i

Phosphorus Loading to Lake Forsyth/

Te Roto o Wairewa

Preliminary Results for a Lake Phosphorus Budget

WCFM Report 2014-006

ii

REPORT: WCFM Report 2014-006

TITLE: Phosphorus Loading to Lake Forsyth/Te Roto o Wairewa

Preliminary Results for a Lake Phosphorus Budget

PREPARED FOR: Environment Canterbury

PREPARED BY: Sean Waters MSc, MAppSci, PhD Candidate

REVIEWED BY: Prof Jenny Webster-Brown , Director WCFM

AFFILIATION: Waterways Centre for Freshwater Management (WCFM)

University of Canterbury & Lincoln University

Private Bag 4800

Christchurch

New Zealand

DATE: 4 July 2014

iii

Executive Summary Te Roto o Wairewa/Lake Forsyth (Wairewa) is a small, shallow, coastal lake on the south side of Banks Peninsula. The lake is currently eutrophic and experiences regular blooms of potentially toxic cyanobacteria. This has had a major impact on biodiversity and recreation values. The Wairewa catchment includes soils rich in phosphorus (P) and, in common with many eutrophic lake systems around the world, P has been identified as a likely limiting nutrient contributing to algal growth and associated water quality issues. Rehabilitation of Wairewa has been identified as a priority by Te Wairewa Runanga, however, effective management interventions require a good understanding of the lake ecosystem. This report presents the preliminary results from research into a P budget for Wairewa, and is part of broader PhD research on the P dynamics in the lake.

Stream flows, direct runoff and P concentrations measured over a 15 month period between 3 December 2012 and 2 March 2014, have been used to estimate P loads transported to the lake during that period. Outflows during artificial lake openings, and as seepage through the beach barrier at Birdlings Flat, have been quantified. The difference between inflows and outflows allow estimation of the amount of P retained in the lake. Analyses of P concentrations in the lake sediments, sediment pore waters, macrophytes and the lake water column, were undertaken to estimate the amount of P stored in each of these potential reservoirs.

P Budget An annual load of 11300kg of P was transported to the lake during the research period. An annual load of 1700kg of P left the lake via the outflow and barrier seepage and hence 9600kg of P was retained in the lake. Of the P inflows, 96% was via the Okana and Okuti Rivers, with >70% of the total from the Okana River alone. More than 70% of the total load over the 15 month research period, was transported during one flood event (June 2013) and approximately 80% of the Okana P load was delivered during this flood event alone. The importance of high flow events in P transport was also evident in the >85% of P load transported while bound to suspended particulate material. Direct runoff, ground water, and ephemeral streams in the catchment, contributed <5% of the P load transported to the lake. A 25% uncertainty was calculated for loads in the Okana River and a similar uncertainty is assumed to apply to all P loads calculated in this report

P Reservoirs More than 430 000 kg of P are stored in the top 25cms of lake sediment with >80 000 kg in the top 5cms of sediment. In contrast, the amount of P stored in the water column is small and highly variable (140-3130kg) with time. This variability is not directly controlled by the rate of P transport from the catchment (external loading), but is commonly controlled by the flux of P from the sediment to the water column (internal loading). The dynamics of this P release are the subject of ongoing research (not reported here). Macrophytes also comprise a very small reservoir of P in the lake (<200kg). A 20% uncertainty was calculated for the sediment P reservoir and a similar level of uncertainty is assumed to apply to all P reservoirs calculated in this report.

Implications for Lake Managment Management interventions aimed at improving water quality in Lake Forsyth/Te Roto o Wairewa will need to address both external and internal P loading. Measures to control external P loading to the lake will need to address the transport of P associated with particulate material during high flow events, especially in the Okana River.

iv

Contents

Page

Section 1 Introduction 1

1.1 Eutrophication in lake systems 1

1.2 P dynamics in lake systems 1

1.3 Lake Forsyth/Te Roto o Wairewa 3

1.4 Rationale for this report 4

Section 2 Methodology 5

2.1 Hydrology 5

2.2 P concentrations and loads 10

2.3 Uncertainties in P loads and P reservoirs 16

Section 3 Results 17

3.1 P loads transported to the lake 17

3.2 P loads transported from the lake 19

3.3 P budget and reservoirs 20

3.4 P variations in the lake 21

Section 4 Discussion 22

4.1 The P budget 22

4.2 The P reservoirs 22

4.3 External vs internal P loading 24

4.4 Implications for management responses 24

References 25

1

Section 1 Introduction

1.1 Eutrophication in lake ecosystems

The eutrophication of freshwater systems refers to the increasing primary productivity that occurs in response to elevated nutrient inputs (Wetzel 2001). These inputs often result from anthropogenic activities including changes in catchment land use patterns, such as deforestation or agricultural activity (Schindler, 2006; Downs, et al 2008). The increased productivity is generally expressed as an increase in phytoplankton concentration and may result in seasonal or permanent algal ‘blooms’ which seriously degrade water quality, with detrimental impacts on biodiversity. Light attenuation and decreased dissolved oxygen levels in particular, have marked impacts on macrophyte, zooplankton, benthic and fish communities (Wetzel, 2001; Sondergaard, 2007; Schallenberg and Sorell, 2009). Such eutrophication is a well-documented phenomenon in lake ecosystems across the world (eg; Jacoby et al, 1982; Kristensen et al , 1992; Kumar Das et al, 2009) and is a significant issue in an large number of New Zealand lakes (Ministry for the Environment, 2010). Primary productivity in lake systems is strongly limited by nutrient availability (Sondergaard, 2007). While numerous nutrients are critical for primary production, the dominant role of the macronutrients nitrogen (N) and phosphorous (P), in controlling phytoplankton biomass in lakes is well established (Wetzel, 2001; Abell et al 2011). P is only bioavailable in the form of the soluble orthophosphate ion (PO4

3-). The concentration of this ion in natural waters is reduced by its affinity for particulate sediment, as it is adsorbed onto clays and oxides minerals as well as organic/inorganic colloidal material. Hence, N generally occurs in lake waters in much higher concentrations than P, and phytoplanktonic biomass will often be limited by the lower availability of dissolved P. In addition N limitation may be circumvented by the fixation of atmospheric N by various bloom-forming cyanobacteria (Wetzel, 2001; Sondergaard, 2007). Although different systems may be limited at times by various nutrients (Berger, 2006), P is considered to be the foremost nutrient in the limitation of primary productivity in most lakes (Reynolds & Davies, 2001; Schindler, 2006; Downs et al, 2007; Kumar Das et al, 2009). Therefore understanding the sources and fate of P is considered to be critical to understanding the environmental state of lake ecosystems (Sondergaard, 2007).

1.2 Phosphorous dynamics in lake systems 1.2.1 External phosphorus loading to lakes P delivery to lake systems via the various external pathways is termed ‘external loading’. Natural P inputs include particulate-associated and dissolved-P species sourced from run off and the resulting erosion in the lake catchment. While these processes may be greatly accelerated by human activities, point source nutrient-rich effluents such as sewage, industrial wastes and detergents have largely been controlled by regulation in developed nations (Carpenter et al, 1998; Schindler, 2006). However diffuse sources of nutrient

2

pollution have been increasingly recognised as a major issue and are much more difficult to control. The strong affinity of P for particulate material means that non-dissolved forms predominate in the transport of P from catchment to lake (eg; Pacini & Gachter, 1999). Erosion, which preferentially mobilises clay-sized particles and organic material, both of which can be strong P adsorbents, is exacerbated by land use activities such as deforestation and agriculture. In addition the intensification of agriculture, often requiring intense fertiliser application, increases the soil- bound store of P which in turn can increase the amount of P carried by eroded soils. These activities, without very careful control, can increase the external P loading of lake systems enormously (Carpenter et al, 1998; Schindler, 2006; Parliamentary Commissioner for the Environment, 2012).

1.2.2 Phosphorous cycling within lakes

Once in the lake system, P may exist in a number of chemical forms (species). Dissolved orthophosphate (PO4

3-) in the water column is directly bioavailable, while particulate-associated P will settle to the bottom of the lake as a result of sedimentation. This sediment can constitute a major reservoir of in-lake P (Wetzel 2001). In most lakes and particularly in shallow lakes there is a constant interaction between the near-surface sediment and the water column. While some sediment bound P will be immobilised permanently, various processes can occur which return some of the P from the sediment, to the lake water (Figure 1). This is referred to as ‘internal P loading’ and can be the major source of P contributing to eutrophication in the lake (Jacoby, 1982; Sondergaard et al, 2003; Gao, 2005; Christophoridis and Fytianos, 2006;). Lake sediments can therefore, be both a source and sink for P and understanding the complex chemical and biological interactions between the sediment and water is critical for understanding eutrophication in a given lake system.

Figure 1. Simplified diagram of lake phosphorous dynamics (adapted from Sondergaard,

2007).

3

1.3 Lake Forsyth/ Te Roto o Wairewa

Lake Forsyth/ Te Roto o Wairewa (Wairewa) is a small (5.6km2), shallow lake on the southern side of Banks Peninsula. The lake catchment covers approximately 110km2, much of which is used for pastoral agriculture. Inflows to the lake are predominantly from the Okana and Okuti Rivers with lesser inflows from smaller stream tributaries and runoff which flows direct to the lake, as well as an unknown, but likely minor component contributed directly from groundwater. The lake is generally less than 2m deep, and is separated from the sea by the eastern end of an active barrier-beach complex known as Kaitorete Spit. This narrow barrier (≤100m) has evolved over 8000 years as an accumulation of material which has been transported north along the coast of the Canterbury Bight (Woodward and Schulmeister, 2005). The timing of the transition of the lake from marine estuary to closed-in lake is uncertain but appears to be least 450 years BP (Woodward and Schulmeister, 2005). In the 19th century a lake outlet still existed and whalers, canoes and coastal traders accessed the lake from the ocean implying that the lake was open at least periodically (Jellyman and Cranwell, 2007). Currently the lake is brackish with a wide range of salinities (salinity=1-11‰ NaCl, mean 6‰ NaCl, Main et al 2003). Artificial opening of the lake to the sea first occurred in 1866 and now occurs approximately once a year in order to manage flooding of local roads and farmland (Soons, 1998; Reid et al, 2004; Woodward and Schulmeister, 2005).

Wairewa is currently a turbid, hypertrophic lake, which experiences regular toxic blooms of cyanobacteria, predominantly Nodularia spumigens and less frequently Anabaena spp. Paleolimnological studies indicate that deforestation of the catchment predominantly after European arrival led to major increases in runoff and sediment to the lake, creating virtually freshwater conditions (Woodward and Schulmeister, 2005). Continuing nutrient input due largely to erosion of the P-rich, volcanic-derived soils of Banks Peninsula (Lynn, 2005), has led to declining water quality and toxic algal blooms. An increase in salinity, probably due to periodic openings, initiated a shift to more saline tolerant algal species such as Nodularia spumigens, the first recorded bloom of which occurred in 1907 (Main et al, 2003; Woodward and Schulmeister, 2005).

Environment Canterbury (ECan) has monitored water quality parameters in the lake since 1993. Nutrient levels are variable and often very high, and these high concentrations are commonly associated with algal blooms. However, the exact causative link between nutrient concentration and bloom formation has not been established (Main et al, 2003). Nutrient speciation and dynamics in the catchment waterways and the lake, are largely un-researched, although seasonal and stream flow related variations in TP and dissolved PO4 have been recorded (Main et al, 2003), and significant changes in in-lake dissolved PO4 levels have been noted over very short time frames (<1hr) (Robertson, 2011). High P concentrations of up to 1600 mg kg-1 have been reported in the lake sediments (Main et al, 2003)

The poor water quality in Wairewa has had a major impact on lake biodiversity (Reid et al, 2004; Woodward and Schulmeister, 2005; Jellyman and Cranwell, 2007). The lake has long been a crucial food source for Ngai Tahu and is one of only two customary lakes in New Zealand. Te Wairewa Runaga have identified the rehabilitation of the lake as the priority issue in their area. A catchment wide approach to nutrient management with the aim of

4

establishing a Mahinga Kai Cultural Park is proposed, but is unlikely to be successful if the current water quality issues cannot be addressed.

1.4 Rationale for this study

The dynamics of external and internal loading of phosphorus in lake ecosystems is complex and the relative contributions and drivers may be lake-specific. Management responses to lake eutrophication are likely to be expensive, controversial and of a long term nature. Hence there is an onus on lake managers to make decisions based on good quality, lake specific research, as well as the outcomes of community consultation.

The purpose of this study was to develop a preliminary phosphorus budget for Wairewa. Inflows and outflows of P to the lake have been determined, as have the size of in-lake reservoirs of P, and the relative importance of external and internal loading to the variability of in-lake P concentrations.

Sensitivity analyses of the results presented in this report are not included in this report. Further work will also identify the form of P being transported to the lake and more details of this work will be submitted for publication in 2014. It should be noted that this work is part of a wider PhD research project investigating the P reservoirs and dynamics within the lake, and the geochemical processes affecting P solubility.

5

2. Methodology

The establishment of a phosphorus budget for Wairewa involved the measurement of hydrological inflows and outflows to/from the lake over a discrete period of time, combined with the concentration of P in these flows. Reservoirs of P in the lake were quantified from estimates of the mass of lake water, sediment and macrophytes, combined with the concentrations of P in the water, sediment and macrophytes respectively.

The budget may be represented as;

ΔP= (Qin+DRin+GWin)-(Qout+GWout)

Where ΔP = change in P stored in the lake

Qin = P input to lake via stream flow

DRin= P input to lake via direct runoff

GWin= P input to lake via groundwater

Qout = P outflow from lake via lake opening discharge

GWout = P outflow from lake via seep through the beach barrier and to ground water.

The methods used for determining each of these components of the budget are presented below.

2.1 Hydrology

2.1.1 Stream flow Odyssey capacitance water level loggers from Dataflow Systems Ltd were installed on the Okana River (100m downstream of SH75 bridge), Okuti River (Kinloch Road bridge), Catons Bay Culvert (at Rail Trail Bridge) and two smaller streams on the SE side of the lake, L1 and L2 (see Fig 2). The Okana, Okuti and Catons loggers recorded at 10 min intervals for the period 3 Dec 2012 - 2 Mar 2014 (15 Months). The L1 and L2 loggers recorded at 10 min intervals for the period 17 May 2013 – 2 Mar 2014 (10.5 months). Manual stream gauging was carried out in the streams mentioned above, to calculate instantaneous stream discharge (Q), where Q (m3/second) = cross sectional area of the stream (A in m2), x water velocity (V in m/sec). Gauging was conducted in accordance with USGS ’6 tenths method’ (Turnipseed & Sauer, 2010). Velocity measurements were taken with a Global Water FP111 flow probe. Discharge was obtained during low to high flow events (Table 1) over the summer and winter periods of 2012-2014. Minimum and maximum flows obtained by flow gauging are presented in Table 1. Technical issues prevented the use of a remote ADV and this, as well as safety considerations

6

associated with manual gauging, prevented very-high flow discharges being obtained. Some discharge gauging was also undertaken at Sites S1 and S2 (Fig 2). Water level loggers were not installed at these sites and gauging was conducted on limited occasions to allow comparison with the sites which had water level loggers installed. Gauged simultaneous discharges (Q), combined with the water level readings from the water level loggers were used to construct low-high flow ratings curves for the individual streams. This allowed continuous discharges to be obtained for low to high flows by the use of the continuous water level readings. To extend these ratings curves to very high flows, very high flow discharges for the Okana River were obtained by first establishing a correlation between the low- high flow discharges for the Okana, and simultaneous 15 minute discharges for the Huka Huka stream at Lathams Bridge (Huka Huka is a NIWA gauging site on an upstream tributary of the Okana River). This correlation (R2=0.86) was then used to estimate very high flow Okana discharges from very high flow water level logger readings for the Okana River, and simultaneous Huka Huka discharges. These very high flow discharges were then used to constrain the very high flow end of the rating curve.

Figure 2. Location of water level loggers (red dots), and spot gauging (yellow dots). NB Okana river level logger is located 100m downstream of SH75 bridge.

The same process was used to gain very high flow estimates and a rating curve for the Okuti River. The low-high flow correlation with the Okana River (R2=0.78) was used instead of with the Huka Huka.

N

2 km

7

For Catons Culvert a linear ratings curve was applied to the flow data above 0.3 m3/sec. L1 and L2 (Fig 2) similarly had linear rating curves applied above moderate flows (0.2 m3/sec). It should be noted that Catons, L1 and L2 are all ephemeral streams which spend a large part of the year dry. Table 1. Maximum and minimum gauged flows obtained by manual stream gauging

Stream Minimum Gauged Q (m3/sec) Maximum Gauged Q (m3/sec)

Okana River 0.26 4.39

Okuti River 0.18 2.25

Catons Culvert 0.003 1.34

L1 0.002 0.33

L2 0.003 0.21

Figure 3. Areas in which direct runoff was modelled using WEPP. Runoff from Areas 1 and 2 is assumed run directly to the lake while Area 3 is considered to recharge the groundwater area b (Fig 4). The position of the National Climate Database ‘Okuti’ weather station is indicated.

2.1.2 Direct Runoff Direct runoff is water flowing from the land surface and subsurface without becoming significantly channelised. For the purposes of this report it is considered to include all runoff which is not captured by the lakes main tributaries (the Okana and Okuti rivers, Catons

8

Culvert and L1, L2 and S2 streams) and this direct runoff has been modelled for the areas indicated in Figure 3. The south-eastern side of the lake has a significant portion of steep hill slopes to the SW of the S2 catchment shown in Fig 2 (area 1, Fig 3). The NW side of the lake also includes hill slopes which do not drain to significant tributaries (area 2, Fig 3). Overland flow from these areas was modelled using the Water Erosion Prediction Project (WEPP) model from the US Department of Agriculture- Agricultural Research Service (downloaded from http://www.ars.usda.gov/Research/docs.htm?docid=10621, May 2014). The Water Erosion Prediction Project (WEPP) model is a process-based, distributed parameter, continuous simulation, erosion prediction model used for predicting hillslope erosion processes (sheet and rill erosion), as well as simulation of the hydrologic and erosion processes in small watersheds. It was used in this research to predict the volumes of direct runoff water reaching Wairewa. The model requires the following data;

Representative hill slope profiles; orientation, length, and steepness. For this research representative profiles were generated for discrete areas (Fig 3). Area 1 was subdivided into 4 discrete areas each with a separate slope profile. Area 2 and 3 were each modelled using one slope profile. Profiles were generated using 20m contours on 1:50000 scale topographical maps using Freshmap mapping program (accessed from http://www.freshmap.co.nz/).

Climate data file. A climate data file was created using CLIGEN software utilising precipitation, temperatures, solar radiation and wind information. For this research the climate file used was that generated by Hammender (2013) for a sediment discharge estimation study in the Wairewa catchment. This climate file covers 1 January-30 June 2013 and uses data from the Okuti station (Agent number 4925) in the National Climate Database (Fig 3).

Soil data file. Physical and hydrological parameters are input for the catchment soils. The New Zealand National Soil Database Spatial Extension identifies four broad soil types in the catchment with steepland soils predominating on the slopes to be modelled. For this research the soil data file for steepland soils from Hammender (2013) is used. Parameters utilised in the data file include particle size distribution, organic matter content, cation exchange capacity, and interrill/rill erodibility and critical shear values.

Land management inputs. These include groundcover and stocking practises. For this research the percentage of bare soil and exposed rock was estimated from oblique photographs and was input as ‘fallow’ and ‘pavement’ respectively. Forest and scrub cover was estimated from 1:50000 topographical maps and aerial photographs, and was input as ‘5yr perennial forest’. ‘Permanent grass’ was input for the remainder of the modelled areas. Stocking practises were determined from discussions with the farm manager of Kinloch Station and were input as 1 cow/5hectares and 1 sheep/hectare.

Model outputs used for this research were ‘event by event’ runoff estimates (mm). These were then multiplied by the area of the slopes of interest to obtain a volume of direct surface runoff.

The model with the input parameters described above, was tested for the Catons Culvert catchment area. Direct runoff resulting from the June 2013 rainfall event was predicted for 17 and 20-22 June 2013 with a total runoff of 509521m3. Over the same time periods, discharge in Catons Culvert was calculated from the level logger data as 381310m3, 25 %

9

less than predicted. However the discharge over the whole June flood event (baseflow to baseflow =15 June-6 July 2013) was 539142m3, only 5% higher than predicted by WEPP. Hence, while the model may be overestimating direct runoff intensity, it is successfully predicting total direct runoff from the catchment for the June rainfall event and on this basis is assumed to accurately predict direct runoff for the other slopes modelled in this study.

2.1.3 Lake volume and outflow

15 minute lake level data (meters above sea level) is recorded by Environment Canterbury at Site 67063, adjacent to State Highway 75. A relationship between lake level (masl) and lake volume (m3) has been determined by Christchurch City Council (G Harrington, CCC. pers comm), as;

Volume= 684034x2+4000000x-211747 (where x=level in masl)

The timing and lake level changes resulting from artificial openings of the lake at Birdlings Flat are apparent in the lake level data. The change in level and hence volume during these events, was used in this research to quantify the volume lost during the openings. Estimates of seepage across the beach barrier (see Section 2.1.4) during these periods were subtracted from the volume lost, to obtain the volume lost due to the artificial opening.

2.1.4 Groundwater

The lake catchment has limited groundwater storage capacity (shown as area “a” and “b” in Fig 4) and no data is available on groundwater movements within these areas. Input to groundwater storage area “a” is likely to be largely captured by flow calculations at the Okana and Okuti logger sites. Inputs to storage area “b” were estimated as direct runoff from Area 3, (Fig 3) by modelling with WEPP as described in Section 2.1.2. All this runoff was assumed to enter the lake via groundwater.

Figure 4. Groundwater storage areas in the lake catchment (map from Berry & Webster-Brown, 2012)

a

b

10

Seepage across the beach barrier at the Birdlings Flat end of the lake has been estimated at 48 L/sec at a lake level of 1.6 masl (G Harrington, CCC. pers comm). A summer hydrological budget for the lake (Berry and Webster-Brown, 2011) also provided data which was used to calculate a seep rate of 337 L/sec at an average lake level of 1.83 masl (over the period 8-9 Dec 2011). These two estimates were used to construct a linear correlation between lake level and seepage, which was then used to estimate seepage across the barrier during the period of this study. Seepage is assumed to be zero when lake level was below 1.55masl. The average seep rate over this period was 434 L/sec at an average lake level of 1.909 masl.

Comparisons can be made with seepage estimates across the barrier in Lake Ellesmere/Te Waihora where considerably more research has been conducted. The lake level vs seepage rate curve is known to be linear, and the average seep rate is 1000L/sec. The actively seeping barrier frontage is approximately 10 x longer than the barrier at Wairewa. However the hydraulic head at Wairewa is significantly higher as the Te Waihora opening trigger levels for summer and winter are 1.05 and 1.13 masl respectively (Horrell, 1992).

2.2 P concentrations and loads

2.2.1 P concentrations in water

Phosphorus concentrations are reported here in two forms;

a- Dissolved reactive phosphate –phosphorus (DRP). This is the elemental P component of the bioavailable, soluble orthophosphate PO4

-3 ion in a filtered (0.45µm) sample.

b- Total phosphorus (TP) which is all P present in an unfiltered water sample.

DRP was analysed by UV/visible spectrophotometric analysis using the ascorbic acid method (APHA 4500-P Method E). TP was analysed by the same method after a persulphate digestion (APHA 4500-P Method B5). The limit of detection was 0.002mg/L.

Samples were collected in new 50ml polypropylene centrifuge tubes, and transported on ice to the laboratory where they were frozen until analysis. Filtration of DRP samples was undertaken on-site.

Stream water P concentrations used in this report were all from samples collected and analysed by the author. However P concentrations used to calculate the P removed from the lake by outflow, and to calculate the P reservoir in the lake water column, were from both samples collected and analysed by the author, and from water monitoring data from Environment Canterbury (provided by R Webster, Environment Canterbury, pers comm).

Direct Runoff Calculation of P loads transported by direct runoff required a P concentration in runoff water. Five samples of direct runoff were collected from slopes on the SE side of the lake during, or immediately after three separate rainfall events. Also, ten samples were analysed from rainfall simulation experiments conducted in the University of Canterbury Fluid Mechanics Laboratory (Miller, 2013), using soils from slopes on the SE side of the lake. These concentrations are likely to represent high intensity runoff at, or near, runoff peaks. As discussed in Section 2.1.2, the WEPP model appears to over-predict runoff intensity, and the predicted runoff volumes are likely to occur over longer time frames than predicted.

11

Hence these high intensity TP concentrations are likely to represent an upper estimate of TP direct runoff load, and a lower estimate TP runoff load was derived by using the average concentration calculated for Catons Culvert during a high rain fall event (June 2013) . The WEPP model utilised a climate database for Jan- June 2013 (6 months). To estimate the TP load over the total period (3 Dec 2012-2 Mar 2104), a multiplier was applied which was derived as follows;

Catons Culvert Load (15 months)/Catons Culvert Load (6 months)

Multiplier for TP = 1.13

Groundwater Calculation of groundwater P loads required a P concentration for groundwater flows. This data was not available, however a single sample was obtained of water seeping from subsoil after a rainfall event in August 2012. The DRP concentration obtained for this sample was used to calculate a lower estimate of the P load entering the lake from groundwater area ‘a’ (Fig 4). The average DRP concentration for Catons Culvert during a high flow event (June 2013) was used to calculate an upper estimate for ground water load. To estimate the DRP load over the total period (3 Dec 2012-2 Mar 2104), a multiplier was applied which was derived as follows;

Catons Culvert Load (15 months)/Catons Culvert Load (6 months)

Multiplier for DRP = 1.24

Lake Opening Outflow Two artificial lake openings occurred during the research period. The first opening was for a period of 6 days from 7-12 June 2013. TP (0.043 mg/L) and DRP (0.005 mg/L) concentrations were used from Environment Canterbury lake monitoring sampling conducted on 10 June 2013. The second opening was for 15 days over the period, 22 June-6 July 2013. No lake sampling was conducted over this period. Instead concentrations for TP (0.130mg/L) and DRP (0.023 mg/L) were utilised from lake sampling in March 2014 which took place during a similarly large flood event, which is assumed to have resulted in similar concentrations of suspended sediment associated P in the lake. Beech Seepage P concentrations used to calculate the P removed from the lake by seepage through the beach barrier are the average DRP concentration from Environment Canterbury monitoring data for the lake over the study period. Beach seepage is likely to only contain dissolved P, as particulate-bound P will be filtered out by passage through the sediment and substrate of the beach barrier.

2.2.2 P concentrations in sediment

Lake sediment samples were collected in 25 cm cores using a Uwitec 90mm X 60cm sediment corer. Cores were sectioned (1-5cm = 1cm increments, 5-15 cm = 2cm increments, 15-25cm = 5cm increments), and samples transferred to new 50ml polypropylene centrifuge tubes for transport, on ice, to the laboratory, Pore water was separated from the sediment by centrifuging at 4000rpm for 40mins. The pore water was then filtered through 0.45µm Millipore membrane filters and frozen until analysis. As volumes were generally too small

12

for analysis by spectrophotometry, the pore water samples were analysed for dissolved phosphorus by ICP-OES at the Department of Soil and Physical Sciences, Lincoln University.

The sediment was also frozen until analysis. A sequential chemical extraction analysis was undertaken on the thawed samples, following the scheme used by Rydin (2000). These sequential extractions allow different P fractions within the sediment to be identified according to their ability to be extracted by various chemical reagents. The total P concentrations resulting from these extractions, along with sediment density figures obtained from the sediment cores, were used to calculate the P reservoir in the lake sediments.

A more detailed chemical analysis of the P associated with the suspended sediment carried by the tributary streams is yet to be conducted.

2.2.3 P concentrations and mass in lake macrophytes

A macrophyte biomass survey was conducted in March 2013. Four sites were surveyed by diving with a ring-net sampler of known area. Triplicate weed samples collected in the sampler at each site were weighed. Two sub-samples were transported to the laboratory for drying and TP analysis. Biomass results from the diving survey were combined with visual observations of macrophyte density delineated by GPS, to provide an estimate of biomass in the lake. The southern end of the lake was not surveyed, but significantly lower macrophyte densities have been observed there over the period of this research. A lower estimate of biomass was obtained by assuming no macrophyte growth at the southern end of the lake, while an upper estimate was obtained by assuming equal macrophyte growth at the southern and northern ends of the lake.

TP concentration of the macrophytes was obtained for duplicate samples which were ground to a fine powder after drying at 1050C for 24 hrs, by first igniting the sample at 5500C for 4 hrs and then boiling in 1 M HCl for 15 minutes (Anderson, 1976). The sample was then analysed by UV/visible spectrophotometric analysis using the ascorbic acid method (APHA 4500-P Method E).

2.2.4 P load curves

Relationships established by correlating stream water P concentrations with simultaneous stream discharge (Fig 5) are referred to in this report as load curves. For the Okana and Okuti rivers, different load curves were evident between summer and winter, with summer TP concentrations being higher for similar flows being than those in winter (Fig 5). These different load curves were used to calculate TP loads for the Okana and Okuti Rivers. Summer low flows (Q<0.57m3/sec) for these rivers showed considerable scatter for TP, possibly reflecting prolonged periods of low flow and the occasional short term and small, increase in low flows ie; low flow ‘fresh’ events. For these summer low flow conditions an average concentration of 0.082 mg/L for the Okana River, and 0.064 mg/L for the Okuti River respectively, were used.

DRP is poorly correlated with flow (Fig 6), and average values were used for summer and winter to calculate loads (Table 2). The exception to these poor correlations was the Okuti summer high flow and Catons Culvert DRP concentrations which showed increasing concentrations with increasing flow (Okuti = R2 =0.53, Catons = R2=0.87). However these

13

regression lines were highly dependent on single, high flow data points. Comparison with concentrations and concentration/flow relationships in other catchment streams indicated that the DRP concentrations that would be derived from the regression relationships for these two data sets would be unrealistically high. Hence, average DRP concentrations were used for the Okuti summer high flows and Catons all season flows, in common with all other catchment streams. For the Okana and Okuti Rivers different DRP concentrations were used for low and high summer flows.

For the smaller tributaries (Catons Culvert, L1, L2), there was either no difference between summer and winter concentrations, or there was not enough data to differentiate. A single load curve for TP was constructed for these streams (Fig 5 for Catons Culvert). A single, all season average was used for DRP concentrations (Table 2)

The level loggers for L1 and L2 streams were not in place for the entire period (15 months) and hence continuous flow data for L1 and L2 streams was limited to 10.5 months. In order to gain an estimated P load for the entire period a multiplier, based on Okana loads was applied to the L1 and L2 loads. This multiplier was based on the following;

Okana load (15 months)/Okana load (10.5months)

multiplier for TP = 1.248

multiplier for DRP=1.313

The load delivered to the lake by Stream S2 (Fig 2) was assumed to be the same as L2, which is the immediately adjacent and similarly sized stream catchment.

Table 2. Average values (where used) for summer low flow TP concentrations, and summer/winter all flows DRP concentrations used in P load calculations.

Stream DRP (mg/L)

summer average

DRP (mg/L)

winter average used

Q<0.57m3/sec Q>0.57m3/sec

Okana River 0.032 0.043 0.025

Okuti River 0.042 0.040 0.023

Catons Culvert All season average= 0.022

L1 All season average= 0.025

L2 All season average= 0.030

14

Figure 5. Constructed TP load curves for the Okana and Okuti Rivers and Catons Culvert. Q= stream flow. Note the separation of summer (Q>0.57m3/sec), summer low flow (Q<0.57m3/sec) and winter flows in the Okana and Okuti Rivers. The Catons all-season load curve is separated into two flow regimes (0.1>Q>0.1 m3/sec)

15

Figure 6. DRP concentrations vs flow for the Okana and Okuti Rivers and Catons Culvert. Note higher summer concentrations in Okana and Okuti Rivers. See Table 2 for average DRP concentrations used for load calculations. See section 2.2.4 for discussion of regression relationships.

16

2.3 Uncertainties in P loads and P reservoirs.

Uncertainties were calculated for both P loads and reservoirs as follows;

Uncertainty in P loads

A ± 25% uncertainty in P load calculation was derived from analytical and measurement errors inherent in both the P concentrations and flows used to calculate the P loads transported by the Okana River. The TP concentration error of ± 15% was based on a 95% confidence interval for replicate analyses of P standards and stream water samples. The error associated with water flow was calculated by applying an ‘uncertainty of measurement’ analysis (0.5 x the smallest feasible measurement unit) to flows up to the highest gauged flow (4.5m3/sec). To flows higher than 4.5m3/sec, a 95% prediction interval was calculated for extrapolated flows. A similar uncertainty is assumed to apply to all P loads in this report.

Uncertainty in the P reservoirs

A ± 20% uncertainty was based on a 95% confidence interval for replicate sediment extractions, testing both experimental uncertainty and variability between sediment samples taken in close proximity. A similar uncertainty is assumed to apply to all P reservoirs in this report.

17

3 Results

3.1 P load transported to the lake

3.1.1 Stream Flows

Table 3 presents the results of P loads transported to the lake by streams (as well as overland flow and groundwater), calculated from the established rating and load curves. In excess of 14000kgs of P were transported into the lake over the 15 month period. Table 3 also presents the P loads delivered by significant flood events during the period. The Okana River delivered by far the greatest P load, and a large proportion of this was delivered in a single flood event (June 2013). The Okana and Okuti Rivers together transport 96% of the TP load, and 72% of the TP load was transported by these two rivers in the June 2013 flood event alone. Fig 7 illustrates the 15 min loads and cumulative loads for Okana and Okuti Rivers and Catons Culvert over the research period. The cumulative load plots confirm the importance of the flood events on the delivery of P to the lake. Even the figures presented in Table 3 and Fig 7 are likely to underestimate the importance of these flood events, because in the June 2013 event, water levels overtopped the level loggers and hence estimated flows reflect the highest reading of the level logger rather than the actual water level. The Okana River level logger was overtopped for approximately 24hrs and the Okuti River level logger was overtopped for approximately 5hrs.

Table 3 also presents the loads delivered as TP, particulate P (the difference between TP and DRP) and DRP. Consistent with the amount of P delivered during flood events, overall 87% of the P load is associated with particulate material.

Table 3. TP and DRP loads delivered to Wairewa by stream flows, overland flow and ground water over the 15 month period 3 Dec 2012-2 Mar 2014.

Stream TP load (kg)

% of total catchment TP load

% of TP load in single flood event (June 2013)

% of TP load in 5 largest flood events

DRP Load (kg)

% Particulate P

% DRP

Okana 10410 74 81 86 1040 90 10

Okuti 3160 22 58 75 660 79 21

Catons 320 2.3 60 88 40 88 12

L1 45 0.3 57 58 13 72 29

L2 35 0.2 88 88 7 79 20

S2 35 0.2 88 88 7 79 20

Direct Runoff 120 0.9 61 - 35 73 29

Ground water - <0.1 - - 10 0 100

Total Catchment

14125 100 73 82 1812 87 13

18

Figure 7. 15 minute and cumulative TP loads for the three main tributaries of the lake. Note the different scale on the Catons Culvert plot axes.

19

3.1.2 Direct Runoff

Table 4 presents calculated P loads transported to the lake in direct runoff from areas 1 and 2 (Fig 3). As discussed in Section 2.2.1, the upper estimate for these loads was derived from high intensity runoff sample concentrations collected during or immediately after rainfall events. The WEPP model used to predict runoff, appears to over-predict runoff intensity but more accurately predicts the total runoff over the longer time frames that ephemeral streams are likely to take to return to pre-flood levels. Hence the lower estimate for the load may be more realistic as it was derived from an average concentration for Catons Culvert discharge over a longer time frame flood event (June 2013). This lower estimate is used in the P budget presented in Section 3.3. Table 3 allows comparison of overland flow loads with loads transported by the catchment streams.

Table 4. Concentrations used to calculate upper and lower limits of direct runoff TP load from areas 1 and 2 (Fig 3). Loads presented are for the 15 month research period. An explanation of the concentrations used is provided in Section 2.2.1

TP concentration (mg/L)

TP load (kg) Area 1 runoff

TP load (kg) Area 2 runoff

TP Load Total (kg)

Upper estimate (runoff sample concentration)

0.378 116 217 334

Lower estimate (Catons concentration)

0.138 42 79 120

3.1.3 Groundwater

The calculated load to the lake from the groundwater storage area “b” (Fig 4) is 4-21 kg. This is based on all the modelled runoff from area 3 (Fig 3) being delivered to the lake, ie: no change in groundwater storage, and DRP concentrations of 0.007 mg/L (seep sample) and 0.037mg/L (Catons Culvert June flood average) respectively. An average of these two values is used in the P budget presented in Section 3.3. Table 3 allows comparison of groundwater flow load with loads transported by catchment streams and overland flow

3.2 P load transported from the lake

A total P load of 1900 kg was calculated to have been transported out of the lake during artificial opening events over the research period. Of this 310 kg (16%) was transported as DRP. The initial 6 day opening event (7-12 June) transported a TP load of 350 kg (DRP= 37 kg) and the second opening (22 June-6 July) transported a TP load of 1550 kg (DRP=310 kg).

Seepage through the beech barrier was calculated to transport 250 kg of P during the research period, all of which is assumed to be DRP. Hence it was apparent that by far the biggest outflow of P (88%) was via the artificial openings.

20

3.3 P budget and reservoirs

Table 5 presents a summary of inflow and outflow P loads over the 15 month research period, which shows that approximately 12000 kg of P were retained in the lake during the period.

Table 5. Phosphorus budget for 15 months 3 Dec 2012- 2 Mar 2014

Budget Component Total P (kg)

Qin Stream flow input 14000

DRin Direct Runoff input 120

GWin Groundwater flow input 10

Qout Lake opening outflow 1900

GWout Beach Barrier Seep 250

ΔP retained in lake 11980

The high and low concentrations of TP recorded in the lake during the period of this research were 0.191 mg/L (sampled and analysed by this author, 10 April 2013) and 0.034mg/L (sampled and analysed by ECan, 14 May 2013) respectively. These analyses illustrate the temporal variability which occurs in the lake and were used, in conjunction with lake level at the time of analysis, to calculate the range of TP in the lake water-column reservoir, as presented in Table 6.

Table 6 also shows estimates for size of the sediment, pore water and macrophyte P storage reservoirs in the lake. The sediment and pore water P ranges reflect the maximum and minimum concentrations seen in three core samples taken at different times. The lake sediment constitutes a very large store of P, relative to the water column or macrophytes. The sediment divisions shown in Table 6 were based on relative enrichment and P mobility, as observed in the cores. Analysis of core profiles showed that the top 5cm of the sediment is enriched in mg P/kg dry weight sediment, relative to the deeper sediments. Comparison between cores showed that P in the top 10cm also appeared to be more mobile than in deeper sediments, and by a depth of 25cm the sediment P appears to be relatively immobile.

Table 6. In-lake reservoirs of P

Reservoir Total P (kg) % of 0-25cm sediment reservoir (sediment only)

Lake water column 140-3130

Sediment Pore Water (0-25cm) 210-1430

Sediment (0-5cm) 79700-101900 18-20

Sediment (0-10cm) 157140-210290 36-41

Sediment (0-25cm) 432300-517100 100

Macrophytes 100-150

21

3.4 P variations in the lake

Fig 8 illustrates a period of time (5 March -10 April 2013) during which the water column TP concentration increased by 0.118mg/L, a 160% increase. This was a period of low stream flows in the catchment when the maximum possible increase in lake TP concentration resulting from external loading was 0.011 mg/L, a 15% increase. Hence the TP concentration increase significantly exceeds that which can be explained by external loading, confirming the internal loading from lake sediment as a major source of P to the water column.

Figure 8. TP concentrations measured in the Wairewa water column over the period 5 March-10 April 2013 as well as the maximum possible increases that could result from the TP load transported to the lake by catchment streams over the same period.

22

4. Discussion and Summary

4.1 The P budget

Figure 9a presents a stylised annual budget for Lake Forsyth/Te Roto o Wairewa. This assumes monthly average P loads to/from the lake based on the 15 month period of this research. While ongoing research, may alter the exact value of these P loads the relative importance of key P transfer processes, or reservoirs, is unlikely to change significantly.

A large amount of P was transported to the lake via the main tributaries, predominantly the Okana and Okuti rivers which together account for >95% of the total load. This is mainly (>85%) associated with particulate material, and hence flood events are of major importance in P transport in the catchment. Over 70% of the annual P load during the 15 month period was delivered to the lake in a single flood event and >80% of load was transported during the five largest flood events during the period. The effect of these flood events on the TP load to the lake can be clearly seen in Fig 7 where the cumulative TP load transported by the Okana River increased by an order of magnitude over the June 2013 flood event. The cumulative load transported by Catons Culvert, similarly underwent a flood associated, order of magnitude increase, in May 2013.

Direct runoff inputs from the steep slopes on the south-east and north-west sides of the lake comprise <1% of P loads to the lake. Even when combined with ephemeral streams (Catons Culvert, L1, L2 and S2), the contribution to the total P load remains very small (4%). Similarly, calculated groundwater contributions to P loads are very small (<0.1%). Only area ‘b’ (Fig 4) has been quantified, and this is based on recharge by modelled runoff from area 3 (Fig 3) and the assumption that there is no change in groundwater storage. No direct measurement of groundwater movements or P concentrations were available, however the limited size of the potential groundwater storage areas (Fig 3), the modelled recharge volumes of area “b”, and the very small effect that the beach barrier seepage had on the budget (<2% of stream inflow), indicates that the contribution of ground water to the budget are likely to be small.

4.2 The P reservoirs

Figure 9b presents stylised P reservoirs in Te Roto o Wairewa/Lake Forsyth. A large proportion (> 80%) of the transported P load is retained in the lake, due to the limited outflow. However the reservoir of P in the lake water column is small relative to P stored in the sediment, even during periods of maximum turbidity resulting from suspended sediment or high primary productivity.

The largest store of P is the sediment reservoir reflecting the high affinity P has for particulate material, combined with sedimentation in the lake. Other research as part of the study of Wairewa P dynamics (not reported here) indicate that the mobility of P in the sediment, and hence the ability of the P to be released to the pore water (and potentially the lake water column), appears to reduce with sediment depth. The upper 5cm of the sediment is enriched in P relative to the deeper sediments, and a ‘dynamic zone’ of mobile P exists above 10cm. By 20-25cm depth P mobility appears to be very limited. Therefore the

23

sediment reservoir calculated in this research has been divided into zones reflecting the P which is potentially available for the enrichment of the lake water column (Table 6).

The reservoir of P in the lake macrophytes is poorly constrained at lake’s southern end, and

is also likely to be variable over time. However relative to the sediment reservoir, the

macrophyte P reservoir is small (<0.1%) and variability is likely to have little effect on the

lake P budget

Figure 9. a. An annualised phosphorus budget for Lake Forsyth/Te Roto o Wairewa. Blue

arrows and figures are inflows, Red arrows and figures are outflows. b. P reservoirs in Lake Forsyth/Te Roto o Wairewa.

24

4.3 External vs internal P loading

Water column concentrations of P were highly variable over time within the lake. This variability did not appear to be the direct result of external loads delivered from the catchment because elevated P concentrations in the lake water column often did not coincide with periods of high P transport from the catchment to the lake. The increase of water column P concentration during late summer 2013, for example, could not be explained by the measured transport of P from the catchment to the lake. Hence much of the variation observed in P concentrations in the lake water column must be due to the flux of dissolved P to/from the lake sediment, referred to as internal loading. The mechanisms that drive this flux, particularly of bioavailable DRP which is available for algal growth, are the subject of ongoing study.

4.4 Implications for lake management

Management of the Wairewa lake environment will need to respond to phosphorus loading issues. This study has highlighted two key problems that will need to be addressed;

the transport of P to the lake by the Okana and Okuti Rivers, especially during high flow events (external loading), and

the release of P from the large reservoir of P-bearing lake sediments to the water column (internal loading).

Management interventions hoping to reduce external P loading need to focus on controlling the transport of P bearing suspended sediment by the Okana River, particularly during large flood events. In addition, future monitoring and research should target these high flow events in order to confirm the preliminary findings of this research over a greater period of time and more variable weather conditions, and/or to gauge the effectiveness of management responses.

A range of management options have been attempted around the world, for the control of internal P loading to lakes. Physical methods include sediment dredging and engineering solutions such as enhancing outlet flows. Biological controls may include harvestable floating wetlands, the permanent establishment of macrophyte beds or increasing the populations of algae-grazing zooplankton. Chemical methods range from the artificial oxygenation of bottom water to sediment capping with various P inactivation agents. The appropriateness of any management intervention is likely to reflect the lake-specific dynamics that drive P release from the lake sediments. Work is currently underway to determine P speciation in Wairewa sediments, and the geochemical dynamics that control this, as well as the in-lake conditions which drive the release of P from the sediment to the water column.

25

5. References

Abell, J.M., Ozkundakcia, D., Hamilton, D.P., Miller, S.D. 2011. Relationships between land use and nitrogen

and Phosphorous in New Zealand lakes. Marine and Freshwater Research, 62, 162–175

Anderson, G. 1976. An ignition method for the determination of total phosphorus in lake sediments. Water

Research 10. 329-331

APHA 4500-P. In Standard Methods for the Examination of Water and Wastewater. 21 Ed. 2005. Eaton, A.D.C.,

Rice, E.W., Greenberg, A.E. Ed’s. American Public Health Association; American Water Works Association;

Water Environment Federation: Washington DC, USA.

Berry, N., Webster-Brown, J. 2012. A summer hydrological budget for Lake Forsyth/ Wairewa: Preliminary

Findings. Unpublished Waterways Center for Freshwater Management Report WCFM Report 2012-004

Burger, D.F. 2006. Dynamics of Internal Nutrient Loading in a Eutrophic, Polymictic Lake (Lake Rotorua, New

Zealand) Unpublished Doctorate Thesis, University of Waikato, New Zealand.

Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, A.N., Smith, V.H. 1998. Nonpoint Pollution

of Surface Waters with Phosphorous and Nitrogen Reviewed work. Ecological Applications 8 (3) 559-568

Christophoridis, C., Fytianos, K. 2006. Conditions Affecting the Release of Phosphorous from Surface Lake

Sediments. Journal of Environmental Quality. 35. 1181-1192.

Downs, T.M., Schallenberg, M., Burns, C.W.2008. Responses of Lake Phytoplankton to Micronutrient

Enrichment: A Study in Two New Zealand Lakes and an Analysis of Published Data. Aquatic Science70. 347-360.

Gao, L., Zhou, J.M., Yang, H., Chen, J. 2005. Phosphorous Fractions in Sediment Profiles and Their Potential

Contributions to Eutrophication in Dianchi Lake. Environmental Geology 48. 835-844.

Hammender, K. 2013. Sediment discharge estimation for steep terrain catchments with loess soil in New

Zealand. Unpublished thesis, Univeristy of Natural Resources and Life Science, Vienna, Austria.

Horrell, G.A. 1992. Lake Ellesmere Water Balance: Variable Analysis and Evaluation. Unpublished Masters

Thesis, University of New South Wales, Australia.

Jacoby, J.M., Lynch, D.D., Welch, E.B., Perkins, M.A. 1982. Internal Phosphorous Loading in a Shallow Eutrophic Lake. Water Research 16. 911-919 Jellyman, D., Cranwell, I. 2007. The Status of Eel Stocks in Wairewa (Lake Forsyth). Ministry of Fisheries. New Zealand Fisheries Assessment Report 2007/11. Kristensen, P., Sondergaard, M., Jeppesen, E. 1992. Resuspension in a Shallow Eutrophic Lake. Hydrobiologia 228. 101-109. Kumar Das, S., Routh, J., Roychoudhury, A.N., Klump, J.V., Ranjan, R.K. 2009. Phosphorous Dynamics in Shallow Eutrophioc Lakes; An Example from Zeekoevlei, South Africa. Hydrobiologia 619. 55-66. Lynn, I. 2005. Lake Wairewa/Lake Forsyth: Potential Sources of Phosphorous to the Lake. Wairewa Programme Report Series 2005/01. Main, M.R., Lavender, R.M., Hayward, S. 2003. The Okana River: Assessment of Water Quality and Ecosystem Monitoring, July 1992 to May 2002 and Water Quality Implications for Lake Forsyth/Wairewa. Environment Canterbury Technical Report U03/20.

26

Miller, J. 2013. Investigating Sources of Sediment Input to Lake Forsyth/Waiwera. Waterways Centre for

Freshwater Management Summer Scholarship Report. University of Canterbury and Lincoln University. Ministry for the Environment. 2010. Lake Water Quality in New Zealand 2010; Status and Trends. Ministry for

the Environment, Wellington.

Pacini, N., Gachter, R. 1999. Speciation of Riverine Particulate Phosphorous During Rain Events.

Biogeochemistry 47. 87-109.

PCE- Parliamentary Commissioner for the Environment. 2004. Growing for Good: Intensive Farming,

Sustainability and New Zealand’s Environment.. Parliamentary Commissioner for the Environment. Wellington.

Parliamentary Commissioner for the Environment. 2012. Water Quality in New Zealand: Understanding the

Science. Parliamentary Commissioner for the Environment. Wellington.

Reid, M., Wybrow, R., Woodward, C. 2004. Managing Te Roto o Wairewa: Lessons from the Past. Water and

Atmosphere 12(4)

Reynolds, C.S., Davies, P.S. 2001. Sources and Bioavailability of Phosphorous Fractions: A British Perspective.

Biological Review 76. 27-64.

Robertson, P. Diurnal Variation in Wairewa Water Quality. Waterways Centre for Freshwater Management Report 2011-003. University of Canterbury and Lincoln University.

Rydin, E. 2000. Potentially Mobile P in Lake Erken Sediments. Water Research 34(7) Schindler, D.W., 2006. Recent Advances in the Understanding and Management of Eutrophication. Limnology and Oceanography 51(1 part2). 356-363. Schallenberg, M., Sorrell, B. 2009. Regime Shifts Between Clear and Turbid Water in New Zealand Lakes; Environmental Correlates and Implications for Management and Restoration. New Zealand Journal of Marine and Freshwater Research 43. 701-712. Sondergaard, M., Jensen, J.P., Jeppesen, E. 2003. Role of Sediment and Internal Loading of Phosphorous in Shallow Lakes. Hydrobiologia506-509. 135-145. Sondergaard, M. 2007. Nutrient Dynamics in Lakes- with Emphasis on Phosphorous, Sediment and Lake Restoration. Unpublished Doctorate Thesis, National Environmental Research Institute, University of Aarhus, Denmark. Soons, J.M., 1998. Recent Coastal Changes in Canterbury- the Case of Lake Forsyth/Wairewa. New Zealand Geographer 54(1) Turnipseed, D.P., Sauer,V.B. 2010. Discharge Measurements at Gaging Stations. U.S Geological Survey. Techniques and Methods Book 3, Chapter A8. Wetzel, R.G. 2001. Limnology; Lake and River Ecosystems. Academic Press, Elsevier. San Diego, U.S.A. Woodward, C.A., Shulmeister, J. 2005. A Holocene Record of Human Induced and Natural Environmental Change from Lake Forsyth (Te Wairewa), New Zealand. Journal of Paleolimnology 34: 481-501.

27

Waterways Centre for Freshwater Management University of Canterbury & Lincoln University Private Bag 4800 Christchurch New Zealand

Phone +64 3 364 2330 Fax: +64 3 364 2365

www.waterways.ac.nz