1 INTRODUCTION

Riverine floodplains are commonly considered among the most species-rich and productive ecosystems (Amoros & Bornette, 2002; Naiman et al., 1993; Tockner & Stanford, 2002; Ward et al., 1999). The complex interaction between topography, sediment transport and flow variations control the biodiversity in inter-connected and dynamic habitats of large river floodplains. Since the second half of the 19th century, alteration of large rivers and their floodplains resulted in ecological discontinuities, habitat fragmentation, morphological alterations, changes in flow and thermal regimes, and sediment processes (Poole & Berman, 2001; Tockner et al., 2008). These alterations also led to large-scale losses of both natural habitats and biodiversity in these systems (Montoya et al., 2012) and to the establishment of novel biotic assemblages (Carling & Petts, 1992; Hobbs et al., 2006; Tockner et al., 2008; Ward, 1989; Ward & Stanford, 1995). As a consequence, floodplain restoration attempts emerged, aimed at achieving more natural geomorphic and hydrological processes according to the flood and flow pulse concepts (Junk et al., 1989; Junk & Wantzen, 2004; Poff et al., 1997; Puckridge et al., 1998; Tockner et al., 2000).

The flood pulse concept emphasises the importance of flooding for floodplain productivity and maintenance of biotic diversity (Junk et al., 1989). Since the application of this concept in large-scale restorations of river floodplains (Lamouroux et al., 2015; Tockner et al., 1999), important efforts to understand how hydrology influences floodplain biodiversity have been carried out. Evaluation of restoration through biological monitoring demonstrated that processes generated by flow increase and floods, such as the establishment of connections with floodplain water bodies (e.g. secondary channels, riverine wetlands, oxbow lakes) and the local increase in hydraulic shear stress, can influence invertebrate densities and species replacement (i.e. turnover; Dolédec et al., 2007; Rader et al., 2008; Wohl et al., 2015; Rolls et al., 2018). At low connectivity, species assemblages are dominated by lenitophilous taxa, such as the majority of Gastropoda. While rheophilous taxa as some mayflies (Ephemeroptera) and caddisflies (Trichoptera), more suited to hydraulic shear stress, increase as lateral connectivity increases (Castella et al., 2015; Marle et al., 2021; Paillex et al., 2007; Reckendorfer et al., 2006). A deeper knowledge of these processes remains necessary to understand the alterations of biodiversity induced by river regulations to improve conservation and restoration. A predictive understanding of hydro-ecological relationships can help mimic key aspects of the natural flow regime, or create hydrological conditions that maximise restoration success within regulated rivers (Palmer & Ruhi, 2019).

A unique feature in semi-connected and disconnected floodplain channels (i.e. secondary channels adjacent to the main river channel) is the fact that the direction of surface water flow can reverse depending on the water level in the main channel and the relative altitude of the upstream and downstream alluvial plugs (i.e. sediment deposited in a neck–floodplain channel). Therefore, two types of connections with the main channel can occur in floodplains, with contrasted sedimentological and biotic consequences. A shear stress-related connection (i.e. condensed flow or active overflow) that usually occurs at the upstream channel entry and generates disturbances through scouring of bed material and biota (Bornette et al., 1994; Bornette et al., 1998). However, surface flow connections can also occur with no generation of shear stress (i.e. diffuse overbank flows or passive overflow, slow connection hereafter) in semi-connected floodplain channels or in disconnected water bodies distant from the river (Amoros & Roux, 1988; Citterio & Piégay, 2009). Slow connections create an entry of river water, suspended solids, and biological propagules in the floodplain channel. This process enhances fine sediment deposition, especially at the downstream parts of floodplain channels (Riquier et al., 2015; Riquier et al., 2017). Although recognised as critical in floodplain functions by Bornette and Amoros (1991) and Amoros (2001), accounting for slow connections remains rare in models of biodiversity changes in floodplains (but see Marle et al., 2021).

In addition to these two types of surface water connection, floodplain water bodies also differ in their degree of vertical connection with the floodplain aquifer. The relative contribution of surface vs. groundwater was shown to influence the function and biota particularly in semi-connected and disconnected floodplain channels (Amoros, 2001; Bornette et al., 1998; Boulton et al., 2008). Indeed, while surface waters generally provide turbid and oxygenated waters and warmer temperatures in summer, groundwater dominated channels are generally characterised by less turbid, more oligotrophic, often less oxygenated waters, buffered temperature variations (i.e. thermal inertia) and a higher yearly persistence of aquatic vegetation (Boulton et al., 2010; Kondolf et al., 2006; Wohl, 2017). Such floodplain channels were shown to maintain stenothermal (i.e. restricted to narrow temperature variations), oligotrophic (i.e. adapted to nutrient-poor waters) and hyporheobiont (i.e. adapted to living in the transitional zone between river and ground water) organisms (Boulton et al., 2008; Brunke & Gonser, 1997; Glanville et al., 2016; Trudgill et al., 2005). Floodplain channel restoration, not only modifies surface water connections, but can also impinge upon groundwater–surface water interactions when lentic pools or channel stretches are deepened by mechanical means primarily to increase the duration of lateral connection with the river (Boulton et al., 2010). To improve models of floodplain biodiversity responses to hydrology and temperature changes under alternative restoration scenarios, it is therefore critical to integrate both types of surface connections as well as the contribution of groundwater. This could apply both at the level of individual channels and of entire floodplains.

We therefore aimed at assessing how lateral surface water connections (both shear stress-related and slow, as defined above) and vertical connection with groundwater (inferred through thermal inertia) contribute and interact to explain variations in aquatic invertebrate taxonomic composition and turnover at the floodplain scale. Aquatic invertebrate assemblages were chosen because: (1) their composition is known to change along the lateral hydrological connectivity gradient in floodplains (Gallardo et al., 2014; Marle et al., 2021; Paillex et al., 2007; Reckendorfer et al., 2006; Waringer et al., 2005); (2) the majority of species have short generation times (lifecycles are often less than 1 year); and (3) they are sensitive to flow variations (Dolédec et al., 2007; Dolédec & Statzner, 1994; Šporka & Nagy, 1998). We used a dataset on floodplain aquatic invertebrates sampled over 10 years (from 2007 to 2016) in 13 floodplain channels located in two bypassed sections of the Rhône River (France) that were subjected to restoration in 2005–2006 (Lamouroux et al., 2015). The two floodplains are adjacent but differ in general morphology, one being strictly braided, the other being wider and incorporating both anastomosed and braided channels. We hypothesised that these morphological differences may influence hydrological processes and therefore the invertebrate assemblages for which divergences in species composition between the two floodplains were already observed (Castella et al., 2021; Castella et al., 2022). We expected the shear stress-related connections to have a prevailing influence upon aquatic biodiversity in the braided floodplain that has a higher overall slope. In more anastomosing floodplains, the lower channel-flow capacity caused by in-channel fluvial deposition (Makaske, 2001) may cause slow connections and thermal inertia as the main drivers of floodplain biodiversity. To test these hypotheses, two biodiversity metrics were considered: (1) taxonomic richness was compared among the three channel types i.e. connected, semi-connected, and disconnected; (2) species turnover was modelled using generalised dissimilarity modelling (Ferrier et al., 2007; Timoner et al., 2020) along three environmental gradients referring to the three components of hydrologic connectivity mentioned above.

中文翻译：

---

热状态与横向连通性一起控制河流洪泛区的水生无脊椎动物组成

1 简介

河流泛滥平原通常被认为是物种最丰富和生产力最高的生态系统之一（Amoros 和 Bornette，  2002 年；Naiman 等人，  1993 年；Tockner 和斯坦福大学，  2002 年；Ward 等人，  1999 年）。地形、泥沙输送和流量变化之间的复杂相互作用控制着大型河流泛滥平原相互连接和动态生境的生物多样性。自 19 世纪下半叶以来，大型河流及其洪泛区的改变导致生态不连续、栖息地破碎化、形态改变、流量和热状态变化以及沉积过程（Poole & Berman,  2001 ; Tockner et al.,  2008）。这些改变还导致这些系统中自然栖息地和生物多样性的大规模丧失（Montoya 等人，  2012 年）以及新生物组合的建立（Carling & Petts，  1992 年；Hobbs 等人，  2006 年；Tockner 等人）等人，  2008 年；沃德，  1989 年；沃德和斯坦福大学，  1995 年）。因此，出现了洪泛区恢复尝试，旨在根据洪水和流量脉冲概念 实现更自然的地貌和水文过程（Junk et al., 1989 ; Junk & Wantzen,  2004 ; Poff et al.,  1997 ; Puckridge et al . .,  1998; Tockner 等人，  2000 年）。

洪水脉冲概念强调洪水对于洪泛区生产力和维持生物多样性的重要性（Junk et al.,  1989）。自从将这一概念应用于河流洪泛区的大规模修复以来（Lamouroux 等人，  2015 年；Tockner 等人，  1999 年）)，为了解水文如何影响洪泛区生物多样性开展了重要工作。通过生物监测进行的恢复评估表明，流量增加和洪水产生的过程，例如与泛滥平原水体（例如二级河道、河流湿地、牛轭湖）的连接以及水力剪切应力的局部增加，会影响无脊椎动物的密度和物种更替（即营业额；Dolédec 等人，  2007 年；Rader 等人，  2008 年；Wohl 等人，  2015 年；Rolls 等人，  2018 年）。在低连通性下，物种组合以嗜嗜性类群为主，例如腹足纲的大多数。虽然一些蜉蝣（蜉蝣目）和石翅目（毛翅目）的流变类群更适合水力剪切应力，但随着横向连通性的增加而增加（Castella 等人，  2015 年；Marle 等人，  2021 年；Paillex 等人，  2007 年；Reckendorfer等人，  2006 年）。仍然需要对这些过程有更深入的了解，以了解河流法规引起的生物多样性变化，以改善保护和恢复。对水文生态关系的预测性理解可以帮助模拟自然流态的关键方面，或创造水文条件，最大限度地提高受管制河流的恢复成功率（Palmer & Ruhi，  2019 年）。

半连通和不连通的洪泛区河道（即与主要河道相邻的次要河道）的一个独特之处在于，地表水流的方向可以根据主要河道的水位和上游的相对高度而反转。和下游冲积栓塞（即沉积在颈部-泛滥平原河道中的沉积物）。因此，在洪泛区可能会出现两种与主河道的连接，具有不同的沉积学和生物学后果。一种与剪应力相关的连接（即冷凝流或主动溢流），通常发生在上游通道入口处，并通过冲刷床材料和生物群产生扰动（Bornette 等人，  1994 年；Bornette 等人，  1998 年））。然而，在半连通洪泛区河道或远离河流的不连通水体中，在不产生剪切应力的情况下，地表水流连接也可能发生（即漫溢的越岸流或被动溢流，此后的缓慢连接）（Amoros & Roux，  1988 年；Citterio和皮耶盖，  2009 年）。缓慢的连接会在洪泛区河道中形成河水、悬浮固体和生物繁殖体的进入。这一过程增强了细沙沉积，特别是在漫滩河道的下游部分（Riquier 等人，  2015 年；Riquier 等人，  2017 年）。虽然被 Bornette 和 Amoros ( 1991 ) 和 Amoros ( 2001 ) 认为在洪泛区功能中至关重要)，在洪泛区生物多样性变化模型中很少考虑缓慢的连接（但参见 Marle 等人，  2021 年）。

除了这两种地表水连接方式外，泛滥平原水体与泛滥平原含水层的垂直连接程度也不同。地表水与地下水的相对贡献被证明会影响功能和生物群，特别是在半连通和不连通的洪泛区河道中（Amoros，  2001；Bornette 等，  1998；Boulton 等，  2008）。事实上，虽然地表水通常在夏季提供混浊和含氧水以及较温暖的温度，但以地下水为主的渠道通常具有混浊度较低、贫营养化、含氧量较少、温度变化缓冲（即热惯性）和水生生物年持久性较高的特点。植被（Boulton 等人，  2010 年；Kondolf 等人，  2006 年；Wohl，  2017 年）。这样的漫滩渠道被证明可以维持狭窄的热（即仅限于狭窄的温度变化）、贫营养（即适应营养贫乏的水域）和低流态生物（即适应生活在河流和地下水之间的过渡地带）生物（Boulton 等人，2009 年）。 ,  2008 年；Brunke & Gonser,  1997; 格兰维尔等人，  2016 年；Trudgill 等人，  2005 年）。洪泛区河道修复不仅会改变地表水连接，而且当通过机械手段加深静水池或河道延伸段时，也会影响地下水-地表水的相互作用，主要是为了增加与河流的横向连接的持续时间（Boulton 等，  2010）。为了改进替代恢复情景下洪泛区生物多样性对水文和温度变化的响应模型，因此整合两种类型的地表连接以及地下水的贡献至关重要。这既适用于个别渠道，也适用于整个洪泛区。

因此，我们旨在评估横向地表水连接（与剪切应力相关的和缓慢的，如上文所定义）和与地下水的垂直连接（通过热惯性推断）如何有助于解释水生无脊椎动物分类组成和洪泛区周转的变化规模。选择水生无脊椎动物组合是因为：(1) 已知它们的组成会沿着洪泛区的横向水文连通性梯度发生变化（Gallardo 等人，  2014 年；Marle 等人，  2021 年；Paillex 等人，  2007 年；Reckendorfer 等人。 ,  2006 年；Waringer 等人，  2005 年); （2）大多数物种的世代时间短（生命周期往往不到1年）；(3) 它们对流量变化很敏感 (Dolédec et al.,  2007 ; Dolédec & Statzner,  1994 ; Šporka & Nagy,  1998 )。我们使用了一个关于泛滥平原水生无脊椎动物的数据集，这些水生无脊椎动物在 10 年（从 2007 年到 2016 年）中采样，位于位于罗纳河（法国）两个绕行段的 13 个洪泛区河道中，这些河道在 2005 年至 2006 年进行了修复（Lamouroux 等人，  2015 年））。这两个洪泛区相邻，但总体形态不同，一个是严格编织的，另一个更宽，并结合了吻合和编织的渠道。我们假设这些形态差异可能影响水文过程，因此已经观察到两个洪泛区之间物种组成差异的无脊椎动物组合（Castella 等人，  2021 年；Castella 等人，  2022 年）。我们预计与剪切应力相关的连接将对整体坡度较高的辫状漫滩中的水生生物多样性产生主要影响。在更吻合的洪泛区，河道内河流沉积导致河道流量降低（Makaske，  2001) 可能导致连接缓慢和热惯性成为洪泛区生物多样性的主要驱动因素。为了检验这些假设，考虑了两个生物多样性指标：（1）比较了三种通道类型（即连接、半连接和断开）之间的分类丰富度；(2) 物种周转使用广义相异模型 (Ferrier et al.,  2007 ; Timoner et al.,  2020 ) 沿三个环境梯度进行建模，参考上述水文连通性的三个组成部分。