Eawag 05141

Doctoral Thesis ETH No. 16997

RIVERINE FLOODPLAIN HETEROGENEITY AS A CONTROLLER OF ORGANIC MATTER DYNAMICS AND TERRESTRIAL INVERTEBRATE DISTRIBUTION

A dissertation submitted to SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH

for the degree of Doctor of Sciences

presented by

SIMONE DANIELA LANGHANS Dipl. Natw. ETH born 18.09.1976 citizen of Nussbaumen (Thurgau) and Jona (St. Gallen)

accepted on the recommendation of

Prof. Dr. Bernhard Wehrli, examiner Prof. Dr. Stuart Findlay, co-examiner Prof. Dr. Klement Tockner, co-examiner Dr. Urs Uehlinger, co-examiner

2006

TABLE OF CONTENTS

Summary 1 Zusammenfassung 5 Introduction 9 Chapter 1: Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems 23 Chapter 2: The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy) 33 Chapter 3: Leaf decomposition across aquatic and terrestrial habitat types in a riverine floodplain mosaic 53 Chapter 4: Cotton strips as a leaf surrogate to measure decomposition in river floodplain habitats 77 Chapter 5: Aerial input, lateral transport, and surface storage of coarse particulate organic matter in a riverine floodplain mosaic (Tagliamento, NE-Italy) 93 Chapter 6: Seasonal variation of riparian arthropods along lateral and vertical gradients in a braided gravel-bed river 119 Synthesis and Outlook 151 Curriculum Vitae 157 Acknowledgements 161

SUMMARY

Rivers and their fringing floodplains are among the most complex, dynamic, and therefore, diverse ecosystems worldwide. Flow and flood pulses are responsible for the formation and structure of different habitats, and these pulses define the degree of connectivity among habitats. Floods and droughts represent two extremes of the flow regime. Between these extreme states, riverine floodplains undergo distinct cycles of expansion and contraction along longitudinal, lateral, and vertical dimensions, and therefore, shift between terrestrial and aquatic phases. In Europe, most rivers and floodplains have been severely altered by human activities, with dramatic consequences for their biodiversity and functioning. To successfully manage riverine floodplains, we need to understand how temporal and spatial heterogeneity, including expansion and contraction dynamics, and the composition and configuration of habitats determine ecosystem processes. Research for this thesis was conducted along the Tagliamento River in northeastern Italy. This system represents a unique opportunity to study ecosystem processes under near-natural conditions, as it has retained an essentially pristine morphological and hydrological character. The present thesis investigates the importance of environmental heterogeneity in determining organic matter dynamics and the distribution of terrestrial invertebrates in a river-floodplain ecosystem (Tagliamento River). The thesis begins with a general introduction that sets the context of the work, followed by a conceptual model that unifies input and storage of organic matter, with leaf quality and decomposition in a complex and dynamic system (Chapter 1). Currently, individual components of organic matter dynamics are mostly treated separately in the literature. My conceptual model takes an advanced and holistic perspective that links natural heterogeneity with ecosystem processes, and provides a framework for the scientific information necessary for the sustainable management of floodplains. The role of the flow regime, defined by its components duration, frequency, and timing of inundation, in controlling leaf decomposition was tested by field manipulation experiments in the Tagliamento River (Chapter 2). Leaf bags (Populus nigra) were exposed to 10 different treatments imitating a dynamic flow regime. Experiments were conducted in two seasons to account for timing of inundation. After 30 days, 12 - 49% of leaf litter was decomposed, with significantly faster rates in winter than in summer. Duration of inundation was

-1- Summary the main hydrological component that determined leaf decomposition, whereas frequency of inundation induced leaf decomposition heterogeneity. Shredding macroinvertebrates played a significant role in leaf decomposition only under permanent aquatic and terrestrial conditions. Fungi were responsible for the faster leaf decay in winter. My results indicate that modifications of the inundation regime will directly modify decomposition heterogeneity. The role of habitat heterogeneity in leaf decomposition was quantified within the floodplain mosaic of the Tagliamento River (Chapter 3). Leaf bags (P. nigra) were placed in seven contrasting habitat types, including the river channel, different pond types, exposed gravel surface, large wood accumulations, vegetated islands, and riparian forest. Based on decomposition rates, habitats separated into three groups: river channel with fast rates, ponds with medium rates, and terrestrial sites with slow rates. Microbes and detritivores drove leaf decomposition in aquatic sites, whereas microbial activity was the main driver in terrestrial sites. Habitat heterogeneity seems to be strongly linked with heterogeneity of leaf decomposition rates. Hence, morphological simplification of the ecosystem, in concert with alterations to the flow regime is expected to homogenize decomposition rates, with subsequent consequences for overall ecosystem functioning. In Chapter 4, I tested the applicability of cotton strips as a surrogate to measure leaf decomposition in contrasting aquatic and terrestrial habitats. Cotton-strip decomposition, measured as loss of tensile strength and mass, was correlated with leaf mass loss. Across river channels, ponds, and terrestrial sites, I found comparable patterns between loss in cotton strip tensile strength and leaf mass. The results suggest that cotton strips have the potential to mimic leaf decomposition in fluvial settings, and therefore, to serve as a functional indicator for stream assessment. In correspondance with leaf decomposition experiments along the Tagliamento River, litterfall and lateral transport of organic matter was quantified with spatially distributed litter traps over a 10 month period. Organic matter standing stock was determined in four habitats and three seasons (Chapter 5). Direct litterfall dominated litter input close to vegetation, whereas lateral transport was the major pathway in the floodplain 10 to 20 m away from vegetation. Islands contributed more than 95% to the annual direct input, and 65% to lateral transport. However, riparian forests were identified as hot spots of litter standing stock.

-2- Summary

In Chapter 6, I explored the connection between environmental heterogeneity and terrestrial invertebrate distribution using a four-dimensional field experiment, including the air space, the sediment surface, the unsaturated zone, and the temporal dimension, along a large gravel bank in the Tagliamento River. Invertebrate abundance, at two taxonomic levels (family and species) responded to all four dimensions. Beetles assemblages were correlated with various environmental parameters indicating that spatial and temporal heterogeneity plays a major role in their behaviour. From my research, I propose that both, spatial and temporal heterogeneity shape and determine organic-matter dynamics and terrestrial invertebrate distribution in riverine floodplains. Therefore, morphological and/or hydrological alterations have consequences for the functioning of organic matter dynamics and the abundance and diversity of terrestrial invertebrates. This can result in changes in nutrient cycling, sediment respiration, primary and secondary production, or decreasing terrestrial and aquatic biodiversity.

-3-

-4-

ZUSAMMENFASSUNG

Flüsse gehören zusammen mit ihren Auenflächen zu den komplexesten, dynamischsten und daher vielseitigsten Ökosystemen der Erde. Schwankungen im Abflussregime mit Hochwasserereignissen bestimmen in erster Linie das Vorhandensein, die Struktur und den Verbindungsgrad der verschiedenen Habitate. Hochwässer die den gesamten Auenbereich überfluten bzw. Phasen in denen komplette Flussabschnitte trocken fallen, stellen Extremereignisse des Abflussregimes dar. Zwischen diesen beiden Extremen durchlaufen flussnahe Auenbereiche einen steten Kreislauf der Expansion und Kontraktion. Da sie diesen Prozessen sowohl in longitudinaler, lateraler als auch vertikaler Dimension ausgesetzt sind, befinden sich Auenbereiche in einem ständigen Wechsel zwischen terrestrischem und aquatischem Zustand. In Europa wurden die meisten Fliessgewässer und Flussauen aufgrund der intensiven Nutzung durch den Menschen stark verändert. Dies hatte dramatische Konsequenzen für deren Biodiversität und Funktionsfähigkeit zur Folge. Um gezielte Massnahmen zur Erhaltung der flusseigenen Dynamik ergreifen zu können, benötigen wir ein präzises Verständnis des Zusammenhangs zwischen zeitlicher und räumlicher Heterogenität und dem Funktionieren von Ökosystemprozessen. Das heisst wir benötigen sowohl ein Verständnis über die Ausdehnung und Kontraktion von Flüssen als auch über die Zusammensetzung und -stellung von Habitaten in mosaikartigen Landschaften, die die ökosystemaren Prozesse bestimmen. Die Untersuchungen für meine Dissertation habe ich entlang des Tagliamento (Italien) durchgeführt. Aufgrund seines ursprünglich gebliebenen morphologischen und hydrologischen Charakters, ermöglicht der Tagliamento das Studium von ökosystemaren Prozessen unter natürlichen Bedingungen. In meiner Dissertation untersuche ich den Einfluss von ökologischer Heterogenität auf die Dynamik organischen Materials und der Verteilung terrestrischer Invertebraten in einem Fluss-Aue-Ökosystem (Tagliamento, Italien). Anschliessend an eine Einleitung, die den generellen Kontext der Arbeit beleuchtet, präsentiere ich ein konzeptionelles Modell das Blattabbau, Eintrag, Ablagerung und Qualität von organischem Material vereinigt (Kapitel 1). Bis anhin wurden die genannten Aspekte der Dynamik von organischem Material in der Literatur separat behandelt. Eine ganzheitliche Sichtweise die eine Verbindung zwischen natürlicher Heterogenität und natürlichen Prozessen

-5- Zusammenfassung herstellt, ist jedoch wichtig für das Verständnis von Ökosystemen und von besonderem Nutzen bei der Planung von Revitalisierungsprojekten. Die Bedeutung der Heterogenität des Abflussregimes für den Blattabbau (insbesondere die Parameter Dauer, Frequenz und Zeitpunkt der Überflutung) wurde mittels Feldexperimenten im Tagliamento untersucht (Kapitel 2). Blätter der Schwarzpappel wurden im Feld exponiert und zehn verschiedenen Kombinationen von Überflutungsdauer und –frequenz ausgesetzt. Um den Einfluss des Überflutungszeitpunktes untersuchen zu können, habe ich das Experiment in zwei verschiedenen Jahreszeiten, im Winter und im Sommer, durchgeführt. Am Ende der Experimente waren 12% bis 49% des Blattmaterials abgebaut, wobei der Abbau im Winter signifikant schneller ablief als im Sommer. Die Überflutungsdauer wurde in diesem Zusammenhang als der Parameter identifiziert, der den Abbau kontrollierte. Wo hingegen die Überflutungsfrequenz für die Heterogenität im Blattabbauprozess verantwortlich war. Zerkleinerer, waren für diesen Prozess von geringer Bedeutung, während Pilze jedoch den Abbau während der Wintermonate massgeblich beschleunigten. Diese Resultate belegen, dass Veränderungen des Abflussregimes etablierte Abbauprozesse direkt verändern können. In Kapitel 3 habe ich die Bedeutung der Habitatheterogenität auf den Blattabbau in der mosaikartigen Auenlandschaft des Tagliamento untersucht. Blätter der Schwarzpappel wurden in sieben unterschiedlichen Habitaten exponiert: Gerinne, Tümpel in der aktiven Aue bzw. im Auenwald, Schotterflächen, Genisten, bewaldete Inseln und Auenwald. Die Habitate konnten aufgrund der Blattabbaurate in drei Gruppen eingeteilt werden: schneller Abbau in Gerinnen, mittlere Raten in Tümpeln und langsamer Abbau in terrestrischen Habitaten. Mikroorganismen und Zerkleinerer waren verantwortlich für den Blattabbau in aquatischen Habitaten, wohingegen in terrestrischen Habitaten Mikroorganismen dominierten. Zudem scheint die Habitatheterogenität eng mit der Heterogenität des Blattabbausprozesses verknüpft zu sein. Daher kann erwartet werden, dass Veränderungen des Abflussregimes oder der Morphologie einer Flussaue Blattabbauraten homogenisieren. Dies hat direkte Konsequenzen für das Funktionieren von Ökosystemen zur Folge. Die Eignung von Cotton Strips als Blattersatz in Blattabbauexperimenten wurde in verschiedenen aquatischen und terrestrischen Habitaten des Tagliamento und in mehreren Flüssen der Schweiz getestet (Kapitel 4). Der

-6- Zusammenfassung

Abbau von Cotton Strips wurde als Verlust der Zugkraft und des Gewichts gemessen und mit dem Gewichtsverlust der Blätter korreliert. In allen Habitaten wurden vergleichbare Muster des Zugkraftverlusts und der Abnahme des Blattgewichtes gefunden. Die Resultate belegen, dass Cotton Strips über das Potenzial verfügen Blattabbau zu imitieren und deshalb als funktioneller Indikator bei der Bewertung von Flusslandschaften herangezogen werden können. Zusammen mit den Blattabbauexperimenten entlang des Tagliamento, habe ich über einen Zeitraum von 10 Monaten die anfallenden Streumengen und der laterale Transport von grobem organischem Material mit räumlich verteilten Streufallen quantifiziert. Die Menge an abgelagertem organischem Material wurde in vier verschiedenen Habitaten, zu drei verschiedenen Jahreszeiten bestimmt (Kapitel 5). Bezüglich des Anteils am Gesamtstreueintrag, lieferte der direkte Eintrag die grössten Mengen in Vegetationsnähe, während in einer Entfernung von 20 Metern von der Vegetation der laterale Transport den wichtigsten Eintragspfad bildete. Gehölztragende Inseln steuerten mehr als 95% des Materials für den jährlichen direkten Eintrag und 65% des Materials für den lateralen Transport bei und stellen daher die Hauptquelle für die Verbreitung von organischem Material in der offenen Aue dar. Auenwald und gehölztragende Inseln wurden als Zentren des Streubestandes identifiziert. In Kapitel 6 habe ich den Zusammenhang zwischen zeitlicher/räumlicher Heterogenität und der Verteilung von terrstrischen Invertebraten untersucht. Hierzu wurde ein vierdimensionales Feldexperiment, mit den Dimensionen Luftraum, Sedimentoberfläche, Sedimentkörper und Zeit, entlang einer Schotterbank am Tagliamento durchgeführt. Die Abundanz der Invertebraten in 2 verschiedenen taxonomischen Klassen (Familie und Art) reagierten in unterschiedlichem Mass auf Veränderungen in allen 4 Dimensionen. Käfergemeinschaften korrelierten mit verschiedenen Umwelvariablen. Dies deutet darauf, dass die zeitliche und räumliche Heterogenität eine Schlüsselrolle im Verhalten und Vorkommen der Käfer spielt. Die Ergebnisse meiner Arbeit lassen den Schluss zu, dass sowohl räumliche als auch zeitliche Heterogenität die Dynamik von organischem Material und das Vorkommen terrestrischer Invertebraten in Auenlandschaften bestimmen. Morphologische und/oder hydrologische Veränderungen im Auensystem haben daher weitreichende Konsequenzen auf die Dynamik des Nährstoffkreislaufs, die Sedimentrespiration, primär und sekundär Produktion und terrestrische bzw.

-7- Zusammenfassung aquatische Biodiversität. Denn diese werden unter anderem durch die Dynamik des organischen Materials und das Vorhandensein terrestrischer Invertebraten beeinflusst.

-8-

INTRODUCTION

Environmental heterogeneity of riverine floodplains In their natural state, riverine floodplains are among the most biologically productive and diverse ecosystems, however, they are also among the most endangered ones on earth (Tockner and Stanford 2002). Riverine floodplains are defined as the entire channel network and valley-bottom area that is exposed to flooding (Stanford et al. 2005, Fig. 1).

Figure 1: 3-D structure of a riverine floodplain emphasizing dynamic longitudinal, lateral and vertical dimensions and recruitment of woody debris (Stanford 1998).

Channel networks within floodplains undergo distinct cycles of expansion, contraction, and fragmentation along longitudinal, lateral, and vertical dimensions (Stanley et al. 1997, Ward et al. 2002, Tiegs and Pohl 2005, Malard et al. 2006) and are, therefore, in a state of constant flux with repeated erosional and sedimentional processes (Junk et al. 1989). The magnitude, frequency, duration, and timing of the expansion and contraction cycles are important variables influencing the size, spatial configuration, and connectivity of aquatic, semi-aquatic, and terrestrial landscape elements (Junk et al. 1989, Sparks et al. 1990, Ward et al. 2002, Doering et al. in press). While habitats change location, size and configuration over time, the overall abundance and diversity of various elements often remains relatively constant (Arscott et al. 2002, Stanford et al. 2005).

Organic matter dynamics in riverine floodplains In functioning as a vital source of carbon and energy to rivers, allochthonous organic matter links terrestrial with aquatic ecosystems, and drives together with local autochthonous production the food web of and along rivers (Hedges et al.

-9- Introduction

1994, Thorp and Delong 1994). The exchange of organic matter between the river and its adjacent riparian zone can be intense, especially during floods (Neatrour et al. 2004). Organic matter is temporarily stored in depositional areas where it is pre-processed prior to its further transfer to other habitats (Mayack et al. 1989, Merritt and Lawson 1992). Decomposition of organic matter is a key ecosystem-level process (Webster and Benfield 1986, Langhans and Tockner 2006). This process has been widely investigated in both aquatic and terrestrial environments (Boulton and Boon 1991, Gessner et al. 1999), but few studies have addressed decomposition of leaf litter along temporal and spatial gradients in complex and dynamic riverine floodplains (Gurtz and Tate 1988, Chergui and Pattee 1990, Hutchens and Wallace 2002, Langhans et al. 2006, Tiegs 2006).

4-Dimensions of terrestrial invertebrate distribution in riverine floodplains The sediment surface Ecologists have long recognized that within riverine floodplains, channels are strongly linked with their adjacent riparian zones by flows of materials, such as nutrients, leaves, and woody debris (Likens and Bormann 1974, Hynes 1975), and the movements of organisms (Paetzold et al. 2005). An important habitat of riverine floodplains are exposed sediments, such as gravel bars. Gravel bars are areas of low productivity, but harbor a highly specific fauna of carnivorous riparian arthropods, primarily spiders, rove beetles, ground beetles, and ants (Manderbach and Hering 2001, Framenau et al. 2002, Sadler et al. 2004, Paetzold et al. 2005). Researchers have recognized the potential importance of naturally active riverine floodplains to support high levels of biodiversity (Ward 1998, Ward and Tockner 2001). However, limited data particularly in relation to the rarity and conservation value of species are available so far. Because of the strong productivity gradient across the gravel bar from the riparian forest to the river shore, riparian arthropods mainly feed on aquatic subsidies, namely emergent aquatic insects (Paetzold et al. 2005).

The air space Emergent aquatic insects form a reciprocal energy flow from rivers back to riparian systems and feed consumers like birds (e.g., McIntosh 2000), bats (e.g., Power and Rainey 2000), lizards (e.g., Sabo and Power 2002), salamanders (Burton 1976), adult odonates (e.g., Sukhacheva 1996), beetles (e.g., Paetzold et al. 2005), and spiders (e.g., Briers et al. 2005). The flux of emergent insects

-10- Introduction varies highly in time and space (Corbet 1964, Iwata 2003). In temperate zones, emergence of individual taxa is usually seasonal and synchronous over a few days to a few months, peaks during the summer (Sweeney and Vannote 1982), but can also provide a low energy flux to the riparian forest during the rest of a year’s cycle (Nakano and Murakami 2001). Spatial variability might be related to characteristics of river and riparian habitats, and insect behavior (Power and Rainey 2000, Power et al. 2004). The air space above the floodplain surface is an important habitat that may vertically extent for 10s of meters and laterally for 100s of meters. Landscape features such as large wood deposits and vegetated islands may play a key role as orientation marks or as structural components for flying insects and birds.

The unsaturated zone The unsaturated zone refers to the geologic media which lies below the sediment surface but above the water table or the shallowest year-round aquifer (Fig. 2, Selker et al. 1999).

Figure 2: A schematic cross-section of a riverine floodplain showing exposed gravel, and the unsaturated and saturated zone separated by the water table.

The evolution of the unsaturated zone is highly dependent on site geomorphology, regional hydrogeology, climate and precipitation, organic matter and nutrient availability, and the natural flow regime (Flint et al. 2001), whereby its dynamics are driven by the sediment and hydrological cycle. Its thickness can, therefore, range from zero, as when a lake or marsh is at the surface, to hundreds of meters, as it is common in arid regions. In floodplains, the unsaturated zone expands and contracts vertically and seasonally. The unsaturated zone is mostly affected by pressure, volume, and temperature, which in turn define soil phenomena such as capillary flow,

-11- Introduction adsorption, chemical interactions, and matric potential (Tindall et al. 1999). It is seen as physically, chemically, and microbiologically heterogeneous (Bone and Balkwill 1988). Despite that fact, it is often treated as a relatively static and segregated entity, with gradients of nutrients, temperature, and microbes from top to the bottom. However, the capillary fringe, which lies on top of the water table and confines the unsaturated zone, is a rich catalytic region characterized by frequent rearrangements of gaseous, liquid, and solid components (Holden and Fierer 2005). Three major functions occur in the unsaturated zone: 1) storage of water and nutrients, 2) transmission of these substances, and 3) physical, chemical and biological activity, such as thermodynamic interactions, chemical reactions and organism activity. Study of processes in the unsaturated zone is of significant interest to a range of professions, such as civil, petroleum, and geotechnical engineering, forest science and engineering, microbiology among others (Selker et al. 1999). However, although the unsaturated zone comprises an extensive part of riverine floodplains, almost nothing is known about its importance for macroorganisms, such as riparian arthropods.

The Tagliamento River: A reference and model system of European importance As a result of widespread and intensive river management in Europe, few examples of complex, unmanaged river systems remain (Gurnell et al. 2000). An exception is the Tagliamento River, which is considered the ´last large natural Alpine river in Europe´ (Müller 1995), and therefore serves as a model river ecosystem of European importance (Tockner et al. 2003, Fig. 3).

Figure 3: Upstream view of the Tagliamento from the Monte di Ragogna showing the dynamic braided channel morphology that characterizes extensive reaches of this river system. This is the section between Venzone and Cornino (river km 60-77) (Photo: S. D. Langhans; March 2005).

-12- futroduction

The Tagliamento in Northeast Italy (Friuli-Venezia Giulia; 46°N, 12°30' E; Fig. 4) originates in the southern Alps from which it flows unimpeded by high dams to the Adriatic Sea traversing an idealized sequence of constrained, braided, and 2 meandering reaches (Ward et al. 1999). The catchment (area: 2580 km ) mainly consists of limestone and flysch occasionally intermixed with layers of gypsum (Tockner et al. 2003). The lowland section is part of the Venetian-Friulian Plain forming a highly permeable aquifer, several hundred meters deep (Fontana et al. in press). Within the central section of the river's length, it receives flow and sediment inputs from several large tributaries including the Degano, But and Fella. The lower parts of the catchment are subject to agricultural and industrial development. Nevertheless, the river is bordered by a floodplain woodland margin throughout almost its entire length. Despite of water abstraction for hydropower, irrigation, and public water supply the major channel-forming discharge is not significantly affected (Gurnell et al. 2000).

Figure 4: Catchment map of the Tagliamento with major tributaries and towns. fuset shows the location of the river in Italy (I), near the borders of Austi·ia (A) and Slovenia (SL) (modified after Ward et al. 1999). The main study area 1s indicated by the black aITow.

Average annual precipitation is 2150 mm, but precipitation increases from W to E and from S to N from 1000 mm to - 3000 mm. The Tagliamento River is influenced by both Alpine and Mediterranean, snowmelt and precipitation

-13- Introduction

3 -1 regimes (Q80 = 72 m s ; Ward et al. 1999). As a result, it exhibits a flashy discharge regime with peaks in spring and autumn. However, high flow and flood pulses can occur at any time of the year (Arscott et al. 2002).

The island-braided reach of the Tagliamento All experiments described in this thesis were conducted in the island-braided reach of the Tagliamento (river-km 79.8–80.8; 135 m a.s.l.). The 2-km long reach has an active area of 187 ha and is 1 km wide. This part of the riverscape is composed of a complex mosaic of aquatic, semi-aquatic, and terrestrial habitats including lotic channels, lentic water bodies in the active plain (parafluvial ponds) and in the riparian forest (orthofluvial ponds), areas of exposed sediment, vegetated islands, and riparian forest (van der Nat 2002, Fig. 5). During baseflow, the study reach consisted of 42% exposed gravel, 35% riparian forest, 15% channels, 7% islands and each 0.5% ponds and large wood. Vegetated habitats within the study site (islands, riparian forest), dominated by Populus nigra and five species of Salix (Karrenberg et al. 2003), ensure supply of woody debris (Gurnell et al. 2000), which is seen as a major factor for the functioning of process dynamics in floodplain systems.

Figure 5: Oblique aerial view of the island-braided section upstream of Pinzano during baseflow conditions illustrating the high habitat diversity in this area. (Photo: S. D. Langhans; July 2005).

The EU-project “Evaluation and improvement of water quality models for application to temporary waters in Southern European catchments” (tempQsim) This PhD was incorporated into the EU-project tempQsim (www.tempqsim.net) whose aim was to improve the understanding of water quality dynamics in temporary waters, and to develop computer models accordingly. Both, temporary streams and riverine floodplains are alternating aquatic and terrestrial

-14- Introduction ecosystems which are expected to provide, in their natural state, multiple services including the provision of clean drinking water, the self purification of waste water, the recharge of ground water, the provision of habitats for a rich terrestrial and aquatic fauna and flora, as well as cultural and aesthetic values. To be able to identify and to quantify such ecosystem services, we need to disentangle the multiple effects of duration, intensity, time, and frequency of surface flow and drying, single and in concert, on both biodiversity and ecosystem processes. Results from the tempQsim-project demonstrate that the effect of inundation duration, following a rainfall pulse, controls the process diversity within the channel network (Fig. 6). The relative extent and the dynamics of the temporary channel network may therefore influence the capacity of the riverine floodplains to produce, transform, and store nutrients and organic matter.

Figure 6: Process diversity as a consequence of the duration of surface wetting. The intensity of a process depends on the status of the surface water (e.g. nutrient concentration) and on the organic matter and nutrient content of the bed sediments (tempQsim-Consortium 2006).

Thesis goals and outline The motivation of this thesis was to improve the understanding of how riverine floodplains are influenced by environmental heterogeneity under near-natural conditions. In the last 10 to 20 years, the need to restore impacted streams in

-15- Introduction

Europe arose from polluted drinking water sources, reduced fish abundance, decreasing biodiversity, and the demand of appealing outdoor recreational places among others. Nowadays, thinking in European dimensions, hydrologically and morphologically intact river systems are very scarce. The Tagliamento River provides the last opportunity to study the basic functioning of ecosystem processes under semi-natural conditions, and therewith to collect information needed for effective river restoration. This work builds up on the experience and knowledge gained in former PhD theses conducted along the Tagliamento River in NE-Italy. Dave Arscott (2002) investigated habitat heterogeneity in relation to macroinvertebrate diversity. Sophie Karrenberg van der Nat (2002) studied patterns of woody pioneer vegetation and regeneration of the main pioneer trees including grey alder, black poplar and five species of willows. Dimitry van der Nat (2002) placed the cornerstone for the understanding of organic-matter dynamics, especially by investigating the importance and turnover of large wood. Edith Kaiser (2002) explored the sources, transformations, and fates of dissolved riverine organic matter. Ute Karaus (2004) surveyed lateral aquatic habitats and explored their ecology. Achim Paetzold (2004) studied the occurrence, composition and the feeding behaviour of riparian arthropods along the shoreline of exposed riverine sediments in the island-braided reach of the Tagliamento. Lastly, Rainer Zah (2001) developed an organic matter budget for the Val Rosegg, a Swiss alpine floodplain.

The presented work contains six interrelated chapters and a summary. In Chapter one, I introduce a conceptual model for braided river-floodplain ecosystems that unifies organic-matter input, quality, storage, and leaf decomposition, and stresses the importance of the flow and inundation regime. In combining these aspects of organic-matter dynamics, which have been treated separately in the ecological literature, the model fosters a more holistic perspective of ecosystem processes in riverine floodplains. Despite growing recognition of the importance of a natural flow regime in riverine floodplains, researchers struggle to quantify ecosystem responses to altered hydrological regimes. How do frequency, timing, and duration of inundation affect fundamental ecosystem processes such as leaf-litter decomposition? This question is addressed in Chapter two.

-16- Introduction

Riverine floodplains consist of a mosaic of aquatic, semi-aquatic and terrestrial habitats which, depending on river flow, are periodically connected and disconnected. Flow and flood pulses also arrange habitats and redistribute organic matter from the river channel to the floodplain, and vice versa, which results in remarkable spatial heterogeneity and temporal dynamics of habitats. Although environmental heterogeneity has been intensively studied, very little is known about how habitat heterogeneity influences ecosystem processes, such as leaf decomposition. The aim of Chapter three was therefore to determine patterns of leaf decomposition across diverse habitat types within a complex floodplain reach in the Tagliamento River system (NE-Italy), and to assess the role that detritivorous invertebrates and fungi play in this system. Chapter four deals with the suitability of cotton strips as a surrogate for natural leaf decomposition. Current understanding of the leaf decomposition process has been advanced through the use of the leaf-litter assay (Petersen and Cummins 1974, Webster and Benfield 1986, Gessner et al. 1999) in which leaf material is incubated in the field and later retrieved to determine leaf-mass remaining. Although this approach is commonly used, it has some shortcomings such as the lack of comparability among different leaf decomposition studies mainly caused by within and/or among leaf species variation. Therefore, we compared cotton-strip decomposition (loss of tensile strength and mass) and leaf decomposition (mass loss) across contrasting floodplain habitats of the Tagliamento River (NE-Italy), to find a relationship between the decomposition rates of leaves and the cotton-strip material. Insights into the functioning of organic-matter dynamics in a riverine floodplain mosaic (Tagliamento River, NE-Italy) are described in Chapter five. Aerial input, lateral transport, and surface storage of particulate organic-matter were quantified over a year, and addressed in an organic-matter budget. Chapter six elaborates on terrestrial invertebrate distribution in four dimensions of a riverine floodplain (air space, sediment surface, unsaturated zone, and time) focusing on the beetle fauna. Although the importance of exposed riverine sediment surfaces for the conservation of a specialized fauna has been acknowledged (Eyre et al. 2002, Eyre 2006, Bates et al. 2006), very little is known about how, when, and even whether the huge volume of unsaturated zone sediments is used by invertebrates. The research of this thesis was funded by the EU community, supported through the tempQsim-project (contract no. EVK1-CT-2002-0012), and

-17- Introduction financed by the Swiss State Secretariat for Education and Research (SBF no. 02.0072).

References Arscott, D. B. 2001. Habitat heterogeneity and aquatic invertebrates along an Alpine floodplain river. PhD dissertation. ETH Zurich. pp. 238. Arscott, D. B., K. Tockner, D. van der Nat, and J. V. Ward. 2002. Aquatic habitat dynamics along a braided alpine river ecosystem (Tagliamento River, Northeast Italy). Ecosystems 5: 802-814. Bates, A. J., J. Sadler, and A. P. Fowles. 2006. Condition-dependent dispersal of a patchily distributed riparian ground beetle in response to disturbance. Oecologia 150: 50-60. Bone, T. L., and D. L. Balkwill. 1988. Morphological and cultural comparison of microorganisms in surface soil and subsurface sediments at a pristine study site in Oklahoma. Microbial Ecology 16: 49-64. Boulton, A. J., and P. I. Boon. 1991. A review of methodology used to measure leaf litter decomposition in lotic environments: time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. Briers, R. A., H. M. Cariss, R. Geoghegan, and J. H. R. Gee. 2005. The lateral extent of the subsidy from an upland stream to riparian lycosid spiders. Ecography 28: 165-170. Burton, T. M. 1976. An analysis of feeding ecology of the salamanders (Amphibia: Urodela) of the Hubbard Brook Experimental Forest, New Hampshire. Journal of Herpetology 10: 187-204. Chergui, H., and E. Pattee. 1990. The processing of leaves of trees and aquatic macrophytes in the network of the River Rhone. International Revue of Hydrobiology 75: 281-302. Corbet, P. S. 1964. Temporal patterns of emergence in aquatic insects. Canadian Entomologist 96: 264-279. Doering, M., U. Uehlinger, A. Rotach, D. R. Schlaepfer, and K. Tockner. 2007. Ecosystem expansion and contraction dynamics along a large Alpine alluvial corridor (Tagliamento River, Northeast Italy). Earth Surface Processes and Landforms. In press. Eyre, M. D., M. L. Luff, and D. A. Lott. 2002. The importance of exposed riverine sediments for phytophagous beetles (Coleoptera) in Scotland and northern England. Aquatic Conservation: Marine and Freshwater Ecosystems 12: 553-566. Eyre, M. D. 2006. A strategic interpretation of beetle (Coleoptera) assemblages, biotopes, habitats and distribution, and the conservation implications. Journal of Insect Conservation 10: 151-160. Flint, A. L., L. E. Flint, G. S. Bodvarsson, E. M. Kwicklis, and J. Fabryka- Martin. 2001. Evolution of the conceptual model of unsaturated zone hydrology at Yucca Mountain, Nevada. Journal of Hydrology 247: 1-30.

-18- Introduction

Fontana, A., P. Mozzi, and A. Bondesan. In press. Alluvial megafans in the Venetian-Friulian Plain (North-eastern Italy): evidence of sedimentary and erosive phases during Late Pleistocene and Holocene. Quaternary International. Framenau, V., R. Manderbach, and M. Baehr. 2002. Riparian gravel banks of upland and lowland rivers in Victoria (south-east Australia): arthropod community structure and life- history patterns along a longitudinal gradient. Australian Journal of Zoology 50: 103-123. Gessner, M. O., E. Chauvet, and M. Dobson. 1999. A perspective on leaf litter breakdown in streams. Oikos 85: 377-384. Gurnell, A. M., G. E. Petts, N. Harris, J. V. Ward, K. Tockner, P. J. Edwards, and J. Kollmann. 2000. Large wood retention in river channels: the case of the Fiume Tagliamento, Italy. Earth Surface Processes and Landforms 25: 225-275. Gurtz, M. E., and C. M. Tate. 1988. Hydrologic influences on leaf decomposition in a channel and adjacent bank of a gallery forest stream. American Midland Naturalist 120: 11-21. Hedges, J. I., G. L. Cowie, J. E. Richey, P. D. Quay, R. Benner, M. Strom, and B. R. Forsberg. 1994. Origins and processing of organic matter in the Amazon River as indicated by carbohydrates and amino acids. Limnology and Oceanography 39: 743-761. Holden, P. A., and N. Fierer. 2005. Microbial processes in the vadose zone. Vadose Zone Journal 4: 1-21. Hutchens, J. J., and J. B. Wallace. 2002. Ecosystem linkages between southern Appalachian headwater streams and their banks: leaf litter breakdown and invertebrate assemblages. Ecosystems 5: 80-91. Hynes, H. B. N. 1975. The stream and its valley. Verhandlungen internationale Vereinigung Limnologie 19: 1-15. Iwata, T. 2003. The roles of fluvial geomorphology in the trophic flow from stream to forest ecosystems. PhD dissertation. Kyoto University. Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river-floodplain systems. In: Proceedings of the International Large River Symposium, Dodge, D. P. (Ed.). Special Publications of Fisheries and Aquatic Sciences. pp. 110-127. Kaiser, E. 2002. Sources, transformation, and fates of riverine organic matter. PhD dissertation. ETH Zürich. pp. 170. Karaus, U. 2004. The ecology of lateral aquatic habitats along river corridors. PhD dissertation. ETH Zürich. pp. 177. Karrenberg van der Nat, S. 2002. Tree regeneration on the flood plain of an Alpine river. PhD dissertation. ETH Zürich. pp. 143. Karrenberg, S., J. Kollmann, P. J. Edwards, A. M. Gurnell, G. E. Petts. 2003. Patterns in woody vegetation along the active zone of a near-natural Alpine river. Basic and Applied Ecology 4: 157-166. Langhans, S. D., and K. Tockner. 2006. The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy). Oecologia 147: 501-509.

-19- Introduction

Langhans, S. D., S. D. Tiegs, U. Uehlinger, and K. Tockner. 2006. Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems. Polish Journal of Ecology 54: 675-680. Likens, G. E., and F. H. Bormann. 1974. Linkages between terrestrial and aquatic ecosystems. BioScience 24: 447-456. Malard, F., U. Uehlinger, R. Zah, and K. Tockner. 2006. Flood-pulse and r iverscape dynamics in a braided glacial river. Ecology 87: 704-716. Manderbach, R., and D. Hering. 2001. Typology of riparian ground beetle communities (Coleoptera, Carabidae, Bembidion spec.) in Central Europe and adjacent areas. Archiv für Hydrobiologie 152: 583-608. Mayack, D. T., J. H. Thorp, and M. Cothran. 1989. Effects of burial and floodplain retention on stream processing of allochthonous litter. Oikos 54: 378-388. McIntosh, A. R. 2000. Aquatic predator-prey interactions. In: New Zealand Stream Invertebrates: Ecology and Implications for Management, Collier, K. J., and M. J. Winterbourn (eds.). New Zealand Limnological Society, Christchurch, New Zealand. pp. 125-155. Merritt, R. W., and D. L. Lawson. 1992. The role of leaf litter macroinvertebrates in stream- floodplain dynamics. Hydrobiologia 248: 65-77. Müller, N. 1995. River dynamics and floodplain vegetation and their alterations due to human impact. Archiv für Hydrobiologie, Supplement 101: 477-512. Nakano, S., and M. Murakami. 2001. Reciprocal subsidies: dynamic interdependence between terrestrial and aquatic food webs. Proceedings of the National Academy of Science, U.S.A. 98: 166-170. Neatrour, M. A., J. R. Webster, and E. F. Benfield. 2004. The role of floods in particulate organic matter dynamics of a southern Appalachian river-floodplain ecosystem. Journal of the North American Benthological Society 23: 198-213. Paetzold, A. 2004. Life at the edge – aquatic-terrestrial interactions along rivers. PhD dissertation. ETH Zürich. pp. 159. Paetzold, A., C. J. Schubert, and K. Tockner. 2005. Aquatic-terrestrial linkages along a braided river: Riparian arthropods feeding on aquatic insects. Ecosystems 8: 748-759. Petersen, R. C., and K. W. Cummins. 1974. Leaf processing in a woodland stream. Freshwater Biology 4: 343-368. Power, M. E., and W. E. Rainey. 2000. Food webs and resource sheds: towards spatially delimiting trophic interactions. In: Ecological Consequences of Habitat Heterogeneity, Hutchings, M. J., E. A. John, and A. J. A. Stewart (eds.) Blackwell Scientific, Oxford, U.K. pp. 291-314. Power, M. E., W. E. Rainey, M. S. Parker, J. L. Sabo, A. Smyth, S. Khandwala, J. C. Finlay, F. C. McNeely, K. Marsee, and C. Anderson. 2004. River to watershed subsidies in an old-growth conifer forest. In: Food Webs at the Landscape Level, Polis, G. A., M. E.

-20- Introduction

Power, and G. R. Huxel (eds.). The University of Chicago Press, Chicago, IL, U.S.A. pp. 217-240. Sabo, J. L., and M. E. Power. 2002. River-watershed exchange: effects of riverine subsidies on riparian lizards and their terrestrial prey. Ecology 83: 1860-1869. Sadler, J. P., D. Bell, and A. Fowles. 2004. The hydroecological controls and conservation value of beetles on exposed riverine sediments in England and Wales. Biological Conservation 118: 41-56. Selker, J. S., C. K. Keller, and J. T. McCord. 1999. Vadose zone processes: An introduction to the vadose zone. CRC Press LLC, Florida, USA. pp. 3-19. Sparks, R. E., P. B. Bayley, S. L. Kohler, and L. L. Osbourne. 1990. Disturbance and recovery of large floodplain rivers. Environmental Management 14: 699-709. Stanford, J. A. 1998. Rivers in the landscape: introduction to the special issue on riparian and groundwater ecology. Freshwater Biology 40: 402-406. Stanford, J. A., M. S. Lorang, F. R. Hauer. 2005. The shifting habitat mosaic of river ecosystems. Verhandlungen der Internationalen Vereinigung für Limnologie 29: 123-136. Stanley, E. H., S. G. Fisher, and N. B. Grimm. 1997. Ecosystem expansion and contraction in streams. BioScience 47: 427-435. Sukhacheva, G. A. 1996. Study of the natural diet of adult dragonflies using an immunological method. Odonatologica 25: 397-403. Sweeney, B. W., and R. L. Vannote. 1982. Population synchrony in mayflies: a predator satiation hypothesis. Evolution 36: 810-821. tempQsim-Consortium. 2006. Enduser Summary of the EU-project: Evaluation and improvement of water quality models for application to temporary waters in Southern European catchments - tempQsim. pp. 71. Thorp, J. H., and M. D. Delong. 1994. The riverine productivity model: an heuristic view of carbon sources and organic processing in large river ecosystems. Oikos 70: 305-308. Tiegs, S. D, and M. M. Pohl. 2005. Planform channel dynamics of the lower Colorado River: 1976-2000. Geomorphology 69: 14-27. Tiegs, S. D. 2006. Landscape controls of litter decomposition in streams. PhD dissertation. ETH Zürich. pp. 187. Tindall, J. A., J. R. Kunkel, and D. E. Anderson. 1999. Unsaturated zone hydrology for scientists and engineers. Prentice-Hall, Inc., New Jersey, USA. Tockner, K., and J. A. Stanford. 2002. Riverine flood plains: present state and future trends. Environmental Conservation 29: 308-330. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: A model ecosystem of European importance. Aquatic Sciences 65: 239-253. van der Nat, D. 2002. Ecosystem processes in the dynamic Tagliamento River (NE-Italy). PhD dissertation. ETH Zürich. pp. 159.

-21- Introduction

Ward, J. V. 1998. Riverine landscapes: biodiversity patterns, disturbance regimes, and aquatic conservation. Biological Conservation 83: 269-278. Ward, J. V., K. Tockner, P. J. Edwards, J. Kollmann, G. Bretschko, A. M. Gurnell, G. E. Petts, and B. Rossaro. 1999. A reference river system for the Alps: the ´Fiume Tagliamento´. Regulated Rivers: Research and Management 15: 63-75. Ward, J. V., and K. Tockner. 2001. Biodiversity: towards a unifying theme for river ecology. Freshwater Biology 46: 807-819. Ward, J. V., K. Tockner, D. B. Arscott, and C. Claret. 2002. Riverine landscape diversity. Freshwater Biology 47: 517-539. Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Review of Ecology and Systematics 17: 567-594. Zah, R. 2001. Patterns, pathways, and trophic transfer of organic matter in a glacial stream ecosystem in the Alps. PhD dissertation. ETH Zürich. pp. 139.

-22-

CHAPTER 1

Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems

Simone D. Langhans, Scott D. Tiegs, Urs Uehlinger, and Klement Tockner 2006 Polish Journal of Ecology (invited paper) 54: 675-680

Environmental heterogeneity is a key regulator of ecological processes. Riverine floodplains are particularly heterogeneous and dynamic systems and loss of their natural environmental heterogeneity and dynamism as a consequence of human impacts constitutes their most serious threat. On river floodplains, flow and flood pulses create a shifting mosaic of channels, ponds, bars, islands, and riparian forest patches. Composition and spatial arrangement of these habitat patches determine their degree of connectivity, which in turn controls the flux of matter and energy among adjacent patches. In light of these attributes, riverine floodplains are model ecosystems for studying the effect of heterogeneity on ecological processes. In this article we introduce a conceptual model for river- floodplain ecosystems that unifies leaf decomposition, organic-matter input, storage and quality, and stresses the importance of the flow and inundation regime. In combining these aspects of organic matter dynamics, which have been treated separately in the ecological literature, this model fosters a more holistic perspective of ecosystem processes on riverine floodplains. We conclude that the linkage between natural heterogeneity and ecosystem processes needs to be considered in future river-floodplain restoration projects.

-23- Heterogeneity controls OM dynamics

Environmental heterogeneity and the shifting habitat mosaic Environmental heterogeneity is defined as variance in patterns and processes over space and time (Kolasa and Rolle 1991). Physical complexity of an ecosystem, i.e., composition and configuration of habitats, is the key component of heterogeneity and one that mediates ecosystem functioning (Cardinale et al. 2002). Natural riverine floodplains are among the most diverse and dynamic of all Earth’s ecosystems (Tockner and Stanford 2002, Tiegs and Pohl 2005). Riverine floodplains consist of a shifting mosaic of aquatic, semi-aquatic, and terrestrial landscape elements. While these elements change their location, size and configuration over time, the overall abundance of various elements often remains constant, a phenomenon referred to as “the shifting habitat mosaic” (Arscott et al. 2002, Stanford et al. 2005). If such a habitat mosaic consists of elements which exchange mass, energy, organisms, or information with one another, as it is the case for riverine floodplain habitats, they can be referred to as interactive (Lovett et al. 2005). Here, we introduce a conceptual model of how environmental heterogeneity controls coarse-particulate organic-matter dynamics. Our model stems from ongoing research of the braided Tagliamento River in northeastern Italy (46° N, 12°30’ E). The Tagliamento is among the last large reference rivers in Europe and offers a rare opportunity to study patterns and processes across different scales under near-natural conditions (Tockner et al. 2003).

Organic matter and decomposition dynamics are heterogeneous In functioning as a vital source of carbon and energy to streams and rivers, allochthonous organic matter links terrestrial with aquatic ecosystems, and decomposition of this matter is a key ecosystem-level process (Webster and Benfield 1986, Langhans and Tockner 2006). Organic matter accumulates and is temporarily stored in depositional areas where it is pre-processed prior to its further transfer to other habitats (Mayack 1989, Merritt and Lawson 1992). Particularly during floods, exchange of organic matter between the river and its adjacent riparian zone can be wide spread (Neatrour et al. 2004). Decomposition of leaf litter has been widely investigated in both aquatic and terrestrial environments (Boulton and Boon 1991, Gessner et al. 1999), but few studies have addressed decomposition of leaf litter across heterogeneous environments such as river-floodplain ecosystems (Chergui and Pattee 1990, McArthur et al.

-24- Chapter 1

1994, Baldy et al. 2002). Physico-chemical conditions, composition and abundance of the decomposer community, leaf-litter quality (Gessner and Chauvet 1994), and the flow regime differ across these environments and control decomposition processes (Langhans and Tockner 2006). Rapid decomposition can occur under permanent inundation (Herbst and Reice 1984) as well as under highly fluctuating water levels (Battle and Golladay 2001). By separating the effects of duration and frequency of inundation, Langhans and Tockner (2006) found that duration of inundation is the main inundation component that controls leaf decomposition in gravel bed rivers. Additionally, leaf decomposition was found to be faster in channels than in ponds in a floodplain reach of the Tagliamento River (S. D. Langhans - unpublished), and faster in humid than in dry riparian habitats (Hutchens and Wallace 2002). In order to understand organic-matter dynamics in complex river-floodplain ecosystems, aspects of aquatic and terrestrial leaf decomposition need to be linked with input, storage, and quality of organic matter, as well as with the character of the flow regime. The conceptual model we present here unifies components of organic-matter dynamics and illustrates how these components, which have been treated separately in the ecological literature, interact on riverine floodplains.

Organic-matter dynamics: A conceptual model for river-floodplain ecosystems Unconstrained river-floodplain ecosystems, such as the Tagliamento River, are characterized by a heterogeneous mosaic of channels, parafluvial and orthofluvial ponds, exposed gravel, vegetated islands and riparian forest (Fig. 1). Composition and spatial arrangement of these habitats determine the interface character among adjacent habitats, which in turn controls the flux of energy and matter through the ecosystem (Naiman and Decamps 1997). In contrast to terrestrial habitats, the length of aquatic-terrestrial interfaces, which determines the interaction potential among adjacent habitats, varies with river stage. The connectivity among and within river-floodplain habitats is controlled by fluvial dynamics which are driven by the flow regime and morphology of the floodplain (Fig. 1A). In regulated compared to natural river systems for example, frequency and duration of inundation often decrease and the exchange of material becomes less frequent and more pulsed. Riparian forests and vegetated islands are highly productive (Naiman and Decamps 1997) and contribute resources to less-

-25- Heterogeneity controls OM dynamics productive aquatic and terrestrial habitats. Spatial arrangement of individual habitat patches governs input and distribution of leaf litter across the floodplain. For example, aerial and lateral litter input declines exponentially with distance from their source (Fig. 1B, Zah and Uehlinger 2001), and varies with wind speed and direction, and slope of the terrain. Due to the heterogeneous nature of river-floodplain ecosystems, the same habitat type, for example parafluvial ponds, can display highly different input-characteristics depending on its position relative to other habitats. Organic matter standing stock (Fig. 1C) integrates input (transport across boundaries), retention, and decomposition. Physical retention depends on relief and vegetation cover, as well as on large wood and channel roughness (Bilby and Likens 1980, Gurtz et al. 1988). Losses include decomposition and flood-related export. A major component of deposited coarse particulate organic matter is leaves (van der Nat 2002), which in temperate zones enter the system during abscission in autumn. Leaf-litter quality is highly variable across a floodplain transect and often decreases from the active zone towards the riparian forest (Fig. 1D) because of the subsequent increasing proportion of hardwood tree species (Ostrofsky 1997). This gradient of leaf-litter quality is related to the successional stage of the vegetation, which in turn is determined by the disturbance regime. The decomposition of leaf litter is strongly affected by its quality (Naiman et al. 2005). For example, lignin content of leaf litter was found to be the best predictor for decomposition in aquatic and terrestrial systems (Gallardo 1993, Gessner and Chauvet 1994). Therefore, floodplain habitats supplied with higher quality leaves (i.e., lower lignin content) experience faster decomposition rates than habitats with input of low quality leaves. However, leaf decomposition potential (i.e., the maximal capacity of a habitat to decompose leaves not limited by environmental factors) in floodplain habitats is highest in channels (i.e., highest decomposition coefficient k), medium in ponds, and lowest in terrestrial habitats (S. D. Langhans - unpublished, Fig. 1E).

-26- Chapter 1

Figure 1: A conceptual model for braided river-floodplains based on observations from the Tagliamento River (NE-Italy). Characterization of aquatic and terrestrial habitats across a river-upland transect: Components of the natural flow regime include duration (% of time) and frequency (number of events) of inundation (A). Input (g dry mass m-2; B) and distribution heterogeneity (g dry mass m-2; C) of particulate organic matter. Leaf-litter quality of donor trees (D). Leaf decomposition potential (k = decomposition coefficient (d-1); E). LW, MW, HW = low, mean and high water level; 100 year = flood event with a 100 years recurrent interval.

-27- Heterogeneity controls OM dynamics

In natural river-floodplain ecosystems, the area of aquatic habitats relative to other floodplain habitats can be small and these less common habitats have a significantly higher decomposition potential than extensive terrestrial habitats (i.e., exposed gravel, riparian forest). Interestingly, despite having the highest decomposition potential, aquatic habitats have low organic-matter input and standing stock. Large areas with relatively low decomposition potential often have high organic-matter input and standing stock (e.g., vegetated islands and riparian forest patches, Fig. 1). This mismatch effect is reduced when floods mix and redistribute organic matter and deposit it during falling water level. Pre- processed organic matter with a more homogenous quality is now more evenly distributed among floodplain habitats. Therefore, at local scales flood pulses control leaf decomposition via the flow regime and more broadly by linking sources of organic matter with areas of high decomposition potential.

Implications Worldwide, the loss of heterogeneity, largely through alteration of the flow regime and channelization, is the most serious threat to the ecological integrity of riverine floodplains (Dobson et al. 1997, Bunn and Arthington 2002). Flow regulation reduces flood intensity, duration and frequency (McMahon and Finlayson 2003). Regulated systems mostly lack channel-forming flows and intense water flux through the alluvial aquifer leading to a loss in dynamics and heterogeneity of the shifting habitat mosaic (Stanford et al. 2005). Parafluvial and orthofluvial ponds, alluvial channels and vegetated islands disappear and habitat diversity decreases. Lateral connectivity among floodplain habitats is severed, and flow and flood pulses no longer function as reorganizers of organic matter which can lead to drastic changes on the ecosystem process level. Additionally, channelization results in a loss of areas with low decomposition potential, such as gravel bars, and in an increase in the relative proportion of high decomposition potential areas, such as channels. As a result, the decomposition process and the transport of organic matter are altered with potential consequences for higher trophic levels in downstream ecosystems. An additional effect of flow regulation and channelization is that the input of leaves into active floodplain habitats decreases, whereas storage of leaf litter in vegetated zones (e.g., riparian forest, islands) increases. As a result, diversity within the litter layer decreases, which can negatively affect litter turnover rates (Hoorens et al. 2003).

-28- Chapter 1

It is crucial to quantify effects of environmental heterogeneity on ecosystem- level processes to understand the links between heterogeneity and the functioning of river-floodplain ecosystems. Knowledge of natural variance in such environments should be integrated in future restoration approaches, which so far are often site specific and therefore do not consider the heterogeneous character of riverine systems. Our model promotes a holistic perspective including interactions among habitats in a mosaic shaped environment. In the future, heterogeneity needs to be seen as an inherent part in river restoration projects.

Acknowledgements We thank the SEFS4 Special Issue PJE Editors for the invitation to contribute to this issue “Advances in European Freshwater Sciences, Krakow 2005”. This work was supported by the EU project “tempQsim” (EVK1-CT2002-0012) and by the Swiss State Secretariat for Education and Research SER (No. 02.0072).

References Arscott, D. B., K. Tockner, D. van der Nat, and J. V. Ward. 2002. Aquatic habitat dynamics along a braided alpine river ecosystem (Tagliamento River, Northeast Italy). Ecosystems 5: 802-814. Baldy, V., E. Chauvet, J. Y. Charcosset, and M. O. Gessner. 2002. Microbial dynamics associated with leaves decomposing in the mainstem and floodplain pond of a large river. Aquatic Microbiology Ecol 28: 25-36. Battle, J. M., and S. W. Golladay. 2001. Hydroperiod influence on breakdown of leaf litter in cypress-gum wetlands. American Midland Naturalist 146: 128-145. Bilby, R. E., and G. E. Likens. 1980. Importance of organic debris dams in the structure and function of stream ecosystems. Ecology 61: 1107-1113. Boulton, A. J., and P. I. Boon. 1999. A review of methodology used to measure leaf litter decomposition in lotic environments: time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. Bunn, S. E., and A. H. Arthington. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30: 455-467. Cardinale, B. J., M. A. Palmer, C. M. Swan, S. Brooks, and N. L. Poff. 2002. The influence of substrate heterogeneity on biofilm metabolism in a stream ecosystem. Ecology 83: 412- 422. Chergui, H., and E. Pattee.1990. The processing of leaves of trees and aquatic macrophytes in the network of the River Rhone. International Revue of Hydrobiology 75: 281-302.

-29- Heterogeneity controls OM dynamics

Dobson, A. P., A. D. Bradshaw, and A. J. M. Baker. 1997. Hopes for the future: restoration ecology and conservation biology. Science 277: 515-522. Gallardo, A., and J. Merino. 1993. Leaf decomposition in two mediterranean ecosystems of southwest Spain: influence of substrate quality. Ecology 74: 152-161. Gessner, M. O., and E. Chauvet. 1994. Importance of stream microfungi in controlling breakdown rates of leaf litter. Ecology 75: 1807-1817. Gessner, M. O., E. Chauvet, and M. Dobson. 1999. A perspective on leaf litter breakdown in streams. Oikos 85: 377-384. Gurtz, M. E., G. R. Marzolf, K. T. Killingbeck, D. L. Smith, and J. V. McArthur. 1988. Hydrologic and riparian influences on the import and storage of coarse particulate organic matter in a prairie stream. Canadian Journal of Fisheries and Aquatic Sciences 45: 655- 665. Herbst, G., and S. R. Reice. 1982. Comparative leaf litter decomposition in temporary and permanent streams in semi-arid regions of Israel. Journal of Arid Environments 5: 305- 318. Hoorens, B., R. Aaerts, and M. Stroetenga. 2003. Does initial litter chemistry explain litter mixture effects on decomposition? Oecologia 137: 578-586. Hutchens, J. J., and J. B. Wallace. 2002. Ecosystem linkages between southern Appalachian headwater streams and their banks: leaf litter breakdown and invertebrate assemblages. Ecosystems 5: 80-91. Kolasa, J., and C. D. Rolle. 1991. Introduction: the heterogeneity of heterogeneity: a glossary. In: Ecological heterogeneity, J. Kolasa, and S. T. A. Pickett (eds.). Springer-Verlag, New York, U.S.A. pp. 1-23. Langhans, S. D., and K. Tockner. 2006. The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy). Oecologia 147: 501-509. Lovett, G. M., C. G. Jones, M. G. Turner, and K. C. Weathers. 2005. Ecosystem function in heterogeneous landscapes. Springer Science and Business Media, Inc., New York, U.S.A., pp. 2-3. Mayack, D. T., J. H. Thorp, and M. Cothran. 1989. Effects of burial and floodplain retention on stream processing of allochthonous litter. Oikos 54: 378-388. McArthur, J. V., J. M. Aho, R. B. Rader, and G. L. Mills. 1994. Interspecific leaf interactions during decomposition in aquatic and floodplain ecosystems. Journal of the North American Benthological Society 13: 57-67. McMahon, T. A., and B. L. Finlayson. 2003. Droughts and anti-droughts: the low flow hydrology of Australian rivers. Freshwater Biology 48: 1147-1160. Merritt, R. W., and D. L. Lawson. 1992. The role of leaf litter macroinvertebrates in stream- floodplain dynamics. Hydrobiologia, 248: 65-77. Naiman, R. J., and H. Décamps. 1997. The ecology of interfaces: the riparian zone. Annual Revue of Ecology and Systematics 28: 621-658.

-30- Chapter 1

Naiman, R. J., H. Décamps, and M. E. McClain. 2005. Riparia: ecology, conservation, and management of streamside communities. Elsevier / Academic Press, San Diego, U.S.A. pp. 430. Neatrour, M. A., J. R. Webster, and E. F. Benfield. 2004. The role of floods in particulate organic matter dynamics of a southern Appalachian river-floodplain ecosystem. Journal of the North American Benthological Society 23: 198-213. Ostrofsky, M. L. 1997. Relationship between chemical characteristics of autumn-shed leaves and aquatic processing rates. Journal of the North American Benthological Society 16: 750-759. Stanford, J. A., M. S. Lorang, and F. R. Hauer. 2005. The shifting habitat mosaic of river ecosystems. Verhandlungen Internationaler Vereinigung für Limnologie 29: 123-136. Tiegs, S. D., and M. Pohl. 2005. Planform channel dynamics of the lower Colorado River: 1976-2000. Geomorphology 69: 14-27. Tockner, K., and J. A. Stanford. 2002. Riverine flood plains: present state and future trends. Environmental Conservation 29: 308-330. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: a model ecosystem of European importance. Aquatic Sciences 65: 239-253. van der Nat, D. 2002. Ecosystem processes in the dynamic Tagliamento River (NE-Italy). PhD dissertation. ETH Zürich. pp. 159. Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Revue of Ecology and Systematics 17: 567-594. Zah, R., and U. Uehlinger. 2001. Particulate organic matter inputs to a glacial stream ecosystem in the Swiss Alps. Freshwater Biology 46: 1597-1608.

-31-

-32-

CHAPTER 2

The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy)

Simone D. Langhans and Klement Tockner 2006 Oecologia 147: 501-509

Despite growing recognition of the importance of a natural flow regime in river- floodplain systems, researchers struggle to quantify ecosystem responses to altered hydrological regimes. How do frequency, timing, and duration of inundation affect fundamental ecosystem processes such as leaf litter decomposition? Along the semi-natural Tagliamento River corridor, located in northeastern Italy, we employed in situ experiments to separate effects of different inundation components on breakdown rates of black poplar (Populus nigra). We used a litter-bag method with two different mesh sizes to investigate how fungi and macroinvertebrates influence leaf breakdown rates. Ten treatments, each representing a specific combination of duration and frequency of inundation, were deployed in two seasons (summer, winter) to mimic complex inundation patterns. After 30 days of exposure, mean percentage of remaining leaf litter (AFDM) ranged between 51% (permanent wet) and 88% (permanent dry). Leaf breakdown was significantly faster in winter than in summer. Duration of inundation was the main inundation component that controlled leaf breakdown rates. Leaf-shredding macroinvertebrates played only a role in the permanent wet treatment. Fungal parameters explained the faster leaf breakdown in winter. Our study suggests that modifications of the inundation regime will directly modify established decomposition processes. Factors reducing duration of inundation will decelerate leaf breakdown rates, whereas a decrease in flow variation will reduce leaf breakdown heterogeneity.

-33- Inundation dynamics and leaf decomposition

Introduction A natural inundation regime is regarded as the key driver of river-floodplain ecosystems (Junk et al. 1989, Poff et al. 1997, Tiegs and Pohl 2005). The magnitude, frequency, duration, and timing of the inundation regime influence biotic communities and ecosystem processes, either directly or indirectly through their effects on other primary regulators (Hart and Finelli 1999, Hession et al. 2000, Tockner et al. 2000, Robertson et al. 2001). Modification of the inundation regime thus has cascading effects on the ecological integrity of river- floodplain systems (Bunn et al. 2002). In a wide range of aquatic and terrestrial environments, decomposition of plant litter is a very important process in ecosystem metabolism and a driving force in nutrient cycling (Wallace et al. 1997, Webster et al. 1999, Cleveland et al. 2004). Much research has been conducted to identify the controlling factors and underlying mechanisms of decomposition (e.g., Robinson and Gessner 2000, Shaw and Harte 2001, Austin 2002, Hieber and Gessner 2002). In wetlands (marshes, swamps, floodplains), the flow regime has been identified as the primary driver of decomposition processes (Molles et al. 1995, Glazebrook and Robertson 1999, Andersen et al. 2003). Drying and wetting define the physical nature of the decomposition environment (Tockner et al. 2000), alter litter quality (Harner and Stanford 2003), and change nutrient conditions (Robertson et al. 1999, Heffernan and Sponseller 2004). However, major inconsistencies exist in inundation-decomposition relationships. Decomposition can be hindered by anoxic conditions due to standing-water conditions or sediment deposition (Cuffney and Wallace 1987, Chauvet 1988) or enhanced by physical (i.e., leaching or fragmentation by stream flow or alternating dry/wet cycles) and biological processes (i.e., microbial activity or invertebrate consumption) (Ryder and Horwitz 1975, Brinson 1977, Molles et al. 1995, Lockaby et al. 1996, Battle and Golladay 2001). Most field studies have used a gradient or a correlative approach (e.g., comparing leaf mass loss with flood duration), which limits the ability to elucidate causal mechanisms (e.g., Brinson 1977, Shure et al. 1986, Graça et al. 2001, Austin 2002). However, hydrologic manipulation experiments are often difficult to interpret because the duration, frequency, and timing of inundation are poorly defined (Lockaby et al. 1996). Microcosm experiments in the laboratory offer limited value because they do not reflect natural conditions (Day 1983, Hagvar 1988).

-34- Chapter 2

In the present study, we designed intensive in situ experiments to mimic complex inundation patterns that were once characteristic for many temperate floodplain rivers (Poff et al. 1997, Tockner et al. 2003). We asked two key questions: First, to what extent do duration, frequency, and timing (season) of inundation influence leaf mass loss? Second, what is the role of microbes (i.e., aquatic fungi) and aquatic and terrestrial macroinvertebrates in controlling rates of leaf litter decomposition under different inundation regimes? We hypothesize that a complex inundation regime maintains high heterogeneity in leaf litter decomposition at the river-floodplain scale.

Methods Study site In situ experiments were carried out in summer 2003 and winter 2004 along the braided reach of the Fiume Tagliamento, a 7th order gravel-bed river in NE Italy (46° N, 12°30’ E) with a total catchment area of 2580 km2. The corridor is fringed by continuous riparian woodland. Despite local water abstraction and a channelized downstream section, the river retains essentially pristine morphological and hydrological characteristics. The flow regime is highly dynamic with frequent flow pulses (sensu Tockner et al. 2000) throughout the year (Arscott et al. 2002, van der Nat et al. 2002). Our study was conducted in an island-braided reach (river-km 79.8 – 80.8; 135 m a.s.l.) where lateral riparian forests and vegetated islands supply the floodplain with allochthonous organic material. Populus nigra and five species of Salix dominate the vegetation (Karrenberg et al. 2003). Average standing stocks of deposited CPOM (mainly leaves) across different floodplain strata ranged from < 1g m-2 ash free dry mass (AFDM) on bare gravel surface to > 1000 g m-2 AFDM on vegetated islands and riparian forests. In aquatic habitats, average annual CPOM standing stock was 10 g m-2 AFDM (van der Nat 2002). Detailed information on the catchment, the main study reach, and the water chemistry is provided by Ward et al. (1999), Tockner et al. (2003), and Kaiser et al. (2004). Water chemistry, air temperature, and relative humidity at the study site, measured during the experiments, are summarized in Table 1.

-35- Inundation dynamics and leaf decomposition

Table 1: Physico-chemical characterization of the study site in summer and winter (n = number of measurements, mean ± 1 SD). Summer Winter Environmental parameter n Mean n Mean Air temperature (daytime) 360 32.8 ± 6.0 380 5.9 ± 6.0 Air temperature (nighttime) 360 19.0 ± 2.8 380 0.0 ± 3.1 Relative humidity (%) 30 55.9 ± 6.2 30 66.6 ± 12.5 Water temperature (°C) 730 16.5 ± 2.4 730 9.0 ± 0.7 Flow velocity (m s-1) 5 0.49 ± 0.11 7 1.05 ±0.11 Specific conductance (µS cm-1) 5 492.0 ± 1.6 3 649.7 ± 5.4 pH 5 8.0 ± 0.0 3 8.2 ± 0.1 Oxygen (%) 5 72.7 ± 11.4 3 132.5 ± 22.8 Total inorganic carbon (mg L-1) 4 28.5 ± 4.0 4 33.5 ± 2.2 Total organic carbon (mg L-1) 4 1.6 ± 0.2 4 0.5 ± 0.2 -1 a NH4-N (µg L ) 4 n.d. 4 4.0 ± 2.0 -1 NO3-N (µg L ) 4 531.2 ± 54.3 4 837.0 ±126.3 -1 NO2-N (µg L ) 4 1.9 ± 1.7 4 10.0 ± 1.5 -1 a a PO4-P (µg L ) 4 n.d. 4 n.d. a n.d. = not detectable

Experimental treatment The experiment was conducted in a 200 m long riffle-run section of a 20 m wide braided channel (detailed description: Paetzold et al. 2005). We used a random- block design in which four blocks (10 treatments each) were established along the riffle-run section (Fig. 1). Each treatment represented a specific combination of duration and frequency of inundation. Therefore, the effect of duration and frequency could be analysed separately. To simulate wet/dry cycles, leaves enclosed in nylon bags were transferred manually from instream to terrestrial (wet to dry) positions and vice versa.

-36- Chapter 2

Figure 1: Experimental design: Ten treatments (T) with different inundation regimes (white: wet; gray: dry conditions). Number: days. D: duration of inundation (days), F: number of dry/wet cycles (frequency).

To partition the effects of shredders and microorganisms, breakdown of leaves enclosed in fine mesh bags, which preclude shredder access, was compared with leaf breakdown in coarse mesh bags. The experiment was designed for 30 days to have a trade-off between considerable leaf breakdown and reduced flood risk (Arscott et al. 2002). Water chemistry (nitrogen, phosphorus, and carbon species) and environmental parameters (temperature, conductivity, oxygen, vegetation cover, and gravel size) were similar among blocks.

Sampling procedures Experiments were carried out in summer 2003 (28 July - 27 August) and winter 2004 (8 January - 7 February). Leaves of Populus nigra (black poplar) were collected during senescence in autumn 2002 (summer experiment) and 2003 (winter experiment) directly from trees adjacent to the study site. Leaves were air-dried and stored under dry conditions. Portions of 5 ± 0.25 g were weighed, moistened and packed in fine mesh (0.5 mm mesh size) and coarse mesh (10 mm mesh size) nylon bags (Boulton and Boon 1991). Pairs of fine and coarse mesh bags were tied to iron bars and exposed in the stream bed and along the terrestrial shore corresponding to the experimental design (treatment). During the summer experiment, one coarse mesh bag out of 80 was lost because of

-37- Inundation dynamics and leaf decomposition vandalism. In both experiments, bags were retrieved after 30 days, placed individually in polyethylene bags and returned to the laboratory in cooling boxes.

Laboratory procedures Leaves were carefully removed from the bags and gently rinsed with tap water to remove adhering debris and invertebrates. The remaining slurry was filtered through a 500 µm mesh and the retained invertebrates were preserved in 70% ethanol. Fifteen leaf discs (diameter: 12 mm) were cut from 5 different leaves (3 discs per leaf) from each bag using a cork borer. Two sets of five discs were placed in a small polyethylene bag and frozen at -20° C for ergosterol analyses. The ergosterol content of decaying litter, an indicator of fungal biomass, was estimated according to Gessner and Schmitt (1996). The third set of five leaf discs was put in a separate aluminum pan, and dried to constant mass at 60° C for 48 h together with the remaining decomposed leaves, before weighing to the nearest 0.1 g. Total dry mass was determined by adding the dry mass of the leaf discs (times three) to the main bulk of leaves. Ground 0.5 g subsamples were ashed (500° C, 6 h) to determine ash free dry mass (AFDM). Subsamples of initial leaves were processed in the same way to develop air dry mass to AFDM conversion factors. The preserved invertebrates were sorted, identified to the lowest possible taxon, and counted. Taxa were assigned to functional feeding groups according to Tachet et al. (2000). Statistical analysis We used analyses of variance (ANOVA) with subsequent Scheffé’s post-hoc tests to test for differences among treatment means (expressed as percent of remaining AFDM; Wieder and Lang 1982), with block, season, duration, frequency, and mesh size as independent variables. If season and mesh size had significant influences on treatment means, we performed an ANOVA for each season and mesh size separately. Data were square root-arcsine transformed to meet ANOVA assumptions. The same ANOVA model was applied to analyze differences among mean fungal biomass (expressed as ergosterol). Leaf breakdown rates (k day-1) were calculated using an exponential decay model (Webster and Benfield 1986). We applied the ANOVA model as described above with k-values as dependent variable. As both analyses provided the same results, only data from percent of remaining AFDM were presented.

-38- Chapter 2

Differences in macroinvertebrate abundance and shredder abundance among treatments 1, 3, and 7 were calculated using a MANOVA model. Wilks’ Lamda test statistics was used to assess treatment effects. Prior to the analyses, data were square root transformed to meet ANOVA assumptions. The relationship between fungal biomass and remaining AFDM was determined by Spearman rank correlation. All statistical analyses were performed using SPSS (ver. 11.0/SPSS Inc., Illinois, USA) with significance levels set at P ≤ 0.05. Our sampling design allowed us to statistically separate the effects of duration and frequency on leaf breakdown based on the calculation of partial residual plots (Ryan 1997). Depending on the treatment combination (duration/frequency), the parameter estimate value for the specific duration was subtracted to evaluate the effect of the corresponding frequency treatment, and vice versa.

Results Effect of timing (season) Leaf breakdown (expressed as percent remaining AFDM after 30 days of exposure) was significantly higher in winter than in summer (ANOVA: F1,147 = 124.16; p < 0.001). In the summer experiment, mean percent of remaining AFDM ranged from 51.4% to 80.5% in coarse mesh bags and from 57.3% to 83.3% in fine bags. In the winter experiment, 41.3% to 82.3% AFDM remained in coarse mesh bags and 49.2% to 88.4% in fine ones (Fig. 2).

-39- Inundation dynamics and leaf decomposition

Figure 2: Mass loss (mean ± 1 SE) of black poplar leaves in coarse and fine mesh bags with different hydrologic regimes in (a) summer and in (b) winter. (Treatments see Fig. 1; n = 3-4).

Effects of duration and frequency In the summer and winter experiment, leaf breakdown was fastest in permanently inundated and slowest in permanently dry treatments (both mesh sizes) (Tab. 2). In both seasons, duration had significant effects on remaining AFDM, while frequency was only significant in the summer experiment (Tab. 3). Subsequent post-hoc tests showed that all duration treatments differed significantly from each other, whereas leaves inundated for 30 days broke down fastest, followed by leaves inundated for 20, 10 and 0 days. In summer, leaf breakdown from treatments with 1 dry/wet cycle was fastest differing significantly from treatments with 5 and with 0 dry/wet cycles. Leaves which were exposed to 10 and 15 dry/wet cycles broke down significantly faster than leaves, which underwent 0 changes (Scheffé’s post-hoc test).

-40- Chapter 2

Table 2: Summary of breakdown coefficients (k day-1) for summer and winter experiments in coarse and fine mesh bags (n = 4, mean ± 1 SE). Summer Winter Treatment Coarse mesh Fine mesh Coarse mesh Fine mesh 1 0.0222 ± 0.0010 0.0185 ± 0.0000 0.0296 ± 0.0014 0.0238 ± 0.0015 2 0.0181 ± 0.0012 0.0152 ± 0.0005 0.0240 ± 0.0006 0.0241 ± 0.0012 3 0.0158 ± 0.0005 0.0155 ± 0.0014 0.0235 ± 0.0011 0.0250 ± 0.0014 4 0.0156 ± 0.0004 0.0135 ± 0.0001 0.0251 ± 0.0008 0.0242 ± 0.0005 5 0.0182 ±0.0005 0.0174 ± 0.0003 0.0229 ± 0.0007 0.0210 ± 0.0007 6 0.0129 ± 0.0004 0.0126 ± 0.0010 0.0169 ± 0.0003 0.0167 ± 0.0003 7 0.0146 ± 0.0002 0.0143 ± 0.0014 0.0155 ± 0.0006 0.0160 ± 0.0004 8 0.0139 ± 0.0006 0.0131 ± 0.0002 0.0158 ± 0.0002 0.0156 ± 0.0003 9 0.0143 ± 0.0009 0.0140 ± 0.0006 0.0167 ± 0.0005 0.0155 ± 0.0004 10 0.0072 ± 0.0004 0.0061 ± 0.0001 0.0065 ± 0.0003 0.0041 ± 0.0001

In summer, we also found a significant 2-way interaction between duration and frequency (Tab. 3). Leaves from treatments with 1 or 15 dry/wet cycles broke down significantly faster when they were inundated 20 instead of 10 days, whereas leaves from treatments with 5 or 10 dry/wet cycles decomposed only slightly faster when inundated 20 days.

Table 3: Values of F from two-way analyses of variance (ANOVA) with percent of remaining AFDM as dependent variable and block, mesh size, frequency, and duration as independent factors for summer and winter experiments (n = 79-80 per season). Degrees of freedom for each effect are given in parentheses. Effect Remaining AFDM (%) Summer Winter Block (3) 3.56* 0.87 Mesh size (1) 8.41** 8.71** Frequency (3) 4.67** 1.42 Duration (2) 210.31*** 530.98*** Frequency X Duration (3) 3.16* 1.87 * P < 0.05; ** P < 0.01; *** P < 0.001

Analyzing coarse mesh bags for both seasons separately, we found significant effects for duration but not for frequency (Tab. 4). Subsequent post-hoc tests

-41- Inundation dynamics and leaf decomposition revealed that leaf breakdown differed significantly among all duration treatments with fastest rates during permanent inundation. Leaf breakdown in fine mesh bags was significantly influenced by duration and frequency in summer and by duration in winter (Tab. 4). In summer, leaves which were inundated 30 days decomposed fastest, those inundated 0 days slowest. Leaf breakdown in treatments with 10 and 20 days of inundation duration was similar to each other. Frequency affected leaf decomposition as follows: leaves from treatments with 0 dry/wet cycles decomposed significantly slower than leaves from treatments with 1, 10, and 15 dry/wet cycles. In winter, 30 and 20 days of inundation caused fastest leaf breakdown differing significantly from 10 and 0 days of inundation.

Table 4: Values of F from two-way analyses of variance (ANOVA) with percent of remaining AFDM as dependent variable and block, frequency, and duration as independent factors for summer and winter experiments, with coarse and fine mesh bags separated (n = 39-40 per season and mesh size). Degrees of freedom for each effect are given in parentheses. Summer experiment Winter experiment Effect Coarse mesh Fine mesh Coarse mesh Fine mesh Block (3) 1.64 2.02 0.78 0.87 Frequency (3) 1.45 3.82* 0.77 2.49 Duration (2) 113.53*** 96.71*** 385.58*** 320.69*** Frequency X Duration (3) 2.27 1.44 1.29 1.29 * P < 0.05; ** P < 0.01; *** P < 0.001

Effects of macroinvertebrates and fungi In both seasons, leaf breakdown was significantly faster in coarse than in fine mesh bags (Tab. 3). Differences between coarse and fine mesh bags were greatest under permanent wet (5.97% of remaining AFDM in summer and 7.86% in winter) and permanent dry conditions (2.75% in summer and 6.15% in winter). Mesh size was only found to have significant effects on leaf breakdown if permanent wet and dry treatments were included in statistical analyses. In both seasons, numbers of macroinvertebrates were high, when, at the end of the experiment, bags were retrieved from wet conditions (treatments 1, 3 and 7) (Tab. 5). Where bags were retrieved from dry conditions (all other treatments), the number of macroinvertebrates was low (< 2 ind. bag-1). In summer, total macroinvertebrate and shredder abundance were similar in treatments 1, 3, and 7

-42- Chapter 2

(MANOVA: Wilks’ λ = 0.952, F4,16 = 0.10, p = 0.981). In winter, total macroinvertebrate and shredder abundance were higher in treatments 1 and 3 than in treatment 7, although not significantly different (MANOVA: Wilks’ λ =

0.348, F4,16 = 2.78, p = 0.063).

Table 5: Total macroinvertebrate abundance (N bag-1) and shredder abundance (N bag-1) in coarse mesh bags for summer and winter experiments (n = 4, mean ± 1 SE). Summer Winter Total Shredder Total Shredder abundance abundance abundance abundance Treatment 1 37.8 ± 10.8 20.8 ± 8.2 243.8 ± 64.0 23.0 ± 2.7 Treatment 3 37.8 ± 4.6 17.8 ± 3.9 149.3 ± 31.1 19.3 ± 3.2 Treatment 7 33.5 ± 5.9 16.5 ± 5.2 42.0 ± 8.3 8.8 ± 1.7

Initial fungal biomass (0.011 mg ergosterol per g AFDM) indicated minor fungal colonization of freshly collected leaves. The ANOVA model showed that the ergosterol content of leaves was significantly higher in winter than in summer (ANOVA: F1,148 = 9.58; p = 0.002). In summer, mesh size (ANOVA:

F1,69 = 26.31; p < 0.001) and duration (ANOVA: F2,69 = 6.38; p = 0.003) had significant effects on the ergosterol content of leaves, whereas frequency was only marginally significant (ANOVA: F3,69 = 2.60; p = 0.059). Fungal biomass was higher in coarse mesh bags, and treatments with permanent inundation were significantly different from all other treatments (Fig. 3a). However in winter, duration (ANOVA: F2,66 = 254.99; p < 0.001), frequency (ANOVA: F3,66 = 7.86; p < 0.001), and the interaction term of both (ANOVA: F1,69 = 3.50; p = 0.002), but not mesh size (ANOVA: F1,66 = 0.002; p = 0.968) significantly influenced fungal biomass. Leaves inundated for 20 and 30 days showed significantly higher fungal biomass than leaves which were inundated for 10 days. Permanent dry leaves showed lowest fungal biomass. Treatments with 5 dry/wet cycles showed highest ergosterol values, significantly different from treatments with 0, 10, and 15 dry/wet cycles (Fig. 3b). The significant interaction between duration and frequency resulted from the fact that the increase of fungal biomass observed from 10 to 20 days of inundation was greater in treatments with 1 or 10 dry/wet cycles than in treatments with 5 or 15 dry/wet cycles. In both seasons, leaf litter breakdown rate (percent of remaining AFDM) was significantly

-43- Inundation dynamics and leaf decomposition correlated with ergosterol concentration (Spearman rank correlation; summer, r = -0.335, P = 0.003; winter, r = -0.849, P < 0.001).

Figure 3: Ergosterol content of leaves (mg gAFDM-1) in coarse and fine mesh bags with different hydrologic regimes in (a) summer and in (b) winter. (Treatments see Fig. 1; n = 4).

Discussion Effects of season and inundation regime In our study, leaf litter breakdown varied considerably between seasons and among treatments. Leaves broke down faster during the winter experiment. Leaves under permanent inundated conditions lost almost 50% of its organic matter within 30 days, indicating a high decomposition rate compared to estimates from other temperate regions (e.g. Baldy et al. 1995, Pereira et al. 1998, Menéndez et al. 2003). Under dry conditions, leaf breakdown was similar

-44- Chapter 2 to previously observed winter rates in the Tagliamento system (S.D. Langhans, unpublished data). Leaf litter processing in aquatic and terrestrial habitats is the result of factors such as leaf-litter quality (Shaw and Harte 2001), temperature (Irons et al. 1994), relative humidity (Kuehn et al. 2004), dissolved oxygen (Chergui and Pattee 1988), dissolved nutrient concentrations (Robinson and Gessner 2000), and abundance and diversity of organisms responsible for decomposition (Dangles and Malmqvist 2004). In our study, oxygen, nutrient concentrations, velocity, and relative humidity were higher, and temperatures were lower in the winter experiment. Increased oxygen and relative humidity levels favour aerobic microbial respiration, and high concentrations of dissolved nutrients accelerate the processing of leaf litter (Robinson and Gessner 2000). Current velocity, ranging from 0.02 to 1.20 m s-1, was found not to alter leaf breakdown rates (V. J. L. Ferreira et al., unpublished data). The effect of differing thermal regimes on leaf litter decomposition shows a clear tendency toward faster breakdown rates with increasing temperatures (e.g., Carreiro and Koske 1992, Menéndez 2003). Apparently, in our experiments, the effect of temperature was most likely masked by other factors such as increased humidity and nutrient concentrations in winter. Generally, mass loss from decaying leaf litter is faster in aquatic than in terrestrial systems because of enhanced leaching and microbial metabolisms (Molles et al. 1995). Similarly in our experiments, duration of inundation rather than frequency was the “master variable” controlling leaf breakdown in both seasons and mesh sizes. Our results correspond to those of Herbst and Reice (1984), who also found faster leaf breakdown rates in a permanent stream relative to a temporary and an intermittent stream (Tanninim and Daliya River, Israel). However, Battle and Golladay (2001) observed greatest decomposition under the most fluctuating hydrologic regime (Dougherty Plain, Georgia, USA). Their study was conducted under standing water conditions, where oxygen depletion may be a controlling factor. Oxygen was never a limiting factor and fine sediment deposition is not prevalent in the braided section of the Tagliamento River where we worked. In both seasons, the effect of duration on percent of remaining AFDM was particularly high if leaves were inundated only for few days. With increasing inundation time, the effect of duration decreased to zero. Frequency showed no correlation with percent of remaining AFDM (Fig. 4). Our winter results are

-45- Inundation dynamics and leaf decomposition comparable with those of Lockaby et al. (1996) who suggested fastest decomposition under single inundation events. Mitsch and Gosselink (1993) stated that decomposition rates are most rapid under aerobic but moist conditions. These conditions existed in the hibernal humid climate in NE Italy, and intensified the effects we have found for one dry/wet cycle. During the hot and dry summer, only frequent dry/wet cycles could enhance decomposition.

Figure 4: The effect of (a) duration (days) and (b) frequency (dry/wet cycles) of inundation on % of remaining AFDM in summer (n = 79) and in winter (n = 80) experiments calculated on the basis of remaining AFDM (%).

Effects of macroinvertebrates and fungi Our second research question focused on the role of fungi and aquatic and terrestrial macroinvertebrates in controlling leaf litter decomposition under different inundation regimes. Shredding macroinvertebrates significantly increased leaf breakdown only during permanent wet conditions (on average by 3.3% to 16.0%). Under permanent dry conditions, faster leaf breakdown in coarse mesh bags might have been caused by leaching because shredders were absent in these bags. Leaves in fine mesh bags seemed to be protected from heavy leaching by the mesh coverage. The minimal difference between coarse and fine mesh bags in all other treatments and both seasons suggested that dry/wet cycles were too frequent to establish feeding effects by macroinvertebrates on leaves. Physical abrasion must have played a minor role in the permanent wet treatment, otherwise, leaves in coarse mesh bags inundated for 20 and 10 days would have shown significantly faster breakdown rates than leaves in fine mesh bags.

-46- Chapter 2

In many aquatic ecosystems, fungi (hyphomycetes) play the predominant role in the microbial decomposition of leaves (Hieber and Gessner 2002). In contrast to bacteria, fungal assemblages are able to respond immediately to rewetting (Fierer et al. 2003). Therefore, alternating dry/wet cycles have only a small influence on the activity of fungal decomposers which might explain why duration was most important in our experiments. In a recent study, Kuehn et al. (2004) demonstrated that microbial activity on Phragmites australis leaves exhibited a diel periodicity with highest activity during night when relative humidity peaked. This finding suggests, why, in our case, fungal biomass was higher in winter than in summer. Even under terrestrial conditions, the humid winter climate in NE-Italy provided enough moisture for microbial activity. Therefore, increased microbial activity accounted for higher leaf breakdown rates during the winter. However, hot and dry summers (as in 2003) may reduce microbial activity, because leaching of polyphenols via rainfall and dew might be necessary before litter can be broken down (Boulton 1991). This explanation agrees with our finding that, if we considered leaf breakdown without feeding effects (fine mesh bags), significant duration effects in the moist season (winter), and significant duration and frequency effects in the dry season (summer) were found. Additionally, fungal biomass explained the effect of duration of inundation on leaf breakdown. However in winter, leaf breakdown behaved differently under the frequency treatments than fungal biomass. We suggest that frequency effects on fungal biomass are too small to influence leaf litter breakdown.

The relationship between hydrology and the decomposition process In headwater streams, leaf litter enters the wetted channel during the autumnal litter fall period, thus input and processing overlap spatially and temporally. In floodplains, however, areas of production, storage and processing of leaves are spatially and temporally separated. Across a river-floodplain transect, standing stock of leaf litter biomass can change over four-orders of magnitude, from < 1 (bare gravel surface) to about 1000 g (vegetated islands) AFDM per m-2 (van der Nat 2002). While vegetated islands and the fringing riparian forest are the primary input and storage areas, aquatic habitats, in particular channels are the fastest processing areas. In such habitats, breakdown rates are several times faster compared to terrestrial areas. Most leaf litter is transferred from input/storage areas to river channels (fast processing areas) during periods with

-47- Inundation dynamics and leaf decomposition over bank flow. A human-regulated hydrologic regime often changes the duration, frequency, and magnitude of flow and flood pulses (Richter 1997, Poff and Hart 2002). In flow controlled systems, not only the amount of conveyed organic matter but also the transport distance and the breakdown dynamics are restricted.

Concluding remarks Most attention on altered flow regimes has focused on their effects upon water quality and biota. There are only a few studies that rigorously tested for their impacts on ecosystem functioning (e.g., leaf litter decomposition) (Bunn et al. 2002). Results from our study showed that riverine floodplains display a remarkable heterogeneity in their ability to process organic matter, similar (albeit more complex) to riparian margins in small headwater streams (Hutchens and Wallace 2002). Modifications of the inundation regime will directly (as shown in this study) and indirectly (i.e., by changing litter composition and quality) influence decomposition processes. Factors reducing duration of inundation will decelerate breakdown rates, whereas, a decrease in flow variation will reduce breakdown heterogeneity. If floodplain inundation changes the riparian vegetation will alter too (e.g., Busch 1995, Tiegs et al. 2005). This can result in dramatic consequences for leaf litter quality and quantity. In addition, the complex mosaic of terrestrial (e.g., bare gravel, islands, riparian forest) and aquatic (e.g., ponds, backwaters, channels) habitats, which ensure large spatial and temporal variation in litter dynamics, might be reduced to a less heterogeneous habitat composition. This decline in heterogeneity will weaken aquatic-terrestrial linkages by reducing the exchange of water, nutrients, and leaf litter among adjacent ecosystems (Gurnell et al. 2005).

Acknowledgements We are grateful to Simone Blaser and Claudio Cruciat for assisting in the field and in the laboratory, Richard Illi for conducting chemical analyses, and Bruno Tona for statistical advice. We also thank Scott Tiegs, Darren Bade, Stuart Findlay, Mark Gessner, and two anonymous reviewers for helpful advice that improved the manuscript. The study was supported by the EU-funded project tempQsim (EVK1-CT2002-00112; http://www.tempqsim.net) and by BBW (No. 02.0072). The conducted experiments comply with the current laws of Italy.

-48- Chapter 2

References Anderson, D. C., S. M. Nelson, and D. Binkley. 2003. Flood flows, leaf breakdown, and plant-available nitrogen in a dryland river floodplain. Wetlands 23: 180-189. Arscott, D. B., K. Tockner, and J. V. Ward. 2002. Geomorphic dynamics along a braided-river corridor in the Alps (Fiume Tagliamento, Italy). Ecosystems 5: 803-814. Austin, A. T. 2002. Differential effects of precipitation on production and decomposition along a rainfall gradient in Hawaii. Ecology 83: 328-338. Baldy, V., M. O. Gessner, and E. Chauvet. 1995. Bacteria, fungi and the breakdown of leaf litter in a large river. Oikos 74: 93-102. Battle, J. M., and S. W. Golladay. 2001. Hydroperiod influence on breakdown of leaf litter in cypress-gum wetlands. American Midland Naturalist 146: 128-145. Boulton, A. J. 1991. Eucalypt leaf decomposition in an intermittent stream in south-eastern Australia. Hydrobiologia 211: 123-136. Boulton, A. J., and P. I Boon. 1991. A review of methodology used to measure leaf litter decomposition in lotic environments: Time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. Brinson, M. M. 1977. Decomposition and nutrient exchange of litter in an alluvial swamp forest. Ecology 58: 601-609. Bunn, S. E., and A. H. Arthington. 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environmental Management 30: 492-507. Busch, D. E. 1995. Mechanisms associated with decline of woody species in riparian ecosystems of the southwestern U.S. Ecological Monographs 65: 347-370. Carreiro, M. M., and R. E. Koske. 1992. Effect of temperature on decomposition and development of microfungal communities in leaf litter microcosms. Canadian Journal of Botany 70: 2177-2183. Chauvet, E. 1988. Influence of the environment on willow leaf litter decomposition in the alluvial corridor of the Garonne River. Archiv für Hydrobiologie 112: 371-386. Chergui, H., and E. Pattee. 1988. The impact of benthic invertebrates on the breakdown of poplar leaves in the network of a large European river. Archiv für Hydrobiologie 113: 15- 25. Cleveland, C., J. C. Neff, A. R. Townsend, and E: Hood E 2004. Composition, dynamics, and fate of leached dissolved organic matter in terrestrial ecosystems: results from a decomposition experiment. Ecosystems 7: 275-285. Cuffney, T. F., and J. B. Wallace. 1987. Leaf litter processing in Coastal Plain streams and floodplains of southeastern Georgia, U.S.A. Archiv für Hydrobiologie Supplement 76: 1- 24. Dangles, O., and B. Malmqvist. 2004. Species richness-decomposition relationships depend on species dominance. Ecology Letters 7: 395-402. Day, F. P. 1983. Effects of flooding on leaf litter decomposition in microcosms. Oecologia 56: 180-184.

-49- Inundation dynamics and leaf decomposition

Fierer, N., J. P. Schimel, and P. A. Holden. 2003. Influence of drying-rewetting frequency on soil bacterial community structure. Microbial Ecology 45: 63-71. Gessner, M. O., and A. L. Schmitt. 1996. Use of solid-phase extraction to determine ergosterol concentrations in plant tissue colonized by fungi. Applied and Environmental Microbiology 62: 415-419. Glazebrook, H. S., and A. I. Robertson. 1999. The effect of flooding and flood timing on leaf litter breakdown rates and nutrient dynamics in a river red gum (Eucalyptus camaldulensis) forest. Australian Journal of Ecology 24: 625-635. Graça, M. A. S., R. C. F. Ferreira, and C. N. Coimbra. 2001. Litter processing along a stream gradient: the role of invertebrates and decomposers. Journal of the North American Benthological Society 20: 408-420. Gurnell, A. M., K. Tockner, P. Edwards, and G. Petts. 2005. Effects of deposited wood on biocomplexity of river corridors. Frontiers in Ecology and the Environment 3: 377-382. Hagvar, S. 1988. Decomposition studies in an easily constructed microcosm: effects of microarthropods and varying soil pH. Pedobiologia 31: 293-303. Harner, M. J., and J. A. Stanford. 2003. Differences in cottonwood growth between a losing and a gaining reach of an alluvial floodplain. Ecology 84: 1453-1458. Hart, D. D., and C. M. Finelli. 1999. Physical-biological coupling in streams: The pervasive effects of flow on benthic organisms. Annual Revue of Ecology and Systematics 30: 363- 395. Heffernan, J. B., and R. A. Sponseller. 2004. Nutrient mobilization and processing in Sonoran desert riparian soils following artificial re-wetting. Biogeochemistry 70: 117-134. Herbst, G., and S. R. Reice. 1982. Comparative leaf litter decomposition in temporary and permanent streams in semi-arid regions of Israel. Journal of Arid Environments 5: 305- 318. Hession, W. C., T. E: Johnson, D. F. Charles, D. D. Hart, R. J. Horwitz, D. A. Kreeger, J. E. Pizzuto, D. J. Velinsky, J. D. Newbold, C. Gianfrani, T. Clason, A. M. Compton, N. Coulter, L. Fuselier, B. D. Marshall, and J. Reed. 2000. Ecological benefits of riparian reforestation in urban watersheds: study design and preliminary results. Environmental Monitoring and Assessment 63: 211-222. Hieber, M., and M. O. Gessner. 2002. Contribution of stream detritivores, fungi, and bacteria to leaf breakdown based on biomass estimates. Ecology 83: 1026-1038. Hutchens, J. J. Jr., and B. J. Wallace. 2002. Ecosystem linkages between southern Appalachian headwater streams and their banks: leaf litter breakdown and invertebrate assemblages. Ecosystems 5: 80-91. Irons, J. G., M. W. Oswood, R. J. Stout, and C. M. Pringle. 1994. Latitudinal patterns in leaf litter breakdown: is temperature really important? Freshwater Biology 32: 401-411. Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river-floodplain systems. Canadian Special Publication of Fisheries and Aquatic Sciences 106, pp. 110-127.

-50- Chapter 2

Kaiser, E., D. B. Arscott, K. Tockner, and B. Sulzberger. 2004. Sources and distribution of organic carbon and nitrogen in the Tagliamento River, Italy. Aquatic Sciences 66: 103-116. Karrenberg, S., J. Kollmann, P. J. Edwards, A. M. Gurnell, and G. E. Petts. 2003. Patterns in woody vegetation along the active zone of a near-natural Alpine river. Basic and Applied Ecology 4: 157-166. Kuehn, K. A., D. Steiner, and M. O. Gessner. 2004. Diel mineralization patterns of standing-

dead plant litter: implications for CO2 flux from wetlands. Ecology 85: 2504-2518. Lockaby, B. G., A. L. Murphy, and G. L. Somers. 1996. Hydroperiod influences on nutrient dynamics in decomposing litter of a floodplain forest. Soil Science Society of America Journal 60: 1267-1272. Menéndez, M., O. Hernández, and F. A. Comín. 2003. Seasonal comparisons of leaf processing rates in two Mediterranean rivers with different nutrient availability. Hydrobiologia 495: 159-169. Mitsch, W. J., and J. G. Gosselink. 1993. Wetlands, 2nd ed. Van Nostrand Reinhold, New York, USA. Molles, M. C., C. S. Crawford, and L. M. Ellis. 1995. Effects of an experimental flood on litter dynamics in the middle Rio Grande riparian ecosystem.. Regulated Rivers 11: 275- 281. Paetzold, A., C. Schubert, and K. Tockner. 2005. Aquatic-terrestrial linkages along a braided- river: Riparian arthropods feeding on aquatic insects. Ecosystems 8: 748-759. Pereira, A. P., M. A. S. Graça, and M. Molles. 1998. Leaf litter decomposition in relation to litter physico-chemical properties, fungal biomass, arthropod colonization, and geographical origin of plant species. Pedobiologia 42: 316-327. Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The natural flow regime: a paradigm for conservation and restoration. BioScience 47: 769-784. Poff, N. L., and D. D. Hart. 2002. How dams vary and why it matters for the emerging science of dam removal. BioScience 52: 659-668. Robertson, A. I., S. E. Bunn, P. I. Boon, and K. F. Walker. 1999. Sources, sinks and transformations of organic carbon in Australian floodplain rivers. Marine and Freshwater Research 50: 813-829. Robertson, A. I., P. Bacon, and G. Heagney G. 2001. The responses of floodplain primary production to flood frequency and timing. Journal of Applied Ecology 38: 126-136. Robinson, C. T., and M. O. Gessner. 2000. Nutrient addition accelerates leaf breakdown in an alpine springbrook. Oecologia 122: 258-263. Richter, B. D., J. V. Baumgartner, R. Wigington, and D. P. Braun. 1997. How much water does a river need? Freshwater Biology 37: 231-249. Ryan, T. P. 1997. Modern regression methods. John Wiley and Sons, New York, USA. pp. 145-147.

-51- Inundation dynamics and leaf decomposition

Ryder, D. S., and P. Horwitz. 1975. Seasonal water regimes and leaf litter processing in a wetland on the Swan Coastal Plain, Western Australia. Marine and Freshwater Research 46: 1077-1084. Shaw, M. R., and J. Harte. 2001. Control of litter decomposition in a subalpine meadow- sagebrush steppe ecotone under climate change. Ecological Applications 11: 1206-1223. Shure, D. J., M. R. Gottschalk, and K. A. Parsons. 1986. Litter decomposition processes in a floodplain forest. American Midland Naturalist 115: 314-327. Tachet, H., P. Richoux, M. Bournaud, and P. Usseglio-Polatera. 2000. Invertébrés d’eau douce: systématique, biologie, écologie. CNRS Éditions, Paris, France. Tiegs, S. D., and M. M. Pohl. 2005. Planform channel dynamics of the lower Colorado River: 1976-2000. Geomorphology 69: 14-27. Tiegs, S. D., J. F. O’Leary, M. M. Pohl, and C. L. Munill. 2005. Flood disturbance and riparian species diversity on the Colorado River Delta. Biodiversity and Conservation 14: 1175-1194. Tockner, K., F. Malard, and J. V. Ward. 2000. An extension of the flood pulse concept. Hydrological Processes 14: 2861-2883. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: a model ecosystem of European importance. Aquatic Sciences 65: 239-253. van der Nat, D. 2002. Ecosystem processes in the dynamic Tagliamento River (NE-Italy). PhD dissertation. ETH Zurich. van der Nat, D., A. P. Schmidt, K. Tockner, P. J. Edwards, and J. V. Ward. 2002. Inundation dynamics in braided floodplains: Tagliamento River, Northeast Italy. Ecosystems 5: 636- 647. Wallace, J. B., S.L. Eggert, J. L. Meyer, and J. R. Webster. 1997. Multiple trophic levels for a forested stream linked to terrestrial litter inputs. Science 277: 102-104. Ward, J. V., K. Tockner, P. J. Edwards, J. Kollmann, G. Bretschko, A. M. Gurnell, G. E. Petts, and B. Rossaro. 1999. A reference river system for the Alps: the ‘Fiume Tagliamento’. Regulated Rivers 15: 63-75. Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Revue of Ecology and Systematics 17: 567-594. Webster, J. R., E. F. Benfield, T. P. Ehrmann, M. A. Schaeffer, J. L. Tank, J. J. Hutchens, and D. J. D’Angelo. 1999. What happens to allochthonous material that falls into streams? A synthesis of new and published information from Coweeta. Freshwater Biology 41: 687- 705. Wieder, R. K., G. E. Lang. 1982. A critique of the analytical methods used in examining decomposition data obtained from litter bags. Ecology 63: 1636-1642.

-52-

CHAPTER 3

Leaf decomposition across aquatic and terrestrial habitat types in a riverine floodplain mosaic

Simone D. Langhans*, Scott D. Tiegs*, Mark O. Gessner, and Klement Tockner Submitted *equally contributed to the work

Riverine floodplains consist of a mosaic of aquatic, semi-aquatic and terrestrial habitats which, depending on river flow, are periodically connected and disconnected. Flow and flood pulses also arrange habitats and redistribute organic matter from the river channel to the floodplain, and vice versa, which results in remarkable spatial heterogeneity and temporal dynamics of habitats. Although environmental heterogeneity has been intensively studied, very little is known about how habitat heterogeneity influences ecosystem processes, such as leaf decomposition. The aim of this study was therefore to determine patterns of leaf decomposition across diverse habitat types within a complex floodplain reach and to assess the role that detritivorous invertebrates and fungi play in this system. Black poplar, Populus nigra L., leaves were placed in litter bags (coarse and fine mesh) and exposed in seven contrasting habitat types (total of 28 sites), ranging from the river channel to the riparian forest, on the floodplain of the Tagliamento River (NE-Italy). Three distinct classes of decomposition rates emerged: channels (fast), ponds (medium), and terrestrial sites (slow). In channels and ponds, leaf decomposition appeared to be driven by both microbial and detritivore activity, whereas in terrestrial habitats microbial activity was likely the main driver. Our results demonstrate that braided floodplain rivers provide a wide range of habitats with different decomposition potentials, resulting in a large spatial heterogeneity in the decomposition of organic matter. Alterations to the natural flow regime (e.g., water abstraction, or retention by dams) and morphological changes (e.g., channelization) of braided rivers strongly decrease habitat diversity and can be expected to homogenize decomposition rates within river floodplains, with consequences for ecosystem functioning.

-53- Leaf decomposition in a floodplain mosaic

Introduction Riverine floodplains are ecosystems of global importance (Tockner and Stanford 2002). Defined as areas of low lying land that are subject to inundation by lateral overflow water from rivers with which they are associated (Junk and Welcomme 1990), floodplains consist of a complex mosaic of aquatic, semi- aquatic, and terrestrial habitats (Ward et al. 2002, Naiman et al. 2005). Interaction between physical and biotic processes produces a continually changing spatial pattern of these habitats, which has been conceptualized as the Shifting Habitat Mosaic (Arscott et al. 2002, van der Nat et al. 2003a, Lorang et al. 2005, Stanford et al. 2006). Periodic flooding, repeated erosion and deposition of inorganic sediments, recruitment of large wood, vegetation development, and ground water-surface water exchange processes determine the dynamic character of riverine floodplains (Richter et al. 1997, Tockner and Stanford 2002, Tiegs and Pohl 2005, Stanford et al. 2006). Typical floodplain habitats include lotic channels, lentic water bodies in the active plain (parafluvial ponds) and in the riparian forest (orthofluvial ponds), areas of exposed sediment, vegetated islands, and riparian forests (Fig. 1). The juxtaposition of multiple landscape elements creates highly heterogenous floodplain structure and function (Tockner et al. in press) and increases the spatial variation in organic matter input, storage, and decomposition (Xiong and Nilsson 1997, Rossi and Constantini 2000, Hutchens and Wallace 2002). Riparian zones and aquatic-terrestrial interfaces (e.g., shorelines) function as transition zones and control the transport of nutrients and energy between adjacent habitats (Malanson 1995, Risser 1995, Naiman and Décamps 1997, Ward et al. 1999). Floodplain forests are often highly productive relative to their adjacent uplands (Naiman and Décamps 1997), which results in large inputs of leaf litter to the riparian forest floor and adjacent floodplain habitats. Exchange of organic material among floodplain habitats, for example, from the riparian forest to the river channel and vice versa, can be extensive and is driven by flow and flood pulses (Tockner et al. 2000, Neatrour et al. 2004, Valett et al. 2005). However, during periods between high-flow events, organic material resulting from lateral movement or direct litter fall accumulates on floodplain soils where it is temporarily stored. Here it is colonized by microbes and partly degraded before it enters aquatic habitats (Merritt and Lawson 1992). The storage of organic material on floodplains is seen as an important mechanism to increase

-54- Chapter 3 the efficiency of organic-matter recycling along river corridors (Mayack et al. 1989).

Figure 1: Map of the main study area during base flow, descriptions of the different habitat types (1 to 7), and locations of the 4 sites per habitat type (e.g., 1-1 to 1-4).

Decomposition of plant litter is an important ecosystem process (Aerts 1997, Webster and Benfield 1986, Gessner et al. 1999) and a driving force in nutrient cycling (Cleveland et al. 2004) across a wide range of aquatic and terrestrial environments. It determines the availability of nutrients for organisms (Aber and Melillo 1991) and is a vital component of ecosystem functioning. Leaf decomposition is influenced by the physico-chemical environment (Aerts 1997, Webster and Benfield 1986), composition and abundance of the decomposer community (Petersen and Luxton 1982, Hieber and Gessner 2002), leaf-litter quality (Melillo et al. 1982, Gessner and Chauvet 1994) and, in aquatic environments, the hydrological regime (Ellis et al. 1999, Langhans and Tockner 2006).

-55- Leaf decomposition in a floodplain mosaic

Leaf decomposition has been widely studied in both aquatic and terrestrial environments (Webster and Benfield 1986, Aerts 1997). However, no studies have examined leaf decomposition across the numerous and diverse habitats that are commonly encountered on riverine floodplains (Chauvet 1988, Chergui and Pattee 1988a). Although it is generally acknowledged that natural ecosystems are spatially and temporally complex, heterogeneity tends to be averaged over time and space in ecosystem analyses (Strayer et al. 2003). In view of the complexity of floodplains, however, studies that quantify natural heterogeneity of decomposition are necessary, if ecosystem functioning is to be understood at the landscape scale (Cardinale et al. 2002, Giller et al. 2004). Here, we examine leaf decomposition and the roles of macroinvertebrates and fungi as decomposers across a range of aquatic and terrestrial floodplain habitats characteristic of braided river sections. Specifically, we ask (1) how variable are decomposition rates across floodplain habitats, and (2) what factors (e.g., invertebrates, fungi, temperature) may drive this variability among habitats? We hypothesized that we would find a wide range of leaf decomposition rates across different habitats with faster decay in aquatic than in terrestrial sites. Furthermore, we expected that leaf decomposition in aquatic habitats would mainly be driven by invertebrates, whereas in terrestrial habitats microbial activity would prevail.

Material and methods Site description The study was conducted in the island-braided reach of the Tagliamento River, a 7th order gravel-bed river located in NE-Italy (46°N, 12°30`E; Ward et al. 1999, Tockner et al. 2003). The Tagliamento has a total catchment area of 2580 km2. The active area (parafluvial floodplain) is fringed by continuous riparian forest dominated by black poplar (Populus nigra L.) and five willow species (Salix spp.) (Karrenberg et al. 2003). Despite local water abstraction and channelization of the most downstream section, the Tagliamento River retains an essentially pristine morphological and hydrological character (Ward et al. 1999). It is characterized by a flashy flow regime driven by intense rainfall events in autumn and snowmelt runoff in spring (Arscott et al. 2002). Long-term average discharge in the study reach is 90 m3 s-1, with floods returning on average at 2, 5, and 10 years, estimated at 1100, 1500, and 2150 m3 s-1, respectively (Gurnell et al. 2001). During high flow, large amounts of organic

-56- Chapter 3 matter, including large wood (van der Nat 2003b) and leaf litter (personal observation) are transported downstream and deposited on the floodplain surface. During baseflow, the 1-km2 study reach consists of 42% exposed gravel, 35% riparian forest, 15% channels, 7% islands and each 0.5% ponds and large wood (Fig. 1). The relative proportion of these habitats changes in response to the water level, and peaks in the hydrograph rearrange them regularly (van der Nat et al. 2003a). Average standing stocks of deposited coarse particulate organic matter (CPOM; mainly leaves and large wood) across the floodplain range from < 1g m-2 ash free dry mass (AFDM) on exposed gravel to 1000 g m-2 AFDM on vegetated islands and in the riparian forest. In aquatic habitats, average annual CPOM standing stock ranges from 5 g m-2 AFDM in ponds to 50 g m-2 AFDM in channels (van der Nat 2002). Detailed information on the catchment, main study reach, and water chemistry is provided by Ward et al. (1999), Tockner et al. (2003), and Kaiser et al. (2004). Physical and chemical water parameters, air temperature, and relative humidity, measured during the experiment, are summarized in Table 1. Temperature was continuously recorded using Vemco Minilog data loggers (MINILOG12-TR-40/+50-064K, Vemco, Nova Scotia, Canada). Relative humidity at the terrestrial sites was measured with HOBO Pro RH/Temp data loggers (Bakrona, Zürich, Switzerland).

Field methods A litter-bag method (Boulton and Boon 1991) was used to study decomposition dynamics across seven floodplain habitats: lotic channels, parafluvial ponds, orthofluvial ponds, exposed gravel, large wood accumulations, vegetated islands, and riparian forest (Tiegs et al. 2007, Fig. 1). Fine- and coarse-mesh bags were used to investigate leaf decomposition, an approach which respectively allows and deters access by macroinvertebrates and affords partitioning of the separate influence of microbes and invertebrates/physical abrasion (e.g. Boulton and Boon 1991). A factorial experiment was designed with habitat (seven levels) and mesh size (two levels) as the main factors and decomposition rate (expressed as half-life, T50) as the dependent variable. Four replicate sites of each habitat were randomly selected on the floodplain (28 sites total) (Fig. 1). The experiment was initiated in December 2002, shortly after peak leaf fall in the area. To minimize the risk of losing litterbags due to floods, the experiment

-57- Leaf decomposition in a floodplain mosaic was designed to run for 3.5 months (Arscott et al. 2002) and conducted during baseflow conditions (Fig. 2). Senescent leaves of black poplar (Populus nigra L.) were collected from trees near the study site in the autumn of 2002. Leaves were air-dried to constant weight and then stored in dry conditions. Portions of 5.00 ± 0.25 g were weighed, re-moistened and packed in fine-mesh (0.5 mm mesh size) and coarse-mesh (10 mm mesh size) nylon bags (Boulton and Boon 1991). Five coarse- and fine- mesh bags were tied in pairs to individual iron bars which were hammered into the ground. The five pairs of litterbags were randomly placed in each site on 16 December 2002. In aquatic habitats, cords tied to iron bars were weighted down to fix litterbags on the bottom of channels and ponds. Litterbags in terrestrial sites were placed on the respective surface, such as pebbles in the exposed gravel habitat, and litter in the riparian forest and on islands. In the large wood habitat, litterbags were fixed in the upper area of the accumulation on miscellaneous material including sand, gravel, litter, small twigs and roots. One litterbag pair was randomly retrieved from each replicated habitat after 18, 32, 50, 62, and 80 days (channels), and after 18, 32, 62, 80, and 102 days (all other habitats) (Fig. 2). Litter bags were carefully placed into polyethylene bags, transported to a field research station near the site, and immediately processed.

Figure 2: Changes in rainfall and water level over the study period in 2002/2003. Dates of bag retrieval are indicated with arrows. The black bars represent total daily rainfall, and the line shows the water level at Villuzza located 500 m downstream of the studied floodplain.

-58- Chapter 3

Laboratory procedures Leaves were removed from bags, individually rinsed with water, and carefully cleaned with a brush to remove macroinvertebrates and adhering debris. The resulting slurry was passed over a 100-µm mesh screen and captured invertebrates were preserved in 80% ethanol. Individuals were identified under a microscope, counted, and sorted into functional feeding groups according to Tachet et al. (2000), Freude et al. (2004), and Kerney et al. (1983). At one occasion (80 days after leaf exposure), immediately following the cleaning of leaf material, 10 leaf discs (diameter: 12 mm) were cut from 5 different leaves (2 discs per leaf) from each bag, using a cork borer. One set of 5 discs each were placed in a small polyethylene bag and frozen at -20° C for ergosterol analyses to provide an estimate of fungal biomass. Ergosterol content of decaying litter was quantified according to Gessner and Schmitt (1996) and converted to fungal biomass based on an average ergosterol content of 5.5 mg per g fungal dry mass (Gessner and Chauvet 1993). The second set was placed in a separate aluminum pan, and dried to constant mass at 60° C for 48 h together with the remaining leaves, before weighing to the nearest 0.1 mg. Total leaf dry mass was determined by adding the bulk leaf mass and two times the disc mass. Subsamples of leaves not placed in the field were processed in the same way to establish an air-dry to oven-dry mass relationship.

Data analysis Leaf-decomposition rates (k) were calculated using a negative exponential model (e.g. Webster and Benfield 1986) setting an initial value of 100% at day 0 (i.e. intercept = 100%). Regressions were calculated using non-linear iterative fitting procedures. To standardize leaf decomposition by temperature, decomposition rates (k’) were also calculated on a degree-day (dd) basis (Boulton and Boon 1991) by substituting degree days for time in the negative exponential decay model. Degree days equaled the sum of mean daily ’ temperature during the study period. Half-lives (T50 in days and T50 in degree- days) were calculated from decomposition rates as ln(2)/k and ln(2)/k’, respectively.

To examine differences in decomposition (T50) among habitats and mesh type, a two-way ANOVA was performed with habitat and mesh size as Fixed factors. If mesh size showed a significant effect, the data set was analyzed separately for coarse- and fine-mesh bags. Subsequently, Tukey’s post-hoc tests

-59- Leaf decomposition in a floodplain mosaic

(Janssen and Laatz 2003) were performed to analyze differences between habitats. The same ANOVA model was used to test for differences in temperature-corrected decomposition rates (T50’) and fungal biomass (mg g leaf dry mass-1) among habitats and between mesh types. When Tukey’s tests did not show differences between habitats, weighted paired contrasts (Lindman 1974) were performed to test for differences among three broader habitat categories: channels, ponds, and terrestrial habitats. Weighting was based on the number of habitats within each habitat category. Repeated measures ANOVA with the between-subjects factors mesh size and habitat types followed by Tukey’s tests were carried out to detect differences in the abundance of total macroinvertebrates and leaf-shredding detritivores.

Abundance data were log10(x+1)-transformed to meet ANOVA assumptions. All analyses were performed using SPSS (version 11.0/SPSS Inc., Illinois, USA). For all tests, an α-value of 0.05 was set to assess statistical significance.

Results Physical and chemical characteristics

Means in water depth (F2,51 = 11.0, P > 0.001), pH (F2,45 = 13.8, P > 0.001), and nitrate concentration (F2,57 = 3.6, P = 0.035), differed among habitats (Table 1a). Ponds were deeper than investigated channels, pH was higher in parafluvial ponds and channels than in orthofluvial ponds, and nitrate concentrations were lower in parafluvial ponds than in channels. All other physical and chemical characteristics were similar among aquatic habitats. In terrestrial habitats, mean air temperature was slightly higher on vegetated islands and relative humidity was lower on exposed gravel than in the other terrestrial habitats (Table 1b).

-60- Chapter 3

Table 1: Physical and chemical characteristics of a) aquatic and b) terrestrial habitat types in the Tagliamento River floodplain during the leaf decomposition experiment (mean ± 1 SD, n = 20). a) Aquatic habitats Parameter Channel Parafluvial Orthofluvial pond pond Water temperature (°C)* 9.3 ± 1.1 8.0 ± 1.8 7.4 ± 1.8 Current velocity (m s-1) 0.24 ± 0.29 0 0 Water depth (m) 0.20 ±0.19 0.53 ± 0.35 0.54 ± 0.20 Conductivity (µS cm-1; at 20 °C) 445 ± 127 451 ± 131 465 ± 158 pH 8.2 ± 0.3 8.1 ± 0.3 7.8 ± 0.1 -1 O2 (mg L ) 12.4 ± 2.3 12.5 ± 4.8 11.5 ± 4.3 -1 NH4-N (mg L ) < 0.01 < 0.01 0.01 -1 NO3-N (µg L ) 656 ± 77 395 ± 118 580 ± 325 Soluble reactive phosphorus (µg L-1) < 5 < 5 < 5 b) Terrestrial habitats Parameter Exposed Large Island Riparian gravel wood forest Air temperature (°C)* 3.5 ± 7.7 3.7 ± 7.0 4.1 ± 6.1 3.3 ± 5.5 Relative humidity, RH (%)*,† 72 ± 28 95 ± 10 97 ± 9 96 ± 9 * continuous records, † a single data logger per habitat type.

Leaf mass loss Mean poplar leaf decay in coarse- and fine-mesh bags was fastest in channels and slowest in the riparian forest (Fig. 3). Decomposition rates ranged from - 0.0245 d-1 in channels (coarse-mesh bags) to -0.0029 d-1 on large wood (fine- mesh bags: Table 2), corresponding to half-lives of 36 and 239 days (Fig. 4a).

-61- Leaf decomposition in a floodplain mosaic

Figure 3: Dry mass remaining of poplar leaves decomposing a) in coarse-mesh and b) in fine-mesh bags in seven different habitat types on the Tagliamento floodplain (mean ± 1 SE, n = 4 sites per habitat type).

Table 2: Summary of decomposition rates (k, mean ± 1 SE, n = 4) of black poplar leaves decomposing in seven different floodplain habitat types and two different mesh types as estimated by nonlinear regression analysis. Habitat type Coarse-mesh bags Fine-mesh bags k (d-1) ± SE r2 k (d-1) ± SE r2 Channel -0.0245 ± 0.0134 0.88 -0.0117 ± 0.0009 0.71 Parafluvial pond -0.0071 ± 0.0002 0.61 -0.0070 ± 0.0002 0.47 Orthofluvial pond -0.0073 ± 0.0004 0.56 -0.0071 ± 0.0002 0.54 Exposed gravel -0.0036 ± 0.0001 0.01 -0.0032 ± 0.0001 0.04 Large wood -0.0044 ± 0.0005 0.44 -0.0029 ± 0.0001 -0.05 Island -0.0039 ± 0.0001 -0.01 -0.0035 ± 0.0001 0.14 Riparian forest -0.0035 ± 0.0001 -0.01 -0.0032 ± 0.0001 0.08

Habitat (F6,56 = 163.2, P < 0.001), mesh size (F1,56 = 35.8, P < 0.001), and the interaction between both (F6,56 = 5.3, P < 0.001) significantly influenced leaf decomposition, suggesting that the effect of invertebrates depended on habitat. In all habitats, leaves decomposed faster in coarse-mesh than in fine-mesh bags,

-62- Chapter 3 although the difference was only marginal in ponds (Fig. 4a). For both mesh sizes, post-hoc tests revealed three distinct categories of decomposition rates: channels (fast: coarse-mesh k = -0.0192 d-1, fine-mesh k = -0.0120 d-1), ponds (medium: coarse-mesh k = -0.0079 d-1, fine-mesh k = -0.0078 d-1) and terrestrial habitats (slow: coarse-mesh k = -0.0053 d-1, fine-mesh k = -0.0049 d-1). In fine- mesh bags, leaf decomposition in terrestrial habitats split into two additional groups with significantly slower decomposition rates on large wood than in all other habitats (Fig. 4a).

Figure 4: Half-lives (T50) of poplar leaves decomposing in seven different habitat types on the Tagliamento floodplain (mean ± 1 SE, n = 4 sites per habitat type); a)

T50 values calculated on a daily

basis (d) and b) T50’ on a degree- day (dd) basis (i.e. temperature corrected).

Mean daily temperature during the study period was higher in channels than in ponds, and higher in aquatic than in all terrestrial habitats (Table 1). When decomposition rates were calculated using degree days to account for among- habitat temperature differences, means neither varied among habitats (F6,52 = 2.0,

P = 0.099) nor between different mesh sizes (F1,52 = 3.0, P = 0.094) (Fig. 4b). However, linear contrasts among habitat categories revealed significant

-63- Leaf decomposition in a floodplain mosaic

differences between channels and ponds (F1,38 = 5.3, P = 0.027), and between channels and terrestrial habitats (F1,38 = 7.3, P = 0.010). The same significant differences were found for contrasts calculated with data from coarse-mesh bags only (channels versus ponds: F1,19 = 8.4, P = 0.009; channels versus terrestrial habitats: F1,38 = 8.5, P = 0.009), but not with data from fine-mesh bags.

Density and composition of macroinvertebrates Total macroinvertebrate abundance was significantly lower in fine-mesh than in coarse-mesh bags (F1,42 = 5.7, P = 0.022) and significantly differed among habitats (F6,42 = 22.1, P < 0.001). Channels showed significantly higher numbers of macroinvertebrates than all other habitats, followed by orthofluvial ponds, which had significantly different densities compared to the least colonized habitats, which were large wood and exposed gravel. Most macroinvertebrates colonizing fine-mesh bags in aquatic habitats were early instars of chironomids. Total number of macroinvertebrates in coarse-mesh bags increased significantly over time (Wilks’ λ: F3,19 = 5.0, P = 0.010) and varied among habitats (F6,21 = 24.7, P < 0.001) (Fig. 5). Litter bags from channels had significantly more macroinvertebrates than bags in all other habitats, mainly because of large numbers of chironomids. Weighted contrasts revealed significant differences in macroinvertebrate numbers among the three habitat categories with the highest abundance in channels, medium in ponds, and lowest in terrestrial habitats (channels versus ponds: F1,21 = 82.3, P < 0.001; channels versus terrestrial: F1,21 = 141.0, P < 0.001; ponds versus terrestrial: F1,21 = 6.22, P = 0.02). Total numbers of macroinvertebrates peaked on day 62 in para- and orthofluvial ponds, but continued to rise in channels (Fig. 5).

Habitat type significantly affected total detritivore abundance (F6,21 = 6.16, P = 0.001), with greatest abundances in channels, which were significantly different from exposed gravel, large wood and vegetated islands. Weighted contrasts revealed significantly higher detritivore abundance in channels than in ponds (F1,21 = 10.3, P = 0.004), and in channels than terrestrial habitats (F1,21 = 26.6, P < 0.001), and slightly greater numbers in ponds than in terrestrial habitats (F1,21 = 4.47, P = 0.05). Total number of detritivores did not significantly change over time (Wilks’ λ: F3,19 = 1.99, P = 0.15).

-64- Chapter 3

Figure 5: Number of detritivorous and other macroinvertebrates associated with coarse-mesh bags in seven different habitat types on the Tagliamento floodplain (mean ± 1 SE, n = 4 sites per habitat type).

The aquatic macroinvertebrate community in litter bags exposed in channels was dominated by chironomids and Baetis sp. Stonefly larvae, such as Leuctra sp. and caddisfly larvae (mainly Limnephilidae), were the main shredders present in these bags. In ortho- and parafluvial ponds, the macroinvertebrate community was more diverse than in channels, consisting of snails (Bythinia sp. and Gyraulus sp.), dragonfly larvae (Boyerina irene), chironomids, amphipods (Gammarus sp. and Echinogammarus sp.), and caddisfly larvae (mainly Limnephilidae). In terrestrial habitats, springtails, spiders, beetles (mostly Carabidae and Staphylinidae) and snails were dominant.

Fungi

Fungal biomass differed among habitats after 80 days of leaf exposure (F6,55 = 4.82, P = 0.001). Specifically, fungal biomass was significantly higher in channels than on exposed gravel and in channels than in orthofluvial ponds, and orthofluvial ponds showed lowest fungal biomass which was significantly different from that on large wood (Fig. 6). Weighted contrasts revealed higher fungal biomass in channels than in ponds (F1,41 = 12.3, P = 0.001) and in

-65- Leaf decomposition in a floodplain mosaic

terrestrial habitats (F1,41 = 20.7, P < 0.001). Separate analyses for fungal biomass data from bags with different mesh sizes revealed that fine-mesh bags contributed mostly to these significant contrasts (channels versus ponds: F1,21 =

24.0, P < 0.001; channels versus terrestrial habitats: F1,21 = 13.4, P = 0.001; ponds versus terrestrial habitats: F1,21 = 4.9, P = 0.04), with none of these contrasts being significant for coarse-mesh bags (P = 0.062 to 0.473). The difference in fungal biomass between mesh sizes was not significant (F1,55 = 3.4,

P = 0.07), nor was the interaction between habitat and mesh size (F6,55 = 2.0, P = 0.095).

Figure 6: Fungal biomass (mg g leaf dry mass-1) after 80 days of poplar leaf decomposition in coarse- and fine-mesh bags in seven different habitat types on the Tagliamento floodplain (mean ± 1 SE, n = 4 sites per habitat type).

Discussion Patterns of leaf decomposition across floodplain habitats Leaf-decomposition rates varied widely across the seven floodplain habitats we examined. Leaves decomposed fastest in channels, at intermediate rates in ponds, and slowest in terrestrial habitats, a result consistent with previous studies that compared channels and adjacent terrestrial habitats (Gurtz and Tate 1988), and ponds and terrestrial floodplain sites (McArthur et al. 1994). However, these studies were conducted in lower-order streams, and can therefore not directly be compared with results from large floodplain systems (Melillo et al. 1983) such

-66- Chapter 3 as the braided reach of the Tagliamento River. Studies conducted in the alluvial corridor of the Garonne River, a 7th order river in south-western France, showed no significant differences in leaf-decomposition rates between some aquatic and terrestrial sites (Chauvet 1988) nor between the mainstem and a parafluvial pond on the floodplain (Baldy et al. 2002). Rapid leaf decomposition in aquatic habitats is the result of physical, chemical and biological ecosystem properties associated with water (Hutchens and Wallace 2002). Especially in flowing waters, leaching (Petersen and Cummins 1974) and fragmentation (Heard et al. 1999, but see Ferreira et al. 2006) may interact to promote leaf decomposition more than in other environments. Furthermore, organisms can benefit from higher temperatures in aquatic than in terrestrial sites during the winter (Hutchens and Wallace 2002). Factors related to chemical properties of water, such as the constant supply of nutrients, can enhance leaf decomposition further (e.g., Elwood et al. 1981, Robinson and Gessner 2000, Gulis and Suberkropp 2003). We found that leaves in para- and orthofluvial ponds decomposed more slowly than in channels, but at similar rates to each other, suggesting similar environmental drivers for leaf decomposition in the two different pond types. In terrestrial environments, the effect of local climate on leaf decomposition has often been summarized by a composite variable, actual evapotranspiration, which predicts faster decomposition in warmer, wetter conditions (Meentemeyer 1978a, Aerts 1997). In our study, decomposition rates across terrestrial habitats were similar and exhibited extremely small within-habitat variability. This suggests that among-habitat differences in environmental conditions were insufficient to measurably influence the decomposition process over the 3.5- month study period. Strong temperature dependence of leaf decomposition has been widely reported (e.g., Robinson and Jolidon 2005). In our study, differences in temperature explained some of the observed variability among habitats, but also revealed new perspectives. Differences in leaf decomposition per day between channels, ponds, and terrestrial habitats did not appear in temperature-corrected analyses when using a degree-day model. The only observed differences were in coarse-mesh bags between channels and ponds and between channels and terrestrial habitats. This overall similarity in temperature-corrected rates suggests that temperature may have played a notable role in determining leaf decomposition across both aquatic and terrestrial habitats. In particular, leaf

-67- Leaf decomposition in a floodplain mosaic decomposition in aquatic habitats was apparently promoted by higher temperatures (cf. Webster and Benfield 1986), whereas lower temperatures in terrestrial habitats (Table 1b) appeared to slow decomposition (cf. Meentemeyer 1978b, Hobbie 1996).

Leaf decomposition and invertebrates Differences in total macroinvertebrate and detritivore abundance in litter bags may partly account for the observed differences in leaf decomposition among habitats. Although after temperature correction no overall significant difference in decomposition rate was found between coarse-mesh and fine-mesh bags, rates were clearly faster in channel habitats in coarse-mesh bags (in both temperature- corrected and uncorrected analyses; Fig. 4). This indicates that leaf consumption by macroinvertebrates was greater in channels than in other habitats. However, as the effect of shredding invertebrates and physical abrasion caused by current velocity on leaves can not be separated, faster decay rates in channels might have been a combination of both factors. Overall, numbers of detritivores in litter bags were low (Fig. 5) compared to values reported in the literature (Chergui and Pattee 1988b, Chauvet et al. 1993), but total macroinvertebrate and detritivore abundance were significantly higher in channels than ponds and higher in channels than terrestrial habitats. Given the typically fast decomposition in flowing waters, the major period for leaf decomposition by stream invertebrates is normally in fall and winter (cf. Merritt and Lawson 1992). In terrestrial habitats, the relationship between soil biota and leaf decomposition depends on site-specific characteristics such as soil type and leaf-litter quality (Swift et al. 1979, Heneghan et al. 1998, Heneghan et al. 1999). Additionally, the activity of soil biota is strongly constrained by seasonal climatic patterns (Heneghan et al. 1999) whereby increasing soil moisture is associated with higher microarthopod numbers (Crossley and Hoglund 1962). However, leaf decomposition in terrestrial habitats of floodplain ecosystems is complex as it is time-constrained by spring floods during which large amounts of leaves are washed out of their original habitats.

Leaf decomposition and fungi Fungi play an eminent role in leaf decomposition in flowing waters (Bärlocher and Kendrick 1974, Gessner and Chauvet 1994, Suberkropp 1998), even in larger rivers where direct input of leaf litter is limited (Baldy et al. 1995, 2002).

-68- Chapter 3

Accordingly, we observed highest fungal biomass in channels, intermediate biomass in terrestrial habitats, and lowest biomass in ponds. This is consistent with the findings of Baldy et al. (2002), who found significantly lower fungal biomass in a parafluvial pond compared to a channel of the Garonne river system. Chergui and Pattee (1988b), who studied the spatial distribution of aquatic fungi associated with decomposing black poplar leaves on floodplain habitats of the Rhone River, France, found faster leaf-decomposition rates in channels with present and active aquatic fungi, compared to still-water habitats where practically no aquatic fungi were found. The current in channels supplies organic material with a continuous source of fungal propagules for inoculation, which in ponds is much more restricted. However, more important may be the fact that the fungi most efficient at degrading leaves in streams and rivers (i.e. aquatic hyphomycetes) are much less prevalent in standing waters (Bärlocher 1992), including floodplain ponds (Baldy et al. 2002), and in terrestrial environments (Bärlocher 1992). Consistent with the higher fungal biomass in channels compared to both ponds and terrestrial habitats, decomposition rates were also greatest in channels, suggesting that part of the faster decomposition in flowing water may be attributable to fungal activity, in addition to detritivore feeding. Interestingly, we found some variation in fungal biomass between coarse- and fine-mesh bags among terrestrial habitats, although environmental conditions, such as mean temperature and relative humidity were not notably different. Microclimatic conditions within litter bags may have been responsible for those differences. In contrast to litter bags in open habitats (such as exposed gravel and large wood), where mean temperature in litter bags can be higher and mean relative humidity lower, leaves in bags exposed in wooded habitats (i.e. vegetated islands and riparian forest) can experience lower mean temperature but higher mean relative humidity. Mean relative humidity in wooded habitats can be higher in fine- than in coarse-mesh bags and vice versa in open habitats (S. D. Langhans - unpublished data). This together with fungal biomass data suggests that fungal growth in terrestrial habitats may have been mainly governed by moisture availability, as suggested by Hutchens and Wallace (2002). Additionally, fungal growth in fine-mesh bags in open habitats could have been curbed by high temperatures (Pietikäinen et al. 2005). Thus, the most rapid leaf decomposition among the four terrestrial habitats in fine-mesh bags deployed in large-wood accumulations could be related to more benign (warm,

-69- Leaf decomposition in a floodplain mosaic moist) conditions for microbial decomposers, which is in agreement with the highest fungal biomass observed in fine-mesh bags in this terrestrial floodplain habitat.

Conclusions Rivers consist of a dynamic mosaic of spatially distinct elements. These elements are linked by hydrological connectivity, the exchange of matter, energy and biota which play a major though poorly understood role in sustaining riverine landscapes (Ward et al. 2002). High spatial heterogeneity is often created and maintained by disturbances, including floods, fire, or strong winds (Ward 1998). Investigating the causes and consequences of ecosystem functioning in heterogeneous landscapes is a challenge (Turner and Chapin III, 2005) as an experimental design has to account for spatial heterogeneity and should concurrently consider environmental constraints. Although we are aware that leaf decomposition in terrestrial habitats is influenced not only across spatial gradients but also along temporal gradients, we chose to conduct shorter but simultaneous experiments in all habitats to account for comparability but minimize flood risks. Our results demonstrate that natural floodplain habitats provide a wide range of decomposition potentials. Leaf decomposition in channels and ponds appeared to be driven by both microbes and invertebrates, whereas in terrestrial habitats microorganisms were more important than invertebrates. Channels were clearly identified as “hot spots” of leaf decomposition. Although the seven habitat types we investigated are spatially distinct during baseflow conditions, they become physically connected during periods of high flow. The dynamic flow regime is not only a main determinant of leaf decomposition in floodplains (Langhans and Tockner 2006), but also increases spatial heterogeneity in leaf decomposition in such systems. Morphologically intact floodplains, together with a natural river flow regime, distribute organic matter and the products resulting from decomposition on a large scale, thus ensuring physical connectivity within river corridors and markedly affecting overall ecosystem functioning (i.e. processes such as organic matter decomposition, primary production, and nutrient transformations; Valett et al. 2005). Accordingly, leaf decomposition is controlled at the floodplain scale by tree species composition in specific habitats, the spatio-temporal arrangement of individual habitats, and a dynamic river flow regime which rearranges habitats, distributes organic matter,

-70- Chapter 3 and sustains physical heterogeneity. Modification of flow regimes, such as river regulation, often decreases habitat heterogeneity in floodplain systems and homogenizes spatially-complex variation in decomposition rates even when riparian species composition is unchanged and litter quality thus remains similar. Due to the lack of storage areas for organic matter in such modified systems (Allan and Flecker 1993), the major input of leaf litter in late autumn is quickly decomposed or exported. As a result, nutrients from litterfall are either unavailable or released in a single pulse, which is likely to have repercussions for organisms by affecting their life histories and for ecosystem processes other than litter decomposition (e.g., primary production, sediment/soil carbon mineralization).

Acknowledgements This research was funded by the European Commission, supported by the tempQsim-project (contract no. EVK1-CT-2002-0012) and the Swiss State Secretariat for Education and Research (SBF no. 02.0072). Many thanks to Claudio Cruciat and Simone Blaser for their assistance in the field, and to Richard Illi, Gabriella Meier Bürgisser and Simone Blaser for analytical assistance in the laboratory.

References Aber, J. D., and M. Melillo. 1991. Terrestrial ecosystems: Decay rates and nutrient dynamics of litter and soil organic matter. Saunders College Publishing, Philadelphia, U.S.A. pp. 227-253. Aerts, R. 1997. Climate, leaf litter chemistry and leaf litter decomposition in terrestrial ecosystems: a triangular relationship. Oikos 79: 439-449. Allan, J. D., and A. S. Flecker. 1993. Biodiversity conservation in running waters. BioScience 43: 32-43. Arscott, D. B., K. Tockner, D. van der Nat, and J. V. Ward. 2002. Aquatic habitat dynamics along a braided alpine river ecosystem (Tagliamento River, northeastern Italy). Ecosystems 5: 802-814. Baldy, V., M. O. Gessner, and E. Chauvet. 1995. Bacteria, fungi and the breakdown of leaf litter in a large river. Oikos 74: 93-102. Baldy, V., E. Chauvet, J. Y. Charcosset, and M. O. Gessner. 2002. Microbial dynamics associated with leaves decomposing in the mainstem and floodplain pond of a large river. Aquatic Microbial Ecology 28: 25-36. Bärlocher, F., and B. Kendrick. 1974. Dynamics of fungal population on leaves in a stream. Journal of Ecology 62: 761-79.

-71- Leaf decomposition in a floodplain mosaic

Bärlocher, F. 1992. Community organization. In: The ecology of aquatic hyphomycetes, F. Bärlocher (ed.), Springer, Berlin, Germany. pp. 38-69. Boulton, A. J., and P. I. Boon. 1991. A review of methodology used to measure leaf litter decomposition in lotic environments: Time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. Cardinale, B. J., M. A. Palmer, C. M. Swan, S. Brooks, and N. L. Poff. 2002. The influence of substrate heterogeneity on biofilm metabolism in a stream ecosystem. Ecology 83: 412- 422. Chauvet, E. 1988. Influence of the environment on willow leaf litter decomposition in the alluvial corridor of the Garonne River. Archiv für Hydrobiologie 112: 371-386. Chauvet, E., N. Giani, and M. O. Gessner. 1993. Breakdown and invertebrate colonization of leaf-litter in 2 contrasting streams - significance of Oligochaetes in a large river. Canadian Journal of Fisheries and Aquatic Sciences 50: 488-495. Chergui, H., and E. Pattee. 1988a. The impact of benthic invertebrates on the breakdown of poplar leaves in the network of a large European river. Archiv für Hydrobiologie 113: 15- 25. Chergui, H., and E. Pattee. 1988b. The dynamics of hyphomycetes on decaying leaves in the network of the River Rhone (France). Archiv für Hydrobiologie 114: 3-20. Cleveland, C. C., J. C. Neff, A. R. Townsend, and E. Hood. 2004. Composition, dynamics, and fate of leached dissolved organic matter in terrestrial ecosystems: results from a decomposition experiment. Ecosystems 7: 275-285. Crossley, D. A. Jr., and M. P. Hoglund. 1962. A litter-bag method for the study of microarthropods inhabiting leaf litter. Ecology 43: 571-573. Ellis, L. M., M. C. Molles, and C. S. Crawford. 1999. Influence of experimental flooding on litter dynamics in a Rio Grande riparian forest, New Mexico. Restoration Ecology 7: 193- 204. Elwood, J. W., J. D. Newbold, A. F. Trimble, and R. W. Stark. 1981. The limiting role of phosphorus in a woodland stream ecosystem - effects of P-enrichment on leaf decomposition and primary producers. Ecology 62: 146-158. Ferreira, V., M. A. S. Graca, J. L. M. P. de Lima, and R. Gomes. 2006. Role of physical fragmentation and invertebrate activity in the breakdown rate of leaves. Archiv für Hydrobiologie 165: 493-513. Freude, H., K. W. Harde, and G. A. Lohse. 2004. Die Käfer Mitteleuropas. Band 2: Adephaga 1: Carabidae (Laufkäfer), Spektrum Akademischer Verlag, Heidelberg, Germany. pp. 521. Freude, H., K. W. Harde, and G. A. Lohse. 2004. Die Käfer Mitteleuropas. Band 4: Staphylinidae I Micropeplinae bis Tachyporinae, Spektrum Akademischer Verlag, Heidelberg, Germany. pp. 263. Freude, H., K. W. Harde, and G. A. Lohse. 2004. Die Käfer Mitteleuropas. Band 4: Staphylinidae II Hypocyphtinae und Aleocharinae, Pselaphidae, Spektrum Akademischer Verlag, Heidelberg, Germany. pp. 381.

-72- Chapter 3

Gessner, M. O., and E. Chauvet. 1993. Ergosterol-to-biomass conversion factors for aquatic hyphomycetes. Applied and Environmental Microbiology 59: 502-507. Gessner, M. O., and E. Chauvet. 1994. Importance of stream microfungi in controlling breakdown rates of leaf litter. Ecology 75: 1807-1817. Gessner, M. O., and A. L. Schmitt. 1996. Use of solid-phase extraction to determine ergosterol concentrations in plant tissue colonized by fungi. Applied and Environmental Microbiology 62: 415-419. Gessner, M. O., E. Chauvet, and M. Dobson. 1999. A perspective on leaf litter breakdown in streams. Oikos 85: 377-384. Giller, P. S., H. Hillebrand, U.-G. Berninger, M. O. Gessner, S. Hawkins, P. Inchausti, C. Inglis, H. Leslie, B. Malmqvist, M. Monaghan, P. J. Morin, and G. O’Mullan. 2004. Biodiversity effects on ecosystem functioning: emerging issues and their experimental test in aquatic environments. Oikos 104: 423-431. Gulis, V., and K. Suberkropp. 2003. Leaf litter decomposition and microbial activity in nutrient-enriched and unaltered reaches of a headwater stream. Freshwater Biology 48: 123-134. Gurnell, A. M., G. E. Petts, D. M. Hannah, B. P. G. Smith, P. J. Edwards, J. Kollmann, J. V. Ward, and K. Tockner. 2001. Riparian vegetation and island formation along the gravel- bed Fiume Tagliamento, Italy. Earth Surface Processes and Landforms 26: 31-62. Gurtz, M. E., and C. M. Tate. 1988. Hydrologic influences on leaf decomposition in a channel and adjacent bank of a gallery forest stream. American Midland Naturalist 120: 11-21. Heard, S. B., G. A. Schultz, C. B. Ogden, and T. C. Griesel. 1999. Mechanical abrasion and organic matter processing in an Iowa stream. Hydrobiologia 400: 179-186. Heneghan, L., D. C. Coleman, X. Zou, D. A. Jr. Crossley, and B. L. Haines. 1998. Soil microarthropod community structure and litter decomposition dynamics: a study of tropical and temperate sites. Applied Soil Ecology 9: 33-38. Heneghan, L., D. C. Coleman, X. Zou, D. A. Jr. Crossley, and B. L. Haines. 1999. Soil microarthropod contributions to decomposition dynamics: tropical-temperate comparisons of a single substrate. Ecology 80: 1873-1882. Hieber, M., and M. O. Gessner. 2002. Contribution of stream detritivores, fungi, and bacteria to leaf breakdown based on biomass estimates. Ecology 83: 1026-1038. Hobbie, S. E. 1996. Temperature and plant species control over litter decomposition in Alaskan tundra. Ecological Monographs 66: 503-522. Hutchens, J. J., and J. B. Wallace. 2002. Ecosystem linkages between southern Appalachian headwater streams and their banks: leaf litter breakdown and invertebrate assemblages. Ecosystems 5: 80-91. Janssen, J., and W. Laatz. 2003. Statistische Datenanalyse mit SPSS für Windows. Springer, Berlin, Germany. pp. 329-332.

-73- Leaf decomposition in a floodplain mosaic

Junk, W. J., and R. L. Welcomme. 1990. Floodplains. In: Wetlands and Shallow Continental Water Bodies, B.C. Patten (ed.). SPB Academic Publishing, The Hague, Netherlands. pp. 491-524. Kaiser, E., D. B. Arscott, K. Tockner, and B. Sulzberger. 2004. Sources and distribution of organic carbon and nitrogen in the Tagliamento River, Italy. Aquatic Sciences 66: 103-116. Karrenberg, S., J. Kollmann, P. J. Edwards, A. M. Gurnell, and G. E. Petts. 2003. Patterns in woody vegetation along the active zone of a near-natural Alpine river. Basic and Applied Ecology 4: 157-166. Kerney, M. P., R. A. D. Cameron, and J. H. Jungbluth. 1983. Die Landschnecken Nord- und Mitteleuropas. Verlag Paul Parey, Berlin, Germany. pp. 384. Langhans, S. D., and K. Tockner. 2006. The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy). Oecologia 147: 501-509. Lindman, H. R. 1974. Analysis of variance in complex experimental designs. Freeman and Company, San Francisco. U.S.A. pp. 79-82. Lorang, M. S., D. C. Whited, F. R. Hauer, J. S. Kimball, and J. A. Stanford. 2005. Using airborne multispectral imagery to evaluate geomorphic work across floodplains of gravel bed rivers. Ecological Applications 15: 1209-1222. Malanson, G. P. 1993. Riparian landscapes: Cascades of material and energy. Cambridge University Press, Cambridge. U.K. pp. 130-177. Mayack D. T., J. H. Thorp, and M. Cothran. 1989. Effects of burial and floodplain retention on stream processing of allochthonous litter. Oikos 54: 378-388. McArthur, J. V., J. M. Aho, R. B. Rader, and G. L. Mills. 1994. Interspecific leaf interactions during decomposition in aquatic and floodplain ecosystems. Journal of the North American Benthological Society 3: 57-67. Meentemeyer, V. 1978a. An approach to the biometeorology of decomposer organisms. International Journal of Biometeorology 22: 94-102. Meentemeyer, V. 1978b. Macroclimate and lignin control of litter decomposition rates. Ecology 59: 465-472. Melillo, J. M., J. D. Aber, and J. F. Muratore. 1982. Nitrogen and lignin control of hardwood leaf litter decomposition dynamics. Ecology 63: 621-626. Melillo, J. M., R. J. Naiman, J. D. Aber, and K. N. Eshleman. 1983. The influence of substrate quality and stream size on wood decomposition dynamics. Oecologia 58: 281- 285. Merritt, R. W., and D. L. Lawson. 1992. The role of leaf litter macroinvertebrates in stream- floodplain dynamics. Hydrobiologia 248: 65-77. Naiman, R. J., and H. Décamps. 1997. The ecology of interfaces: riparian zones. Annual Revue of Ecology and Systematics 28: 621-658.

-74- Chapter 3

Naiman, R. J., H. Décamps, and M. E. McClain. 2005. Riparia: ecology, conservation, and management of streamside communities. Elsevier Academic Press, San Diego. U.S.A. pp. 430. Neatrour, M. A., J. R. Webster, and E. F. Benfield. 2004. The role of floods in particulate organic matter dynamics of a southern Appalachian river-floodplain ecosystem. Journal of the North American Benthological Society 23: 198-213. Petersen, R. C., and K. W. Cummins. 1974. Leaf processing in a woodland stream. Freshwater Biology 4: 343-368. Petersen, H., and M. Luxton. 1982. A comparative analysis of soil fauna population and their role in decomposition process. Oikos 39: 287-338. Pietikäinen, J., M. Pettersson, and E. Bååth. 2005. Comparison of temperature effects on soil respiration and bacterial and fungal growth rates. FEMS Microbiology Ecology 52: 49-58. Richter, B. D., J. V. Baumgartner, R. Wigington, and D. P. Braun. 1997. How much water does a river need? Freshwater Biology 37: 231-249. Risser, P. G. 1995. The status of the science examining ecotones. BioScience 45: 318-325. Robinson, C. T., and M. O. Gessner. 2000. Nutrient addition accelerates leaf breakdown in an alpine springbrook. Oecologia 122: 258-263. Robinson, C. T., and C. Jolidon. 2005. Leaf breakdown and the ecosystem functioning of alpine streams. Journal of the North American Benthological Society 24: 495-507. Rossi, L., and M. L.Constantini. 2000. Mapping the intra-habitat variation of leaf mass loss rate in brackish Mediterranean lake. Marine Eoclogy Progress Series 203: 145-159. Stanford, J. A., M. S. Lorang, and F. R. Hauer. 2006. The shifting habitat mosaic of river ecosystems. Verhandlungen Internationaler Vereinigung für Limnologie 29: 123-136. Strayer, D. L., H. A. Ewing, and S. Bigelow. 2003. What kind of spatial and temporal details are required in models of heterogeneous systems? Oikos 102: 654-662. Suberkropp, K. F. 1998. Microorganisms and organic matter decomposition. In: River Ecology and Management: Lessons from the Pacific Coastal Ecoregion, R. J. Naiman, and R. E. Bilby (eds.). Springer, New York, U.S.A. pp. 120-143. Swift, M. J., O. W. Heal, and J. M. Anderson. 1979. Decomposition in terrestrial ecosystems. Blackwell scientific publications, Oxford, U.K. pp. 71-73. Tachet, H., P. Richoux, M. Bournaud, and P. Usseglio-Polatera. 2002. Invértebrés d’eau douce: systématique, biologie, écologie. CNRS Éditions, Paris. pp. 588. Tiegs, S. D., and M. Pohl. 2005. Planform channel dynamics of the lower Colorado River: 1976-2000. Geomorphology 69: 14-27. Tiegs, S. D., S. D. Langhans, K. Tockner, and M. O. Gessner. 2007. Cotton strips as a leaf surrogate to measure decomposition in river floodplain habitats. Journal of the North American Benthological Society 26: 70-77. Tockner, K., F. Malard, and J. V. Ward. 2000. An extension of the flood pulse concept. Hydrological Processes 14: 2861-2883.

-75- Leaf decomposition in a floodplain mosaic

Tockner, K., and J. A. Stanford. 2002. Riverine flood plains: present state and future trends. Environmental Conservation 29: 308-330. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: A model ecosystem of European importance. Aquatic Sciences 65: 239-253. Tockner, K., S. E. Bunn, G. Quinn, R. Naiman, J. A. Stanford, and C. Gordon. In press. Floodplains: critically threatened ecosystems. In: Aquatic Ecosystems: Trends and Global Perspective, N.V.C. Polunin (ed.). Cambridge University Press, Cambridge, U.K. Turner, M. G., and F. S. Chapin III. 2005. Causes and consequences of spatial heterogeneity in ecosystem function. In: Ecosystem function in heterogeneous landscapes, Lovett, G. M., C. G. Jones, M. G. Turner, and K. C. Weathers (eds.), Springer Science and Business Media, Inc., New York, USA. pp. 9-30. Valett, H.M., M. A. Baker, and J. A. Morrice, C. S. Rawford, M. C. Molles, C. N. Dahm, D. L. Moyer, J. R. Thibault, and L. M. Ellis. 2005. Biogeochemical and metabolic responses to the flood pulse in a semiarid floodplain. Ecology 86: 220-234. van der Nat, D. 2002. Ecosystem processes in the dynamic Tagliamento River (NE-Italy). PhD dissertation. ETH Zurich. pp. 159. van der Nat, D., K. Tockner, P. J. Edwards, J. V. Ward, and A. M. Gurnell. 2003a. Habitat change in braided flood plains (Tagliamento, NE-Italy). Freshwater Biology 48: 1799- 1812. van der Nat, D., K. Tockner, P. J. Edwards, and J. V. Ward. 2003b. Large wood dynamics of complex Alpine river floodplains. Journal of the North American Benthological Society 22: 35-50. Ward, J. V. 1998. Riverine landscapes: Biodiversity patterns, disturbance regimes, and aquatic conservation. Biological Conversation 83: 269-278. Ward, J. V., K. Tockner, P. J. Edwards, J. Kollmann, G. Bretschko, A. M. Gurnell, P. E. Petts, and B. Rossaro. 1999. A reference system for the Alps: the “Fiume Tagliamento”. Regulated Rivers-Research & Management 15: 63-75. Ward, J. V., K. Tockner, D. B. Arscott, and C. Claret. 2002. Riverine landscape diversity. Freshwater Biology 47: 517-239. Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Revue of Ecology and Systematics 17: 567-594. Webster, J. R., E. F. Benfield, T. P. Ehrmann, M. A. Schaeffer, J. L. Tank, J. J. Hutchens, and D. J. D’Angelo. 1999. What happens to allochthonous material that falls into streams? A synthesis of new and published information from Coweeta. Freshwater Biology 41: 687- 705. Xiong, S., and C. Nilsson. 1997. Dynamics of leaf litter accumulation and its effects on riparian vegetation: a review. Botanical Review 63: 240-264.

-76-

CHAPTER 4

Cotton strips as a leaf surrogate to measure decomposition in river floodplain habitats

Simone D. Langhans*, Scott D. Tiegs*, Klement Tockner, and Mark O. Gessner 2007 Journal of the North American Benthological Society 26: 70-77 *equally contributed to the work

Leaf-litter assays have advanced our understanding of decomposition processes in both terrestrial and aquatic ecosystems. Some shortcomings inherent in the technique may be overcome through use of a cotton-strip assay. Key assumptions for using cotton strips as proxies for natural leaves are: 1) decomposition rates of the 2 materials are related, and 2) the materials decay in a similar way when exposed to the same environmental conditions. These assumptions were tested by comparing cotton-strip decomposition (loss of tensile strength and mass) and leaf decomposition (mass loss) across different floodplain habitats of the Tagliamento River (northeastern Italy). Patterns of loss of cotton-strip tensile strength and leaf mass were broadly comparable across river channels, ponds, and terrestrial sites. Differences between river channels and ponds were greater for loss of cotton-strip tensile strength than leaf mass, indicating that, in some situations, loss of cotton-strip tensile strength may be more sensitive to differing environmental conditions than loss of leaf mass. Loss of cotton-strip mass was less sensitive than loss of either tensile strength or leaf mass. Although combined data from all floodplain sites and additional sites in Swiss streams yielded a curvilinear relationship between loss of cotton-strip tensile strength and mass, the slope was extremely steep in the range of 20 to 30% mass loss (corresponding to 0 to 95% loss in tensile strength), indicating that inferring one variable from the other is unreliable. Leaf mass loss was significantly correlated with loss of tensile strength in fine- and coarsemesh bags in ponds and in coarse-mesh bags in terrestrial sites. However, these correlations were relatively weak (r = 0.50–0.63), suggesting that loss of tensile strength did not accurately reflect leaf mass loss. Thus, the

-77- Cotton strips cotton-strip assay should not be used uncritically as a surrogate for leaf-litter assays, but it has potential as a standardized method to measure organic-matter decomposition in fluvial settings and as a functional indicator for stream assessment.

Introduction Leaf decomposition is a vital process in forested stream and floodplain ecosystems (Cummins et al. 1989, Wallace et al. 1997). Current understanding of this process has been advanced mainly through use of the leaf-litter assay (e.g., Petersen and Cummins 1974, Gessner et al. 1999, Webster and Benfield 1986) in which leaf material is incubated in the field and later retrieved to determine leaf mass remaining and other response variables related to decomposition. The leaf-litter assay, adopted by stream ecologists from soil scientists (e.g., Bocock and Gilbert 1957), attempts to mimic decomposition of naturally occurring litter in streams (Cummins et al. 1980, Webster et al. 2001). However, the approach has some important shortcomings (Webster and Benfield 1986, Boulton and Boon 1991). Limitations of the leaf-litter assay can stem from variability in leaf quality (Wallace et al. 1996, Benfield et al. 2001). Pronounced differences in leaf quality result in a wide continuum of decomposition rates across species (Petersen and Cummins 1974, Webster and Benfield 1986, Gessner and Chauvet 1994). Individual studies can control for this variability by using one or several model species (e.g., Wallace et al. 1996, Robinson and Gessner 2000, Benfield et al. 2001, Lecerf et al. 2005); however, broader synthesis is hampered because different species are used across studies. Even within a given species, variation in leaf quality and decomposition rate can be significant; it can result from variation among genotypes (Leroy et al. 2006), differences in edaphic or meteorological factors (Austin 2002), or herbivore-induced plant defenses (Irons et al. 1991). Light gradients in the canopy may lead to variation in leaf quality even among leaves from individual trees (Sariyildiz and Anderson 2003). Regardless of its cause, variability among batches of leaves creates statistical noise in data sets, and it can call the validity of a study into question when care is not taken to ensure that a homogeneous batch of leaf material is used across treatments (e.g., at different sites or in different years). In addition, variability in leaf quality can mask an ecologically meaningful effect if within-species variability is large and the level of replication is constrained by practical

-78- Chapter 4 considerations. This problem may be particularly acute when leaf decomposition is used in bioassessment (Maltby and Booth 1991, Wallace et al. 1996, Gessner and Chauvet 2002, Lepori et al. 2005) where costs tend to limit the number of experimental units. Cotton strips, which have been used widely in soil science (Treonis et al. 2002, van Gestel et al. 2003), may offer a solution to the problems of within- and among-leaf species variability. However, their potential has not been fully explored in fluvial environments (but see Hildrew et al. 1984, Boulton and Quinn 2000, Claret et al. 2001). Cellulose, a major constituent of both cotton strips and leaf litter (Roberts and Rowland 1998), is a suitable substrate for leaf- colonizing fungi (Singh 1982) and bacteria (Rabinovich et al. 2002) and can serve as a food source for some leaf-shredding stream invertebrates (Sinsabaugh et al. 1985). Decay of cotton strips can be measured by determining mass loss (Egglishaw 1972), as is usually done for leaf litter, but more commonly, loss in cotton-strip tensile strength is used as the response variable (Boulton and Boon 1991). The standard quality of cotton strips, in contrast to natural leaves, could be especially useful in studies that compare different sites or detect change through time, including functional bioassessment of stream ecosystems. Cotton strips also offer other advantages. Loss of cotton-strip tensile strength tends to occur much faster than mass loss of leaf litter from typical riparian trees. As a result, incubation times in the field can be kept short, making cotton strips less susceptible to damage or loss by vandalism or flooding and especially useful in environments where leaves decompose slowly (Newman et al. 2001). Moreover, cotton strips are less prone to fragmentation than leaves (Egglishaw 1972), are smaller than the leaf packs and leaf bags usually used in decomposition experiments, more transportable, and free of nutrients such as N and P. This lack of nutrients may be particularly useful when the effects of dissolved nutrients on litter decomposition are of interest (e.g., Newman et al. 2001). Thus, cotton strips may be a good surrogate for assessing patterns of organic-matter decomposition in fluvial environments. An important consideration for using cotton strips as a surrogate for natural leaves is that a strong relationship exists between decay rates of leaves and cotton strips (Howard 1988), but the extent to which cotton-strip assays reflect decomposition of natural litter is not currently known. Therefore, the goal of our study was to ascertain whether patterns of cotton-strip decomposition mimic those of natural leaves. We compared leaf mass loss to losses of cotton-strip

-79- Cotton strips tensile strength and mass across a range of river floodplain habitats to address the questions: 1) Does cotton-strip decomposition match leaf-litter decomposition across differing environmental conditions? 2) Are losses of leaf mass and cotton-strip tensile strength correlated? 3) Can loss of cotton-strip tensile strength be predicted from loss of cotton-strip mass?

Methods The main portion of our study was conducted on the Tagliamento River, a 7th- order gravel-bed river that drains the Julian Alps of northeastern Italy. The experiment was initiated in December 2002 shortly after peak leaffall in the area. The island-braided reach of the river (river km 79.8-80.8, 135 m asl) selected for study has a variety of distinct habitats (Tockner et al. 2003, Langhans et al. 2006, Langhans and Tockner 2006; Fig 1). Sites were chosen in 7 habitat types (4 replicate sites in each habitat type, 28 sites total; described in Tables 1, 2). Habitat types were classified into 3 broader categories: river channels, ponds, and terrestrial habitats.

Table 1: Mean (±1 SE) physical and chemical characteristics of aquatic habitat types on the Tagliamento River floodplain during the experiment. Temperature was measured continuously at each site. Variable River channel Parafluvial ponds Orthofluvial ponds Water temperature (°C) 9.3 ± 0.5 8.0 ± 0.8 7.4± 0.8 Conductivity (µS/cm; at 20°C) 445 ± 56 451 ± 58 465 ± 70 pH 8.2 ± 0.1 8.1 ± 0.1 7.8 ± 0.0

O2 (mg/L) 12.4 ± 1.0 12.5 ± 2.1 11.5 ± 1.9

NH4-N (mg/L) <0.01 <0.01 0.01

NO3-N (µg/L) 656 ± 34 395 ± 52 580 ± 143 Soluble reactive P (µg/L) <5 <5 <5

-80- Chapter 4

N •I

D Vegetation D Water - - D Exposed Gravel 0--- 0.5km

Habitat Desc1iption Habitat catego1y 1 Channel Riffle habitats characte1ized by flowing water Channel 2 Parafluvial pond Isolated still-water habitats located on the active floodplain Pond 3 01thofluvial pond Isolated still-water habitats located in the 1iparian forest Pond 4 Island Forest floor of vegetated islands TeITestrial 5 Riparian forest Florest floor of ripa1ian zone TeITestrial 6 Large wood Accumulations of large wood located on the active floodplain TeITestrial 7 Exposed gravel Areas of non-vegetated gravel located on the active floodplain TeITestrial Figure 1: The braided-island reach of the Tagliamento River showing locations of sites. Numbers coITespond to habitat types, and the letters (a-d) coITespond to replicates within habitat types (n = 28 sites). The small dark features in the active channel indicate patches of large wood and vegetation, whereas those in the riparian forest indicate orthofluvial ponds.

Table 2: Mean (±1 SE) physical and chemical characteristics of te1Testrial habitat types on the Tagliamento River floodplain during the experiment Temperature was recorded continuously with at each site. Relative humidity was measured once at each site.

Variable Exposed gravel Large wood Island Riparian forest Mean air temperature (°C) 3.5 3.5 4.1 3.4 Temperature range (°C) -6.8-36.3 -6.8-36.3 -5-36.1 -3.7-26.6 Mean relative humidity (%) 72 95 97 96 Humidity range (%) 20-100 63-100 35-100 26-100

-81- Cotton strips

Coarse- and fine-mesh bags (mesh sizes = 10 mm and 0.5 mm, respectively) were used to allow or exclude macroinvertebrates and to facilitate estimation of decomposition attributable to invertebrates or microbes. Two types of substrate were placed in each leaf bag: 1) air-dried, recently senesced leaves (5.00 ± 0.25 g) from black poplar, Populus nigra L., the dominant tree species on the Tagliamento floodplain (Karrenberg et al. 2002), and 2) a single cotton strip. Cotton strips (4 cm × 6 cm) were cut from Shirley Soil Burial Test Fabric (Shirley Institute, Manchester, UK). This standardized fabric consists of 100% unbleached cotton (96% cellulose). Groups of 10 cotton strips were wrapped in aluminum foil and autoclaved (Howson 1988) for 20 min at 120°C before being placed in the mesh bags. Pairs of coarse- and fine-mesh bags were randomly assigned to the 28 sites. Five pairs of bags were fixed at each site with rope and an iron rod (280 mesh bags total). After 17, 31, 49, 61, or 80 d of incubation, 1 pair of bags was removed from each site, placed in a plastic bag, and transported to the laboratory. Each bag was emptied into a shallow plastic tray that contained several centimeters of tap water. Each side of the cotton strip and leaves was cleaned with a soft-bristled paint brush to remove adhering debris. Cleaned leaves were oven-dried to constant mass and weighed. Cleaned cotton strips were soaked in 70% ethanol to inhibit microbial decay during storage (Correll et al. 1997), then air dried, and stored individually in small plastic bags until tensile strength and mass were determined at a later time. A complementary study was conducted in 20 low-order streams in Switzerland with similar general characteristics. Streams were 2nd or 3rd order, had elevations ranging from 350–700 m asl, and drained calcareous watersheds. Fifteen streams drained fully forested watersheds and 5 flowed through pastures. Cotton strips were deployed in a manner similar to that used in the study on the Tagliamento floodplain. Cotton strips were retrieved from streams after 42–51 d and processed as described above. Tensile strength of all cotton strips was measured with an Instron Series IX Tensiometer (Instron Corporation, Canton, Ohio). Measurements were made at 20°C and 65% relative humidity in a climate-controlled room. Five randomly selected cotton strips were used to determine the mean and standard deviation of preincubation tensile strength (631 ± 17 kg) and mass (497.0 ± 4.3 mg). Strips were soaked in ethanol and air dried before measurements were made as a procedural control. Individual cotton strips were not weighed initially because of

-82- Chapter 4 very low variability among strip masses (coefficient of variation < 1%). Some strips were not included in analyses because they were too decomposed to determine tensile strength with confidence. Exponential decay coefficients (k) were calculated for leaves (mass loss) and cotton strips (tensile-strength loss) by fitting data to a simple exponential decay –kt model, Xt = X0e , where Xt is the cotton mass, leaf mass, or tensile strength upon removal of the litter bags from the field, X0 is the initial value, and t is the elapsed time in days. Pearson correlation coefficients were calculated between loss of cotton-strip tensile strength and leaf mass loss. Separate 2-way analyses of variance (ANOVA) were used to test for differences in leaf and cotton-strip decay coefficients between the 2 mesh types (fixed factor) and among the floodplain habitat types (fixed factor). When a significant habitat effect was detected, weighted contrasts were used to test for differences between pairs of the 3 broad habitat categories (i.e., river channels vs. ponds, ponds vs. terrestrial sites, and river channels vs. terrestrial sites). Contrasts were weighted by the number of habitat types in each habitat category. Paired 1-tailed t-tests were used to test the hypothesis that decay coefficients of cotton strips and leaves in individual habitat categories were greater in coarse mesh than fine-mesh bags. Data did not perfectly meet test assumptions of homoscedasticity and normality (Kolmogorov–Smirnov test), but in light of the robustness of ANOVA and t- tests to moderate violations of these assumptions (Box 1954), their use was deemed acceptable. Analyses were done using Systat 10 (Systat Software, Point Richmond, California) except for t-tests which were run in Microsoft Excel 2003.

Results Decay rates of cotton strips (tensile strength and mass) and leaves (mass) were faster in river channels than in ponds and terrestrial sites (Fig. 2A, B). Leaf mass loss was the only measure that differed significantly between ponds and terrestrial sites. Significant interactions between mesh size and habitat type were also observed for both types of cotton-strip decay rates, but not for leaf mass loss (p = 0.07). Both cotton-strip decay rates were significantly higher in coarse- mesh than fine-mesh bags in river channels, and leaf decay rates tended to be higher in coarse-mesh than in fine-mesh bags (p = 0.052; Fig 2A, B). None of the decay rates differed between mesh types in ponds (Fig 2A, B). In terrestrial habitats, rates of leaf and cotton-strip mass loss differed between mesh types,

-83- Cotton strips but rates of loss of cotton-strip tensile strength did not (Fig 2). Statistical detection of the very minor differences in mean decay rate between coarse- and fine-mesh bags in terrestrial habitats (Fig. 2A, B) were a consequence of extremely low within-habitat variability and large total sample size.

Figure 2: Mean (±1 SE) exponential decay rates (k) of Populus nigra leaves (mass loss) and cotton strips (loss of tensile strength and mass) in coarsemesh (A) and fine-mesh (B) bags in 3 habitat categories (4 river sites, 8 ponds, and 20 terrestrial [Terr] sites) on the Tagliamento floodplain. Horizontal lines over bars indicate no significant difference between habitat categories. Vertical black lines between bars indicate no significant difference between coarse- and fine-mesh bags within habitats. (*) = p = 0.052, * = p < 0.05, ** = 0.001 < p < 0.01, *** = p < 0.001.

Leaf mass loss and loss of cotton-strip tensile strength were not consistently correlated across habitat categories and mesh types except in pond habitats (Table 3). Even in pond habitats, significant correlation coefficients were low (ranging between 0.50 and 0.63) and strong curvilinear relationships were not detected either. This result indicates that loss of cotton-strip tensile strength captured only a modest amount of variability in leaf mass loss across floodplain habitats.

-84- Chapter 4

Table 3: Correlation coefficients (r), number of leaf bags (n) and significance levels (p) for comparisons of rates of loss of leaf mass and cotton-strip tensile strength in coarse- and fine- mesh bags across habitat categories and mesh types. Some cotton strips were not included because they were too decomposed to permit confidently determine tensile strength. Habitat category Coarse mesh Fine mesh r n p r n p River 0.45 14 0.099 0.48 14 0.068 Pond 0.50 24 0.012 0.63 24 <0.001 Terrestrial 0.63 69 <0.001 0.07 69 0.60

Combined data from the habitats on the Tagliamento floodplain and the Swiss streams showed that the range of losses of cotton-strip tensile strength (0- 99%) corresponded to a much narrower range of losses of cotton-strip mass (most values between 20-60%; Fig. 3). Loss of cotton-strip tensile strength at Swiss sites was rapid, and as a result, most data points clustered in the upper portion of the graph. Overall, a curvilinear relationship emerged that was characterized by an extremely steep increase in tensile-strength loss between ~20% and 30% mass loss and a nearly 100% loss of tensile strength when only ~45% of mass had been lost.

100

80

60

River Pond 40 Terrestrial Figure 3: Relationship between loss of tensile + Swiss streams strength and mass of 20 Cotton Tensile Strength Loss (%) cotton strips across all sites from the 0 Tagliamento floodplain 0 20406080100and 20 low-order streams Cotton Mass Loss (%) in Switzerland.

-85- Cotton strips

Discussion Variation in cotton-strip decay broadly matched the variation in leaf decay across different habitat categories on the Tagliamento floodplain. Decomposition in the river channel was significantly faster than decomposition in ponds or terrestrial sites, regardless of whether cottonstrip tensile strength, cotton-strip mass, or leaf mass were used as an indicator variable. This broad match of patterns across habitats is consistent with the findings of Baldy et al. (2002), who observed faster decay of P. nigra leaves in the main river channel compared to a floodplain pond (oxbow lake) in southwestern France, and of Gurtz and Tate (1988), who measured faster decay of hackberry leaves in a stream channel relative to a nearby terrestrial site in Kansas, USA. In coarsemesh bags, the ratio between rates of decay in river channels and ponds was >6× higher for cotton strips (measured as loss of tensile strength) than for P. nigra leaves (measured as mass loss) in our study and 12× higher than for P. nigra leaves in the study by Baldy et al. (2002). However, cotton-strip mass loss was less sensitive than leaf mass loss to differences between habitats, with the ratio of decay rates in channels and ponds >4× greater for cotton-strip tensile strength than for leaf mass. Thus, loss of tensile strength, but not mass, of cotton strips can be a more sensitive measure of differences in environmental conditions than leaf mass loss. These results suggest that, in some instances, cotton strips can be an appropriate and sensitive surrogate for natural leaf material in decomposition experiments done in fluvial environments. A consideration for using cotton strips as a surrogate for natural leaf material in decomposition studies is that decomposition rates of both materials are related. In several instances loss of cotton-strip tensile strength and leaf mass were correlated in the present study. However, loss of cotton-strip tensile strength failed to predict loss of leaf mass precisely, suggesting that cotton strips and leaves capture different aspects of organic-matter decomposition. Cotton strips were typically in a more advanced state of decay than the leaf material when our experiment was terminated. Because leaf species vary widely in their decay rates (Petersen and Cummins 1974, Webster and Benfield 1986, Gessner and Chauvet 1994), strength of the relationship between loss in leaf mass and cotton tensile strength could hinge on the particular leaf species used, and correlations with cotton-strip decay might have been stronger had a faster- decomposing species been used. However, even if stronger relationships exist in other circumstances, loss of cotton-strip tensile strength should not be equated

-86- Chapter 4 indiscriminately with leaf mass loss, but rather should be used as an indicator of cellulose decomposition in its own right (Hildrew et al. 1984, Boulton and Quinn 2000). Leaf-shredding invertebrate taxa were rare or absent in a survey across the aquatic habitats within our study reach (Arscott et al. 2005). This fact, the modest or nonsignificant differences in rates of decay of leaves and cotton strips in coarse- and fine-mesh bags, and the low abundance of macroinvertebrates in litter bags in all habitats except the river channel (SDL, SDT, MOG, and KT, unpublished data) suggest that shredding by invertebrates was not a major cause of decay in most habitats on the Tagliamento floodplain. In other situations, invertebrates can be important agents of litter decompostion (Webster and Benfield 1986). For example, leaf decay rates in fine-mesh bags were ~½ the rates in coarse-mesh bags in a polluted river (Pascoal et al. 2005), and in a more natural mountain stream, shredders were estimated to contribute to >½ of the overall leaf mass loss (Hieber and Gessner 2002). Likewise, Cuffney et al. (1990) observed up to 74% reduction in decomposition rate when shredder abundance was experimentally reduced with insecticide in a small headwater stream. The palatability of cotton strips and extent to which stream invertebrates feed on them is not well known, but both may depend on the degree to which cotton strips are conditioned by microorganisms, as is the case with leaf litter in streams (Suberkropp 1992, Graça 2002). The short time that cotton strips typically reside in the field may limit accrual of micobial biomass even when microbial activity is pronounced. Thus, cotton strips could prove most useful as an integrator of decomposition through microbial activity rather than through invertebrate shredding. The idea that cotton strips are more appropriate for determining microbial than invertebrate activity is corroborated by a terrestrial study in which loss of cotton-strip tensile strength was not correlated with invertebrate density (van Gestel et al. 2003). Cotton strips consist almost entirely of cellulose, whereas leaves contain a complex of structural and other compounds whose full degradation requires a suite of enzymes. As a consequence, different mechanisms are likely to be involved in the decomposition of cotton and leaves. For example, leaf decomposition in streams is mediated to a significant extent by pectinases (Jenkins and Suberkropp 1995), which degrade the middle lamellae of plant tissues and result in leaf disintegration, a process that would not occur with

-87- Cotton strips cotton strips. Microbial leaf decomposition in streams is largely caused by leaf- degrading fungi (aquatic hyphomycetes). The degree to which aquatic hyphomycetes colonize and degrade cotton is not currently known, but these fungi are capable of degrading cellulose (Singh 1982), suggesting that they may colonize and degrade cotton strips as well. One drawback of the cotton-strip assay is that a tensiometer, an instrument required to measure tensile strength, is not readily available to many researchers. Therefore, measuring cotton-strip mass loss instead of loss of tensile strength could be useful, provided that loss of mass is correlated with loss of tensile strength. Our study and others in streams (Egglishaw 1972) and soils (Latter and Howson 1977) indicate that the relationship between loss of cotton-strip mass and tensile strength is curvilinear, although the specific relationship may depend on the particular conditions of the study (e.g., soil, stream, experimental procedures; Latter and Howson 1977). In our study, the fact that the steep slope of the relationship in the narrow range of 20 to 30% mass loss corresponds to 0 to 95% loss in tensile strength is a matter of concern. Thus, sensitivity of the cotton-strip assay may be greatly diminished if it is based on mass loss rather than loss of tensile strength, as has been observed elsewhere (A. J. Boulton, University of New England, Armisdale, Australia, personal communication). Last, cotton strips could be useful in comparative studies, especially in stream bioassessment where a need exists to include functional criteria as a complement to traditional structural criteria (Wallace et al. 1996, Boulton 1999, Gessner and Chauvet 2002). The cotton-strip assay has been shown to be sensitive to human activities (Chew et al. 2001, Rapp et al. 2001) and to different environmental conditions (our study). It has been advocated as a means to assess soil health (Trettin et al. 1996, Chew et al. 2001, Rapp et al. 2001), and it was also useful to assess effects of siltation in a New Zealand stream (Boulton and Quinn 2000). Thus, although the cotton-strip assay should not be used uncritically as a surrogate for natural leaf decay, it has potential as a standard method for measuring some aspects of organic-matter decomposition in fluvial environments, and it may prove useful for assessing the health of river floodplain systems.

Acknowledgements We thank Simone Blaser for her help in the field and for measuring tensile strength of cotton strips. Pamela Woods provided helpful editing on an earlier

-88- Chapter 4 version of this manuscript. EMPA St. Gallen kindly provided access to the tensiometer for measuring cotton-strip tensile strength. Two anonymous referees, A. J. Boulton, and P. Silver provided thorough reviews or editing that improved the manuscript. This research was supported through the RivFunction and TempQsim projects funded by the European Union Commission (contracts EVK1-CT-2001-00088 and EVK1-CT- 2002-0012) and the Swiss State Secretariat for Education and Research (SBF no. 01.0087 and 02.0072).

References Arscott, D. B., K. Tockner, and J. V. Ward. 2005. Lateral organization of aquatic invertebrates along the corridor of a braided floodplain river. Journal of the North American Benthological Society 24: 934-954. Austin, A. T. 2002. Differential effects of precipitation on production and decomposition along a rainfall gradient in Hawaii. Ecology 83: 328-338. Baldy, V., E. Chauvet, J.-Y. Charcosset, and M. O. Gessner. 2002. Microbial dynamics associated with leaves decomposing in the mainstem and floodplain pond of a large river. Aquatic Microbial Ecology 28: 25-36. Benfield, E. F., J. R. Webster, J. L. Tank, and J. J. Hutchens. 2001. Long-term patterns in leaf breakdown in streams in response to watershed logging. International Review of Hydrobiology 86: 467-474. Bocock, K. L., and O. J. W. Gilbert. 1957. The disappearance of leaf litter under different woodland conditions. Plant Soil 9: 179-185. Boulton, A. J. 1999. An overview of river health assessment: philosophies, practice, problems and prognosis. Freshwater Biology 41: 469-479. Boulton, A. J., and P. I. Boon. 1991. A review of methodology used to measure leaf litter decomposition in lotic environments: time to turn over an old leaf? Australian Journal of Marine and Freshwater Research 42: 1-43. Boulton, A. J., and J. M. Quinn. 2000. A simple and versatile technique for assessing cellulose decomposition potential in floodplain and riverine sediments. Archiv für Hydrobiologie 150: 133-151. Box, G. E. P. 1954. Some theorems on quadratic forms applied in the study of analysis of variance problems. Annals of Statistics 25: 290-302. Chew, I., J. P. Obbard, and R. R. Stanforth. 2001. Microbial cellulose decomposition in soils from a rifle range contaminated with heavy metals. Environmental Pollution 111: 367- 375. Claret, C., A. J. Boulton, M.-J. Dole-Olivier, and P. Marmonier. 2001. Functional processes versus state variables: interstitial organic matter pathways in floodplain habitats. Canadian Journal of Fisheries and Aquatic Sciences 58: 1594-1602.

-89- Cotton strips

Correll, R. L., B. D. Harch, C. A. Kirkby, K. O'Brien, and C. E. Pankhurst. 1997. Statistical analysis of reduction in tensile strength of cotton strips as a measure of soil microbial activity. Journal of Microbiological Methods 31: 9-17. Cuffney, T. F., J. B. Wallace, and G. J. Lugthart. 1990. Experimental evidence quantifying the role of benthic invertebrates in organic-matter dynamics of headwater streams. Freshwater Biology 23: 281-299. Cummins, K. W., G. L. Spengler, G. M. Ward, R. M. Speaker, R. W. Ovink, D. C. Mahan, and R. L. Mattingly. 1980. Processing of confined and naturally entrained leaf litter in a woodland stream ecosystem. Limnology and Oceanography 25: 952-957. Cummins, K. W., M. A. Wilzbach, D. M. Gates, J. B. Perry, and W. B. Taliaferro. 1989. Shredders and riparian vegetation. BioScience 39: 24-30. Egglishaw, H. J. 1972. An experimental study of the breakdown of cellulose in fast-flowing streams. Memorie dell'Istituto Italiano di Idrobiologia 29: 405-428. Gessner, M. O., and E. Chauvet. 1994. Importance of stream microfungi in controlling breakdown rates of leaf-litter. Ecology 75: 1807-1817. Gessner, M. O., and E. Chauvet. 2002. A case for using litter breakdown to assess functional stream integrity. Ecological Applications 12: 498-510. Gessner, M. O., E. Chauvet, and M. Dobson. 1999. A perspective on leaf litter breakdown in streams. Oikos 85: 377-384. Graça M. A. S. 2002. The role of invertebrates on leaf litter decomposition in streams - a review. International Review of Hydrobiology 86: 383-393. Gurtz, M. E., and C. M. Tate. 1988. Hydrologic influences on leaf decomposition in a channel and adjacent bank of a gallery forest stream. American Midland Naturalist 120: 11-21. Hieber, M., and M. O. Gessner. 2002. Contribution of stream detrivores, fungi, and bacteria to leaf breakdown based on biomass estimates. Ecology 83: 1026-1038. Hildrew, A. G., C. R. Townsend, J. Francis, and K. Finch. 1984. Cellulolytic decomposition in streams of contrasting pH and its relationship with invertebrate community structure. Freshwater Biology 14: 323-328. Howard, P. J. A. 1988. A critical evaluation of the cotton strip assay. In: Cotton Strip Assay: An Index of Decomposition in Soils, A. F. Harrison, P. M. Latter, and D. W. H. Walton (eds.). Institute of Terrestrial Ecology, Cumbria, UK. pp. 34-32. Howson, G. 1988. Current method for preparation, insertion and processing of cotton strips. In: Cotton Strip Assay: An Index of Decomposition in Soils Pages, A. F. Harrison, P. F. Latter, and D. W. H. Walton (eds.)., Institute of Terrestrial Ecology, Cumbria, UK. pp. 166-171. Irons, J. G., J. P. Bryant, and M. W. Oswood. 1991. Effects of moose browsing on decomposition rates of birch leaf litter in a sub-Arctic stream. Canadian Journal of Fisheries and Aquatic Sciences 48: 442-444. Jenkins, C. C., and K. Suberkropp. 1995. The influence of water chemistry on the enzymatic degradation of leaves in streams. Freshwater Biology 33: 245-253.

-90- Chapter 4

Karrenberg, S., P. J. Edwards, and J. Kollmann. 2002. The life history of Salicaceae living in the active zone of floodplains. Freshwater Biology 47: 733-748. Langhans, S. D., S. D. Tiegs, U. Uehlinger, and K. Tockner. 2006. Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems. Polish Journal of Ecology. In press. Langhans, S. D. and K. Tockner. 2006. The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy). Oecologia 147: 501-509. Latter, P. M., and G. Howson. 1977. Use of cotton strips to indicate cellulose decomposition in the field. Pedobiologia 17: 145-155. Lecerf, A., M. Dobson, C. K. Dang, and E. Chauvet. 2005. Riparian plant species loss alters trophic dynamics in detritus-based stream ecosystems. Oecologia 146: 432-442. Lepori, F., D. Palm, and B. Malmqvist. 2005. Effects of stream restoration on ecosystem functioning: detritus retentiveness and decomposition. Journal of Applied Ecology 42: 228-238. Leroy, C. J., T. G. Whitham, P. Keim, and J. C. Marks. 2006. Plant genes link forests and streams. Ecology 87: 255-261. Maltby, L., and R. Booth. 1991. The effect of coal-mine effluent on fungal assemblages and leaf breakdown. Water Research 25: 247-250. Newman, S., H. Kumpf, J. A. Laing, and W. C. Kennedy. 2001. Decomposition responses to phosphorus enrichment in an Everglades (USA) slough. Biogeochemistry 54: 229-250. Pascoal, C., F. Cassio, and L. Marvanova. 2005. Anthropogenic stress may affect aquatic hyphomycete diversity more than leaf decomposition in a low-order stream. Archiv für Hydrobiologie 162: 481-496. Petersen, R. C., and K. W. Cummins. 1974. Leaf processing in a woodland stream. Freshwater Biology 4: 343-368. Rabinovich, M. L., M. S. Melnik, and A. V. Boloboba. 2002. Microbial cellulases. Applied Biochemistry and Microbiology 38: 305-321. Rapp, J., T. Shear, and D. Robison. 2001. Soil, groundwater, and floristics of a southeastern United States blackwater swamp 8 years after clearcutting with helicopter and skidder extraction of the timber. Forest Ecology and Management 149: 241-252. Roberts, J. D., and A. P. Rowland. 1998. Cellulose fractionation in decomposition studies using detergent fibre pre-treatment methods. Communications in Soil Science and Plant Analysis 29: 2109-2118. Robinson, C. T., and M. O. Gessner. 2000. Nutrient addition accelerates leaf breakdown in an alpine springbrook. Oecologia 122: 258-263. Sariyildiz, T., and J. M. Anderson. 2003. Decomposition of sun and shade leaves from three deciduous tree species, as affected by their chemical composition. Biology and Fertility of Soils 37: 137-146.

-91- Cotton strips

Singh, N. 1982. Cellulose decomposition by some tropical aquatic hyphomycetes. Transactions of the British Mycological Society 79: 560-561. Sinsabaugh, R. L., A. E. Linkins, and E. F. Benfield. 1985. Cellulose digestion and assimilation by 3 leaf-shredding aquatic insects. Ecology 66: 1464-1471. Suberkropp, K. 1992. Interactions with invertebrates. In: The Ecology of Aquatic Hyphomycetes, F. Bärlocher (ed.)., Spinger-Verlag, Berlin, Germany. pp. 118-134. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: a model ecosystem of European importance. Aquatic Sciences 65: 239-253. Treonis, A. M., D. H. Wall, and R. A. Virginia. 2002. Field and microcosm studies of decomposition and soil biota in a cold desert soil. Ecosystems 5: 159-170. Trettin, C. C., M. Davidian, M. F. Jurgensen, and R. Lea. 1996. Organic matter decomposition following harvesting and site preparation of a forested wetland. Soil Science Society of America Journal 60: 1994-2003. van Gestel, C. A. M., M. Kruidenier, and M. P. Berg. 2003. Suitability of wheat straw decomposition, cotton strip degradation and bait-lamina feeding tests to determine soil invertebrate activity. Biology and Fertility of Soils 37: 386-386. Wallace, J. B., S. L. Eggert, J. L. Meyer, and J. R. Webster. 1997. Multiple trophic levels of a forest stream linked to terrestrial litter inputs. Science 277: 102-104. Wallace, J. B., J. W. Grubaugh, and M. R. Whiles. 1996. Biotic indices and stream ecosystem processes: Results from an experimental study. Ecological Applications 6: 140-151. Webster, J. R., and E. F. Benfield. 1986. Vascular plant breakdown in freshwater ecosystems. Annual Review of Ecology and Systematics 17: 567-594. Webster, J. R., E. F. Benfield, J. J. Hutchens, J. L. Tank, S. W. Golladay, and J. C. Adams. 2001. Do leaf breakdown rates actually measure leaf disappearance from streams? International Review of Hydrobiology 86: 417-427.

-92-

CHAPTER 5

Aerial input, lateral transport, and surface storage of coarse particulate organic matter in a riverine floodplain mosaic (Tagliamento, NE-Italy)

Simone D. Langhans, Janine Rueegg, Urs Uehlinger, and Klement Tockner In preparation

Quantifying temporal and spatial dynamics of organic matter is a critical step in understanding river-floodplain ecosystem functioning. We studied aerial input, lateral transport, and standing stock of coarse particulate organic matter (CPOM) over an annual cycle in a pristine braided river-floodplain mosaic (Tagliamento, NE-Italy). Annual CPOM input (area-weighted mean) to the entire floodplain was 126.7 g AFDM m-2 y-1 ranging from 7.4 g along shorelines to 446.6g on vegetated islands. Mean annual lateral transport (AFDM m-1 y-1) was 19.7 g and varied from 12.7 g (islands) to 41.3 g (riparian forest). Direct litter fall dominated the input close to the vegetation, while lateral transport was the major pathway in the open tract. Vegetated islands contributed more than 95% to the annual direct input and 65% to lateral transport in the open tract. Total CPOM input (AFDM m-2 y-1) into the channel network ranged from 1.4g (lateral input) to 7.4 g (aerial input). Vegetated islands and the riparian forest were key CPOM storage areas accounting for 97% of total floodplain storage. Within the investigated floodplain, stored CPOM is mainly composed of wood, while miscellaneous material and leaves dominate aerial input and lateral transfer across the open tract. This study identifies aerial input as the major CPOM input pathway, emphasises the role of vegetated islands as major sources of CPOM for the active floodplain tract, and highlights the influence of wind on the distribution and transport of organic matter in riverine floodplains.

-93- CPOM dynamics in a floodplain mosaic

Introduction Organic matter produced in terrestrial habitats is an important energy source in rivers and streams, and forms together with in-stream primary production the energy supporting heterotrophic organisms (Minshall 1978, Neatrour et al. 2004). The relative importance of allochthonous versus autochthonous energy varies significantly along the river continuum and across geographical regions (Vannote et al. 1980). Small forested headwater streams in temperate regions are often heterotrophic depending on CPOM as their primary energy base (Fisher and Likens 1973, Cummins 1974). For forested mid-order reaches, a shift from heterotrophy to autotrophy is predicted (Vannote et al. 1980). In headwater streams pathways of allochthonous organic matter are well studied, including direct aerial input and lateral transport (Fisher and Likens 1973, Benfield 1997, Pozo et al. 1997), lateral transport of deposited organic matter by wind (Teeri and Barrett 1975, Benson and Pearson 1993), input of organic matter by bank erosion (Zah 2001), and import and export of organic matter by hydrological processes (Dance et al. 1979, Maamri et al. 1994). Riverine floodplains are defined as the entire channel network and valley- bottom area that are exposed to flooding (Stanford et al. 2005). Although floodplains are prominent features along river corridors, little is known about their functioning as ecosystems (Thoms et al. 2005, Valett et al. 2005, Malard et al. 2006). Along corridors where extensive river-floodplain interactions occur, productivity of floodplain vegetation and processing of organic matter within the ecosystem can alter the longitudinal patterns predicted by the river continuum concept (Junk et al. 1989). Thorp and Delong (1994) and Hedges et al. (1994) suggested that the volume of organic matter entering lowland rivers from the floodplain and riparian zone, combined with some local autochthonous production, is the principal source of fine and coarse particulate organic matter driving the food web in large rivers. The semiterrestrial plain is strongly connected to the permanent channel through periodic and often predictable flooding. Lateral exchange of CPOM might be extensive (Junk et al. 1989, Ward and Stanford 1995, Neatrour 2004), and the quality and quantity of organic matter entering the aquatic system can significantly be altered by truncated connectivity. Floodplains can serve as sources or sinks for CPOM depending on the flow regime, degree of connectivity, floodplain topography, and sediments loads among others (Cuffney 1988). Floodplains are known as important storage areas

-94- Chapter 5

(Merritt and Lawson 1979), and organic matter may be largely pre-processed on the floodplain and the resulting particles are entrained by the channel during high flows (Smock 1990). Overall, it remains a major challenge to determining CPOM dynamics in riverine floodplains characterized by multiple channel networks and complex vegetation structures. Therefore, only few studies that quantified organic matter dynamics of large river-floodplain systems were carried out (e.g., Cuffney 1988, Webster et al. 1995, Meyer et al. 1997, Zah 2001). Hydrological connectivity, including the fluvial exchange of matter and organisms, is the key component for river-floodplain functioning (Ward et al. 1999, Pringle 2003). Hence, reduced connectivity induces severe changes to ecosystem dynamics and processes. In disconnected floodplains, organic matter and nutrients produced or released in the floodplain are either recycled in situ or are rapidly transported downstream during extreme floods (Tockner et al. 1999, Thoms et al. 2005). Therefore, the biogeochemical cycling in most temperate river-floodplain systems has been greatly modified as a consequence of river regulation and floodplain reclamation (Gergel et al. 2005). Information derived from studies of near natural river-floodplain systems helps to understand how lateral connectivity and habitat complexity may control the main transfer and transformation pathways of organic matter and nutrients. The main objectives of the study were to quantify spatial and temporal dynamics of aerial input, lateral transport, and standing stock of CPOM in a large braided river-floodplain mosaic (Tagliamento River), and to upscale CPOM dynamics to the entire floodplain area. We hypothesize i) that aerial input is the major input pathway, ii) that vegetated islands are key sources of CPOM for the open tract of the floodplain, iii) that wind plays a substantial role for the redistribution of CPOM within the open tract, and iv) that the riparian forest and vegetated islands are predominant storage areas for organic matter.

Material and methods Site description The study was carried out from February to December 2005 in the island- braided section of the Tagliamento River, a seventh-order gravel-bed river in NE Italy (46°N, 12°30’E) with a total catchment area of 2580 km2. The river flows 172 km unimpeded from the Carnian Alps to the Adriatic Sea and connects two biomes, the Alps with the Mediterranean (Ward et al. 1999, Tockner et al. 2003).

-95- CPOM dynamics in a floodplain mosaic

The Tagliamento River escaped severe river engineering for almost its entire length. Therefore, it provides the opportunity to study ecosystem processes under natural conditions and at different spatial scales. The Tagliamento is characterized by a flashy flow regime with bankfull floods (water levels exceeding 3 m) in spring (Alpine snowmelt) and in autumn (heavy rainstorms). However, in 2005 the maximum water level was 1.40 m (flow pulse in October) that inundated 62% of the active tract (Fig. 1).

Figure 1: Water level (m; measured at Villuzza) at the study site during the study period.

Our study area (135 m a.s.l.) was a 2 km long and 1.1 km wide reach in the island-braided section (Fig. 2). This reach is characterized by a complex mosaic of aquatic, semi-aquatic, and terrestrial habitats (Gurnell 2001, Langhans et al. 2006). At base flow (about 40 m3s-1), the active tract contains exposed gravel sediments (89.6 ha, 47% of the total area), vegetated islands (14.9 ha, 8%), large wood accumulations (0.6 ha, 0.3%), the channel network (30.3 ha, 15.9%), and numerous ponds (0.8 ha, 0.5%). Shorelines, if considered as a 1m-strip along the channel network, account for 1.1% (2.0 ha) of the total area. The river is fringed by the riparian forest on the right bank (50.8 ha, 27%) and by hillslope forest of the Monte di Ragogna on the left bank (Fig. 2). The dominant riparian tree species are Populus nigra and five species of Salix (Karrenberg et al. 2003). Detailed information on the catchment and on the study reach has been provided by Ward et al. (1999) and Tockner et al. (2003).

-96- Chapter 5

Figure 2: The island-braided reach of the Tagliamento River. Three transect types were established to investigate direct input and lateral transport: transects perpendicular to the long axes of islands (ip), transects parallel to the main axis of islands (il), and transects perpendicular to the border of the riparian forest (f). Sampling sites for standing stock are marked with circles. Small letters indicate habitats (f = riparian forest, s = shoreline, i = island, g = exposed gravel surface), numbers indicate replicates (1-12). Flow is from right to left.

Environmental factors Air temperature, relative humidity, rainfall, wind speed, and wind direction, were continuously measured during the experiment (Tab. 1). Rainfall, wind speed, and wind direction were recorded using a HOBO micro weather station (onset, Pocasset, MA, USA). Relative humidity and air temperature were measured two meters above ground with a HOBO Pro RH/Temp data logger (onset, Pocasset, MA, USA) and a Vemco Minilog data logger (MINILOG12- TR-40/+50-064K, Vemco, Nova Scotia, Canada).

Sampling design We quantified direct (i.e. aerial) input of CPOM (coarse particulate organic matter; > 1.0 mm) with 96 traps. Traps consisted of white buckets (diameter = 0.25 m, height = 0.3 m, area = 0.049 m2) placed on the floodplain surface and fixed by iron sticks hammered into the ground. Lateral transport of CPOM was collected with 96 lateral traps which were designed to simultaneously collect material from the two opposite directions (Fig. 3).

-97- CPOM dynamics in a floodplain mosaic

Table 1: Physical characteristics of the Tagliamento floodplain during the experiment in 2005 (mean ± 1 SD; continuous records). Relative humidity data were collected from the meteorological station at Fagagna (~10 km southeast from the study area). Meteorological wind direction is defined as the geographic compass direction (0° = North, 90° = East, 180° = South, 270° = West) wind is coming from. n.r. = not recorded. Temperature Relative humidity Rainfall Wind speed Wind direction (°C) (%) (mm month-1) (m s-1) (ø) February 2.3 ± 1.4 44.0 ± 17.6 11.9 0.42 ± 0.38 n.r. March 7.4 ± 3.7 59.7 ± 15.9 31.3 0.38 ± 0.37 n.r. April 12.2 ± 6.3 67.9 ± 66.4 208.6 0.99 ± 1.02 166.6 ± 108.0 May 18.7 ± 7.6 62.3 ± 8.9 82.8 0.63 ± 0.75 182.0 ± 102.1 June 22.4 ± 7.4 64.5 ± 11.7 135.8 0.65 ± 0.67 182.6 ± 101.5 July 23.8 ± 6.9 68.9 ± 6.2 169.4 0.60 ± 0.67 179.1 ± 103.3 August 21.2 ± 6.5 73.6 ± 9.3 202.6 0.46 ± 0.56 174.5 ± 105.0 September 19.4 ± 6.8 73.6 ± 71.3 253.8 0.42 ± 0.55 166.5 ± 106.7 October 13.8 ± 1.2 76.1 ± 12.2 1.8 0.22 ± 0.41 162.0 ± 101.1 November 6.6 ± 6.5 71.0 ± 68.3 84.0 0.43 ± 0.79 168.7 ± 122.4 December 3.1 ± 1.7 74.0 ± 21.2 84.4 0.50 ± 0.28 n.r.

Figure 3: Lateral CPOM transport traps: A) Side view and B) top view. White buckets (25 cm diameter), separated in two equal parts by a flat piece of wood to distinguish between lateral transport from the vegetation and from the open tract, respectively. Pieces of wood (30 cm x 10 cm) on each side of the bucket prevented input from other directions. B) Buckets were dug into the ground with the rim plain to the sediment surface, the top covered to avoid aerial input. Metal grids were fit in the opening of the buckets to exclude amphibians and mammals but to trap organic matter all the same. Traps contained four holes at the bottom of the bucket, covered with 1.0 mm mesh, to avoid slack flow.

-98- Chapter 5

Pairs of traps (one direct and one lateral trap; the distance between the two traps was 5 m) were deployed along 70 m long transects from the riparian forest and from vegetated islands into the open floodplain tract (Fig. 2). The first trap pair for both the parallel and the perpendicular transects was located inside the forest or islands, 5 m from the vegetation edge. The remaining pairs were positioned in the open tract at 1, 5, 10, 20, 35 and 70 m distance from the vegetation edge. Of the 12 transects established, 4 extended perpendicularly from the riparian forest edge into the active channel area, 4 from the islands parallel to the longitudinal island axis, and 4 from the islands perpendicular to the longitudinal island axis (Fig. 2). The exact trap locations were determined with a differential global positioning system (dGPS; TCS1, Trimble, Sunnyvale, California, USA). The traps deployed in February 2005 were sampled in monthly intervals from March to October and bi-weekly in November and December 2005. The trapped organic matter was collected with a brush and a small shovel, and stored in polyethylene bags -20°C until processing. To determine stored surface CPOM, we placed a 0.5 x 0.5 m metal frame on the ground and picked the CPOM with forceps. We sampled four different habitats namely vegetated islands, riparian forest, exposed gravel, and shorelines (1 m wide strip along the water/land boundary). Each habitat was sampled at 12 sites, which were selected using a GIS-based random generator. Sampling took place in April, May, September, October, and November 2005. Each date the sampling frame was placed at a slightly different location around the pre- selected sampling point. The samples were stored at -20 °C until processing.

Sample processing CPOM from input and storage samples were thawed, rinsed with tap water to remove sand and adhering particles smaller than 1mm, and separated into four fractions: (1) leaves, (2) wood which included twigs and bark, (3) grass, and (4) miscellaneous such as catkins, fruits, seeds, flowers. The fractions were dried to constant weight at 60°C for 48 h, weighted, ashed (500°C for 6 h), and reweighed to calculate ash-free dry mass (AFDM).

Data analysis Direct input was expressed as g AFDM m-2 year-1 and lateral transport as g AFDM m-1 year-1 (Zah 2001). Values for January and February 2005 were calculated by assuming that input and transport data of December 2004 equaled

-99- CPOM dynamics in a floodplain mosaic those of December 2005, and that input and transport decreased linearly from December until March. For October and November, means of direct input and lateral transport from the two sampling dates were used for further analyses. Differences among means of direct input along the three different transect types were analyzed using repeated measures ANOVA with transect type (three levels), and CPOM fraction (four levels) as independent variables and month as repeated measures factor. Lateral transport data were analyzed in the same way with the direction of the opening (two levels) as an additional dependent variable. Results from ANOVA models were compared with mixed models, which accounted for missing values (Krueger 2004), with autoregressive variance-covariance structures for direct input and compound symmetry structures for lateral transport. As no differences were found between the two models, only results from the repeated measures ANOVA are discussed. We used repeated measures ANOVA to test for differences among means of CPOM storage (expressed as g AFDM m-2) over time, with habitat type (four levels), and fraction (four levels) as independent variables and month (six levels) as repeated measures factor. Input, transport, and storage data were log(x+1)-transformed before ANOVAs were performed. Although data did not perfectly meet assumptions of homoscedasticity and normality (Kolmogorov-Smirnov test), ANOVA deemed to be an acceptable method due to its robustness to moderate violations of test assumptions (Box 1954). The relationship between wind speed (expressed in m s-1) and wind direction (expressed in degrees of meteorological wind direction) with direct input and lateral transport of CPOM was determined by non-linear regression and curve estimation. Parameters of exponential functions, which described patterns of CPOM input (g AFDM m-2 y-1) or lateral transport (g AFDM m-1 y-1) with distance from the vegetation edge of islands and the riparian forest, were determined using non-linear regression. All statistical analyses were performed using SPSS (vers. 14.0/SPSS Inc., IL, USA) with significance levels set at P ≤ 0.05.

CPOM modeling at the ecosystem (floodplain) level In February 2005, the entire study reach was mapped in detail with a differential global positioning system (dGPS; TCS1, Trimble, Sunnyvale, California, USA),

-100- Chapter 5 and different habitats were vectorised as polygon coverage with ArcGis (vers. 9.1, Environmental Systems Research Institute (ESRI), Redlands, California, USA). We identified parameters of exponential functions using non-linear regression to describe annual rates of CPOM input (g AFDM m-2 y-1) or lateral transport (g AFDM m-1 y-1) as function of distance from the riparian forest or vegetated islands. Parameters of the exponential models were estimated separately for the riparian forest and for islands. Two different cost distance layers (cell size of 1m2), one from the edge of the riparian forest and one from the edge of vegetated islands, both with channel distances ≥100’000 were created to account for downstream transport of CPOM which was imported into the channel network. Direct input and lateral transport were calculated for each cell of the floodplain using the estimated functions. Values of input and transport from islands and riparian forests were added in order to calculate total CPOM input and total CPOM transport to exposed sediment, large wood, and ponds. To upscale direct and lateral CPOM input to the channel network, cost distance layers (cell size of 1m2) including channels were created for distances from the edge of riparian forests and vegetated islands. For each cell, direct inputs and lateral transports were recalculated. Total CPOM input to channels was calculated by summing up inputs from the forest and from islands. Values of lateral transport were extracted at each meter along shorelines. To calculate lateral transport from the forest or from islands, corresponding shorelines were considered. Lateral transport was divided by means of channel widths to estimate lateral input (methodological details: see Zah 2001).

Results Seasonal dynamics of aerial input and lateral transport Along individual transects, mean aerial CPOM inputs (AFDM m-2 year-1) ranged from 39.8 g ± 27.6 g (parallel transects) to 70.0 g ± 49.6 g (transects perpendicular to islands). Inputs varied over time (season) (Wilks’ λ: F9,316 =

17.2, P ≤ 0.001) and among CPOM fractions (F3,324 = 27.0, P ≤ 0.001) but did not differ among transect types (F2,324 = 2.9, P = 0.056) (Fig. 4).

-101- CPOM dynamics in a floodplain mosaic

Figure 4: Aerial CPOM input (g AFDM m-2 d-1) over an annual cycle in 2005 along transect types (n = 32, mean ± 1SE).

Leaves were the dominant CPOM fraction in aerial traps (average: 78.8%), and miscellaneous material was more abundant than wood and grass. The significant interaction between time (month) and CPOM fraction (Wilks’ λ:

F27,924 = 11.6, P ≤ 0.001) emerged as composition changed seasonally from miscellaneous material (dominant from March to May) to leaves (July to December) (Fig. 5).

Figure 5: Relative contribution of CPOM fractions in aerial input traps along transects: A) perpendicular to the riparian forest, B) perpendicular to islands, and C) parallel to islands. All figures n = 32.

Mean lateral CPOM transport along transects (AFDM m-1 year-1) ranged seasonally (Wilks’ λ: F9,594 = 58.3, P ≤ 0.001) from 12.2 g ± 4.9 g (riparian forest)

-102- Chapter 5 to 20.5 g ± 8.9 g (transects orthogonal to islands). Mean lateral transport was different among transect types (F2,6023 = 4.7, P = 0.010), between direction of trap opening (F7, 602 = 6.2, P = 0.013), and among CPOM fractions (F3,602 = 172.9, P ≤ 0.001). Transport was significantly higher along transects orthogonal to islands than orthogonal to the riparian forest (Fig. 6). Traps facing towards the vegetation collected significantly more CPOM (Fig. 7B, D) than traps facing towards the open tract (Fig. 7A, C). Litter composition varied seasonally (Wilks’

λ: F27,1735 = 50.9, P ≤ 0.001), being dominated by miscellaneous material from March to August, and by leaves from September to December (Fig. 7).

Figure 6: Seasonal lateral CPOM transport (g AFDM m-1 d-1) within the vegetation (upper panel; A, B) and in the open tract (lower panel; C, D). (A) open towards the open tract, B) open towards the vegetation) (mean ± 1SE, n = 4), (C) open towards the open tract, D) open towards the vegetation) (mean ± 1SE, n = 28).

-103- CPOM dynamics in a floodplain mosaic

Figure 7. Relative composition of laterally transported CPOM along transects. Upper panel: perpendicular to riparian forests (A = open towards the open tract, B = open towards the vegetation); middle panel: perpendicular to islands (C = open towards the open tract, D = open towards vegetation); lower panel: parallel to islands (E = open towards the open tract, F = open towards vegetation). All figures n = 32.

Spatial patterns of aerial input and lateral transport Direct aerial CPOM inputs decreased exponentially from the edge of the riparian forest and vegetated islands towards the open tract (0.059 < k < 0.088, Fig. 8). At the forest edge, aerial inputs (AFDM m-2 y-1) averaged 213.8 g ± 150.0 g, and at the island’s edges 323.5 g ± 119.5 g (perpendicular transects) and 175.2 g ± 82.0 g (parallel transects), respectively. In the open tract (1 m to 70 m distance from vegetation edge) average aerial CPOM input (AFDM m-2 y-1) was 15.4 g ± 10.9 g (riparian forest transects), 27.8 g ± 21.4 g (perpendicular island transects), and 17.3 g ± 14.2 g (parallel island transects). Aerial input was dominated by leaves (-5 to 20 meters) and by miscellaneous material (20 to 70 meters from the vegetation edge) (Fig. 9).

-104- Chapter 5

Figure 8. Decrease in aerial input of CPOM (g AFDM m-2 y-1) with increasing distance from the vegetation (m) (n = 40, mean ± 1SE). Functions: forest (perpendicular): input = 11.868 * e(-0.059*distance), island (perpendicular): input = 44.160 * e(-0.088*distance), island (parallel): input = 8.801 * e(-0.073*distance).

Figure 9: Relative composition of CPOM (direct input) in relation to the distance from vegetation. A) Transects perpendicular to the riparian forest, B) transects perpendicular to islands, and C) transects parallel to islands. All figures n = 40.

-105- CPOM dynamics in a floodplain mosaic

Wind speed and total aerial input were negatively correlated along all transect types (logarithmic functions: 0.24 ≥ R2 ≤ 0.38). Correlations between wind direction and aerial input were significant for all transect types fitting negative logarithmic and linear functions equally well (logarithmic: 0.16 ≥ R2 ≤ 0.37, linear: 0.16 ≥ R2 ≤ 0.36). Lateral CPOM transport decreased with distance to the vegetation following logarithmic functions (Fig. 10).

Figure 10: Decreasing lateral transport of CPOM (g AFDM m-1 year-1) with increasing distance from the vegetation boundary (m) (n = 40, mean ± 1SE). A) Traps open towards the open tract, B) traps open towards the vegetation. Functions: forest (perpendicular): transport = -5.03 * log(distance) + 23.13, island (perpendicular): transport = -4.27 * log(distance) + 26.84, island (parallel): transport = -3.28 * log(distance) + 12.81. Note different scale.

At the vegetation edges, lateral transport (AFDM m-1y-1) ranged from 10.3 g ± 4.74 g (islands) to 21.0 g ± 5.3 g (riparian forest) and slightly declined along transects to minimum average values of 5.0 g ± 3.7 g (riparian forest) and 16.3 g

-106- Chapter 5

± 4.2 g (islands). The relative composition of CPOM transported laterally was dominated by leaves and remained relatively constant with distance to the vegetation, among transect types, and between trap sides (Fig. 11).

Figure 11: Lateral transport: relative composition of CPOM with increasing distance from vegetation. A) Traps perpendicular to the riparian forest (A = open towards the open tract, B = open towards vegetation), perpendicular to islands (C = open towards the open tract, D = open towards vegetation), and parallel to islands (E = open towards open tract, F = open towards vegetation) (n = 40).

Total lateral CPOM transport was not significantly correlated with wind speed or wind direction. Significant, although weak, correlations were found between lateral transport of leaves and wind speed following negative linear functions (0.16 ≥ R2 ≤ 0.23).

-107- CPOM dynamics in a floodplain mosaic

Spatial and temporal heterogeneity of CPOM standing stock CPOM standing stock (AFDM m-2) ranged from 6.7 g (exposed gravel) to 468.3 g (riparian forest), with an area-weighted mean for the entire floodplain of 154.6 ± 40.5 g (Tab. 2A, B).

Table 2: A) Area specific annual standing stock (g AFDM m-2), CPOM input (g AFDM m-2 y-1), and CPOM transport (g AFDM m-1 y-1) in individual floodplain habitats and area- weighted mean for the entire floodplain. B) Total amount of CPOM (t AFDM m-1 y-1) in each habitat and on the entire floodplain. Habitat composition was calculated for baseflow conditions. n.s. = not sampled, * data from van der Nat 2002, ** from van der Nat et al. 2002. A) Riparian Vegetated Large Exposed Shoreline Channel Pond Mean total forest island wood gravel floodplain Area covered by each habitat type (ha) 50.8 14.9 0.6 89.6 2.0* 30.4 0.8 189.1 CPOM standing stock (g AFDM m-2) 468.3 315.6 n.s. 6.7 16.6 4.5** 50.0** 154.6 Direct CPOM input (g AFDM m-2 y-1) 316.6 446.6 16.1 10.7 7.4 7.4 15.7 126.7 Lateral CPOM transport (g AFDM m-1 y-1) 41.3 25.4 14.3 12.7 15.6 --- 20.6 19.7 Lateral CPOM input (g AFDM m-2 y-1) ------0.7 15.6 1.4 ------B) Riparian Vegetated Large Exposed Shoreline Channel Pond Total forest island wood gravel floodplain Habitat composition (contribution to total floodplain area, %) 27.0 8.0 0.3 47.0 1.3 15.9 0.5 100 Total CPOM standing stock 237.9 47.0 n.s. 5.4 0.3 1.4 0.4 292.4 Total direct CPOM input 160.8 66.5 0.1 9.6 0.2 2.3 0.1 239.5 Total lateral CPOM transport 21.0 3.8 0.1 11.4 0.3 --- 0.2 36.8 Total lateral CPOM input ------1.1 0.3 0.4 ------

-108- Chapter 5

CPOM standing stock changed with time (Wilks’ λ: F4,173 = 8.1, P ≤ 0.001) and differed among habitats (F3,176 = 211.4, P ≤ 0.001) with highest amounts in the riparian forest significantly higher than on vegetated islands (Fig. 12B). On exposed gravel and along the shoreline, standing stock was significantly lower than in vegetated habitats (Fig. 12A). In vegetated habitats, standing stock decreased from spring to summer, and increased again in autumn. In spring, litter standing stock in the riparian forest was twice as high compared to vegetated islands. However in autumn, CPOM standing stock on islands increased to equally high levels as in the riparian forest. Standing stock on exposed gravel and along shorelines did not exhibit a distinct seasonal pattern.

Figure 12: CPOM standing stock (g AFDM m-2, n = 12, mean ± 1SE) on A) exposed sediment and along shorelines, and B) in vegetated habitats.

The composition of stored CPOM varied with time (Wilks’ λ: F12,458 = 5.3, P ≤ 0.001). The amount of wood and leaves was significantly higher than the amount of miscellaneous material and of grass. The significant interaction between habitats and CPOM composition (F9,176 = 17.1, P ≤ 0.001) arose from the fact that CPOM composition remained relatively constant over time on exposed gravel and along shorelines (Fig. 13B). However, on vegetated islands and in the riparian forest the relative amount of leaf litter decreased from spring to summer and increased towards the end of the year due to leaf abscission, while wood was a fairly constant background (Fig. 13A).

-109- CPOM dynamics in a floodplain mosaic

Figure 13: Composition of CPOM standing stock A) on exposed sediment, B) along shorelines, C) on islands, and D) in the riparian forest (n = 12). Misc. = miscellaneous material.

Upscaling CPOM dynamics to the entire floodplain Total aerial input (AFDM y-1) to the entire floodplain was 239.5 t (area- weighted mean: 126.7 g AFDM m-2 y-1, Fig. 14A, Tab. 2A, B). Lowest input was to large wood accumulations (0.1 t AFDM y-1) and highest input to the riparian forest (160.8 t AFDM y-1, Tab. 2B). Total lateral transport (AFDM y-1) was estimated as 36.8 t (area-weighed mean: 19.7 gAFDM m-1 y-1, Tab. 2A, B). Lateral transport (AFDM y-1) ranged from 0.1 t on large wood accumulations to 21.0 t in the riparian forest (Tab. 2B). Islands contributed more than 95% to the total aerial input in the active tract, and more than 65% to the total lateral transport. Annual lateral inputs to the open tract and into the channel network were 1.1 t AFDM y-1 (0.7 g AFDM m-2 y-1), and 0.4 t AFDM y-1 (1.4 g AFDM m-2 y-1) (Fig. 14B, Tab. 2A, B), whereby lateral input from islands to the open tract was two-orders-of-magnitude higher than input from the riparian forest. On average, 292.4 t AFDM were stored on the entire floodplain surface, whereof 81% were stored in the riparian forest and 16% on vegetated islands (Tab. 2B).

-110- Chapter 5

Figure 14: A) Annual direct input (g AFDM y-1) and B) annual lateral transport (g AFDM y-1) modelled for the entire floodplain, considering forest and vegetated islands. Color shade from black to white signifies in A) 55 (black) - 0 (white) g AFDM y-1, and in B) 50 (black) - 0 (white) g AFDM y-1.

Discussion We did a detailed year-round study on coarse particulate organic matter (CPOM) dynamics for a ~2 km2 large braided river-floodplain ecosystem. Rates of input, lateral transfer, and surface storage - all important components of organic matter dynamics - have been quantified, and spatially-explicit data on these rates are provided. Together with detailed decomposition data already available for the same floodplain system (Langhans and Tockner 2006, Langhans et al. submitted), we now are able to provide comprehensive insight into a major part of CPOM cycling for an entire floodplain ecosystem. Similar detailed studies are rare with exceptions being investigations on the pro-glacial Val Roseg floodplain, Switzerland (Zah 2001), and on the Ogeechee River, USA (Cuffney 1988). The potential pathways for CPOM input into the floodplain, including the open tract, are direct input from forested areas (239.5 t AFDM y-1) and fluvial input from upstream sections (60 t y-1, determined by van der Nat 2002 in a previous study). The key outputs are lateral and longitudinal transfer by wind and water, and decomposition. In the present study, we did not quantify fluvial input and output rates for CPOM. However, van der Nat (2002), who sampled suspended and floating organic matter at the up- and downstream end of the floodplain over an annual cycle, found no significant differences between input and export rates. In addition, frequent flood pulses typical for the Tagliamento

-111- CPOM dynamics in a floodplain mosaic

River were absent in 2005, which means that fluvial transport was not a key component of CPOM cycling during the present study period.

Seasonal dynamics of aerial input and lateral transport Across the investigated floodplain system, aerial CPOM input varied significantly among seasons (Fig. 4). While input rates in the riparian forest and on vegetated islands were similar to those reported from other temperate deciduous forests (e.g., Weigelhofer and Wahringer 1994, Pozo et al. 1997), direct inputs to the open tract were very low and restricted to areas close to vegetation (see also Zah et al. 2001). As expected, aerial input of CPOM peaked during leave abscission in autumn. The seasonal pattern of lateral CPOM transport was less variable than direct input (Fig. 5) suggesting that lateral transport is important for CPOM redistribution during periods with low direct input. Higher lateral transport close to islands than to the riparian forest might be explained by their higher productivity (CPOM input is significantly higher on islands than in the riparian forest) and by their higher exposure to wind. Correlations of wind speed and lateral CPOM transport were similar between islands and the riparian forest suggesting that higher leaf productivity was the dominant factor governing the higher lateral transport from islands. Standing stock dynamics supported this finding as from October to November, which is the time of leaf abscission, CPOM increased twice as much on islands as in the riparian forest. In Mediterranean regions, the autumnal periods of leaf abscission often coincide with floods or are followed by floods (e.g., Gonzalez and Pozo 1996) resulting in a massive downstream export and lateral exchange of organic matter. Heavy autumnal rains are also a common phenomenon in the Tagliamento catchment but were absent during our study year in 2005.

Spatial patterns of aerial input and lateral transport Mean area-specific aerial CPOM input (g AFDM m-2 y-1) to the total floodplain was much lower than in headwater streams and in meandering temperate floodplain systems (e.g., Bell 1978, Neatrour et al. 2004). Braided river- floodplains, such as the Tagliamento contain a wide range of habitat types, from large unproductive gravel fields to highly productive island patches (Fig. 1). It is the spatial vicinity of habitats with contrasting functional capacities that makes floodplain ecosystems so sensitive to patch composition and configuration. The

-112- Chapter 5 relative composition of aerial CPOM input rapidly changed with distance to the vegetation. Leaves were the dominant fraction close to the vegetation edge (0-20 m), whereby miscellaneous material prevailed from 20 m on. However, laterally transported CPOM was dominated by leaves at all distances. Consequently, leaves provide the main energy source in the habitat of their origin (island and riparian forest) but are also the predominant component of large-scale energy transfer across the open tract of the floodplain. Lateral transport of CPOM across the floodplain also declined with distance from vegetation. However, this decrease was less pronounced compared to direct aerial input and therefore, lateral transport was identified as the prevailing CPOM pathway for large parts of the open tract. Estimates of the contribution of lateral input to total CPOM input were found to range from negligible (Fisher 1977) to 70% (northwestern USA) (Cushing 1988). In our study, 84% and 89% of total litter input into the channel network and to exposed sediments derived from direct input while only 16% and 10% were from lateral input, respectively. Therefore, direct input was identified as the primary CPOM pathway of litter input on the floodplain scale. However, this heavily depends on the location of channels, as lateral transport gains importance with increasing distance from the vegetation edge. This shift in the dominant input pathway might be of general validity as it was also observed in coniferous forests (Zah and Uehlinger 2001). On exposed gravel, vegetated islands contributed two orders of magnitude higher amounts of CPOM to lateral inputs than the riparian forest which identifies them as key sources of CPOM for the open floodplain tract despite their relative small area.

CPOM dynamics and wind Studies investigating the influence of wind on CPOM dynamics are scarce. Teeri and Barrett (1975) found that wind was a substantial factor in the transport of organic matter deposited on snowbanks in a high arctic lowland ecosystem. However, in the riparian zones of Birthday Creek, a small upland rainforest stream in north Queensland (AUS), wind was not correlated with lateral CPOM transport (Benson and Pearson 1993). We found significant, although weak, correlations between wind speed and wind direction with direct input and lateral transport rates of leaves. The Tagliamento has a wide open floodplain tract (average width in this reach: 800 m) which means strong wind exposure for vegetated islands and the riparian forest edge. Therefore, we concluded that

-113- CPOM dynamics in a floodplain mosaic wind plays a role in the distribution of CPOM in the investigated riverine floodplain.

Spatial and temporal heterogeneity of CPOM standing stock CPOM standing stock in vegetated habitats was similar to other floodplains such as at the Ogeechee River, USA (e.g., Cuffney 1988), and was approximately two-orders-of-magnitude higher than in non-vegetated habitats. Therefore, the riparian forest and vegetated islands were clearly identified as hot spots of litter standing stock. Although CPOM standing stock on islands was 1.5 times lower than in the riparian forest in spring, equally high levels were reached in autumn. This finding implies higher annual productivity on vegetated islands, but also higher CPOM export in the form of lateral transport which might be induced by a higher perimeter-area ratio in these habitats. Consequently, the presence of vegetated islands in riverine floodplains increases organic-matter availability to non-vegetated areas which has major impacts on the nutrient input and nutrient cycling (van der Nat et al. 2002). Islands along braided rivers can be considered as “islands of fertility”, similar to vegetation patches in desert systems (e.g., Schade and Hobbie 2005). We recognize that vegetated islands not only serve as major source of and retention areas for CPOM, as demonstrated in the present study, but they may also function as landscape “diffusers” that release nutrients and dissolved organic matter to surface and subsurface flowpaths. Compared to vegetated habitats, litter standing stock along shorelines and on exposed gravel surface was low and not affected by leaf abscission. Along shorelines, CPOM was washed ashore during higher flow in April and September 2005 which was indicated by higher standing stock in these months. On exposed sediments, CPOM storage peaked in October after the main flood of year 2005 receded. These results underpin the importance of the flow regime for distributing organic matter in the open tract of the floodplain. In all habitats, CPOM was mainly stored in the form of wood, whereas largest amounts of transported CPOM (direct and lateral) consisted of leaves.

Implications for stream ecology The island-braided floodplain is composed of a complex mosaic of aquatic and terrestrial landscape elements. These elements play different roles in the cycling of coarse particulate organic matter. Islands and the riparian forest are key production and storage areas while aquatic habitats, in particular channels, are

-114- Chapter 5 areas with a high decomposition potential (Langhans et al. submitted). Channels are also important for the fluvial transfer of CPOM to further downstream sections. However, the open tract serves as a transfer and filter zone for organic matter, both through wind and water activity, and therefore links different habitat types. Our study shows that leaves, seeds and fruits are primarily transferred laterally, while large wood is a relatively stable CPOM component (and exhibits low decomposition rates), as had been demonstrated by van der Nat et al. (2003) for the same study area. Hence, the relative composition of stored CPOM changed over an annual cycle due to rapid decomposition and massive lateral transfer of leaves (see Figures 7 and 12). Large wood may form the matrix structure of stored CPOM in riverine floodplains, while other fractions are subject to transfer and transformation.

Acknowledgements Many thanks to Claudio Cruciat for his assistance in the field, and to Richard Illi, Erika VanDaalen, and Denise Weibel for analytical assistance in the laboratory. We also thank Lukas Indermaur and Fabienne Suter for providing ArcGis maps of the studied floodplain reach, and Diego Tonolla and Rosi Siber for helping with Arcmap. This research was funded by the EU commission, supported through the tempQsim-project (contract no. EVK1-CT-2002-0012), and by the Swiss State Secretariat for Education and Research (SBF no. 02.0072).

References Bell, D. T., F. L. Johnson, and A. R. Gilmore. 1978. Dynamics of litterfall, decomposition, and incorporation in the streamside forest ecosystem. Oikos 30: 76-82. Benfield, E. F. 1997. Comparison of litterfall input to streams. Journal of the North American Benthological Society 16: 104-108. Benson, L. J., and R. G. Pearson. 1993. Litter inputs to a tropical Australian rain-forest stream. Australian Journal of Ecology 18: 377-383. Box, G. E. P. 1954. Some theorems on quadratic forms applied in the study of analysis of variance problems. Annals of Statistics 25: 290-302. Cuffney, T. F. 1988. Input, movement and exchange of organic matter within a subtropical coastal blackwater river-floodplain system. Freshwater Biology 19: 305-320. Cummins, K. W. 1974. Structure and function of stream ecosystems. BioScience 24: 631-641. Cushing, C. E. 1988. Allochthonous detritus input to a small, cold desert spring stream. Verhandlungen internationaler Vereinigung für Limnologie 23: 1107-1113.

-115- CPOM dynamics in a floodplain mosaic

Dance, K. W., H. B. N. Hynes, and N. K. Kaushik. 1979. Seasonal drift of solid organic matter in 2 adjacent streams. Archiv für Hydrobiologie 78: 139-151. Fisher, S. G., and G. E. Likens. 1973. Energy flow in Bear Brook, New Hampshire - integrative approach to stream ecosystem metabolism. Ecological Monographs 43: 421- 439. Fisher, S. G. 1977. Organic matter processing by a stream-segment ecosystem: Fort River, Massachusetts, USA. Internationale Revue der Gesamten Hydrobiologie. 62: 701-727. Gergel, S. E., S. R. Carpenter, E. H. Stanley. 2005. Do dams and levees impact nitrogen cycling? Simulating the effects of flood alterations on floodplain denitrification. Global Change Biology 11: 1352-1367. Gonzalez, E., and J. Pozo. 1996. Longitudinal and temporal patterns of benthic coarse particulate organic matter in the Aguera stream (northern Spain). Aquatic Sciences 58: 355-366. Gurnell, A. M., G. E. Petts, D. M. Hannah, B. P. G. Smith, P. J. Edwards, J. Kollmann, J. V. Ward, and K. Tockner. 2001. Riparian vegetation and island formation along the gravel- bed Fiume Tagliamento, Italy. Earth Surface Processes and Landforms 26: 31-62. Hedges, J. I., G. L. Cowie, J. E. Richey, P. D. Quay, R. Benner, M. Strom, and B. R. Forsberg. 1994. Origins and processing of organic matter in the Amazon River as indicated by carbohydrates and amino acids. Limnology and Oceanography 39: 743-761. Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river-floodplain systems. In: D. P. Dodge (ed.). Proceedings of the International Large River Symposium. Canadian Special Publication of Fisheries and Aquatic Sciences 106. pp. 110-127. Karrenberg, S., J. Kollmann, P. J. Edwards, A. M. Gurnell, and G. E. Petts. 2003. Patterns in woody vegetation along the active zone of a near-natural Alpine river. Basic and Applied Ecology 4: 157-166. Krueger, C. 2004. A comparison of the general linear mixed model and repeated measures ANOVA using a dataset with multiple missing data points. Biological Research for Nursing 6: 151-157. Langhans, S. D., S. D. Tiegs, U. Uehlinger, and K. Tockner. 2006. Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems. Polish Journal of Ecology 54: 675-680. Langhans, S. D., S. D. Tiegs, M. O. Gessner, and K. Tockner. Leaf decomposition across aquatic and terrestrial habitat types in a riverine floodplain mosaic. Submitted. Maamri, A., H. Chergui, and E. Pattee. 1994. Allochthonous input of coarse particulate organic matter to a Moroccan mountain stream. Acta Oecolgica 15: 495-508. Malard, F., U. Uehlinger, R. Zah, and K. Tockner. 2006. Flood-pulse and riverscape dynamics in a braided glacial river. Ecology 87: 704-716. Merritt, R. W., and D. L. Lawson. 1979. Leaf litter processing in floodplain and stream communities. Strategies for Protection and Management of Floodplain Wetlands and other

-116- Chapter 5

riparian ecosystems, Johnson, R. R., and J. F. McCormick (eds.). United States Department of Agriculture Forest Service General Technical Report WO-12. pp. 93-105. Meyer, J. L., A. C. Benke, R. T. Edwards, and J. B. Wallace. 1997. Organic matter dynamics in the Ogeechee River, a blackwater river in Georgia, USA. Journal of the North American Benthological Society 16: 82-87. Neatrour, M. A., J. R. Webster, and E. F. Benfield. 2004. The role of floods in particulate organic matter dynamics of a southern Appalachian river-floodplain ecosystem. Journal of the North American Benthological Society 23: 198-213. Pozo, J., E. Gonzáles, J. R. Díez, J. Molinero, and A. Elésegui. 1997. Inputs of particulate organic matter to streams with different riparian vegetation. Journal of the North American Benthological Society 16: 602-611. Pringle, C. 2003. What is hydrologic connectivity and why is it ecologically important? Hydrological Processes 17: 2685-2689. Schade, J. D., and S. E. Hobbie. 2005. Spatial and temporal variation in islands of fertility in the Sonoran Desert. Biogeochemistry 73: 541-553. Smock, L. A. 1990. Spatial and temporal variation in organic matter storage in low-gradient headwater streams. Archiv für Hydrobiologie 118: 169-184. Stanford, J. A., M. S. Lorang, and F. R. Hauer. The shifting habitat mosaic of river ecosystems. Verhandlungen internationaler Vereinigung für Limnologie 29: 123-136. Teeri, J. A., and P. E. Barrett. 1975. Detritus transport by wind in a high arctic terrestrial ecosystem. Arctic and Alpine Research 7: 387-391. Thoms, M. C., M. Southwell, and H. M. McGinness. 2005. Floodplain-river ecosystems: Fragmentation and water resources development. Geomorphology 71: 126-138. Thorp, J. H., and M. D. Delong. 1994. The riverine productivity model: an heuristic view of carbon sources and organic processing in large river ecosystems. Oikos 70: 305-308. Tockner, K., D. Pennetzdorfer, N. Reiner, F. Schiemer, and J. V. Ward. 1999. Hydrological connectivity, and the exchange of organic matter and nutrients in a dynamic river floodplain system (Danube, Austria). Freshwater Biology 41: 521-535. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: A model ecosystem of European importance Aquatic Sciences 65: 239-253. Valett, H. M., M. A. Baker, J. A. Morrice, C. S. Crawford, M. C. Molles, C. N. Dahm, D. L. Moyer, J. R. Thibault, and L. M. Ellis. 2005. Biogeochemical and metabolic responses to the flood pulse in a semiarid floodplain. Ecology 86: 220-234. van der Nat, D., A. P. Schmidt, K. Tockner, P. J. Edwards, and J. V. Ward. 2002. Inundation dynamics in braided floodplains: Tagliamento River, Northeastern Italy. Ecosystems 5: 636-647. van der Nat, D. 2002. Ecosystem processes in the dynamic Tagliamento River (NE-Italy). PhD dissertation. ETH Zürich. pp. 159.

-117- CPOM dynamics in a floodplain mosaic van der Nat, D., K. Tockner, P. J. Edwards, and J. V. Ward. 2003. Large wood dynamics of complex Alpine river floodplains. Journal of the North American Benthological Society 22: 35-50. Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell, and C. E. Cushing. The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences 37: 130-137. Ward, J. V., K. Tockner, P. J. Edwards, J. Kollmann, G. Bretschko, A. M. Gurnell, G. E. Petts, and B. Rossaro. 1999. A reference river system for the Alps: The ‘Fiume Tagliamento’. Regulated Rivers – Research & Management 15: 63-75. Ward, J. V., and J. A. Stanford. 1995. Ecological connectivity in alluvial river ecosystems and its disruption by flow regulation. Regulated Rivers – Research & Management 11: 105- 119. Webster, J. R., J. B. Wallace, and E. F. Benfield. 1995. Organic processes in streams of the eastern United States. In: Ecosystems of the world: river and stream ecosystems. Cushing, C. E., K. W. Cummins, and G. W. Minshall (eds.). Elsevier, New York. U.S.A. pp. 117- 187. Weigelhofer, G., and J. A. Waringer. 1994. Allochthonous input of coarse particulate organic matter (CPOM) in a first to fourth order Austrian forest stream. Internationale Revue der gesamten Hydrobiologie 79: 461-471. Zah, R. 2001. Patterns, pathways, and trophic transfer of organic matter in a glacial stream ecosystem in the Alps. PhD Dissertation. ETH Zürich. pp. 139. Zah, R., and U. Uehlinger. 2001. Particulate organic matter inputs to a glacial stream ecosystem in the Swiss Alps. Freshwater Biology 46: 1597-1608.

-118-

CHAPTER 6

Seasonal variation of riparian arthropods along lateral and vertical gradients in a braided gravel-bed river

Simone D. Langhans, Barbara Keller, Manfred Kahlen, and Klement Tockner In preparation

Braided river-floodplain ecosystems are composed of a complex habitat mosaic including exposed riverine sediments. Although the importance of the sediment surface for the distribution of riparian arthropods has been widely acknowledged, the ecological role of the unsaturated zone beneath the surface is mostly unknown. We simultaneously investigated terrestrial arthropod distribution across a large gravel bank of a braided floodplain (Tagliamento, NE-Italy) in four dimensions, i.e. the air space up to a height of 2 m above the sediment/water surface, the surface of exposed riverine sediments, and the unsaturated zone (down to 1 m beneath the sediment surface), each over a 10- month time period (fourth dimension). Additionally, beetle assemblages on the sediment surface and in the unsaturated zone were examined to identify habitat variables governing their distribution at the family and species level. Terrestrial arthropods were most abundant at interfaces (sediment-air, water-land), as well as in the riparian forest. We identified a total of 307 beetle species from 34 families, whereof 260 species (31 families) inhabited the sediment surface and 128 species (22 families) the unsaturated zone. Carabids and staphylinids accounted for 31% and 34%, respectively of the total species composition on the sediment surface and for 24% and 42% in the unsaturated zone e. At both taxonomic levels, the environmental variables ‘sediment depth’ and’ temperature’ explained most of the variation. Our study underpins the importance of environmental gradients for riparian arthropod composition and density in riverine floodplains, and the need to include the unsaturated zone (and the air space) in future investigation to get a complete view of factors controlling their populations.

-119- Riparian arthropods along environmental gradients

Introduction Natural rivers are dynamic, and physically and biologically complex (Tockner and Stanford 2002). They are characterized by a set of fluvial styles including straight, braided, wandering, and meandering channels (Richards et al. 2002). Braided gravel-bed rivers are widespread in temperate piedmont and mountain- valley areas (Tockner et al. 2006). In their pristine state, braided rivers are characterized by a shifting mosaic of channels, ponds, bars, and islands, since both flow and flood pulses create a diversity of habitats with fast turnover rates (Ward et al. 2002, Naiman et al. 2005). Braided rivers serve as excellent model systems to elaborate upon the complex relationship between habitat complexity and biodiversity. Braided rivers are also key areas for conservation and restoration since they provide habitat for a highly endangered fauna and flora (Tockner et al., 2003; Sadler et al., 2004). The understanding of their natural complexity and dynamics, however, is a prerequisite for developing sustainable management schemes (Ward et al. 2001). Ecological research in braided rivers has mainly focused on the sediment surface and to a lesser extent on the subsurface areas. While the hyporheic zone beneath the wetted channel is well known as an active component of the stream ecosystem, the unsaturated zone (i.e. hypogeic zone) is a “black box”, at least from the ecological point of view. However, extensive layers of unsaturated gravel, often several meters thick, are a key feature of braided rivers. Even if only parts of the extensive hypogeic system are accessible for riparian arthropods it is likely to be the most extensive habitat within braided rivers (Plachter and Reich 1998). This unsaturated zone may be crucial as a habitat and for the survival and the rapid recolonisation of terrestrial arthropods after flood and drought events. Dieterich (1996) exposed sediment cages at different depths within a gravel bar, and found a diverse invertebrate community comprising aquatic (oligochaeta, larvae of midges and stoneflies) and terrestrial (mites, rove beetles, ground beetles) species. High densities of terrestrial arthropods occurred in winter, which underpins the potential role of unsaturated sediments as refuge areas during the cold season (e.g. to overwinter). In addition, the area immediately above the water and sediment surface, i.e. the air space, is an important but virtually neglected since not specifically defined habitat for emerging aquatic insects, terrestrial arthropods, as well as for birds and bats. Landscape features such as large wood deposits and vegetated islands may play a key role as orientation marks or as structural components (e.g. for mating and

-120- Chapter 6 resting) for flying insects as well as for birds. While the reciprocal linkage between aquatic and terrestrial habitats has recently emerged as an important research topic (Baxter et al. 2005, Paetzold et al. 2005) little is known on the interaction between the unsaturated zone (i.e. unsaturated gravel), the sediment surface, and the air space. Exposed riverine sediment surfaces are inhabited by a diverse and specialized riparian arthropod fauna including spiders, rove beetles, and ground beetles (Hering and Plachter 1997, Sadler et al. 2004, Paetzold et al. 2006), which are subsidized by emerging aquatic insects (at least along gravel-bed rivers). As a consequence, especially carabid beetles reach high densities along gravel-bed rivers (Hering and Plachter 1997, Paetzold et al. 2005). In the present study, we systematically sampled terrestrial arthropods along environmental gradients in a braided floodplain complex over a 10-month period. Vertically, we sampled the air space, the sediment surface, and the unsaturated zone. Horizontally, we sampled from the channel edge into the riparian forest. We applied comparable sampling devices that allowed for direct comparisons in density, activity, and composition of terrestrial arthropods. Coleopterans from the sediment surface and from the unsaturated zone were identified to species level since beetles are a highly abundant and species-rich group in braided rivers, and they are excellent indicators of the environmental conditions of river- floodplain complexes (e.g. Manderbach and Reich 1995, Bonn et al. 2002, Sadler et al. 2004). The main objectives of this study were i) to simultaneously quantify terrestrial arthropod distribution in the air space, on the exposed sediment surface, and within the unsaturated zone (vertical gradient) across a large gravel bank (lateral gradient) over a 10-month period (temporal aspect), ii) to identify habitat variables governing the distribution of beetle assemblages (at the family and species level) inhabiting the sediment surface and the unsaturated zone, and iii) to establish the extent to which the sediment surface and the unsaturated zone provide habitats for the beetle fauna. We hypothesized that riparian arthropods accumulate at the interface between water and land (shoreline), between air and sediments (sediment surface), and between the open tract and the riparian forest. Further, we predicted a distinct separation of the beetle assemblage along lateral and vertical gradients. Finally, we hypothesized that the unsaturated zone provides habitat for a diverse and specialized fauna.

-121- Riparian arthropods along environmental gradients

Material and methods Site description Experiments were conducted from February to November 2005 in the braided section of the 7th-order Tagliamento River in northeastern Italy (46° N, 12°30’ E). The Tagliamento originates at 1000 m.a.s.l. in the southern fringe of the European Alps and drains a catchment of 2580 km2. It flows almost unimpeded by dams for 172 km to the Adriatic Sea. The river has a flashy flow regime with frequent flow and flood pulses (sensu Tockner et al. 2000) throughout the year (Arscott et al. 2002, van der Nat et al. 2002). Our study was conducted in an island-braided reach (river-km 79.8-80.8; 135 m a.s.l.), where maximum annual water level fluctuations are about 2 m (Tockner et al. 2003). This reach contains a spatially complex and temporally dynamic habitat mosaic dominated by extensive areas of exposed riverine sediments (Petts et al. 2000) (Fig. 1). The 800 m wide active tract is fringed by a ribbon of intact riparian forest, with Populus nigra and five willow species as the dominant tree species (Karrenberg et al. 2003).

Experimental design Experiments were conducted across a 220 m wide gravel bank, bordered by a 20 m wide channel on the left side and by a small alluvial groundwater channel (width: ≤ 5 m) and the riparian forest on the right bank (Fig. 1). Sediments on the gravel bank consisted of gravel and pebble (Tab. 1C, Ward et al. 1999, Tockner et al. 2003) with patches of sand along the alluvial groundwater channel. Terrestrial arthropods, including the adult stages of aquatic insects, were sampled at multiple locations along three transects that extended from the channel edge into the riparian forest (Fig. 1). Three different methods that allowed quantitative comparisons among them were applied: (i) window traps (so-called ‘combi-trap’; Gygax, 1999) to sample the airscape, (ii) pitfall traps to collect arthropods at the sediment surface (Spence and Niemelä 1994), and (iii) tube traps (modified after Dieterich 1997) for quantifying the mobility and density of arthropods in the unsaturated zone (detailed description in Fig. 2). Combi-traps have the advantage that they are easy to maintain, empty and service. Furthermore, catches do not depend on wind direction. Pitfall traps, although they measure activity x density rather than true abundance (Luff 1975, Baars 1979), are considered as a standard method in terrestrial ecology (Spence and Niemelä 1994).

-122- Chapter 6

100 m

Channel

Figure 1: The island-braided reach of the Tagliamento River, and the three transects established across the investigated gravel bar. The river flows from right to left.

Along the transects, tube- and pitfall traps were installed in the riparian forest and at 1 m, 5 m, 20 m, and 60 m -100 m from the channel edge. Window traps were placed at three distances from the channel edge: - 1 m (above the water surface), 5 m, and 60-100 m, plus an additional set of traps in the riparian zone. At each location, two traps, 0.5 m and 2.0 m above the sediment surface were set 5 m apart to account for spatial independence. Catch data were converted to units based on the collecting area of each type of trap and their exposure time, to be comparable across the three trapping methods (i.e. insect abundance m·2 day" 1 ). The collecting areas for the different trap types included the surface of the 2 plexiglass windows for window traps (0.8 m ) , the circular opening area of the 2 funnel for pitfall traps (0.018m ), and the area of two square holes for tube traps 2 (0.048m ).

-123- Riparian arthropods along environmental gradients

Sampling procedures At monthly intervals (February until November 2005) pitfall and window traps were exposed for one week, tube traps for two weeks. The content of the window traps was sieved through a 100 µm-mesh, and arthropods were stored in 80% ethanol. Plastic bottles from tube- and pitfall traps were closed with caps and transported to the laboratory. Samples were sieved through a 100 µm-mesh screen, and the arthropods retained by the screen were preserved in 80% ethanol. Pitfall- and tube-trap samples were sorted to order level and counted. Coleopterans from pitfall- and tube traps were identified to species levels according to Freude et al. (1964-1983), Lucht (1987), Lohse and Lucht (1989- 1994), Lucht and Klausnitzer (1998), and Assing and Schülke (1999, 2001). Window trap samples from April, July, and October 2005 were sorted to order level (Chinerey 1984) and counted.

Figure 2: Trap types used to collect terrestrial invertebrates in the three vertical layers. A) Combi-trap (Gygax 1999): combination of a window trap and a yellow pan trap. Two plexiglass windows, functioning as the sampling device, are fixed on a yellow funnel, which is filled with 60% ethanol as a preservative (plus a drop of detergent). The funnel is mounted on two wooden bars, which were hammered into the ground. B) Pitfall trap: outer tube (Ø = 15 cm, length: 30 cm) to stabilize the sediment, and the sampling device composed of a funnel (Ø = 15 cm) with the rim flush with the sediment surface and an attached plastic bottle (250 ml, part-filled with 70% ethanol as a preservative and a drop of detergent) covered by a plastic roof. C) Tube trap: outer plastic tube (inner Ø = 15 cm) with square, trellised holes (20

-124- Chapter 6 cm x 12 cm, mesh size = 1 cm) a pair each at 20 cm, 60 cm, and 100 cm sediment depth. Three inner tube segments each consisting of a place holder with two opposite openings (20 cm x 12 cm), and a sampling device composed of a white funnel attached to a white plastic bottle (250 ml, 70% ethanol and a drop detergent). Tube traps were installed with an excavator four months before the experiment started to diminish disturbance effects on the invertebrate fauna.

Environmental factors Air temperature, sediment temperature in three different depths, rainfall, and wind speed were continuously measured during the experiment. Rainfall and wind speed were recorded using a HOBO micro weather station (onset, Pocasset, MA, USA). Relative humidity and air temperature were measured two meters above ground with a HOBO Pro RH/Temp (onset, Pocasset, MA, USA) data logger and a Vemco Minilog data logger (MINILOG12-TR-40/+50-064K, Vemco, Nova Scotia, Canada), respectively. Temperatures within the sediment were measured at depths of 0.2 m, 0.6 m, and 1.0 m using Vemco Minolog data logger. Sediment water content (%), organic matter content within the sediment (kg AFDM m-3), and grain size distribution (%) were examined once in November 2004 while tube traps were installed. To measure sediment water content, a know amount of moist sediment was weighed, dried (48 h, 60°C), and reweighed. The dry mass was divided by the moist weight of the sediment to calculate percentage of water content. To calculate sediment organic-matter contents, sediment (< 2 mm, 1 l) was rinsed with deionized water to separate organic matter from the inorganic fraction. A known amount of the water/organic-matter mixture was passed through a fiber glass filter, which was dried (5 d, 60°C), weighed and burnt in a muffle furnace to calculate AFDM. The same sediment was passed through different sieves to obtain five grain size classes (1: 8 mm – 4 mm, 2: 4 mm – 2 mm, 3: 2 mm – 1 mm, 4: 1 mm – 0.063 mm, 5: ≤0.063 mm). All environmental factors are listed in Table 1 A-C.

-125- Riparian arthropods along environmental gradients

Table 1: Physical characteristics of the study site at the Tagliamento River floodplain during the experiment (mean ± 1 SD). A) Meteorological characteristics of the air space (continuous records), B) sediment surface and subsurface (unsaturated zone) temperatures, and C) sediment characteristics. Grain size fractions: 1 = 8 mm - 4 mm, 2 = 4 mm - 2 mm, 3 = 2 mm - 1 mm, 4 = 1 mm - 0.063 mm, 5 = < 0.063). n.r. = not recorded. A) Air temperature RH Rain Wind speed (°C) (%) (mm) (m s-1) February 2.3 ± 1.4 44.0 ± 17.6 n.r. n.r. March 7.4 ± 3.7 59.7 ± 15.9 n.r. n.r. April 12.2 ± 6.3 67.9 ± 66.4 128.6 1.09 May 18.7 ± 7.6 62.3 ± 8.9 35.0 0.67 June 22.4 ± 7.4 64.5 ± 11.7 37.6 0.71 July 23.8 ± 6.9 68.9 ± 6.2 101.0 0.55 August 21.2 ± 6.5 73.6 ± 9.3 133.2 0.39 September 19.4 ± 6.8 73.6 ± 71.3 165.8 0.42 October 13.8 ± 1.2 76.1 ± 12.2 0.6 0.36

B) Temperature (°C) Surface 0.2 m 0.6 m 1.0 m February 1.3 ± 6.0 0.8 ± 2.6 1.9 ± 0.8 3.1 ± 0.5 March 9.0 ± 8.1 7.9 ± 3.7 6.7 ± 2.2 6.1 ± 1.6 April 11.8 ± 4.8 11.6 ± 2.9 11.7 ± 0.8 11.5± 0.4 May 17.3 ± 7.1 17.4 ± 3.9 16.2 ± 1.1 15.1± 0.4 June 21.7 ± 8.3 21.5 ± 5.0 20.5 ± 1.7 19.2± 0.9 July 25.5 ± 7.2 26.1 ± 3.8 24.2 ± 0.8 22.1± 0.8 August 21.8 ± 6.5 21.7 ± 3.2 20.8 ± 1.0 20.5± 0.5 September 21.4 ± 7.6 21.0 ± 2.9 20.8 ± 0.8 19.9± 0.6 October n.r. 12.6 ± 2.6 13.6 ± 0.6 14.0± 0.4 November 3.7 ± 5.9 3.8 ± 2.9 6.5 ± 1.9 8.5 ± 1.5

-126- Chapter 6

C) Distance from Depth Water content Grain size fractions Organic matter channel edge 1 2 3 4 5 content (AFDM) (m) (m) (%) (%) (kg m-3) 1 0.2 5.2 ± 2.9 27.4 19.5 17.4 35.4 0.3 9.1 ± 1.2 5 0.2 3.0 ± 2.2 28.4 19.5 15.4 36.5 0.2 8.7 ± 1.5 20 0.2 2.2 ± 0.7 27.3 17.8 16.9 37.7 0.3 9.4 ± 1.9 20 0.6 2.1 ± 0.7 31.9 21.0 14.0 32.9 0.2 8.4 ± 1.1 100 0.2 1.6 ± 0.4 29.9 18.4 12.1 39.3 0.3 12.3 ± 7.4 100 0.6 2.2 ± 0.5 21.4 15.5 12.7 50.1 0.3 9.2 ± 0.8 100 1.0 2.1 ± 0.4 35.1 23.9 15.0 25.9 0.1 6.7 ± 1.5 Forest 0.2 25.7 ± 10.3 0.1 0.1 0.7 80.1 19.0 40.9 ± 15.6

Data analysis Influence of spatial and temporal heterogeneity on terrestrial arthropod abundances Analyses were based on transects as the unit of replication, and within transects, traps at similar locations (distance to the channel edge and distance to the riparian forest) were treated as replicates. Catches were considered to be independent within sites, given the small surface of each trap compared to the investigated area. Two-way ANOVAs were performed to test for differences in total terrestrial arthropod abundance (N m-2 d-1) along three different gradients: Vertical gradient (air, sediment surface, and unsaturated zone), lateral gradient (distance to the channel edge: -1 m or 1 m, 5 m, 60-100 m, and forest), and temporal gradient (three seasons, April, July, October) with separate analyses for the two different distances. If main effects were the same for both analyses, only results from the analysis including distance to the channel edge were shown. Subsequent Tukey’s post-hoc tests were used to determine the significance of pairwise differences within the three gradients. For each dimension, separate two-way ANOVAs were used to test for differences in total terrestrial arthropod abundance among the lateral and the temporal gradient. Trap height and sediment depth (vertical gradient) were included in two-way ANOVAs regarding the air sapce and the unsaturated zone, respectively. Subsequent Bonferroni’s (for unequal sample sizes) and Tukey’s (for equal sample sizes) post-hoc tests were used to determine the significance of pairwise differences

-127- Riparian arthropods along environmental gradients between season, distance, and depth. Abundance data were log-transformed (ln(x+1)) to control for heteroscedasticity. ANOVAs were performed using SPSS (ver. 13.0/SPSS Inc., Illinois, USA) with significance levels set at P ≤ 0.05.

Influence of spatial and temporal heterogeneity on beetle abundances Repeated measures ANOVA was used to test for differences among beetle abundances along the vertical and the lateral gradients over time. The vertical gradient consisted of four levels including the sediment surface (0 m), and 0.2 m, 0.6 m and 1.0 m sediment depth. Lateral gradients, including distances to the channel edge and to the riparian forest, were analyzed in separate repeated measures ANOVA.

Spatial and temporal heterogeneity in beetle assemblages One goal in investigating beetle assemblages was to relate faunal distributions to measured environmental, spatial, and temporal gradients. These relationships were summarized by analyzing environmental, spatial, and temporal covariance using multivariate techniques. All analyses were performed using the 4.5 CANOCO package (Ter Braak and Šmilauer 2002). All explanatory variables were log(x+1)-transformed in order to normalize and bring them to a common scale as were arthropod abundance data. Abundance data were used as they provide a better overall measure of species-environment relationships than do presence/absence data (Cushman and McGarigal 2004). The arthropods were analyzed at two different taxonomic levels (i.e., family, species) to identify fidelity to habitat conditions. Detrended Correspondence Analysis (DCA) was carried out on both, the family and the species data set. The data sets only included species which were captured in at least ten samples to be able to establish correlation patterns. The resulting gradient lengths were in both datasets > 3.5, suggesting that unimodal analytical techniques were more appropriate (Ter Braak and Šmilauer 1998). We used partial CCA (described by Borcard et al. 1992) to partition how much of the variation in the two data sets was caused by spatial, temporal, and environmental gradients, or by spatially and temporally structured environmental variation (Cushman and McGarigal 2004). For this analyses, we used three sets of explanatory variables: spatial variables (SV: distance to the channel edge, and distance to the riparian forest), environmental variables (EV:

-128- Chapter 6 depth, rainfall, water level, temperature, sediment water content, sediment organic matter content, and five different grain size classes), and temporal variables (TV: month and seasons). Forward selection procedures (Ter Braak and Šmilauer 2002) with 999 Monte Carlo permutations and α ≤ 0.05 on all three explanatory variable sets were used for the selection of variables to include in the models. Selected variables for the family and the species data sets which were included in later analyses are listed in Table 2A and B. Next, we hierarchically partitioned the variance in the family and species data that was explainable by the three different explanatory variables to quantify the strength of species-environment relationships. The hierarchical variance partitioning approach explicitly separated the effects of spatial-, environmental-, and temporal variables on the structure of the arthropod community. The details of the approach are given by Cushman and McGarigal (2003) and Anderson and Gribble (1998). Further, separate CCAs were run to detect patterns in the family and species data sets and environmental, spatial, and temporal variables. Statistical significance of the entire CCAs and of the first canonical axes was determined by Monte Carlo tests (999 permutations, Lepš and Šmilauer 2003).

-129- Riparian arthropods along environmental gradients

Table 2: Environmental, spatial, and temporal variables determined by forward selection in CCA in order of explained variability at A) the family level, and B) the species level (in order of their contribution to the entire model). EV = environmental variable, SV = spatial variable, TV = temporal variable. Marginal effect: lists the individual variables in order of the variance they explain singly; conditional effect: shows the environmental variables in order of their inclusion in the model and the additional variance they explain at the time they were included in the model. A) Families Marginal effect s Conditional effects Eigenvalue (λ) F P Eigenvalue (λ) F P EV: depth 0.09 11.53 0.001 0.09 11.53 0.001 SV: distance to the forest 0.09 11.47 0.001 0.09 12.32 0.001 EV: temperature 0.07 9.48 0.001 0.06 10.04 0.001 EV: water level 0.05 6.38 0.001 0.04 5.70 0.001 EV: grain size (4 mm - 2 mm) 0.04 4.47 0.001 0.03 4.56 0.001 EV: rain 0.05 6.46 0.001 0.03 4.27 0.001 TV: season 0.04 4.98 0.001 0.03 4.45 0.001 EV: grain size (8 mm - 4 mm) 0.04 5.15 0.001 0.02 3.07 0.001 TV: month 0.06 7.41 0.001 0.01 2.43 0.003 EV: sediment water content 0.08 10.10 0.001 0.02 2.17 0.011 SV: distance to the channel 0.04 5.71 0.001 0.00 0.87 0.577 B) Species Marginal effects Conditional effects Eigenvalue (λ) F P Eigenvalue (λ) F P TV: month 0.35 11.26 0.001 0.35 11.26 0.001 EV: depth 0.33 10.54 0.001 0.32 11.17 0.001 SV: distance to the forest 0.32 10.5 0.001 0.31 11.19 0.001 EV: temperature 032 10.45 0.001 0.18 6.70 0.001 EV: grain size (2 mm - 1mm) 0.21 6.73 0.001 0.13 4.86 0.001 EV: rain 0.24 7.77 0.001 0.11 4.21 0.001 TV: season 0.12 3.64 0.001 0.13 5.07 0.001 EV: water level 0.23 7.24 0.001 0.07 3.12 0.001 SV: distance to the channel 10.16 0.32 0.001 0.08 3.10 0.001 EV: grain size (4 mm - 2 mm) 0.19 5.98 0.001 0.04 1.79 0.010 EV: sediment water content 0.32 10.45 0.001 0.04 1.62 0.030 EV: grain size (8 mm - 4 mm) 0.21 6.78 0.001 0.02 0.77 0.759

-130- Chapter 6

Results Seasonal dynamics of arthropod densities along lateral and vertical gradients Total invertebrate catches (expressed as density x activity values; individuals m- 2 d-1) ranged from 4 to 40 individuals in the air space, from 120 to 1’100 ind. on the sediment surface, and from 21 to 118 ind. in the unsaturated zone. In the air space, adult aquatic dipterans were the dominant order (all seasons, Fig. 3A). On the sediment surface, collembolans dominated in April, coleopterans in July, and spiders and coleopterans in October (Fig. 3B). In the unsaturated zone, collembolans dominated in April and July, and mites in October (Fig. 3C). In all three habitats, coleopterans accounted for approximately 20% of the total arthropod abundances, except for window- and tube traps in April and October, respectively, when coleopterans were absent. Catches from the sediment surface and the unsaturated zone were dominated by ground-dwelling arthropods. In window traps, catches were composed of aerial (70%) and ground-dwelling (22%) arthropods, mainly spiders and coleopterans.

Figure 3: Terrestrial arthropod composition A) in the air space, B) on the sediment surface, and C) in the unsaturated zone in three different seasons (means of transects, n = 18-39). Only orders which contributed ≥ 5% to the total arthropod abundance were included. Ac = Acarina, Ar = Araneae, Au = Auchenorrhyncha, C = Collembola, Co = Coleoptera, Di = Diptera, He = Hemiptera, Hy = Hymenoptera, O = Others, Op = Opiliones, Tr = Trichoptera. All individuals are adult arthropods.

-131- Riparian arthropods along environmental gradients

Total arthropod catches varied significantly along the vertical (F2,142 = 322.9,

P < 0.001) and the lateral gradient (distance to the channel edge: F2,142 = 6.9, P =

0.001; distance to the riparian forest: F5,127 = 8.1, P < 0.001), as well as with season (F2,142 = 24.0, P < 0.001). Vertically, density was always highest at the sediment surface. In the air space, overall density was mostly similar in traps installed at 0.5 and 2.0 m above the sediment surface (only close to the channel density was significantly higher at 0.5 m height; in the riparian forest at 2.0 m). In the unsaturated zone, density was similar in the three depth layers. Laterally, arthropod density peaked at 1 m and 5 m from the channel edge, and again in the riparian forest (all layers). Seasonally, highest abundance occurred in spring and summer (all layers).

Lateral, vertical, and seasonal distribution of coleopterans During the experiment, a total of 7’660 coleopterans were caught (sediment surface and unsaturated zone). Abundances (individuals m-2 d-1) ranged from 0 to 45 individuals in the unsaturated zone, and from 7 to 792 individuals on the sediment surface (Fig. 4). Beetle abundances varied significantly with season

(Wilks’ λ = 0.89, F9,32 = 36.4, P ≥ 0.001), among vertical layers (depth) (F3,40 =

182.8, P < 0.001), and laterally with distance to the channel edge (F3,49 = 11.0, P

< 0.001) and to the riparian forest (F7,40 = 9.5, P < 0.001). Vertically, abundances were significantly higher on the sediment surface than in the unsaturated zone. Laterally, abundance decreased from the channel edge towards the center of the gravel bar and was again higher in the riparian forest.

Species composition of coleopterans A total of 307 beetle species from 34 families were identified. On the sediment surface, 260 species from 31 families were found (Appendix 1), whereby carabids and staphylinids accounted for 31% and 34% of the total species richness, respectively. In the unsaturated zone, 128 species from 22 families were caught. Carabids (24%) and staphylinids (42%) were the most species-rich families (Appendix 1). Each of the remaining families contributed less than 5% to total species richness. Eighteen families with 89 species were found both at the sediment surface and in the unsaturated zone.

-132- Chapter 6

Figure 4: Beetle abundances (density x activity; N m-2 d-1, mean ± 1SD) on exposed sediments (surface and subsurface) at different locations across the gravel bank and in the riparian forest from February (F) to November (N) 2005. m = distance from the channel edge.

Environmental heterogeneity and coleopteran assemblages Partial CCAs with spatial (position along vertical and lateral gradients), temporal (season), and environmental data as independent variables revealed a similar variance partitioning at the family and the species level. The amount of explained variance was 28.3% (family data set) and 30.5% (species data set) (Fig. 5A, B). Most variance in the two data sets was captured by environmental variables (families: 13.9%, species: 15.4%). In both data sets the pure combined spatial/temporal component, and the combined spatial/temporal component of environmental variation explained very little of the variance (ST, and the shared fraction of all three variables, Fig. 5A, B).

-133- Riparian arthropods along environmental gradients

Figure 5: Percent variance partitioning of A) beetle family data, and B) all beetle species following partial CCA (E = pure environmental variation, S = pure spatial variation, T = pure temporal variation, TE = pure temporal component of environmental, SE = pure spatial component of environmental, ST = pure combined spatial/temporal component, and in the center the combined spatial/temporal component of environmental variation.

At the family level, CCA discriminated between two major groups: one correlated with ‘water level’ and one with ‘distance to the riparian forest’ (Fig. 6). An additional group was built by carabids and staphylinids which were correlated with sediment depth. Overall, the variables ‘depth’, ‘temperature’, and ‘distance to the riparian forest’ were the most assertive variables (Tab. 2A). ‘Depth’ and the ‘sediment water content’, and ‘temperature’ and ‘month’ were strongly correlated. The first axis of the CCA constrained by environmental, spatial and temporal variables explained 9.7% (P = 0.001) and remaining axes accounted for 18.5% of the variance in the distribution of beetle families.

-134- Chapter 6

Figure 6: CCA biplot of beetle family assemblages and environmental (EV), spatial (SV), and temporal variables (TV) in samples across the investigated gravel bank. Arrows point in the direction of increasing values for the respective variables; arrow angles relative to axis and other variables indicate correlation strengths (SV: distance f (from the riparian forest), distance ch (from the channel edge); EV: depth, temperature, water level, sediment water content (SWC), grain size 1 (GS 1), grain size 2 (GS 2); TV: season, month). Grain size fractions from Table 1. Circled families are: Chrysomelidae, Dryopidae, Dytiscidae, Haliplidae, Histeridae, Hydrophilidae, Lathrididae, Leiodidae, Limnichidae, Nitidulidae, Pselaphidae, Ptilidae, Scarabaeidae, Silphidae, Throscidae. Significance of the model was P = 0.001 (F = 15.67).

At the species level, ‘month’, ‘depth’, and ‘distance to the forest’ affected species distribution most (Tab. 2B). ‘Depth’ and the ‘sediment water content’ as well as ‘grain size fraction 1’ and ‘2’ were strongly correlated with each other. The first axis explained 8.6% (P = 0.001) and the remaining axes accounted for 21.9% of the total explained variance. Carabids dominated in spring and staphylinids later in the season, whereby they were also correlated with ‘depth’ and ‘distance to the forest’ (Fig. 7).

-135- Riparian arthropods along environmental gradients

Figure 7. CCA biplots of beetle species assemblages and environmental (EV), spatial (SV), and temporal variables (TV) in samples across the investigated gravel bank (grouped in families). Arrows point in the direction of increasing values for the respective variables; arrow angles relative to axis and other variables indicate correlation strengths (SV: distance f (from the forest), distance ch (from the channel edge); EV: depth, temperature, rain, water level, sediment water content (SWC), grain size 1 (GS 1), grain size 2 (GS 2), grain size 3 (GS 3); TV: season, month). Grain size fractions from table 1. Significance of the model was P = 0.001 (F = 18.72).

Discussion We investigated the arthropod community in three vertical habitat layers across a river-floodplain transect. To our knowledge this is the first study that simultaneously focused on the air space, the sediment surface, and the unsaturated zone. While the sediment surface of floodplains has extensively been studied for terrestrial arthropods (Boscaini et al. 2000, Bonn and Kleinwächter 1999, Sadler et al. 2004, Paetzold et al. 2005) relatively little is known about the ecological role of air- and unsaturated zone (but see Dieterich 1996, Loeser et al. 2006). The use of the same methodological approach to investigate all three layers allowed us to directly compare the diversity and the density of riparian arthropods across lateral and vertical environmental gradients.

-136- Chapter 6

Our results clearly emphasize the importance of ecotones as important habitats for a diverse and abundant fauna (Ward et al. 1999). While the entire riparian zone can be defined as an ecotone, i.e. a transition zone between relatively homogenous areas, additional ecotonal habitats within the riparian zone between the permanent aquatic and the permanent terrestrial system separate or link individual habitat types (Ward and Wiens 2001). In our study, arthropod abundance peaked at the sediment surface (sediment-air interface), at the channel edge (water-land interface) and in the riparian forest. The riparian forest is a highly productive area and therefore can support higher densities of terrestrial arthropods than the desert-like open tract of the braided river. Riparian vegetation provides resources for a variety of phytophagous beetles and detritivores. In addition, riparian vegetation functions as a retreat area during flood events (Adis and Junk 2002). Shoreline ecotones, on the other hand, are known to be suitable for predacious carabids and staphylinids, which depend on stranding and emerging aquatic insects as a food resource (Hering and Plachter 1997, Paetzold and Tockner 2005, Paetzold et al. 2006). High beetle abundances in May and June can be explained by the specific life cycle patterns of many riparian species that are synchronized with the emergence patterns of aquatic insects (e.g., Manderbach and Plachter 1997, Hering and Plachter 1997). In contrast to the sediment surface, where abundance decreased with distance to the shoreline, abundances within the sediments did not differ along lateral nor along vertical gradients. Lateral restrictions of arthropods at the sediment surface are mainly due to food availability and habitat conditions. Many beetles require moist sediment conditions for survival. However, the entire unsaturated zone was used as a habitat without distinct lateral and vertical trends in abundance. There, habitat conditions (moisture, temperature) remain rather constant in time and space, while limited food resources may restrain arthropod density. Due to sampling constrains, we could not sample the interface between the unsaturated and the saturated zone, which is another ecologically important ecotone. We may expect higher food availability at this interface which again could favor terrestrial arthropods. In the unsaturated zone, arthropod abundances were one-order-of-magnitude lower than on the sediment surface, but did not differ among vertical layers of sediments. As presumed by Dieterich (1996) the unsaturated zone provides a year-round habitat for ground-dwelling arthropods down to a depth of at least 1

-137- Riparian arthropods along environmental gradients m. In the present study we found that the arthropod composition in the unsaturated zone was similar but less divers than the assemblage at the sediment surface, which implies that the unsaturated zone does not contain a distinct fauna but provides habitat for a subset of the surface community. In the air space, arthropod abundances were much lower compared to the sediment surface and the unsaturated zone. Jackson and Fisher (1986) reported emergence rates of 700 to 156’000 insects m-2 year-1. However, temporal variation in aerial arthropod density along river margins can be distinct as aquatic insects exhibit synchronized emergence patterns in temperate zones triggered by temperature or by flow and flood pulses (Sweeney and Vannote 1982). In our study, adult stages of aquatic insects often accounted for < 5% of the total catch indicating that we might have missed main emergence events, or that emerging aquatic insects are primarily restricted to the air space above the water surface. Indeed, we found highest densities of emerging aquatic insects close to the river margin. Total biomass of emerging aquatic insects was found to consist of 60-99% dipterans (Jackson and Fisher 1986, Gray 1989). In our study, mean arthropod composition along transects was also dominated by dipterans in all seasons. This highlights that dipterans are major inhabitants of the air space not only at the channel edge where they emerge (Paetzold and Tockner 2005) but also across the whole exposed gravel field. Additionally, a high proportion of ground-dwelling arthropods collected in window traps, including spiders and coleopterans, indicate that the sediment surface and the air space of exposed riverine sediments are strongly linked by an exchange of organisms. However, we only sampled a layer up to 2 m above the sediment surface. This represents a small volume compared to the potential air space, which might actively or passively be exploited by insects. For example, Geerts and Miao (2005) have shown that in the Great Plains (US) the Convective Boundary Layer extents to > 1500 m above the land surface, and is dominated by well-defined regions of high insect concentrations. In another study, Loeser et al. (2006) found that flood-deposited clumps of intertwined plant material and inorganic debris attached to trees and elevated above the ground by past high water events present a unique, persistent, and unstudied habitat type for spiders. In all three vertical layers (air space, sediment surface, unsaturated zone), adult beetles were abundant and accounted for 17 to 29% of the collected arthropods. On the sediment surface, beetle abundances were up to one order of magnitude higher than within the sediment. However, in both habitats, channel

-138- Chapter 6 edges and the riparian forest provided key habitats as has been shown by Hering and Plachter (1997), and Sanzone et al. (2003). Spatial, temporal, and environmental variables all explained a significant amount of variation on the beetles’ family and species levels, although the largest part of the variance was explained by the environmental variables ‘depth’ and ‘temperature’. A large fraction of variation in both data sets remained unexplained, which is not uncommon in ecological studies (e.g., Borcard et al. 1992, Okland and Eilertsen 1994). From all families, only carabids and staphylinids showed a correlation with depth suggesting that these highly mobile carnivorous beetles not only forage on the sediment surface (Paetzold and Tockner 2005, Paetzold et al. 2005), but also actively frequent the unsaturated zone. The strong correlation of the species distribution with all three variable types (environmental, spatial, and temporal) highlights the importance of different heterogeneity gradients for the distribution of riparian beetles. Carabid species, for instance, were strongly correlated with the temporal variable ‘month’ indicating that most species of this family are spring breeders (Cárdenas and Hidalgo 2004) which might hibernate as adults in the riverine sediments. Staphylinids were more broadly distributed and influenced by additional variables such as ‘distance to the forest’, and ‘depth’ which highlights that their distribution depends on different heterogeneity gradients.

Implications Braided rivers are considered as harsh environments due to their high habitat turnover rate, their low productivity, and the extreme thermal variability (e.g. Ward and Uehlinger 2003, Tockner et al. 2006). At the Tagliamento, for example, temperature at the sediment surface can vary from 10°C during night to more than 40°C during daytime (K. Tockner, unpubl. data). Despite these harsh conditions, our results clearly demonstrate that braided rivers provide habitats for a highly diverse and abundant arthropod community, with a high proportion of threatened riparian specialists (Sadler et al. 2004). Our results indicate that floodplains are multidimensional ecosystems including the air space, the sediment surface, and the unsaturated zone. Especially the unsaturated zone, whose extent can be extensive, is a highly underestimated habitat for terrestrial arthropods including a diverse and abundant beetle fauna. Future investigations should clarify what services the

-139- Riparian arthropods along environmental gradients unsaturated zone provides for terrestrial arthropods, and how the different layers (air space, sediment surface, unsaturated zone) are linked with each other.

Acknowledgements We thank Claudio Cruciat and Janine Rueegg for their assistance in the field, and Simone Blaser for her help identifying invertebrates. We are grateful to Peter Gäumann and the Eawag Werkstatt team, who constructed the tube traps, and to Peter Duelli from the Swiss Federal Institute for Forest, Snow and Landscape Research (WSL, Switzerland), who provided the pitfall and window traps. Thanks to Achim Paetzold for commenting an earlier version of the manuscript. This research was funded by the European Commission, and supported through the tempQsim-project (contract no. EVK1-CT-2002-0012) and by the Swiss State Secretariat for Education and Research (SBF no. 02.0072).

References Adis, J., and W. J. Junk. 2002. Terrestrial invertebrates inhabiting lowland river floodplains of Central Amazonia and Central Europe: a review. Freshwater Biology 47: 711-431. Anderson, M. J., and N. A. Gribble. 1998. Partitioning the variation among spatial, temporal and environmental components in a multivariate data-set. Australian Journal of Ecology 23:158-167. Arscott, D. B., K. Tockner, and J. V. Ward. 2002. Geomorphic dynamics along a braided-river corridor in the Alps (Fiume Tagliamento, Italy). Ecosystems 5: 803-814. Assing, V., and M. Schülke. 1999. Supplemente zur mitteleuropäischen Staphylinidenfauna (Coleoptera, Staphylinidae). Entomologische Blätter 95: 1-31. Assing, V., and M. Schülke. 2001. Supplemente zur mitteleuropäischen Staophylinidenfauna (Coleoptera, Staphylinidae) II. Entomologische Blätter 97: 121-176. Baars, M. A. 1979. Catches in pitfall traps in relation to mean sensities of carabid beetles. Oecologia 41: 25-46. Baxter, C. V., K. D. Fausch, W. C. Saunders. 2005. Tangled food webs: reciprocal flows of invertebrate prey link streams and riparian zones. Freshwater Biology 50: 201-220. Bonn, A., and M. Kleinwächter. 1999. Microhabitat distribution of spider and ground beetle assemblages (Araneae, Carabidae) on frequently inundated river banks of the River Elbe. Zeitschrift für Ökologie und Naturschutz 8: 109-123. Bonn, A., K. Hagen, and D. Wohlgemuth-von Reiche. 2002. The significance of flood regimes for carabid beetle and spider communities in riparian habitats - a comparison of three major rivers in Germany. River Research and Applications 18: 43-64.

-140- Chapter 6

Boscaini, A., A. Franceschini, and B. Maiolini. 2000. River ecotones: carabid beetles as a tool for quality assessment. Hydrobiologia 422/423: 173-181. Borcard, D., P. Legendre, and P. Drapeau. 1992. Partialling out the spatial component of ecological variation. Ecology 73: 1045-1055. Cárdenas, A. M, and J. M. Hidalgo. 2004. Life cycles of riparian ground beetles (Coleoptera, Carabidae): strategies for reproduction in the bank of a temporary stream in the southern Iberian peninsula. Vie Milieu 54: 47-56. Chinerey, M. 1984. Insekten Mitteleuropas. Collins Publishers, London. pp.444. Cushman, S. A., and K. McGarigal. 2003. Hierarchical, multiscale decomposition of species- environment relationships. Landscape Ecology 17: 637-646. Cushman, S. A., and K. McGarigal. 2004. Patterns in the species-environment relationship depend on both scale and of response variables. Oikos 105: 117-124. Dietrich, M. 1996. Methods and preliminary results from a study on the habitat functions of the gravel bar interior in alluvial floodplains. Verhandlungen der Gesellschaft für Ökologie 26: 363-367. Freude, H., K. W. Harde, and G. A. Lohse. 1964-1983. Die Käfer Mitteleuropas, 11 Bände. Goecke and Evers, Krefeld, Germany. Geerts, B., and Q. Mia. 2005. Airborne radar observations of the flight behavior of small insects in the atmospheric convective boundary layer. Environmental Entomology 34: 361-377. Gray, L. J. 1989. Emergence production and export of aquatic insects from a tallgrass prairie stream. Southwestern Naturalist 34: 313-318. Gygax, R. 1999. Methodenoptimierung zur Schätzung der lokalen Biodiversität von Arthropoden. Diplomarbeit Zoologisches Museum Unviersität Zürich. Hering, D., and H. Plachter. 1997. Riparian ground beetles (Coleoptera, Carabidae) preying on aquatic invertebrates: a feeding strategy in alpine floodplains. Oecologia 111: 261-270. Jackson, J. K., and S. G. Fisher. 1986. Secondary production, emergence, and export or aquatic insects of a Sonoran desert stream. Ecology 67: 629-638. Karrenberg, S., J. Kollmann, P. J. Edwards, A. M. Gurnell, and G. E. Petts. 2003. Patterns in woody vegetation along the active zone of a near-natural Alpine river. Basic and Applied Ecology 4: 157-166. Lepš, J., and P. Šmilauer. 2003. Multivariate Analysis of Ecological Data using CANOCO. Cambridge University Press, U.K. pp. 269. Loeser, M. R., B. H. McRae, M. M. Howe, and T. G. Whitham. 2006. Litter hovels as havens for riparian spiders in an unregulated river. Wetlands 26: 13-19. Lohse, G. A., and W. H. Lucht. 1989-1994. Die Käfer Mitteleuropas, 1.-3. Supplementband mit Katalog. Goecke and Evers, Krefeld, Germany. Lucht, W. H. 1987. Die Käfer Mitteleuropas, Katalog. Goiecke and Evers, Krefeld, Germany. pp. 342.

-141- Riparian arthropods along environmental gradients

Lucht, W. H., and B. Klausnitzer. 1998. Die Käfer Mitteleuropas, 4. Supplementband. Goecke and Evers, Krefeld, Germany. pp. 398. Luff, M. L. 1975. Some features influencing the efficiency of pitfall traps. Oecologia 19: 345- 357. Manderbach, R., and M. Reich. 1995. Auswirkungen grosser Querbauwerke auf die Laufkäferzönosen (Coleoptera, Carabidae) von Umlagerungsstrecken der Oberen Isar. Archives of Hydrobiology Supplement 101: 573-588. Manderbach, R. and H. Plachter. 1997. Lebensstrategie des Laufkäfers Nebria picicornis (FABR. 1801) (Coleoptera, Carabidae) an Fliessgewässern. Beiträge der Gesellschaft für Ökologie 3: 17-27. Naiman, R. J., H. Décamps, and M. E. McClain. 2005. Riparia: Ecology, Conservation and Management of Streamside Communities. Elsevier/Academic Press, San Diego. pp.430. Okland, R. H., and O. Eilertsen. 1994. Canonical correspondence-analysis with variation partitioning - some comments and an application. Journal of Vegetation Science 5: 117- 126. Paetzold, A., and K. Tockner. 2005. Effects of riparian arthropod predation on the biomass and abundance of aquatic insect emergence. Journal of the North American Benthological Society 24: 395-402. Paetzold, A., C. J. Schubert, and K. Tockner. 2005. Aquatic-terrestrial linkages along a braided river: Riparian arthropod predation affects biomass and abundance of aquatic insect emergence. Ecosystems 8: 748-759. Paetzold, A., J. Bernet, and K. Tockner. 2006. Consumer-specific responses to riverine subsidy pulses in a riparian arthropod assemblage. Freshwater Biology 51: 1103-1115. Petts, G. E., A. M. Gurnell, A. J. Gerrard, D. M. Hannah, B. Hansford, I. Morrissey, P. J. Edwards, J. Kollmann, J. V. Ward, K. Tockner, and B. P. G. Smith. 2000. Longitudinal variations in exposed riverine sediments: a context for the ecology of the Fiume Tagliamento, Itlay. Aquatic Conservation: Marine and Freshwater Ecosystems 10: 249- 266. Plachter, H., and M. Reich. 1998. The significance of disturbance for populations and ecosystems in natural floodplains. Proceedings of the International Symposium on River Restoration. Tokyo, Japan. pp. 29-38. Richards, K., J. Brasington, and F. Hughes. 2002. Geomorphic dynamics of floodplains: ecological implications and a potential modelling strategy. Freshwater Biology 47: 559- 579. Sadler, J. P., D. Bell, and A. Fowles. 2004. The hydroecological controls and conservation value of beetles on exposed riverine sediments in England and Wales. Biological Conservation 118: 238-250. Sanzone, D. M., J. L. Meyer, E. Marti, E. P. Gardiner, J. L. Tank, and N. B. Grimm. 2003. Carbon and nitrogen transfer from a desert stream to riparian predators. Oecologia 134: 238-250.

-142- Chapter 6

Spence, J. R., and J. K. Niemelä. 1994. Sampling carabid assemblages with pitfall traps - the madness and the method. Canadian Entomologist 126: 881-894. Sweeney, B. W., and R. L. Vannote. 1982. Population synchrony in mayflies: a predator satiation hypothesis. Evolution 36: 810-821. Ter Braak, C. J. F., and P. Šmilauer. 1998. Reference manual and user’s guide to Conoco for Windows: Software for canonical community ordination (version 4). Microcomputer Power. Ithaca, U.S.A. Ter Braak, C. J. F., and P. Šmilauer. 2002. CANOCO Reference manual and CanoDraw for Windows User’s guide: Software for Canonical Community Ordination (version 4.5). Microcomputer Power. Ithaca. U.S.A. Tockner, K., and F. Malard, J. V. Ward. 2000. An extension of the flood pulse concept. Hydrological Processes 14: 2861-2883. Tockner, K., and J. A. Stanford. 2002. Riverine flood plains: present state and future trends. Environmental Conservation 29: 308-330. Tockner, K., J. V. Ward, D. B. Arscott, P. J. Edwards, J. Kollmann, A. M. Gurnell, G. E. Petts, and B. Maiolini. 2003. The Tagliamento River: a model ecosystem of European importance. Aquatic Sciences 65: 239-253. Tockner, K., A. Paetzold, U. Karaus, C. Claret, and J. Zettel. 2006. Ecology of braided rivers. IAS Special Publication. van der Nat, D., A. P. Schmidt, K. Tockner, P. J. Edwards, and J. V. Ward. 2002. Inundation dynamics in braided floodplains: Tagliamento River, Northeast Italy. Ecosystems 5: 636- 647. Ward, J. V., K. Tockner, P. J. Edwards, J. Kollmann, G. Bretschko, A. M. Gurnell, G. E. Petts, and B. Rossaro. 1999. A reference river system for the Alps: the ‘Fiume Tagliamento’. Regulated Rivers: Research and Management 15: 63-75. Ward, J.V., and J. A. Wiens. 2001. Ecotones of riverine ecosystems: Role and typology, spatio-temporal dynamics, and river regulation. Ecohydrology and Hydrobiology 1: 25-36. Ward, J.V., K. Tockner, U. Uehlinger, and F. Malard. 2001. Understanding natural patterns and processes in river corridors as the basis for effective river restoration. Regulated Rivers-Research & Management 17: 311-323. Ward, J.V., K. Tockner, D. B. Arscott, and C. Claret. 2002. Riverine landscape diversity. Freshwater Biology 47: 517-539. Ward, J.V., and U. Uehlinger. 2003. Ecology of a glacial floodplain. Kluwer Academic Publ. Dordrecht, NL. pp. 306.

-143- Riparian arthropods along environmental gradients

Appendix 1. Taxa list, and beetle abundances (individuals) on the sediment surface (surface) and in the unsaturated zone (subs.). Pooled data.

Taxa surface subs. Taxa surface subs.

Anthicidae Anisodactylus nemorivagus Anthicus bimaculatus (Duftschmid, 1812) < 10 Illiger, 1801 0 Asaphidion caraboides Anthicus luteicornis (Schrank, 1781) >100 1 Schmidt, 1842 < 10 Asaphidion flavipes Endomia occipitalis (Linnaeus, 1761) < 100 1 (Dufour, 1843) < 10 Asaphidion pallipes Endomia unifasciata (Duftschmid, 1812) 1 (Bonelli, 1812) 1 < 10 Bembidion azurescens Mecynotarsus serricornis Dalla Torre, 1877 < 10 (Panzer, 1796) < 10 Bembidion bugnioni Omonadus floralis Daniel, 1902 < 100 < 100 (Linnaeus, 1758) 1 Bembidion cruciatum bualei Aphodiidae Jacqelin du Val, 1852 < 100 < 10 Rhyssemus limbolarius Bembidion decorum Petrovitz, 1963 < 10 (Zenker, 1801) < 10 < 10 Byrrhidae Bembidion distinguendum Chaetophora spinosa (Jacqelin du Val, 1852) < 100 < 10 (Rossi, 1794) 1 Bembidion elongatum Cantharidae Dejean, 1831 < 100 < 10 Cantharis montana Bembidion fasciolatum Stierlin, 1889 1 (Duftschmidt, 1812) >100 < 100 Carabidae Bembidion foraminosum Abax ater ssp. Cf. Lombardus Sturm, 1825 < 10 < 10 Fiori, 1896 < 10 Bembidion fulvipes Abax carinatus porcatus Sturm, 1825 >100 < 100 (Duftschmied, 1812) < 100 1 Bembidion lampros Agonum duftschmidti (Herbst, 1784) 1 -cluster < 10 Bembidion properans Agonum fuliginosum Stephens, 1829 1 (Panzer, 1809) 1 Bembidion pseudascendens Agonum muelleri (Müller-Motzfeld, 2004) >100 < 100 (Herbst, 1784) < 10 Bembidion punctulatum Agonum sexpunctatum Drapiez, 1820 < 10 (Linnaeus, 1758) 1 Bembidion pygmaeum Agonum viduum (Fabricius, 1792) < 10 (Panzer, 1796) < 10 Bembidion quadrimaculatum Amara communis (Linnaeus, 1761) < 10 (Panzer, 1797) < 10 Bembidion ruficorne Amara similata Sturm, 1825 < 10 < 10 (Gyllenhal, 1810) < 10 Bembidion scapulare Anchomenus dorsalis oblongum Dejean, 1831 < 100 < 10 Pontoppidan, 1763 < 100 Bembidion testaceum Anisodactylus binotatus (Duftschmid , 1812) < 100 (Fabricius, 1787) < 100 Bembidion tetracolum

-144- Chapter 6

Say, 1813 < 100 < 10 Harpalus luteicornis Bembidion tibiale (Duftschmid , 1812) 1 (Duftschmid, 1812) 1 Limodromus assimilis Bembidion varicolor (Paykull, 1790) < 100 (Fabricius, 1803) 1 Lionychus quadrillum Broscus cephalotes (Duftschmid, 1812) >100 < 100 (Linnaeus, 1758) < 100 Nebria brevicollis Calathus erratus (Fabricius, 1792) < 10 (Sahlberg, 1827) < 100 Nebria picicornis Carabus cancellatus emarginatus (Fabricius, 1801 ) >100 Duftschmid, 1812 < 10 Notiophilus palustris Carabus granulatus (Duftschmid, 1812) 1 Interstitialis Duftschmid, 1812 < 100 Omophron limbatum Carabus violaceus ssp. 1 (Fabricius, 1776) >100 1 Chlaenius nitidulus Oodes helopioides (Schrank, 1781) < 100 (Fabricius, 1792) < 10 Chlaenius vestitus Paranchus albipes (Paykull , 1790) < 10 (Fabricius, 1796) < 100 Cicindela hybrida transversalis Paratachys bistriatus Dejean, 1822 < 100 (Duftschmid, 1812) 1 Clivina collaris Paratachys micros (Herbst, 1784) < 100 (Fischer-Waldheim, 1828) < 100 > 100 Clivina contracta Perileptus areolatus (Fourcroy, 1785) < 10 (Creutzer, 1799) < 10 < 10 Cylindera arenaria Poecilus lepidus Leske, 1785 < 100 < 10 (Fuesslin, 1775) < 10 Porotachys bisulcatus Cylindera germanica (Nicolaï, 1822) 1 < 10 (Linneaus, 1758) < 10 Pseudophonus griseus Diachromus germanus (Panzer, 1797) < 10 (Linnaeus, 1758) 1 Pseudoophonus rufipes Drypta dentata (De Geer, 1774) 1 (Rossi, 1790) 1 Pterostichus melanarius Dyschirius abditus (similis) (Illiger, 1798) < 100 < 10 (Fedorenko , 1993) 1 Pterostichus melas italicus Dyschirius gracilis (Dejean, 1828) 1 (Heer, 1837 ) < 10 Pterostichus niger Dyschirius intermedius (Schaller, 1783) < 100 < 10 (Putzeys, 1846) 1 Pterostichus nigrita Dyschirius minutus (Paykull, 1792) < 10 (Putzeys, 1867) 1 Pterostichus strenuus Dyschirius substriatus (Panzer, 1796) 1 (Duftschmied, 1812) < 10 Stenolophus teutonus Elaphropus diabrachys (Schrank, 1781) < 10 (Weipert, 1996) < 10 Stomis pumicatus Elaphropus parvulus (Panzer, 1796) 1 < 10 (Dejean, 1831) 1 1 Thalassophilus longicornis Elaphropus sexstriatus (Sturm, 1825) < 10 < 100 (Duftschmid, 1812) < 100 < 100 Chrysomelidae Elaphrus aureus Altica tamaricis Müller, 1821 >100 < 10 (Schrank, 1785) 1

-145- Riparian arthropods along environmental gradients

Chaetocnema conducta Hypothenemus eruditus (Motsch., 1838) 1 Westwood, 1836 < 10 Chaetocnema hortensis Lepyrus palustris (Geoffroy, 1785) 1 (Scopoli , 1763) < 10 Chaetocnema semicoerulea Neophytobius quadrinodosus (Koch, 1803) < 10 < 10 (Gyllenhal , 1813) < 10 Crepidodera aurata Phyllobius sinuatus (Marsham 1802) 1 (Fabricius, 1801) 1 Longitarsus nigrofasciatus Tanymecus palliates (Goeze, 1777) < 10 (Fabricius, 1787) < 10 Pachnephorus tesselatus Trachyphloeus asperatus (Olivier, 1792) 1 1 Boheman 1843 1 Phaedon armoraciae Ditiscidae (Linnaeus, 1758) < 10 Potamonectes depressus Phaedon cochleariae elegans (Panzer, 1795) 1 (Fabricius, 1792) < 10 Dryopidae Phaedon laevigatus Dryops ernesti (Duftschmid, 1825) < 100 Des Gozis, 1886 < 10 Psylliodes napi Dryops nitidulus (Heer, 1841) < 10 (Fabricius , 1792 ) 1 Dryops striatopunctatus Psylliodes picina (Heer, 1841) < 10 < 100 (Marsham, 1802) < 10 Dryops subincanus Coccinnellidae (Kuwert, 1890) < 100 1 Hippodamia variegata Dryops viennensis (Goeze, 1777) 1 (Laporte de Castelnau, 1840) < 10 < 10 Cryptophagidae Dytiscidae Atomaria gottwaldi Agabus congener Johnson, 1971 1 (Thunberg 1794) 1 Atomaria gravidula Agabus didymus Erichson, 1846 < 100 (Oliver, 1795) 1 Atomaria impressa Platambus maculatus Erichson, 1846 1 (Linnaeus, 1758) < 10 Atomaria lewisi Potamonectes eleganss Reitter , 1877 < 10 (Panzer, 1794) 1 Atomaria nigrirostris Elateridae Stephens, 1830 >100 > 100 Adrastus binaghii Atomaria plicata Leseigneur, 1969 < 10 < 10 Reitter 1875 1 Adrastus lacertosus Atomaria pusilla Erichson, 1841 < 10 1 (Paykull, 1798) 1 Adrastus pallens Cucujidae (Fabricius, 1792) 1 Cryptolestes ferrugineus Agriotes litigiosus (Stephens, 1813) < 10 (Rossi, 1792) 1 1 Curculionidae Betarmon bisbimaculatus Acalyptus carpini (Fabricius, 1803) < 10 (Fabricius, 1792) 1 Drasterius bimaculatus Bagous glabrirostris (Rossi, 1790) < 10 (Herbst, 1795) < 100 Negastrius sabulicola Dorytomus taeniatus (Boheman, 1853) 1 (Fabricius, 1781) < 10 Paracardiophorus musculus

-146- Chapter 6

(Erichson, 1840) < 10 Colon sp. 1 Synaptus filiformis Leiodes carpathica (Fabricius, 1781) < 10 < 10 (Ganglbauer, 1896) < 100 Zorochros alysidotus Leiodes pallens (Kiesenwetter, 1858) < 100 < 10 (Sturm, 1807) < 10 Zorochros meridionalis Leiodes rotundata (Laporte de Castelnau, 1840) >100 < 10 (Erichson, 1845) < 100 Zorochros minimus Limnichidae (Lacordaire, 1835) 1 Limnichus incanus Zorochros stibicki Kiesenwetter, 1851 < 100 Leseigneur, 1970 < 100 < 10 Limnichus sericeus Elmidae (Duftschmied, 1825) < 100 Esolus parallelepipedus Lucanidae (Müller, 1806) 1 Dorcus parallelepipedus Gyrinidae (Linnaeus, 1758) 1 Orectochilus villosus Monotomidae (Mueller, 1776) 1 Monotoma longicollis Haliplidae (Gyllenhal, 1827) < 100 Brychius glabratus Nitidulidae (Villa 1833) 1 Epuraea luteola Haliplus laminatus Erichson, 1843 < 10 Schaller 1783 1 Glischrochilus quadrisignatus Haliplus lineatocollis (Say, 1835) < 10 < 10 (Marsham, 1802) < 10 < 10 Stelidota geminata Histeridae (Say) < 100 Hypocaccus rugifrons Oedemeridae (Paykull, 1798) < 100 Anogcodes ferruginea Hydraenidae (Schrank, 1776 ) 1 Hydraena sp. 1 Pselaphidae Ochthebius nobilis Bibloplectus < 10 Villa, 1835 < 10 Brachygluta haematica Gruppe < 10 < 100 Hydrophilidae Brachygluta sp. 1 Georissus caelatus Brachygluta trigonoprocta Erichson, 1847 < 10 (Ganglbauer, 1895) < 10 Georissus laesicollis Brachygluta xanthoptera Germar, 1831 < 10 < 10 (Reichenbach, 1816) < 100 < 100 Laccobius alternus Bryaxis glabricollis Motschulsky, 1855 < 10 (Schmidt-Goebel, 1838) 1 Laccobius striatulus striatulus Bythinus reichenbachi (Fabricius, 1801) < 10 (Machulka, 1928) 1 < 100 Lathridiidae Meliceria sp. < 10 Corticaria pubescens Tychobythinus sp. < 100 Gyllenhall, 1827 1 Ptiliidae Melanophthalma curticollis Ptenidium longicorne (Mannerheim, 1844) < 10 (Fuss, 1868) < 100 < 10 Leiodidae Ptinella britannica Agathidium atrum Matthews, 1858 < 10 (Paykull, 1798) 1 Smicrus filicornis Colon affine (Fairmaire & Laboulbène, Sturm, 1839 1 1855) < 10

-147- Riparian arthropods along environmental gradients

Rhizophagidae Anotylus nitidulus Monotoma brevicollis (Gravenhorst, 1802) 1 Aubé, 1837 1 Anotylus rugosus Monotoma longicollis (Fabricius, 1775) >100 > 100 (Gyllenhal, 1827) >100 Anotylus tetracarinatus Rhizophagus picipes (Block, 1799) < 10 < 10 (Olivier, 1790) < 10 < 10 Apimela macella Scarabaeidae (Erichson, 1839) < 100 Hoplia brunnipes Atheta autumnalis Bonelli, 1807 < 10 (Erichson, 1839) < 10 Psammodius asper Atheta fungi (Fabricius, 1775) < 10 < 10 (Gravenhorst, 1806) 1 Psammodius pierottii Atheta pertyi Pittino, 1978 1 (Heer, 1839) 1 Rhyssemus limbolarius Atheta sodalis Petrovitz, 1963 < 100 (Erichson, 1837) < 10 Serica brunnea Atheta triangulum (Linnaeus, 1758) 1 (Kraatz, 1856) < 10 Scolitidae Bledius erraticus Xyleborus dispar Erichson, 1839 1 (Fabricius, 1792) < 10 1 Bledius litoralis Xyleborus germanus Heer, 1838 1 Blandford, 1894 < 10 1 Carpelimus bilineatus Xyleborus saxeseni (Stephens, 1834) < 100 < 10 (Ratzeburg, 1837) < 100 < 10 Carpelimus gracilis Scydmaenidae (Mannerheim, 1831) < 100 > 100 Chelonoidum latum Carpelimus opacus (Motschulsky, 1851) 1 < 100 (Baudi, 1848) < 10 Silphidae Chilopora rubicunda Phosphuga atrata Erichson, 1837 < 10 (Linnaeus 1758) < 10 Cypha pirazzolii Staphylinidae (Baudi, 1869) < 100 < 100 Aleochara haematoptera Deleaster dichrous Kraatz, 1856 < 10 < 10 (Gravenhorst, 1802) < 10 1 Aloconota appulsa Drusilla canaliculata (Scriba, 1867) 1 (Fabricius, 1787) < 10 Aloconota cambrica Gabrius nigritulus (Wollaston, 1855) < 10 (Gravenhorst, 1802) 1 Aloconota eichhoffi Gabrius osseticus (Scriba, 1867) 1 (Kolenati, 1846) < 10 Aloconota insecta Gabrius splendidulus (Thomson, 1856) 1 < 10 (Gravenhorst, 1802) < 10 Aloconota pfefferi Gabrius tirolensis (Roubal, 1929) 1 1 (Luze, 1903) 1 1 Aloconota planifrons Geodromicus suturalis (Waterhouse, 1864) 1 (Lacordaire, 1835) < 10 Aloconota sulcifrons Hydrosmecta fluviatilis (Stephens, 1832) < 10 (Kraatz, 1854) 1 Amischa analis Hydrosmecta gracilicornis (Gravenhorst, 1802) 1 (Erichson, 1839) 1 1

-148- Chapter 6

Hydrosmecta haunoldiana Gravenhorst, 1806 < 10 Bernhauer 1914 < 10 Omalium rivulare Hydrosmecta quadraticeps (Paykull, 1789) 1 Scheerpeltz, 1943 < 10 < 10 Paederidus rubrothoracicus Hydrosmecta valdieriana (Goeze, 1777) >100 < 100 Scheerpeltz, 1944 1 Paederidus ruficollis Ilyobates sp. 1 (Fabricius, 1781) < 100 < 100 Ischnosoma longicorne Paederus caligatus (Maeklin, 1847) < 10 Erichson, 1840 1 Ischnosoma splendidum Paederus limnophilus (Gravenhorst, 1806) < 10 Erichson , 1840 < 100 Lathrobium brunnipes Paederus riparius (Fabricius, 1792) 1 (Linnaeus, 1758) 1 < 10 Lathrobium castaneipenne Parocyusa cingulata (Kolenati, 1846) < 10 (Kraatz, 1856) 1 Lathrobium dilutum Parocyusa longitarsis (Erichson, 1839) 1 (Erichson, 1837) 1 < 10 Lathrobium ripicola Parocyusa rubicunda (Czwalina, 1888) 1 < 100 (Erichson, 1837) < 100 < 10 Leptacinus batychrus Philonthus rotundicollis (Gyllenhal, 1827) < 10 (Ménétriés, 1832) < 100 < 10 Leptacinus sulcifrons Philonthus rubripennis (Stephens, 1833) 1 Stephens, 1832 < 10 Liogluta longiuscula Plathysthetus nitens (Gravenhorst, 1802) < 10 (Sahlberg) 1 Medon brunneus Platydomene bicolor (Erichson, 1839) 1 (Erichson, 1840) < 100 Medon ripicola Platydomene picipes (Kraatz, 1854 ) < 10 (Erichson, 1840) 1 < 100 Neobisnius lathrobioides Platydomene springeri (Baudi, 1848) 1 < 10 (Koch, 1937) 1 < 10 Neobisnius procerulus Proteinus brachypterus (Gravenhorst, 1806) 1 (Fabricius , 1792 ) < 10 Neobisnius prolixus Proteinus ovalis (Erichson, 1840) 1 < 10 Stephens, 1834 < 100 < 10 Neobisnius villosulus Pseudomedon obsoletus (Stephens, 1832) < 100 < 100 (Nordmann, 1837) 1 1 Ochthephilus angustatus Quedius fuliginosus (Erichson, 1840) < 10 < 100 (Gravenhorst, 1802) 1 Ochthephilus angustior Rabigus tenuis (Bernhauer 1943) < 10 < 10 (Fabricius, 1792) < 100 Ochthephilus omalinus Scopaeus debilis (Erichson, 1840) < 100 (Hochhuth, 1851) 1 < 10 Ochthephilus praepositus Scopaeus gracilis Mulsant & Rey, 1878 1 < 10 (Sperk, 1835) < 10 Ochthephilus rosenhaueri Scopaeus laevigatus (Kiesenwetter, 1850) < 10 (Gyllenhal, 1827) < 10 Ocypus nitens Scopaeus sericans (Schrank 1781) < 10 (Mulsant & Rey, 1855) < 10 Omalium caesum Sepedophilus constans

-149- Riparian arthropods along environmental gradients

(Fowler, 1888) < 10 1 Tachyporus atriceps Sepedophilus marshami Stephens, 1832 1 (Stephens, 1832) < 10 Tachyporus austriacus Sepedophilus obtusus Luze, 1901 1 (Luze, 1902) 1 Tachyusa balteata Sepedophilus pedicularius Erichson, 1839 < 100 (Gravenhorst, 1802 ) 1 Taxicera deplanata Sepedophilus testaceus (Gravenhorst, 1802) 1 (Fabricius, 1792) < 10 Taxicera dolomitana Stenus boops ludmilae (Bernhauer 1900) < 100 Hromádka, 1979 < 10 Thecturota marchii Stenus carbonarius (Dodero, 1922) 1 Gyllenhal, 1827 < 10 Thinobius crinifer Stenus fossulatus Smetana, 1959 < 10 < 10 Erichson, 1840 1 Thinobius petzi Stenus guttula Müller, 1821 < 10 Scheerpeltz, 1957 < 10 < 100 Stenus latens Puthz, 2003 < 10 Thinobius sp. 1 Stenus longipes Heer, 1839 < 100 Thinodromus dilatatus Stenus palposus (Erichson, 1839) < 10 < 10 Zetterstedt, 1838 1 Xantholinus laevigatus Stenus phyllobates miscellus Jacobson, 1847) 1 Benick, 1925 1 1 Xantholinus linearis Stenus planifrons misael (Olivier, 1794) 1 Bondroit, 1912 < 10 Zyras limbatus Stenus ruralis Erichson, 1840 < 10 (Paykull, 1789) < 100 Tachyporus abdominalis Throscidae (Fabricius, 1781) < 100 Trixagus dermestoides Linnaeus 1766 < 1

-150-

SYNTHESIS AND OUTLOOK

Understanding spatial heterogeneity has been referred to as “the final frontier” within ecology (Kareiva 1994), and documentation of how environmental heterogeneity affects ecosystem processes by characterizing the patterns of heterogeneity and investigating the processes which underlie those patterns has been identified as an essential task (Pastor 2005). Natural riverine floodplains are, in particular, highly dynamic and complex landscapes and are therefore predestined for studying these issues. In the first part of my work, I conceptually established and experimentally investigated the influence of spatial and temporal heterogeneity on organic- matter dynamics, including leaf decomposition, aerial input, lateral transport and surface standing stock of coarse particulate organic matter. In the conceptual model for organic-matter dynamics in riverine floodplains, I combined leaf decomposition, organic-matter input, storage and quality, by stressing the importance of the flow regime. I predicted that the area of aquatic habitats relative to terrestrial floodplain habitats can be small, however, they exhibit a significantly higher decomposition potential than the dominant habitats such as exposed gravel or riparian forest. Conversely, aquatic habitats have low organic-matter input and standing stock, whereas large areas with relatively low decomposition potential receive large amounts of organic matter and standing stock (e.g., vegetated islands and the riparian forest). This mismatch effect is reduced when floods redistribute organic matter. Pre-processed organic matter with a more homogenous quality is then distributed among floodplain habitats. I concluded that at local scales temporal variability in the form of flow and flood pulses controls leaf decomposition via the flow regime and more broadly by linking sources of organic matter with areas of high decomposition potential. I found that duration of inundation was the main inundation component that controlled leaf breakdown rates, whereas frequency of drying and rewetting was responsible for fine-scale differences in the processing of leaves. Leaf-shredding macroinvertebrates significantly influenced breakdown rates only under permanent wet and dry conditions. Fungal biomass explained the faster leaf breakdown in winter. Hence, I concluded that the flow regime substantially influences leaf breakdown heterogeneity and accordingly has the potential to directly modify established decomposition processes either through a change in the duration of inundation or flow variation.

-151- Synthesis and Outlook

According to what I have predicted in the conceptual model, spatial heterogeneity separated leaf decomposition rates across the riverine floodpain in three distinct groups: fast decomposition rates in channels, medium rates in ponds, and slow rates in terrestrial habitats. In channels and ponds, leaf decomposition was driven by both microbial and detritivore activity, whereas in terrestrial habitats microbial activity was the main driver. My work shows that riverine floodplains provide a wide range of habitats with different decomposition potentials, resulting in a large spatial heterogeneity in the decomposition process of organic matter. Simplification in floodplain morphology decreases habitat diversity and homogenizes decomposition rates with consequences for the whole ecosystem functioning. Tracking the response of leaf decomposition in relation to several variables of environmental heterogeneity is time-consuming if the common leaf bag approach is used. Therefore, we tested the suitability of cotton-strips as an alternative, less laborious method to measure leaf decomposition. Cotton strip decay, measured as loss of tensile strength, broadly followed patterns of leaf decay (measured as leaf mass loss) across terrestrial and aquatic floodplain habitats. Consequently, cotton strips may be used as surrogates for leaf material in decomposition experiments. Previously predicted patterns of litter input and standing stock were supported by my field experiments. Additionally, I found that in the active tract of the floodplain vegetated islands contributed more than 95 % to the total aerial input and 65 % to the lateral transport or organic matter. Moreover, I identified vegetated islands and the riparian forest as key storage areas of CPOM. Litter is stored in form of wood, but transferred across the open tract of the floodplain in form of miscellaneous material in spring and leaves in autumn. Hence, I concluded that vegetated islands are not only key sources of coarse particulate organic matter in the open tract of the floodplain, but also increase, together with the riparian forest, the storage capacity of riverine floodplains. In the second part of my work, I examined the subject of how spatial and temporal heterogeneity determines the distribution of terrestrial floodplain invertebrates. I found that terrestrial floodplain invertebrates were distributed along vertical and lateral environmental gradients, and temporal gradients. Two key areas of high abundance could be identified, first the sediment surface and second the channel edge and the riparian forest. Vertically, from 2 m in the air to 1 m in the sediment, a clear differentiation of the terrestrial invertebrate fauna

-152- Synthesis and Outlook was found. Beetles were most abundant on the sediment surface 1 m from the channel edge and in the riparian forest. Both, ground beetles and rove beetles were negatively correlated with increasing sediment depth, whereas most other families were positively related with temperature and found near the riparian forest. Therefore, I concluded that spatial and temporal heterogeneity are important determinants for terrestrial invertebrate distribution. Additionally, the volume of unsaturated sediment and the air space contribute to overall riverine habitats and should be included in future investigations. From this research, I propose that both, spatial and temporal heterogeneity shape and determine organic-matter dynamics and terrestrial invertebrate distribution in riverine floodplains. Morphological and/or hydrological alterations have consequences for their functioning that can comprise changes in nutrient cycling, sediment respiration, primary and secondary production, and terrestrial and aquatic biodiversity.

Outlook Riverine floodplains provide multiple services, including the provision of clean drinking water, the self purification of waste water, the recharge of ground water, flood protection, the provision of habitats for a rich terrestrial and aquatic fauna and flora, as well as cultural and aesthetic values. There is a great need to identify and to quantify these ecosystem services. In order to achieve this, we need to disentangle the multiple effects on ecosystem processes governed by spatio-temporal heterogeneity. Within spatial heterogeneity, the composition, configuration, and connectivity of aquatic and terrestrial floodplain habitats, single and in concert, on both biodiversity and ecosystem processes must be considered (Fig. 1). In the situation where process rates vary little among habitats or where there is reduced exchange among them, ecosystem functions reflect habitat composition; while habitat configuration is less important (left side of Fig. 1). However, both composition and configuration play a role in determining the rate of an ecosystem process, depending on the nature of the boundary among habitats with different process rates and on the patterns of connection among habitats (right half of Fig. 1). The incorporation of the flow regime with duration, intensity, time, and frequency of inundation and drying will account for temporal heterogeneity. Finally, different scales must be addressed and compared by combining small- and large-scale field

-153- Synthesis and Outlook investigations and laboratory experiments to gain a holistic view of how spatio- temporal heterogeneity affects ecosystem processes.

Figure 1. Illustration of how habitat composition and configuration impact ecosystem function. Influence is indicated by arrows, and ellipses enclose aspects of configuration that influence process rates (adapted from Meyer 2005).

Established knowledge of the underlying principles determining how spatio- temporal heterogeneity relates to ecosystem functioning, to biodiversity and water quality will help linking river management and river restoration. Nowadays, national and European legal frameworks (e.g., Water Framework Directive) mandate that water resource management identifies and achieves quantifiable measures of water quality and ecological status. However, the integration of water resource use and sustainable ecosystem management is a complex challenge, especially for freshwater systems in highly utilized landscapes like riverine floodplains. By this means, feedback mechanisms between hydrological, ecological and morphological processes must be understood to an extent that allows for process-based prognostic modeling of key coupled processes at relevant scales.

References Kareiva P. 1994. Space: the final frontier for ecological theory. Ecology: 75: 1. Meyer, J. L. 2005. Heterogeneity and ecosystem function: enhancing ecological understanding and applications. In: Ecosystem function in heterogeneous landscapes, Lovett, G. M., C. G. Jones, M. G. Turner, and K. C. Weathers (eds.). pp. 451-461.

-154- Synthesis and Outlook

Pastor, J. 2005. Thoughts on the generation and importance of spatial heterogeneity in ecosystems and landscapes. In: Ecosystem function in heterogeneous landscapes, Lovett, G. M., C. G. Jones, M. G. Turner, and K. C. Weathers (eds.). pp. 49-66.

-155-

-156-

CURRICULUM VITAE

Simone Daniela Langhans

Department of Aquatic Ecology, Eawag/ETH Zürich P.O. Box 611, Überlandstrasse 133, 8600 Dübendorf, Switzerland Tel: +41 (0)44 823 5157; Fax: +41 (0)44 823 5315 Email: [email protected]

Date of Birth: 18 September 1976 Nationality: Swiss

RESEARCH and EDUCATION 2007 Postdoc, Department of Aquatic Ecology, Eawag/ETH Zürich 2002-2006 Dr. Sc. nat., Department of Aquatic Ecology, Eawag/ETH Zürich, ‘Riverine floodplain heterogeneity as a controller of organic matter dynamics and terrestrial invertebrate distribution’, under the supervision of Prof. Dr. Bernhard Wehrli 2000-2002 Teacher’s diploma in Biology, ETH Zürich 2000 Master thesis in Biology, Department of Aquatic Ecology, Eawag/ETH Zürich, ‘Floating organic matter: an ecological indicator of a riverine floodplain system (Tagliamento, northeastern Italy), under the supervision of Prof. James V. Ward and Prof. Dr. James Ward 1996-2000 Master Degree in Biology at ETH Zürich Major subjects: stream ecology and marine biology

TEACHING and other RESEARCH EXPERIENCE 2002-2006 Teaching assistant, Department of Aquatic Ecology, Eawag/ETH Zürich. 2005-2006 M.S. thesis supervisor: Erika Van Daalen, Yvonne Kunz, Janina Pohl 2004-2006 Applied-statistics workshops addressing ANOVA to undergraduate and M.S. students

-157- Curriculum Vitae

2002-2006 Teaching systematics and ecology of benthic invertebrates, ETH Zürich 2003-2007 Applied ecosystem study (Systempraktikum), engaged students in an ecological assessment of the Thur River, wrote protocols explaining field procedures, demonstrated and oversaw their implementation (e.g., macroinvertebrate / terrestrial arthropod sampling, identification), and supervised subsequent statistical analysis/interpretation

PEER REVIEWED PUBLICATIONS Kunz, Y., J. Pohl, and S. D. Langhans. 2007. Uferbezogene Indikatoren - Neue Ansätze zur Fliessgewässerbewertung. Wasser Energie Luft 99, no. 1. Tiegs, S. D., S. D. Langhans, K. Tockner, and M. O. Gessner. 2007. Cotton strips as a leaf surrogate to measure decomposition in river floodplain habitats. Journal of the North American Benthological Society 26: 70-77. Langhans, S. D., S. D. Tiegs, U. Uehlinger, and K. Tockner. 2006. Environmental heterogeneity controls organic-matter dynamics in river-floodplain ecosystems. Polish Journal of Ecology 54: 675- 680. Langhans, S. D., and K. Tockner. 2006. The role of timing, duration, and frequency of inundation in controlling leaf litter decomposition in a river-floodplain ecosystem (Tagliamento, northeastern Italy). Oecologia 147: 501-509. Trottmann, N., S. D. Langhans, and K. Tockner. 2006. Schwemmgut als Ausbreitungsmedium. Das Innenleben eines unterschätzten Naturstoffs. Wasser Energie Luft 98, no. 3: 207-213. Trottmann, N., S. D. Langhans, and K. Tockner. 2005. Schwemmgut, ein wichtiger Weg der Ausbreitung. Natur und Mensch 5: 8-11. Tockner, K., and S. D. Langhans 2003. Die ökologische Bedeutung des Schwemmgutes. Wasser Energie Luft 95, no. 11/12: 353-354.

-158- Curriculum Vitae

IN PREPARATION/SUBMITTED Langhans, S. D., S. D. Tiegs, M. O. Gessner, and K. Tockner. Leaf decomposition across aquatic and terrestrial habitat types in a riverine floodplain mosaic. Langhans, S. D., J. Rueegg, K. Tockner, and U. Uehlinger. Aerial input, lateral transport, and surface storage of coarse particulate organic matter in a riverine floodplain mosaic (Tagliamento, NE-Italy). Langhans, S. D., B. Keller, M. Kahlen and K. Tockner. Seasonal variation of riparian arthropods along lateral and vertical gradients in a braided gravel-bed river.

NON-REFEREED publications Langhans, S. D., and K. Tockner. 2005. Limnological characteristics of the Fiume Tagliamento. In: Tagliamento due sponde sul fiume - Guida storica e technica die un tratto del medio corso. Spilimbergo, Italy. Trottmann N., S. D. Langhans, and K. Tockner. 2004. Schwemmgut als Ausbreitungsmedium – das Innenleben eines unterschätzten Naturstoffs. Eawag Jahresbericht.

COMPETETIVE RESEARCH FUNDING and TRAVEL AWARDS 2006 Commission for Oceanography and Limnology, Swiss Academy of Sciences, grant for young scientists 2006 ETH travel grant 2005 Petersen Travel Award (NABS endowment fund) for non-North American student’s

JOURNAL REFEREE EXPERIENCE Hydrobiologia, Archiv für Hydrobiologie, Oecologia

CONFERENCE PRESENTATIONS 2006 Environmental heterogeneity controls organic matter dynamics in river-floodplain ecosystems. Annual Aquatic Ecology Symposium, Dübendorf, Switzerland (oral)

-159- Curriculum Vitae

2005 Differential effects of inundation on leaf litter decomposition. 6th International Gravel-Bed Rivers. Workshop Lienz, Austria (poster) 2005 Spatial heterogeneity of leaf litter decomposition in a complex mosaic of floodplain habitats. 4th Symposium for European Freshwater Sciences Krakow, Poland (oral) 2005 Spatial heterogeneity of leaf litter decomposition in a complex mosaic of floodplain habitats. 53st Annual Meeting of the North American Benthological Society. New Orleans, LA, USA (poster) 2004 The role of the inundation regime in controlling leaf litter decomposition in a braided river. 52st Annual Meeting of the North American Benthological Society. Vancouver, Canada (oral) 2003 Floating organic debris: a functional link in a river corridor. 51st Annual Meeting of the North American Benthological Society. Athens, GA, USA (poster)

-160-

ACKNOWLEDGEMENTS

Over the last four years, many people have accompanied my way and helped me both professionally and personally. Without their support I would certainly not be at the point where I am now. I would like to thank Klement Tockner for making this PhD possible, for inspiring me with his sweeping ideas, and for always being a supportive supervisor; Bernhard Wehrli, for taking the obligation to lead my dissertation; Urs Uehlinger, my second supervisor, who always knew an answer to my questions; and Stuart Findlay for being my external co- examiner. Lisa Shama and Scott Tiegs for helping me out with language problems, but more so for being fun travel mates and even better friends; Achim Paetzold, Ute Karaus, Michael Döring, and Lukas Indermaur, members of the Tagliamento-project group: a supportive team that always had a helpful advice at hand, and was fun to work with; Barbara Keller for explaining the mystery of multivariate statistics; Christian Rellstab for editing the final version of this work; Manfred Kahlen for identifying the beetles; Simone Blaser, Richi Illi, Janine Rüegg, Erika VanDaalen, and Denise Weibel, who assisted in the field and in the lab; Chris Robinson, Mark Gessner, Christiane Rapin, and Christoph Tellenbach, who were always willing to help with minor problems or to spend time with me chatting in the hallway; Herr Gäumann and his Werkstatt team for inventing and constructing the tube traps. Many many thanks to Ombretta and Claudio Mingolo Cruciat, my second family in Italy. They supported me not only with constant field work assistance, but especially with delicious food and social entertainment. My work would of course, not have been possible without the love and dedication of my parents, who introduced me to nature, taught me to respect it, and inspired my curiousness about scientific phenoma. And finally, I thank Sascha Kardaetz for learning me how to use words to express my thoughts and for being near.

-161-