Metamorphosis in river ecology: from reaches to macrosystems

Authors

  • James H. Thorp

    Corresponding author
    1. Kansas Biological Survey and Department of Ecology and Evolutionary Biology, University of Kansas, Higuchi Hall, Lawrence, KS, U.S.A
    • Correspondence: James H. Thorp, Kansas Biological Survey and Department of Ecology and Evolutionary Biology, University of Kansas, Higuchi Hall, 2101 Constant Ave., Lawrence, KS 66047-3759, U.S.A.

      E-mail: thorp@ku.edu

    Search for more papers by this author

Summary

  1. A metamorphic shift in the focus of aquatic ecological research is needed to (i) better understand the functioning of river ecosystems at large spatiotemporal scales; (ii) respond to threats from climate change, species invasions and increased pressure for more river dams and interbasin water transfer; and (iii) meet critical needs for catchment-level management.
  2. Implementing changes in research and management necessitate new spatiotemporal and research domains and a partial shift from reach-level studies to macrosystem ecology.
  3. Macroecology, as defined here, and associated processes usually cannot be understood simply by scaling up information from smaller extent and finer grain samples because of problems of connectivity and spatial heterogeneity of crucial system components and entanglements from cross-scale interactions producing threshold responses and other nonlinear dynamics.
  4. An important driver of macrosystem research in terrestrial and aquatic ecosystems is the immediate and long-term danger from global climate change, which will alter flow regimes, water temperatures and dissolved oxygen in riverine landscapes.
  5. An initial step in implementing metamorphic change in river science is to integrate the newest macroecological perspectives with current large-scale riverine concepts, especially ones emphasising the importance of tributary patterns and junctions (Network Dynamics Hypothesis) or the hydrogeomorphic patch nature of rivers at multiple scales (Riverine Ecosystem synthesis).
  6. The hierarchical nature, system boundaries and research domains of riverine macrosystem ecology are examined, and twelve sample research questions are provided to help direct future research.
  7. Current research impediments to riverine macroecology are discussed, and a partial solution based on the need for Long-Term Ecological Research-like sites (each with a central organisational hub and a diffuse, multi-institutional network of data collection stations at strategic catchment points) is proposed.

Introduction

Most historic changes in the science of aquatic ecology have been gradual and often resulted from either development of new techniques (e.g. analytical tools and geospatial models) or simple waning of interest in one subject countered by rising curiosity about another topic (e.g. shifting focus on roles of competition, predation and patch dynamics the last half of the 20th century). In contrast, more abrupt or dramatic metamorphic development in ecology periodically manifests with little warning. Like the process of arthropod ecdysis, metamorphosis in any scientific discipline can be hazardous if the subject stagnates between stages but is ultimately beneficial and necessary if allowed to proceed normally.

I suggest that an ongoing, metamorphic change in the broad field of ecology and environmental sciences relates to a growing emphasis on macroecology. For example, Lawton (1999) stated that: ‘… ecological patterns and laws, rules and mechanisms that underpin them are contingent upon the organisms involved and their environment’. He argued that understanding this contingency is possible at the level of populations and a few species, intractable at the broader community level, but again manageable for ‘… large sets of species, over large spatial scales, or over long time periods… ’ and recommended a focus on macroecology. Ricklefs (2008) went further by suggesting that the ‘… local community is an epiphenomenon that has relatively little explanatory power in ecology and evolutionary biology’, a view sharply disputed by Brooker et al. (2009). Whether one agrees that community ecology is generally an intractable focal research area, I suggest that aquatic ecologists need to give greater heed to macroecological approaches to understanding and managing natural systems.

At least four environmental drivers now seem to be merging in aquatic ecology, each of which requires broader spatiotemporal scales of research to address adequately urgent environmental questions. Three of these relate to concerns over immediate to long-term effects of climate change, interbasin-to-intercontinental species transfers and major modifications of the hydrological and geomorphic structure of rivers in response to heightened water and energy needs (e.g. reservoir construction, channelisation and inter-basin water transfer). A fourth is responsive to an increasing recognition of the critical need for catchment-level management. Some drivers have only recently begun receiving attention, while others have been prominent for decades but have not always been analysed at appropriate scales. This metamorphic change should not diminish in any way the value of prior research nor limit its continued usefulness but, instead, offers opportunities for examining new questions and sometimes improving our ability to analyse older ones.

Implementing the needed shifts in fundamental aquatic ecology and applied research and management will often require ecologists to employ new empirical and modelling techniques using a more unique suite of independent and dependent variables collected at larger spatiotemporal scales of grain and extent. In the last case, adapting to this new epistemological state will necessitate a change from a strong emphasis on reach-level studies to an increasing focus on macrosystem ecology.

Unfortunately, ‘macroecology’ is a discipline currently lacking a fully developed body of theory, as a group of macrosystem ecologists concluded in a 2012 workshop funded by the US National Science Foundation (Heffernan et al., in press). Therefore, I define here the relevant domains of this research field for aquatic ecologists, emphasise interactions between external processes and macrosystems, analyse the applicability of current aquatic theories to future macrosystem research, identify impediments to expanded macroecological research, provide some optional research directions and suggest major research initiatives to overcome some research impediments. I show why scaling up from a smaller spatiotemporal scale is often ineffective and macro-ecological approaches are needed to understand biodiversity, food-web complexity and ecosystem function.

The emphasis of this article is on fundamental river macroecology, although some applied aspects are briefly mentioned. In another paper (McCluney et al., in press), my coauthors and I discussed applied riverine macrosystems ecology in greater detail, especially as it relates to effects of dam construction and changes to the hydrogeomorphic structure of rivers.

Macroecology defined

‘Macrosystems ecology’ (or macroecology) is an evolving ecological discipline promising better understanding of the functioning of terrestrial and aquatic landscapes in response to regional through global-scale processes. Macroecological research encompasses a broad range of studies from species-level research at ‘biogeographical scales’ to global climate effects on nutrient spiralling in rivers.

As a relatively new ecological discipline, macroecology simultaneously benefits and suffers from occasionally vague, overly malleable definitions that are inherently scale (spatiotemporal) and process dependent. Nonetheless, recent applications in mostly terrestrial macroecology emphasise subcontinental to continental spatial extents, with concomitant interactions among environmental parameters operating from regional through global scales. Although this emphasis certainly encompasses a portion of aquatic macroecology, it may constrain important research initiatives in our discipline. Therefore, I define here macrosystems based on their physical size, temporal period of interactions and large-scale processes affecting them.

A macrosystem is a hierarchically organised, integrated terrestrial, inland aquatic and/or marine ecological unit of large spatial extent (c. 102–106 km2 or more depending on the types and sizes of ecosystems present) whose temporal interactions within the unit and with regional through global processes are especially significant over periods of decades to millennia.

It follows then that

Macroecology is the study of internal and external factors influencing ecosystem structure and function in macrosystems at large spatiotemporal scales.

Macroecological properties and processes usually cannot be understood simply by scaling up information from smaller extent and finer grain samples, in part because of problems of connectivity and spatial heterogeneity of crucial system components (Peters, Bestelmeyer & Turner, 2007; Peters et al., 2008; Heffernan et al., in press) and entanglements from cross-scale interactions producing nonlinear dynamics, such as threshold responses (Carpenter & Turner, 2000; Peters et al., 2008).

Fundamental riverine macroecology focuses on some combination of biological, geological, chemical and hydrological processes in natural environments, whereas applied macroecology examines anthropogenic effects, although some overlap occurs between the two foci. Some examples of fundamental macroecological research are studies of factors regulating material fluxes (e.g. nutrient cycling and carbon sequestration; such as Hadwen et al., 2009), system metabolism (e.g. Dodds et al., 2013), large-scale diversity patterns (e.g. Muneepeerakul et al., 2008; Geheber & Piller, 2012), carbon sources (e.g. Sullivan, 2013), food-web complexity (e.g. Romanuk et al., 2006) and overall ecosystem health (Sheldon et al., 2012). Such studies could emphasise within-ecosystem processes at large spatiotemporal scales or include studies on effects of regional through global processes. Multiple-ecosystem interactions (e.g. riverscape and floodscape within the riverine landscape) are especially relevant. As valuable as other ecological studies will continue to be, ecophysiological, autecological, population and small-scale community research rarely fit within macroecology.

Applied riverine macroecology might focus on ecological effects of large-scale changes to the hydrogeomorphic complexity of rivers from, for example, channelisation (e.g. Sabter, 2008), dam construction (Vörösmarty et al., 2010; Delong et al., 2011) and interbasin water transfer (e.g. Grant et al., 2012). Other studies might focus on general catchment management (e.g. Vaughan et al., 2009; McCluney et al., in press) or examine macroecological responses to climate change (e.g. Palmer et al., 2008) and species invasions (e.g. Strayer, 2010).

Rivers as macrosystems

System boundaries

Most macrosystem research has focused on subcontinental terrestrial systems and their interactions with regional through global processes, especially climate change. This has caused some scientists to question whether rivers are macrosystems – an attitude possibly reflecting misunderstanding about the nature of riverine landscapes.

A riverine macrosystem can be analysed as multi-ecosystem landscapes (cf., Sorrano et al., 2010) composed of a ‘riverscape’ (main and side channels plus true backwaters) and a ‘floodscape’ (wetlands, oxbows, floodplain lakes and normally dry terrestrial floodplains; Thorp, Thoms & Delong, 2008) interacting across longitudinal, lateral, vertical and temporal dimensions (cf., Ward, 1989). Floodscapes interact directly with riverscapes during flood pulses and indirectly at lower flows via the riparian zone, catchment run-off and groundwater flows. This link is integral to their macrosystem nature and may be greater than between any other two adjacent types of ecosystems. Although the smaller ‘active catchment zone’ interacts most significantly with the main channel, the entire catchment contributes to a riverine macrosystem through surface or groundwater pathways.

Riverine macrosystems are influenced not only by processes internal to the riverine landscape but also by long distance, organic exchanges – a form of telecommunication – with the ocean, atmosphere and other basins. While fresh water–ocean exchanges are clearly critical for diadromous species, these nutrient interactions also affect general aquatic and riparian production and food webs (Morris & Stanford, 2011). Carbon sequestration in enriched soils of adjacent ‘bottom lands’ of the floodscape (Hoffman, Glatzel & Dikau, 2009; Noe, Hupp & Rybicki, 2013) impacts carbon cycling on a global scale. Denitrification and methane production occurring in non-flowing backwaters of anastomosing rivers influence atmospheric gas exchange and seaward movement of nitrates, thereby exacerbating problems like hypoxia in the Gulf of Mexico offshore from the Mississippi River (Scavia et al., 2003).

Hierarchical nature

Riverine landscapes are formed of a mosaic of longitudinal, lateral and vertical patches such that ecological structure and function are influenced most intensely by patch attributes at scales immediately above and below the initial scale of interest (Wu & Loucks, 1995; Thorp, Thoms & Delong, 2006; Thorp et al., 2008). While the hierarchical organisation of rivers is essential to their nature as macrosystems, the arrangement of patches – especially on a river's longitudinal dimension – can also have dramatic effects from species distributions to ecosystem function (Thorp et al., 2008). For example, the ecological effects of a large waterfall on the Hudson River (New York State) would probably have been dramatically different on this otherwise shallow rocky Adirondack Mountain stream if the intervening waterfalls had been located farther upstream instead of at Glenn Falls, NY, where the river descends into a flooded estuarine valley. The need to understand the hierarchical nature of rivers to predict system responses contrasts with the less constraining site- or reach-based studies relying on single spatial and temporal scales (e.g. Woodward & Hildrew, 2002).

Generic responses to regional through global patterns

Perhaps most important with respect to recent macro-system definitions, hierarchical riverine landscapes (Frissell et al., 1986; Sorrano et al., 2010) are dramatically affected by regional through global patterns of temperature and precipitation. Every river has a somewhat unique natural flow regime (Poff et al., 1997) which is subject to normal flow fluctuations over months to decades or longer which produce droughts and flow/flood pulses. These patterns affect, for example, material exchanges with the floodscape, reproductive patterns and genetic selection of species for drought tolerance. When precipitation patterns become more intense and erratic from climate changes – as predicted for the US Great Plains (Dodds et al., 2004) and thereby western portions of the Mississippi River macrosystem – the altered flow regime is likely to alter natural aquatic biodiversity, food-web complexity and ecosystem function.

The temporal scale of flow and flood pulses and the subsequent ecological responses to altered flow regime will vary with the nature of the hydrogeomorphic patch examined (Thorp et al., 2006). For example, patches associated with constricted versus anastomosing channels should respond to external drivers at different temporal scales depending on the nature of the patch (for examples, see Fig. 1 in Thorp et al., 2008). This could alter, for instance, the diversity and nature of ‘flow habitats’ (Geoffrey Poole, pers. com.), with resulting impacts on species and processes, such as carbon sequestration and denitrification.

Figure 1.

Hierarchical organization of hydrogeomorphic patches within a riverine macrosystem composed of the riverine landscape with its riverscape and floodscape elements. The patches at one scale are nested within the level of organization (L-1 to L-6) above and are affected most substantially by the level below. The grain and extent for sampling at each level are indicated, based on hierarchy theory. The likely most important hydrologic or hydraulic factors are shown for each level, as are the general ecological units of primary focus. Flow pulses may occur multiple times within a year, while above-bank flood pulses (not shown) may be limited to ones occurring every year or so. The flow history records flood and drought events occurring 10–50 or perhaps 10–100 years (depending on the river and ecoregion), while the flow regime reflects longer term flow events.

Aquatic macrosystem domains

Pickett, Kolasa & Jones (2007) emphasised the vital importance of defining research domains through scientific dialogue, and Heffernan et al. (in press) identified this critical step for advancing macroecology. Given that a domain is ‘… the set of objects, relationships and dynamics at specified spatial and temporal scales that are the subject of scientific inquiry’ (Pickett et al., 2007), what are appropriate domains for riverine macroecology? Unfortunately, a single, all-encompassing domain cannot be specified because of the process-dependent nature of macroecology. Instead, domains must be matched to the scientific question at hand.

Spatiotemporal domains

From a spatial-scale perspective, the current focus on subcontinental to continental scales suggests that catchments will most often be the appropriate maximum spatial extent of riverine landscape domains. The catchment would then be impacted from outside most strongly by processes occurring at the next superior level (e.g. regional or subcontinental), including by climatic factors and invasive species. To understand within-catchment processes, one would primarily need to decipher effects at the adjacent subordinate level (Fig. 1) – the valley segment of the riverine landscape (Thorp et al., 2013), as can be delineated statistically using GIS models (Williams et al., 2013).

From a temporal perspective, macroecological domains can vary widely depending on the focal process (geological, climatic-hydrological, or biological; Fig. 1) and the local ecoregion-biome. While most stream ecological studies have focused on general seasonal differences or on flood pulses occurring seasonally or yearly, macroecological river studies would benefit by focusing on much longer time periods. The temporal extent of a river's flow history and longer flow period can vary extensively among biomes, with the latter often varying upwards to 100–500 years or more (Thorp et al., 2008).

Research domains

The domain involving breadth of research questions can be broad and may only partially overlap with historical emphases in aquatic ecology. Most fundamental research in ‘stream ecology’ has until very recently focused on main channel studies of small spatial extent (typically the reach level or smaller) and fine grain, often spanning a year or less (but see Heino, 2009). Research questions have usually involved, for example, new species descriptions, autecological research, response to habitat conditions, population biology, community interactions (e.g. food webs, predator–prey, competition, small-scale patch dynamics) and relative roles of stochastic versus deterministic factors. Ecosystem functional studies (stream metabolism and nutrient spiralling) have until recently been limited almost exclusively to headwater streams, as in the successful LINX projects (Lotic Intersite Nitrogen Experiment). Even the newest STREON research (STReam Experimental and Observatory Network) is an intersite study of mostly headwater streams. Fundamental ecological research on interactions between floodscapes and riverscapes is uncommon (but see Junk, Bayley & Sparks, 1989; Hamilton, Lewis & Sippel, 1992; Junk & Wantzen, 2004).

Applied/fundamental macroecology: influences of global climate change

As discussed previously, riverine macrosystem studies can involve fundamental and applied research questions as well as problems linking both research worlds. Within the realm of terrestrial and aquatic macroecological research, questions related to detrimental effects of climate change are often most prominent (cf., Heffernan et al., in press). Because effects of periodic and directional climate change have such pervasive effects, I briefly discuss here their relationship to riverine macro-ecology.

Climatic shifts over time should influence processes and interactions among and within ecosystems, including, for example, rates of species transfer among basins and continents (Diez et al., 2012). Effects within rivers will be most prominent in altered flow regimes (e.g. Palmer et al., 2008), but changes to temperature and dissolved oxygen will also manifest at multiple scales. Increased temperatures will negatively impact many rivers directly and indirectly because of depleted oxygen levels. Species confined to low temperature, high oxygen conditions in montane streams face a ‘summit trap’ (Sauer et al., 2011) preventing escape to alternate geographical areas other than possibly higher elevations in the local mountainous region; a comparable quandary faces aquatic species in high latitudes. Species living in low elevation, east–west flowing rivers (e.g. many US Great Plains rivers) face a comparable ‘latitudinal trap’ where migration to higher latitude, cooler streams are impeded, leaving them no choice but to adapt physiologically or migrate if possible to unfamiliar but cooler habitats, such as larger and deeper rivers or shallow springs and any shaded headwaters. Unfortunately, other ecological challenges will face them in those novel environments, and the general ecological plasticity of aquatic species is unclear. Altered thermal conditions may also lead to divergence in sex ratios throughout the basin in species with temperature-dependent sex determination, such as some reptiles (cf. Refsnider & Janzen, 2012). The probability of successful species invasions from other basins would also be affected by climate change.

Although temperature-oxygen shifts may be harmful, the more significant climate change problem will probably be altered flow regimes, which are critical drivers of riverine macrosystems often modified by humans (Poff et al., 1997). Increased variability in precipitation should negatively impact rivers by making droughts and flow pulses increasingly erratic and severe. Again, species in east–west flowing rivers in more arid ecoregions are probably to be most severely impacted, especially because of migration barriers. Changes in the magnitude of snow pack and timing of snow melt are also critical for some rivers. At larger spatial extents, access to diverse flow habitats will be less dependable, thereby affecting species diversity and ecosystem function. For example, increased flows could more frequently eliminate anoxic habitats in backwaters of anastomosing rivers, thereby decreasing rates of bacterial denitrification and increasing nutrient spiralling lengths. Fish reproduction by riverine and floodplain specialists could be negatively impacted if flows and floods are more erratic and temporally shifted. Riverscape–floodscape exchanges would be altered for species and detrital matter, and reproductive processes would be interrupted. Climate change could also impact processes typically exhibited at smaller spatial extent, finer grains and shorter temporal periods. For instance, the relative importance of stochastic and deterministic factors in controlling benthic species diversity could change within stream patch mosaics at the reach level and below.

Integrating macroecological approaches with current basin-level concepts

An initial step to advancing metamorphic changes in river science is to integrate, where applicable, the newest macroecological perspective with current large-scale riverine concepts. Three basin-level concepts mostly encompass the breadth of ecological perspectives by emphasising either (i) a theoretical clinal nature of rivers (River Continuum Concept [RCC]; Vannote et al., 1980); (ii) the hypothetical importance of tributary patterns and junctions (Network Dynamics Hypothesis; Benda et al., 2004); or (iii) the proposed hydrogeomorphic patch nature of rivers at multiple scales (as in the Riverine Ecosystem Synthesis [RES]; Thorp et al., 2006, 2008).

In my opinion, an especially useful riverine theory for macrosystem analyses will include at least three criteria: (i) it should encompass an entire catchment (headwaters to river terminus) and apply to all river types; (ii) it will be inherently hierarchical; and (iii) it should account for nonlinear dynamics and emergent properties by focusing on larger-scale processes and not rely on scaling up from data of smaller extent and finer grain.

River continuum concept

Although criticised by many for its hydrogeomorphic foundation (Statzner & Higler, 1985; Perry & Schaeffer, 1987; Poole, 2010) and ecological predictions (Townsend, 1989; Lau, Leung & Dudgeon, 2009), the RCC clearly advanced our understanding of connectedness in river systems and spawned a major shift in how we viewed rivers. However, the RCC is inapplicable as a macroecological model based on the three criteria cited above even though it focuses on headwaters to large river channels. In particular, the RCC does not apply to many dryland rivers, some of whose flow disappears downstream (not meeting criterion #1). More importantly, its emphasis on gradual and continuous longitudinal changes in physical and biological features and resulting lack of hierarchical components (failing criterion #2) means that you should theoretically be able to predict ecological structure and functions at larger spatiotemporal scales (e.g. valley or basin scale) by merely scaling up (contradicting criterion #3) from finer grain and smaller spatial extent data (e.g. reach level), thereby negating the need for macroecological approaches.

Network dynamics hypothesis

The Network dynamics hypothesis (NDH) (Benda et al., 2004) highlights effects of basin size, network shape, drainage density, network geometry, longitudinal position and relative size of intersecting streams on stream bed features, riparian attributes and ecological habitat heterogeneity. The size, spacing and confluence angles of tributaries reflect underlying geologic structure, topography and erosion history, which in turn alter habitat structure of the receiving stream. Regional and climate-change-induced effects alter the network's amount and variability of water flow and sediment supply. The NDH focuses on ecological patterns and effects within channels (the riverscape), but the network structure is directly responsive to basin characteristics. While the temporal dimension is a relatively minor element, a model component explains how stochastic disturbances affect the morphology and age of fluvial habitat features prominent at confluences.

The NDH shows some promise as a basin-level model for macroecological research. It is a universally applicable model (criterion #1) of river channels emphasising a nested, branching hierarchy (criterion #2) as opposed to the linear RCC. It can potentially help decipher how spatial arrangement of tributaries in a network interacts with stochastic catchment processes that influence patterns of habitat heterogeneity – which cannot be determined by scaling up from subordinate hierarchical levels (criterion #3).

Hydrogeomorphic patches models and the riverine ecosystem synthesis

A growing emphasis on the discontinuum nature of rivers (Perry & Schaeffer, 1987; Townsend, 1989; Rice, Greenwood & Joyce, 2001) and the importance of hierarchical patch dynamics in ecosystems in general (Wu & Loucks, 1995) led to development of river concepts emphasising hydrogeomorphic patches at multiple scales (Montgomery, 1999; Poole, 2002). The RES (Thorp et al., 2006, 2008) is, thus far, the most comprehensive treatment of this body of theory and applies to riverine landscapes in all biomes (meeting criterion #1).

The RES asserts that rivers resemble hierarchical (criterion #2) mosaics of patches at varying spatiotemporal scales, with each patch type possessing intrinsic hydrological and geomorphic attributes (Fig. 1). Patches can vary in number and permanency of lateral channels, spatial diversity of current velocities, temporal variability in flow/flood pulse rate and extent, substrate size and variability, riparian-channel interactions and riverscape-floodscape exchanges. Patches occur at multiple sites (repeatable) along the longitudinal path where similar hydrological and geomorphic conditions occur (see Fig. 1 in Thorp et al., 2008). Key, but non-exclusive, focal units are patches at the valley-to-reach scale (termed functional process zones, or FPZs; Thoms et al., 2004) where large-scale hydrogeomorphic changes are prominent. FPZs are only partially predictable in position, especially when comparing among ecoregions. Resulting ecological traits are hierarchical (Fig. 1), thereby allowing integration of hydrological, geomorphic and ecological characters appropriate for the scale of interest. Ecological attributes are more comparable in similar patch types than in adjacent patches of fundamentally different types (cf., Poole, 2002). Because patches are spatially and temporally scaled hierarchically in the RES framework, process occurring at higher scales cannot be ascertained by merely scaling up from subordinate hierarchical levels (matching criterion #3).

Macroecological challenges and future research avenues

Emphasis on four components of the research domain

Although a great diversity of research challenges exist in riverine macroecology, I believe that fundamental riverine macroecology could benefit significantly by increasingly emphasising four components of its research-question domain (ordered by increasing spatial extent): (i) nature and importance of riverscape–floodscape interactions over different temporal scales; (ii) structural and functional importance of hydrogeomorphic patch structure at the valley level; (iii) interactions (e.g. invasive species ecology, meta-community interactions and biogeographical species patterns) between multiple riverine macrosystems over regional to continental scales; and (iv) impacts of global to regional climate change. It would be useful to examine these topics across multiple spatiotemporal scales and where possible examine interacting effects of multiple drivers and feedback loops.

Many fundamental ecological research questions could benefit from a macrosystems approach in rivers. I have listed twelve in Table 1, with most having a link to the RES. The list is not meant to be exhaustive, nor are the questions necessarily ranked by importance or feasibility. Instead, I have included these questions to give all investigators – especially the emerging new generation of scientists – some potential macroecological pathways to explore.

Table 1. Example questions for riverine macrosystem studies
A. External Interactions –Climate Changes to Flow and Temperature
A.1: Does a river basin's directional orientation influence the resilience of biotic diversity to climate change?
A.2: Do river network patterns (e.g. trellis versus dendritic) and associated diversity of thermal habitats and refuges influence fragility of community structure when temperatures rise?
A.3: Does valley-patch complexity in flow and thermal habitats influence the response of community structure (species and trait-based diversity) to increased temperatures and variability in flow and flood pulses?
A.4: Will more erratic and larger flow and flood pulses reduce rates of bacterial denitrification in the riverscape and alter carbon sequestration rates in the floodscape?
B. External Interactions – Inter-Macrosystem Exchanges
B.1: Does a catchment's network pattern influence overall susceptibility to species invasion from other basins and do similarities between the old and potentially new network patterns influence likelihood of invasion success?
B.2: Is there a relationship between network pattern and: (i) frequency of freshwater-marine exchanges (e.g. diadromous fishes); and (ii) food web structure and system metabolism?
B.3: Do the types of valley-scale hydrogeomorphic patches and spatial pattern influence the importance of diadromous species exchanges with marine systems?
C. Intra-Macrosystem Interactions at the Valley-Level of Hydrogeomorphic Patches
C.1: Are community diversity, food web complexity and length, system metabolism and nutrient spiraling length directly proportional to: (i) flow habitat complexity within the riverscape; and (ii) frequency and predictability of riverscape-floodscape connections?
C.2: Are resistance and resilience of community structure and ecosystem function related to habitat complexity of valley-scale patches and network pattern?
C.3: To what degree are biodiversity, food web complexity and ecosystem function in similar patch types affected by longitudinal patch location, proximity to similar patches, longitudinal extent of continuous patch types and nature of intervening patches. [Note: while some types of valley patches are more predictably sited in a basin, location of others are less predictable, especially among ecoregions (Thorp et al., 2006, 2008)].
C.4: Does the importance of a given patch type to biodiversity, food web complexity and ecosystem function vary with network pattern and location?
C.5: Which structural and functional aspects of riverine landscapes can be ascertained most and least effectively by scaling up from comparable attributes at smaller spatiotemporal scales and are these relationships affected by hydrogeomorphic patch type?

Problems accessing and interpreting macroecological data

An immediate problem that will arise when one attempts to answer these or similar large-scale questions is as follows: ‘How comprehensive does the study need to be?’ From a temporal perspective, scientists are limited to their ability to collect macroecological data over long time periods because of short research funding periods (typically 2–5 year) and publication pressures on individuals. One possible solution is to access long-term data sets in museums; for example, stable isotope analyses can provide clues to effects of hydrological and hydrogeomorphic changes in rivers over time and space (e.g. Delong et al., 2011). Another approach is to access data at specific, long-term research sites. These include, for example, data collected on Germany's Breitenbach River from 1969 to at least 2005 by a series of scientists at Max Planck's Limnological River Station Schlitz (e.g. Wagner, 2011) and currently by the Rhein-Main Observatory. Research stations and programmes collecting long-term data include, for example, the Long-Term Ecological Research (LTER) sites in the U.S.A, the expanding International LTER program, the US Long-Term Research Monitoring Program (LTRMP) on the Upper Mississippi and various other multicountry research efforts on several continents to study international rivers such as the Danube and Mekong Rivers.

An inherent problem faced by most of these sites, however, is that the temporal data are generally adequate to very good but the spatial extent data is often severely limited because of the site-specific nature of the studies. This frequently means that scientists are forced to draw conclusions about processes occurring at larger spatial extent by extrapolating data from one small portion of the riverine landscape (typically headwater stream channels) – a scaling-up process fraught with significant dangers, as described earlier. However, the negative impact of scaling up probably varies substantially with the ecological scale of the dependent variable. For example, as one moves up the ecological organisational ladder from organisms to ecosystems, scaling up becomes more problematic and alternative independent and dependent variables are required. More research on scaling up procedures and limitations is needed.

Need for larger-scale research approaches and funding

Although there are common hurdles that must be overcome by researchers working in all macrosystems, riverine macroecologists face some unique challenges. Aside from the depauperate ranks of large river scientists in general, the major challenges are minimal funding for large-scale research and the dearth of scientific groups collecting data at large spatial scales for extended periods. Long-term and large-scale federal research funding in the U.S.A is small compared with funding for lakes and smaller lotic systems, although apparently river research is more adequately funded in Europe. Major ecosystem and macrosystems programmes in the U.S.A, such as NSF's LTER and NEON (National Ecological Observatory Network), have ignored large rivers completely or for at least a quarter century except for studies of small basins with their headwater streams. One logical explanation for the LTER decision may be that current LTER sites have strong place-based foci, which challenges researches to collect data over large basins. This place-based criterion should be modified for rivers by recognising merits of a central organisational hub in an integrated riverine macrosystem LTER with a diffuse, multi-institutional network of data collection stations at strategic catchment points. Such an LTER should be supported by multiple government agencies to maximise use of data to answer both fundamental and applied macroecological questions.

Acknowledgments

Manuscript production benefitted from NSF grants DEB 0953744 and 1249370 and from EPA-ORD through their ‘Intermittent Expert’ program. However, views expressed here do not necessarily reflect views or policies of those agencies. Comments on an earlier draft from Andy Casper, Joe Flotemersch, Jurek Kolasa, Cliff Ochs, Geoff Poole, four of my graduate students (Rachel Bowes, Logan Luce, Brian O'Neill and Sarah Schmidt) and two anonymous reviewers are greatly appreciated.

Ancillary