The importance of plant water use on evapotranspiration covers in semi-arid Australia

Introduction Conclusions References


Introduction
Evapotranspiration (ET) cover systems, also referred to as monolithic cover, alternative cover or phytocaps, are increasingly accepted for the closure of mining waste facilities, such as waste rock dumps and tailings storage facilities, as well as municipal landfills (Benson et al., 2002;Knoche et al., 2006;Madalinski et al., 2003;Rock, 2010;Introduction Conclusions References Tables Figures

Back Close
Full  Waugh et al., 1994;Yunusa et al., 2010).The objective of those covers is to minimise drainage into the underlying hazardous wastes by maximising vegetation rainfall interception, soil water storage, and ET, the combination of direct evaporation from the soil surface and transpiration from vegetation, and thereby minimising surface runoff and deep drainage (Salt et al., 2011).After rainfall ceases, the loss of stored soil water through ET increases the soil water storage capacity for future rainfall events (e.g.Hauser et al., 2001;Rock, 2010).Cover materials as well as cover designs vary greatly in their characteristics and complexity and may incorporate geotextile liners, multiple soil layers with selected hydrological properties, and compacted clay layers, depending on local climate and material availability (Benson et al., 2002).
Most previous ET studies did not directly measure plant water usage and therefore did not consider partitioning between plant transpiration and evaporation from bare soil.The plant contribution to ET from cover systems may vary from 44-93 % in temperate eastern Australia (Yunusa et al., 2010) to about 22 % in Mediterranean Western Australia (Gwenzi, 2010).Furthermore, plant community maturity and composition are likely to substantially affect the water usage by vegetation (Barnswell and Dwyer, 2011;Smesrud et al., 2012).In contrast, information about the effect of vegetation composition on ET losses from cover systems in semi-arid climate is limited, but it is necessary to identify plant species that can both reliably minimise drainage of water through cover systems and maintain robust vegetation cover during prolonged unfavourable weather conditions.Introduction

Conclusions References
Tables Figures

Back Close
Full Therefore, the aim of this study was to investigate ET rates of two native shrub species on an ET cover design in semi-arid Australia to identify the main drivers of ET under semi-arid conditions, quantify water use characteristics of key plant species, and develop indicators of vegetation cover and species composition for the management of ET cover systems.

Study site
The study site was located near Cobar, western New South Wales, Australia (31 • 33 S, 145 • 52 E).The climate of this area is described as BSh (B = arid, S = steppe, h = hot) in the K öppen and Geiger climate classification (Kottek et al., 2006), with a long term average annual rainfall of approximately 400 mm (Bureau of Meteorology, 2012).Rainfall distribution does not follow a seasonal pattern but tends to be uniformly distributed throughout the year.However, rainfall can be highly variable, especially in early spring and summer.Mean annual minimum and maximum temperatures are 12.8 • C and 25.2 • C, respectively (Bureau of Meteorology, 2012).
Evapotranspiration and evaporation was investigated at a 2.0 m thick test plot cover system (35 × 35 m plot area) constructed from benign (non-acid-forming) waste rock (O'Kane Consultants Inc., 2002), overlain by 150 mm loamy topsoil.A water retention curve was determined for the topsoil by core sampling and laboratory analysis.From the water contents ranging between saturation (0.45 m 3 m −3 ) and residual (0.0 m 3 m −3 ), the van Genuchten parameters (van Genuchten, 1980) α and n were derived as 0.0468 and 1.22, respectively.Scattered shrub, herb and grass species had established on the cover in the three years between topsoil application and measurement.Vegetation cover was determined by a modified Braun-Blanquet method (Mueller-Dombois and Ellenberg, 1974) and by foliage projective cover (FPC) (Specht, 1981).Up to 30 species provided average total vegetation coverage (FPC v ) of 26 %.Based on a vegetation Introduction

Conclusions References
Tables Figures
The two species differ markedly in shoot and leaf morphology.Senna artemisioides is an erect shrub with a pinnate arrangement of terete leaves resulting in a feathery appearance of the canopy.Therefore, shadows cast by individual plants are diffuse.In contrast, Sclerolaena birchii individuals are low-growing but develop a dense system of hairy shoots with small persistent elliptical leaves and spiny fruits that more completely shade a greater area of the soil surface than does Senna artemisioides (Fig. 1).

Experimental design
A field experiment was conducted over a period of 5 days in April 2011 to establish the relationships between atmospheric vapour pressure deficit (VPD), and the evapotranspiration of an individual plant (ET i ).The latter was used to predict the actual evapotranspiration of the test plot (ET plot ) in relation to various scenarios of coverage and potential evapotranspiration (pET).Two individuals of each species (Senna artemisioides (Sen) and Sclerolaena birchii (Scl)) were selected to estimate ET i .Likewise, one spot with no vegetation was selected to estimate evaporation from bare soil (E b ), providing a total of five measurement locations on the test plot.Furthermore, a representative replicate of each Senna artemisoides, Sclerolaena birchii and the bare soil were selected for destructive soil moisture sampling in the upper 5 cm.Initially, all measurement locations and replicates were watered to simulate a 17 mm rainfall event, which was sufficient to raise the near-surface soil water content of each location markedly.ET i measurements were conducted with an open top chamber (OTC, detailed below) over the course of a day from dawn (06:00 h) to dusk (18:30 h) and then integrated to estimate daily values of ET i .The OTC was removed after 10 min of measurement to minimise micro-climatic disturbances.Evapotranspiration was assumed to be negligible at night although we are aware of studies that indicate the occurrence of nocturnal evapotranspiration for other vegetation types (Donovan et al., 1999).The Figures

Back Close
Full soil water content of the top soil layer (5 cm) was measured both in the morning and evening.A weather station at the experimental site provided meteorological information.The weather situation prior to the experiment was characterised by dry conditions with precipitation less than 33 mm over the previous 6 weeks and typical for the climate in this semi-arid region.

Atmospheric variables
The hydrological boundary conditions of ecosystems, particularly with regard to the plant available soil water balance, are set by the atmospheric demand for water vapour, expressed through the vapour pressure deficit (VPD): where e s and e a denote the saturated and actual vapour pressures [kPa], respectively.The saturated vapour pressure was calculated according to Murray (1967) as: e s = 0.611 exp 17.27(T a − 273.16) where T a is the actual air temperature [K].The actual vapour pressure was calculated according to Monteith and Unsworth (1990) as: where H r denotes the relative humidity [%].
In addition to VPD we used the potential evapotranspiration (pET [mm d −1 ]) to estimate the atmospheric water demand; pET was calculated with the FAO56 Penman-Monteith method (Allen et al., 1998):

Conclusions References
Tables Figures

Back Close
Full where ∆ denotes the slope of the vapour pressure curve [kPa T is the mean daily air temperature at 2 m height [ • C], VPD is the vapour pressure deficit [kPa], and u denotes the wind speed at 2 m height [m s −1 ].

Evapotranspiration
Evapotranspiration measurements were conducted with an open top chamber (OTC), similar to the systems described by Hutley et al. (2000).The total volume of the chamber was 1.42 m 3 and ground area was 0.64 m 2 .The metal frame was covered by a clear PVC foil of 0.2 mm thickness.Measurements of photosynthetic active radiation (PAR) showed that the attenuation due to the foil was 9 % under cloud-free skies and 20 % under cloudy conditions, which is in the range of attenuation reported by other studies (Garcia et al., 1990;M üller et al., 2009).Air was pumped into the chamber through a centrifugal fan at the base of the chamber.Air flow was measured at the chamber outlet through a vane thermo-anemometer (RS Components Ltd., Smithfield, NSW, Australia).Air temperature and relative humidity inside and outside the chamber were measured with a Vaisala HUMICAP ® Humidity and Temperature Probe (HMP155, Vaisala, Inc., Helsinki, Finland), and PAR with a LI-190 quantum sensor (LI-COR, Lincoln, Nevada, USA).The evaporation rate from bare soil (E b [mm s −1 ]) was calculated according to Hutley et al. ( 2000): where subscript "b" denotes measurements of bare ground, V is the volumetric air flow rate [m 3 s −1 ], ∆ρ is the difference in vapour densities outside and inside the chamber The evapotranspiration rate from the area covered by an individual plant i ET i [mm s −1 ]) was then calculated as: where the subscripts "v" and OTC refer to vegetation and chamber measurement, respectively.The chamber was tested under controlled conditions prior to the field experiment by comparing the gravimetric water loss from a tray (E g ) to the evaporational water loss measured by using the chamber (E OTC ).The OTC was tested for two different wind speeds to account for a possible range of speeds of air flow during the experiment.For both high (0.8 m s −1 ) and low (0.4 m s −1 ) values of air flow, highly significant positive linear relationships were found (R 2 = 0.94 and R 2 = 0.97, respectively) between E OTC and E g .This emphasises the reliability of evapotranspiration rates estimated using the chamber.However, since E OTC slightly underestimated E g , the measured values of ET i and E b during the field experiment were corrected accordingly.We used the estimated values of ET i to predict the actual evapotranspiration from the test plot (ET plot [mm d −1 ]) under different scenarios of species composition as: where

Results
Climatic conditions on all measurement days were similar with sunny mornings and slightly overcast, windy afternoons.Maximum VPD occurred around 14:00 h and varied between 4.4 kPa on 8 April and 3.9 kPa on 6 April.Air temperature fluctuated between 10-15 • C in the mornings and 30-36 • C in the afternoons.Soil moisture was markedly increased through watering and decreased rapidly thereafter.However, volumetric soil moisture seven days after watering exceeded moisture conditions prior to watering (Fig. 3).
Figure 4 illustrates the integrated daily values of ET for each individual plant and bare soil.Generally, for all OTC measurements, ET decreased over the period of observation.Daily ET was most elevated for Senna artemisioides and Sclerolaena birchii, and lowest for bare soil on all three days of measurement.However, three different temporal characteristics of daily ET/E were observed for the two plant species and bare soil.For Senna artemisioides, daily ET was highest (e.g.4.8 mm d −1 for Sen1) three and five days after watering, but dropped to 3.1 mm d −1 at seven days, while for Sclerolaena birchii no such distinctive pattern of daily ET was detected (e.g., 2.6, 2.4, and 2.0 mm d −1 for Scl1).Contrary to both plant species, evaporation from bare soil (E b ) was elevated on the third day only (1.2 mm d −1 ) and dropped substantially to 0.7 mm d −1 five and seven days after watering (Fig. 4).ET i , E b , and VPD followed diurnal courses as shown in Fig. 5.For both Senna artemisioides and Sclerolaena birchii, ET i exceeded E b as early as 07:00 h, indicating the role vegetation plays for water extraction.Senna artemisioides showed markedly higher values of diurnal ET three and five days after watering compared to Sclerolaena birchii (Fig. 5a, b).Seven days after watering, the differences between the two species were only apparent in the morning, while afternoon ET values were similar (Fig. 5c).
The diurnal courses in Fig. 5 indicate the occurrence of different relationships between ET i or E b and VPD for observations before and after midday.Therefore, we estimated the strength of the linear relationship [coefficient of correlation (R 2 )] between

Conclusions References
Tables Figures

Back Close
Full both metrics and their slope for observations in the morning and afternoon (Fig. 6).
The distinction between morning and afternoon values was made based on the maximum value of ET i or E b for each day, which occurred between 13:00 and 14:00 h.In general, a strong positive linear relationship was identified between ET i or E b and VPD with values of R 2 ranging between 0.87 and 0.76.For all measurements the relationship was stronger in the morning than in the afternoon.The slopes of the regression lines were most elevated for Senna artemisioides (Fig. 6b) with values of 5.2 × 10 −5 mm s −1 kPa −1 and 7.5 × 10 −5 mm s −1 kPa −1 , in the morning and afternoon, respectively.For Sclerolaena birchii the slopes were markedly lower (Fig. 6c) with values of 2.5 × 10 −5 mm s −1 kPa −1 and 3.4 × 10 −5 mm s −1 kPa −1 .The steeper slopes of the regressions for both species in the afternoon are associated with the cessation of ET at much greater VPD values near sunset than in the morning.The lowest values (9.3 × 10 −6 mm s −1 kPa −1 and 8.5 × 10 −6 mm s −1 kPa −1 ) were derived for the slope of the regression line between E b and VPD (Fig. 6c).
As both plant species show differences in their plant specific ET, we investigated whether the species composition, given the observed vegetation coverage of 26 %, is influencing ET on a plot scale.In Fig. 7 we plotted estimated time series of ET plot for three scenarios of species composition, as well as pET.For all three scenarios, ET plot was markedly below pET.Moreover, no pronounced difference was found for ET plot between the three scenarios.
In order to investigate the influence of vegetative coverage (FPC v ) on the results derived in Fig. 7, we compared the projections of ET plot given the ET i and E b values observed five days after watering (Fig. 4) for the same scenarios of species composition in relation to FPC v .For this theoretical projection it is assumed that ET would not be affected by a possible depletion of soil water.The results in Fig. 8 indicate that below vegetation coverage of 50 % the influence of species composition on ET plot was relatively small, whereas for values of FPC v above 80 % species composition critically influenced ET plot .In the scenario where Senna artemisioides was the dominant species and vegetation cover was 100 %, pET could be used as a predictor of ET plot .

Conclusions References
Tables Figures

Back Close
Full

Ecohydrology of vegetation on ET cover systems
For water-limited ecosystems, evapotranspirational losses constitute the most critical ecohydrological variable (Hultine and Bush, 2011;Rodriguez-Iturbe et al., 2001).The results of this field study confirm that both evapotranspiration from individual plants ET i ) and evaporation from bare soil (E b ) are increased substantially under wet soil conditions, i.e. shortly after initial watering (Fig. 4) which simulated a rainfall event of 17 mm.However, the maximum diurnal rates of water loss into the atmosphere are markedly higher for both plant species than from bare soil (Fig. 5), resulting in overall higher daily ET i compared to E b (Fig. 4).This emphasises the critical role of transpiration when partitioning evapotranspiration into evaporation from bare soil and plant transpiration (Cavanaugh et al., 2011;Raz-Yaseef et al., 2012;Xu et al., 2011;Zhang et al., 2011).
Daily E b from the soil surface is already stable five days after watering, indicating that E b of a semi-arid environment quickly converges to its minimum (Paruelo et al., 1991;Wythers et al., 1999), when a minimum water content has been reached in a generally dry soil profile.The lack of a marked drop in ET i from Senna artemisioides before the seventh day after watering (Fig. 4) indicates limited but continuing plant-available water in the root zone (Denmead and Shaw, 1962;Williams and Albertson, 2004).Not only are daily patterns of ET i and E b influenced by soil water availability in combination with low soil hydraulic conductivity, but these factors may also be more important than atmospheric water demand (Liu et al., 2010;Pingintha et al., 2010;Stoy et al., 2006;van Heerwaarden et al., 2010;Wilske et al., 2010).For ecosystems with no soil water limitation evapotranspiration is mainly governed by the atmospheric water demand (Laio et al., 2001;Mackay et al., 2007;Rejskova et al., 2012;Takagi et al., 1998;Tang et al., 2006) and radiation (Mackay et al., 2007;Tang et al., 2006;Wolf et al., 2011).However, with decreasing soil water potential, i.e. drying of soil, evapotranspiration is more and more limited by plant-available soil water, associated with a declining unsaturated hydraulic conductivity of the soil (Seneviratne et al., 2010).The Introduction

Conclusions References
Tables Figures

Back Close
Full distinctively different relationships between ET i or E b and VPD for the morning and afternoon, as well as for the two plant species and bare soil (Fig. 6), indicate that ET i or E b are not solely determined through atmospheric demand, but also by soil water flow characteristics.The similar responses of both species to increasing VPD in the morning of all three measurement days (Fig. 5) suggests that rehydration occurred throughout the plant on all nights during the observation period.However, the steeper slope of the regression line describing ET i of Senna artemisioides (Fig. 6b) indicates faster water transport to the evaporation surfaces within leaves which could be due to a higher ratio of vascular conducting surface to leaf area surface, a closer connection between water storage and conducting tissues, or a higher leaf area conductance under optimum condition (Huber, 1956;Zimmermann and Brown, 1971).The distinctively different correlations between ET i and VPD for morning and afternoon demonstrate that in the afternoon, transpiration is limited by factors that retard water re-supply to the transpiring interface and potentially lead to stomatal closure.The main limiting factor in the afternoon is likely to be perirhizal conductance, which is a function of low soil water availability and low hydraulic conductivity.While the slopes of the regression lines for bare soil evaporation are markedly lower, the relationships of bare soil evaporation to VPD are generally similar to those of the two plant species.This implies a replenishment of the soil surface overnight akin to the rewetting of vegetation.As opposed to vegetation with roots potentially tapping into deeper and moister layers, topsoil rewetting has to be based on 1-dimensional upward flow of water, the hydraulic conductivity of the soil, water vapour movement and potential dew formation.Similar daily patterns of water loss from bare soil and the plant species may be due to condensation of water vapour, transported from moister layers at depth, within the dry surface soil layer at times of lowest soil temperatures.This moisture is subsequently available to VPD driven evaporation in the following morning (Yamanaka and Yonetani, 1999).Stronger limitations for E b compared to ET i arise from the fact that the soil surface dries to much lower water potentials than the deeper root zone of the vegetation.This leads to a strongly reduced unsaturated hydraulic conductivity which is further pronounced Introduction

Conclusions References
Tables Figures

Back Close
Full through conditions at the soil to atmosphere interface while the 3-dimensional water flow to roots and water depletion in the soil space around roots is less strongly affecting the actual hydraulic properties.
In contrast to Senna artemisioides, the pattern of daily ET i for Sclerolaena birchii was much more uniform (Fig. 4), indicating different water balance characteristics for the two plant species compared to evaporation from bare soil.Distinguishing between intensive and extensive water exploiters (Rodriguez-Iturbe et al., 2001), Senna artemisioides can be considered an intensive water exploiter, showing a strong increase of ET i after a rain pulse, while the extensive exploiter Sclerolaena birchii mainly relies on deeper soil water and only shows a slight, delayed, or no reaction to rain events (Burgess, 2006;Rodriguez-Iturbe et al., 2001).Root excavations examined for the two plant species after the field campaign confirmed the hypothesis of a deeper root zone for Sclerolaena birchii.This species explores the cover system to a depth of 1.6 m although most of the root biomass was located between 0.1 and 1.0 m depth.Most of the Senna artemisioides root biomass was found to be in the upper 0.1 m of soil while no roots occurred below a depth of 1.1 m.
Given the critical role plant transpiration imposes for ET of areas covered with vegetation, it is reasonable to question the relevance of this result for plot evapotranspiration under conditions of low vegetation coverage.Many studies have concluded that in arid and semi-arid ecosystems evaporation from bare soil is a much more important contributor to plot ET than plant transpiration (e.g.Lauenroth and Bradford, 2006;Paruelo et al., 1991;Stannard and Weltz, 2006).Indeed, the projections of three scenarios of species composition plotted in Fig. 7 suggest that for a vegetation cover percentage as low as 26 % (as estimated at the study site), vegetation composition is irrelevant for daily plot evapotranspiration.Furthermore, for all scenarios, the actual plot evapotranspiration remains markedly below the potential evapotranspiration, indicating that the latter is a rather weak predictor for plot evapotranspiration under conditions of low vegetation coverage.Introduction

Conclusions References
Tables Figures

Back Close
Full However, the opposite is the case as vegetation coverage approaches 100 % and species with high transpiration rates such as Senna artemisioides, dominate the plant community (Fig. 8).In general, vegetation composition plays a critical role in water loss from a cover system if vegetation coverage exceeds 50 %.Nevertheless, it should be noted that for the sustainable establishment of plant communities, vegetation density has to be in equilibrium with soil moisture availability (Specht and Specht, 1999;Josa et al., 2012).In semi-arid Australia, vegetation coverage on rehabilitated waste rock material is generally below that of natural sites and rarely exceeds 50 % (Morrison et al., 2005;Vickers et al., 2012).In addition, vegetation on rehabilitated areas shows a higher sensitivity to environmental fluctuations (Vickers et al., 2012) and may therefore be more susceptible to death during prolonged periods of drought.

Implications for evapotranspiration cover design
All the above findings have critical implications for the design of any ET cover system (be it waste rock covers in the mining industry or municipal landfill covers), particularly with regard to species selection and composition of plant communities in the context of climatic conditions.Keeping in mind the fundamental objective of ET cover systems, which is to minimise drainage by, for example, maximising evapotranspiration (Salt et al., 2011), it becomes apparent that site-specific conditions, such as the climate regime, are crucial.Both plant species considered here seem to be well adapted to the climatic conditions of the site although they exhibit different water usage characteristics.The intensive water exploiter Senna artemisioides facilitates higher water loss through ET shortly after rainfall events and should therefore be more suitable for areas with rainfall occurring predominantly in frequent and low intensity events, eventually resulting in high soil infiltration rates.If the climate is characterised by prolonged drought periods and if rainfall events occur as high intensity storms, a more extensive water user with deeper roots and hence extended long-term water use may be preferable, such as Sclerolaena birchii.It is clear that, under the current climate of western New South Wales, both intensive and extensive water users can survive.However, Introduction

Conclusions References
Tables Figures

Back Close
Full the initial re-establishment of those ecosystems might be challenging since intensive water users with elevated initial biomass production may out-compete extensive water users if both species show similar drought adaptations (Smesrud et al., 2012;Zea-Cabrera et al., 2006).Nevertheless, the coexistence of species with those contrasting water use strategies is favoured by a spatial segregation of soil moisture (Zea-Cabrera et al., 2006), which is likely to occur on reconstructed ecosystems due to the high likelihood of spatial heterogeneity of soil physical properties (Gwenzi et al., 2011;Kr ümmelbein et al., 2010;Mazur et al., 2011;Schneider et al., 2010).Regardless of the water use characteristics of individual species, the plant community tends to converge towards a stable equilibrium with long-term availability of soil water (Specht and Specht, 1999).It is therefore questionable if water-limited ecosystems in arid and semi-arid regions are able to sustain vegetation coverage of above 50 %.This value, however, is critical to employ plant transpiration as the main driver of water loss from soil rather than evaporation (Fig. 8).Therefore, the application and management of ET cover systems in water-limited environments remains a challenging task for both industry and science.
Analysing the soil-vegetation-atmosphere continuum in the context of ET cover systems denotes an important future research task (Arora, 2002), which requires (a) further on-site measurements with regard to vertical soil water dynamics, (b) controlled manipulative experiments with regard to ET characteristics of plants under limited soil water conditions, and (c) modelling frameworks aiming to establish an optimal soil-vegetation design with regard to soil texture and thickness, and climatically welladapted plant communities and their species composition.Built on those premises, this study denotes the starting point of further investigations to identify optimal ET cover designs that are robust and reliable under given climatic conditions.Introduction

Conclusions References
Tables Figures

Back Close
Full

Conclusions
This study has exemplified that a well-considered selection of plant species can make the difference between success and failure of operational ET cover systems.The dynamic interplay between climate, plant community and soil water denotes the crucial foundation of designing robust and reliable ET cover systems in the face of extreme weather events of semi-arid and arid areas such as prolonged drought periods and intense rainfall events.Built on this premise, the distinct ET characteristics of plant species can be utilised to design an optimal barrier that maximises ET and thereby minimises drainage into the underlying hazardous wastes.However, considering the minor role vegetation ET plays under conditions of low vegetation coverage, we stress the need for thorough evaluation of the trade-offs between total plant-available water in dryland ecosystems and the relatively large vegetation coverage that is required to make plant community composition a critical determinant of water loss through evapotranspiration.Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | [g m −3 ], A b is the ground area [m 2 ] (Fig. 2).Discussion Paper | Discussion Paper | Discussion Paper | FPC v and FPC b denote the projective cover of vegetation and bare soil [%], respectively, ω i denotes the fractional coverage within FPC v , n is the total number of individual plants, and E b is the daily evaporation rate of bare soil.To investigate the influence of each species on ET plot we considered three scenarios of species composition (ω Slc /ω Sen ): (1) ω Slc /ω Sen = 0.5/0.5, (2) ω Slc /ω Sen = 0.7/0.3, and (3) ω Slc /ω Sen = 0.3/0.7.Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |