The Australian cotton industry is a world leader in the production of quality cotton with high levels of water efficiency. Australian cotton yields have increased steadily over the past 20 years, and are now two and a halftimes the global average with higher levels of water use efficiency than other key production regions in the world. In recent years, the industry has proactiveIy improved the management of its water resources. It has invested substantial resources in research and extension programs to improve the cotton produced per unit of water input. However, the industry has never investigated water use using a supply chain assessment methodology such as life cycle assessment or water footprinting. These methods have grown in popularity and importance as tools to investigate and communicate the impacts of producing a product such as cotton throughout the supply chain. IdealIy, such analyses provide results for a retail product and the use of that product by consumers, Recently, such studies have been commissioned for iconic cotton product manufacturers such as Levi Strauss.

A review of supply chain water use assessment methods identified two main methodological approaches; water footprinting or WF (which supersedes the term 'virtual water' or VW) and life cycle assessment (LCA). A third method, water balancing, is not strictly a supply chain assessment tool, though it could be applied that way. However, because it is the most commonly used way to quantify water use for cotton production at the farm level it is reviewed in detail here also. Results from water balance studies (at the farm level, for example) can be readily used as inventory data for either a WF or LCA study.

Differences between the LCA and WF methods are numerous. Water footprinting/virtual water was developed as a method of assisting countries to reduce water use by importing products that require a high amount of irrigation water to grow locally. Hence, water stressed regions such as the Middle East could reduce production of wheat and livestock that was grown locally using irrigation, and import these products from other countries. This would amount to water savings forthe importing country. The key feature of the method was that it is based on the theoretical water requirement for production regardless of the source of that water. Initial WFNF studies reported, for example, that Australian wheat 'used' 1588 L I kg of wheat produced, without identifying the fraction of water sourced from 'soil stored

moisture derived from rainfall' (or green water) compared to the fraction sourced from irrigation. For a nation such as Australia that grows much of our cereal product from dryland areas, this approach was easily misinterpreted and equated to the volume of Irrigatibn water required. Later studies clarified that only 25% of the water used for Australian wheat was derived from irrigation (still thought to be an overestimate compared to other Australian research). Even for a predominantly irrigated crop such as cotton, inclusion of green water in the assessment would lead to a considerably higher level of reported water use. The WF method also includes 'grey water' or dilution water, which is the estimated volume of water required to assimilate pollutants released from a production system to ensure these are below threshold levels. Hence, releases of chemicals and nutrients would require an estimated volume of 'dilution' water to ensure these were below environmental and health thresholds, and this water 'use' would be attributed to the production system. This could also be quite significant for cotton production in some situations.

In the field of LCA there have been a number of advances aimed at accurately quantifying and interpreting water use data. Most of this work has been done in the pastthree years, and studies done prior (or even during) this time are quite variable in quality and rigour. LCA research is divided into two important stages relevant to this discussion; the inventory stage (data collection) and the impact assessment stage, when the inventory results are interrogated and interpreted. Different methods apply to each stage. At the inventory stage, an LCA study may use data collected for other purposes (i. e. irrigation water balance research or WF research), but at the impact assessment stage, these data are used to

provide insight into the impacts of using water on competitive users or the environment using a number of methods. It is this impact assessment method that is more advanced than a simple 'inventory' of water use (which could be done with a series of water balances) or a WF study, which doesn't extend beyond the inventory stage either. This helps address the problem that we can understand intuitively, that the impact of water use will differ greatly depending on where it is used and particularly, if it is being drawn from a depleted, oven allocated (or stressed) source. InterestingIy, while Australia may deem some catchments (notably the Murray Darling) as being over allocated or stressed, on a global scale the stress weighting is not compareble to severely stressed regions such as North Africa or India.

State-of-the-art LCA methods for water use specify the use of detailed inventory methods such as water balancing, specification of water quality inputs and outputs, and methods to quantify the impact of using water. However, WF methods provide no insight to these differences. Recent LCA methods have proposed ways to define water use in terms of the stress created on a catchment by using the water. 'Water stress has been defined globalIy in order to allow comparisons with different production regions of the world. These methods are operational and have been applied to Australian beef production (by CSIRO and the authors) and for pork production (by the authors). To date, no study of an irrigated crop has been made to the Author's knowledge.

The major differences between WF and LCA can be summarised as follows:

i) The inclusion I exclusion of green water. This is included in the WF method and

generally excluded from LCA methods at the impact assessment level.

ii) The inclusion I exclusion of grey water. This is included in the WF method and

excluded from LCA methods at the impact assessment level.

iii) Inclusion of methods to assess the impact of using water(on competitive users

and the environment) rather than simply the total volume used. This is excluded from the WF method and included in LCA methods.

On review and comparison of these methods, the authors felt LCA to be the most robust and useful method for conducting supply chain water use assessments in the Australian cotton industry. The reasons for this were:

i)

ii)

iii) iv)

v)

State-of-the-art LCA research specifies the use of a detailed water balance to identify flows of water at each stage in the supply chain. This is a robust approach for quantitying water use in cotton production. Data are readily available and results can be communicated easily with the industry and the consumer.

LCA has a robust methodology and framework for handling water 'uses such as green water and grey water. This may be done excluding these from the impact assessment and including additional impact assessment methods that deal directly with the issue of concern (such as contaminant release). This results in a more readily understandable and meaningful result.

Taking point ii) into account, LCA is able to include green and grey water use at the inventory level in order to provide a compareble result with a WF method if this is desired.

Impact assessment methods are available in LCA that can quantify not only the total water used, but also the impact of using this water on either the environment or on other competitive users, This is an important advance on the water footprint method.

LCA is able to incorporate additional impact assessment areas such as energy use and GHG emissions to provide a broader assessment.

Considering most of the advances in LCA water methodology have been made recently, the cotton industry is in a good position to provide a robust study based on well-grounded methods that can be used as a benchmark for future research in the cotton supply chain and

in the textiles industry more broadly. A number of recommendations are provided for future research in this area.