Large floodplain river restoration in New Zealand: synthesis and critical evaluation to inform restoration planning and research

New Zealand (NZ) has a diversity of large river ecosystems that provide essential ecosystem services but are impaired by multiple ecological impacts. River restoration is an active field worldwide and there is good potential for river restoration practitioners in NZ to draw on lessons from elsewhere, although there is also a need to tailor approaches to national and local contexts. Here, we provide a critical review of large floodplain river restoration to guide environmental management in NZ and inform research and management priorities. The review is structured using a driver-pressure-state-impact-response framework, with a focus on responses, i.e. large river restoration methods. Thirty-one river restoration methods aligned with 14 broad restoration goals were evaluated collaboratively and semi-quantitatively. Evaluation outcomes are presented to inform regional and national scale restoration planning. Recommendations were identified to address eight key knowledge or policy gaps: (1) understanding cumulative impacts facing large river systems, (2) prioritising restoration measures at the landscape-scale, (3) promoting lateral connectivity in large river floodplains, (4) incorporating knowledge of geomorphology into river management, (5) enhancing understanding of cultural priorities and community aspirations, (6) assessing how costs and benefits of river restoration vary among timescales, (7) understanding the feasibility of restoration methods that have received limited application in NZ and (8) improving protection of threatened native fish species.


Introduction
Large rivers and their floodplains provide essential ecosystem services that support a diversity of life and have allowed human civilisations to prosper on fertile alluvial soils (Postel and Carpenter 1997;Schindler et al. 2014). Globally, aquatic ecosystems face multiple stressors that degrade ecological services (Dudgeon et al. 2006;Vörösmarty et al. 2010;Arthington et al. 2016;Reid et al. 2019). Large rivers and their floodplains are particularly threatened, reflecting their status as focal points for human development, as well as lowland receptors for upstream changes to water quality and quantity (Tockner et al. 2008). For example, only 37% of rivers > 1000-km-long remain free-flowing over their entire length (Grill et al. 2019) and most large rivers in North America, Europe, Africa and Australasia are 'strongly affected' by flow regulation (Nilsson et al. 2005). In recent decades, there have been notable ecological improvements in some large rivers in areas such as western Europe (e.g. Lamouroux et al. 2015). However, the rate at which stressors are causing adverse changes is increasing among large rivers in general, requiring increasingly urgent action to avoid crossing resilience thresholds and facing ecosystem collapse (Best 2019;Best and Darby 2020). The applied science of river restoration has emerged to address ecological degradation Communicated by James D. Ford in rivers worldwide and involves human interventions to river channels, riparian zones, floodplains and inflows that are designed to enhance hydrological and biogeochemical processes, and restore environmental elements that have been degraded (Wohl et al. 2005(Wohl et al. , 2015. New Zealand (NZ) is an environmentally diverse island country, 268,000 km 2 in area, with the two largest islands spanning subtropical and temperate zones (34-47°S; Fig. 1). Prominent mountain ranges are present in the centres of the North Island (maximum elevation = 2797 m) and South Island (maximum elevation = 3764 m) that strongly influence climate and headwater characteristics (Winterbourn et al. 1981;Snelder and Biggs 2002). Annual rainfall varies from < 600 mm for parts of the east coast to > 4000 mm on the west coast of the South Island (NIWA 2012). Dominant land uses by area are pastoral grassland (39%), indigenous forest (23%) and exotic forest (7%), whereas urban areas account for only ~ 1% of the land area (Landcare Research 2020).
There are 27 large river systems in NZ ( Fig. 1, Table 1) classified as stream order 7 or greater (Johnson et al. 1995), based on the 2010 NZ River Environment Classification (Snelder et al. 2010). Here, we expand our focus to include 6th order rivers to reflect that these larger medium-sized rivers include numerous nationally important waterways with large floodplains and important socio-ecological values. Rivers with order ≥ 6 comprise ~ 9000 river km (Table 1) and vary substantially with geology and land cover ( Table 2).  Table 1 The frequency and length of river systems in New Zealand with stream order ≥ 6. Values are based on analysis of the number of rivers that drain to the ocean, and the associated length of river reaches, using the River Environment Classification (Snelder et al. 2005) † Based on rivers that drain to the ocean. Thus, tributaries are not counted separately, even if they have stream order of ≥ 6 Dominant geologies underlying medium-large river systems include hard sedimentary (47%), which characterises many southern and central South Island rivers such as central reaches of the Clutha River (8th order). Volcanic acidic geology (25%) characterises central North Island rivers such as the upper reaches of the Waikato River (8th order), whereas alluvial geology (7%) characterises the lower reaches of many braided rivers, such as the Rakaia River (7th order) and the Waitaki River (8th order) on the east coast of the South Island. Land adjacent to large rivers comprises a higher proportion of pastoral land cover (50%) and a lower proportion of indigenous forest (23%; Table 2) than in NZ generally, reflecting the lowland setting of large rivers.
As is the case for developed regions generally, lotic ecosystems face considerable pressures in NZ, and there is growing demand there for enhanced restoration to support adaption to environmental change and address degraded ecological values (Gluckman 2017; Ministry for the Environment and Stats NZ 2020). Proponents of enhanced restoration include indigenous Māori, who highlight ongoing impacts of freshwater degradation to their traditional values and ability to enact traditional customary practices (Te Aho 2019; Stewart-Harawira 2020). The increasing emphasis on river restoration in NZ reflects greater awareness of the fundamental importance of freshwater and its protection; this is a component of Te Mana o te Wai, which was recently adopted as a concept that underpins national freshwater management policy (Ministry for the Environment 2020).
There is extensive experience of undertaking river restoration in developed northern temperate regions (e.g. see reviews by Wohl et al. (2005Wohl et al. ( , 2015) but examples of large river restoration projects and associated literature are more limited for ecosystems in the Southern Hemisphere. While river restoration is an active field in NZ (e.g. Caruso 2006), restoration projects have tended to focus on smaller rivers or upland areas, and there is a need for greater focus on the restoration of large floodplain rivers, albeit with a catchment-scale (mountains to the sea) perspective. Designing and implementing appropriate restoration actions for large rivers is particularly challenging due to multiple interacting stressors (Wohl et al. 2015), stakeholders with conflicting viewpoints or aspirations (McLain and Lee 1996), constraints imposed by upstream flow regulation (Palmer and Ruhi 2019), the requirement to undertake habitat restoration across large areas (Schmutz et al. 2014), the importance of considering socio-economic aspects in densely populated lowland floodplains (Woolsey et al. 2007) and challenges in measuring success . When planning large river restoration projects in NZ, there are benefits to drawing on lessons learned elsewhere, but there is uncertainty about how transferable approaches used overseas are to catchments in NZ with different physical, sociocultural and ecological contexts. While larger rivers in NZ have much in common with other rivers at similar latitudes, there are several unique or unusual biophysical and sociocultural characteristics within or among classes of NZ rivers. These features mean it is appropriate to screen or adapt restoration methods developed and applied elsewhere before applying them in NZ. Key biophysical and sociocultural characteristics of NZ rivers are discussed in the Supplementary Information, summarised as follows: • Morphologically, NZ's island status and physical geography limit the maximum size of lowland rivers relative to continental regions such as mainland Europe or North America. • Ecologically, NZ has a unique fish fauna characterised by high levels of endemism, particularly among nonmigratory species (McDowall 2010). Over one-third of NZ's native fish species are diadromous, i.e. they have a life history that involves migration between marine and freshwater environments (McDowall 1998). Life history characteristics of native NZ fishes generally differ from those of fish species in the northern hemisphere and require careful consideration during restoration planning. • Māori are the indigenous people of NZ and recognising values of tangata whenua (people of the land connected to a place through ancestral linkages; Harmsworth et al. 2016) and associated traditional perspectives is critical for freshwater management in NZ (Brierley et al. 2019; Ministry for the Environment 2020). Integration of indigenous knowledge can enhance river restoration projects (Fox et al. 2017;Collier 2017), and acknowledg- ing and respecting the value that indigenous worldviews and knowledge bring, and integrating socio-ecological values into river restoration, are imperative for restoration practitioners in NZ and elsewhere . Such integration needs to consider local contexts (Collier 2017) and the potential for co-governance and co-management arrangements to support recognition of Māori rights and interests (Fisher and Parsons 2020). • Land development and related pressures on aquatic ecosystems have increased in NZ since European colonisation in the mid-1800s and have therefore occurred relatively recently when compared to temperate regions in the northern hemisphere (King 2003). This has led to commonalities in land development pressures throughout NZ catchments (discussed further below).
Multiple studies have evaluated river restoration methods either generally or with focus on regions in the northern hemisphere (e.g. Buijse et al. 2002;Bernhardt et al. 2005;Wohl et al. 2005Wohl et al. , 2015Lamouroux et al. 2015;Mondal and Patel 2018;Erős et al. 2019;Palmer and Ruhi 2019). Considering the above, there is need for a critical review of large floodplain river restoration to guide environmental management in NZ and to inform the strategic direction of research nationally. Accordingly, this review addresses the following research questions in a NZ context: 1) What are the main threats to the ecological values of large floodplain rivers? 2) What methods are available to restore large floodplain rivers? 3) What is the potential for restoration methods to enhance large floodplain river ecosystems? 4) What are key research priorities to support improved river restoration outcomes?

Methods
The review focuses on larger medium-sized and large rivers, defined as rivers with stream order ≥ 6 (Johnson et al. 1995), although much information is relevant to smaller rivers. Furthermore, the review focuses on mainstem and adjacent floodplain habitats in large rivers, rather than smaller tributaries in upland areas; this reflects the key research gap identified, although we recognise that successful river restoration requires a catchment-scale approach (Fausch et al. 2002;Wohl et al. 2005). This review is structured based on a driver-pressure-stateimpact-response (DPSIR) framework, which is an effective way to present information to support environmental management and decision-making (OECD 2003;Ness et al. 2010;Tscherning et al. 2012). This suitability of the DPSIR framework applies to floodplain river management (Schindler et al. 2016), for which the DPSIR framework is wellsuited for parsing the complex causal interactions that apply to large river catchments. All components of the DPSIR framework were reviewed, although the focus here is on responses, i.e. methods to restore large floodplain rivers in NZ. Consequently, detailed information about other components of the DPSIR framework is presented in Supplementary Information. All applicable river types were considered, although we attempted to identify whether responses were more relevant to certain river types. Braided rivers were included in the review; however, we recognise that braided rivers have distinctive physical and ecological characteristics that can warrant unique management approaches (Piégay et al. 2006) and this review is not intended to be a comprehensive treatment of braided river restoration.
Literature concerning drivers, pressures, states, and impacts was first compiled and reviewed (research question 1). Key river restoration methods were then identified based on the literature Hornung et al. 2019) to address research question 2. Next, evaluation criteria were developed by researchers with experience of ecological restoration in NZ. These criteria were then used to evaluate river restoration methods to assess the potential for each approach to address the impacts identified (research question 3). Evaluation was undertaken at a workshop in December 2020 attended by 11 scientists and environmental managers who have experience of freshwater restoration and are predominantly affiliated with academic and government organisations in NZ (see author list and acknowledgements). Relative costs and the timescale to achieve benefits were also assessed for individual methods. Discussions at the workshop led to identification of key uncertainties and research priorities (research question 4).

Drivers: root causes of impacts
Drivers are the social, economic or environmental developments that exert pressures on the environment (Tscherning et al. 2012) and may therefore be considered the root causes of ecological impacts (Fig. 2). Key drivers of ecological changes to large floodplain rivers are agricultural intensification (including changing land cover to agricultural land), climate change, flood protection, hydropower, invasion of non-indigenous species, and urbanisation (Joy and Death 2013;Matthaei and Piggott 2019;Collier et al. 2019), as described in Table S1. Matthaei and Piggott (2019) identify the following key pressures to rivers in NZ: water contaminants; changes to flow regimes associated with abstraction, climate change, urbanisation or flow regulation; competition by non-indigenous species with native biota; and warming of rivers. In addition, floodplain disconnection is a major pressure in large lowland rivers generally (Tockner et al. 2010), affecting several large rivers in NZ due to riverbank modifications, drainage or channel alterations caused by sand extraction (e.g. Collier et al. 2019). Changes to sediment supply can have major effects on geomorphology, particularly in association with dam construction, which reduces coarse sediment supply and can lead to bed armouring and channel incision downstream (Tonkin and Death 2014). Fish passage obstruction due to impoundments such as dams, weirs or floodgates is also a pressure in many NZ river catchments (Allibone 1999). Key pressures (Fig. 2) are described further in Table S2.

Pressures: mechanisms that cause impacts
Pressures can interact in synergistic, additive or antagonistic ways. In NZ, such interactions have not been widely studied for large floodplain rivers (but see Collier et al. 2019), although work undertaken in streams has highlighted the potential for interactions between water contaminants and increased water temperatures to affect invertebrate and periphyton communities in complex ways, with synergistic interactions possible (Piggott et al. 2012(Piggott et al. , 2015. For large rivers in the North Island such as the Waikato River, water contaminants and climate change are expected to interact synergistically to enhance the colonisation of invasive fish species such as cyprinids and Gambusia affinis that are tolerant of poor water quality and increased water temperatures (Collier et al. 2015;Pingram et al. 2021).

States: ecosystem services provided by large floodplain rivers
Ecosystem services can be defined as the goods and services provided by natural ecosystems to sustain human life (Daily 1997). In the context of the DPSIR framework, 'states' can be defined in different ways (e.g. Ness et al. 2010); however, for this review, we contend that states can be considered to be the quality of the ecosystem services that large floodplain rivers provide.
Large floodplain rivers provide a range of ecosystem services that can be broadly grouped into three categories: (1) provisioning, (2) regulating and maintenance and (3) cultural services (Schindler et al. 2014). Key ecosystem services provided by large floodplain rivers in NZ (Fig. 2) are described in Table S3, which focuses on the services that are generally most associated with large rivers and are often the focus of restoration. Large rivers provide other

Responses
Aesthetics/educationremove litter, education (e.g., signage, school groups) Bank stabilisationhard engineering, bioengineering Enhance fish passage -↑ fish passage, fish pumps Environmental planningfloodplain protection Fisheries managementfish stocking, quotas, translocation Floodplain reconnectiondike removal, diversion or openings Enhance flow regime -↓ abstraction, manage reservoir releases, sustainable drainage systems, restore peat/soil C Instream habitat creationchannel reconstruction/realignment, side arms Instream habitat improvement -↑ gravel, ↑ structure Invasive species controlcontrol riparian/aquatic invasive plants and pest fish Restore free-flowing riverdam/weir removal Riparian/floodplain improvementwetland creation, riparian fencing and stock exclusion, riparian/floodplain planting, water level management Water quality management -WWTP improvements, ↓ agricultural loads, urban stormwater management Thermal managementmanage reservoir withdrawals Fig. 2 Summary of drivers, pressures, states, impacts and responses applicable to large floodplain river restoration in New Zealand ecosystem services such as maintenance of soil quality and atmospheric regulation (Schindler et al. 2014); these are generally sensitive to similar pressures as the values listed, and they may be enhanced by common restoration approaches. Ecosystem services are aligned with the three broad categories listed above, although some services align with two categories, e.g. mahinga kai-defined as traditional resource harvesting (Phillips et al. 2016) and freshwater species traditionally used as food or another resource (Ministry for the Environment 2020)-provides nutrition (a provisioning service), as well as fulfills social, cultural and spiritual needs (a cultural service) (King et al. 2013).

Impacts: effects to large floodplain river ecosystems
There is wide variability in the extent to which ecological services provided by large floodplain river ecosystems in NZ have been affected by pressures. Numerous large rivers such as the Buller (7th order) and Rakaia are renowned for their 'wild and scenic' characteristics, protected by statutory Water Conservation Orders that are designed to preserve their 'outstanding' status (Hughey et al. 2014). By contrast, other rivers have suffered extensive ecological degradation due to multiple pressures, as is particularly the case for several rivers in the North Island such as the Whanganui and the Waikato (Gluckman 2017;Brierley et al. 2019;Collier et al. 2019). Despite this variability, to some degree all large rivers in NZ face a common set of key impacts (Fig. 2), described in Table S4: loss of biodiversity, water quality decline, reduced production of mahinga kai species, increased flood risk, loss of spiritual values, loss of recreational opportunities and reduced scenic values. The interactions between pressures and impacts to aquatic ecosystems are particularly well-researched in NZ (Matthaei and Piggott 2019). Notably, there have been several studies that describe the decline in endemic freshwater fish abundance and distribution (e.g. Weeks et al. 2016;Joy et al. 2019), as well as trends in river water quality (e.g. Ballantine and Davies-Colley 2014;Larned et al. 2016). Nonetheless, there are several key uncertainties, especially in relation to how impacts to large rivers compare with impacts to other freshwater ecosystems. Furthermore, there has been limited study of cumulative impacts on NZ rivers, including the potential for impacts to interact to undermine the resilience of large river ecosystems, as well as to initiate regime shifts that result in accelerated decline in ecosystem services (Tockner et al. 2010;Angeler et al. 2014;Grantham et al. 2019). These gaps partly reflect shortcomings in understanding reference (natural baseline) conditions in medium-large NZ rivers, as well as gaps in understanding of the causal links between land use/ land change and ecological impacts (Larned et al. 2020).

Responses: large floodplain river restoration methods
In total, 31 restoration methods were identified, aligned with 14 restoration goals ( Fig. 2; Table 3). For each method, NZ context and examples were discussed and identified during workshop discussions (Table 3). Key themes of these discussions included: • The unique importance to freshwater management in NZ of the concept of Te Mana o te Wai, which refers to restoring and protecting the integrity of water (Te Aho 2019), and is a fundamental concept that underpins national policy (Ministry for the Environment 2020); • The importance of considering how the function of a river restoration project changes through time, recognising that a floodplain is a time-dependent concept (Junk et al. 1989), and floodplains continue to respond to historical changes. Certain restoration actions can provide short-term benefits but they can also 'fossilise' morphology and prevent geomorphic changes necessary for floodplain functioning (Biron et al. 2014). A pertinent example is riparian planting, which has long been promoted as a stream restoration tool in NZ (e.g. McKergow et al. 2016), but can be incompatible with the need to leave space for lateral channel migration in floodplains (Biron et al. 2014) in areas where introduced willow (Salix spp.) and poplar (Populus spp.) species are used, which are superior to native riparian vegetation for stabilising river banks (Phillips and Daly 2008). The need for wide riparian buffers and more natural floodways is now being considered for floodplain management in regions such as Wellington (Death 2018); • Increasing severity of flooding in NZ due to climate change (Table S1) necessitates river restoration projects that increase lateral connectivity to provide greater floodwater storage, while also providing ecological enhancements (Hutchings et al. 2019); • The fundamental importance of providing 'room for the river' to support effective floodplain functioning has been established internationally (e.g. Buffin-Bélanger et al. 2015) and there are multiple examples in NZ of river restoration projects that increase lateral connectivity (Table 3). Nonetheless, the importance of promoting lateral connectivity has not been well recognised in NZ river management policy. Managing coordination among multiple landowners and stakeholders in large river floodplains can be a challenge to enhancing lateral connectivity, e.g. as experienced in systems such as the lower Rakaia River. However, this challenge is not unique to NZ (e.g. Hand et al. 2018).  Reduce sediment and nutrient loads from agriculture ↓ agricultural loads Widely promoted to manage nutrient inputs from agriculture, which is a major pressure facing NZ rivers (Table S2) The potential for each restoration method to address key impacts (Table S4) was qualitatively assessed as either 'no potential', 'low to moderate' or 'high'. River restoration methods were then further evaluated based on six criteria ( Table 4) that encompassed effectiveness, achievability and sustainability. The potential for implementation to be hindered by socio-political impediments was considered (criterion A2; Table 4) but criteria to evaluate social or cultural benefits of river restoration methods were not included, although social (Druschke and Hychka 2015) and cultural (see above) aspects are nonetheless critical considerations to ensure successful restoration outcomes. These issues were not included in the scope of evaluation because it was recognised that social and cultural considerations and priorities can vary widely among projects and are not necessarily specific to restoration methods. Furthermore, there was uncertainty regarding cultural priorities, reflecting a knowledge gap among workshop attendees. For each method, overall confidence in the evaluation scores was also assessed to help identify knowledge gaps.
Expected outcomes and evaluation scores varied among restoration methods, as summarised in Table 5, which provides a reference for NZ river restoration practitioners (individual scores are reported in Table S5). Some restoration methods (e.g. wastewater treatment plants) address a narrow range of impacts, whereas others (e.g. wetland creation) address multiple impacts (Table 5); clearly, the performance and scope of restoration methods are important to consider during restoration planning. Furthermore, cost and the timescale to achieve benefits were identified as crucial considerations as summarised in Fig. 3, which provides an additional tool to support NZ river restoration practitioners in identifying appropriate restoration methods.
Given that all large river systems in NZ face a range of impacts, Table 5 highlights that a suite of restoration methods is generally required to achieve multiple objectives. When planning restoration at the catchment scale, factors such as spatial scale, economic cost, timescales of restoration trajectories, sustainability, expected benefits and sociocultural values need to be strategically evaluated to ensure successful and enduring outcomes. Detailed spatial planning and analysis are therefore required to optimise where and when restoration actions are implemented. To this end, it was identified that there is a need to enhance large river restoration planning in NZ by considering tools developed and successfully applied elsewhere to screen the suitability of restoration action and guide strategic planning (Guida et al. 2015(Guida et al. , 2016Remo et al. 2017). Outputs from such tools can potentially inform participatory processes that allow managers, tangata whenua and stakeholders to explore trade-offs among potentially conflicting objectives, such as restoring ecosystem services and protecting assets from floods (Halbe et al. 2018).

Conclusion
This review has described drivers, pressures, states, impacts and responses relevant to large floodplain river restoration in NZ (Fig. 2). As such, this review can inform regional and national scale restoration planning by clarifying the key threats to the ecological values of large floodplain rivers in NZ (Research Question 1), as well as identifying restoration methods available to restore large floodplain rivers (Research Question 2), aligned with 14 broad restoration goals (Table 3, Fig. 2). The potential for restoration methods to enhance large floodplain river ecosystems in NZ (Research Question 3) varies depending on the impacts facing a catchment (Table 5), and there are marked differences among methods in the potential for landscape scale and transformative benefits to occur, as well as the timescale and response trajectory to achieve improvements (Fig. 3). The variability among restoration methods in factors such as scope and timescales, as well as the crucial importance of recognising sociocultural values (Harmsworth and Awatere 2013), underline the well-established importance of adopting a catchment-based approach to restoration planning (Fausch et al. 2002;Wohl et al. 2005). Outcomes of our evaluation  Fig. 3) are intended to support catchment-scale restoration planning, potentially supported by applying spatial planning tools such as those developed in NZ (e.g. spatial conservation prioritisation software applied by West et al. 2019) or elsewhere (e.g. Erős and Bányai 2020) to inform decision-making. We expect therefore that our research can directly inform work by river restoration practitioners in NZ. We also hope our work can benefit practitioners in other jurisdictions, including regions such as temperate areas in South America, Australia and British Columbia (Canada) Benefits are localised to the immediate vicinity of project site. Benefits are either temporary or restore a narrow range of ecological processes and ecosystem services 2 Benefits are expected to be substantial but are confined to the project site (reach level or smaller) 3 Benefits extend beyond the site level but are confined to a portion of a sub-catchment and restore a narrow range of ecological processes and ecosystem services 4 Benefits extend beyond the site level and provide moderate benefits to multiple ecological processes and ecosystem services 5 Method involves landscape-scale interventions to greatly enhance a wide range of ecological processes and the provision of multiple ecosystem services Achievability A1: Transferability 00 Unknown Is the method appropriate for NZ ecosystems? 0 Method is inappropriate for NZ due to biophysical or social-cultural incompatibility 1 Method would require major adaptation to be suitable for NZ ecosystems and/or is rarely applicable to NZ rivers 2 Method would require minor adaptation to be suitable for NZ ecosystems and/or is applicable to most large rivers in NZ 3 Method is well suited throughout NZ large river ecosystems A2: Socio-political acceptability 00 Unknown Are there socio-political impediments to acceptance? 0 Major socio-political impediments to implementation; obtaining regulatory and social approval is highly challenging. E.g., method presents a structural risk to major infrastructure or substantially impairs the livelihoods of communities.
1 Moderate socio-political impediments to implementation; obtaining approval is uncertain 2 Minor socio-political impediments to implementation that can be addressed with planning 3 The method is widely supported by stakeholders A3: Certainty 00 Unknown How certain is it that the method will provide the ecological benefits that are expected if it is implemented? 0 High probability that the method will not provide ecological benefits 1 The method will likely provide some ecological benefits but could fail 2 The method will provide ecological benefits, although there is a risk that ecological outcomes could be poorer than expected 3 The method will fully provide the expected ecological benefits Sustainability S1: Sustainability/longevity 00 Unknown Will the method provide long-term ecological benefits with minimal requirement for maintenance or further input? 0 Method is short term (e.g. < 1 year) and needs to be continually repeated to provide ongoing benefits 1 Method is medium-term (e.g. 1-10 years) and may require maintenance 2 Method is long-term (e.g. 10 + years) but requires ongoing (e.g. annual) maintenance  Table 5 Evaluation of restoration methods relevant to large floodplain rivers in New Zealand. Shading under 'evaluation' is based on the average of scores assigned to multiple criteria (Table 4); see Table S5 for individual scores Fig. 3 Goals, time scale, relative costs and ecological benefits (based on 'effectiveness' defined in Table 4 and presented in Table 5) for restoration methods for large floodplain rivers in New Zealand Page 11 of 17 18 Regional Environmental Change (2023) 23:18  (Kopf et al. 2015) 1. Undertake research to better understand how impacts facing medium-large rivers in NZ combine and interact in the face of current and future pressures 2. Further develop benchmarks to guide restoration goals that reflect natural variability and the legacy of historical impacts that constrain the feasibility of achieving restoration endpoints Screening restoration methods and prioritising where and when they should be implemented to maximise effectiveness Spatial analysis frameworks developed elsewhere (e.g. Remo et al. 2017) could be applied in NZ to strategically screen the suitability of restoration methods identified in Table 5 to enhance restoration planning at the catchment scale. Such tools can support with evaluating trade-offs and synergies among objectives, such as enhancing lateral connectivity of riverine ecosystems and protecting assets such as farmland from flooding Outcomes of spatial analysis can potentially inform participatory planning processes such as structured decision-making. Structured decision-making is a tool that has been used for catchment planning in regions such as western Canada by providing a rigorous, transparent and values-based approach to explore restoration options, and to guide restoration planning as part of a framework that integrates scientific knowledge with meaningful consultation (Gregory et al. 2012;Martin et al. 2018) 3. Further develop and apply spatial analysis tools to identify and prioritise effective and integrated approaches for landscape-scale restoration in floodplains Lateral connectivity The ecological functioning of most large floodplain rivers in NZ has been adversely affected by floodplain disconnection; however, action to restore lateral connectivity has been limited nationwide, reflecting that the importance of promoting and maintaining lateral connectivity has not been well recognised in NZ environmental policy. The need to maintain floodplain storage is made more pressing by the increasing severity of flooding due to climate change. Spatial analysis tools (see above) could assist with evaluating options to enhance lateral connectivity in floodplains. Benefits for multiple ecosystems services (e.g. biodiversity, flood protection, mahinga kai) need to be highlighted. Improved technology (e.g. automated floodgates coupled with hydrological sensors) supported by ecohydrological modelling could potentially be used to optimise flood management infrastructure to maximise environmental benefits (e.g. providing lateral connectivity during key life stage periods), without increasing flood risk (see Pingram et al. 2021) 4. Further integrate the need to promote and maintain lateral connectivity into NZ environmental and infrastructure policy Geomorphology Knowledge of river geomorphology is poorly incorporated into river management policy in NZ. Sediment transport dynamics are inadequately considered in relation to managing the effects of impoundments and other infrastructure on geomorphology. Understanding of linkages between geomorphology and biology is also generally lacking Balancing sediment supply and stream erosive power can stabilise channels and reduce risks to infrastructure. Achieving that balance requires understanding natural features that reduce erosive power (e.g. meanders, increasing floodplain storage) to better protect infrastructure outside the floodplain 5. Develop catchment-scale knowledge of geomorphic controls on river character and behaviour, e.g. by undertaking detailed sediment budgeting to assess potential for channel adjustment, or by undertaking assessments, such as those based on the River Styles Framework (Brierley and Fryirs 2005;Wheeler et al. 2022) to evaluate geomorphic resilience and inform efforts to enhance lateral connectivity and restore floodplain habitats 6. Develop consistent approaches, guidelines and/or policy nationally to improve consideration of sediment transport dynamics during resource consent processes, including for managing the effects of impoundments and other instream structures 7. Develop and refine integrated approaches that seek to balance sediment supply and stream erosive power, while still protecting key infrastructure Cultural priorities Cultural priorities can vary widely among projects and are important considerations during restoration planning. Nonetheless, the participants in this study had poor understanding of the cultural priorities that potentially apply to restoration methods. This limitation reflects that cultural priorities regarding large river restoration are not generally well understood by the river restoration community in NZ In addressing this knowledge gap, it is necessary to recognise that restoration practitioners have a duty to engage in restoration that respects the intellectual property and data sovereignty rights that apply to traditional ecological knowledge (Robinson et al. 2021). Furthermore, restoration practitioners need to recognise that existing approaches to river management that rely on engineering interventions can reflect historical marginalisation of Māori values (Parsons et al. 2019). Structured decision-making (Failing et al. 2013) potentially provides a tool to better integrate cultural priorities into restoration planning 8. Develop better understanding of cultural priorities for river restoration among the general restoration community through ethical engagement and by engaging and involving mana whenua at the earliest stages of river restoration information gathering, planning, and goal setting that share commonalities with NZ such as a rapid yet relatively recent history of land development, rich indigenous cultures, prevalence of diadromy among fish communities and dependence on hydroelectric power. This work has highlighted several gaps or shortcomings that could be addressed with research or new policy to improve river restoration outcomes (research question 3). Broadly, we identified 11 recommendations to address the following eight key knowledge or policy gaps: (1) understanding cumulative impacts facing large river systems, (2) prioritising restoration measures at the landscape-scale, (3) promoting lateral connectivity in large river floodplains, (4) incorporating knowledge of geomorphology into river management policy, (5) enhancing understanding of cultural priorities and community aspirations, (6) assessing how costs and benefits of river restoration vary among timescales, (7) understanding the feasibility of restoration methods that have received limited application in NZ and (8) improving protection of threatened native fish species (Table 6). Recommendations and research priorities (Table 6) are intended to provide direction for researchers and policy makers, although they are not intended to be exhaustive. In particular, the recommendations and research priorities identified here reflect the experiences and worldviews of workshop participants. Although we sought to include experts with a breadth of experience, we recognise that participants' backgrounds were grounded in conventional science and there are major benefits to better including Indigenous knowledge systems into river restoration (Harmsworth et al. 2016;Fox et al. 2017;Wilkinson et al. 2020). We also recognise that we predominantly focused on fish and aquatic ecology, reflecting the authors' experience; however, additional research that further evaluates restoration from the perspective of other important aspects of biodiversity such as wildfowl and riparian flora is important.
Acknowledgements Kevin Collier participated in the workshop and we are grateful for his insights. We thank an anonymous reviewer for providing constructive comments.
Funding This review was funded by Waikato Regional Council (New Zealand).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The costs and timescales associated with restoration methods vary substantially (Fig. 3) and there is poor understanding of how costs and benefits of river restoration vary among timescales. Accounting for such variability is challenging during restoration planning, which often fails to sufficiently consider long (multi-decadal) timescales that are required to achieve benefits from some restoration methods such as increasing soil carbon in peatlands or reducing agricultural nutrient loads. Furthermore, evaluating long timescales is important to appreciate the potential for restoration methods to 'lock-in' geomorphological or management regimes for decades, potentially constraining future ecosystem functioning or restoration options 9. Assess restoration outcomes over a range of time scales (months, years, decades, centuries) during restoration planning. Such assessment may be qualitative or involve quantifying economic and environmental costs/benefits over different periods

Feasibility of restoration methods not widely applied in NZ
There is uncertainty about the feasibility of several restoration methods to restore large floodplain rivers in NZ-see methods with low 'certainty' scores in Table 5. Such uncertainty reflects a lack of monitoring and assessment of the application of these methods in contexts relevant to large floodplain rivers in NZ. Examples of such methods for which certainty is low include restoring soil organic carbon to benefit ecosystems by modifying hydrology, as well as managing the depth of reservoir releases to enhance downstream thermal regimes 10. Undertake research to better understand the feasibility of restoration methods that have received limited application in NZ but could provide valuable restoration outcomes Statutory protection of native fish Despite the documented decline in vulnerable endemic freshwater fish populations in NZ (Joy and Death 2013;Joy et al. 2019) the extinct NZ grayling (Prototroctes oxyrhynchus) is the only freshwater fish species with full legal protection (Miskelly 2016). This lack of legal protection contrasts with many other jurisdictions overseas where a range of legal instruments are employed to protect threatened fish populations (e.g. threatened fish species are afforded specific legal protection, or reserves are created to protect freshwater biodiversity), traversing multiple levels of government, and involving several agencies (e.g. Collares-Pereira and Cowx 2004). An example of a locally targeted solution is the introduction of bylaws to support the eel fishery in the Waikato-Tainui Fisheries Area, under the Waikato-Tainui (Waikato River Fisheries) Regulations 2011. These bylaws prohibit the taking of migrant female longfin eel, set more protective size limits and implement seasonal closures of priority catchments during eel migration 11. Collaboratively identify mechanisms to enhance protection of native fish populations. Options to consider include additional legal protection, incentives to manage problematic introduced species, additional protection for catchments with threatened populations, and prioritising restoration of critical habitats (including lateral habitats) to enhance recruitment, especially for species with migratory requirements