Environmental Flows and Abstraction Reform in the UK

www.gov.uk/defra

Environmental flows and abstraction reform in the UK

January 2013

A report of research carried out by Centre for Ecology and Hydrology, on behalf of the Department for Environment, Food and Rural Affairs

Authors: Acreman, M.C, Old,G and Laize, C. (Eds)

With contributions from many UK stakeholders (see Annex 1) plus Jackie King (south Africa), Eloise Kendy (USA) and Ian Overton (Australia)

The views expressed in this report are not necessarily those of the authors Report to Defra.

Publication organisation Department for Environment, food and Rural Affairs Flood Risk Management Division, Noble House, 17 Smith Square London SW1P 3JR

Copyright in the typographical arrangement and design rests with the Crown. This publication (excluding the logo) may be reproduced free of charge in any format or medium provided that it is reproduced accurately and not used in a misleading context. The material must be acknowledged as Crown copyright with the title and source of the publication specified. The views expressed in this document are not necessarily those of Defra. Its officers, servants or agents accept no liability whatsoever for any loss or damage arising from the interpretation or use of the information, or reliance on views contained herein.

Summary

Existing abstractions of watermean that many rivers are already defined as over-abstracted, some with flow related impacted aquatic systems and water is often not available for new abstractions when it is needed. This may be suppressing economic growth or affecting other policies, such as location of new housing or industries. The government Water White Paper calls for reform of water abstraction licensing and the Department of Environment Food and Rural Affairs is seeking a more flexible and responsive means of licensing, within a risk- based framework. At present knowledge does not exist to make anything more than broad statements about flow needs of river ecosystems. The Environment Agency has an existing procedure through its Catchment Abstraction Management Strategies (CAMS) that include the definition of Environmental Flow Indicators (EFIs). However, the suitability of EFIs and the underlying science to support reform has been questioned.

The overall goal of the work programme is to develop, through several phases, an improved environmental flow method. This report is focused on the first phase which involves undertaking a scoping study. The four objectives of Phase 1 are to: (1) collate stakeholder views on the pros and cons of current methods and expectations of the future; (2) review methods used in three countries and their relevance to England and Wales; (3) identify options for potential evolution of the current EFI method or possible new methods; and (4) recommend activities for Phase 2.

HERE WE WILL SUMMARISE THE KEY FINDINGS OF THE FOUR PHASE 1 OBJECTIVES LISTED ABOVE.

Objectives 1 to 3 were met by considering the following:

 Revision or replacement?  Holy grail or pointless?  Broad eco-hydrological approach  FDC vs sequencing of flows  Heavily Modified Water Bodies and natural flow?  Standards for protected areas  General vs local  Multiple stressors  Thresholds  Resilience and recovery  Baseline conditions and env change  Knowledge and uncertainty  National accounting  Maximum thresholds  Adaptive management  Implementation  National skills in hydro-ecology

Objective 4: Summarise activities for Phase 2. This includes Project A: Hydro-ecological understanding and Project B: Implementation.

Background

Many view the UK as that green and pleasant land where rainfall is plentiful and rivers flow naturally. However, there is great spatial and temporal variation in water availability and on a global scale of water stress the Thames catchment is equivalent to Ethiopia. Water availability also varies over time with periodic droughts and floods. Water resources need careful management to ensure the needs of abstractors are met for public supply, growing food, supporting industry and generating power, whilst conserving the environment and its benefits to people. This is required under various policies and national and European legislation. The European Water Framework Directive (WFD) aims to achieve good ecological status in all water bodies, the European Habitats and Birds Directives (HBD) require designated sites to be in favourable condition, Sites of Special Scientific Interest need to be in favourable condition and the National Ecosystem Assessment (NEA) demonstrates how functioning ecosystems deliver services that support human well-being and contribute to the national economy. All of these requirements may not be achievable; indeed some may be in conflict. For example the River Itchen is designated under the Habitats Directive for its largely man-made habitats, whilst the WFD seeks to return those habitats closer to natural conditions. Furthermore many ecosystem services result from anthropogenic manipulation of the environment. Nevertheless, there is a clear desire to protect and improve the environment and make it a better place for people and wildlife (the role of the Environment Agency (EA)) and it is widely recognised that river ecosystems requires sufficient water flow to achieve this. The terms environmental flows is now widely used to describe the flow regime required by a river to maintain its ecosystem and the benefits they provide to people. The challenge is to define what we want to maintain and what flow regime is required to maintain it.

Many rivers are failing to meet Good Ecological Status (GES), a key target under the WFD. In England and Wales only 28% of surface water bodies (rivers, lakes, transitional and coastal waters) are at good ecological status/potential or better1. In many cases this may be accounted for by hydrological pressures: 15% of river water bodies (14% in England, 1% in Wales) have a flow regime that may not support GES (this includes 4% in flow regulated rivers2). Failure to meet GES may also result from poor water quality, and past and ongoing degradation of river habitats from channelisation and poor land management. However, in the vast majority of cases nothing has been reported by Agency staff (Entec, 2005), in the few cases reported approximately 80% of the limitations relate to water quality issues. The direct link between water quantity and quality through dilution is significant here. In addition, river ecosystems may not be adjusted to current conditions because of slow reaction times, so the actions of recent management are not detected.

Existing abstraction licences mean that many rivers are already defined as over-abstracted and water for new licences is not available. This may be suppressing economic growth or affecting other policies, such as location of new housing or industries. Pressure on water resources is likely to increase. There is growing demand for water as our population increases, more people chose to live alone; climate change may lead to warmer wetter winters, but hotter and drier summers, potentially reducing availability and increasing demand at critical periods. Future food security may also depend on increased irrigated agriculture.

The government Water White Paper calls for reform of water abstraction licensing and the Department of Environment Food and Rural Affairs (DEFRA) is seeking a more flexible and responsive means of licensing, within a risk-based framework. This offers both threats and opportunities in the management of freshwater ecological status; at present our knowledge

1 Based on 2009 data 2 Based on 2011 data

5 simply does not exist to make anything more than broad statements about flow needs of river ecosystems.

The EA has an existing procedure through its Catchment Abstraction Management Strategies (CAMS) that include the definition of Environmental Flow Indictors (EFIs). However, the Agency has raised three headline questions: i) Are EFIs fit for purpose for the reform we are looking for?; ii) Is the current science good enough for us to achieve that and iii) If not, how can we develop the approach to EFIs and the underlying science to support reform?

This report provides a compilation of issues and ideas from agencies, academics, consultants and NGOs (Annex 1) and identifies some potential steps towards improving environmental flows approaches over the next 20 years. (Abstractors were not included in the initial contact group as their participation was felt to require a specially designed process in a follow-up phase).

Objectives

The overall goal of the work programme is to develop an improved environmental flow method that defines sufficient water to protect the aquatic and related wetland ecosystems using best available science, but recognising uncertainty, whilst being fair, transparent, understandable, flexible, practicable and affordable. Where possible, long term (>10 years) development of any new method should support interim improvements by delivery of partial results in the short to medium term (<3 years).

The overall objectives of the programme of work are to:  Build collaboration with relevant agencies to understand current issues and support future requirements  Engage with different stakeholders e.g. water companies, farmers, industry, whilst recognising roles and responsibilities, positions and interests  Review state of the art of the science of environmental flows to identify what, if any, changes are needed to make the EFI method suitable for protecting the environment under future pressures.

This programme will be delivered through several phases of work.

This report covers Phase 1 of the work programme, which involves undertaking a scoping study. The objectives of Phase 1 are to:  Collate views of the pros and cons of current methods and expectations of the future from agencies, academics, NGOs and consultants  Review methods used in three countries (with advanced environmental flow expertise) and assess their relevance to England and Wales  Identify options for potential evolution of the current EFI method or possibly new methods  Recommend activities for Phase 2

Later phases of the work programme may include the development of a research programme with the participation of a wider range of stakeholders, such as water companies.

Approach

The following four steps will be undertaken within this scoping and development project (Phase 1) to assess the feasibility of new approaches that will deliver new strategy and methods.

(i) Review the current use of environmental flow indicators and discuss with key agencies, academics and consultants, plus external interest parties (e.g. Scottish Environment Protection Agency (SEPA), Scottish National Herritage (SNH)) the pros and cons of current methods and expectations for the future.

Activities Establish focal contact points and staff to provide input to the process Produce a brief background document to introduce the topic to contacts Define a consistent set of questions Hold one-to-one meetings, small group meetings or telephone discussions with contacts Record pros and cons of current methods and expectations for the future

(ii) Review methods used in other countries.

Activities Select a small set of countries that have developed novel methods Distil the essential elements that are different from UK methods Discuss specific aspects with overseas contacts Produce brief summaries of key elements of the methods used

Summaries of environmental flow expertise and experience from South Africa and USA will be included as Annex 3 and 4, respectively. Key points will be selected and discussed in the main body of this report along with contributions of experience and expertise from Australia. South Africa, USA and Australia were chosen as they are leading countries in embracing environmental flow considerations. It is essential to recognise that not all experience from overseas will be applicable to the UK. For example, extrapolations from one hydroclimatic regime to another need to be treated with caution, particularly when ecological dependencies on river flows are considered.

(iii) Identify possible future methods that could be developed to meet future challenges.

Activities Assess the advantages and disadvantages of novel methods and their application in England and Wales Compare novel methods against advantages and disadvantages of current methods, and against expectations of agencies and Defra criteria Discuss key aspects of methods with agency focal points including operational obstacles Anticipate wider stakeholder reactions Assemble views, alternative options or new ideas stimulated by discussions Select a best set of method options with short and long term gains

(iv) Define a Phase 2 project that will engage wider stakeholder expectations.

Activities Distil an action list and feedback Define wide stakeholder group and methods of engagement Produce Phase 2 plan that will fully scope methods with estimated time scales and costs

Project outputs This report represents the primary project output. All key outcomes from the above activities will be fully documented within it. Included will be the main output, a plan for Phase 2 with agreed principles, ideas and roles for a full scoping of the development of a new environmental flow approach. Secondary project outputs will be a raised awareness within government agencies of possible new approaches to environmental flows for abstraction management and greater interaction between agencies on environmental flow issues.

Current use of environmental flow indicators

Annex x gives the description of EFIs and how they are used in the Environment Agency. EFIs are used to indicate where abstraction pressure may start to cause an undesirable effect on river habitats and species. They don't indicate where the environment is damaged from abstraction.

The current abstraction licensing practice for England and Wales aims to maintain an appropriate flow regime in each river. The primary driver for this is to meet the WFD. However, other objectives are also important at many sites such as delivering requirements of the European HBD, Sites of Special Scientific Interest (SSSI) and more generally ecosystem services that support human well-being and contribute to the national economy.

Under the WFD all river water bodies have ecological status objectives - primarily GES, in addition to a no deterioration objective. Ecological status for rivers is defined in terms of biological quality elements (fish, macro-invertebrates, plants and algae). Maintenance of an environmental flow is not required except for pristine/natural water bodies (where the target is High Ecological Status); abiotic quality elements (chemical, physico-chemical and hydromorphological) play a supporting role.

“Good” is defined as slight deviation from the reference (natural) state. However, surface water bodies can be designated as heavily modified (HMWB) if i. substantially changed in character due to physical modifications ii. the modifications have existing uses (such as navigation, water storage etc) iii. restoring to GES would adversely impact on those uses iv. there is no better option to provide the use In HMWB the WFD default objective is to reach good ecological potential (GEP). This is the best ecology for that water body, without compromise the uses to which the modifications relate.

It is noteworthy that GES and GEP are the primary objective options. Alternative objectives can be set through the river basin planning process, which involves stakeholder engagement. There is often a mis-match between flow assessments and biological classifications and there is a need for additional supporting ecological information (SNIFFER, 2012). Although the targets are biological, it is recognised that these are unlikely to be met in rivers whose flow regime is highly altered. The EU Blueprint is likely to suggest flow standards and guidance for WFD implementation Europe-wide.

Derivation of current EFIs Table 1. Recommended standards for UK river types for achieving Good Ecological Status given as % allowable abstraction of natural flow (thresholds are for annual flow statistics)

Type or sub Season flow > Qn60 Flow > Qn70 flow > Qn95 flow < Qn95 type

Apr – Oct 30 25 20 15 A1

Nov – Mar 35 30 25 20

Apr – Oct 25 20 15 10 A2 (ds), B1, B2, C1, D1 Nov – Mar 30 25 20 15

Apr – Oct 20 15 10 7.5 A2 (hw), C2, D2 Nov – Mar 25 20 15 10

Salmonid spawning & Jun – Sep 25 20 15 10 nursery areas (not flow > Q flow < Q Chalk rivers) Oct – May 20 15 80 80 10 7.5

Environmental flows required to support GES in UK rivers were defined through expert consensus (Acreman et al., 2008). These are presented as maximum abstractions that are permissible under different flow conditions for different rivers (Table 1). Types A1, A2, etc. were initially produced from a river classification based on macrophytes (Holmes et al., 1998). Macro-invertebrate experts confirmed the relevance of the classification then the table was adapted by fish experts to include salmonid spawning and nursery areas as an additional river type. It can be seen from Table 1 that the maximum abstraction for any river type depends on the season and the flow at the time of abstraction (given as a natural flow duration statistic e.g. Qn95). It is noteworthy that the experts involved in producing Table 1 felt that insufficient information existed to define thresholds with certainty. The consensus was rather more ‘we have no evidence that abstraction up to 10% will have a negative impact’ than ‘we have evidence that 10% is the clear threshold above which impacts will be significant’. Additional tables exist for special cases such as for designated rivers (e.g. SSSI).

Typology for water resources standards for rivers

Type Gradient Altitude Description (Metres per (metres) kilometre) Clay and/or Chalk; low A1 0.8+/- 0.4 36 +/- 25 Predominantly clay. altitude; low slope South East England, Eutrophic; silt-gravel bed East Anglia and Cheshire plain

Type A Type A2* Slightly low altitude Chalk catchments;

steeper 55 +/- 38 predominantly gravel 1.7 +/- 0.8 beds; base-rich Hard limestone and B1 4.1 +/-9.9 93 +/- 69 Hard sandstone, sandstone; low-medium Calcareous shales; altitude; low-medium Predominantly South slope; typically and South West mesotrophic with gravel- England and South boulder or pebble-cobble) West Wales bed B2 Shallower 71 +/- 58 Predominantly North than B1 West and East

2.7 +/- 10.7 Scotland Type B Type Non-calcareous shales, C1 5.4 +/- 6.5 101 +/-84 Hard limestone; more hard limestone and silt and sand than C2; sandstone; medium mesotrophic altitude; medium slope; C2 Steeper than 130 +/- 90 Non-calcareous oligomeso-trophic with C1 shales; pebble, cobble and/or 7.3 +/- 10.8 pebblebedrock;

Type C Type boulder bed Oligomeso-trophic Granites and other hard D1 Medium low altitude Oligotrophic, rocks; low and high gradient 93+/- 92 substrate finer than altitudes; gentle to steep 11.3+/- 15.6 D2 (including silt and slopes; ultra-oligo sand); more slow flow Oligo-trophic, with cobble, areas than D2

boulder, bedrock, and/or D2 High gradient High Altitude Stream order 1 and 2

pebble bed 25.5 +/- 33 178 ± 131 bed rock and boulder; ultra-oligo trophic

Type D Type torrential * To reflect the different sensitivities of the headwaters of chalk streams to the downstream reaches, type A2 was split into two – A2 (headwaters) and A2 (downstream)

Figure 1 RAM Framework flow screening using EFIs to assess abstraction pressures and resource availability

EFIs in Catchment Abstraction Management Strategies

A procedure has been developed by the EA (the regulatory body for England and Wales that issues abstraction licences) to assess the pressure on river ecosystems from abstraction. This is called the Resource Assessment and Management (RAM) framework and is used to develop the Agency’s CAMS. In this procedure, the natural3. Flow Duration Curve (FDC) is determined for an Assessment Point (AP) on a river (thin black line in Figure 1). The EFI (green line in Figure 1) is then defined using a simplified version of Table 1 (green numbers in Figure 1 given in Table 2) that does not include seasonal variations and replaces the river types (A1-D2) with three levels of sensitivity to abstraction (high, medium or low).

This overall abstraction sensitivity band is determined by combining the individual sensitivity of three elements. The first element is called physical characterisation and relates to predicted river types or sub-types in the left hand column of Table 1, which are based on river macrophyte communities. The expected sub-type is defined using a model (Acreman et al, 2008) and GIS-derived parameters (catchment area, rainfall, and baseflow index). The second element is expected macro-invertebrate LIFE (Lotic Index for Flow Evaluation) scores and the third expected fish communities. FDCs incorporating the influence of abstractions and discharges at their current typical rates (the Recent Actual Scenario), and considering the potential pressures associated with licensed abstraction (the Fully Licensed Scenario) are then added (the blue lines on Figure 1).

The scenario and EFI flows are then compared to indicate where and under what conditions abstraction pressure may start to cause an undesirable effect on river habitats and species; they don't necessarily indicate where the environment is damaged from abstraction. In the example in Figure 1 the recent actual FDC is below the objective curve at all flows below Q80, but Fully Licensed abstractions would result in flows below the EFI for around 50% of the time. The CAMS resource availability colour shown on Figure 1 summarises the ‘actual’ and ‘potential’ abstraction risk relationships between the scenario flows and the EFI and is reported for four summary flow conditions (Q30, Q50, Q70 and Q95) for all WFD river, lake and estuarine water bodies, as well as for CAMS APs in the Water Resources GIS. These colours summarise whether, and under what flow conditions new consumptive licences can be issued, or if further investigation is needed to understand the extent to which the abstractions are causing an environmental impact. WFD Q95 Recent Actual Flow Compliance colours (also defined on Figure 1) focus more closely in the low flow risks associated with current actual abstraction. If abstraction is contributing to a failure to meet GES, measures may need to be put into place to restore sustainable abstractions.

The end result is a calculation of resource availability and identification of where and when water is available for further abstraction, where a water body may be over licensed but not over abstracted and where the water body is considered over abstracted.

The RAM framework is intended as a generic, broad-scale approach and a common process applied across England and Wales.Table 1 has been applied directly, such as in Northern Ireland where there is no equivalent to the RAM framework. In some cases, the analysis of abstraction impacts and environmental flows setting is supplemented by application of more detailed methods, such as PHABSIM, DRIED-UP, FHAT (APEM, 2011), HEFT (Atkins, 2005) that incorporate local conditions, particularly channel morphology and hydraulics and provide finer-scale resolution of pressures. These detailed studies may allow for different environmental flows to be considered in particular river reaches that still meet legislative requirements. However, in practice this has not been done.

3 Here natural is the flow without abstractions and returns; it does not attempt to define flows from a catchment under natural land cover. It defines the WFD reference flow regime

Table 2. EFI abstraction impact limits as % natural flow (green figures from Figure 1)

Abstraction Point on Flow Duration curve Sensitivity Band (ASB) Q30 Q50 Q70 Q95 3 (most sensitive) 24 20 15 10 2 26 24 20 15 1 (least sensitive) 30 26 24 20

EFI in WFD assessments

In England and Wales the EFI is used in compliance assessments of the hydrological classification for WFD. The compliance is used to identify the water bodies where reduced river flows may be causing or contributing to a failure of good ecological status.

The compliance assessment shows where specific scenario flows are below the EFI, and indicates by how much. This has been used to identify areas where flows may not be supporting good ecological status and is being used to help target measures or further investigation. For WFD purposes we have assessed compliance at low flows (Q95) recent actual scenario.

The degree of non-compliance has been split into three compliance bands to help prioritise actions to deal with abstraction pressures, where the risk of not supporting GES is greatest.

Flow adequate Flow not adequate to support Not adequate to support GES GES – Low - Moderate to support GES Confidence (uncertain) – High Confidence (quite certain)

Abstraction Compliant with Non-compliant Non-compliant Non-compliant Sensitivity EFI Band 1 Band 2 Band 3 Band (up to 25% (25-50% below (up to 50% below the EFI the EFI at Q95) below the EFI at at Q95) Q95)

ASB3 ‘high’ <10% lower <35% lower <60% lower >60% lower than than natural flow than natural than natural natural flow flow flow

ASB2 <15% lower <40% lower <65% lower >65% lower than ‘moderate’ than natural flow than natural than natural natural flow flow flow

ASB1 ‘low’ <20% lower <45% lower <70% lower >70% lower than than natural flow than natural than natural natural flow flow flow

General views on the aim of the project

Two general issues are worthy of discussion before assessing comments on detailed aspects of environmental flow science and methods.

Revision or replacement?

One school of thought advocates “if it isn’t broken, don’t fix it”. UKTAG and the environment agencies have already invested in development of environmental flows and EFIs and tools to implement abstraction licensing. Like or dislike it, abstractors have become accustomed to the approach and have focused on local studies to refine specific abstraction issues, rather than complain about the EFI method itself. The manifestation of this approach is to assess evidence of the success of the current EFI method, by analysing the status of rivers where EFIs have been implemented. If it is found to be lacking, it should be revised. In a review of EFIs (Atkins 2010), the EA felt that there is no obvious alternative to EFIs as a national screening of flow pressures. EA expert consensus is that we should not ‘throw out the baby with the bath water’ and should build on and improve what we have to make environmental flows more certain. Additionally UKTAG concluded in a recent review (UKATG 20124) that there is no new quantitative information that can be used to refine the standards for low flows (SNIFFER, 2012). Consequently, no changes to current recommendations on standards for low flows, on which EFIs are based, were proposed.

The alternative school of thought is that future work should not be constrained by old or current approaches, that although based on our best knowledge are not considered to be scientifically robust. We should not waste time looking for retrospective justification and be surprised when we don’t find it. The review of EFIs (Atkins 2010) concluded that EFIs are not currently effective in indicating whether flows support GES, because there was not a good correlation between compliance with the EFIs and the ecological status, even when confounding pressures are removed from the analysis. Past use of a method does not justify continual investment in improving it, if it will never deliver what we need. We should start from scratch and think about the right approach that both protects the environment and is practical to apply.

There is some opportunity for a middle ground here. We should evaluate the current EFI method, but not be constrained in our thinking about a better way to do things. For the UK government to invest in new methods and tools, there must be a measurable gain in terms of improved water management, provision and environmental protection, particularly in water stressed areas.

Holy grail or wholly pointless?

One school considers that science may eventually describe precise relationships between flow, other stressors and river ecosystem status. This reductionist approach would lead us to develop deterministic methods based on flow-ecology response curves that incorporate, seasonal aspects of flow regimes and species life cycles, including biofilms, food-webs and energy flows based on scientific findings. Thresholds may be a management concept and should not be forced on scientists since many flow-ecology relationships are smooth curves (although some clear thresholds do exist, such as the start of overbank flow to floodplains). Flow targets can then be defined to meet specific ecological or wider specific objectives.

The alternative school of thought is that the natural environment is too complex and that we will never be able to understand it fully. Lack of knowledge is too often being used as an

4 http://www.wfduk.org/sites/default/files/Media/UKTAG%20Summary%20Report_final_260412.pdf

15 excuse to avoid developing methods and making decisions. Seasonal differences in flow needs are insignificant when compared to flow changes caused by abstraction and so are not worth including in flow setting. We have no real clue as to whether upland or lowland rivers are more sensitive to flow change, which is the basis of the physical element of sensitivity bands. We would be better to focus on ‘getting some water in our rivers roughly in sympathy with natural flows’ with a broad objective of delivering a good ecosystem that maintains some important habitats and species, and delivers some ecosystem services. Related to this, but synonymous with it, is the suggestion to focus effort on other factors, such as irrigation efficiency, public supply leaks and river channel restoration and to make best use of the flow we have.

There is also middle ground here. We should seek to improve our scientific understanding and base methods on best science (recognising it will always be limited) and include an appreciation of between and within year variability in flow regimes. Working with experts can provide guidance provided uncertainty is recognised and results are placed in a risk-based framework.

Views of the current EFI method and wider scientific issues around use of environmental flow indicators

The following review has been produced through discussions with a range of interested parties including Agency staff, academics, NGOs and consultants listed in Annex 1. These are not necessarily the views of the report authors or the organisations by whom the individuals are employed. The views expressed are from professional experience and differing opinions and practices regarding the use of EFIs also be found in the hydroecology community in the UK.

Broad eco-hydrological approach

There is a strong view that despite any limitations of the EFI method, it has put environmental flow requirements on the water management agenda and has been a major step forward for balancing the needs of abstractors with the environment. It explicitly recognises riverine and wetland ecosystems, and their legitimate water requirements. It is consistent with the NEA that quantifies the benefits of natural ecosystems in providing goods and services to people that have a high economic value and contribute to quality of life and well-being. Furthermore, there have been very few abstraction licence appeals from using the existing standards and there is no evidence to suggest that a deterioration from GES has occurred since using the EFI, so there is some merit in what we are doing now.

Look-up tables (Tables 1 & 2) meet many higher-level criteria, such as being simple and easy to apply, understandable, transparent, consistent across all rivers of England and Wales (and Scotland and Northern Ireland), even if the precise figures used are slightly different. SNIFFER (2012) concluded that no new quantitative information is available that can be used to refine Table 1 and thus by implication the standards derived from it (Figure 1).

Australia is considering a move towards assessing ecosystem services outcomes from environmental flows as these are explicitly referred to in the Water Act of 2007 which emphasises the notion that water for ecosystems is water indirectly for people.

In the Murray-Darling basin, the entire water budget is allocated between its users and the government owns the water allocated to the environment. This is called ‘Commonwealth environmental water’, defined as the water made available for the purposes of protecting or restoring the environmental assets of the Murray-Darling Basin. The Commonwealth

Environmental Water Holder (CEWH) makes decisions on the use of the environmental water and can ask for the water to be released from reservoirs to produce watering events such as inundating floodplains to support bird breeding. It can also sell its water to other abstractors or increase its holding by buying water. Water is allocated according to average water availability conditions to enable long term planning and investment by abstractors. In any year the allocations are scaled proportionally to account for the actual total water available.

Strategic level environmental flow assessment is well developed in UK compared to other countries. Some countries, e.g. Germany and France use detailed habitat modelling at specific locations, rather than using a national level screening tool with local refinements.

The flow hydrograph

River flows vary continually from high flows to low flows, or in some groundwater fed cases, such as winter-bournes, the river may seasonally dry-up naturally. Furthermore, the length of the hydrological cycle (driven by seasons) can vary between years. The flow hydrograph presents a complex signal of variations over different time-scales from minutes to decades and hydrologists have sought indices that summarise the key characteristics of the regime. One approach used since about 1915 by water resources engineers is the flow duration curve (FDC), which is a cumulative frequency curve that shows the percentage of time specified flows were equalled or exceeded in a given period. Abstraction allowances to support GES (Table 1) and EFIs (Table 2) are defined in terms of points on the FDC. The use of the FDCs is consistent with water resource management within regulatory and abstractor organisations. However, FDCs do not record temporal sequencing of flows. For example, the allowable abstraction figures were originally expressed in terms of flow on the day and takes no account of whether the flow has been low or high before or is likely to rise or fall after that day. It also assumes that any flow is equally likely at any time during the period for which data are analysed. However, we know that clustering of different flows (e.g.low flows) occurs in particular years and particular seasons. It is also more difficult to account for different implications of surface water and groundwater abstractions. A hydrological time series approach provides the best opportunity to understand links between hydrology and freshwater ecosystems (as the hydrograph better represents those flows to which instream biota are likely to respond to) and also provides flexibility in licensing, for example water companies could license-trade with farmers by releasing water from reservoirs for downstream irrigation. Although this may be relevant to only a small number of river reaches and the costs of retrofitting dynamic release controls to reservoirs may be prohibitively high.

Time series of daily flows and hydrograph analysis already underpin much of the FDC summaries of ‘Complex Impacts’, which represent the impact of many of our larger reservoirs and flow regulation/transfer schemes in the WRGIS and CAMSLedgers. Some Agency areas apply EFIs to flow time series to assess implications for hydrograph characteristics, because the impact of EFIs is not always clear from the FDC. However, some feel that currently there is too much focus on flow statistics such as Q95. In practice, Q95 gives very different conditions in different rivers, although there is European wide recognition of the need to maintain a low flow and flow variability for a healthy water environment. The EA have developed tools for hydrograph analysis for use alongside FDCs to aid the understanding of the ecological impact of changing flows.

Direct river and groundwater abstractions rarely have much influence on the timing of high and low flow events. Taking a constant amount of water tends to lower the whole flow regime leaving the temporal variability (highs, lows, changes in flow rate) of the natural hydrograph. Analysis of the flow hydrograph is much more important when considering flow releases from reservoirs, where the entire flow regime needs to be defined. Some studies

17 have shown that the shape of hydrograph, such as the rate of rise and fall, is an important characteristic for river ecology and morphology. For example, salmonids may be stimulated to migrate as flow rises, bream move off of inundated floodplain habitats as the flow drops. This concept is important in regulating releases from hydropower stations, where flow downstream may pulse or alter vary rapidly (this is called flow ramping), causing washout of juveniles and stranding of adults under sudden low flow conditions. These issues are currently being investigated in Heavily Modified Water Bodies. The duration of periods of flow under or over specific thresholds could be included with an improved risk assessment method, but is not readily applicable using FDCs. Another piece of evidence is from DRUWID, where it’s shown that LIFE score relates to antecedent flow over several years (as was predicted by Extence et al. 1999)

In South Africa and Australia, FDCs are not extensively used; the focus of environmental flow assessment is on the flow regime, defined by river flow time series and its link to elements of the river ecosystem. In particular, timing and duration of floodplain inundation are extremely important and cannot be characterised by flow duration curves. In the USA physical habitat models are used extensively, which employ flow time series, although importantly these are not time series models, in that each time is treated independently. Where FDCs are used they use seasonal rather than annual data This incorporates within- year variability without having to estimate and deal with time series everywhere. In Australia environmental flows are often identified through an accumulation of the needs of a range of species and habitat suitability that can be defined through their hydrological needs such as flow magnitude, frequency, duration, inter-flood period, etc. Often these are simplified into environmental water requirements expressed as the return frequency and duration of particular flow bands. Here the limitation is understanding the hydro-ecological relationships rather than the river hydrology.

In Connecticut, USA, seasonal flow statistics are used, which improve the ability of managed flows to mimic natural flows, whereas in Florida a percent-of-flow approach is employed. In South Africa flow time is almost exclusively used for environmental flow assessment. In Australia flood timing can be critical to ecosystem response and condition and is incorporated into simple environmental flow requirements and in more complex ecosystem response modelling.

Targets for heavily-used, modified rivers and impoundments

Where water bodies are designated as heavily modified (HMWB), due to modifications that are economically important, they are set an objective of GEP, rather than GES. EFIs have not been defined for GEP. Where water bodies have a series of impoundments, such as the River Thames, back-water effects may make parts of the river ecosystem water-level dependent rather than flow dependent. In such cases it is difficult to define an appropriate flow regime and most likely the natural flow would not support a natural ecosystem. HMWBs, designated for uses such as flood protection or navigation, are set objectives to reach GEP by 2027. To reach GEP these water bodies require all the relevant mitigation measures to be put in place. Target flows could be set to maintain this condition.

The concept of linking EFIs with abstraction licensing is based on the premise that a direct river abstraction has a direct ecologically-relevant impact on the flow regime downstream. In the case of large impoundments, such as dams, the abstraction from a reservoir may have little relationship with the flow downstream (apart from long-term volume reduction). The flow regime instead will be dependent on releases from the impoundment, which are unlikely to be related to abstraction and will have their own legal agreement sometimes enshrined within an Act of Parliament associated with construction of the dam that may limit releases to avoid downstream flooding, to save water, or may enhance releases for downstream water users. Impoundment flow releases need to be linked to the abstraction licence (except for

18 hydropower and regulation for downstream abstraction). Abstraction from reservoirs is accounted for to some extent in WRGIS and CAMSLedgers through the use of ‘Complex storage support’ often from time series analysis of the reservoirs (though a full true time series analysis is not essential).

In the Trent, water in the river has been abstracted from the Severn catchment and returned as treated water to the Trent. Effluent discharge rates are thus very important for meeting EFIs but are regulated by separate consents related to water quality standards. Abstraction licences and discharge consents may need to be integrated more closely, with discharges representing ‘deposits or credits’ in a future trading system.

An approach to define flow releases was recommended under WFD 82 (Acreman et al., 2009) which has been further developed in WFD21d (SNIFFER, 2012). This is being tested in a number of experimental flows releases from reservoirs by water companies, but has not yet been implemented. WFD 82 recognised that many impoundments can totally alter the flow regime, whereas most direct abstraction reduces the flow but maintains the temporal variation in flow. Thus there was potential to define the whole flow regime by specifying ecologically-important elements of the flow hydrograph. This could allow ecological- optimisation of a flow release regime under total flow release volume constraints agreed with impoundment owners. The EA are undertaking investigative work into flows from over 300 reservoirs as part of their HMWB investigations. The Rivers Agency in Northern Ireland is considering using the 40% deviation from natural flow rule (recommend by WFD 82) to define reservoir releases.

Much of the conceptual thinking on environmental flows is based on the natural river paradigm that suggests an optimum river ecosystem requires a natural flow regime. Such a concept is easier to consider in large wilderness areas, such as in the USA, than in a crowded, modified, country such as the UK. The natural flow regime may not be the appropriate flow regime to match with an altered or modified river channel. In some cases there may be an intention to re-engineer the river to more natural conditions, thus the EFI should be still be based on the naturalised flow regime. If these modifications cannot be removed, it would be sensible to define alternative objectives based on a typology that reflects the effect of the modification. For example, it has been argued that the natural flow regime should not be used as a baseline for the River Itchen because a very long history of river management has created a nationally important ecosystem, designated under the Habitats Directive. Many of the engineering features, such as mills and weirs, are listed historical monuments are protected from removal. In such cases EFIs could be based on abstraction allowances related to modified rather than the natural flow regime.

In many other countries environmental objectives are determined separately for different rivers, rather than having generic targets for all rivers, such as GES or GEP. In Connecticut, USA, every river reach is assigned a condition goal class ranging from 1 to 4. Class 1 streams support habitat conditions and biological communities typical of free-flowing streams. Class 2 and 3 streams support “minimally altered” and “moderately altered” biological communities, respectively, compared to free-flowing streams of similar types. Class 4 streams are recognized as being substantially modified. Reservoir release rules are tailored to each class. Dams on Class 1 streams are not allowed to manipulate their reservoir storage actively; in other words, they must operate as run-of-the-river, flow-through systems. Dams on Class 2 streams must release at least 75% of their reservoir inflows at all times. A similar procedure is followed in South Africa, where the 1998 National Water Act states that only basic human needs and the river ecosystem have rights to water. Every river is assigned a management class through a process of research, stakeholder consultation and negotiation: Class 1 (Minimally used), Class 2 (Moderately used) or Class 3 (Heavily used). An ecological reserve is then defined to meet the ecological objectives.

The linkage of the flow hydrograph to specific elements of the river ecosystem is a central part of the environmental flow in Australia and South Africa. In many rivers in Australia little is known about hydrological and ecological relationships of life cycle stages of floodplain, wetland and in-channel ecosystems. The maintenance of adult stages, in species such as riparian vegetation, fish and birds is often not restricted by the timing of peak flows. However the timing of flows is considered critical to certain life stages involved in regeneration and recruitment and these are estimated to determine release regimes from storages. In Connecticut USA flow releases are related to bio-periods (biologically-based seasons lasting between one and four months). It is recognized that highly modified rivers can differ considerably in their flow requirements, so setting of environmental flows is undertaken on a river-by-river basis.

Standards for high status water bodies protected areas

Standards and tests have been developed to assess water bodies that are candidates to achieve High Ecological Status (cHES). The tests work by checking that the deviation of Q95 from natural is in the range of 5%. It is noteworthy that even in a long record hydrometrically good station it difficult to gauge to this level of accuracy at Q95. Although a few protected areas, such as SSSI rivers will be in cHES water bodies, others will not.

Standards for protected areas

WFD is not directly compatible with designation legislation, so Natural England has been working to establish standards for rivers to meet favourable condition under the Habitats Directive. These need to be finalised and implemented. UKTAG is currently developing a framework for each UK region to use to better align protected area flow standards with WFD flow standards.

Countries outside the EU do not have generic flow targets for designated rivers. Environmental flow assessment is related river-by-river to ecological requirement, often to maintain specific ecosystems, communities or species that may have explicit conservation status. In Washington State, USA, instream flow rules typically cover long river reaches that include a mix of modified and more natural stretches. State biologists develop instream flow rules to protect the high-quality habitat, and those rules also apply to the degraded habitat within these reaches. In Australia, a series of icon sites have been defined with associated flow requirements. These aim to meet a range of ecological objectives that may be species- based (e.g. conservation of the Murray Cod), ecosystem-based (maintaining flow through the mouth of the Murray), stimulating wetland bird breeding or conserving red gum or black box floodplain forests. The designation of the icon sites in the Murray-Darling Basin was based on the identification of large, relatively natural, floodplain areas in good condition. The river estuary and the river channel itself were added as extra icon sites. The assumption is that these key areas are representative of the ecosystem as they are the high water users, being the large floodplain, channel and river mouth.

General v local

To implement European legislation such as the WFD a consistent approach is required across the continent. This concept extends to the UK where similar methods are used in Northern Ireland, Scotland, England and Wales. The UK TAG is helping with unification of approaches. A hierarchical approach could be employed that has a single screening tool applicable across all the UK, together with a set of subsidiary methods that provide higher local resolution. Furthermore, since rivers in the Lake District will be ecological more similar to those in Scotland than those in Dorset, methods and sub-methods should be based on biogeohydrology (a typology based method) rather than political boundaries. In addition, the use of expertise from across the UK will help ensure that standard development is robust

20 enough for use on water bodies throughout the country, the risk of challenges to regulatory decision making is reduced and clarity is provided for UK wide operators/interested parties.

Due to different legislative frameworks, the WFD 48 flow limits (Table 1) have been used more directly in Scotland and Northern Ireland than in England and Wales. For some the 3- way typology is too simplified and does not cover the range of river types in the UK and differences in sensitivities between them. Thus in most circumstances the EFI is used as a screening tool with models calibrated on local data employed to set environmental flow needs in specific river reaches. Similarly, the optimisation framework proposed by SNIFFER (2012) develops local solutions using local data. A range of locally applicable methods has been developed that can be individually justified, but produce different results and has led to a lack of consistent response. Such models often focus on river hydraulics to define physical habitat for fish (e.g. HEFT, FHAT, PHABSIM). Other approaches, such as HEV or DRIED- UP can be used to asses macro-invertebrate - flow relationships. There is a lack of guidance on the advantages and disadvantages of different approaches or the measures of river status; currently choice and results may be very dependent on which consultant is employed. Some work on generalisation of hydraulic habitat has been undertaken (RAPHSA) and this could be developed into a true physical component of EFI e.g. based on wetted width, to complement macrophyte, fish and macro-invertebrate elements. Further development of the underpinning science would also be advantageous.

Many US states require by law the use of detailed models, such as PHABSIM for hydropower project relicensing.

In South Africa, ecological reserve assessments are undertaken at one of four levels of complexity depending on the ecological and social importance of the water body and the level of potential conflict over water: desktop (2 days), rapid (2 weeks) with some field assessment, intermediate (8 weeks) and comprehensive (32 weeks), both with detailed field assessment.

In the Murray-Darling basin Australia, the new Basin Plan is attempting to set basin scale objectives for aquatic ecosystem condition and therefore priorities for environmental water management. At smaller scales of river regions and local river reaches more detailed modelling and expert panels are used to set environmental flow management rules. This is often undertaken in conjunction with infrastructure use and in some cases land management options.

Multiple stressors

Flow itself (discharge – m3s-1) has little direct influence on the river ecosystem. It has indirect impacts on biota through dilution and its interaction with channel morphology that gives rise to physical habitat (e.g. depth, velocity, turbulence, wetted area) and associated processes of sediment transport, interaction with floodplains and water temperature. This makes flow regulation a crude tool for ecosystem management. The focus on flow as an indicator relates to volumetric licensing of abstraction and the availability of flow at gauging stations at other locations from flow models. Evidence from natural river catchments in USA shows consistent relationships between flow and these other variables, such that flow is a good surrogate for general river ecosystem conditions. In less natural systems it may be necessary to survey habitat state. Several abiotic ecological indicators of habitat state have been proposed (SNIFFER, 2012) that may be easily measured in the field or derived from existing data.

Few UK rivers are very natural, River Habitat Survey (RHS) data suggest that more than 50% of rivers in England and Wales have been modified physically i.e. straightened, deepened or widened for flood management or other purposes, strengthened with concrete revetments, fragmented by weirs and other impoundments, or separated from their

21 floodplains by embankments. Other factors such as simplification of riparian zones may influence river habitat through changes in temperature and light levels. Connection with out- of-channel habitats, such as on floodplains may be very important for certain life stages, such as fish spawning. Although some of these may be remedied as measures to achieve GES, many will not. In many cases the natural flow regime may not be the appropriate flow regime to match with the altered physical nature of the channel. For example, natural high flows can sweep organisms from a river reach where refugia have been removed. The EFI could vary with river modification or restoration to improve habitat. This has led to calls for a more focused effort on channel modification to make best use of flow.

At an expert workshop where 54 ecological indicators of severe effects of abstraction and impoundment were proposed (SNIFFER, 2012) which included biotic, abiotic, multi-taxa and multi-trophic level indicators. Thus, a multiple pressures approach is required that integrates flow with habitat, water quality, etc. with wider environmental indicators, such that funds are used for measures that will achieve desired results. Resources spent on improving flows may be wasted if other conditions are not met. Furthermore, it is suggested that given our uncertain knowledge of river flow-ecology relationships at the scale of river management a risk based approach is likely to be most appropriate. This would prioritise risks and flow needs of key habitats/ecological elements to enable identification of risk areas rather than focussing on formal standards (e.g. WFD).

Although there is good evidence of examples of impacts of different pressures on the river ecosystem, this tends to have been assessed in specific analyses for different disciplines, such as fisheries or control of point source pollution. A systematic review is required to assess the relative scale of different pressures and their combined impact to provide a comprehensive knowledge base on which to build a future research programme.

Physical habitat models (such as PHABSIM) in the USA are used because biota respond to the availability of habitat (depth, velocity, substrate, cover, etc.) not to the flow per se, so in morphologically altered rivers managing the flow regime requires case-by-case EFI determination.

In the USA, other models of sediment transport, water quality, etc., are used as required in association with habitat models in considering environmental flow requirements.

In Australia, environmental flow assessments include the management of water quality issues which includes managing the risk of black water events (very low oxygen levels) that cause fish kills, salinity levels that modify ecosystem response and in extreme cases lead to their decline and cause algal blooms.

Hydrological and ecological thresholds

The definition of limits to abstraction to maintain ecological targets such as GES, depend on the existence of thresholds at which the ecosystem alters significantly. Some thresholds may exist such as bank-full flow that separates water only in the main river and water also on the floodplain flow. However, ecological responses to flow tend to be in the form of smooth curves, such that ecosystem state changes seamlessly as flow changes. In such circumstances, defining flow limits becomes a political decision, rather than a scientific decision.

In the USA thresholds are only applied to certain species or guilds. In South Africa, the DRIFT method does not explicitly include ecological thresholds because river ecosystems gradually decline in condition rather than going through a series of thresholds. Noteworthy exceptions where thresholds are used in Australia and South Africa are complete de- watering of the river bed, loss of river-floodplain connectivity and closing of a river mouth and

22 isolation of estuary from sea. The use of environmental water requirements as indicators of good ecological condition introduces the concept of thresholds when reporting whether flow metrics are met or not. A risk based approach would help predictons of ecological condition from hydrological outcomes that do not meet flow targets. The Strategic Adaptive Management (SAM) approach used in South Africa to define management goals incorporates the concept of thresholds of ‘potential concern’.

Resilience, recovery and abstraction allowance flexibility

The limits in Tables 1 & 2 define allowable abstraction on any day (as % of flow) to the flow on that day (expressed a point on the FDC). Although attractive from a water regulation view point, this does not account for some important hydro-ecological processes and functions. Although the abstraction limit for each FDC point is effectively a threshold, these limits have been produced by expert panel pooling experience, rather than based on specific tipping points at which ecological conditions abruptly alter, such as inflection points in the habitat- flow relations (e.g. floodplain-river connectivity). EFIs could reflect hydro-ecological thresholds more explicitly, such as explicit links to channel morphology, bank-full discharge and floodplains, but this would need site specific data and some thresholds may be outside the range relevant to day to day abstraction management.

In natural river channels there is often a diversity of habitat at different flows that can support organisms with different requirements. Resectioning, including widening, deepening, straightening and general loss of morphological diversity reduce this diversity and make some flow conditions intolerable for some species. This suggests that modified river channels may be more sensitive to flow change than natural channels. It also suggests against undue focus on one target species, as this may be to the detriment of other species. Generally the argument for a specific sentinel species (typically salmonids) is that provision of flows for them would also protect other ecosystem components.

The flow-on-the-day approach does not incorporate a time series approach including the importance of antecedent conditions or recovery. For example, the impact of reduced low flow may depend on whether the past few weeks or months have been wet or dry. Duration of periods under given thresholds may be more important than whether the threshold has been crossed on any day. Many riverine systems are quite resilient to high/low flows and can survive natural floods/droughts provided they are given time to recover, although resilience may change according to the time of year although information on this is not currently available. If we are interested in whether long-term damage occurs impacts of short term events may not be significant.

The greatest demand for water is often for agriculture and public supply when flows are low (during hot dry summers) but this is also when allowable abstraction is at its lowest. Whilst the emphasis is on abstraction when water is plentiful (during floods) and storage during dry periods, consideration could be given to higher abstractions for particular periods provided that the river ecosystem is allowed to recover during the following wetter period. However, care would need to be taken here as at the time the length of the period of low flow would not be known. Vegetation studies have shown that Chalk rivers may take 2-4 years to recover after drought; other rivers recover more quickly. The rate of recovery probably depends on other stressors such as water quality or level of morphological alteration. Future work and data collection are needed to help answer this question. A key point in this respect is that natural variability and abstraction impact are not the same i.e. a major abstraction during low flows is not the same as a natural drought.

There is no experience in the USA and South Africa of managing abstractions during drought, based on recovery periods. In Australia the proportion of water allocated for environmental flow releases is determined by the type of environmental water holdings. Most

23 environmental water is ‘general security’ which means that water for human critical needs are allocated first, then high security water entitlements are allocated mostly to agricultural use, before allocations are given to general security. During periods of drought environmental water is usually the first to be cut. During the recent 10-year drought some environmental flow releases were allocated but not made for political reasons because farmers were suffering from insufficient water.

In semi-arid South Africa it is noted that high abstractions during some part of the dry season could result in the riverbed becoming de-watered or at least experiencing a drastic change in hydraulic habitat and living space for aquatic organisms. Thus their river communities of plants and animals would probably not survive intact through an annually repeated period of high abstractions in the dry season. This is probably only relevant to a few emphemeral streams, such as Chalk rivers.

It is noteworthy that most countries have overriding public health safeguards. For example in Connecticut, USA, drought contingency plans include a provision for zero releases during water supply drought emergencies, which ensures that the “last drop of water” goes to people rather than to the stream.

Baseline conditions and environmental change

Most thinking has focused on the natural flow paradigm as the baseline against which to determine environment flows. This is consistent with using reference conditions for natural water bodies under the WFD. For HMWBs it may be more appropriate to consider the issue from a different starting point, such as defining the volume of water available in a reservoir (that is not essential for purposes for which the reservoir is used) and making best use of it through environmental flow releases that optimise ecological benefits along the downstream river and within the reservoir itself.

A decision support framework has been proposed for optimising benefits as water is released from impoundments (SNIFFER, 2012). This framework is based upon the Building Block Methodology (Acreman et al., 2009). Such approaches are used in, for example, South Africa (Tharme and King, 1998), but we must be mindful that they are operating under different laws, policies and environmental conditions and the hydro-ecological science base is different. For example, in South Africa different objectives (pristine through to ‘working rivers’) are defined through stakeholder participation. They recognise that rivers do not need to be in a near natural state to provide important ecosystem services. We cannot necessary transfer this experience directly to the UK to implement the WFD, but we need to assess overseas approaches and operations in the light of their policies.

Climate change is anticipated to alter natural flow regimes and reference conditions in the UK. The Future Flows project provides a consistent set of future flow scenarios. Similar work for water temperature is needed, and current ecological reference condition models need to be adapted to be fit for the future. Benchmark (natural) river flow regimes are defined either from naturalised flow data from gauging stations or models such as Low Flows Enterprise, that are based on data from broadly natural catchments. It is probably unreasonable to consider that flow regimes are stable and current climate change scenarios suggest that flow regimes will change in the future. The latest thinking on the WFD is that reference conditions should change as the climate changes, although this is not an approved policy yet. The implication is that benchmark flows should also change. Abstraction allowances could remain the same in percentage terms, but against a new baseline that is not that of the current natural flow regime. Much will depend on the nature of the change, e.g. whether the probability of occurrence for clusters of low flow events alters. Thus, a new system would need to be flexible in being able to incorporate the best evidence as it becomes available as we will never have all the answers at any one point in time.

As noted above, the sensitivity to abstraction is partly based on prediction of expected communities of macrophytes, fish and macro-invertebrates. The RIVPACS IV model (within RICT) is based on macroinvertebrates samples taken from reference sites in the 1970s and 80s. There has been a general improvement in river water quality since then, marked by a change of reference communities. Consideration should be given to how these communities have changed and are likely to change. In lakes, organisms in sediment cores have been used to explore reference communities. It is critical that the baseline conditions used are representative.

Many countries consider that flow alteration caused by climate change is dwarfed by the alteration caused by dams, withdrawals, and land-use change. The baseline used for environmental flow assessment in Michigan, USA, is current, rather than natural, conditions. Flow-ecology relationships for the Susquehanna and Potomac River basin also are expressed in terms of flow alteration rather than as flow depletion from a natural state.

South Africa now routinely assesses ordinary scenarios of management options and then the same scenarios with climate change super-imposed. It is recognized that it might be eventually necessary to re-assess not only the baseline (reference) flows but also the abstraction licenses themselves in terms of whether or not water is being put to best use.

Environmental flows in Australia are based on available water during any year and often targeted at specific ecological outcomes, such as maintaining floodplain forests for stimulating a wetland bird breeding event. The Australian Government through the Murray- Darling Basin Authority and the Commonwealth Environmental Water Holder have developed an environmental flow decision framework that identifies the water availability and defines a range of ecological goals from maintaining refugia during low water availability to promoting regeneration and building resilience during high water availability times.

Australia is considering moving towards assessing ecosystem services outcomes from environmental flows as these are explicitly referred to in the Water Act of 2007.

Knowledge and uncertainty

Relative to many other countries, the UK has good hydrological and biological data and EFIs are well developed compared to other European countries.. Yet hydro-ecological relationships remain poorly understood due to lack of research and paucity of sites where both biological and hydrological data exist, although flows can be modelled retrospectively. The UK agencies have funded a series of projects and workshops to define and improve the flow needs for riverine biota under the assumption that researchers have improved their knowledge in the meantime e.g. RFOs (1998), WFD 48 (2006), WFD 82 (2007), SNIFFER (2012). Yet little research has been funded in the intervening periods, by EA, NERC or other funding bodies. Each of these has also highlighted knowledge gaps which then haven’t been addressed. For example, in the workshops for WFD 48, the experts felt that they could not define flow standards because the information did not exist. A key recommendation was that long-term data are collected to refine the standards. With the possible exception of macroinvertebrates and abstracted systems, through the CAMS monitoring, this has not happened. So for fish, macrophytes and algae we are no closer to understanding appropriate standards than we were six years ago. The collection and analysis of long term datasets is critical and requires funding.

It is recognised that hydro-ecological relationships will never be perfect no matter how many data are collected. We need to work on the basis of best available evidence, as we cannot wait for a final solution. We need staged development of science with interim consensus positions that we can adopt in practical work until the next step forward comes along -

25 waiting for improved (or perfect) scientific information could be at the cost of inadequate environmental care. This can be incorporated within a consensus, best belief e-flows framework, now, but which is continuously subject to challenge and revision. Accepted best knowledge or practice can remain so until it gets replaced, by the decision of an appointed expert group. In this context, information may be 'fairly sure' as well as 'precautionary'.

The available scientific literature has been reviewed many times with uncertainties, inconsistencies and gaps highlighted. This information needs to be recorded in a more systematic manner, such as through a Systematic Review and formal database, so that new work builds-on and does not repeat past work. Norris et al. (2012) provide a method for using published literature to determine the support for cause-effect hypotheses in environmental assessments. In most environmental flow studies in Australia the uncertainties of hydrological modelling and the inferred ecological response are noted but not explicitly calculated.

Uncertainty is included in a qualitative manner in EFI definition. For each of the three elements (macrophytes, macroinvertebrates and fish) a three-level confidence score is given in addition to the sensitivity score. The models for predicting the sensitivity types of each element vary in their ability to predict each type. Consideration should be given to a more risk-based approach taking uncertainty more explicitly into account, worked through to flow regimes to allow sensitivity of adopting EFI number or slight variations. This will help to build trust between regulators and abstractors and to allow flexibility of abstraction allowances.

Many hydrological and ecological monitoring networks were designed for specific purposes, such as water resources assessment, flood forecasting and monitoring of point source pollution. Many records are still quite short and may not adequate capture time series structure, especially for extremes of the flow regime. Many networks are being reviewed and revised under drives for cost reduction. Staff in the EA have worked hard to maintain networks for hydro-ecological purposes and the WFD Ecological Status Indicator network has been expanded. A comprehensive overview of biological and ecological monitoring and data requirements is needed to assess whether they are fit for purpose to support development of future improve hydro-ecological relationships. This could be an internal exercise as agency staff have a good grasp of what the current data are good for and where the gaps are. Investment in monitoring should be commensurate with investment in flow management and restoration, but there are often very few baseline data available, so designing and measuring the success of restoration schemes is often difficult.

In Michigan, USA, environmental flow criteria are applied in a risk-based context that provides ‘early warning flags of adverse resource impact’. Prospective water users use an online Water Withdrawal Assessment Tool (WWAT) to determine the level of risk associated with their proposed withdrawals.

In South Africa, revisions to the DRIFT method may include establishing a band around the 50th percentile to allow for some flexibility regarding daily climate and water-use activities as well as addressing the uncertainty in the ecological reserve values recommended.

National accounting

Under WFD all river water bodies must meet GES or GEP individually unless an alternative objective is set. In Sweden, a national accounting approach has been taken. The Swedish government has calculated that if environmental flows are implemented at its five largest hydropower dams (presumably to achieve GEP), a further 22 new small dams would need to be built to fill the power production short-fall. It is estimated that the total environmental degradation would be far worse with 22 new dams. Consequently, it would be better in terms of national environmental accounting to designate five ‘sacrificial’ rivers where no

26 environmental flow is released, but to ensure environmental flows on all other rivers. National environmental accounting could be considered as part of EFIs. However, UK abstraction has many small abstractions across many rivers that in combination have major impacts as well as individual major abstractions. This may be something that could be addressed by the Natural Capital Initiative.

In the Penobscot River basin Maine USA, agreement was reached to remove two dams to improve the river ecosystem whilst increasing storage in remaining dams which slightly increased hydropower production.

Some South African environmental flow assessments have recognised at the basin level the concept of sacrificing some part of a river system in order to keep development low on the remainder of the system.

In Australia the National Water Initiative is addressing the perceived over-allocation of rivers in the Murray-Darling Basin. The ecosystem is considered degraded and the balance of water allocation is too far towards consumptive use rather than environmental water. Rather than trying to restore all rivers to natural conditions, a number of key environmental assets have been identified based on condition, uniqueness and representation.

Maximum thresholds

Most historical environmental flow work has focused on the problems of insufficient water at low flows. Indeed, Tables 1 and 2 give only maximum abstractions. The natural flow regime paradigm stresses the importance of all aspects of the flow regime including floods. However, too much flow at the wrong time can be as detrimental to the ecosystem as not enough flow. Urban research (Booker et al., 2003 for example) shows that high flows are the most critical for the river ecosystem, for example washing fry from river reaches due to lack of refuges.

Releases of water which are not part of the normal hydrological regime can also have detrimental effects. Inter-basin transfers provide further examples of where rivers may have elevated flows. (e.g. Ely-Ouse transfer). This transfer may be supply-side, e.g. a diversion upstream of a hydropower station, or discharge-side, where effluent from a treatment is discharge into a different river than it was abstracted from. Discharges are taken into account in actual flows in CAMS ledgers but maximum targets are not considered.

Many abstraction schemes involve releases from regulating reservoirs for use downstream e.g. the River Dee where flows during dry periods are much higher than they would be naturally (although it is a Special Area of Conservation). Furthermore, in the River Ehen recent freshets were released that resulted in high freshwater pearl mussel mortalities.

Many studies across the US relate adverse ecological impacts to increased flashiness of high flows due to impervious land surfaces and agricultural drainage. It is recognised that, for example, high flows that once brought fish onto the floodplain of a broad, meandering river might only flush them downstream now that the channel is deep and narrow. However, to date these stressors are not being regulated to restore more natural high flow regimes.

In South Africa, the Inkomati Catchment Management Agency flow strategy includes definition of upper and lower environmental flow limits per month.

In Australia droughts are more frequent than floods but these do occur. During these natural floods there is little to no control over flows and they are seen as a natural part of the environment. In highly regulated rivers with large storages there are problems with irrigation

27 demand releases producing high flows during the agricultural growing season which is not the natural high flow season.

Adaptive management

In the light of inevitable uncertainty in hydro-ecological relationships and thus in setting abstraction limits, alternative strategies are required such as explicitly incorporating uncertainty in a risk-based framework.

Another strategy is adaptive management approach where simple measures are applied, such as returning some flow to a dried-up river, combined with simple habitat restoration and, most importantly, monitoring. Further interventions and flow modifications are then based on the knowledge gained by monitoring the response. The importance of adaptive management is emphasised in identifying medium and long term solutions (SNIFFER, 2012). This approach does not seem particularly easy in the current Asset Management Planning (AMP) system where final answers appear to be required in short timescales.

Furthermore, whilst EFIs may show abstraction pressure they should not necessarily be used directly as a target for restoration or resolving unsustainable abstractions. Adaptive management can be a better restoration approach, re-introducing water and monitoring response. Restoration may even require unnatural flows to kick-start recovery processes and reintroduction of lost species. Flow recovery from water bodies that are non-compliant with the EFIs should only occur when supported by additional investigations to provide ecological justification and where costs are not disproportionate.

Implementation

Although this project focuses on the science of EFIs, it is important to put this in the context of implementation, because although a scientifically elegant new EFI methods may be designed, it would be merely of academic interest if they could not be implemented. Some scientists involved in previous method development have felt that the science has been compromised in making the method simple to implement, such as removing the seasonal element from Table 1 to derive Table 2. However, many practitioners believe seasonal differences in EFI are insignificant compared to the current patterns of abstraction pressure.

It is recognised that there are wider requirements for an EFI method beyond scientific excellence. Abstractors require water for their uses, so licenses ultimately relate to volumes of water than can be abstracted. Licenses have to comply with the law and be enforceable.

A major challenge in implementation is application of the complexity in our scientific understanding with the need for extreme simplicity and clarity in our policy and procedural formulations. How decisions are made on the trade-offs when coming up with an almost certainly compromised method that is practical is important.

There are also very practical limits to implementation. Environmental flow releases can be designed for a dam but implementation relies on the existence of and the ability to operate release valves to deliver the flow. It may be impossible or too expensive to retrofit valves into old dams. The fact that water bodies are failing to meet GES may not be the fault of the EFI method, but more due to the fact that the EFIs are not implemented.

Environmental flow approaches tend to focus on specific sites or parts of catchments (with the exception of major studies such as of the Itchen). Water companies have been pushing the EA to give them ‘probable’ and ‘confirmed’ abstraction change scenarios for their water resources planning. They want to look at pressures on catchments and not sites as this would cost less. However, the EA prefer to look at pressures on one confirmed site at a time.

For licenced abstraction volumes to be related to flow, the abstractor must know the flow. Currently, this is related to an assessment point, normally a flow gauging station. Ideally the flow would be available at all abstraction points, which is possible with real-time flow computer model based on the river network. (A computer model is already in place that provides real-time flood forecasts for major river reaches.) Since abstraction allowances are based on naturalised flows, these would need to be computed, requiring real-time collection of abstraction and effluent discharge data. A further challenge would be incorporating the impacts of groundwater abstractions that may affect rivers in a delayed manner.

One possibility in the long-term might be the application of a continuous national time-series model that provides now-casts of flow at any point on the river network and is constantly updated by flow data from gauging stations and water quality data. This would be linked to water depths and velocities by hydraulic rating data, based on hydrometric survey data, flows and hydraulic characteristics would be linked back in time to a hydro-ecological model updated two to three times per year as ongoing WFD biological monitoring data are assimilated. This model would be dynamically linked to water management procedures both for setting licenses and implementing any conditions, such as hands-off flows. Furthermore, it has been suggested that the natural variability of river flows and ecological resilience could be included by specifying hands-off flows for naturally dry, wet or normal years (SNIFFER, 2012). However, predicting natural or influenced time series of river flows, particularly everywhere is most definitely not a trivial matter. It would need strong justification both from three perspectives (1) that such as model will be beneficial to water resources allocation (2) that it is feasible hydrologically (3) that sufficiently strong hydro-ecological relationship exist to benefit from the model.

In Connecticut, USA, all regulated water supply reservoir operators are required to gauge inflow to their reservoirs. On the other hand, the Susquehanna River Basin Commission has devised a procedure for estimating flows based on nearby gauges.

Many parts of South Africa have water resource models up and running. For example, the Inkomati Catchment Management Agency, has daily flow data of the whole river system shown on a website.

In Australia water managers release water from storage to meet water demands downstream which includes irrigation demands and environmental water demands. This allows water managers to optimise flow releases to produce desired flow magnitudes at desired times. Flows are measured at points downstream. Return flows that come from floodplains and wetlands are placed back into the general consumptive pool and can be re- allocated to users downstream.

National skills in hydro-ecology

The UK undoubtedly has some excellent experts in the field of hydro-ecology, including those in academia, consultants, NGOs and agencies. Many of these work in specialist areas such as water policy, licensing, fish ecology, river hydraulics and water resources. Yet the pool of expertise is small. To be well-equipped for the future the UK needs an increased pool of hydro-ecology experts. A review is required of undergraduate and post graduate training and UK expertise in general. This may lead to a request to NERC for MSc or PhD fee support. The Agency should be talking to NERC (e.g. through WSKEP) about skills, any programme would likely need joint funding by the Agency, NERC and possibly other stakeholders (e.g. UK Water Industry Research UKWIR)). It is important to note that there is a relatively narrow window now for PhD research to kick-off if it is not to deliver too late for FAAW. Another angle to investigate is more intensive training e.g. a week or two week long course in hydro-ecology to increase expertise of current staff.

It is important to recall that a key recommendation from WFD 48 was that the experts be involved in an ongoing basis. There is also a need for better involvement of scientists in policy development to help understand and make trade offs.

In South Africa and Australia, academics have been more closely involved with environmental flow setting than in the UK. Much of the research on environmental flow requirements is undertaken in academic and government research organisations and this knowledge is incorporated with other others during expert panels that include local knowledge and more detailed knowledge of hydrology and land use. In South Africa the Department of Water Affairs has had very strong support from the national Water Research Commission to fund multi-year research programmes and post-graduate studies feeding into under-graduate teaching and a range of training courses. Twenty years on they have an excellent national base of informed water professionals who have produced a large body of scientific output and now work hands-on with the water managers.

In the USA, regulatory agencies have more scientific expertise to develop environmental flow methods. Basic science tends to be developed by academics, with other organizations, such as NGOs playing intermediate development roles.

Recommendations for Phase II

Basic principles

Any new method must be compatible with the wider objectives of licensing:  Meeting the needs of the abstractors whilst conserving the environment  Transparent & understandable  Broadly acceptable to abstractors and regulators  Adaptable to climate change  Contributing to sustainable management  Proportionate to environmental &/or abstraction requirements  Risk-based tier approach incorporating uncertainty  Considering the effects of all relevant environmental stresses  Based upon scientific principles and best available evidence, using expert concensus  Adaptable in the long term and responsive to shorter term changes in conditions  Flexible in terms of licence trading  Providing certainties for the user  Recognising that flow impacts of abstractions and discharges accumulate down the catchment

The degree to which all the above can be met is uncertain. For example, some abstractors may never fully accept a method. Some abstractors have needs that are difficult to meet, such as requiring a constant abstraction allowance regardless of flow in the river e.g. fish farms (although the depleted reach may be very short). In many parts of the public water supply system there is currently insufficient storage to allow abstraction rates to vary significantly with flow (this applies to most PWS groundwater abstractions, and many ‘licence of right’ large surface water abstractions located towards the bottom of our rivers).

Projects

The above discussion highlights a series of activities that could be undertaken in Phase 2. These fall into two broad projects: A. Hydro-ecological science and B. Implementation.

Phase 2 project A: Hydro-ecological understanding

This project focuses on improving hydro-ecological science to support setting environmental flow to licence abstractions. It is about improving our understanding of the relationship between river flow and biota and how this relationship is mediated by other factors, such as channel morphology, water quality, instream macrophytes, embankments and weirs. The project will include several activities, the main (activity A) being a workshop to define specific research for Phase III, supported by other activities (B-G), required as input to the workshop.

Principles This project will be planned to ensure scientifically excellent methods are based upon cutting-edge scientific principles and best available evidence, they are adaptable to climate change and are risk based and consider the effects of all relevant environmental stresses. Outputs will be adaptable in the long term and responsive to shorter term changes in conditions. The research will be guided by implications for implementation but influenced by implications for resulting abstraction allowances.

Roles The project will be led by research scientists with guidance from regulatory agencies and NGOs.

Activity A. Hydro-ecology science workshop £20,000

The main purpose of the workshop is to design a series of specific research projects for Phase III. Participants will be from UK academic institutions and consultants who would undertake the research plus agencies and NGOs who would advise on technical aspects of the work. In preparation for this workshop a set of other activities (B-G) should be put in place.

In recognition that hydro-ecological relationships will never be perfect, participants could be founder members of a task group to test available evidence and knowledge and build interim consensus positions, which will be continuously subject to challenge and revision as the science base develops.

Activity B. Review of data and monitoring £25,000

This activity would assess whether the current monitoring of biology, hydrology and related variables is fit for the purpose of developing a new EFI method, implementing and assessing it. The benefits of surveying/calculating refined ecological indicators (e.g. SNIFFER 2012) should be considered in how they could increase confidence in biological classification, prioritise sites for remediation and validate environmental flows. Hydrology should include continuous river flow gauging stations and sites where data are dis-continuous (spot gauging sites), groundwater, inter-basin transfers, effluent discharges, reservoir releases and associated data for hydrological models such as rainfall and soil moisture. It would need to include both quantity and quality. Biological data should include all macrophytes, algae, macro-invertebrate and fish surveys, plus other sources, such as angling returns, records of invasive species. Matches between monitoring sites need to be established to support cross- disciplinary assessment. Related variables should include land cover information, and those collected through River Habitat and Corridor Surveys, Fluvial Audits and other morphology surveys, such as river cross-sections derived for flood management.

Activity C. Hydro-ecology synthesis and database £15,000

Most projects include a review of literature and this is often repeated, re-inventing the wheel. Information and data on flow and other related habitat requirement of riverine and floodplain

31 species needs to be recorded in structured database that can be searched for species or river type and used as a resource for all future projects. The Synthesis Review and Systematic Review can provide an objective structure for this process. Future literature reviews can then be restricted to adding to this database. Current work funded by Defra at CEH will consider a structure for the review process. The database should be put on the internet to allow wide access. Work funded by the EA at Atkins-Water provides a start to the review activity. The funds suggested would design and develop the infrastructure and incorporate test literature. More funds would be needed to fully populate the database (indeed it is proposed as an on-going task, so long-term curation needs to be assessed).

Activity D. Assessing the effectiveness of the current EFI approach £50,000

Some 400 investigations have been undertaken (with reporting by the end of the year) on compliance of water bodies with the Water Framework Directive. A meta-analysis was undertaken to assess the relationship between WFD compliance and EFI compliance. This needs to be repeated and the analysis extended to new methods of assessing effectiveness of and related issues). It could be extended to a type of systematic review to assess the relative scale of different pressures and their combined impact on river ecosystems. An assessment is already planned by the Agency.

Activity E. Hydraulic-habitat method comparison (see concept note Annex 2) £100,000

There are numerous methods available to assess the flow needs of river ecosystems, many of these are based on hydraulic rating relationships (e.g. between flow and depth, or flow and wetted width). The advantages and disadvantages of these options at different river reaches are not known, so guidance on selection cannot be given. As part of the review of EFIs (Atkins, 2010) Agency staff identified the need to learn from ongoing WR investigations and the application of complementary hydro-ecological assessments and tools such as the Hydro-Ecological Validation (HEV) and DRIED-UP. This activity would establish a river reach test-bed and would compare methods. This would have good flow measurement and many biological surveys. Detailed morphological and hydraulic data would be collected, perhaps on a 5-10 m grid. Various methods can be applied with a consistent target defined by all available data.

Activity F. Flow time series implications of EFIs £15,000

Preliminary analysis needs to be undertaken on the benefits of using flow time series as opposed to FDCs for environmental flow setting, taking account of the uncertainties involved. This will include other options such as seasonal FDCs and time-series sampling strategies of FDCs. It should incorporate the implications of current EFIs for flow hydrograph elements, including analysis of temporal sequencing of flows, durations-under-threshold, ecological recovery and seasonal aspects. It could also include assessment of the utility of different indices of the hydrological regime and the idea of licensed abstraction allowance being conditional on flow in previous years/months and/or likely flows in months to come. Work already underway in Agency

Activity G. Hydro-ecological experiments £15,000

A series of specific hydro-ecological experiments needs to be designed in outline that can address fundamental questions. These could include a series of flow releases from reservoirs and abstraction scenarios. In addition it should include use of experimental channels where manipulations are possible. Real-world experiments need to be tied-into long terms monitoring. Some effort has already been invested by water companies, UKWIR projects, etc. It is important that data collection is comparable so that integrated analysis can be undertaken across trials.

Phase 2 Project B: Implementation

This project focuses on the implementation of hydro-ecological understanding to achieve effective abstraction licensing. The project will include several activities, the main (activity A) being a workshop to define specific research for Phase 3, supported by other activities (B-E), required as input to the workshop.

Principles This project will operate under the general criteria, of being practical, transparent, understandable, proportional, broadly acceptable and providing certainties to abstractors and regulators, proportionate to environmental &/or abstraction requirements, and flexible in terms of licence trading.

Roles The project will be led by regulatory agencies with support from researchers, but will involve abstractors in the process to achieve the above principles throughout the activities.

Activity A. Implementation workshop £20,000

The main purpose of the workshop is to design a series of specific research projects for Phase 3. Participants will be from UK agencies and abstractors plus academics and consultants who would undertake supporting roles. It will elucidate the degree to which new ideas for scientific methods meet the overall criteria of what is practical, transparent, etc. It will reconcile licensing needs with hydro-ecological evidence. It will consider a range of other issues including the use of flow time series, the practicality of monthly variations in abstraction allowances, the benefits of now-casting flows and flexibility introduced by permitting abstraction during dry period in return for hands-off flows during flow recovery. The workshop should take place after the science workshop detailed above. In preparation for this workshop a set of other activities (B-G) should be put in place.

The project could include development of new water management approaches that link hydro-ecological models to licence granting, operationalisation and review. The development of adaptive management (including monitoring) approaches to defining and implementing measures to deliver good river health could also be included.

Activity B. The use of flow time series for EFIs £20,000

This activity will look at the implications, costs and benefits of moving from annual flow duration curves. It will consider hybrid approaches such as using FDCs for licensing, but running flow time series to get risk of failure. Assessments of scenario based time series could be considered. It needs to consider the hydrological uncertainties and their influence quantifiable flow-ecology relationships

Activity C. Data needs of implementation £15,000

This activity would assess whether the current monitoring of biology, hydrology and related variables are fit for purpose to manage abstractions through EFIs. Consideration should be given to monitoring/calculating the additional ecological indicators proposed in SNIFFER (2012). Improved ecological indicators would allow a more robust biological classification and prioritisation of remediation resources. This needs to consider how monitoring may

33 change in the future as we go through RBM cycles, as Agency monitoring may be very different by the time abstraction reform happens.

Activity D. Feasibility of flow now-casting £15,000

This will consider the feasibility, costs and benefits of developing and running a model to now-cast flows at any point on the river network in England and Wales where abstraction takes place. This would explore a range of challenges, such as real-time collection (or simulation) and analysis of abstractions and effluent returns to determine both actual and naturalised flows. Options for a way forward would be assessed, such as piloting the approach on a catchment where increased flexibility in abstraction management is most needed and hence where investment in a now-casting system would have most benefit.

Activity E Feasibility of linking abstractions with effluent discharges and dam releases £10,000

River flow in many water bodies is dependent to a great extent on effluent discharges and releases from reservoirs. This activity will assess the potential for, costs and benefits of integrating abstraction licences and discharge consents, plus quantifying their hydroecological benefits would be useful.

References

Acreman, M.C., Dunbar, M.J., Hannaford, J., Wood, P.J., Holmes, N.J., Cowx, I., Noble, R., Mountford, J.O., King, J., Black, A., Extence, C., Crookall, D. & Aldrick, J. 2008. Developing environmental standards for abstractions from UK rivers to implement the Water Framework Directive Hydrological Sciences Journal, 53, 6, 1105-1120. Acreman, M.C., Aldrick, J., Binnie, C., Black, A.R., Cowx, I., Dawson, F.H., Dunbar, M.J., Extence, C., Hannaford, J., Harby, A., Holmes, N.T., Jarrett, N., Old, G., Peirson, G., Webb, J., Wood, P.J. 2009 Environmental flows from dams; the Water Framework Directive Engineering Sustainability. 162, ESI, 13-22. Acreman, M.C., Ferguson, A. 2010 Environmental flows and European Water Framework Directive. Freshwater Biology 55, 32-48 AMEC 2011 Water demand for energy generation. Report to Environment Agency. AMEC, London. APEM 2011 Functional Habitat Assessment Tool (FHAT) APEM, Stockport. Atkins 2010 Initial Review of Environmental Flow Indicators. Final Report to Environment Agency, October 2010 Atkins 2005 Hands-off Flows for cSAC Rivers Study: Phase 2 Methodology Report Environment Agency, North West Region. Booker, D.J., Dunbar, M.J., Shamseldin, A., Durr, C.S. & Acreman, M.C 2003 Physical habitat assessment in urban rivers under future flow scenarios. Journal of the Chartered Institution of Water and Environmental Management. 17, 4, 251-256. Cowx, I.G., Noble, R.A., Nunn, A.D., Harvey, J.P. 2004 Flow and level criteria for coarse fish and conservation species. Report to Environment Agency W6-096 University of Hull Dunbar, M. J., Warren, M Extence, C, Baker, L, Cadman, D, Mould, D.J., Hall, J., Chadd, R 2010 Interaction between macroinvertebrates, discharge and physical habitat in upland rivers. Aquatic Conservation, 20 (S1). S31-S44. 10.1002/aqc.1089 Entec UK Ltd 2005 RAM Framework Review Report to Environment Agency. Environment Agency 2010 Managing water abstraction Environment Agency, Bristol. Environment Agency 2012 Anglian in-river needs pilot project. NWN Inception Report 5084305 012. Environment Agency, Peterborough. Extence, C., Balbi, D.M. and Chadd, R.P.. 1999 River flow indexing using British benthic macro-invertebrates: a framework for setting hydro-ecological objectives. Regulated Rivers, 15, 543-574.

Norris, R.H., Webb, J.A., Nichols, S.J., Stewardson, M.J., Harrison, E.T. 2012 Analysing cause and effect in environmental assessments using weighted evidence from the literature. Freshwater Science 31, 2, 5-21. SNIFFER 2012 Ecological indicators of the effects of abstraction and flow regulation; and optimisation of flow releases from water storage reservoirs. Project WFD 21D Final report. SNIFFER, Edinburgh Soley, R. 2008 Water Resource Assessments for the Water Framework Directive, CAMS & Abstraction Licensing: Methods summary. Environment Agency, Bristol. Soley, R. 2012 UK groundwater resources in the spotlight AMEC Groundwater, AMEC, London Tharme R. E. and King J. M., Development of the building block methodology for instream flow assessments and supporting research on the effects of different magnitude flows on riverine ecosystems. Report to Water Research Commission, 576/1/98. 1998, Cape Town, South Africa. Williams, E. 2012 Review of international water abstraction regulation. Report to Defra AEA, Didcot.

Annex 1 Contacts made during the project

Environment Agency: Megan Klaar, Mark Warren, Jane Atkins, Mike Dunbar, Niall Jones, John Murray-Bligh, Lucy Bolton, Chris Extence, Miran Aprahamian, Karen Saunders, Nicola Poole, Sheena Engineer

Defra: Stuart Kirk

SEPA: Willie Duncan, Richard Gosling,

Northern Ireland: Deirdre Quinn, Peter Close

Natural England: Chris Mainstone, Anna Wetherell

Countryside Council for Wales: Tristan Hatton-Ellis

British Waterways: Peter Birch

RSPB: Rob Cunningham

WWF: Rose Timlett

Angling Trust: Mark Owen

Salmon and Trout Trust: Janina Gray

Atkins: Andy Gill

APEM: David Bradley

AMEC: Rob Soley, Nick Jarritt

University of Loughborough: Paul Wood

University of Hull: Ian Cowx

University of Cardiff: Steve Ormerod

University of Oxford: Dustin Garrick

University of Adelaide: Mike Young

MWH: Evan Dollar

Wallingford Hydro-Solutions: Andy Young

River Restoration Centre: Jenny Mant and Di Hammond

Annex 2

Concept note

Comparison of site-based environmental flow assessment methods

Background

Within CAMS, the RAM framework is intended as a generic, broad-scale approach to setting EFIs. In some cases, the analysis of abstraction impacts and environmental flows setting is supplemented by the application of more detailed methods, such as PHABSIM, DRIUD or HEFT, that incorporate local conditions and provide finer-scale resolution of pressures. The selection of method depends on many issues, some objective, such as the suitability of the method for particular circumstance, others more subjective, such as the relationship between EA area staff and particular consultants. Some methods have developed by consultants whilst others have been developed through Agency contracts to produce specific tools. No objective assessment has been undertaken to clarify the advantages and disadvantages of different methods.

Project objective

To test the advantages and disadvantages of different methods of environmental flow setting.

Approach

Select a set of river reaches that have key characteristics for which the EA would need to set EFIs. These would have good biological data, such as fish, invertebrate and macrophyte surveys and high quality flow gauging stations. The reaches could be where actual e-flow results are required, but this is not essential as the project is strategic rather than operational.

Collect detailed morphological and hydraulic data, possibly on a 1m grid at 2 or 3 different flows. These baseline data would provide the key input to most methods.

Select and apply different e-flow methods. Where possible each method could be applied by different teams to assess user-induced uncertainty.

Assess the advantages and disadvantages different methods.

Consortium members

A range of organisation with appropriate methods would be invited to apply method and possibly each other’s methods.

Output

Guidance on the advantages and disadvantages different methods agreed by method developers/users.

Approximate costs

£100,000

Annex 3 Experience from South Africa relevant to reform of water abstraction licensing policy for England and Wales

Jackie King, Water Matters, South Africa

Executive Summary This document summarises environmental flow expertise and experience from South Africa.

After a short introduction, Section 2 focuses on the approach used in South Africa for setting the Ecological Reserve, which is the country’s term for EFlows. It explains the legislative background and how protection of aquatic ecosystems fits into the South African Department of Water Affairs’ vision for integrated water resource management. It then outlines the main methods used to assess the ecological water requirements, all of which have been developed within South Africa.

Section 3 provides feedback on ten issues or ideas provided by the Centre for Ecology and Hydrology for discussion. These are:  The use of natural or present day river condition to guide water allocations  Environmental Flow Indicators for achieving Good Ecological Potential  The use of flow duration curves  The timing of abstractions  Climate change  Dealing with uncertainty  Ecological thresholds  Sacrificial rivers  Real-time flow data  Capping unnaturally high low flows

This section also addresses two subsequent questions from CEH on how allocated EFlows should vary through dry and wet years, and the use of EFlow methods for dam releases and for limiting abstractions.

Section 4 outlines how South African methods could inform the setting and implementation of EFlows in the U.K.

Section 5 provides some suggestions on relevant research and data collection.

Table of Contents Executive Summary ...... 38 1. Introduction ...... 41 2. Methods used in South Africa to set environmental flows ...... 41 2.1. The legislative background ...... 41 2.2. Measures for ecosystem protection ...... 42 2.3. The scientific methods used to assess the Ecological Reserve...... 44 3. Comment on the range of issues identified in the UK for consideration in its Phase 1 final report ...... 49 3.1. Natural versus present condition ...... 49 3.2. Environmental flow indicators to meet GES or GEP ...... 50 3.3. The use of flow duration curves ...... 50 3.4. Timing of abstractions ...... 50 3.5. Climate change...... 51 3.6. How to deal with uncertainty ...... 51 3.7. Thresholds ...... 52 3.8. National environmental accounting and sacrificial rivers ...... 52 3.9. Real-time flow data ...... 53 3.10. Capping unnaturally high flows...... 53 3.11. Subsequent questions from client ...... 54 4. South African EFlow methods that could inform the setting and implementing of EFlows in the UK ...... 54 4.1. Setting of environmental flows ...... 54 4.2. Implementing environmental flows ...... 55 5. Advice on data collection and research ...... 57 6. References ...... 59 Appendix A1: Detailed explanation of DRIFT ...... 61

List of Tables Table 2.1 Levels of determination of the Ecological Reserve, with estimates of the work involved ...... 45

List of Figures Figure 2.1 DWA’s plans for Integrated Water Resource Management, showing the positioning of the Management Class, Reserve and

RQOs. WDM = water demand management. ISP = Internal Strategic Perspectives. BHN = Basic Human Needs...... 43 Figure 2.2 The compulsory licensing process ...... 44

1. Introduction

This document summarises South African expertise and experience of environmental flows.

The Terms of Reference are to: 1. briefly describe the methods used in South Africa to set environmental flows with reference to other documents where complex methods are mentioned. These should be methods that are, or have been, used by governments or regulatory agencies, rather than theoretical methods or those used for purely academic study 2. provide views on ten issues provided in the ToR by the client 3. highlight where the South African methods, or aspects of them, may be employed to improve environmental flow setting and implementation in the UK 4. provide advice on data collection or research that is needed to improve environmental flow assessments or implementation. The book Sustainable use of South Africa’s inland waters: a situation assessment of Resource Directed Measures 12 years after the 1998 National Water Act by King and Pienaar (2011) is the main reference used throughout this document.

2. Methods used in South Africa to set environmental flows

2.1 The legislative background The South African approach to setting environmental flows is designed to meet the requirements of the 1998 National Water Act (NWA). The NWA evolved from the National Water Policy (NWP) drawn up in 1997 by the incoming first democratic government. Of the 28 principles established in this document, four were of special relevance for aquatic ecosystems. Principle 7: The objective of managing the quantity, quality and reliability of the Nation’s water resources is to achieve optimum, long term, environmentally sustainable social and economic benefit for society from their use. Principle 8: The water required to ensure that all people have access to sufficient water shall be reserved. Principle 9: The quantity, quality and reliability of water required to maintain the ecological functions on which humans depend shall be reserved so that the human use of water does not individually or cumulatively compromise the long term sustainability of aquatic and associated ecosystems. Principle 10: The water required to meet the basic human needs referred to in Principle 8 and the needs of the environment shall be identified as “The Reserve” and shall enjoy priority of use by right. The use of water for all other purposes shall be subject to authorisation. Thus, the two components of the Reserve (for Basic Human Needs and for ecosystem maintenance) are the only two rights to water in the NWA, with all other uses allocated afterwards through a licensing process. The ecological part of the Reserve is the South African version of an environmental flow (EFlow) and the NWA requires that it be set for all significant water bodies (called water resources in the NWA) including rivers, estuaries, wetlands, lakes and groundwater systems5. Its purpose is ‘to protect the ecosystems that underpin our water bodies’ (NWP 1997). Chapter 3 of the NWA focuses on how this will be done, through the use of three Resource Directed Measures:

5 A timescale of about 15-20 years is envisaged, with review at intervals thereafter.

1) a Classification of every major water resource in the country as either Management Class 1 (Minimally used), Management Class 2 (Moderately used) or Management Class 3 (Heavily used), through a process of research, stakeholder consultation and negotiation. 2) A Reserve of water from that water resource for Basic Human Needs and a further Ecological Reserve of water for the ecosystem itself to ensure its continued health and efficient functioning at the agreed Management Class level. 3) Resource Quality Objectives, which are measurable targets that can be monitored in order to ensure compliance to the agreed Management Class.

2.2 Measures for ecosystem protection The Resource Directed Measures are central to the overall strategy of the South African Department of Water Affairs (DWA) as it moves toward Integrated Water Resources Management (IWRM

Water availability taking into account ecological Water use requirements taking into account BHN and water requirements economic growth imperatives

Water balance / imbalance including need and

justification for strategic advocacy Stakeholder Inputs Stakeholder

Water resource

management options National Water Water resource

Resource Strategy classification Biodiversity overlays scenarios

Future use projections including Catchment Management water for equity, rural and Strategy (including ISPs) economic development needs Public Participation Public

Recommended Class Allocation Schedule

Compulsory Water New water Updated modeling Management Class, licensing conservation resource system and decision Reserve and Resource

Appeals and WDM development support tools Quality Objectives Publication, Publication,

Monitoring, evaluation and enforcement (water resource quality audit, source directed controls, and more)

Adaptive Adaptive management management

Figure 0.1). DWA’s vision for IWRM comprises four key phases. First, water availability (top left box) and water use requirements (top right box) provide information on the existing water balance for each catchment. Second, in-depth investigations and activities linked to water resource classification, with extensive stakeholder involvement, will result in a recommended Management Class for each of the various parts of a catchment and the consequent setting

42 of the Reserve and Resource Quality Objectives for each part. This will pave the way for a proposal on how water resource use will be allocated after the two parts of the Reserve have been set. Third, the ensuing information will be itemised and made available for publication and appeals. The information will include details of the Management Class, Reserve and other RQOs set for various parts of the catchment, any conditions and other details regarding compulsory licencing emanating from the allocation schedule, and any proposed new water resource structures or the revision of design and operating rules for existing ones. The fourth and final phase will be the monitoring, enforcement and evaluation activities that will provide for on-going management of the process.

Water availability taking into account ecological Water use requirements taking into account BHN and water requirements economic growth imperatives

Water balance / imbalance including need and

justification for strategic advocacy Stakeholder Inputs Stakeholder

Water resource

management options National Water Water resource

Resource Strategy classification Biodiversity overlays scenarios

Future use projections including Catchment Management water for equity, rural and Strategy (including ISPs) economic development needs Public Participation Public

Recommended Class Allocation Schedule

Compulsory Water New water Updated modeling Management Class, licensing conservation resource system and decision Reserve and Resource

Appeals and WDM development support tools Quality Objectives Publication, Publication,

Monitoring, evaluation and enforcement (water resource quality audit, source directed controls, and more)

Adaptive Adaptive management management

Figure 0.1 DWA’s plans for Integrated Water Resource Management, showing the positioning of the Management Class, Reserve and RQOs. WDM = water demand management. ISP = Internal Strategic Perspectives. BHN = Basic Human Needs.

Although Reserve assessments have been ongoing since the early 1990s, these are all labelled ‘preliminary’ as they are not yet set in the context of the consultative process of Water Resource Classification. Rather, the scientists have provided the volume, timing,

43 magnitude frequency and variability of water needed for maintaining the ecosystem of concern at various ecological health levels and then DWA has taken it upon itself to decide which of these will in fact be the ‘preliminary’ future condition. This has allowed them to move ahead speedily and make decisions on water licences. The Water Resource Classification System was promulgated in September 2010 and so classification, with substantial stakeholder involvement, is now proceeding two or three catchments at a time. Each classification procedure will lead to a Management Class and an Ecological Reserve (that is no longer preliminary) being set for each part of a catchment. These will be legally binding. Compulsory licensing is an innovative part of the NWA. Once the Management Class and Ecological Reserve are set, it will drive a complete re-think about who has a licence for the remaining water or water-resource use, and will inter alia redress past inequities in who has access to water. Compulsory licensing involves calling for licence applications from existing users (who should previously have registered their existing uses) and prospective new users, followed by the preparation of, and consultation on, a series of allocation schedules. This will culminate in the issue of water-use licences per catchment, not just for water abstraction but also for effluent disposal and other uses of water resources. At present, a nationwide process of validation and verification of present use is under way to provide the baseline information on which compulsory licensing will proceed6.

Figure 0.2 The compulsory licensing process

2.3 The scientific methods used to assess the Ecological Reserve South Africa initially tested international methods available for setting EFlows in the early 1990s (King and Tharme 1994) and then went on to develop its own approaches. The EFlows/Reserves had to meet the following specifications:  be legally defensible, since they had to serve as a basis for issuing legally valid water-use licenses  be scientifically defensible, and in line with IWRM  match administrative requirements, i.e. the information had to be provided to the licensing agencies in a format that could be used as a basis for drawing up water-use

6 The plans are good but execution is more of a problem. Validation and verification are slow and fraught with uncertainty. Hydrological water resource models are not set up for most catchments and licenses are presently issued with fair to poor understanding of how much water is available or how a potential upstream user would impact an actual or potential downstream user. Monitoring and enforcement are rudimentary.

allocation plans and catchment management strategies, and for setting individual water-use license conditions  provide conservative estimates of the water quantity and quality required to meet the Ecological Reserve, since they are intended to prevent irreversible degradation of water resources, in line with DWA’s protection policy  include options for rapid determinations in order to meet the projected demands for licences. Reserve assessments are done at one of four levels of complexity depending on the ecological and social importance of the water body and the level of potential conflict over water (Table 0.1). Guidelines exist to help DWA decide which level to use for any specific catchment. Various locally-accepted approaches can be used at any one level of complexity. For instance, the Desktop Method of Hughes and Hannart (see below) was originally used for all desktop assessments but DRIFT (see below) now has routines for all levels of complexity.

Table 0.1 Levels of determination of the Ecological Reserve, with estimates of the work involved Level of Resources Time Resolution of Field activities complexity required required results Desktop Low 2 days none Low field assessment of Rapid Low 2 weeks Low ecological condition Intermediate Medium 8 weeks detailed field assessment Medium Comprehensive High 32 weeks and data collection Medium/High

The initial EFlows assessment method developed in South Africa was the Building Block Methodology (King et al. 2000). This first identifies the desired future condition for the river (i.e. the trade-off point between water resource development and ecosystem protection) and then describes the flow regime that would be expected to achieve this. Because the South African government had no mechanism in the early 1990s for comprehensive stakeholder consultation, and thus was not able to identify this desired condition, the scientists were asked to recommend such a condition as well as then defining the flows to attain and maintain it. Time showed that the BBM had three major weaknesses. First, it was essentially prescriptive. A river condition is specified, and then the recommended flow regime to achieve it is described. The outputs do not lend themselves to negotiation, because effort is mostly directed to justifying a single flow regime, and information on the implications of not meeting it are not readily apparent from application of the method. Second, a range of stakeholders began to question why scientists were making the decision about the future condition of their rivers, when surely a more broad-based stakeholder process should be employed. Third, the BBM did not adequately address the impacts of flow changes on subsistence users. These social impacts are part of the costs of water-resource developments in many developing countries, but at that time were still rarely described in a structured way. The BBM was the fore-runner of a suite of more modern holistic approaches that has developed globally over the last two decades. Their main purpose is to provide neutral (neither pro-development nor anti-development) information that allows stakeholders to make more informed input to decision makers, and governments to make more balanced and informed decisions. The holistic approaches do this by providing scenarios of various future water management options for discussion and choice of a preferred way forward. Two main such scenario-based approaches have evolved in South Africa: DRIFT (Downstream Response to Imposed Flow Transformations: King et al. 2003, 2004, 2010; Box 1 and

Appendix 1) and HBSF (Habitat Flow Stressor Response: O’Keeffe et al. 2002; IWR 2004; Hughes and Louw 2010; Box 2). Both approaches are recognised by the South African government and their outputs are used in setting the Ecological Reserve. In parallel, a method that links closely with DRIFT has evolved for setting the Reserve for estuaries (Adams et al. 2002; DWAF 2008; Turpie et al. 2011). Methods in an earlier state of development are available for wetlands (Rountree and Malan in press), ephemeral rivers (Rossouw et al. 2005; Seaman et al. 2010), the water quality component of the Ecological Reserve (Scherman Consulting in prep.) and groundwater systems (Colvin et al. 2004; Parsons and Wentzel 2007). A suite of other techniques, guidelines and procedures also exist for: hydrological simulation and analyses, analysis of hydraulic habitat, strategic adaptive management (SAM), assessment of present ecological condition, assessment of ecological importance and sensitivity, catchment delineation in terms of ecoregions and geomorphological river zones, the biophysical descriptors of Resource Quality Objectives, assessment of socio-economic conditions, and more. All of these are described in some detail in King and Pienaar (2011). The two EFlow methods fit into the larger IWRM framework as providers of information on the current ecological state, sensitivity and importance of the water body, and on the required water allocation to maintain that condition or some alternative conditions that always include one ecological state higher and one state lower than present day (scenarios). One of these states/scenarios is recommended from an ecological perspective as the future state. DWA combines this input with other relevant information (including from consultation with stakeholders) before deciding on the Management Class. The Reserve and other RQOs are then set. These two EFlow methods have provided a body of knowledge on the links between flow and ecosystem condition. This led to creation of the Desktop Reserve model that provides a coarse set of rules for strategic planning purposes on the volume of water required to maintain rivers in the different hydrological regions of the country at specified ecological condition levels (Hughes and Munster 2000; Hughes and Hannart 2003). Many Reserves have been assessed and set at this level (Table 0.1).

The term EFlows is not often used in South Africa, partly because the law uses the term Reserve and partly because ‘flows’ are not necessarily an appropriate concept when allocating water to maintain wetlands, lakes and groundwater systems. Instead the term Ecological Water Requirement (EWR) is used to express the required water allocation required for ecosystem maintenance in each scenario. The term Reserve is subsequently used for whichever of these scenarios/water allocations eventually emerges at the end of the process of negotiation to maintain the agreed Management Class.

Box 1: Describing flow-biota-people relationships – DRIFT (details in Appendix 1) DRIFT predicts the ecological, social and economic consequences of proposed water- resource actions.

DRIFT uses the concept of response curves to describe flow-biota relationships. Response curves show the relationship between one responding indicator, such as a group of organisms that is perceived to respond to flow in the same way (in the graphic below, fish guild A) – and one driving indicator (in the graphic, minimum dry-season discharge). Present Day conditions are represented by the 100% fish abundance line and the red line superimposed on this that shows the recorded range of dry season low flows. The blue line (the response curve), with its uncertainty band (lighter blue lines) is a prediction, based on data and expert opinion, of how fish guild A would respond to years of higher and lower minimum dry-season discharges. Hundreds of response curves of this nature are created by the multidisciplinary team of specialists for one river to capture their understanding of the nature of its functioning, and housed in custom-built DRIFT-DSS software that can be queried on the predicted outcomes of any flow-related scenario of interest.

140

120

100

Abundance Abundance offish guild A

10 20 30 40 50 60 70 80 90 Dry Season, Minimum Five Day Flow m3/s

Box 2: Describing flow-biota relationships – Habitat Flow Stressor Response approach The Habitat Flow-Stressor Response (HFSR) approach addresses the problem of few data on flow-biota relationships by using hydraulic habitat as a surrogate for the biota and evaluating how this would change under a range of low-flow conditions.

It assumes that changes in habitat conditions (stress) will be reflected in a change in the biotas (response) as their degree of discomfort or damage changes with changes in low flows. It calculates a stress index for each relevant low flow, allowing conversion of flow duration curves to stress duration curves. It uses these to advise on flows for specified ecological conditions, or to evaluate low-flow scenarios in terms of the resulting ecological conditions. Floods are addressed using flood classes, with those required for each Ecological Category added to the low-flow requirement.

As at the end of 2010, 43% of the total area of South Africa had had some level of Reserve assessment done for surface waters and 59% of the country had been covered by groundwater Reserve assessments. Some of these scientific results have progressed further to decisions on Management Class (or its earlier rudimentary equivalent) being made and Reserves set that hold legal status; the rest are gradually moving through the planning and

48 decision making process within DWA. The problem that remains is operationalisation of these legal Reserves at regional level. The Reserve requirements promulgated at national level are not provided in a form that is easily transformed into management actions at regional level and local capacity to perform the work is low, so delivery of the right flows, and monitoring and enforcement of them, are largely not happening. The problem has been acknowledged and actions are being taken to address the situation but it is recognised that this could take some time (years) to resolve.

3. Comment on the range of issues identified in the UK for consideration in its Phase 1 final report

This section has been difficult to respond to because the issues have some specifically UK/Europe/WFD perspectives that this author may not fully grasp, and also because the issues were not presented as specific questions that could be focused on. For instance, Table 1 shown in the ToR is of percentage of natural flow allowable as an abstraction but it is unclear if this table is meant to describe the ‘appropriate flow regime in terms of environmental flow indicators (EFIs)’. A later email requested that the issues be looked upon as ideas, and comment be made regarding whether or not they are sensible, scientifically justifiable and workable. This is responded to for each issue below.

3.1 Natural versus present condition The issue/idea: A muted version of the natural flow regime may not be a useful target for highly modified rivers. The response: South Africa recognises that rivers may be in a degraded condition for many reasons other than flow regime changes and that setting suitable flows will have limited success in terms of ecosystem health if the other drivers in the catchment (such as land use, existence of a riparian belt of vegetation etc.) are not also managed appropriately. That aside, RSA does not try to achieve Good Ecological Status (GES) in all water bodies but rather recognises that different water bodies can be maintained at different levels of ecological health depending on the aspirations for that catchment. DRIFT acknowledges that the natural condition might not be known and that predicted changes from the present condition would be understood more easily as that is the condition that scientists and stakeholders see and measure. DRIFT therefore takes the present day flow regime and river condition as its starting point and describes how potential management options could alter this state (away from natural if a development is involved; toward what is thought to be natural if restoration is being considered) . Perhaps an option for the UK could be to take such a starting point and describe a number of scenarios – each with their flow regime and river condition. Several of these could achieve GES in different ways perhaps, and then government and stakeholders together could choose which one they want. This approach does suggest that it may not be possible to set a blanket set of criteria for the whole country (as per Table 1 in the ToR) but rather each river system/catchment has to be addressed individually. Conclusion: UK and South Africa have different starting points:  UK: what must I do to achieve GES?  South Africa: here are some options for possible futures for the river and its users – which one does government and stakeholders want (some could achieve something akin to GES in various ways, some will not)? The South African approach avoids the traps linked to reference condition and natural flow regimes. Both of these concepts are used in the EFlows assessment, and data on them are

49 produced, but they are not central to the process, other than in the Desktop Model, and are rather used as on-the-side information that guides thinking.

3.2 Environmental flow indicators to meet GES or GEP The issue/idea: Environmental flow indicators have not yet been defined for water bodies that must maintain Good Environmental Potential. The response: The South African approach is to develop a number of scenarios that describe different options for dam design and operation. One of these is chosen and its pattern of water releases becomes the EFlow downstream of the dam. Again, this requires that each situation be assessed individually rather than relying on country-wide numbers. Conclusion: Highly modified rivers can differ considerably in their permutations of water- resource structures. Setting EFIs for the downstream rivers would best be done river-by- river. Criteria to be considered when setting the flows could be set countrywide.

3.3 The use of flow duration curves The issue/idea: Flow duration curves lose the temporal sequencing of flows. The response: Flow duration curves can be very useful but can also be deceptively un- useful. A questionable use of them may be in planning. The annual FDCs for the natural, present day and future scenarios could be provided and then a statement made that the development linked to the future scenario will not have much impact because ‘it only changes the end bits of the FDC a little’ or ‘it only drops the middle of the FDC a little’. Such an analysis patently ignores the fact that it cannot be distinguished with any accuracy if low flows or floods would be impacted or, even worse, how the timing, duration and frequency of different magnitude flows would be affected. All of these characteristics of a flow regime are important in maintaining river health, with the links between different flows and ecosystem functioning now well recognised (King and Brown 2006). The problem of poor use of FDCs can be partially avoided by using monthly FDCs, which will probably reveal the key months when most flow change is expected. The UK issue on FDCs as stated in the ToR appears to be more to do with operationalisation of an agreed EFI than in planning. A possible way forward could be to create not only a FDC of the full environmental flow (as shown in the UK Figure 1) but also monthly FDCs of just the low flow parts of the EFlows hydrograph. The target could then be to aim for the 50th (or other) percentile each month, with some stated variation allowed. Conclusion: Monthly FDCs of the low flow part of the EFI could help guide appropriate EFlows.

3.4 Timing of abstractions The issue/idea: When should water be abstracted from the river for high-demand times in the dry season? The response: The UK issue is that most water is needed from a river during the hot dry season when river flows are lowest. It is suggested that higher abstractions could be allowed for particular periods, with lower or no abstractions at other times to allow ecosystem recovery. It is not clear if the suggested ‘particular period’ of higher abstractions would be in the wet season or at some restricted time in the dry season. If it was in the wet season, then – yes – modest amounts of flow could be abstracted with less impact than abstractions in the dry season. This water would have to be stored somewhere for use in the dry season. Off-channel storage is best for the river as sediment, chemical and thermal regimes will be less impacted than by a large in-channel water-storage structure.

If it was in the dry season, then the impact on the river would be far greater. River ecosystems have many plant and animal communities that appear and disappear at different times of the year, and any species that do exist throughout the year have different life stages in different seasons. The sets of species or life-stages are adapted to cope with specific environmental conditions. High abstractions during some part of the dry season could result in the riverbed becoming de-watered or at least in a drastic change in hydraulic habitat and living space for aquatic organisms. This could result in a decline in numbers of individuals thereby disrupting the food chain and, if repeated year after year, eventually the disappearance of species and perhaps whole communities. Some species would return (or appear for the first time) after flow increased again, but those species that would tend to persist over time would be pioneering, robust, opportunistic ones with short life cycles that appear very soon when conditions improve and go through their life cycles quickly. For instance, aquatic midges with a life cycle of three weeks could persist over the years while dragonflies, which live for a year or more as aquatic larvae, might gradually disappear. The characteristics of pioneering species – fast response after disturbance – also makes them potential pest species (e.g. some algae, mosquitoes) that could proliferate to nuisance proportions in the absence of the competing more sensitive species. Conclusion: I do not think a river community of plants and animals can survive intact an annually repeated period of high abstractions in the dry season. There will be plant and animal community changes that will usually encompass the loss of sensitive and rare species and the sporadic appearance - and in some cases proliferation - of hardy potentially pest species. Biodiversity will decrease.

3.5 Climate change The issue/idea: Climate change could bring new reference conditions as even pristine systems will be affected. The response: South Africa now routinely assesses ordinary scenarios of management options and then the same scenarios with climate change super-imposed. The understanding this brings is incorporated into the setting of the Management Classes for a basin. When the Management Classes, Ecological Reserves and RQOs have been set, it is with the understanding that the whole process will be re-done every 5-10 years, which implies that a new ‘natural’ flow regime could be set in the future. If climate change is predicted to have substantial impacts on rainfall and temperatures in the UK, then it might be necessary eventually to re-assess not only the baseline (reference) flows but also the abstraction licences themselves in terms of whether or not water is being put to best use. This re-assessment might happen by default if abstraction allowances are proportionally pulled down: farmers might shift to other crops (as has happened in the Western Cape) and municipalities may introduce and enforce water demand management more strictly through increased tariffs, mending burst pipes etc. (as has happened all over RSA). Conclusion: So, yes, the UK reference conditions for flow should periodically be re-visited and amended using simulations that incorporate climate change and/or measurements of rainfall/runoff. And yes, future abstraction allowances could then be proportionally reduced, but maybe also a wider reaching assessment of best water use should be considered.

3.6 How to deal with uncertainty The issue/idea: Recommended environmental flows encompass various degrees of uncertainty. The response: See last sentence in Response part of Section 3.3. We are investigating:  Setting the Ecological Reserve using DRIFT or similar

 Using the long-term simulated hydrograph (ex DRIFT) of the agreed Reserve flows to produce monthly FDCs of the low flow part of the hydrograph only  Identifying the 50th percentile on each FDC as the low flow to aim for in any one month  Establishing a band around the 50th percentile that will differ from river to river and month to month, within which flow could fluctuate from day to day within that month. This would allow for some flexibility regarding daily climate and water-use activities as well as address the uncertainty in the Reserve values recommended. Conclusion: This kind of operationalisation of environmental flows is in its early days. King and Pienaar (2011) give further details of some work of this kind in South Africa.

3.7 Thresholds The issue/idea: Current EFIs do not reflect hydro-ecological thresholds. The response: To a large extent, based on our present knowledge, it seems that with land and water development river ecosystems gradually decline in condition rather than going through a series of thresholds where conditions abruptly worsen. There are one or two obvious exceptions to this where major state changes can occur abruptly due to a single significant change in flow. The most obvious of these are:  complete de-watering of the river bed  loss of river-floodplain connectivity  closing of river mouth and isolation of estuary from sea. At the moment DRIFT and HFSR do not specifically identify when such state changes occur, but the various scenarios they produce will show the degree to which such state changes would be achieved. In parallel, work continues on developing and implementing Strategic Adaptive Management (SAM) as a means for defining management goals, measuring success in achieving them and modifying management as needed (learning by doing). SAM incorporates the concept of thresholds of potential concern (TPC), which are auditable limits on any ecosystem indicator of interest. TPCs form the basis of the monitoring programme that feeds back into SAM and could include the major state changes mentioned above and many lesser thresholds such as a percentage loss of riffle habitat, species abundance or biodiversity. Thresholds (state change or lesser) could be built in to DRIFT simply by adding them as an indicator so that each scenario specifies more clearly whether or not they are at risk. Conclusion: If the UK were to consider inserting thresholds into their EFIs, then they may have to move toward more river-specific EFIs as the characteristics of floodplain connectivity, riffles etc. would differ from river to river.

3.8 National environmental accounting and sacrificial rivers The issue/idea: National environmental accounting would allow a countrywide approach to water resource development that recognises sacrificial and conserved rivers. The response: Some South African environmental flow assessments have recognised at the basin level the concept of sacrificing some part of a river system in order to keep development low on the remainder of the system. Such a practice is not part of a national accounting system as yet. However, the newly forming Catchment Management Agencies (CMAs), of which there will be nine in South Africa, will be able to extend the concept of sacrificial and conserved rivers beyond basins to wider regions, if they wish. I suspect it will be a difficult concept to put into practice because memories are short and decision makers

52 are replaced often, so soon after a river has been ‘sacrificed’ focus might still turn to developing the so-called conserved ones. One of the problems with sacrificial rivers is that no two rivers are alike. This is an excerpt from the chapter Inland water ecosystems in Grafton and Hussey (2011). … it is no accident that each community of animals and plants occurs where it does. Samples of riverine vegetation (Reinecke et al. 2007), aquatic invertebrates (King and Schael 2001) and periphyton (algae on rocks) (J. Ewart-Smith, University of Cape Town. pers. comm.) from Western Cape rivers in South Africa, for instance, are each easily distinguishable from those collected from neighbouring catchments, rivers or part of rivers. In other words, each sample ‘knows’ which river and catchment it came from, suggesting that every catchment and river functions slightly differently. We have called this phenomenon ‘catchment and river signatures’ and there seems to be no single environmental driver, but rather a complex of interactions operating over geological time that determines which group of species lives where (Schael and King 2005). Clearly, it cannot be assumed that all rivers within any one ecoregion are much the same and thus that a random selection of them can be sacrificed to development with no implications for biodiversity in all its forms. At present we manage catchments mainly as water-supply entities, with little or no recognition or understanding of their individuality or of the complexity of their structures and functioning. At the very least, conservation planning (e.g. Groves et al. 2002) needs to be inserted as an overlay to national and regional water-development plans to guide which water ecosystems should be preserved and which could be degraded to a managed extent where other goals such as food production are a priority.

My main point here is that even within one ecoregion of similar rivers there are differences that we can describe but not yet explain. Sacrificing some of these rivers and assuming other nearby ones will adequately represent them may be erroneous. We may be losing kinds of ecosystem functioning that we do not yet recognise and understand. The same applies if the sacrificial rivers are in different ecoregions to the conserved ones. Conclusion: Recognising sacrificial and conserved rivers is an idea deserving of further investigation, with a view to better understanding diversity from gene level to landscape level before naming sacrificial rivers. They would then have to be established in a way that provides adequate long-term protection for the conserved ones.

3.9 Real-time flow data The issue/idea: In the UK, EFlows are monitored at river flow gauge stations. For better monitoring the continuous flow at all points of the river system needs to be known in real time. The response: Many parts of South Africa have water resource models up and running. For example, one of the new CMAs, the Inkomati Catchment Management Agency, has daily flow data of the whole river system shown on a website. It alerts farmers to when they must reduce or stop abstractions based on near real-time flows. The contact is ICMA Executive Manager Brian Jackson ([email protected]). One of their features to look out for is CROC (the Crocodile River Operations Committee and their Decision Support System for ‘evaluating, recording, making and auditing decisions’). Conclusion: It is probably impractical to expect flows to be recorded above and below all abstraction points. Setting up some kind of model that can show the general pattern of flows in a river in near real time will give a general idea of whether there is a problem. Problem areas can be zoned in on with on-the-ground detailed monitoring and enforcement.

3.10 Capping unnaturally high flows The issue/idea: The release of water from dams for downstream use, or the discharge of effluents into rivers can cause abnormally high flows. These have the potential to be as damaging to the ecosystem as flows that are abnormally low.

The response: The monthly low-flow FDCs mentioned in Section 3.6 provide a means of defining the upper and lower limits per month. The kind of catchment-wide computerised flows set up by ICMA (Section 3.9) capture dam releases and effluent discharges as well as abstractions. Conclusion: The EFI can and should define upper limits to flow as well as lower limits. These can be monitored in near real-time.

3.11 Subsequent questions from client Question 1: In South Africa, once the flow requirements are set do they vary over time, for example year to year? This could be because natural flow variability means more or less water is available (or perhaps the environmental reserve is fixed). Our experience in the UK is that a range of species live in any river and do not all have the same flow requirements. One year is good for one species, another year is good for another species. So some natural year to year variability could be a good thing. Response 1: The Ecological Reserve is set as a percentage of natural flow for every day/month/year. Thus, it varies over time from days to decades, reflecting drier and wetter periods. As King and Pienaar (2011) explain “The Reserve flow on any day is calculated by ascertaining what the natural flow would have been on that day and reading off the equivalent percentile value on the Reserve flow duration curve”. They add “Though simple in concept, it is more complex to implement in real time as it requires knowledge of the natural flow that would be occurring at any specific time and place”. Techniques to estimate natural low flows in developed catchments that are being developed include (King and Pienaar 2011):  Use a neighbouring undeveloped catchment that is measuring natural flow as a guide to natural flow in the catchment of concern  Set up a near real-time hydrological model of the rainfall-runoff process  Estimate natural flow from observed flows and a good knowledge of water use in the catchment. Question 2: Does South Africa use a different EFlows method/approach for direct abstraction than for releases from reservoirs? Response 2: The question confuses EFlows assessment and implementation. South Africa uses the kinds of EFlow assessment methods described above to inform the SETTING of an EFlow (via the Management Class and the Ecological Reserve) for the different parts of a river system. How this legally-required Reserve flow is then ACHIEVED is an implementation problem. Most river systems will have a combination of dams and direct abstraction points on them, and water managers have to work out, at the basin level if possible, the releases from dams, the allocation of water-use abstraction licences and the conditions linked to each that they believe will achieve the day-by-day Reserve in the river. And then they have to MONITOR and MANAGE the process to ensure success.

4. South African EFlow methods that could inform the setting and implementing of EFlows in the UK

4.1 Setting of environmental flows The UK efforts in this regard seems to be confounded by the WFD rather than helped by it as a lot of effort appears to be invested in trying to define a possibly unknown ( and possibly unwanted) reference condition. DRIFT uses the present condition as its starting point and

54 predicts how that would change with any proposed management action (development or restoration). The reference condition concept is used somewhat loosely in DRIFT, in that the present ecological state (PES) is assessed on a scale of A to E, so some awareness of an assumed natural condition (i.e. A) is employed. The future scenarios then describe how this PES would change. This works quite well and experienced ecological teams seem to agree quite easily on the PES of a river.. In order to ensure that not all rivers are chosen to be maintained at the lowest possible ecological level, South Africa has a country-wide atlas of the locations of FEPAs (Freshwater Ecosystem Protected Areas); the FEPAs are presently being factored into the decision making on Management Classes. Perhaps, rather than using the reference condition in the UK, rivers could be assessed on their degree of functionality. Criteria could include: are floodplains connected; are seasonal flow differences maintained; are all size floods still present and so on. Rivers that meet most criteria are in GES; and GES could actually then take several different forms, to suit the requirements of different stakeholders and different parts of the country7. If it were to come to choosing which kind of GES is required, DRIFT can produce dozens of scenarios quickly once it is set up with the initial specialist inputs (in the form of response curves).

4.2 Implementing environmental flows Many see EFlows implementation as simply ensuring the right flow moves down the river but it is much more complex. Key aspects of successful implementation in South Africa are seen as: 1. Development of the appropriate policy, legislation and transboundary basin agreements 2. Re-organisation of water management institutions to meet the requirements of the new Act 3. Structured and continual engagement with stakeholders 4. Development of holistic flow-assessment methods 5. Design of new kinds of infrastructure and operating rules to deliver and monitor environmental flows 6. Management of water quality, and of the quality of the ecosystems that form the water resource base of the country 7. Development of catchment management strategies and regional regulatory mechanisms for the authorisation of water resource use 8. Creation of awareness among government and other stakeholders 9. Continual investment in research and capacity building 10. Delivery of the environmental water allocation 11. Monitoring , enforcement and adaptive management. DRIFT contributes to almost all of the above items.

At present, South Africa has made significant progress in items 1-5 and 8-9. These are all reported on in King and Pienaar (2011).The remaining items needing more vigorous attention are:  Management of water quality, and of the quality of the ecosystems that form the water resource base of the country  Development of catchment management strategies and regional regulatory mechanisms for the authorisation of water resource use  Delivery of the environmental water allocation

7 This author and Cate Brown are presently preparing a paper on this concept.

 Monitoring , enforcement and adaptive management.

All of these remaining items need to be achieved at the regional/local level. There is a general recognition within South Africa now that while the other items will continue to be developed, effort should now move to operationalising the agreed EFlows (Reserve) at the local level , that is, focusing on items 6,7, 10 and 11.

5. Advice on data collection and research

As at 2010 South Africa has produced the following deliverables specifically on EFlows since 1990:

 172 peer reviewed scientific articles  32 chapters in books  1 complete book  131 technical refereed research reports  196 other publications  59 invited/keynote presentations at international conferences  126 other contributions to international conferences  170 contributions at national conferences, meetings  54 postgraduate theses  22 national training courses

This was achieved through the commitment by the Department of Water Affairs to establish ecosystem protection measures as per the NWA and a consistent, very strong support from the national Water Research Commission. Senior water professional in universities were called in early in the process to contribute to EFlows workshops where they quickly learnt that they did not have quite the understanding and data needed. This triggered multi-year research programmes and post-graduate theses, while all that they were learning fed into under-graduate teaching and a range of training courses. Twenty years on we have an excellent national base of informed water professionals who work hands-on with the water managers. My advice on research is to strengthen the scientist-manager link as much as possible so that the scientists can see where and how they can help. Organise a dedicated source of research funding so that academics, consultants and post-graduates are attracted to enter the field. Make presentations at every conference possible within the country, including ecologists at engineering conferences and vice versa, in order to raise awareness. Liaise with universities to launch relevant post-graduate programmes. We have drawn up specific guidelines for EFlows data collection and analysis, which I can forward if wished. Basically, all data have to be collected with their links to flow or inundation clearly shown (e.g. if riparian trees form seeds in May, what is the link with flow; if fish guild A migrates along the river in October, what is the link to flow; if fish guild B moves on to floodplains in February what is the link with inundation). If this link is not made in every case, then it becomes impossible to predict how these ecosystem elements would be affected by flow/inundation changes and so also difficult to justify the flows/inundations you request as an EFlow. Other research that is needed is, to the extent that it is not yet done:  how to delineate catchments (split them into smaller homogeneous units), which guides the choice of representative EFlow sites, and defines the area over which any data set can be extrapolated)  how to estimate present ecological condition and present ecological sensitivity and importance  develop a data management system to collate all available knowledge and data on the river system in a form that allow predictions of development/restoration driven changes (this is what DRIFT does)

 a targetted ‘parsimonious’ (i.e. not a shopping list) monitoring system for EFlows monitoring  how to valuate ecosystem services so that the value of the present ecosystem and costs of developing its water resources can be explicitly described  how to transform the technical information into an accessible form for politicians, stakeholders and decision makers.

6. References

Adams J.B., G.C. Bate, T.D. Harrison, P. Huizinga, S. Taljaard, L. van Niekerk, E. Plumstead, A.K. Whitfield and T.H. Wooldridge 2002. A method to assess the freshwater inflow requirements of estuaries and application to the Mtata Estuary, South Africa. Estuaries 25(6B): 1382-1393. Colvin C., L. Cave and I. Saayman 2004. A functional approach to setting resource quality objectives for groundwater. Water Research Commission Report No. 1235/1/04. Water Research Commission, Pretoria. Department of Water Affairs and Forestry 2008. Water Resource Protection and Assessment Policy Implementation Process. Resource Directed Measures for protection of water resources: Methodology for the Determination of the Ecological Water Requirements for Estuaries (Version 2). Department of Water Affairs and Forestry, Pretoria. Grafton R.Q. and K. Hussey (eds). Water resources planning and management. Cambridge University Press, Cambridge UK. 777pp. Hughes D.A. and F. Munster 2000. Hydrological information and techniques to support the determination of the water quantity component of the Ecological Reserve for rivers. Water Research Commission Report No. 867/3/2000. Water Research Commission, Pretoria. Hughes D.A. and P. Hannart 2003. A desktop model used to provide an initial estimate of the ecological instream flow requirements of rivers in South Africa. Journal of Hydrology 270(3-4): 167-181. Hughes D.A. and D. Louw 2010. Integrating hydrology, hydraulics and ecological response into a flexible approach to the determination of environmental water requirements for rivers. Environmental Modelling & Software 25(8): 910-918. IWR Source-to-Sea (eds) 2004. A comprehensive Ecoclassification and habitat flow stressor response manual. Project No. 2002-148. Department of Water Affairs and Forestry, Pretoria. King, J. and H. Pienaar (eds) 2011. Sustainable use of South Africa’s inland waters: A situation assessment of Resource Directed Measures 12 years after the 1998 National Water Act. Water Research Commission Report No. TT 491/11. Water Research Commission, Pretoria. ISBN 978-1-4312-0129-7. 259 pp. King, J.M. & R.E. Tharme 1994. Assessment of the Instream Flow Incremental Methodology and initial development of alternative instream flow methodologies for South Africa. 590 pp. Water Research Commission, Pretoria. ISBN 1-86845-056-2. King J.M., R.E. Tharme and M.S. De Villiers (eds) 2000. Environmental Flow Assessments for Rivers: Manual for the Building Block Methodology. Report No. TT 131/00. Water Research Commission, Pretoria. King, J.M. C. A. Brown & H. Sabet 2003. A scenario-based holistic approach for environmental flow assessments. Rivers Research and Applications 19 (5-6): 619-639. King, J.M., C.A. Brown, B.R. Paxton and R.J. February 2004. Development of DRIFT, a scenario-based methodology for environmental flow assessments. Water Research Commission Report 1159/1/04. Water Research Commission, Pretoria. 159 pp. King, J.M. & C.A. Brown 2006. Environmental flows: striking the balance between development and resource protection. Ecology and Society 11(2): 26 (online). King, J.M. and C.A. Brown 2010. Integrated Basin Flow Assessments: concepts and method development in Africa and South-east Asia. Special Issue of Freshwater Biology 55(1):127-146. O'Keeffe J., D. Hughes and R.E. Tharme 2002. Linking ecological responses to altered flows, for use in environmental flow assessments: the Flow Stressor-Response method. Verh. Int. Ver. Limnol 28: 84-92. Parsons R. and J Wentzel 2007. Groundwater resource directed measures manual. Water Research Commission Report number TT 299/07. Water Research Commission, Pretoria. Rossouw L., M.F. Avenant, M.T. Seaman, J.M. King, C.H. Barker, P.J. du Preez, A.J. Pelser, J.C. Roos, J.J. van Staden, G.J. van Tonder and M. Watson 2005. Environmental water

59 requirements in non-perennial systems. Water Research Commission Report No. 1414/1/05. Water Research Commission, Pretoria. Rountree M.W. and H. Malan (eds) in press. Rapid Ecological Reserve Determination methods for wetlands (Version 2.0). Joint Department of Water Affairs and Water Research Commission Study. Water Research Commission, Pretoria. Submitted to DWA in 2010. Scherman Consulting in prep. Methods for determining the water quality component of the Ecological Reserve. First draft submitted to DWA in 2008. Seaman M.T., M.F. Avenant, M. Watson, J.M. King, J. Armour, C.H. Barker, E. Dollar, P.J. du Preez, D. Hughes, L. Rossouw, and G. van Tonder 2010. Developing a method for determining the Environmental Water Requirements for non-perennial systems. Water Research Commission Report No. TT459/10. Water Research Commission, Pretoria. Turpie J., S. Taljaard, J. Adams, L. van Niekerk, N. Forbes and B. Weston 2011. Methods for the determination of the Ecological Reserve for estuaries. Version 3. Water Research Commission Draft Report No. K5/1930. Water Research Commission, Pretoria.

Appendix A1: Detailed explanation of DRIFT

Background DRIFT was developed in South Africa in response to the shortcomings recognised in the Building Block Methodology. It had become clear that a more appropriate approach would be one of scenario development and analysis. This would provide both government and other stakeholders with a range of options to consider rather than the prescription of one future. Purpose In 1997, DRIFT began development through application in real water resource projects in South Africa and Lesotho. It focuses on an individual or complex of inland aquatic ecosystem(s) (usually at the scale of a river basin), where it is used to create scenarios of any management option that could affect the quantity of water in that/those ecosystem(s). The ecosystem(s) can be perennial or non-perennial rivers, wetlands, floodplains or groundwater, or any combination of these, and the water quantity change can be caused by abstractions, diversions, construction of dams, changes in operating rules of infrastructure, change in land use such as deforestation, or similar. Thus, DRIFT can address small and large catchments, potential new developments, potential changes in current practices or potential restoration projects. An allied estuarine approach with DRIFT features can provide similar information for estuaries. In this explanation, rivers are focused on for simplicity. Underlying concept DRIFT is essentially a knowledge-management tool, allowing data and knowledge to be used to their best advantage in a structured process (King et al. 2004). Its central rationale is that different parts of the flow regime maintain the river ecosystem in different ways (

Table A1.1). Removal of one part of the flow regime will affect the river differently than removal of another part. Furthermore, it is assumed that:  it is possible to identify and isolate these different parts of the flow regime within a long-term hydrological data set of daily flows  it is possible to describe in isolation the probable biophysical consequences of changing the present day nature of any one of these parts  the parts of the flow regime and their linked consequences can be re-combined in various ways, to describe the river condition of any flow regime of interest  the social and socio-economic impacts of each river condition can be described. Moving from concept to practice, DRIFT outputs scenarios that predict how the hydrological, geomorphological, chemical, botanical, zoological, social and socio- economic nature of the aquatic system could change from the present day condition, and also provides a measure of whether that change will take the ecosystem closer to or further from its natural condition. The mix of disciplines involved in any one investigation can vary from study to study. National or regional economic information is generated outside DRIFT for each scenario if needed, and the combined predictions of change are used in stakeholder consultations, to illustrate possible pathways into the future, and in eventual decisions on water and land management. The scenarios also provide the foundation information on which monitoring and adaptive management activities can be based.

Table A1.1 Different kinds of river flow, and their importance to ecosystem functioning (King et al. 2004) Flow Importance to ecosystem Low flows These are the daily flows that occur outside of high-flow peaks. They define the basic hydrological nature of the river: its dry and wet seasons, and degree of perenniality. The different magnitudes of low-flow in the dry and wet seasons create more or less wetted habitat and different hydraulic and water-quality conditions, which directly influence the balance of species at any time of the year. Small Small floods are ecologically important in semi-arid areas in the dry season. floods They stimulate spawning in fish, flush out poor-quality water, mobilise and sort gravels and cobbles thereby enhancing physical heterogeneity of the riverbed, and contribute to flow variability. They re-set a wide spectrum of conditions in the river, triggering and synchronising activities as varied as upstream migrations of fish and germination of riparian seedlings. Large Large floods trigger many of the same responses as do the small ones, but floods additionally provide scouring flows that influence the form of the channel. They mobilise coarse sediments, and deposit silt, nutrients, eggs and seeds on floodplains. They inundate backwaters and secondary channels, and trigger bursts of growth in many species. They re-charge soil moisture levels in the banks, inundate floodplains, and scour estuaries thereby maintaining links with the sea. Flow Fluctuating discharges constantly change conditions through each day and variability season, creating mosaics of areas inundated and exposed for different lengths of time. The resulting physical heterogeneity determines the local distribution of species: higher physical diversity enhances biodiversity.

Detail of operating procedures DRIFT consists of several modules, some of which operate outside the model but provide input into it and some of which operate internally. The DRIFT process of scenario creation is initiated by hydrological information emanating from an external hydrological systems model (Figure ). This model simulates three or more daily hydrological data sets for many consecutive years for the selected number of representative sites along the river system. The three or more time series are of naturalised (historical flow), present day flow and the flows under one or more scenarios. The daily flow sequences are then converted into summary statistics of flow indicators that have been chosen by the ecologists. Early flow indicators used represented just changes in the magnitude of low flows or the number of floods (Table ), but later ones recognized changes in the timing of different flows also (

Table A1.2). These latter indicators were developed for studies of large flood-pulse rivers with extensive floodplains but have since also been successfully used for smaller rivers with flashier hydrographs. Each flow indicator has a value for each study site under each scenario, allowing stakeholders and decision makers to easily understand the flow conditions they would experience with different management options (Table ).

Figure A1.1. The process of scenario creation in DRIFT. Shaded area is the DRIFT Decision Support System, custom built for each river system.

Table A1.2 Example of early DRIFT flow indicators: summary data for the natural state and for four scenarios of increasing development at Site 2 on the Malibamatso River downstream of Katse Dam, Lesotho (Metsi 2000; 2002). Flow indicators Natural Minimum Design Extra Treaty degradation limitations scenario Wet season low flows (m3 s-1) 0.05-30.85 0.07 – 25.00 0.07 – 0.00 – 0.50 – 0.50 1.90 1.90 Dry season low flows (m3 s-1) 0.08 – 0.05 – 9.00 0.05 – 0.00 – 0.50 – 0.50 23.01 1.20 1.20 Class 1 floods (# annum-1) 8 3 3 2 1 Class 2 floods (# annum-1) 2 2 1 0.5 0 Class 3 floods (# annum-1) 2 2 2 0.5 0 Class 4 floods (# annum-1) 1 1 0 0 0 1:2 floods (present/absent) Present Present Absent Absent Absent 1:5 floods (present/absent) Present Present Absent Absent Absent 1:10 floods (present/absent) Present Present Absent Absent Absent 1:20 floods (present/absent) Present Present Present Present Present Mean annual runoff (Mcm) 554 367 184 97 22 % natural mean annual runoff 100% 66% 33% 18% 4% Total System Yield (m3s-1) n/a 18.3 22.8 25.2 26.8 Class 1-4 floods are intra-annual, with Class 1 floods being the smallest, and the larger floods are inter- annual with return periods of 2,5,10 and 20 years.

Table A1.3 Ecologically-relevant flow indicators adopted for EFlows assessments of large flood-pulse rivers, which incorporate timing as well as magnitude of flows Flow indicator Units Mean annual runoff Millions of cubic metres Dry season onset Calendar week Dry season duration Days Dry season minimum flow m3s-1 Flood season onset Calendar week 3 -1 3 -1 Flood season peak (m s ) m s Flood season volume (Mcm) Millions of cubic metres Flood season duration (days) Days Flood type 1 to 4

Table A1.4 Values for one of the indicators (Dry season minimum flow) recognised in

Table , for Present Day (PD) and three investigative planning pathways with Low, Medium and High levels of water resource development on the Okavango River system (King et al. 2010)

Site PD Low Medium High Comment (m3s-1) (m3s-1) (m3s-1) (m3s-1) 1 12 0.4 0.3 0.3 All Scenarios similar. Drastic drop from PD Minimum flow would drop to 50% (Low), 38% (Medium) of PD and then under High increase 2 32 16 12 24 to 75% because of dam releases in the dry season Decline through Low and Medium to 43% of 4 35 20 15 19 PD then increases for High to 54% Progressive decline from PD to very large 5/6 114 101 93 21 drop for High: 89%, 82%, 18%

The flow indicator summary statistics for each scenario/site are entered into the DRIFT database, where they link into the biophysical module. This consists of hundreds, possibly thousands, of response curves drawn by the discipline specialists. Each response curve describes, to the best of the specialists’ understanding, the relationship between one flow indicator and one biophysical indicator at one representative site8. The biophysical indicators are chosen by the specialists and must be attributes of the ecosystem that respond to flow changes ( Table ). The response curves are drawn by the discipline specialists, using best available knowledge, in workshops led by experienced DRIFT process specialists (Figure ). Figure illustrates the specialists’ understanding of how the abundance of Fish Guild A will increase or decrease in abundance on a yearly basis if the minimum dry season discharge is greater or less than the median value. Abundance is on a scale of change from 0 to 5 (Table ), which allows the specialists to express their understanding of probable change from present day in a semi-quantitative way if data are few. These ratings can be converted into percentage changes from present day for use in economic calculations, for instance, if required (Table ). The outcome of the link-up between flow indicators and biophysical indicators is an indicator- by-indicator description of how the river ecosystem is predicted to change under the scenario posed (Figure ). The next module uses these outputs to predict how human users of the river’s resources would be affected by the changes in the river (Figure ). Social indicators that represent the links between the river ecosystem and its human users (Table A1.5) are chosen and response curves drawn, in the same way as in the biophysical module.

Table A1.5 Examples of ecosystem indicators used in an EFlows assessment for the Neelum River in Pakistan (Southern Waters (C.A. Brown) 2011) Discipline Indicator 1 Average depth over riffles Hydraulics 2 River width 1 Embeddedness of riffles 2 Depth of pools Geomorphology 3 Area of backwaters 4 Overbank sedimentation Water Quality 1 Dilution of pollution loads

8 More recent applications also include response curves linking biophysical indicators to each other, e.g. the response of a fish guild to changes in a vegetation indicator

2 Temperature (diel range) 1 Algae Vegetation 2 Marginal vegetation 3 Natural flood terrace vegetation Ephemeroptera, Trichoptera and Plecoptera (EPT) 1 Score Macroinvertebrates 2 Simuliidae 3 Other flies and midges 1 Brown trout 2 Tibetan Snow Trout Fish 3 High Altitude Loach 4 Kashmir Hillstream Loach 5 Himalayan Cat Fish

-1

-2

-3

-4

Abundance of Fish Guild A A Guild Guild Fish Fish of of Abundance Abundance

-5

0 0

10 10 20 20 30 30 40 40 50 50 60 60 70 70 80 80 90 90

100 100 110 110 120 120 130 130 140 140 150 150 160 160 170 170 180 180 190 190 200 200 Minimum dry-season discharge (cumecs)

Figure A1.2 An example of a response curve – in this case of the relationship between the abundance of Fish Guild A and the magnitude of dry season minimum discharge. The open square indicates the median value of present day dry season minimum discharge (100 m2s-1) and the present day status of Fish Guild A (always taken as zero) (Brown 2011), and the understanding reflected is that Fish Guild A will decline in abundance if the dry season minimum flow increases above about 150 m2s-1 or decreases below about 80 m2s-1.

Table A1.6 DRIFT ratings of change, their associated description of change and the conversion to percentage change from present day – a negative score means a loss in abundance relative to present day, a positive means a gain Ratings of Severity of change change % abundance change 5 Critically severe 501% gain to ∞ up to pest proportions 4 Severe 251-500% gain 3 Moderate 68-250% gain 2 Low 26-67% gain 1 Negligible 1-25% gain 0 = Present day None no change -1 Negligible 80-100% retained -2 Low 60-79% retained -3 Moderate 40-59% retained -4 Severe 20-39% retained -5 Critically severe 0-19% retained includes local extinction

Table A1.7 Social indicators used in an EFlows assessment for the Okavango River system (King et al. 2010) Social indicator Household income — fish Household income — reeds Household income – floodplain gardens Household income and wealth – livestock Household income – tourism Potable water/water quality Wellbeing/welfare from intangible river resources Indirect use e.g. carbon sequestration, groundwater recharge, flood attenuation Non-use e.g. existence, bequest, and option values for preservation

Outside of the DRIFT application and parallel to it, an assessment of the wider consequences of each flow scenario should be done (Figure ). This would provide information on each scenario in terms of, for instance, the loss or gain of irrigated agricultural land, the generation of hydropower, the potential for industrial and urban development and the cost of water to offstream users. The combined output of this and the DRIFT application is three streams of information representing the three columns of sustainable development: ecological integrity, social equity and economic wealth (Figure ), provided both as technical, quantitative predictions and, where wished, in a more accessible, user-friendly narrative format. The three streams of information are given equal weight in the information provided to governments and other stakeholders.

Software developed

DRIFT Software has developed over a period of about ten years, driven by the need to standardise and streamline data manipulation. The most important software products are as follows, most being Excel based at present but with the whole DSS is presently being written in Delphi.  DRIFT-SOLVER distributes a specified volume of river flow into an annual pattern of different magnitude flows that optimises river condition (Brown and Joubert 2003).  DRIFT-HYDRO converts simulated daily flow time series into ecologically relevant summary statistics of selected flow indicators (Brown et al. 2005).  The DRIFT Decision Support System (DSS), which links hydrological system model outputs with ecological consequences of flow regimes and socio-economic outcomes, to produce predictions of ecological and social change for multiple scenarios (many examples – see next section).  The DRIFT DSS in Delphi – under construction, due for completion in 2012 (Beuster et al. 2008).

Together, these software packages allow: the hydrological simulations to be transformed into statistics that DRIFT can use; the indicators to be selected and the response curves drawn and calibrated; the predictions of biophysical and social change detailed; and the overall change in integrity of the ecosystem to be predicted. Detailed guidelines are available for all data collection and analysis steps along the way, as are user manuals for the software.

History of Application DRIFT began development in South Africa in about 1996, and has evolved through numerous applications both within its country of origin and internationally. It works equally well in flood-pulse rivers or those with flashy hydrographs and is presently the foundation of an environmental flow method being developed for ephemeral rivers as well as one input to the design of a method for estuaries. Within South Africa, it is one of the two main methodologies used for setting the Ecological Reserve (of water for ecosystem maintenance) as required by the 1998 National Water Act. DRIFT has been used to complete EFlows assessments for the South African Department of Water Affairs of the:  Palmiet River  Breede River  Olifants and Doring Rivers  Groot Brak River  Gwaing River  Maalgate River  Kaaimans River  Goukamma River  Berg River (lower).

Internationally, it has been and is being used mainly in Africa and Asia, and is specifically designed to address the conditions (possibly data poor) and needs (full ecosystem health and subsistence use) of these areas, although it works equally well in data-rich and developed world situations:  1997-8: used in an early form to assess the predicted impacts of the Lesotho Highlands Water Project (Katse and Mohale Dams)  2004-7: used in partial form for the Lower Mekong Basin (Cambodia, Lao PDR, Thailand, Viet Nam), funded by the World Bank and the Mekong River Commission  2006: the Zambezi Delta (Mozambique), funded by the Liz Claiborne and Art Ortenberg Foundation, the International Crane Foundation and the Carr Foundation  2007: the Mzingwane River (Zimbabwe), funded by IUCN  2007: the Phuthiatsana River (Lesotho), funded by Lesotho Department of Water Affairs  2006-9: the Pangani Basin (Tanzania), funded by IUCN and the Tanzanian government  2008-9: the Nile River (Sudan), funded by the Sudanese Dams Implementation Unit  2008-10: the Okavango Basin (Angola, Namibia, Botswana), funded by GEF/UNDP  2009-10: the Lower Zambezi River (Mozambique), funded by Riversdale Mining  2009-10: the Kunene River (Namibia and Angola), funded by the Angolan and Namibian Governments through the Permanent Joint Technical Commission  2010-11: the Neelum River (Pakistan), funded by the Government of Pakistan  2011: the Huaura River (Peru), funded by the International Finance Corporation  2012: the Neelum-Jhelum River in Pakisatan, funded by the Government of Pakistan.

DRIFT has been recognised as a good practice methodology by the World Bank, IUCN, OKACOM, and the South Africa and Tanzanian governments.

Strengths Some of these strengths are shared with most holistic EFlows approaches, but the items in bold appear to be stronger in DRIFT than in comparable international methods. All the strengths are reproduced here to summarise the nature of holistic EFlows approaches. 1. Designed for use in data poor situations. 2. Can be used to predict the ecological consequences of any contemplated action that would alter the water regime of an inland aquatic ecosystem. 3. Has a strong social and resource-economic module, and a structured interface with macro-economic models. 4. Can be used for perennial and non-perennial rivers, wetlands, floodplains, groundwater and estuaries. 5. Is comprehensive, addressing all parts of the ecosystem, all parts of the flow (or inundation in wetlands) regime, and social-ecological-economic interactions. 6. Combines predictions of hydrological, biophysical, social and economic changes, and produces both complex, technical outputs and accessible, simple summaries. 7. Through the scenarios, provides neutral (not pro-development or anti-development) information for consideration by governments and other stakeholders. 8. It has its own software to standardise routine activities. 9. It is well documented in the international literature, with guidelines for data collection and analysis, and user manuals for software. 10. It is continually developing within the same basic concept, to meet the needs of new clients and situations. 11. Capacity building is promoted within countries by employing national discipline specialists to collect and analyse the data, create the response curves and interpret the scenarios; DRIFT process specialists can also train interested people and countries to run the process. 12. It provides semi-quantitative or quantitative predictions of change, for use in planning. 13. It can be used to assess the effects of both development and rehabilitation plans. 14. It can help guide the decisions about issuing of water-use licences, and the writing of operating rules for dams. 15. It provides detailed and transparent predictions of change that can inform the design and targets of monitoring programmes and consequent adaptive management strategies. 16. It can provide predictions of the impact of climate change (when used with a climate change model). 17. It strongly supports, and its results are an integral part of, Integrated Water Resource Management.

Weaknesses All of the following are shared with other comparable holistic methods. 1. It needs to develop a user-friendly science-management interface to better convey its outputs (this is underway for DRIFT). 2. It requires simulated daily hydrological data, which may mean setting up a systems model for basins.

3. It works with IWRM and ecosystem concepts that are still relatively poorly understood by clients, funders, governments and other stakeholders, although this situation improves yearly as these concepts enter mainstream thinking.

References Brown C.A. and A. Joubert 2003. Using multicriteria analysis to develop environmental flow scenarios for rivers targeted for water resource development. Water SA 29 (4): 365-374. Brown C., C. Pemberton, A. Greyling and J. King 2005. DRIFT User Manual: Volume 1: biophysical module for predicting overall river condition in small to medium sized rivers with predictable flow regimes. Water Research Commission Report No. 1404/1/05, Pretoria. 105pp. Beuster H., J.M. King, C.A. Brown and A. Greyling 2008. Feasibility Study: DSS software development for integrated flow management: conceptual design of the DSS and criteria for assessment. South African Water Research Commission Report Project K5/1404, Pretoria. 23pp. King, J.M., C.A. Brown, B.R. Paxton and R.J. February 2004. Development of DRIFT, a scenario-based methodology for environmental flow assessments. Water Research Commission Report 1159/1/04. Water Research Commission, Pretoria. 159 pp. King J.M., C.A. Brown and J.I. Barnes 2010. Final IFA Project Report. Report 08-2009 EPSMO/BIOKAVANGO Okavango Basin Environmental Flows Assessment Project, OKACOM, Maun, Botswana. 42 pp. Metsi Consultants 2000. Consulting services for the establishment and monitoring of instream flow requirements for river courses downstream of LHWP dams – LHDA 648. Final Report: Summary of main findings. Lesotho Highlands Development Authority, Lesotho. 74pp. Metsi Consultants 2002. Development of a mitigation and compensation framework for communities downstream of LHWP Phase 1 structures. Lesotho Highlands Development Authority. Report No. LHDA 678-004.

Annex 4

Experience from USA relevant to reform of water abstraction licensing policy for England and Wales

Eloise Kendy, The Nature Conservancy, USA

This contribution briefly describes some methods used in the United States to set environmental flows, focusing on the 10 issues described below; highlighting where these methods, or aspects of these methods, may be employed to improve environmental flow setting and implementation in the UK; and providing advice on data collection or research that is needed to improve environmental flow assessment or implementation. Because every state and federal agency uses different approaches to assess environmental flow needs, a comprehensive review of all methods is well beyond the scope of this memo. Moreover, many of the methods used are considered outdated because they focus only on minimum flows or only on one species. Still, several states have recently advanced new, holistic approaches for determining and applying environmental flows. Therefore, this memo attempts to address the 10 issues by discussing them in the context of these specific cases. For a more detailed account of innovative methods being used in the US, please see: Kendy, E., C. Apse, and K. Blann. 2012. A practical guide to environmental flows for policy and planning, with nine case studies from the United States. The Nature Conservancy. http://conserveonline.org/workspaces/eloha/documents/practical-guide- to-environmental-flows-for-policy/view.html.

Issue #1: Setting EFIs for highly modified river reaches

EFIs currently assume that the river under assessment is in natural condition other than its flow regime. However, few UK rivers are natural. Most have been straightened, deepened or widened for flood management or other purposes, weirs and other impoundments have been built and banks have been strengthened with concrete revetments. Furthermore, trees have been removed from riparian zones and embankments prevent flood water reaching floodplains. Although some of these may be remedied as measures to achieve Good Ecological Status, many will not. In many cases, the natural flow regime may not be the appropriate flow regime to match with the altered physical nature of the channel. EFI could vary with river modification.

This issue pertains to all rivers affected by anthropogenic change other than flows. In natural rivers, flows are the “master variable” controlling habitat, biological processes, and sediment, nutrient, and temperature regimes. Yet, natural ecological systems are resilient to change. Even modified stream channels harbor some aspect of natural ecology, and flows remain critical to the health of those ecosystems. Only in the most heavily modified rivers, such as cement channels, may channel morphology have replaced flows as the major control of ecological processes. The appropriate flow regime for a modified reach depends on its management objective, which may be generalized as either:

(1) retain the existing modifications and optimize flow conditions for certain species, communities, or processes within the modified reach; or (2) restore natural ecological functions to the modified reach by replacing or removing the modifications; or (3) protect or restore natural functions in reaches downstream from the modified reach.

The first objective acknowledges that riverine habitat need not be totally natural to sustain healthy ecosystems. Although a highly modified channel might never achieve excellent ecological status, it can potentially provide usable habitat for certain species or for what may be considered a ‘new normal’ of good, but not excellent, condition.

However, a natural flow regime may no longer be appropriate. For example, low flows that once provided winter refuge in a deep natural channel might freeze solid now that the channel is wide and shallow. High flows that once brought fish onto the floodplain of a broad, meandering river might only flush them downstream now that the channel is deep and narrow.

So rather than asking “what is the natural flow regime and which flow components are most important to maintain full ecological function,” one might ask “what are the most important resources in this stretch of river and do they benefit from, or are harmed by, any particular type of flow level or event?” Perhaps, for example, an important fish species prefers a certain type of habitat for spawning, with habitat defined in terms of depth, velocity and cover.

By requiring the use of tools such as Physical Habitat Simulation (PHABSIM) for hydropower project relicensing, the U.S. Federal Energy Regulatory Commission (FERC)9 in essence pursues this management objective. PHABSIM models flows and habitat conditions along measured river cross sections. It does not assume a natural channel or flow regime; it simply determines what conditions of depth, cover, and velocity are produced by different flow levels at each cross section. Beyond fish models such as PHABSIM, other models of sediment transport, water quality, etc., are used as warranted.

It may be that even in a highly modified channel there are fish or other species that could thrive if flow conditions promoted certain types of habitat features at certain times of the year. And the results could be unexpected. For example, a fish species that spawns on submerged vegetation at channel margins may have benefitted under natural conditions from a freshet that brought water onto floodplain. But in a channelized river, a freshet may only create uniform high velocity flow and scour, whereas a lower flow actually exposes some channel margins. The fish would respond to the availability of habitat, not to the flow level per se. Importantly, an analysis that begins with the natural flow regime would conclude that fish spawn during freshets, whereas a study of that particular species in that specific river reach would determine the flow level that exposes the largest amount

9 http://www.ferc.gov/industries/hydropower.asp

75 of its preferred spawning habitat. Therefore, the managed flow regime to achieve this objective requires case-by-case EFI determination.

The Nature Conservancy’s Ecologically Sustainable Water Management (ESWM) approach (Richter et al. 2003, 2006) has been used across the US to determine environmental flows for both modified and natural rivers. The six-step process includes: (1) developing initial numerical estimates of key aspects of river flow necessary to sustain native species and natural ecosystem functions; (2) accounting for human uses of water, both current and future, through development of a computerized hydrologic simulation model that facilitates examination of human- induced alterations to river flow regimes; (3) assessing incompatibilities between human and ecosystem needs with particular attention to their spatial and temporal character; (4) collaboratively searching for solutions to resolve incompatibilities; (5) conducting water management experiments to resolve critical uncertainties that frustrate efforts to integrate human and ecosystem needs; and (6) designing and implementing an adaptive management program to facilitate ecologically sustainable water management for the long term.

The reliance of this approach on interdisciplinary expert opinion makes it well suited to customizing a flow regime to a modified river reach to achieve specific ecosystem goals. The experts are well qualified to design flow prescriptions that protect the species, communities or processes identified as targets of restoration. A practitioner website10 details step 1 (Richter et al. 2006) and provides documented case studies of how this collaborative, adaptive approach has been applied to EFI determination.

UK streams almost universally support anadromous salmonids and some other migratory fishes, so it makes sense to manage flows to support their migration through highly modified reaches. The lower Thames around London is inhabited by cyprinids that comprise a significant rough fishery, so this reach could be managed as a migration corridor for salmonids and as habitat for cyprinids. Again, flows that are managed for specific species may be quite different from the natural flow regime.

The second objective – enhancing natural ecological functions within the modified reach –requires a suite of management actions beyond environmental flows. For example, built infrastructure such as levees and revetments that limit lateral flows may be set back away from the channel and replaced with natural infrastructure such as wetlands and floodplains. In this case, a natural flow regime will help restore the sediment, nutrient, and temperature regimes that support natural ecological functions and maintain the restored channel morphology.

The third objective -- managing for natural ecological functions downstream from the modified reach – suggests a natural flow regime, even if the modified reach may not directly benefit. When unaltered environments upstream or downstream are priorities, the ecological function of the modified reach is to convey environmental flows to or from them. This management objective may be appropriate for extremely modified reaches such as concrete-lined channels. For example, the State of Washington’s instream flow rules typically cover long river reaches that include a mix of modified and more natural stretches. State biologists

10 http://www.conservationgateway.org/content/savannah-process-0

76 develop instream flow rules to protect the high-quality habitat, and those rules also apply to the degraded habitat within these reaches. In cases where the entire reach is confined with riprap (revetments), they conduct site-specific instream flow studies to establish flow rules that maximize benefits to salmon and steelhead fish within the existing physical constraints (Hal Beecher, Washington Department of Fish and Wildlife, written communication, 9 July 2012), thereby pursuing our first objective.

The State of Connecticut’s Streamflow Standards and Regulations11 apply to every river and stream reach in the state. For the vast majority of these that have not had environmental flow studies, the presumptive standards are based on the natural flow regime, regardless of the degree of channel modification.

That said, Connecticut’s flow standards do explicitly address modified stream reaches. Through a public process, every river reach in the state is assigned a condition goal class ranging from 1 to 4. Among the considerations for goal classification is the “practicality of, and potential for, restoring stream flow patterns to achieve consistency with Stream Flow Standards and Regulations due to the extent of prior channel modification or current impact of development and impervious land cover in the watershed.”

Class 1 streams support habitat conditions and biological communities typical of free-flowing streams. Class 2 and 3 streams support “minimally altered” and “moderately altered” biological communities, respectively, compared to free-flowing streams of similar types. Class 4 streams are recognized as being substantially modified.

As in the UK, different EFIs are defined for each class. EFIs for Classes 1, 2, and 3 are designed to mimic or minimally deviate from the natural flow regime. Class 4 acknowledges that the natural flow regime has been significantly altered: “A river or stream segment classified as “Class 4” pursuant to the Stream Flow Standards and Regulations may exhibit substantially altered stream flow conditions caused by human activity to provide for the legitimate needs and requirements of public health and safety, flood control, industry, public utilities, water supply, agriculture and other lawful uses.” Therefore, the EFI for Class 4 is to achieve Class 3 flow requirements to the maximum extent practicable. The premise, based on the US Clean Water Act, is that although every river reach may not attain excellent ecological status, all reaches do deserve to meet a good standard of quality.

Not every highly modified reach in Connecticut necessarily will be classified as a Class 4. If the objective is to restore ecological function to that reach, then it is managed as a Class 3. Or, if the objective is to protect or restore a less modified downstream reach, then its EFI can be based on the goal class of the downstream reach. The standards and regulations also contain provisions to raise or lower a river’s goal class in the future if information, societal needs, or other circumstances change. To reiterate, regardless of the degree of modification, under the Connecticut statute, flow criteria for all rivers strive to mimic the natural flow regime to the extent possible.

11 Connecticut Department of Energy and Environmental Protection Stream Flow Standards and Regulations Section 26-141b-1 to 26-141b-8, inclusive, of the Regulations of Connecticut State Agencies, Effective December 12, 2011, http://www.ct.gov/dep/lib/dep/regulations/26/26-141b-1throughb-8.pdf.

The degree of modification is considered when determining “the extent possible.” In some cases, a water user may conduct a study to show that another flow regime is warranted. Therefore, the presumptive standard is to use natural flows, unless an alternative flow regime is sufficiently documented.

Issue #2: Defining EFIs for Good Ecologic Potential (GEP)

Related to 1 above, the current EFIs are to achieve Good Ecological Status (GES). Where rivers cannot reach GES due to modifications that are economically important, they should meet GEP. Where major dams impound a river and the flow regime is potentially completely altered, an approach to define flow releases has been recommended (Acreman et al., 2009). This is being tested, but has not yet been implemented. EFIs have not been defined for GEP.

The State of Connecticut recently adopted reservoir release rules that address this issue. As mentioned above, a condition goal class of 1-4 will be assigned to each stream or river reach in the state. Reservoir release rules are tailored to each Class. Dams on Class 1 streams are not allowed to manipulate their reservoir storage actively; in other words, they must operate as run-of-the-river, flow-through systems. Dams on Class 2 streams must release at least 75% of their reservoir inflows at all times.

Connecticut’s Class 3 parallels the EU’s GEP. Dams on Class 3 streams, which include almost all dams of any significant size, have additional release requirements to help mimic natural flow patterns. First, the volume of the required release depends on the bioperiod in which it occurs. Bioperiods are biologically-based seasons lasting between one and four months. The regulations define six bioperiods based on the flow needs of the range of river species typically found in Connecticut (see Table 1). Varying the release requirements according to bioperiod improves their accuracy in mimicking natural seasonal flows. Second, short-term larger releases are required when flow is higher than normal for the bioperiod (see Issue #3, below). Although the technical committee that recommended this two-level release system discussed its application year-round, the final regulation applies it only during the rearing and growth bioperiod (summer).

For more information about the development and implementation of Connecticut’s new Streamflow Standards and Regulations, see Kendy et al. (2012, pages 29-32 by Mark P. Smith).

Table 1. Effective dates and minimum release requirements by bioperiod, as stipulated by Connecticut’s stream flow regulations. Minimum Required Release Effective Bioperiod Dates Antecedent Antecedent Period Dry Period Wet Dec 1- Feb Overwinter Bioperiod Q99 28/29 Habitat Forming Mar 1 – Apr 30 Bioperiod Q99 May 1 – May Clupeid Spawning Bioperiod Q95 31 Resident June 1 – June Bioperiod Q90 Spawning 30 Rearing and Bioperiod Bioperiod July 1- Oct 31 Growth Q80 Q50 Salmonid Nov 1 – Nov 30 Bioperiod Q90 Spawning

Issue #3: Varying EFIs to mimic natural temporal sequencing

EFIs are based on a flow duration curve (Table 2). This approach loses the temporal sequencing of flows. For example, on any day during April to October, if the flow is at Q95, 20% can be abstracted from an A1 river. This takes no account of whether the flow has been low or high before or is likely to rise or fall after that day. EFIs could vary according to durations of flow at different levels.

Table 2. EFIs for UK river types and sub-types for achieving Good Ecological Status, expressed as percent allowable abstraction of natural flow (thresholds are for annual flow statistics).

Type or sub Season flow > Qn60 Flow > Qn70 flow > Qn95 flow < Qn95 type

Apr – Oct 30 25 20 15 A1

Nov – Mar 35 30 25 20

Apr – Oct 25 20 15 10 A2 (ds), B1, B2, C1, D1 Nov – Mar 30 25 20 15

Apr – Oct 20 15 10 7.5 A2 (hw), C2, D2 Nov – Mar 25 20 15 10

Salmonid spawning & Jun – Sep 25 20 15 10 nursery areas (not flow > Q flow < Q Chalk rivers) Oct – May 20 15 80 80 10 7.5

The UK’s EFIs are based on flow duration curves. Connecticut’s reservoir release rules also are based on flow duration curves (Table 1). However, Connecticut uses seasonal flow statistics, whereas the UK uses annual flow statistics to create the curves. Seasonal calculation of flow duration statistics improves the ability of managed flows to mimic natural flows.

Connecticut further addresses temporal sequencing by defining flow levels based on average inflows over the preceding two weeks. If inflow to the reservoir during the preceding two weeks exceeded the bioperiod Q25 exceedance value, then the flow level is considered “wet”; if inflow was less, then it is considered “dry” (see Table 1.) This was a practical compromise between aquatic ecologists, who preferred daily flow adjustments, and reservoir operators, who have operational and capacity issues to consider. This approach adequately addresses one of the difficult issues regarding release policies: how to account for wet and dry periods as they occur to prevent either overly augmenting streamflow that would naturally be low or unduly depleting streamflow that would naturally be high. By tying releases to immediately preceding conditions, they mimic the natural flow regime.

The interstate Susquehanna River Basin Commission, which has regulatory authority over water abstraction, recently adopted flow criteria12 based on flow duration

12 SRBC Low Flow Protection Policy and Technical Guidance for Low Flow Protection Related to Withdrawal Approvals, http://www.srbc.net/pubinfo/businessmeeting.htm

80 curves, supplemented by additional criteria to maintain the historic natural variability of ecologically important flow components.

Ten types of flow statistics describe the magnitude and frequency of large and small floods, freshets, median monthly flow, and monthly low flow conditions in the Susquehanna River basin: magnitude and frequency of 20‐ year (large) flood, 5‐ year (small) flood, and bankfull (1‐ 2 year high flow) events; frequency of high flow pulses in summer and fall; high pulse magnitude (monthly Q10); monthly median (Q50); typical monthly range (area under monthly flow duration curve between the Q75 and Q10); monthly low flow range (area under monthly flow duration curve between Q75 and Q99); monthly Q75 and monthly Q95. In addition, monthly range and monthly low-flow range statistics quantify changes in flow-duration curve shape, complementing analyses of changes in individual flow metrics to assess seasonal impacts of water use on ecological flow regimes. EFIs are expressed in terms of acceptable ranges of these flow statistics (Kendy et al. 2012, pages 38-41; DePhilip and Moberg 2010).

In the State of Florida, the Southwest Florida Water Management District uses a percent-of-flow approach to mimic natural temporal sequencing of streams used for water supply and to protect downstream estuaries from impacts of large freshwater withdrawals during the ecologically vulnerable dry season (Flannery et al. 2002). This approach limits withdrawals to a certain percentage of streamflow at the time of withdrawal. Low-flow cutoffs protect the rivers from over-extraction. During high flow periods, when allowable withdrawals exceed demand, withdrawals are diverted to offstream storage reservoirs; during low flow, that water may be used to meet demands or released to meet flow standards.

Issue #4: Managing water abstraction during low flow periods

Water is often most needed for agriculture and public supply when flows are low (during hot dry summers) but this is also when allowable abstraction is at its lowest. Consideration could be given to higher abstractions for particular periods provided that the river ecosystem is allowed to recover afterwards, perhaps allowing less or no abstraction.

To my knowledge, no US river is intentionally managed for very large abstractions followed by a recovery period, as described above. Jurisdictions dealing with low- flow issues are more likely to store wet-season water to supply dry-season irrigation and to implement urban conservation measures, rather than to allow over-extraction. Beyond basic ecosystem support, low flows are often needed to dilute contaminant loads in order to meet water quality standards.

Studies by Tufts University (Vogel et al. 2007) demonstrated that if municipal water supply customers would reliably reduce their water usage during droughts – for example by forgoing lawn watering and car washing -- then water supply reservoir operators could safely release water from storage to meet low-flow needs downstream. This demonstration was instrumental in negotiating Connecticut’s reservoir release rules.

During drought conditions, Connecticut’s release requirements for public water supply reservoirs may be reduced systematically (Table 3). These reductions in releases are tied to the water suppliers’ required drought contingency plans so that the reduced releases to streams occur somewhat simultaneously with the implementation of water-use restrictions that suppliers impose as they approach an emergency declaration. This includes a provision for zero releases during water supply drought emergencies, which ensures that the “last drop of water” goes to people rather than to the stream. Reducing releases during drought significantly decreases the impact of the release requirements on the security of water supplies for human use.

Table 3. Reduced release requirements for Connecticut’s public water supply reservoirs during drought. Percentage of Required Dry-Period Release Water Supply Plan Rearing & Growth Trigger All Other Bioperiods Bioperiod Drought Advisory 100% 75% Drought Watch 50% 50% Drought Warning 25% 25% Drought Emergency No Release Required No Release Required

Although most UK water supplies are withdrawn directly from rivers (or closely connected shallow aquifers) rather than from reservoirs as in Connecticut, the same principle would apply: water conservation measures enacted during drought can free up water for river ecosystems without threatening the security of water supplies for basic human needs.

Issue #5: Adjusting “baseline” to account for climate change

It is probably unreasonable to consider that flow regimes are stable and current global warming suggests that flow regimes will change in the future. The latest thinking on the WFD is that reference conditions should change as the climate changes. The implication is that baseline or benchmark flows should also change. Abstraction allowances could remain the same in percentage terms, but against a new baseline that is not that of the current natural flow regime.

To date, the degree of flow alteration caused by climate change is dwarfed by the alteration caused by dams, withdrawals, and land-use change. Climate change is just the latest stressor. The protection and restoration of natural flow patterns is viewed as a crucial step toward building ecosystem resiliency to climate change. Allowing relative adjustments to baseline might hasten the demise of species that otherwise would survive climate change. If the plant and animal communities that historically inhabited a river still exist there, then it seems imprudent to assume a new baseline for flow management.

The question of what to use as “baseline” for environmental flow policy is often answered by political expediency as much as by scientific merit. Michigan was one of the first states to systematically link ecological outcomes to water withdrawal licensing. The baseline that Michigan uses is current, rather than natural, conditions,

82 which essentially limits the water licensing program goal to flow protection, and not restoration. As disappointing as this may be for ecologists, without this agreement upfront, current water users would not have joined the effort that ultimately succeeded in integrating ecological criteria into water licensing (Kendy et al. 2012, pages 13-17; Ruswick et al. 2010).

Various efforts are currently underway to predict the impacts of climate change on water availability for public supplies. However, relatively few efforts are attempting to predict impacts on aquatic ecosystems from flow alteration brought on by climate change for purposes of water withdrawal licensing. By comparing regional hydrology under climate change to current and pre-development conditions, water managers can plan for future water supply and ecosystem service provision under various scenarios that also account for likely future changes in land and water use as well as climate. The Interstate Commission on the Potomac River Basin (ICPRB) plans to make this assessment, using regional rainfall-runoff models linked to ecological response models, to inform water management. Stakeholders in the Connecticut River basin are using a newly developed hydrologic model and decision–support system to evaluate environmental and economic outcomes of various multi-dam management scenarios under climate change to support FERC re-licensing (Kendy et al. 2012).

Issue #6: Presenting EFIs in a risk-based context

EFIs are very uncertain due to lack of precise scientific knowledge. Yet they presented as fixed values with no latitude. EFIs should be presented in a risk- based context.

The State of Michigan applies environmental flow criteria in a risk-based context. Scientists modeled fish response curves (Figure 1) that relate population and density changes in fish communities to percentage reductions in mean August flow for each of the 11 river types that occur in the state (Zorn et al. 2009). Curves for thriving species (those expected to be especially abundant) can be considered “early warning flags” of Adverse Resource Impact, which the Michigan legislature defined in terms of characteristic species (expected to be more abundant than the state mean abundance).

Figure 1. Typical fish-response curves. ARI indicates Adverse Resource Impact, depicted here as 90% of characteristic fish species remaining. Light lines indicate thresholds between water management zones associated with different degrees of ecological risk. A = register water use, B = notify local water users, C = form a water user committee.

To compensate for uncertainties in the models, the 2008 Michigan law created “management zones” representing increasing levels of risk to the environment (Figure 1), and prescribed a suite of water management actions for each level. Because the curve for each river type is different (Figure 2), the flow removal associated with a given change in fish assemblage -- and therefore the boundaries between management zones -- differ by river type.

Prospective water users use an online Water Withdrawal Assessment Tool (WWAT) to determine the level of risk associated with their proposed withdrawals. Users enter the location, timing, quantity, and if relevant, the screen depth of their proposed groundwater or surface water withdrawals. The WWAT calculates flow depletion of the nearest stream segment during summer low flow due to the proposed withdrawal, added to the cumulative withdrawals from upstream segments. The WWAT uses the appropriate fish response curve for the affected river type to determine the ecological risk level associated with that depletion. If the risk level is low, then the withdrawal may be registered online with no further analysis. If the risk level is high -- meaning the withdrawal would likely cause an Adverse Resource Impact -- then site-specific review by Department of Environmental Quality staff is required, using local flow and fish data and expert opinion instead of the less accurate statewide model. After site review, the withdrawal will be registered, registered with modifications, or rejected.

1 1

0.8 0.8

0.6 0.6

Cold 0.4 0.4

0.2 0.2

0 0 0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1

1 1 1

0.8 0 .8 0 .8

0.6 0 .6 0 .6

C-Trans 0 .4 0 .4 0.4

0 .2 0 .2 0.2

0 0 0 0 0. 25 0 .5 0 .7 5 1 0 0.2 5 0.5 0. 75 1 0 0.25 0.5 0.75 1

1 1 1

0 .8 0 .8 0.8

0.6 0 .6 0 .6

0 .4 W-Trans 0.4 0 .4

0 .2 0.2 0 .2

0 0 0 0 0.2 5 0 .5 0 .75 1 0 0.25 0.5 0.75 1 0 0. 25 0 .5 0 .7 5 1

1 1 1

0.8 0.8 0.8 Warm 0.6 0.6 0.6 0.4 0.4 0.4

0.2 0.2 0.2

0 0 0.25 0.5 0.75 1 0 0 0 0.25 0.5 0.75 1 0 0.25 0.5 0.75 1

Streams Sm Rivers Lg Rivers

Figure 2. Michigan’s fish response curves, showing how each type of river has different curves, and therefore different water withdrawals associated with each management zone. X- and y-axis labels are as shown in Figure 1. From Zorn et al. (2009).

As a result of this risk-based approach, registration of new water abstractions is expedited when environmental risk is low. Government staff time focuses on withdrawals that pose the most risk and stream segments that are most highly valued by society (because anyone can request a site review). Justifiably, the Michigan Water Withdrawal Assessment process has won three national awards for streamlining government programs.

Issue #7: Reflecting hydro-ecological thresholds in EFIs

EFIs vary consistently along a flow statistic gradient; higher percentage at Q60, lower at Q95. This does not explicitly include any thresholds or tipping, such as where ecological conditions abruptly become more sensitive to small changes in flow. Such thresholds could be at inflection points in the habitat-flow relations at low flows or high flows (e.g. floodplain-river connectivity). EFIs could reflect hydro-ecological thresholds.

In the rare cases in which hydro-ecological thresholds exist in U.S. rivers, they only apply to certain species or guilds. But ideally, environmental flow criteria meet the needs of all plants and animals that use a river reach.

That said, many states and interstate river basin commissions in the US are quantifying flow-ecology relationships similar to those shown in Figure 2 to provide a

85 scientific basis for environmental flow criteria. Using an expedient combination of quantitative and qualitative approaches, relationships can be developed for all major ecosystem components, as well as for physical and chemical processes, for different types of rivers (see, for example, Susquehanna River basin case study in Kendy et al. 2012, pages 38-41). If a threshold exists, then such relationships can define it. However, it is quite unlikely that it will apply to every species with in the river or river type.

Issue #8: Optimizing multiple benefits nationally

Under WFD all river water bodies must meet GES or GEP individually. In Sweden, a national accounting approach has been taken. The Swedish government has calculated that if environmental flows are implemented at its 5 largest hydropower dams, a further 22 new small dams would need to be built to fill the power production short-fall. It is estimated that the total environmental degradation would be far worse with 22 new dams. Consequently, it would be better in terms of national environmental accounting to designate 5 sacrificial rivers where no environmental flow is released, but to ensure environmental flows on all other rivers. National environmental accounting could be considered as part of EFIs.

National environmental accounting is a great idea, although environmental flows may play a supporting, rather than a leading role. A system-level approach that optimizes outcomes for hydropower, other water management sectors, and conservation of environmental and social resources is desirable and technically achievable. While integrated regional environmental and hydropower planning is being pursued in various less-developed basins around the world, it is actually being implemented in one developed basin, where ecosystem restoration -- not just protection – is the primary goal.

The 2.2-million-hectare Penobscot River basin in Maine historically supported culturally and economically significant populations of migratory fish. These fish populations declined dramatically following the construction of a series of hydropower dams on the main stem river and major tributaries in the early 20th century. Restoring the fishery means restoring not only environmental flows, but also the very migration route that the dams now block. The solution, negotiated between a privately owned power company, the Penobscot Indian Nation, resource agencies, and NGOs, features the removal of two main stem dams and improved fish passage and environmental flow releases at the dams that remain. By increasing storage and re-operating the remaining dams, total hydropower production from the basin will be maintained or increase slightly. The proportion of the basin accessible to migratory fish will increase significantly and contribute to resurgence in fish populations (Opperman et al. 2011). The first dam removal was recently completed.13

In order for “accounting” to balance environmental and economic values at the national or large basin scale, restoration activities such as dam removal or re-

13 Documented in these videos: http://vimeo.com/44282039, http://www.ustream.tv/recorded/23506070.

86 operation should focus on connecting the most ecologically valuable sub-basins and river reaches. In many cases aquatic habitat connectivity, rather than environmental flows, will be the primary currency of trade-off. Cascades of dams that maximize energy production from other parts of the basin can terminate with re-regulation reservoirs that release environmental flows. Synergies such as optimization between flood risk management and hydropower also can be achieved. Opperman and Harrison (2008) explain through several case studies how integrated planning yields multiple benefits such as these.

Of course, technical solutions are not the only challenges to system-scale optimization. When dams are removed or re-operated, a significant challenge is to devise a system that financially compensates dam owners who currently profit from producing power. In the Penobscot case, one company owns most of the relevant hydropower dams in the basin, so its balance sheet is unaffected by shifting power production between dams.

Issue #9: Gauging flows at abstraction points

EFI implementation is based on reference to river flow gauging stations at assessment points. For better implementation flow data are required for all abstraction points and continuously if issue 3 above is to be addressed. EFI implementation requires a knowledge of flow in real-time at all locations on the river network (a computer model is already in place that provide real-time flood forecasts for major river reaches).

To enforce EFI compliance, flow data are indeed required for all abstraction points. Connecticut, for example, requires all regulated water supply reservoir operators to gauge inflow to their reservoirs. On the other hand, the Susquehanna River Basin Commission has devised a procedure for estimating flows based on nearby gauges.14

To inform regional water management decisions, knowledge of where, when, and how much water occurs in all water bodies across the landscape – not just at abstraction points -- is also needed. Ideally, both current and “baseline” daily flow data should be modeled for everywhere ecological data have been collected, where flow management actions such as water diversion may be taken, where streamflow will be monitored to ensure compliance with flow standards, and above and below major river confluences. This information may be used to assess flow characteristics, classify river types, quantify flow alteration, evaluate ecological responses to flow alteration, develop EFIs, and evaluate the status of sites relative to EFIs.

For further guidance and case studies of modeling flow data for national EFI development, please see Kendy et al. (2012, pages 50-54). The Sustainable Yield Estimator (pages 18-19 and 52) is a new technique that has shown particular

14 SRBC Technical Guidance for Low Flow Protection Related to Withdrawal Approvals, http://www.srbc.net/pubinfo/businessmeeting.htm

87 promise for this application; several northeastern states are currently investing in its development.

Issue #10: Managing high flows

Most historical environmental flow work has focused on the problems of insufficient water at low flows. Indeed, Table 1 above gives only maximum abstractions. However, the recent definitions of e-flows stress the importance of all aspects of the flow regime. Urban research for example shows that high flows are the most critical for the river ecosystem. Many abstraction schemes involve releases from regulating reservoirs for use downstream, e.g. River Dee, due to which flows during dry periods are much higher than they would be naturally. However, too much flow at the wrong time can be as detrimental to the ecosystem as not enough flow. This requires consideration of discharges as well as abstractions in eflow setting.

For regional applications, environmental flow regimes should be expressed as percent alteration from natural, rather than as a specific flow level or volume (Richter et al. 2011). For example, an optimization model is being used to integrate multiple reservoir re-operations in the Connecticut River basin. Environmental flow targets are input to the model in the form of penalty functions. Penalties accrue regardless of whether the modeled flows are above or below target flows; the closer the modeled flows are to target flows, the lower the penalties. The model finds scenarios that minimize penalties across the basin. Flow-ecology relationships for the Susquehanna and Potomac River basin also are expressed in terms of flow alteration rather than as flow depletion (Kendy et al. 2012). Many studies across the US relate adverse ecological impacts to increased flashiness of high flows due to impervious land surfaces and agricultural drainage. However, to date these stressors are not being regulated to restore more natural high flow regimes.

Improving environmental flow setting and implementation in the UK

Because no regional environmental flow policy can cover all circumstances, it is important to provide for site-specific study in special cases. Some resources are worth the additional time and expense to assess individually, rather than defaulting to a regional standard. Moreover, the lessons that are learned in the process are likely to inform and improve the regional standard, as well.

Research and data needs

It is likely that in the UK, as in the US, the greatest data need is the location and quantification of water abstraction and return flows, including seepage through aquifers. Without knowing how cumulative depletions affect a river reach, it is extremely difficult to plan and manage future abstractions. Water abstraction and return flow data also are needed to model the basic streamflow data described above, which are a fundamental building block of EFIs.

Flow-ecology relationships have proven to be very helpful in supporting EFI decisions in the US. Previously, EFIs in many states were vulnerable to accusations

88 of arbitrariness. A scientific understanding of how flow alteration affects different components of aquatic ecosystems helps guide the development of EFIs and provides substantive defense of their implementation. Poff et al. (2010) and Kendy et al. (2012) explain qualitative and quantitative approaches for developing flow- ecology relationships and using them to inform EFI assessment.

References Apse, C., M. DePhilip, J. Zimmerman, and M. P. Smith. 2008. Developing instream flow criteria to support ecologically sustainable water resource planning and management. Final report to the Pennsylvania Instream Flow Technical Advisory Committee, 196 p. http://www.portal.state.pa.us/portal/server.pt/document/440033/pa_inst ream_flow_report-_tnc_growing_greener-_final.pdf. DePhilip, M., and Moberg, T. 2010. Ecosystem flow recommendations for the Susquehanna River basin. Report to the Susquehanna River Basin Commission and U.S. Army Corps of Engineers. The Nature Conservancy. 96 p + appendices http://www.nature.org/media/pa/tnc-final-susquehanna-river-ecosystem- flows-study-report.pdf. Flannery, M. S., E. B. Peebles, and R. T. Montgomery. 2002. A percent-of-flow approach for managing reductions of freshwater inflows from unimpounded rivers to southwest Florida estuaries. Estuaries 25: 1318-1322. Kendy, E., C. Apse, and K. Blann. 2012. A practical guide to environmental flows for policy and planning, with nine case studies from the United States. The Nature Conservancy. http://conserveonline.org/workspaces/eloha/documents/practical-guide- to-environmental-flows-for-policy/view.html. Opperman, J. J., and D. Harrison. 2008. Pursuing sustainability and finding profits: integrated planning at the system level. HydroVision 2008, Paper No. 093, 20 p. HCI Publications, Sacramento. http://www.conservationgateway.org/file/integrated-hydropower- planning-system-scale. Opperman, J. J., J. Royte, J. Banks, L. R. Day, and C. Apse. 2011. The Penobscot River, Maine, USA: a basin-scale approach to balancing power generation and ecosystem restoration. Ecology and Society 16: 7. http://www.ecologyandsociety.org/issues/view.php?sf=61. Poff, N. L., Richter, B. D., Arthington, A. H., Bunn, S. E., Naiman, R. J., Kendy, E., Acreman, M., Apse, C., Bledsoe, B. P., Freeman, M. C., Henriksen, J., Jacobson, R. B., Kennen, J. G., Merritt, D. M., O'Keeffe, J. H., Olden, J. D., Rogers, K., Tharme, R. E., and Warner, A. 2010. The ecological limits of hydrologic alteration (ELOHA): a new framework for developing regional environmental flow standards. Freshwater Biology 55:147-170. http://www3.interscience.wiley.com/cgi- bin/fulltext/122588390/PDFSTART Richter, B. D., M. Davis, C. Apse, and C. P. Konrad. 2011. A presumptive standard for environmental flow protection. River Research and Applications. DOI: 10.1002/rra.1511. Richter, B. D., A. T. Warner, J. L. Meyer, and K. Lutz. 2006. A collaborative and adaptive process for developing environmental flow recommendations. River Research and Applications 22: 297-318.

Richter, B. D., R. Mathews, D. L. Harrison, and R. Wigington. 2003. Ecologically sustainable water management: managing river flows for ecological integrity. Ecological Applications 13: 206-224. Ruswick, F., Allan, J., Hamilton, D., and Seelbach, P. 2010. The Michigan Water Withdrawal Assessment process: science and collaboration in sustaining renewable natural resources. Renewable Resources Journal 26:13-18. Vogel, R. M., J. Sieber, S. A. Archfield, M. P. Smith, C. D. Apse, and A. Huber-Lee. 2007. Relations among storage, yield and instream flow. Water Resources Research 43. http://www.weap21.org/downloads/RelationsAmongStorage.pdf. Zorn, T. G., Seelbach, P. W., Rutherford, E. S., Wills, T. C., Cheng, S.-T., and Wiley, M. J. 2009. A regional-scale habitat suitability model to assess the effects of flow reduction on fish assemblages in Michigan streams. Michigan Department of Natural Resources, Fisheries Research Report 2089, Ann Arbor, Michigan, 46 pp. http://www.michigan.gov/documents/dnr/RR2089_268570_7.pdf.

…