Agricultural Water Management 96 (2009) 15171524

Contents lists available at ScienceDirect

Agricultural Water Management

journal homepage: www.elsevier.com/locate/agwat

Increasing productivity in irrigated agriculture:

Agronomic constraints and hydrological realities
Chris Perry a,*, Pasquale Steduto b, Richard. G. Allen c, Charles M. Burt d
a

Consultant water resources economist, 17 Storey Court, St Johns Wood Road, London NW8 8QX, United Kingdom
Division of Land and Water, FAO, United Nations, viale delle Terme di Caracalla, 00153 Rome, Italy
c
Departments of Civil Engineering and Biological and Agricultural Engineering, University of Idaho, Kimberly, ID, USA
d
Irrigation Training and Research Center, BioResource and Agricultural Engineering Department, California Polytechnic State University, San Luis Obispo, CA, USA
b

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 16 December 2008
Accepted 30 May 2009
Available online 5 July 2009

Irrigation is widely criticised as a proigate and wasteful user of water, especially in watershort areas.
Improvements to irrigation management are proposed as a way of increasing agricultural production
and reducing the demand for water. The terminology for this debate is often awed, failing to clarify the
actual disposition of water used in irrigation into evaporation, transpiration, and return ows that may,
depending on local conditions, be recoverable. Once the various ows are properly identied, the
existing literature suggests that the scope for saving consumptive use of water through advanced
irrigation technologies is often limited. Further, the interactions between evaporation and transpiration,
and transpiration and crop yield are, once reasonable levels of agricultural practices are in place, largely
linearso that increases in yield are directly and linearly correlated with increases in the consumption of
water. Opportunities to improve the performance of irrigation systems undoubtedly exist, but are
increasingly difcult to achieve, and rarely of the magnitude suggested in popular debate.
2009 Elsevier B.V. All rights reserved.

Keywords:
Irrigation efciency
Evaporation and transpiration
Water use efciency
Irrigation technology
Productivity of water

1. Introduction
Competition for scarce water resources is already widely
evidentfrom the Murray Darling basin in Australia to rivers of the
middle East, southern Africa and the Americas, and from the
aquifers of northern India, to the Maghreb and the Ogallala in the
central US. Where water use is not regulated and controlled,
imbalances between demand and supply are evident in falling
water tables, drying estuaries, inadequate supplies to lower
riparians and damaged aquatic ecosystems.
Climate change is generally expected to worsen the situation.
While some areas will receive higher rainfall, most of the currently
water-scarce regions will become drier and warmer. These two
changes will exacerbate scarcity: reduced rainfall means less ow
in rivers; higher temperatures mean increased evaporation and
water consumption by natural vegetation (compounding the
reduction in runoff), and higher water demand for agricultural
use. So tensions between supply and demand are likely to be
aggravated.
Irrigation is the dominant user of water in virtually all the
situations where scarcity is signicant, and demand for irrigation is
likely to increase under the basic drier and warmer scenario

* Corresponding author.
E-mail address: [email protected] (C. Perry).
0378-3774/$ see front matter 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.agwat.2009.05.005

outlined above. Improved irrigation management is thus often

portrayed as the key issue in coping with current and future
scarcity.
Irrigation has long been described as a wasteful, low value
water use (Bhatia and Falkenmark, 1992). More recent examples
are not difcult to nd. The quotation below is from a guest column
by Lester Brown, founder of the Earth Policy Institute, in The Indian
Financial Express dated December 1, 2008. We quote here at some
length because the article brings together a number of the issues
that this paper will address:
Water policy analysts Sandra Postel and Amy Vickers found
that surface water irrigation efciency ranges between 25%
and 40% in India, Mexico, Pakistan, the Philippines and
Thailand; between 40% and 45% in Malaysia and Morocco;
and between 50% and 60% in Israel, Japan, and Taiwan.
Raising irrigation water efciency typically means
shifting from the less efcient ood or furrow system to
overhead sprinklers or drip irrigation, the gold standard of
irrigation efciency. Switching from ood or furrow to lowpressure sprinkler systems reduces water use by an
estimated 30%, while switching to drip irrigation typically
cuts water use in half. A drip system also raises yields because
it provides a steady supply of water with minimal losses
to evaporation.

1518

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

Any measures that raise crop yields on irrigated land also raise
the productivity of irrigation water.
Often, as will be discussed in detail, terminology obscures
rather than claries the discussionthus Cooley et al. (2008), in a
major report on water use in California, state:
In this analysis, we quantitatively estimate changes in water
withdrawals and provide qualitative estimates of changes in
consumptive use where possible. (p. 16)
However, the apparently vague qualitative estimates of
consumptive use are rmly translated into specic volumes
of water:
For example, the savings we nd in these scenarios can be
compared using dam equivalents. Assuming that a dam yields
174,000 acre-feet (215 million m3) [of new water, our
efciency scenarios save as much water as provided by 320
dams of this size. (p. 6)
Later, while detailing the most substantial anticipated saving
(through better irrigation scheduling) it is noted that:
Water-use efciency is dened here as yield divided by applied
water. (p. 27, emphasis added)
The purpose of this paper is to bring together key messages from
the existing, peer-reviewed science. On this basis, we will challenge
the fundamental assumptions embodied in the quotations set out
above. We will argue that local estimates of irrigation efciency
bear no necessary relationship to what is happening at the basin
scale, and are consequently often misleading; that advanced
irrigation techniques can sometimes result in increased water
consumption; and that improving yields does not necessarily mean
increased water productivity; and that these indicators should in
any case not be measured in terms of water applied. Most
importantly, we argue that impacts must be carefully assessed
though the application of those proven scientic principles of
hydrology, irrigation technology, energy balances, and plant
physiology that actually dene and constrain the options available.
The paper is in six parts. Following this introduction, the second
section sets out widely reviewed, unambiguous terminology for
describing water use in irrigation and other sectors. This will
already highlight some existing confusion, and also provide the
terminology for the subsequent sections. The third section deals
with evaporation and transpirationthe critical components of
water demand in situations of scarcity.
The fourth section turns to the crop production resulting from
the use of more or less water in irrigated agriculturehow yield
responds to water availability, how yield is affected by water
stress, and how climate (temperature, humidity, etc.) affect water
demand and crop production.
The fth section considers the engineering options that affect
how irrigation water is applied at project and farm level, and how
these affect water use and the demand for water.
Finally, the interactions between these factors are brought
together, and the conclusions summarized.
2. Water accounting
Water accounting has developed from two distinct perspectivesirrigation engineering and hydrology (which term we use to
include hydrogeology).
Hydrology is the study of the movement, distribution, and
quality of water throughout the Earth. Hydrology provides a

general physical framework for understanding the ow of water,

especially at large scales. The hydrological cycle comprises
evaporation from land and water surfaces and transpiration from
vegetation, which together move water into the atmosphere, and
precipitation onto land and sea. Hydrologists usually take a basin
perspective, analyzing precipitation, evaporation and transpiration, runoff to streams and rivers, recharge to aquifers and outows
to the sea.
The irrigation engineering perspective looks more narrowly at
interventions designed to utilize surface water ows or groundwater for application to crops where rainfall is inadequate to meet
crop demand. Obviously not all the water diverted at the source
reaches the eldthere is canal seepage, evaporation from wet
soil, and so onand these considerations led to the development
some 60 years ago of concepts of efciency (Israelsen, 1950) that
dened ratios between the quantity of irrigation water that meets
its intended purpose and the diverted quantity as efciencyfor
example conveyance efciency (the ratio of water volume
delivered to the elds to the water diverted into the canals) and
eld application efciency (the ratio of irrigation water stored in
the root zone to irrigation water applied to the eld).
Both the hydrological and the engineering approaches have
merit: the hydrological approach is a general framework for
understanding water ows, the engineering approach is essential
for design and operation of irrigation systems and municipal water
supply systems.
However, to understand and interpret the impacts of changes in
water management, analysis must conform to the law of
conservation of mass, where water is neither created nor
destroyed, but only stored or transferred spatially as liquid or
latently as evaporation. The use of the terminology and analytical
framework of irrigation engineering is often unhelpful at times
even misleading and perverse in the context of water scarcity at
scales beyond the irrigation project itself. Concern at this issue is
not novel: in 1979, the US Government completed a report,
Irrigation Water Use and Management (ITF, 1979). Regarding
irrigation efciency, they stated:
[A]ny report dealing with irrigation efciencies must rst
dene efciency with a great deal of care. Many different and
sometimes conicting denitions have been published. It is
frequently assumed that because irrigation efciency is
low, much irrigation water is wasted. This is not necessarily
so. (p. 22).
In other words, losses at the scale of an individual eld or an
irrigation project are not necessarily losses in the hydrological
sense because, according to the law of conservation of mass, the
lost water may be available for use at some other point in the
basin, or from an aquifer. A simple example of this is in a monsoon
climate, where excess water is available in one season (but cannot
otherwise be stored) and water is scarce in another season.
Irrigation losses that recharge the aquifer in the wet season will
improve ground-water availability in the dry season.
This is not to advocate excessive irrigation applications in
general: in arid and semi-arid climates, excess irrigation water can
cause waterlogging, and, when that water evaporates, leave the
soil salinized. In this case, water is wasted, and the productivity of
the land is reduced.
It is thus essential to evaluate every case carefully and to
follow the water to determine its fate and disposition. Often,
such careful analysis will not fully support simple assumptions
that using less water is the same as saving water. The
conclusion that improving irrigation efciency and reducing
losses might not result in increasing the availability of water for
other uses has not proved easy for non-specialists to grasp. In

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

every other context, if we improve efciency, we consume less:

energy-efcient boilers consume less gas or electricity; fuel
efcient cars consume less petrol, and so on. Utilizing the term
efciency in and of itself carries the implication of being
good. It is a value-laden term.
While the recognition that improved irrigation efciency must
be carefully interpreted is growing (for an excellent recent
example, see Ward et al., 2008), several researchers have
advocated the use of fractions terminology to describe the
hydrologic outcome of water use (Willardson et al., 1994; Molden
and Sakthivadivel, 1999; Molden et al., 2001; Jensen, 2007;
Clemmens et al., 2008). Following a review of these and many
other papers and reports, and consultation throughout the
National Committees of the International Commission on Irrigation and Drainage, Perry (2007) set out such terminology, stressing
that it is suitable for unambiguous application to all types of water
use, including hydro-power, domestic, industrial, environmental
and irrigation. The terms are:
Water use: any deliberate application of water to a specied
purpose. The term does not distinguish between uses that
remove the water from further use (evaporation, transpiration,
ows to saline sinks) and uses that have little quantitative
impact on water availability (navigation, hydropower, most
domestic uses). Water use goes to the:
1. Consumed fraction (evaporation and transpiration) comprising:
a Benecial consumption: Water evaporated or transpired for the
intended purposefor example evaporation from a cooling
tower, transpiration from an irrigated crop.
b Non-benecial consumption: Water evaporated or transpired
for purposes other than the intended usefor example
evaporation from water surfaces, unwanted riparian vegetation, wet soil.
2. Non-consumed fraction, comprising the:
a Recoverable fraction: water that can be captured and reused
for example, ows to drains that return to the river system and
percolation from irrigated elds to aquifers; return ows from
sewage systems.1
b Non-recoverable fraction: water that is lost to further usefor
example, ows to saline groundwater sinks, deep aquifers that
are not economically exploitable, or ows to the sea.
The law of conservation of mass dictates that the consumed
fraction and non-consumed fraction sum to 1.0. Usage of this set of
terms in describing water use and impacts of water management
change has a number of benets:
 It is entirely consistent with the science of hydrologyand the
critical importance of hydrological context is obvious: canal
seepage over a utilizable aquifer becomes recharge and is
eventually recoverable; drainage ows into an estuary near the
sea are most probably non-recoverable.
 The terminology is consistent across sectors, allowing unambiguous discussion of the hydrological impacts of alternative
technologies and interventions.
 The terminology can be applied at any scale without modication.
 The terminology is value-neutral, i.e., it is not value laden by
the word efciencybenecial use is whatever society denes
it to be; riparian vegetation may have no value in some areas,
while constituting a valuable wetland in another.
1
The ICID paper setting out these terms addresses the issue of pollution at some
length, and recommends, because the signicance of any particular type or
concentration of pollution depends entirely on the proposed use, that pollution
should be a separate analysis.

1519

 Attention is directed clearly to what we want (benecial

consumption), what we do not want (non-benecial consumption and non-recoverable ows), and what is of secondary interest
(recoverable ows).
It may seem trivial to rename these various ow components
that occur during water resources management. However, the
following examples are presented to illuminate the signicance of
the problem:
 Efcient domestic supply systems involve virtually no consumptive use; outows are captured, treated and returned to the
water resources system. Efcient irrigation systems result in
substantial consumptive use, as 85% or more of the diversions go
to crop evapotranspiration. When specialists in these two sectors
promote improved efciency in papers, conferences and workshops, what they are advocating will have quite different
indeed often opposite hydrological impacts with regard to
downstream availability of water.2
 The latest Intergovernmental Panel on Climate Change report on
Climate Change and Water (Bates et al., 2008) mentions waterhungry appliances such as dishwashers, washing machines, lawn
sprinklers, etc.failing to distinguish between non-consumptive uses (dishwashers and washing machines, where virtually
all the water used can be recovered, treated if necessary, and
reused) and consumptive uses (lawn sprinklers), and to understand the fundamental signicance of that distinctionlet alone
to realise, as will be described later, that a really well designed
lawn sprinkler will result in higher water consumption.
 In Yemen, a World Bank project (World Bank, 2003) has claimed
that millions of cubic meters of water can be saved through the
introduction of improved irrigation technology. The claim is based
entirely on reductions in water use, while no measurements were
taken of water consumed with the earlier or the upgraded
technology. The appraisal report indicated that the areas being
upgraded are underlain by a shallow aquifer, so the question of
how much of the non-consumed fraction is recoverable and is
already being re-extracted and put to additional use is of
fundamental importance to assessing the projects impact.
 The Pacic Institute (2007) quotes Egypts annual renewable
water resource as 86.8 km3a surprisingly high gure, given that
Egypts agreed share of the Nile is 55.5 km3, and rainfall is
negligible. Meanwhile, Earthtrends (2007) reports a gure of
58 km3, with Internal renewable resources adding an additional
1 cubic kilometer. Both sources refer to the FAOs AQUASTAT as
a source of information. The source of confusion, of course, is that
the higher gure is a measure of water use, and much of that use is
in the form of re-use of the recoverable fraction.
Use of the terms proposed by ICID adds clarity to the analysis of
water resources management: it becomes clear, for example that
minimizing return ows is progressively more important as the
use nears the outow to the sea. Return ows in tidal areas such as
the Thames in London, or in narrow coastal strips exploiting runoff
from inland hills, often cannot be recovered for further use. In such
cases, minimizing return ows (through the use of advanced
irrigation technology, or low ow showers and toilets) will
typically save water for productive use elsewhere.
One nal addition is needed to complete the terminology for
this paper, and again it is designed to clarify an area of confusion.
Conventionally, Water Use Efciency is dened as a productivity
termoutput of crop per unit of water (Jones, 2004). However, the
term is so widely misused and confused with the old term of
2
Both types of use reduce streamow between offtake points and the point
where return ows arrive, with possible negative impacts on sh, navigation, etc.

1520

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

irrigation efciency (the proportion of water used that is

consumed by the crop) as to have become meaningless. Here
we use the term water productivity, dened as output of crop per
unit of water consumed.
3. Transpiration and evaporationdenitions, interactions
and typical relationships
Transpiration is the ow of water vapor from stomates of leaves
that causes replacement liquid water to move from soil to roots,
through stems and on to leaves. Water vapor exits through the
same stomates that carbon dioxide enters. The water vapor lost by
transpiration in exchange for carbon dioxide is the primary process
for plant growth and development. Other nutrients are delivered
from the soil by the water used in transpiration.
Evaporation is the direct conversion of water into water vapor
when wet leaves or soil are exposed to drier air and radiant heat.
Evapotranspiration (ET) is dened as the sum of direct
evaporation (E) and transpiration (T) of soil water through plant
systems and into the atmosphere.
Evaporation and transpiration are the components of the
hydrologic cycle where liquid water does disappear from the local
hydrological system in the form of vapor to return via precipitation
at some other location, and some other time. Thus, and because of
its generally large magnitude, ET is an important part of the
hydrologic cycle, and of water balances. Understanding, evaluating
and inuencing ET are critical elements of basin-scale water
resources management.
This process of converting liquid water to vapor consumes large
amounts of energy (approximately 2.5 megaJoules per kg). The
environmental energy available at any location at any specic time
determines the maximum rate of evaporation, and is dened by
available radiation energy from the sun and to lesser degrees by
heat from the air, moderated by wind speed and air humidity.
These four parameters (solar radiation, air temperature, air
humidity and wind speed) are the four weather parameters
employed by many ET equations.
The concept of a nearly xed energy requirement to evaporate a
unit of water is the foundation of all modern ET calculation
methods, for example the Penman-Monteith equation used by FAO
to estimate ET from weather data (Allen et al., 1998). The FAO
reference ET approximates a near upper limit for ET from an
extensive surface of fully watered vegetation with specied
characteristics at a specic location. This reference ET, adjusted
for crop characteristics, denes potential ETthe maximum rate of
water evapotranspiration for that location and crop.
Reference evaporation rates tend to be greatest in areas of the
world having the greatest hydrologic water scarcity due to the
negative feedback between general water scarcity and climatic
aridity; if less water is available, less actual ET takes place,
humidity is decreased, and evaporative demand and reference ET
increase. Conversely, if more water is available, more ET takes
place, humidity is increased (i.e., aridity is reduced) and
evaporative demand decreases.
Actual ET (as distinct from potential ET) is the managed
outcome of irrigation of agricultural crops and an unmanaged
outcome of rainfall on rainfed crops and natural vegetation. If
the water supply is fully adequate to meet the demand implied by
the FAO reference value, then actual and potential ET will be the
same.
The objective of irrigation is to make sufcient soil water
available to meet transpiration requirements that are dictated by
the local climate, the amount of plant cover, and the stage of
growth. In general, it is the fulllment of potential rates for T that
govern crop yield, with E as an incidental and often unavoidable
associate. Under water decit conditions, crops respond by rst

reducing canopy growth and then closing stomata, which in turn

reduces T, carbon assimilation, biomass production and harvestable yield below potential.
For a fully watered crop there is a maximum potential value for
ET dened by environmental energy. Thus any increase in E must
result in an offsetting decrease in T, and reduced production per
unit of consumed water.
Strategies to obtain shifts from E to T include organic and plastic
surface mulching, reduction in frequency and spatial extent of
surface wetting by irrigation, reduced tillage, and higher plant
densities.
4. Engineering options: how irrigation water is applied at
project and farm level, and how this affects water use,
consumption, and the demand for water
There is no simple answer to the question what irrigation
method is best?. The Irrigation and Drainage Division of the
American Society of Civil Engineers (Burt et al., 1999) provides a
reference that discusses the complexities.
Irrigation technology is often a farm-level choice, and it is
appropriate to consider the farmers perspective carefully in
understanding options and impacts. Thereafter, as recommended
in Section 3, we will wish to follow the water to understand the
wider water implications.
Farmers invest in improved irrigation technology for a variety
of reasons, including:
 Increased income: if yield tonnage, quality, or alternative, high
value crops will more than adequately pay for investment, there
is an incentive to improve.
 Risk aversion/food security: Farmers may shift from rainfed
agriculture to irrigation to reduce the uncertainties associated
with variable rainfall patterns. Similarly, farmers may shift from
public, surface-delivery systems to well water because the
surface water is delivered in an inexible and unreliable manner.
 Convenience: This is primarily seen in commercial farming. As an
example, a farmer may not want to have to wake up in the middle
of the night to receive project water deliveries, or he may be able
to deliver fertilizers more precisely and cheaply through
fertigation systems.
 Reduced costs: A farmer may save pumping costs if delivery
losses are reduced; he may save labor by installing equipment
that does not require constant eld presence.
Not all of these motives are water-related saving labor,
growing higher value crops, reducing uncertainty and these
factors, together with cost, credit availability, extension advice,
technical support and various other factors will inuence a
farmers selection of technology. But T is what drives biomass
production, and biomass production (subject to important
qualications addressed in the next section) determines how
much produce the farmer will have to sell. From the farmers
perspective, in the terminology of Section 2 maximizing benecial
consumption (Tact) is a legitimate and high priority objective.
Similarly, the farmer will generally wish to minimize all other
ows (non-benecial consumption, and ows of non-consumed
water to drains and groundwater). From the farmers perspective
such ows are lost.
For example, the simplest single intervention to improve
production is often improved land leveling. Ensuring that crops are
not over-irrigated in low areas (resulting in percolation) or underirrigated in high areas (resulting in below-potential yield) is a clear
benet to the farmer because it increases benecial consumption
of water and reduces ows that produce no benet to his
operations.

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

1521

increased T is at the expense of what were previously return ows,

then it is critical to understand whether these were previously
benecially consumed elsewhere. Third, regarding improvements
to the distribution system, the higher proportion of diverted water
that arrives at the farm gate will sometimes allow the farmer to
extend his irrigated area, which almost certainly will represent an
increase in overall consumption of water.
5. Crop productionhow yield responds to water availability;
how yield is affected by water stress; and how climate affects
water demand and crop production
Fig. 1. Comparison of yield vs. irrigation water applied for furrow and drip irrigation
of processing tomatoes in California (Burt and ONeill, 2007).

There are decades of experience with almost every irrigation

technology that can be imagined. Nevertheless, surprises are the
norm rather than the exception when rst applying those
technologies in a new setting. Often apparently promising
technologies are simply not suitable. Although it is desirable that
eld irrigation improve for many reasons (environmental, crop
quality, crop yield, energy consumption, etc.) it is difcult to nd
the most appropriate combination of sustainable modernization
of eld irrigation hardware and management for a location.
Fig. 1 illustrates the complexity involved with making simple
predictions of water saving (on the eld level) and yield increases.
The gure shows results from a large scale study by the Irrigation
Training and Research Center (ITRC) of commercial results in a
modern environment with excellent ow measurement, water
delivery exibility, and availability of technical support. The study
examined water applied and yields for processing tomato on 187
furrow-irrigated elds and 164 drip irrigated eldswith typical
eld sizes of 50 ha. It is clear from comparing yields and applied
water depths in the gure that it would be risky to assume that
there are immediate major benets across-the-board when drip is
implemented.
Note too, that the measure of water use here is water applied (as
preferred by Cooley et al., 2008), not water consumed: the former
term usually shows improved irrigation technology in the best
possible light.
In a large scale study for the California Dept. of Water Resources
(Burt et al., 2002; Mutziger et al., 2005), it was determined that in
general with good management and equipment, the transpiration (T)
component of evapotranspiration (ET) increases somewhat under
drip irrigation. This is due to decreased stress, and some transfer of
environmental energy not consumed by E into T and indicates that,
while E has an indirect impact in reducing transpirative demand, the
better approach is to reduce E and maximize T.
However, whether the evaporation (E) component increases or
decreases is highly dependent upon the irrigation method,
management and specic crop. For example, with subsurface drip
irrigation (a complex drip technology) the evaporation is minimized. But for many tree crops, irrigation with conventional drip
irrigation increases the E component because part of the soil
surface is wetted fairly frequently. The specic differences depend
upon the relative frequencies of irrigation and the percentage of
soil that is wet.
There are many valid reasons to improve irrigation technology.
From the farmers perspective the potential to increase the
proportion of water available at the gate that ends up as crop
transpiration (T) is the most likely source of increased revenues,
and savings in labor are also potentially signicant. Several issues
must be analyzed in each situation to evaluate the wider impact of
increasing T.
First, to the extent this represents a decrease in non-productive
E, this is a clear gain to the systembut savings in E are not
guaranteed, and must be critically evaluated. Second, if the

Of the water used in crop production, the benecial consumption is represented by that fraction of water use that is transpired
by the crop itself. The water consumed by transpiration is in
exchange for the assimilation of carbon dioxide (through plant
stomata) leading to the production of biomass (that is, the total
volume of vegetative matter produced) of which usually only a part
is harvested as yield (that is, the grain, fruit or vegetable that the
farmers can harvest for sale or consumption).
Often, as already described, when analyzing crop production no
distinction is made between benecial (transpiration) and nonbenecial (evaporation from soils and other surfaces) consumption
of water due to practical difculties to estimate or measure them
separately. Therefore, often the water consumption of the crop is
reported as evapotranspiration (ET).
To better understand how different crops, different agroclimatic environments, and different management practices may
inuence the relationship between crop production and water
consumption, we dene the water productivity of a crop (WP) as the
ratio between the amount of crop produced and the amount of
water consumed to obtain such production. Moreover, in dening
water productivity, we need to be specic in indicating which
product (biomass or yield) and which consumption (transpiration or
evapotranspiration) we are referring to. Thus, we can express the
water productivity of a crop (WP) as:





Biomass WP(T) = (kg of biomass)/(m3 of water transpired).

Biomass WP(ET) = (kg of biomass)/(m3 of water evapotranspired).
Yield WP(T) = (kg of usable yield)/(m3 of water transpired).
Yield WP(ET) = (kg of usable yield)/(m3 of water evapotranspired).

Over time, plants have evolved optimal stomatal behavior

(Cowan, 1982) in regulating the exchange of water and carbon
dioxide. As a consequence a strong, relatively constant biomass
water productivity is found when considered in terms of
transpiration [Biomass WP(T)], occurring under the same evaporative demand of the atmosphere, the same air carbon dioxide
concentration, and the same nutritional status (Steduto et al.,
2007). The essentially constant (or linear) relationship between
biomass and transpiration has been observed experimentally
since the early 1900s and has wide agreement in the scientic
literature (Tanner and Sinclair, 1983; Howell, 1990). It is
important to note from the above points that while the water
productivity of a plant is essentially linear for a given situation,
the slope of that relationship varies from location to location and
season to season.
We now consider what options are really available to improve
the biomass water productivity of a eld crop. As already noted,
reducing the non-benecial water consumption (evaporation from
soil and foliage) can lead to water savings. In these circumstances,
production can be maintained as long as T is maintained. To the
extent that evaporation is reduced, ET will be reduced and the
overall consumptive (ET) crop water productivity is increased. This
is valid for both biomass and yield water productivity.

1522

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

It is also evident that climatic conditions with reduced

evaporative demand (ETpot) will reduce both benecial (T) and
non-benecial (E) consumption of water. A reduced evaporative
demand from the atmosphere is usually associated with low wind,
low solar radiation, lower temperature and low vapor pressure
decit in the atmosphere (i.e., high relative humidity). When the
reduced evaporative demand is associated predominantly with
low radiation and temperature, it is important to account for
possible impacts on the biology of the crop. Crops like maize and
cotton, for example, require warm climatic conditions to grow
vigorously and productively. In addition, minimum thresholds of
solar radiation are needed for photosynthesis.
Options available to grow crops with reduced evaporative
demand are offered when it is possible to shift crop seasons, e.g.,
moving spring crops into winter periods. This management
practice has been shown to be successful in many cases (see
Soriano et al., 2004 for sunower). Another option is offered by
controlling the climate variables as it is done when using
greenhouses. Evaporative demand in greenhouses is usually of
the order of 60% of the evaporative demand in open elds.
Similarly, higher concentration of carbon dioxide inside greenhouses can increase assimilation of certain crops. The overall result
is a signicant increase in water productivity. Improved crop
varieties that are resistant to cold can also offer an option to grow
them in climates with lower evaporative demand.
Apart from the cases indicated above, the biomass water
productivity in terms of benecial consumption (T) is extremely
stable and difcult to improve for a given evaporative demand and
for a given carbon dioxide concentration. So far, the variability of
biomass water productivity observed between different varieties
of the same species is low (around 10% when normalized for
climate) and selection of high water productivity cultivars (such as
wheat) has been of practical application in very arid dryland
systems. Variation in biomass WP(T), for a given climate and nonlimiting nutrient conditions, is observed mainly between different

large crop categories species (due to difference in their photosynthetic pathways) and between cereals, legumes, ber crops and
oil crops where a signicant difference in biomass composition
exists.
Upper limits to productivity, under non-limiting nutrient
conditions, can only be improved through genetics.
Genetic engineering that reduces susceptibility to pests and
diseases has a clearly benecial impact on water productivity,
because crop loss after signicant consumption of water can be much
reduced. Similarly, drought tolerance (that is, capacity of the plant to
survive severe water stress and continue to maturity) also increases
production from scarce and unreliable rainfall. However, there is
very limited evidence to date of improvements in the fundamental
relationship between transpiration and biomass, WP(T).
Fig. 2 summarizes the relationship between biomass and water
consumption (E or ET). Changes to biomass water productivity
(Biomass WP) are indicated by changes in the slope of the
relationship. All cases presented assume no pest or disease
occurrence. In each case, the initial Biomass WP is the same,
represented by line A.
Fig. 2(a) shows three specic cases:
 The basic conceptual relationship for a given crop under nonlimiting nutrient conditions (A) in a specic climate. For a given
transpiration (T1) a corresponding biomass production (M1) is
obtained; the relationship is linear, and intersects the origin.
 If nutrient status of the crop is poor, the slope of the relationship
is lower (A0 ) producing less biomass (M2) for the same amount of
water transpired (T1).
 Genetically improved varieties, or crops with more efcient
photosynthetic processes, may show a higher slope (B) producing more biomass (M3) for the same transpiration (T1).
Fig. 2(b) represents typical eld conditions in which the total
water consumption (ET1) is split into its benecial (T1) and non-

Fig. 2. Biomass vs. water consumption relationships.

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

benecial (E1) components. If mulching is introduced (to reduce

evaporation from the soils) then E1 can instead contribute to crop
transpiration and for the same amount of total water consumption
(up to T = ET1) a higher biomass production is obtained (up to M4).
The same effect is produced when a crop grows vigorously in the
early stages, and the leaf canopy reduces the exposure of the soil to
sunlight.
Fig. 2(c) shows the case of a crop having a deeper root growth
that captures more water from the soil proleeither by reducing
deep percolation or utilising residual moisture from antecendent
rainfall or irrigation. The Biomass WP is not changed but a higher
transpiration (T2) leads to a higher biomass production (M5), i.e., a
higher production commensurate with a higher water consumption.
Fig. 2(d) shows the case of a crop that can be grown in areas,
or seasons, having (i) higher atmospheric evaporative demand
(A00 ), or (ii) lower atmospheric evaporative demand (A000 ). In
these cases, the biomass water productivity is lowered or
increased due to the climatic conditions, and not to the crop
characteristics and status. When the differences in climate are
accounted for (e.g., climatic normalization to the reference
evapotranspiration ETo, as illustrated in Steduto et al., 2007)
slopes will then represent the Biomass WP specic to the crop
and its nutritional status.
The key conclusion that can be drawn from the illustration of
the cases in Fig. 2 is that the increase in productivity per unit area,
or a water saving for the same productivity, can be derived by:
(i) increasing water productivity (i.e., higher slope of the biomass
vs. transpiration relationship);
(ii) diverting the non-benecial consumption (E) into benecial
consumption (T);
(iii) capturing more water available in the soil.
To increase the Biomass WP, options include selection of crop
varieties that are well adapted to the local environmental
conditions; maintaining a non-limiting nutritional status of the
crop; and, to the extent possible, to grow the crop in seasons and/or
areas with lower atmospheric evaporative demand.
To divert E into crop T, some of the options are to use mulching
opportunities, weed control, early canopy cover of the ground (e.g.,
early vigor type of varieties). To capture more water from the soil,
options include the use of varieties with deeper rooting system and
soil moisture control practices.
Crops that withstand disease and pest attacks effectively
increase the productivity of water by reducing crop failure and
related unproductive water consumption up to the point of failure.
Similarly, crops that tolerate drought, and recover to produce some
yield also increase production per unit of water consumed. But
changing the fundamental relationships between transpiration
and biomass production have not yet shown signicant progress.
When moving from biomass to yield water productivity, some
additional considerations are needed to understand the options
available for improvement. For those crops where the harvestable
part of the biomass is represented by grains, fruits, or tubers, the
variability in yield water productivity is quite high, even if
referenced to transpiration only [Yield WP(T)], because of the
variability in harvest index (HI), i.e., the proportion of biomass
allocated to the reproductive organs, which varies among species.
Increasing the HI was a major strategy of the Green Revolution,
when grain varieties with short stems and far more biomass
concentrated in the grain were bred. However, the free ride
provided by increasing HI may be ending. Since about 1980, only
minor increases in the harvest index have been achieved. . . it
appears unlikely that further major yield increases in cereals can
result from further major increases in HI. (Sinclair and Gardner,
1998.) This conclusion is conrmed in Zwart and Bastiaanssen

1523

(2004), whose data show that crop water productivity for wheat,
rice, and maize has not changed appreciably in twenty-ve years.
While water stress has very limited impact on biomass WP(T), it
may have a large impact on yield (e.g., by reducing the number of
reproductive organs). In simple terms, a crop subjected to water
stress at critical times may have a considerably lower harvestable
yield than a crop subjected to the same total stress (expressed as a
percentage of seasonal potential demand) but at less critical
periods.3 Other crops (e.g., processing tomatoes, wine grapes) must
be stressed at specic growth stages to obtain the required fruit
quality.
Regulated decit irrigation (RDI) consists of reducing water
deliveries when yield is relatively insensitive to stress, while
ensuring full deliveries at critical periods. While undoubtedly an
area for improved management and additional research, it is
important to note that the level of management required is high,
and that the risk of damaging the crop by accidental over-stressing
is signicant. Fereres and Soriano (2007) provide a comprehensive
overview of the potential of RDI for various crops and situations.
Two important points emerge from their review: rst, that
providing limited irrigation water forces the crop to make
maximum use of all sources of waterdeeper rooting structures
capture any residual moisture from antecedent rainfall or
irrigation events, and keeping the soil relatively dry ensures that
rainfall during the growing season is more likely to inltrate into
the root zone. Where such water would otherwise evaporate, this
is a production gain based on increased consumption of water, not
a shift in the WP(T) relationship (as set out above in Fig. 2(b)).
Second, to the extent that irrigation water can be restricted
while the crop yield is less sensitive to stress, and fully ensured
during periods of maximum sensitivity, minor increases (10% or
so) in WP(T) can be achieved but this requires very precise control
and timing of irrigation.
Although tree and fruit crops have a much more complex
production process than eld crops, the fundamental relationship
with consumptive water use are similar. Again, however, reference
to the literature (for example, Goldhamer et al., 2006) serves to
demonstrate that very high management levels are required to
achieve modest gains in water productivity.
6. Conclusions
The formulation of much of the popular criticism of irrigation is
simplistic and misleading. Clear terminology distinguishing
between consumptive and non-consumptive uses, benecial and
non-benecial consumption, and identifying the extent to which
water that is not consumed can be recovered for productive use
elsewhere is fundamental to rational analysis of the performance
of irrigation systems and the scope for improvement.
Water demand in an irrigated area depends on the climate, and
the crops being grown. Those areas of the world with most limited
water resources tend to be those with highest water demand.
Potential demand is determined by the energy balance between
the incoming radiation from the sun and the energy consumed in
converting liquid water into water vapor. The balance is achieved
through a combination of transpiration through the crop and
evaporation from wet surfaces. Transpiration is benecial in the
sense of contributing to the formation of biomass. It is therefore
desirable to maximize the transpiration component and minimize
evaporation.

3
Where water is abundant, the options for improving yield water productivity
are offered only through genetics. However, all crops have been improved for higher
harvest index through breeding and existing genetic opportunities. Further
increases seem to be limited mainly to tuber and root crops, while for crops like
cereals, the genetic opportunities may have been already exploited to the limit.

1524

C. Perry et al. / Agricultural Water Management 96 (2009) 15171524

Improved irrigation management and appropriate technology

offer scope to achieve this, though both are required for success.
Poorly managed hi-tech systems can be as wasteful and
unproductive as poorly managed traditional systems. Furthermore, careful analysis, in the terminology proposed, will often
reveal that nominal savings in terms of water applied
overstate actual reductions in consumption. Indeed, better
irrigation tends to increase transpiration and studies indicate
that evaporation in reasonably well managed systems is
generally rather low, so that offsetting savings from this source
are often insignicant. This is not to deny the many reasons that
farmers opt for and benet from improved irrigation systems
increased income from higher value crops, convenience, labor
saving, lower pumping costs, etc. Real water savings are
possible, but generally limited.
In assessing how productively water is used it is necessary to
distinguish between biomass (the total production of vegetative
matter) and yield (the production of grain, fruit, or tuber). The
relationship between biomass and transpiration is essentially
linear for a given crop and climateprovided nutrients are
adequate. Increasing the biomass productivity of water can be
achieved through improving nutrient status, growing the crop
during a cooler, more humid season, or through genetic improvements.
Yield, on the other hand, can be inuenced by ensuring that
water is not restricted during critical growth stages, while
curtailing supply at other times. Again, careful accounting must
be applied, but productivity gains much in excess of 10% seem
unlikely, and the level of management required is high.
Plant breeding has delivered substantial productivity gains over
recent decadesincreasing the proportion of biomass going to
grain, reducing the growing period, and breeding crops that
tolerate lower temperatures. Genetic modication offers further
promises certainly in terms of reducing the impact of diseases
and pests, possibly in terms of increased tolerance of drought but
to date the fundamental relationship between transpiration and
biomass has not been changed.
Researchers will continue to identify new opportunities and
approaches to improve the productivity of scarce resources, and
the purpose of this paper is certainly not to deny the scope for such
progressbut it is also appropriate to benet from the substantial
body of existing peer-reviewed knowledge. The opportunities are
often to make incremental gains through a variety of interventions,
and identifying and implementing these requires careful analysis,
supportive conditions, and careful monitoring for unintended
consequences.
The interactions identied in this paper suggest that where
agricultural management is moderately good and water is
scarce, interventions in water management sometimes
approaches a zero sum gamereduced losses in one place
reduce availability elsewhere, and increases in crop production
are accompanied by commensurate increases in water consumption.
Demonstrating that an intervention is not a zero sum game is
certainly an appropriate test to apply to any proposed intervention.
Acknowledgements
The authors acknowledge the valuable comments of two
anonymous reviewers.

References
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements. United Nations FAO, Irrigation
and Drainage Paper 56. 1998. Rome, Italy. 300 p. http://www.fao.org/docrep/
X0490E/X0490E00.htm (accessed Feb. 5, 2007).
Bates, B.C., Kundzewicz, Z.W., Wu, S., Palutikof, J.P. (Eds.), 2008. Climate Change
and Water. Technical Paper of the Intergovernmental Panel on Climate Change.
IPCC Secretariat, Geneva, p. 210.
Bhatia, R., Falkenmark, M., 1992. Water resource policies and the urban poor:
innovative approaches and policy imperatives. Background paper for the Working Group on Water and Sustainable Development. International Conference on
Water and the Environment, Dublin.
Burt, C.M., ONeill, B.P., 2007. Drip and furrow on processing tomatoeld performance. In: 28th Annual Irrigation Assoc. Technical Conference, San Diego, CA,
Dec. 9.
Burt, C.M., Mutziger, A., Howes, D.J., Solomon, K.H., 2002. Evaporation from Irrigated
Agricultural Land in California. ITRC Report R 02-001.
Burt, C.M., Clemmens, A.J., Bliesner, R.D., Merriam, J., Hardy, L.A., 1999. Selection of
Irrigation Methods for Agriculture. On-Farm Irrigation Committee. Water
Resources Division. ASCE. N.Y., N.Y. ISBN 0-7844-0462-3. 129 pp.
Clemmens, A.J., Allen, R.G., Burt, C.M., 2008. Technical concepts related to conservation of irrigation and rainwater in agricultural systems. Water Resources
Research, 44, W00E03, doi:10.1029/2007WR006095.
Cooley, H., Christian-Smith, Juliet, Gleick, Peter, H., 2008. More with Less: Agricultural Water Conservation and Efciency in California. Pacic Institute.
Cowan, I.R., 1982. Encyclopedia of Plant Physiology 12B, 589612.
Earthtrends, 2007. http://earthtrends.wri.org/searchable_db/index.php?step1/
4countries&cID%5B%
5D1/455&theme1/42&variable_ID1/4693&action1/
4select_years, accessed 1 May 2007.
Fereres, Elias, Soriano, Mara Auxiliadora, 2007. Decit irrigation for reducing agricultural water use. Journal of Experimental Botany. 58(2), 147159. http://jxb.oxfordjournals.org/cgi/content/full/58/2/147 viewed December 5, 2008.
Goldhamer, D.A., Viveros, M., Salinas, M., 2006. Regulated decit irrigation in
almonds: effects of variations in applied water and stress timing on yield
and yield components. Irrigation Science 24, 101114.
Howell, T.A., 1990. Grain dry matter yield relationships for winter wheat and grain
sorghumsouthern high plains. Agronomy Journal 82, 914918.
Indian Financial Express, 2008, http://www.nancialexpress.com/news/low-pricesof-water-and-drop-in-productivity/392609/0, viewed December 5, 2008.
Interagency Task Force, 1979. Irrigation Water Use and Management. US Govt.
Printing Ofce, Washington, DC, USA, 143 pp.
Israelsen, O.W., 1950. Irrigation Principles and Practices. John Wiley and Sons, Inc.,
New York, 471 pp.
Jensen, Marvin, E., 2007. Beyond irrigation efciency. Irrigation Science, OdI
10.1007/s00271-007-0060-5, 2007.
Jones, H., 2004. In: Bacon, M. (Ed.), Water Use Efciency in Plant Biology. Chapter 2
in Water Use Efciency in Plant Biology. Blackwell, Oxford, UK.
Molden, D., Sakthivadivel, R., 1999. Water accounting to assess use and productivity
of water. International Journal of Water Resources Development 15 (1), 5571.
Molden, D., Sakthivadivel, R., Keller, J., 2001. Hydronomic Zones for Developing
Basin Water Conservation Strategies. International Water Management Institute (IWMI) Research Report No. 56.
Mutziger, A.J., Burt, C.M., Howes, D.J., Allen, R.G., 2005. Comparison of measured and
modied FAO 56 modeled bare-soil evaporation. Journal of Irrigation and
Drainage Engineering 131 (1), 5972.
Pacic Institute, 2007. http://www.worldwater.org/data.html, accessed 1 May 2007.
Perry, Chris, 2007. Efcient irrigation; inefcient communication; awed
recommendations. Irrigation and Drainage, vol. 56. Wiley Interscience, pp.
367378.
Sinclair, T.R., Gardner, F.P., 1998. Environmental limits to plant production. Principles of ecology in plant production. CAB International.
Soriano, M, Auxiliadora, Orgaz, Francisco, Villaboso, Francisco, J., Fereres, Elias,
2004. Efciency of water use of early plantings of sunower. European Journal
of Agronomy 21:44, 465476, Elsevier Science.
Steduto, P., Hsiao, T.C., Fereres, E., 2007. On the conservative behaviour of biomass
water productivity. Irrigation Science 25, 189207.
Tanner, C.B., Sinclair, T.R., 1983. Efcient water use in crop production: research or
re-search? p. 128. In H.M. Taylor et al. Limitations to Efcient Water Use in
Crop Production. Am. Soc. Agron., Madison, WI.
Ward, Frank A., Pulido-Velazquez, Manuela, 2008. Water Conservation in irrigation
can increase water use, Proceedings of the National Academy of Sciences, vol.
105 no. 47, 1821518220, www.pnas.org_cgi_doi_10.1073_pnas.0805554105.
Willardson, L.S., Allen, R.G., Frederiksen, H., 1994. Eliminating irrigation efciencies.
In: USCID 13th Technical Conference, vol. 1922, Denver, Colo., October, p. 15.
Zwart, S.J., Bastiaanssen, W.G.M., 2004. Review of measured crop water productivity values for irrigated wheat, rice, cotton, and maize. Agricultural Water
Management.