community ecology research and a fundamental challenge in ensuring targeted protection of biodiversity [51]. Numerous researchers have conducted extensive studies to identify keystone species in communities. Paine [17] was the first to identify keystone species using a controlled analog removal method. This method is theoretically feasible; however, directly removing or introducing species into natural ecosystems is challenging. Subsequently, Yeaton [18] and Wootton [12] proposed the competitive advantage hindrance and species interaction strength method, respectively. Through field experimental simulations, these methods have been shown to better compensate for the shortcomings of controlled simulation removal methods. However, both methods require long-term experimental observations to determine the key species in a community, which is not a practical approach. In the field of community ecology, scholars such as Power et al. [14], Hurlbert [15] and Khanina [19] proposed a series of methods, including the community importance index, functional importance index, and equivalent dominant species methods. Since then, community structure analysis has been employed to identify key species in communities, and it has gained considerable popularity in the study of plant communities. However, this method has been criticized for its limited scope of application in simple biological communities with limited mobility [15,20]. In recent years, topological ecological network analysis has emerged as a powerful framework, integrating graph theory and ecology to assess species interactions. By quantifying degree strength, centrality indices, keystone indices, and network stability metrics (e.g., key player problem), this approach evaluates species’ structural and functional roles within food webs [21]. It subsequently determines the key species in the topological network that are instrumental in establishing and maintaining the network structure and stability [22]. Consequently, ecological network analysis has emerged as a potent tool to identify key species within communities. For example, Xing et al. [23] used this method to confirm that Exopalaemon modestus and Bellamya aeruginosa were keystone species in Xingkai Lake, China. Similarly, Wang et al. [24] explored the impact of the removal of key species, such as Muraenesox cinereus and Leptochela gracilis on the stability of the food web in the East China Sea. In summary, ecological network analysis quantitatively describes the interactions between species in a food web through topological network indices, thereby providing new perspectives for research on the stability of food web structures, simplifying food webs, and protecting species diversity. Thus, this is an effective method for screening keystone species.

The study constructed a mangrove benthic food web by analyzing species’ feeding relationships and employed ecological network analysis to quantify the topological indices of individual species. The analysis revealed that species exhibiting higher degree values consistently demonstrated correspondingly higher out-degree values. This correlation aligns with findings from keystone species identification studies in other aquatic ecosystems [51,52]. In ecological network analysis, degree value typically reflects the total number of trophic interactions for a target species, while out-degree value specifically quantifies the number of prey species for which the target species acts as a predator [35]. The observed parallel trends between a species’ out-degree and degree values suggest that its contribution to food web stability primarily stems from predatory relationships. Conversely, when in-degree (representing prey relationships) and degree values show similar patterns, this indicates the species serves as a crucial prey resource in the ecosystem. A notable example of this latter pattern was documented by Wang et al. [24], who identified Trichiurus lepturus (degree = 39, in-degree = 32) as a keystone species in the East China Sea based on its high degree and in-degree values.

The ecological network analysis demonstrated that C. cingulata occupies a keystone position in the food web, while L. scabra plays a critical role in regulating interspecific information exchange. Pereira et al. [53] demonstrated that species occupying intermediate trophic levels frequently function as keystone species within their communities because of the complexity of their predation and prey relationships. Plant-consuming animals such as C. cingulata and L. scabra play a pivotal role in the transfer of substances to consumers within the mangrove ecosystem [54]. Both species function as primary consumers in the mangrove benthic food web, occupying an intermediate trophic position, while serving as prey for higher-level predators such as B. sinensis and S. histophorus. The trophic levels of connection for these species were 10 and 9, respectively, indicating their roles as predators. Numerous species have trophic relationships, suggesting a complex interaction network within the ecosystem that facilitates the transfer of substances and energy produced by primary producers to consumers at higher trophic levels. Consequently, they play a pivotal role in regulating predator density through bottom-up effects, thereby affecting information exchange and control. This finding is consistent with the conclusions of most previous studies. Gasalla et al. [55] demonstrated the Loligo plei inhabiting the waters of the inner shelf of the South Brazilian Bay functions as a keystone species, exerting control over the transfer of matter and energy from lower to higher trophic levels through bottom-up effects. The vertical migration of C. cingulata and L. scabra in the mangrove area, synchronized with tides, may serve as the primary mechanism regulating the transfer of organic matter in mangroves, thereby significantly impacting community information exchange [56]. Furthermore, C. cingulata was identified as a prevalent species in the mangrove ecosystem of Yanpu Bay. The notable environmental resilience and expeditious reproductive cycle of this species facilitate the sustenance of predators, such as B. sinensis, S. histophorus, and E. rossii. Therefore, C. cingulata serves as a vital keystone species within mangrove benthic communities, exhibiting ecological significance in maintaining food web structure.

The study revealed that S. histophorus stabilizes community structure via bottom-up effects and possesses the greatest capacity for information propagation, whereas P. magnuspinnatus dominates as the most efficient information disseminator. Notably, both S. histophorus and P. magnuspinnatus are classified as high-trophic-level fishes, occupying the pinnacle of the mangrove food web as the top predators within the benthic food web. Downward effects have been identified as pivotal factors influencing the community structure and function of mangrove ecosystems [57]. When high-trophic-level predators are present, their impact on the density of food sources through predation can lead to alterations in the community structure and function. The skin and oral mucosa of S. histophorus and P. magnuspinnatus are characterized by a high density of capillaries. These vascularized structures facilitate direct exchange with the atmosphere, enabling respiration and permitting movement between aquatic and terrestrial environments [58]. In addition, S. histophorus and P. magnuspinnatus possess distinct pectoral fin structures and exhibit remarkable jumping and climbing abilities, which are indicative of their well-developed motor skills. These physiological and morphological adaptations collectively underpin the species’ amphibious survival strategies, providing competitive advantages in mangrove food web information transfer and explaining their elevated criticality indices within the benthic trophic network.

During KPP-1 operation, the benthic food web experienced a loss of SOM, phytoplankton, and zooplankton. However, there was minimal change in network fragmentation and the selected nodes were predominantly primary producers or consumers. This observation suggests that changes in primary producers have a negligible effect on communities. When the food web selected four nodes, network fragmentation abruptly increased, and nodes manifested benthos, indicating that the loss of predators, such as S. histophorus and B. sinensis, exerts a substantial impact on food web structure. Mangrove ecosystems represent one of the most productive marine ecosystems in terms of primary productivity, supporting abundant and diverse primary producers [59], thereby the loss of any single primary producer species typically has minimal impact on the overall ecosystem stability. In contrast, predators are typically identified as keystone species given their high trophic level, broad prey spectrum, and low abundance, making them sensitive indicators of community shifts. Consequently, the absence of S. histophorus and B. sinensis would result in a decline in the interconnectedness of the food web as a whole, leading to an escalation of network fragmentation. These findings underscored the critical role of these predators in preserving the structural integrity and stability of mangrove benthic food web.

The results of the KPP-2 algorithm demonstrated that the incorporation of SOM and phytoplankton nodes enhanced information transmission by 77.8% and 92.6%, respectively. These findings suggest that the primary food sources of benthos in mangroves are SOM and phytoplankton. The incorporation of M. longipes resulted in 100% distance-weighted accessibility to the entire community, signifying its potential for information transmission to all nodes within the community. The present study determined that the primary food sources of M. longipes were leaf litter, SOM, and mangrove hermit crab limpets. These food sources are preyed upon by high-trophic-level fish species such as S. histophorus and B. sinensis. M. longipes occupies a central position within the food web, serving as a bridge in the mangrove benthic ecosystem. In other marine ecosystems, species occupying intermediate trophic levels are often identified as keystone species due to their disproportionate influence on community structure. For example, Wang et al. [60] also demonstrated this phenomenon in the southern Zhejiang Sea, where the removal of striped bass (an intermediate trophic-level species) led to an 86.8% increase in distance-weighted arrival rates, confirming its keystone role. Similarly, M. longipes exhibited a strong capacity for mediating energy flow and facilitating information transfer within mangrove benthic food web, underscoring its pivotal ecological function.

In the context of keystone species research, studies have historically prioritized the identification of rare and endangered species within a community, overlooking the significance of common species. This bias can lead to certain limitations in the research findings [61,62]. However, in actual communities, the key role of species in food web structure and function is not directly related to their abundance. For instance, Wang et al. [24,60] identified keystone species in the fish community off the coast of southern Zhejiang province and found that the ribbon fish was both the dominant and keystone species in the fish community. The present study also found that the key species of the community included both common species such as M. longipes and C. cingulata [63,64] and rare species such as B. sinensis and S. histophorus [65], indicating that species abundance was not the main factor affecting the composition of key species in the community. Additionally, keystone species research has gradually evolved from focusing on key predators to studying species at different trophic levels in communities [66]. For instance, Leander modestus is a pivotal species in the Xingkai Lake food web despite occupying a low trophic level [23]. The current study reinforced this standpoint by identifying six ecologically critical species in mangrove benthic food web: C. cingulata, L. scabra, and M. longipes (middle to low trophic levels), along with P. magnuspinnatus, S. histophorus, and B. sinensis (higher trophic levels).

5. Conclusions

This study employed systematic analysis of mangrove benthic food web trophic networks in Yanpu Bay, identifying six keystone species (C. cingulata, L. scabra, P. magnuspinnatus, S. histophorus, B. sinensis, and M. longipes) through quantitative keystone index evaluation. These ecologically significant species maintain structural stability in benthic communities by mediating energy flows through trophic cascades, regulating interspecific interactions, and enhancing ecological network connectivity. The findings provide valuable insights for future research directions, including the following: (1) developing a prioritized conservation framework for keystone species, (2) elucidating the spatial distribution patterns of keystone species across different habitats, and (3) systematically evaluating the ecological consequences of keystone species loss on both community biodiversity and ecosystem stability.