陆海交互作用下滨海湿地造就了独特的高程、水盐等环境梯度，同时形成了盐沼、泥滩、近海水域等景观异质、功能多样的不同类型生态系统，对生境维持、功能调节、生物多样性保护等均发挥着重要作用。近年来，气候变化、风暴潮等自然因素，及上游水利工程密集建设、大量围填海等人类活动的双重胁迫，直接改变了滨海湿地固有的水分、盐分梯度分布特征，使得湿地水盐平衡变得更加敏感且脆弱；而水盐等环境梯度的变化进一步影响了滨海湿地生物种群的空间分布和物种交互作用，导致部分湿地生态系统结构与功能退化日益严重，亟需开展滨海湿地的生态保护。目前已有诸多学者针对滨海湿地植被、浮游生物、底栖动物、鱼类等不同生物群落的空间分布开展了广泛研究，为滨海湿地的生态保护提供了科学支撑。然而现有研究往往关注变化环境下单一生物群落的生态响应机制，忽略了不同生物群落物种之间的相互作用，鲜有考虑环境梯度上滨海湿地生态系统结构和功能的动态反馈，且生态系统的稳定维持机制尚不清晰。
食物网耦合了多营养层级生物群落及其物种之间的相互作用，对揭示生态系统结构功能的响应机制提供良好的研究手段。本研究聚焦于黄河三角洲的滨海湿地，采用野外调查、室内检测、模型构建和过程模拟等多手段，阐明黄河三角洲湿地食物网营养结构特征、能量流动特性和食物网稳定性的空间变化规律，揭示变化环境下湿地食物网的稳定维持机制，提出促进物种共存、群落稳定的有效策略。取得以下主要研究结果：
（1）分析了黄河三角洲陆海梯度上多营养层级生物群落的空间分布特征，明确了多营养层级生物群落在环境梯度上差异化的响应规律。
沿陆海梯度开展了黄河三角洲淡水湿地、高地、盐沼、泥滩和近海水域等不同生境环境要素和生物群落的野外调查，阐明了多营养层级生物群落（植物、大型底栖动物和鱼类群落）的空间差异特征及响应规律。研究发现黄河三角洲陆海梯度上植物群落结构具有明显的空间差异性，由陆至海优势植物依次为芦苇、柽柳、盐地碱蓬、互花米草和海草。淡水湿地和高地的大型底栖动物群落以昆虫纲为主（分别占比92.9%和72.9%），盐沼湿地大型底栖动物群落组成较为多样化，泥滩和近海水域湿地双壳纲占据主要优势（分别占比86.7%和62.5%）。陆海梯度上鱼类群落结构的组成差异较小。植物和大型底栖动物群落的物种丰富度沿陆海梯度变化显著（植物：p=0.047；大型底栖动物：p=0.005）。植物群落的物种丰富度沿陆海梯度显著下降，大型底栖动物群落的物种丰富度由淡水湿地到盐沼湿地逐渐减小，从盐沼湿地到近海水域逐步增加，与Remane曲线一致。陆海梯度上鱼类群落的物种丰富度差异较小（p=0.32）。由陆向海，黄河三角洲湿地植物、大型底栖动物和鱼类的生物量均具有显著性差异（植物：p=0.001；大型底栖动物：p<0.001；鱼类：p<0.001），其中，植物生物量呈降低趋势，大型底栖动物生物量呈增加趋势，近海侧湿地鱼类的生物量较高于近岸侧鱼类的生物量。
（2）构建了黄河三角洲陆海梯度上湿地食物网的营养结构关系模型，发现陆海梯度上湿地食物网营养结构的定性拓扑值逐渐增大，食物网的营养结构逐渐复杂。
进一步选取黄河三角洲陆海梯度上分别位于高地、盐沼、泥滩和近海水域的芦苇、盐地碱蓬、互花米草和海草生境为研究区域，野外采集食物网要素并测定碳氮稳定同位素含量，构建了食物网的营养结构关系模型，定量化食物网营养结构的定性拓扑指标。研究发现，除芦苇和海草区域碳源的δ15N值外，不同区域碳源（包括沉积碎屑、水体悬浮颗粒物、植物、底栖微藻和浮游植物）、消费者（包括浮游动物、大型底栖动物和鱼）的δ13C和δ15N值均有显著性差异（p<0.05）。芦苇区碳源和消费者的δ13C值分布范围分别为-25.74至-6.84‰和-25.06至-20.54‰，δ15N值为1.61至3.90‰和2.66至8.58‰。盐地碱蓬区域碳源和消费者的δ13C值分布范围分别为-28.99至-5.46‰和-20.01至-12.24‰，δ15N值为4.05至6.26‰和5.54至11.89‰。互花米草区域碳源和消费者的δ13C值分布范围分别为-19.90至-7.00‰和-22.44至-10.46‰，δ15N值为4.95至6.65‰和6.46至11.72‰。海草区域碳源和消费者的δ13C值分布范围分别为-20.61至-5.62‰和-27.01至-11.24‰，δ15N值为5.02至6.68‰和4.32至9.61‰。芦苇、盐地碱蓬、互花米草和海草区域消费者营养位置范围分别为1.36~3.25、1.91~3.94、1.65~3.32、1.35~3.62。芦苇区域营养位置最高的是虾虎鱼，其他区域均为鲈鱼。双壳类、腹足类、蟹类和鲈鱼在不同区域的食源偏好差异较大。海草和互花米草区域食物网的链接数和链接密度均高于盐地碱蓬和芦苇区域。
（3）发展了平衡态下滨海湿地食物网的能量流动模型，发现黄河三角洲陆海梯度上湿地食物网均主要由碎屑食物链贡献、符合金字塔结构，但由陆向海总碳流通量逐渐增加。
在食物网营养结构关系模型的基础上，构建了平衡态下食物网的能量流动模型，获得了维持食物网能量流转和物种共存的最低碳流通量和营养传递效率，定量化基于碳流通量加权的拓扑特征指标值。研究发现，黄河三角洲陆海梯度上食物网的总碳流通量逐渐增加，依次为10.52 g m-2 yr-1（芦苇区）、75.58 g m-2 yr-1（盐地碱蓬区）、337.73 g m-2 yr-1（互花米草区）、336.64 g m-2 yr-1（海草区）。四个区域食物网的能量流动模式均符合金字塔结构，初级消费者（植食者和碎屑食者）的碳流通量占消费者总碳流通量的81%（芦苇区）、72%（盐地碱蓬区）、92%（互花米草和海草区）。芦苇、盐地碱蓬和海草区碎屑食物链对能流的贡献比例约55%，然而互花米草区碎屑食物链贡献了84%的能量流动。芦苇区底栖碳源对食物网能流的贡献比例（55%）略高于浮游碳源（45%），其他区域浮游碳源的贡献比例较高（59~64%）。各区域食物网要素的生物量和生产量随营养级的分布均符合金字塔结构，芦苇、盐地碱蓬、互花米草和海草区食物网的营养传递效率均值分别为12.9%、11.2%、7.7%和9.3%。芦苇、互花米草和海草区食物网的加权链接密度和加权连通度均低于其对应定性拓扑指标，盐地碱蓬区域食物网相反。
（4）建立了滨海湿地食物网稳定性的定量化模型，发现海草区域食物网稳定性最高，互花米草区域稳定性最低，食物网稳定性主要由集团内捕食模块形成的杂食性环决定，且黄河三角洲湿地实际食物网的稳定性显著高于随机食物网。
在食物网营养结构和能量流动研究的基础上，构建了基于捕食者与被捕食者之间相互作用强度矩阵的滨海湿地食物网稳定性定量化模型，识别了食物网的相互作用环并阐明食物网稳定性与相互作用环的环重之间的关系，探究滨海湿地实际食物网与随机食物网的稳定性、环重的差异特征。研究发现，芦苇、盐地碱蓬、互花米草和海草区食物网的相互作用强度（绝对值，log10）分布范围分别是-6.7~0.8 yr-1、-5~1.6 yr-1、-4.2~1.5 yr-1、-4.9~1.8 yr-1。海草区食物网稳定性最高（Re(λmax)=0.009 yr-1），其次为芦苇（Re(λmax)=0.07 yr-1）和盐地碱蓬区域（Re(λmax)=0.09 yr-1），互花米草区食物网稳定性最低（Re(λmax)=0.37 yr-1）。食物网稳定性随食物网节点数、链接密度的增加逐渐降低（R2=0.49，p<0.001和R2=0.22，p<0.001），随加权连通度的增加逐渐增加（R2=0.46，p<0.001）。食物网的Re(λmax)值和食物网内由集团内捕食（IGP）模块形成的杂食性环的最大环重有显著正相关关系（R2=0.84，p<0.001），即最大环重值越高，食物网稳定性越低。芦苇、盐地碱蓬区域最抑制食物网稳定的相互作用环均由浮游植物、浮游动物和虾类组成，互花米草区域由浮游植物、浮游动物和鲫鱼组成，海草区域为浮游植物、浮游动物、双壳类、腹足类、其他甲壳类、蟹类和多毛类组成的相互作用环。对相互作用强度矩阵随机化处理发现黄河三角洲实际食物网的稳定性显著高于随机食物网，实际食物网的最大环重显著低于随机食物网（p<0.001）。
（5）基于能量学参数改进了食物网相互作用环的环重算法，构建了集团内捕食模块的种群动力模型，揭示了物种能量特性和食物网能量流动平衡耦合约束下的稳定维持机制。
阐明了实际食物网和随机食物网正反馈杂食性环内相互作用强度的差异性分布特征，揭示了相互作用强度的分布模式对食物网稳定的作用机制；基于能量学参数改进正反馈杂食性环的环重算法，揭示了食物网稳定维持的能量学机制；构建了IGP模块的种群动力模型，验证稳定维持机制的普适性。研究发现，在实际食物网中，正反馈杂食性环内捕食者对被捕食者的负相互作用强度（αit*αbi, 其中，b, i, t分别为IGP模块内的基础食源、中间消费者和杂食者）和被捕食者对捕食者的正相互作用强度（αtb）之间存在负相关关系，该负相关关系可约束环重，维持食物网稳定。随机食物网的低稳定性正是由于破坏了这一约束效应。改进的环重算法创新地关联了环内物种的P/B（生产量/生物量）值与相互作用强度，并解析到环重随着(Pt/Bt * Pi/Bi * Pt/Bt * Bt/Bb)^(1/3)项的增加而增加。进一步发现（Pt/Bt * Pi/Bi）项（表征αit*αbi的大小）和（Pt/Bt * Bt/Bb）项（表征αtb）之间也存在如同相互作用强度之间的负相关关系，对环重具有约束效应，促进食物网稳定。该负相关关系是由于杂食性环内物种的能量学特性（如生物量、生产量/生物量）和食物网内能量流动的平衡约束形成的。进一步模拟了2000个具有不同生物量和能流结构的IGP模块，IGP模块形成的正反馈杂食性环证实了上述约束效应的普遍存在。
（6）构建了考虑物种功能性状和适应行为的适应性集团内捕食模型，发现物种的功能多样性和对功能性状的调节速率促进了集团内捕食模块的物种共存和群落稳定性。
进一步构建了融入物种功能性状和适应行为的适应性IGP模型，分析了适应性IGP模块和非适应性IGP模块在物种共存和群落动态上的差异特征，探究了环境梯度上IGP模块的群落动态和多稳态响应机制。研究发现，在由环境容纳量K和杂食者对基础食源的最大取食率aRPmax组成的二维参数空间上，非适应性IGP模块内3个物种共存的区域很小，中间消费者较易被去除，K和aRPmax均很高时，基础食源和杂食者振荡共存。选取非适应性IGP模块的四种稳态，探究反映物种功能多样性的功能性状范围w（0.1~1）和对功能性状调节的速率v（0.001~0.01）的作用，在四种稳态的基础上均发现w对物种共存具有促进作 用、v对群落稳定性具有促进作用，并且发现，随着w的增大，由于杂食者对中间消费者存在竞争和捕食的双重压力，杂食者的持久抑制了中间消费者的生物量，同时由于级联效应，释放了基础食源的生物量。不同的是，在较低的环境容纳量时，需要较高的w促进物种共存，而在较高环境容纳量时，只有局域的w（0.2<w<0.7）和较高的v（v>0.03）可以支撑3个物种的共存。且当环境容纳量较高时，三个物种振荡共存，但随着v的增加，3个物种生物量的CV值（标准差/均值）逐渐减小，群落稳定性增加。
综上所述，本研究对黄河三角洲陆海梯度上芦苇、盐地碱蓬、互花米草和海草区域不同湿地类型的生物群落、食物网营养结构、能量流动、稳定性的变化规律进行了深入的系统性解析，揭示了滨海湿地食物网稳定维持的能量学机制，强调了物种功能多样性对群落稳定的促进作用，可为黄河三角洲湿地的生物多样性保护、生态过程维持和生态功能发挥提供重要科学支撑。

Under the interactions between land and sea, coastal wetlands create unique environmental gradients such as elevation, water and salinity, and form different types of the ecosystem with different landscape heterogeneity and diverse functions, such as salt marsh, mudflat, and offshore waters, which play an important role in habitat maintenance, functioning regulation and biodiversity conservation. In recent years, global change, storm surge and other natural factors, as well as human activities such as intensive constructions of hydraulic engineering and large reclamations, have directly changed the inherent water and salinity spatial characteristics of coastal wetlands, resulting in a more sensitive and fragile balance between water and salinity. The changes of environmental gradients further affect the spatial distribution and species interactions, resulting in increasingly serious degradation of ecosystem structure and function. Therefore, it is very urgent to carry out the ecological protection of coastal wetlands. In recent years, a large number of relevant studies have been carried out on the spatial characteristics of different biological communities such as vegetation, plankton, benthos, and fish on the land-sea gradient of coastal wetlands. However, existing studies often only focused on a single biological community in a changing environment, ignoring the interactions among the different biological communities, and rarely considering the dynamics at the ecosystem level, in addition, it is still unclear on the responses and maintenance mechanisms of ecosystem stability in coastal wetlands.
The development of food web theory, combing multiple biological communities and complex trophic interactions among them, provides a good approach to reveal the structural and functional changes of ecosystems. Taking the land-sea gradient of the coastal wetland in the Yellow River Delta as the research object, combined with field investigations, indoor experiments, model construction, and process simulations, this study aims to explore the changes of food web trophic structure, energy flows, and quantitative food web stability on the land sea-gradient. This study aims to reveal the maintenance mechanism of food web stability under a changing environment, and on this basis, we would put forward effective strategies to enhance food web stability. The main research results are as follows:
(1) Clarifying the spatial characteristics of multiple biological communities on the land-sea gradient of the Yellow River Delta wetland. It is found that the different biological communities have asymmetric responses to the land-sea gradient.
We conducted field investigations of the spatial characteristics of environmental factors in freshwater wetland, upland, salt marsh, mudflat, and offshore areas on the land-sea gradient in the Yellow River Delta wetland. We also investigated the spatial characteristics of the biological communities, including aquatic plants, macrobenthos, and fish. There are obvious differences in the composition of vegetation community on the land-sea gradient in the Yellow River Delta wetland. The dominant plants are Phragmites australis, Tamarix chinensis, Suaeda salsa, Spartina alterniflora, and seagrass Zostera japonica along the land-sea gradient. The macrobenthos community in the freshwater and upland areas are dominated by insects (accounting for 92.9% and 72.9% respectively). The macrobenthos community is dominated by bivalvia in mudflat and seashore areas, accounting for 86.7% and 62.5% respectively. There is little difference in the community structure composition of fish. The species richness of vegetation and macrobenthos communities changed significantly along the land-sea gradient (vegetation, p=0.047; macrobenthos, p=0.005). The species richness of the vegetation community decreased significantly along the land-sea gradient, and the species richness of the macrobenthos community decreased gradually from freshwater wetland to high salt marsh wetland, and then increased gradually, which is in agreement with the Remane curve. There is little difference in species richness of fish on the land-sea gradient (p=0.32). There were significant differences in the biomass of plants, macrobenthos, and fish on the land-sea gradient of the Yellow River Delta wetland (plants, p=0.001; macrobenthos, p<0.001; fish, p<0.001). Generally, the biomass of plants from land to sea decreased gradually, whereas, the biomass of macrobenthos increased gradually. In addition, the biomass of fish near the sea was higher than that on the nearshore.
(2) Constructing the models of food web trophic structure along the land-sea gradient in the Yellow River Delta wetland. It is found that the qualitative topological indicators of food web trophic structure increased gradually on the land-sea gradient, implying that food webs became more complex from land to sea.
We selected the habitats of Phragmites australis (PA), Suaeda salsa (SS), Spartina alterniflora (SA), and seagrass Zostera japonica (SG), located in upland, salt marsh, mudflat, and offshore, respectively, along the land-sea gradient in the Yellow River Delta wetland as our study areas. We conducted field investigations to collect typical species in the food web and then detected the content of carbon and nitrogen stable isotopes. Finally, we constructed the food web trophic structure model and quantify its qualitative topological index. It is found that except for the δ15N values of the carbon sources in the PA and SG habitats, there were significant differences in the δ13C and δ15N values among the carbon sources and also among all consumers (for all, p<0.05). In PA habitat, the δ13C and δ15N values of the carbon sources (consisting of sediment detritus, suspended particulate matter, plants, microphytobenthos, and phytoplankton) ranged from -25.74 to -6.84‰ and 1.61 to 3.90‰, respectively. The δ13C and δ15N values of consumers (zooplankton, macrobenthos and fish) in PA habitat ranged from -25.06 to -20.54‰ and 2.66 to 8.58‰, respectively. The δ13C and δ15N values of the carbon sources in SS habitat ranged from -28.99 to -5.46‰ and 4.05 to 6.26‰, respectively. For consumers in the SS habitat, the ranges of δ13C and δ15N values are from -20.01 to -12.24‰ and from 5.54 to 11.89‰, respectively. The δ13C and δ15N values in the SA habitat ranged from -19.90 to -7.00‰ and 4.95 to 6.65‰, respectively, for carbon sources and ranged from -22.44 to -10.46‰ and 6.46 to 11.72‰, respectively, for consumers. The δ13C and δ15N values in the SG habitat ranged from -20.61 to -5.62‰ and 5.02 to 6.68‰, respectively, for carbon sources and ranged from -27.01 to -11.24‰ and 4.32 to 9.61‰, respectively, for consumers. The ranges of trophic positions of consumers in the PA, SS, SA, and SG habitats were 1.36~3.25, 1.91~3.94, 1.65~3.32, and 1.35~3.62, respectively. The highest trophic position in PA habitat was Synechogobius hasta, whereas, in other habitats were all Lateolabrax japonicus. Besides, bivalves, gastropods, crabs, and L. japonicus had obvious differences in dietary preferences in different habitats. The number of trophic links and linkage density of the food webs in the SA and SG habitats were significantly higher than those in SS and PA habitats.
(3) Constructing the equilibrium energy flow model of food webs. It is found that the energy of the food webs in the Yellow River Delta wetland was mainly contributed by the detritus food chain, which all had the energy pyramid structure. The total energy flows of the food webs gradually increased from land to sea.
Based on the analysis of the food web trophic structure, the energy flow model of the food web in an equilibrium state was constructed. Thus we quantified the energy flows and trophic transfer efficiencies in the food webs in terms of maintaining energy transfer and species persistence. Then we calculated quantitative topological metrics based on the weight of energy flows. Results showed that the food web energy flows gradually increased from land to sea in the Yellow River Delta wetland, specifically, the energy flows in the PA habitat ranged from 0.0002 to 2.63 g m-2 yr-1, and that in the SS habitat ranged from 0.004 to 15.50 g m-2 yr-1, and that in the SA habitat ranged from 0.001 to 225.97 g m-2 yr-1, and that in the SG habitat are from 0.003 to 143.62 g m-2 yr-1. The energy flows of the four food webs all showed an energy pyramid structure. In addition, the ingestions of the primary consumers (herbivores and detritivores) accounted for a large number of total ingestions, e.g. 81% in PA habitat, 72% in SS habitat, and both 92% in SA and SG habitats. Detritus food chains contributed about 55% of the total energy flows in the PA, SS, and SG habitats, and the grazing food chains contributed about 37%. However, the detritus food chains in the SA habitat contributed 84% of the total energy flows. Regarding the pelagic and benthic sources, we found that the contributions of the benthic carbon sources to the total energy flows in the PA habitat (55%) was slightly higher than that of the pelagic sources (45%), while, in other habitats, the contributions of the pelagic carbon sources were higher, i.e. 59~64%. The biomass and production of species along trophic levels in the four food webs all showed pyramidal structure. The average trophic transfer efficiencies of the food webs in PA, SS, SA, and SG were 12.9%, 11.2%, 7.7%, and 9.3%, respectively. The weighted linkage density and the weighted connectance of the food webs in PA, SA, and SG habitats were lower than the unweighted qualitative topological metrics, whereas, the food web in SS habitat showed a contrast pattern. 
(4) Constructing food web stability quantitative model. It is found that the stability of the food web in the seagrass habitat is highest, whereas, the food web in Spartina alterniflora habitat had the lowest stability. Omnivorous loops generated by the intraguild predation module determine food web stability. The stability of the empirical food webs in the Yellow River Delta wetland was significantly higher than that of random food webs.
On the basis of the food web trophic structure and energy flows, we further quantified the trophic interaction strengths between predator and prey for each trophic link and obtained the trophic interaction strength matrix to quantify food web stability. We also analyzed the trophic interaction loops and thus analyzed the relationship between loop weight and food web stability. We finally investigated the difference in food web stability and loop weight between an empirical food web and random food webs. Our results showed that the absolute of all trophic interaction strengths (log10) within the four food webs in PA, SS, SA, and SG habitats ranged from -6.7 to 0.8 yr-1, from -5 to 1.6 yr-1, from -4.2 to 1.5 yr-1, and from -4.9 to 1.8 yr-1, respectively. The absolute value of the interaction strength of the predator on the prey and the interaction strength of the prey on the predator of the food web in the PA habitat was 0.81 ± 1.47 yr-1 and 0.14 ± 0.24 yr-1, respectively; for the SS habitat, 1.39 ± 4.50 yr-1 and 2.38 ± 7.21 yr-1; for the SA habitat, 1.11 ± 5.02 yr-1 and 1.57 ± 3.65 yr-1; for the SG habitat, 0.52 ± 0.78 yr-1 and 2.61 ± 10.64 yr-1. The food web in the SG habitat had the highest quantified stability (Re(λmax)= 0.009 yr-1), followed by the food web in the PA and SS habitats (Re(λmax)= 0.07 and 0.09 yr-1, respectively). The stability of the food web in the SA habitat was the lowest (Re(λmax)= 0.37 yr-1). We found that the stability of food webs decreased with the increase of species richness and linkage density (R2 = 0.49, p<0.001 and R2 = 0.22, p<0.001, respectively), and gradually increased with the increase of weighted connectance (R2 = 0.46, p<0.001). The values of the Re(λmax) of food webs were positively correlated with the maximum loop weights of the omnivorous loops (R2=0.84，p<0.001), which means that the higher the maximum loop weight, the lower the stability of the food web. That is, the heaviest omnivorous loop has a greater suppressive effect on stability. The most dangerous loop that inhibits food web stability was all the loop consisting of phytoplankton, zooplankton, and shrimp in the PA, SS. In the SA habitat, the dangerous loop was composed of phytoplankton, zooplankton, and Carassius auratus. In the SG habitat, the dangerous loop was composed of phytoplankton, zooplankton, bivalves, gastropods, other crustaceans, crabs, and polychaetes. The stability of the empirical food webs in the Yellow River Delta wetland was significantly higher than that of the random food webs (T-test, p<0.001).
 (5) Inventing the algorithm of loop weight of positive feedback omnivorous loops based on energetic parameters, and constructing a population dynamic model of intraguild predation, revealing the stability maintenance mechanisms under the coupled constraints between species energy characteristics and food web energy flow balance.
The different distribution characteristics of the interaction strengths between the empirical food webs and the random food webs were clarified, and the mechanism of the interaction strength distribution on the stability of the food web was revealed. Based on the energetic parameters, the algorithm of positive feedback omnivorous loop weight was invented to reveal the energetic mechanisms of food web stability maintenance. A population dynamic model of the IGP module was constructed to verify the generality of the revealed stability maintenance mechanisms. The results showed that the product of the negative interaction strengths (αit*αbi, in which, b, i, and t mean the basic, intermediate, and top species in the IGP module) was negatively correlated with the value of the positive interaction strength (αtb). That is, there is no loop where both the negative and positive interaction strengths are high which would result in a really high loop weight, and thus enhance food web stability. The low stability of random food webs is due to that randomizations destroy the negative correlation. The invented loop weight algorithm innovatively proposed that the loop weight, as well as species interaction strengths, are positively correlated with species-specific production rates, i.e. production to biomass ratios. We further found that the loop weight increases with the increase of the term of (Pt/Bt * Pi/Bi * Pt/Bt * Bt/Bb)^(1/3). Furthermore, it was found that there is also a negative relationship between the term (Pt/Bt * Pi/Bi) (referring to the value of αit*αbi) and the term of (Pt/Bt * Bt/Bb) (referring to the value of αtb), which is similar with the negative correlation found among the interaction strengths. Based on the constructed IGP theoretical ecological model, results showed that there was a significant correlation between the stability Re(?max) of the IGP module and the weight of the positive feedback omnivorous loop derived from the IGP module. It further showed that the energetic constraints at the trophic group and food web level create a pattern in interaction strengths within trophic interaction loops that enhances food web resilience. The IGP modeling confirmed the generalization of the mechanisms of food web stability found in the Yellow River delta wetland. 
(6) Constructing the adaptive intraguild predation model with species functional traits and adaptative behaviors coupled. It was found that the species functional diversity and the adaptive behaviors to adjust functional traits jointly promote the species coexistence and community stability.
The study further emphasized the adaptive behaviors of species and their ability to regulate their functional traits through building an adaptive IGP model integrating species functional diversity and adaptive behaviors. We then compared the differences between adaptive and non-adaptive IGP models in species biomasses, persistence, and community dynamics. Further, we aimed to explore the adaptive behaviors of species in the IGP module along an environmental gradient. It is found that only a small region allowed the coexistence of the three species in the non-adaptative IGP module in the parameter space defined by carrying capacity K and the maximum attack rate of the intraguild predator on the basal prey aRP,max, and the intermediate consumer is easy to be removed from the system. When K and aRP,max are very high, the basic resource and the omnivore oscillatory coexist. We selected four parameter combinations representing four regimes of the non-adaptive IGP module in terms of species persistence and species biomasses to further investigate the effects of trait adaptation. For each condition, we simulated the dynamics of the adaptive IGP in the grid gradient of the speed of trait adaptation v and the range of trait variation w indicating functional diversity. It was all found that w promotes species coexistence and v promotes community stability. With the increase of w, the persistence of the omnivore inhibits the biomass of the intermediate consumer and then releases the biomass of basic resources. The difference is that at lower environmental capacity, higher w is needed to promote species coexistence, while at higher environmental capacity, only constrained w (0.2<w<0.7) and higher v (v>0.03) enable to support the coexistence of the three species. When K is high (K=16), the three species only oscillatory coexist. While, with the increase of v, the variability of the biomasses of the three species gradually decreases, and thus the community stability increases.
In conclusion, this study systematically and deeply analyzed the biological communities, food web trophic structure, energy flows, and ecosystem functions, and food web stability in the Phragmites australis, Suaeda salsa, Spartina alterniflora, and seagrass Zostera japonica habitats. Importantly. this study revealed the energetic mechanism of food web stability and emphasized the importance of species functional diversity and adaptive behaviors in enhancing community stability. In general, this study provides important scientific supports for wetland biodiversity conservation, ecological process maintenance, and ecological functioning promotion in the Yellow River Delta wetland.