Cover Image
close this bookPriorities for Water Resources Allocation (NRI)
View the document(introduction...)
View the documentAcknowledgments
View the documentIntroduction
close this folderPriorities and conflicts in water resource development
View the documentPaper 1 Demographic trends: implications for the use of water
View the documentPaper 2 Fortunately there are substitutes for water: othetwise our hydropolitical futures would be impossible
close this folderIssues in water resources management
View the documentPaper 3 Managing water resources versus managing water technology: prospects for institutional change
View the documentPaper 4 Water as an economic resource
close this folderDomestic water use
View the documentPaper 5 Domestic water use: engineering, effectiveness and sustainability
View the documentPaper 6 Domestic and community water management
close this folderUrban and industrial water use
View the documentPaper 7 Pollution alleviation issues: a case study on the River Ganges
View the documentPaper 8 Wastewater treatment and use for irrigation
close this folderWatershed management and land use
View the documentPaper 9 Institutional aspects of watershed management
View the documentPaper 10 The hydrological impact of land-use change (with special reference to afforestation and deforestation)
close this folderIrrigation
View the documentPaper 11 Small-scale irrigation in sub-Saharan Africa: a balanced view
View the documentPaper 12 Environmental and health aspects of irrigation
close this folderAquatic resources
View the documentPaper 13 Water management for aquaculture and fisheries; irrigation, irritation or integration?
View the documentPaper 14 Managing systems not uses: the challenges of waterborne interdependence and coastal dynamics
close this folderThe wider environment
View the documentPaper 15 World food production: the past, the present and the future
View the documentPaper 16 Climate change and the future of agriculture

Paper 16 Climate change and the future of agriculture

Martin Parry

Environmental Change Unit, University of Oxford

Summary: The purpose of this paper is to review our current knowledge of the potential effects of climate change on agriculture and world food supply. Attention is first directed to the types of climate change, particularly reductions in soil water availability, likely to be most critical for agriculture. Secondly, a number of types of effects on agriculture are considered: 'direct' effects of elevated CO2, shifts of thermal and moisture limits to cropping, effects on drought, heat stress and other extremes, effects on pests, weeds and diseases, and effects on soil fertility. Thirdly, a summary is presented of likely overall effects on crop and livestock production. The conclusion is that, while global levels of food production can probably be maintained in the face of climate change, the cost of this could be substantial. There could occur severe negative effects at the regional level. Increases in productive potential at higher latitudes are not likely to open up large new areas for production. The gains in productive potential here are unlikely to balance possible reductions in potential in some major grain-exporting regions at mid-latitudes.

Introduction

This review considers, firstly, those systems, sectors and regions of agriculture that are most sensitive to anticipated changes of climate that may result from emissions of greenhouse gases, primarily carbon dioxide. Secondly, it summarises present knowledge about the potential socioeconomic impact of changes of climate on world agriculture. Thirdly, it considers the adjustments in agriculture that are most likely to occur. Finally, it establishes research priorities for future assessments of impact.

The most significant aspects of climatic change

The potentially most important changes of climate for agriculture include: changes in climatic extremes; warming in the high latitudes; poleward advance of monsoon rainfall; and reduced soil water availability (particularly in mid-latitudes in midsummer, and at low latitudes).

Climatic extremes

It is not dear whether changes in the variability of temperature will occur as a result of climate change. However, even if variability remains unaltered, an increase in average temperatures would result in the increased frequency of temperatures above particular thresholds. Changes in the frequency and distribution of precipitation are less predictable, but the combination of elevated temperatures and drought or flood probably constitutes the greatest risk to agriculture in many regions from global climate change.

Warming in high latitudes

There is relatively strong agreement among GCM predictions that greenhouse-gas induced warming will be greater at higher latitudes (IPCC, 1990). This will reduce temperature constraints on high-latitude agriculture and increase the competition for land here (Parry and Duinker, 1990). Warming at low latitudes, although less pronounced, is also likely to have a significant impact on agriculture.

Poleward advance of monsoon rainfall

In a warmer world the inter-tropical convergence zones and polar frontal zones might advance further poleward as a result of an enhanced ocean-continent pressure gradient. If this were to occur then total rainfall could increase in some regions of monsoon Africa, monsoon Asia and Australia, though there is currently little agreement on which regions these might be (IPCC, 1990). Rainfall could also be more intense in its occurrence, so flooding and erosion could increase.

Reduced soil water availability

Probably the most important consequences for agriculture would stem from higher potential evapotranspiration, primarily due to the higher temperatures of the air and the land surface. Even in the tropics, where temperature increases are expected to be smaller than elsewhere and where precipitation might increase, the increased rate of loss of moisture from plants and soil would be considerable (Parry, 1990; Rind et al., 1989). It may be somewhat reduced by greater air humidity and increased cloudiness during the rainy seasons, but could be pronounced in the dry seasons.

Types of effect on agriculture

There are three ways in which increases in greenhouse gases (GHG) may be important for agriculture. Firstly, increased atmospheric carbon dioxide (CO2) concentrations can have a direct effect on the growth rate of crop plants and weeds. Secondly, GHG-induced changes of climate may alter levels of temperature (as well as temperature gradient), rainfall and sunshine and this can influence plant and animal productivity. Finally, rises in sea level may lead to loss of farmland by inundation and to increasing salinity of groundwater in coastal areas.

Effects of CO2 enrichment

· Effects on photosynthesis

Carbon dioxide is vital for photosynthesis, and the evidence is that increases in CO2 concentration would increase the rate of plant growth (Cure, 1985; Cure and Acock, 1986). There are, however, important differences between the photosynthetic mechanisms of different crop plants and hence in their response to increasing CO2. Plant species with the C3 photosynthetic pathway (e.g. wheat, rice and soybean) tend to respond more positively to increased CO2 because it tends to suppress rates of photorespiration. However, C4 plants (e.g. maize, sorghum, sugar-cane and millet) are less responsive to increased CO2 levels (Figure 1). Since these are largely tropical crops, and most widely grown in Africa, there is thus the suggestion that CO2 enrichment will benefit temperate and humid tropical agriculture more than that in the semi-arid tropics. Thus, if the effects of climate changes on agriculture in some parts of the semi-arid tropics are negative, then these may not be partially compensated by the beneficial effects of CO2 enrichment as they might in other regions. In addition we should note that, although C4 crops account for only about one-fifth of the world's food production, maize alone accounts for 14% of all production and about three-quarters of all traded grain. It is the major grain used to make up food deficits in famine-prone regions, and any reduction in its output could affect access to food in these areas (Morison, 1990).

The C3 crops in temperate and subtropical regions could also benefit from reduced weed infestation. Fourteen of the world's 17 most troublesome terrestrial weed species are C4 plants in C3 crops. The difference in response to increased CO2 may make such weeds less competitive. In contrast, C3 weeds in C4 crops, particularly in tropical regions, could become more of a problem, although the final outcome will depend on the relative response of crops and weeds to climate changes as well (Morison, 1990). Many of the pasture and forage grasses of the world are C4 plants, including important prairie grasses in North America and central Asia and in the tropics and subtropics. The carrying capacity of the world's major rangelands are


Figure 1 Typical photosynthesis response of plants to CO2

Little is also known about possible changes in yield quality under increased CO!. The nitrogen content of plants is likely to decrease, while the carbon content increases, implying reduced protein levels and reduced nutritional levels for livestock and humans. This may, however, also reduce the nutritional value of plants for pests, so that they need to consume more to obtain their required protein intake.

· Effects on water use by plants

Just as important may be the effect that increased CO2 has on the closure of stomata. This tends to reduce the water requirements of plants by reducing transpiration (per unit leaf area) thus improving what is termed 'water-use efficiency' (the ratio of crop biomass accumulation to the water used in evapotranspiration). A doubling of ambient CO2 concentration causes about a 40% decrease in stomata! aperture in both C3 and C4 plants which may reduce transpiration by 23 46% (Morison, 1987; Cure and Acock, 1986). This might well help plants in environments where moisture currently limits growth, such as in semi-arid regions, but there remain many uncertainties, such as to what extent the greater leaf area of plants (resulting from increased CO2) will balance the reduced transpiration per unit leaf area (Allen et al., 1985; Gifford, 1988). In summary, we can expect that a doubling of atmospheric CO2 concentrations from 330 to 660 ppmv might cause a 10-50% increase in growth and yield of C3 crops (such as wheat, rice and soybean) and a 0-10% increase for C4 crops (such as maize and sugar-cane) (Warrick et al., 1987). Much depends, however, on the prevailing growing conditions. Our present knowledge is based on experiments mainly in field chambers and has not yet included extensive study of response in the field under suboptimal conditions. Thus, although there are indications that, overall, the effects of increased CO2 could be distinctly beneficial and could partly compensate for some of the negative effects of CO2-induced changes of climate, we cannot at present be sure that this will be so.

Effects of changes of climate

· Changes in thermal limits to agriculture

Increases in temperature can be expected to lengthen the growing season in areas where agricultural potential is currently limited by insufficient warmth, resulting in a poleward shift of thermal limits of agriculture. The consequent extension of potential will be most pronounced in the northern hemisphere because of the greater extent here of temperate agriculture at higher latitudes. There may, however, be important regional variations in our ability to exploit this shift. For example, the greater potential for exploitation of northern soils in Siberia than on the Canadian Shield may mean relatively greater increases in potential in northern Asia than in northern N America (Parry, 1990). A number of estimations have been made concerning the northward shift in productive potential in mid-latitude northern hemisphere countries (see, for example, Figure 2). These relate to changes in the climatic limits for specific crops under a variety of climatic scenarios, and are therefore not readily compatible (Newman, 198(); Blasing and Solomon, 1983; Rosenzweig, 1985; Williams and Oakes, 1978; Parry and Carter, 1988; Parry et al., 1989). They suggest, however, that a 1°C increase in mean annual temperature would tend to advance the thermal limit of cereal cropping in the mid-latitude northern hemisphere by about 150-200 km, and to raise the altitudinal limit to arable agriculture by about 15() 200 m.

While warming may extend the margin of potential cropping and grazing in mid-latitude regions, it may reduce yield potential in the core areas of current production, because higher temperatures encourage more rapid maturation of plants and shorten the period of grain filling (Parry and Duinker, 1990). An important additional effect, especially in temperate mid-latitudes, is likely to be the reduction of winter chilling (vernalization). Many temperate crops require a period of low temperatures in winter either to initiate or accelerate the flowering process. Low vernalization results in low flower bud initiation and, ultimately, reduced yields. A 1°C warming has been estimated to reduce effective winter chilling by between 10 and 30%, thus contributing to a poleward shift of temperate crops (Salinger, 1989). Increases in temperature are also likely to affect the crop calendar in low latitude regions, particularly where more than one crop is harvested each year. For example, in Sri Lanka and Thailand, a 1°C warming would probably require a substantial re-arrangement of the current crop calendar which is finely tuned to present climatic conditions (Kaida and Surarerks, 1984; Yoshino, 1984).

· Shifts of moisture limits to agriculture

There is much less agreement between GCM-based projections concerning GHG-induced changes in precipitation than there is about temperature - not only concerning changes of magnitude, but also of spatial pattern and distribution through the year. For this reason it is difficult to identify potential shifts in the moisture limits to agriculture. This is particularly so because relatively small changes in the seasonal distribution of rainfall can have disproportionately large effects on the viability of agriculture in tropical areas, largely through changes in growing period when moisture is sufficient and thus through the timing of critical episodes such as planting, etc. However, recent surveys for the IPCC have made a preliminary identification of those regions where there is some agreement amongst 2 x CO2 experiments with general circulation models concerning an overall reduction in crop-water availability (Parry, 1990; Parry and Duinker, 1990). It should be emphasized that coincidence of results for these regions is not statistically significant. The regions are:


Figure 2 Grain maize limit under the GISS transient response Scenario in the 1990s, 2020s, and 2050s (relative to the limit for the current climate)

Decreases of soil water in December, January and February:

Africa :

north-east Africa, southern Africa

Asia :

western Arabian Peninsula; SE Asia

Australasia :

eastern Australia

N America :

southern USA

S America :

Argentine Pampas

Decreases in soil water in June, July and August:

Africa :

north Africa; west Africa

Europe :

parts of western Europe

Asia :

north and central China; parts of Soviet central Asia and Siberia

N America :

southern USA and Central America

S America :

eastern Brazil

Australia :

western Australia

- Regions affected by drought, heat stress and other extremes

Probably most important for agriculture, but about which least is known, are the possible changes in climatic extremes, such as the magnitude and frequency of drought, storms, heat waves and severe frosts (Rind et al., 1989). Some modelling evidence suggests that hurricane intensities will increase with climatic warning (Emanuel, 1987). This has important implications for agriculture in low latitudes, particularly in coastal regions.

Since crop yields often exhibit a nonlinear response to heat or cold stress, changes in the probability of extreme temperature events can be significant (Mearns et al., 1984; Parry, 1976). In addition, even assuming no change in the standard deviation of temperature maxima and minima, we should note that the frequency of hot and cold days can be markedly altered by changes in mean monthly temperature. To illustrate, under a 2 × CO2 equilibrium climate, the number of days in which temperatures would fall below freezing would decrease from a current average of 39 to 20 in Atlanta, Georgia (USA), while the number of days above 90°F would increase from 17 to 53 (EPA, 1989). The frequency and extent of area over which losses of agricultural output could result from heat stress, particularly in tropical regions, is therefore likely to increase significantly. Unfortunately, no studies have yet been made of this. However, the apparently small increases in mean annual temperatures in tropical regions (c. 1-2°C under a 2 × CO2 climate) could sufficiently increase heat stress on temperate crops such as wheat so that these are no longer suited to such areas. Important wheat producing areas such as N India could be affected in this way (Parry and Duinker, 1990).

There is a distinct possibility that, as a result of high rates of evapotranspiration, some regions in the tropics and subtropics could be characterised by a higher frequency of drought or a similar frequency of more intense drought than at present. Current uncertainties about how regional patterns of rainfall will alter mean that no useful prediction of this can at present be made. However, it is clear in some regions that relatively small decreases in water availability can readily produce drought conditions. In India, for example, lower-than-average rainfall in 1987 reduced food grains production from 152 to 134 million tonnes, lowering food buffer stocks from 23 to 9 million tonnes. Changes in the risk and intensity of drought, especially in currently drought-prone regions, represent potentially the most serious impact of climatic change on agriculture both at the global and the regional level.

· Effects on the distribution of agricultural pests and diseases

Studies suggest that temperature increases may extend the geographic range of some insect pests currently limited by temperature (EPA, 1989; Hill and Dimmmock, 1989). As with crops, such effects would probably be greatest at higher latitudes. The number of generations per year produced by multivoltine (i.e. multigenerational) pests would increase, with earlier establishment of pest populations in the growing season and increased abundance during more susceptible stages of growth. An important unknown, however, is the effect that changes in precipitation amount and air humidity may have on the insect pests themselves and on their predators, parasites and diseases. Climate change may significantly influence interspecific interactions between pests and their predators and parasites. Under a warmer climate at mid-latitudes there would be an increase in the overwintering range and population density of a number of important agricultural pests, such as the potato leafhopper which is a serious pest of soybeans and other crops in the USA (EPA, 1989). Assuming planting dates did not change, warmer temperatures would lead to invasions earlier in the growing season and probably lead to greater damage to crops. In the US Corn Belt increased damage to soybeans is also expected due to earlier infestation by the corn earworm.

Examination of the effect of climatic warming on the distribution of livestock diseases suggests that those at present limited to tropical countries, such as Rift Valley fever and African Swine fever, may spread into the mid-latitudes. For example, the horn fly, which currently causes losses of $ 730.3 million in the US beef and dairy cattle industries might extend its range under a warmer climate leading to reduced gain in beef cattle and a significant reduction in milk production (Drummond, 1987; EPA, 1989). In cool temperate regions, where insect pests and diseases are not generally serious at present, damage is likely to increase under warmer conditions. In Iceland, for example, potato blight currently does little damage to potato crops, being limited by the low summer temperatures. However, under a 2 x CO2 climate that may be 4°C wanner than at present, crop losses to disease may increase to 15% (Bergthorsson et al., 1988). Most agricultural diseases have greater potential to reach severe levels under warmer and more humid conditions (Beresford and Fullerton, 1989). Under warmer and more humid conditions cereals would be more prone to diseases such as Septoria. In addition, increases in population levels of disease vectors may well lead to increased epidemics of the diseases they carry. To illustrate, increases in infestations of the Bird Cherry aphid (Rhopalosiphum padi) or Grain aphid (Sitobian avenae) could lead to increased incidence of Barley Yellow Dwarf Virus in cereals.

Effects of sea-level rise on agriculture

Greenhouse gas-induced warming is expected to lead to rises in sea level as a result of thermal expansion of the oceans and partial melting of glaciers and ice caps, and this in turn is expected to affect agriculture, mainly through the inundation of low-lying farmland but also through the increased salinity of coastal groundwater. The current projection of sea-level rise above present levels is 20±10 cm by c. 2030, and 30±15 cm by 2050 (Warrick and Oerlemans, 1990).

Preliminary surveys of proneness to inundation have been based on a study of existing contoured topographic maps, in conjunction with knowledge of the local 'wave climate' that varies between different coastlines. They have identified 27 countries as being especially vulnerable to sea-level rise, on the basis of the extent of land liable to inundation, the population at risk and the capability to take protective measures (UNEP, 1989). It should be emphasised, however, that these surveys assume a much larger rise in sea levels than is at present estimated to occur within the next century under current trends of increase of GHG concentrations. On an ascending scale of vulnerability (1 to 10) experts identified the following most vulnerable countries or regions: 10, Bangladesh; 9, Egypt, Thailand; 8, China; 7, western Denmark; 6, Louisiana; 4, Indonesia. The most severe impacts are likely to stem directly from inundation. South East Asia would be most affected because of the extreme vulnerability of several large and heavily populated deltaic regions. For example, with a 1.5 m sea-level rise, about 15% of all land (and about one-fifth of all farmland) in Bangladesh would be inundated and a further 6% would become more prone to frequent flooding (UNEP, 1989). Altogether 21% of agricultural production could be lost. In Egypt, it is estimated that 17% of national agricultural production and 20% of all farmland, especially the most productive farmland, would be lost as a result of a 1.5 m sea-level rise. Island nations, particularly low-lying coral atolls, have the most to lose. The Maldive Islands in the Indian Ocean would have one-half of their land area inundated with a 2 m rise in sea level (UNEP, 1989).

In addition to direct farmland loss from inundation, it is likely that agriculture would experience increased costs from saltwater intrusion into surface water and groundwater in coastal regions. Deeper tidal penetration would increase the risk of flooding and rates of abstraction of groundwater might need to be reduced to prevent re-charge of aquifers with sea water. Further indirect impacts would be likely as a result of the need to relocate both farming populations and production in other regions. In Bangladesh, for example, about one-fifth of the nation's population would be displaced as a result of the farmland loss estimated for a 1.5 m sea-level rise. It is important to emphasize, however, that the IPCC estimates of sea-level rise are much lower than this (about 0.5 m by 2090 under the IPCC 'Business-As-Usual case').

Summary of plant, pest and sea-level effects

Potential impacts on yields vary greatly according to types of climate change and types of agriculture. In general, there is much uncertainty about how agricultural potential may be affected. In the northern mid-latitudes where summer drying may reduce productive potential (e.g. in the US Great Plains and Corn Belt, Canadian Prairies, southern Europe, south European former USSR) yield potential is estimated to fall by c 10-30% under an equilibrium 2 x CO2 climate (Parry, 1990). However, towards the northern edge of current core-producing regions (e.g. the northern edge of the Canadian Prairies, northern Europe, northern states of the former Soviet Union and Japan, southern Chile and Argentina) warming may enhance productive potential, particularly when combined with beneficial direct CO2 effects. Much of this potential may not, however, be exploitable owing to limits placed by inappropriate soils and difficult terrain, and on balance it seems that the advantages of warming at higher latitudes would not compensate for reduced potential in current major cereal producing regions.

Effects at lower latitudes are much more difficult to estimate because production potential is largely a function of the amount and distribution of precipitation and because there is little agreement about how precipitation may be affected by GHG warming. Because of these uncertainties the tendency has been to assert that worthwhile study must await improved projection of changes in precipitation. Consequently very few estimates are currently available of how yields might respond to a range of possible changes of climate in low-latitude regions. The only comprehensive national estimates available are for Australia where increases in cereal and grassland productivity might occur (except in western Australia) if warming is accompanied by increase in summer rainfall (Pearman, 1988).

The impacts described above relate to possible changes in potential productivity or yield. It should be emphasised that such potential effects are those estimated assuming present-day management and technology. They are not the estimated future actual effects, which will depend on how farmers and governments respond to altered potential through changes in management and technology. The likely effects on actual agricultural output and on other measures of economic performance such as profitability and employment levels are considered in the next section.

Effects on production and land use

To date (1992) six national case studies have been made of the potential impact of climatic changes on agricultural production (in Canada, Iceland, Finland, the former Soviet Union, Japan and the United States) (Parry et al., 1989; Smit, 1989; EPA, 1989). These studies are based on results from model experiments of yield responses to altered climate and the effects that altered yields might have on production. Other countries have conducted national reviews of effects of climate change, basing these on existing knowledge rather than on new research. [he most comprehensive of these are for Australia and New Zealand (Pearman, 1988; Salinger et al., 1990). Brief surveys have also been completed in the UK and West Germany (SCEGB, 1989). Several other national assessments are currently in progress but not yet complete. This section provides a summary of results from the most detailed of these surveys. These provide us with an array of assessments for three world regions: the northern and southern mid-latitude grain belts and northern regions at the current margin of the grain belt. We shall take these regions in turn. Unless otherwise stated the estimated effects are for climates described by 2 x CO2 GCM experiments. No national assessments have been completed using climates described by transient response GCM experiments. Some of the estimates relate to the effect only of altered climate, others the combined effect of altered climate and the direct effect of increased atmospheric CO2.

Effects on production in the northern and southern mid-latitude grain belts

· United States

Increased temperatures and reduced crop water availability projected under the GISS and GDFL 2 x CO2 climate experiments are estimated to lead to a decrease of yields of all the major unirrigated crops (EPA, 1989) (Figure 3). The largest reductions are projected for the south and south-east. In the most northern areas, however, where temperature is currently a constraint on growth, yields of unirrigated maize and soybeans could increase as higher temperatures increase the length of the available growing season. When the direct effects of increased CO2 are considered, it is evident that yields may increase more generally in northern areas but still decrease in the south where problems of heat stress would increase and where rainfall may decrease. Production of most crops is estimated to be reduced because of yield decreases and limited availability of suitable land. The largest reductions are in sorghum (-20%), corn (-13%) and rice (-11%), with an estimated fall in net value of agricultural output of $ 33 billion. If this occurred, consumers would face slightly higher prices, although supplies are estimated to meet current and projected demand. However, exports of agricultural commodities could decline by up to 70%, and this could have a substantial effect on the pattern of world food trade.

· Canada

On the Canadian prairies, where growing season temperatures under the GISS 2 x CO2 equilibrium climate would be about 3.5°C higher than today, average potential yields could decrease 10 to 30%. Spring wheat yields in Saskatchewan are estimated to fall by 28% (Williams et al., 1988). Since Saskatchewan at present produces 18% of all the world's traded wheat, such a reduction could well have global implications. Assuming (unrealistically) that the present-day relationship between production and profit in Saskatchewan holds in the future, average farm household income is estimated to fall by 12%, resulting in a reduction in expenditure by agriculture of Can$ 277 million on the goods and services provided by other sectors, leading to a Can$ 250 million (6%) reduction in provincial GDP in sectors other than agriculture and a 1% loss of jobs. In Ontario, precipitation increases of up to 50% would be more than offset by increases in evapotranspiration with consequent increased moisture stress on crops (Smit, 1987). Maize and soybean would thus become very risky in the southern part of the province. In the north, where maize and soybean cannot currently be grown commercially because of inadequate warmth, cultivation may become profitable but this is not expected to compensate for reduced potential further south and, if there were no adjustment of current land use and farming systems, the overall cost in lost production is reckoned at Can$ 100 million to 170 million.


Figure 3 Estimated maize yields in the USA under the GISS and GFDL 2 x CO2 with and without the direct effects (DE) of CO2 (a) dryland and (b) irrigated

· Japan

Under a warming of 3.0-3.5°C and a 5% increase in annual precipitation (the GISS 2 x CO2 climate), rice yields are expected to increase in the north (Hokkaido) by c. 5%, and in the north-central region (Tohoku) by c. 2%, if appropriate technological adjustments are made (Yoshino et al., 1988). The average increase for the country overall is c. 2-5%. Cultivation limits for rice would rise about 500 m and advance c. 100 km north in Hokkaido. Yields of maize and soybeans are both estimated to increase by about 4%. Sugar-cane yields in the most southern part of Japan could decrease if rainfall was reduced. The northern economic limit of citrus fruits would shift from southern Japan to northern Honshu Island (Yoshino, personal communication, 1989). Net primary productivity of natural vegetation is expected to increase by c. 15% in the north, c. 7% in the centre and south of Japan (Yoshino et al., 1988).

· Australia and New Zealand

In Australia and New Zealand national assessments have been based on a thorough review of existing knowledge and on use of expert judgement rather than on model experiments (Pearman, 1988; Salinger et al., 1990). Overall, it is reckoned that wheat production in Australia could increase under a 2 x CO2 climate, assuming a quite simple scenario of increased summer rainfall, decreased winter rainfall and a general warming of 3°C. Increases are expected in all states except Western Australia, where more aridity might cause a significant reduction in output (Pittock, 1989). More generally, the major impact of production would probably be on the drier frontiers of arable cropping. For example, increases in rainfall in subtropical northern Australia could result in increased sorghum production at the expense of wheat. Increased heat stress might shift livestock farming and wool production southward with sheep possibly replacing arable farming in some southern regions. Many areas currently under fruit production would no longer be suitable under a 3 C warming, and would need to shift southwards or to higher elevations in order to maintain present levels of production. All of these changes would also be affected by changes in the distribution of diseases and pests.

Effects on production in northern marginal regions

Some of the most pronounced effects on agriculture would be likely to occur in high-latitude regions because GHG-induced warming is projected to be greatest here and because this warming could remove current thermal constraints on farming. Inappropriate terrain and soils are, however, likely to limit the increase in extent of the farmed area and, in global terms, production increases would probably be small (Parry, 1990). A summary of available information is given below.

· Iceland

With mean annual temperatures increased by 4.0°C and precipitation 15% above the present average (consistent with the GISS 2 x CO2 climate), the onset of the growing season of grass in Iceland would be brought forward by almost 50 days, hay yields on improved pastures would increase by about two-thirds and herbage on unimproved rangelands by about a half (Bergthorsson et al., 1988). The numbers of sheep that could be carried on the pastures would be raised by about 250% and on the rangelands by two-thirds if the average carcass weight of sheep and lambs is maintained as at present. At a guess, output of Icelandic agriculture could probably double with a warming of 4°C.

· Finland

Assuming in Finland an increase in summer warmth by about a third and precipitation by about half (consistent with the GISS 2 x CO2 climate) barley and spring wheat yields increase about 10% in the south of the country but slightly more in the north (due to relatively greater warming and lower present-day yields) (Kettunen et al., 1988). The area under grain production in Finland might increase at the expense of grass and livestock production as a consequence of raised profitability, with the greatest extension being in winter crops such as wheat rather than spring crops such as barley or oats.

· Northern states former Soviet Union

The only other region for which an integrated impact assessment has been completed is in north European former Soviet Union. In the Leningrad and Cherdyn regions, under climates that are 2.2-2.7°C warmer during the growing season and 36-50% wetter (consistent with the GISS 2 x CO2 climate), winter rye yields are estimated to decrease by about a quarter due to faster growth and increased heat stress under the higher temperatures (Pitovranov et al., 1988a and b). However, crops such as winter wheat and maize, which are currently low yielding because of the relatively short growing season in these regions, are better able to exploit the higher temperatures and exhibit yield increases in Cherdyn of up to 28% and 6% respectively with a 1°C warming.

The differential yield responses described above are reflected in substantial changes in production costs incurred in meeting production targets. Thus, while production costs for winter wheat and maize in the Central Region around Moscow are estimated to be reduced by 22% and 6% under a 1°C warming and with no change in precipitation, they increase for most other crops, particularly quick-maturing spring-sown ones which are the dominant crops today. This would suggest that quite major switches of land use would result and the land allocation models used in the study indicate that, to optimise land use by minimising production costs, winter wheat and maize would extend their area by 29% and 5% while barley, oats and potatoes would decrease in extent.

· Summary of potential effects in mid-latitude regions

The effects of possible climatic changes on regional and national production have not yet been investigated in any great detail, nor for more than a few case studies. The effects are strongly dependent on the many adjustments in agricultural technology and management that undoubtedly will occur in response to any climatic change. So numerous and varied are these potential adjustments that it is extraordinarily difficult to evaluate their ultimate effect on aggregate production. In this section we have therefore considered the effects on production that are likely to stem directly from changes in yield, unmodified by altered technology and management. Adjustments in technology will be considered, briefly, in the last section of this paper. In summary, it seems that overall output from the major present-day grain producing regions could well decrease under the warming and possible drying expected in these regions. In the USA grain production may be reduced by 10-20% and, while production would still be sufficient for domestic needs, the amount for export would probably decline. Production may also decrease in the Canadian prairies and in the southern former Soviet Union. In Europe production of grains might increase in the UK and the Low Countries if rainfall increases sufficiently, but may fall in southern Europe substantially if there are significant decreases in rainfall as currently estimated in most GCM 2 × CO: experiments (Parry, 199()). Output could increase in Australia if there is a sufficient increase in summer rainfall to compensate for higher temperatures. Production could increase in regions currently near the low-temperature limit of grain growing: in the northern hemisphere in the northern Prairies, Scandinavia, north European (former) Soviet Union; and in the southern hemisphere in southern New Zealand, and southern parts of Argentina and Chile. But it is reasonably clear that, because of the limited area unconstrained by inappropriate soils and terrain, increased high-latitude output will probably not compensate for reduced output at mid-latitudes. The implications of this for global food supply and food security are considered in the next section.

Implications for global food security

Although, on average, global food supply currently exceeds demand by about 10-20, its year-to-year variation (which is about ±10%) can reduce supply in certain years to levels where it is barely sufficient to meet requirements. In addition, there are major regional variations in the balance between supply and demand, with perhaps a billion people (about 15% of the world's population) not having secure access to sufficient quantity or quality of food to lead fully productive lives. For this reason the working group on food security at the 1988 Toronto Conference on 'The Changing Atmosphere' concluded that:

"While averaged global food supplies may not be seriously threatened, unless appropriate action is taken to anticipate climate change and adapt to it, serious regional and year-to-year food shortages may result, with particular impact on vulnerable groups .

Statements such as this are, however, based more on intuition than on knowledge derived from specific study of the possible impact of climate change on the food supply. No such study has yet been completed, although one is currently being conducted by the US Environmental Protection Agency and is due to report in 1992. The information available at present is extremely limited. It has for example, been estimated that increased costs of food production due to climate change could reduce per caput global GNP by a few percentage points (Schelling, 1983). Others have argued that technological changes in agriculture will override any negative effects of climate changes and, at the global level, there is no compelling evidence that food supplies will be radically diminished (Crosson, 1989). Recent reviews have tended to conclude however that, at a regional level, food security could be seriously threatened by climate change, particularly in less developed countries in the semi-arid and humid tropics (Parry, 1990; Parry and Duinker, 1990). Analyses conducted for the IPCC, designed to test the sensitivity of the world food system to changes of climate, indicate what magnitudes and rates of climatic change could possibly be absorbed without severe impact and, alternatively, what magnitudes and rates could seriously perturb the system (Parry, 1990; Parry and Duinker, 1990). These suggest that yield reductions of up to 20% in the major mid-latitude grain exporting regions could be tolerated without a major interruption of global food supplies. However, the increase in food prices (7% under a 10% yield reduction) could seriously influence the ability of food-deficit countries to pay for food imports, eroding the amount of foreign currency available for promoting development of their non-agricultural sectors. It should be emphasized that these analyses are preliminary and more work is necessary before we have an adequate picture of the resilience of the world food system to climatic change.

Conclusions

Our assessment of possible effects has, up to this point, assumed that technology and management in agriculture do not alter significantly in response to climatic change, and thus do not alter the magnitude and nature of the impacts that may stem from that change. It is certain, however, that agriculture will adjust and, although these adjustments will be constrained by economic and political factors, it is likely that they will have an important bearing on future impacts. On balance, the evidence is that food production at the global level can, in the face of estimated changes of climate, be sustained at levels that would occur without a change of climate, but the cost of achieving this is unclear. It could be very large. Increases in productive potential at high mid-latitudes and high latitudes, while being of regional importance, are not likely to open up large new areas for production. The gains in productive potential here due to climatic warming would be unlikely to balance possible large-scale reductions in potential in some major grain-exporting regions at mid-latitude. Moreover, there may well occur severe negative impacts of climate change on food supply at the regional level, particularly in regions of high present-day vulnerability least able to adjust technically to such effects. The average global increase in overall production costs could thus be small (perhaps a few percent of world agricultural GDP). Much depends however, on how beneficial are the so-called 'direct' effects of increased CO2 on crop yield. If plant productivity is substantially enhanced and more moisture is available in some major production areas, then world productive potential of staple cereals could increase relative to demand with food prices reduced as a result. If, on the contrary, there is little beneficial direct CO2 effect and climate changes are negative for world agricultural production could increase significantly, these increased costs amounting to perhaps over 10% of world agricultural GDP.

Although we know little, at present, about how the frequency of extreme weather events may alter as a result of climatic change, the potential impact of concurrent drought or heat stress in the major food-exporting regions of the world could be severe. In addition, relatively small decreases in rainfall or increases in evapotranspiration could markedly increase both the risk and the intensity of drought in currently drought-prone (and often food-deficient) regions. Change in drought-risk represents potentially the most serious impact of climatic change on agriculture both at the regional and the global level. The regions most at risk from impact from climatic change are probably those currently most vulnerable to climatic variability. Frequently these are low-income regions with a limited ability to adapt through technological change. This paper has emphasised the inadequacy of our present knowledge. It is clear that more information on potential impacts would help us identify the full range of potentially useful responses and assist in determining which of these may be most valuable.

Some priorities for future research may be summarised as follows:

· Improved knowledge is needed of effects of changes in climate on crop yields and livestock productivity in different regions and under varying types of management; on soil-nutrient depletion; on hydrological conditions as they effect irrigation-water availability; on pests, diseases and soil microbes, and their vectors; and on rates of soil erosion and salinisation.

· Further information is needed on the range of potentially effective technical adjustments at the farm and village level (e.g. irrigation, crop selection, fertilizing, etc.); on the economic, environmental and political constraints on such adjustments; and on the range of potentially effective policy responses at regional, national and international levels (e.g. re-allocations of land use, plant breeding, improved agricultural extension schemes, large-scale water transfers).

References

AKITA, S. and MOSS, D. N. (1973) Photosynthetic responses to CO2 and light by maize and wheat leaves adjusted for constant stomata! apertures. Crop Science, 13, 234-237.

ALLEN, L. H. Jr., JONES, P. and JONES, J. W. (1985) Rising atmospheric CO2 and evapotranspiration. Advances in evapotranspiration. Proceedings of the National Conference on Advance in Evaporation, American Society of Agricultural Engineers, St Joseph, Michigan, USA.

BERESFORD, R. M. and FULLERTON, R. A. (1989) Effects of climate change on plant diseases. Submission to Climate Impacts Working Group, May 1989.

BERGTHORSSON, P., BJORNSSON, H., DYRMUNDSSON, O.,GUDMUNSSON, B., HELGADOTTIR, A. and JONMUNDSSON, J. V. (1988) The effects of climatic variations on agriculture in Iceland. In: The Impact of Climatic Variations on Agriculture, Vol. 1, Cool Temperature and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Klawer, Dordecht, The Netherlands.

BIASING, T. J. and SOLOMON, A. M. (1983) Response of North American Corn Belt to Climatic Warming. Prepared for the US Department of Energy, Office of Energy Research, Carbon Dioxide Research Division, DOE/N88-004, Washington, DC.

CARTER, T. R., PARRY, M. L. and PORTER, J. H. International]ournal of Climatology, 2, 251-269.

CROSSON, P. (1989) Greenhouse warming and climate change: why should we care? Food Policy, 14-2, 107-118.

CURE, J. D. (1985) Carbon dioxide doubling responses: a crop survey. In: Direct Effects of Increasing Carbon Dioxide on Vegetation. STRAIN, B. R. and CURE, J. D. (eds), US DOE/ERO238, Washington, DC.

CURE, J. D. and ACOCK, B. (1986) Crop responses to carbon dioxide doubling: a literature survey. Agricultural and Forest Meteorology, 38, 127-145.

DRUMMOND, R. O. (1987) Economic aspects of ectoparasites of cattle in North America. In: Symposium on the Economic Impact of Parasitism in Cattle, Twenty-third World Veterinary Congress, Montreal.

EMANUEL, K. A. (1987)The dependence of hurricane intensity on climate: mathematical simulation of the effects of tropical sea surface temperatures. Nature, 326, 483-485.

ENVIRONMENTAL PROTECTION AGENCY (1989) The Potential Effects of Global Climate Change on the United States. Report to Congress.

GIFFORD, R. M. (1988) Direct effect of higher carbon dioxide levels on vegetation. In: Greenhouse. Planning for Climate Change. PEARMAN, G. I. (ed) CSIRO, Australia.

HILL, M. G. and DIMMOCK, J. J. (1989) Impact of Climate Change: Agricultural/Horticultural Systems. DSIR Entomology Division, submission to New Zealand Climate Change Programme, Department of Scientific and Industrial Research, New Zealand.

INTERGOVERNMENTAL PANEL on CLIMATE CHANGE (IPPC) (1990) Scientific Assessment of Climate Change: Policymakers Summary. WMO and UNEP, Geneva and Nairobi.

KAIDA, Y. and SURARERKS, V. (1984) Climate and agricultural land use in Thailand. In: Climate and Agricultural Land Use in Monsoon Asia. YOSHINO, M. M. (ed), University of Tokyo Press, Tokyo.

KETTUNEN, L. MUKULA, 1. POHJONEN, V. RANTANEN, O. and VARJO, U. (1988) The effects of climate variations in Finland. In: The Impact of Climate Variations on Agriculture, Vol. 1 Assessments in Cool Temperate and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Kluwer, Dordecht, The Netherlands.

MEARNS, L. O., KAATAZ, R. W. and SCHNEIDER, S. H. (1984) Extreme high-temperature events: changes in their probabilities with changes in mean temperatures. Journal of Climatic and Applied Meteorology, 23, 1601-1613.

MORISON, J. I. L. (1987) Intercellular CO2 concentration and stomata! response to CO2. In: Stomatal Function. ZEIGER, E., COWAN, I. R. and FARAQUAR, G. D. (eds), Stanford University Press, Stanford.

MORISON, J. I. L. (1990) Direct effects of elevated atmospheric CO2 and other greenhouse gases. In: the Potential Effects of Climate Change on Agriculture and Forestry. PARRY, M. L. and DUINKER, P. N. (eds), Working Group II Report, IPCC.

NEWMAN, J. E. (1980) Climate change impacts on the growing season of North American corn belt. Biometeorology, 7-2, 128-142.

PARRY, M. L. (1976) Climatic Change, Agriculture and Settlement. Dawson, Folkestone, UK.

PARRY, M. L. (1990) Climatic Change and World Agriculture. Earthscan, London.

PARRY, M. L. and CARTER, T. R. (1988) The assessments of the effects of climatic variations on agriculture: aims, methods and summary of results. In: The Impact of Climatic Variations on Agriculture, Vol. 1, Assessments in Cool Temperate and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Klawer, Dordecht, The Netherlands

PARRY, M. L., CARTER, T. R. and PORTER, J. H. (1989) The greenhouse effect and the future of UK agriculture. Journal of the Royal Agricultural Society of England, 120-131.

PARRY, M. L. and DUINKER, P. N. (1990) The Potential Effects of Climatic Change on Agriculture and Forestry. Working Group II Report, IPCC.

PEARMAN, G. I. (ed) (1988) Greenhouse. Planning for Climate Change, CSIRO, Melbourne, Australia.

PITOVRANOV, S. E., IAKIMETS, V., KISLEV, V. E. and SIROTENKO, O. D. (1988a) The effects of climatic variations on agriculture in the subarctic zone of the USSR. In: The Impact of Climatic Variations on Agriculture, Vol. 1, Assessments in Cool, Temperate and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Kluwer, Dordecht, The Netherlands.
PITOVRANOV, S. E., IAKIMETS, V., KISLEV, V. E. and SIROTENKO, O. D. (1988b) The effects of c:limatic variations on agriculture in the semi-arid zone of the USSR. In: The Impact of Climatic Variations on Agriculture, Vol. 2, Assessments in Semi-and Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Kluwer, Dordecht, The Netherlands.

PITTOCK, A. B. (1989) The Greenhouse Effect, Regional Climate Change and Australian Agriculture. Paper presented to the Australian Society of Agronomy, Fifth Agronomy Conference, Perth, Australia.

RIND, D., GOLDBERG, R. and RUEDY, R. (1989) Change in climate variability in the 21st century. Climatic Change, 14, 5-37.
xxxxxxxxxxxxxxxxxxxxx
ROSENZWEIG, C. (1985) Potential CO2-induced climate effects on North American wheat producing regions. Climatic Change, 10, 367-369.

SALINGER, M. J. (1989a) The Effects of Greenhouse Gas Warming on Forestry and Agriculture. Draft Report for WMO Commission in Agrometeotology.

SALINGER, M. J. (1989b) Carbon Dioxide and Climate Change: Impacts on New Zealand Agriculture. (personal communication).

SALINGER, M. J., WILLIAMS, W. M., WILLIAMS, J. M. and MARTIN, R. J. (1990) Carbon Dioxide and Climate Change: Impacts on Agriculture, New Zealand. New Zealand Meteorological Service, DSIR Grasslands Division, MAFI ech.

STUDY COMMISSION OF ELEVENTH GERMAN BUNDESTAG (SCEGB) (1989) Protecting the Earth's Atmosphere: an International Challenge. Bonn University, Bonn.

SCHELLING, T. (1983) Climate change: implications for welfare and policy. In: Changing Climate: Report of the Carbon Dioxide Assessment Committee. National Academy of Sciences, Washington, DC.

SMIT, B. (1987) Implications on climate change for agriculture in Ontario. Climate Change Digest, 87-02, Environment Canada.

SMIT, B. (1989) Climate warming and Canada's position in agriculture. Climate Change Digest, 89-01, Environment Canada.

UNITED NATIONS ENVIRONMENT PROGRAMME (UNEP) (1989) Criteria for Assessing Vulnerability to Sea-level Rise: a Global Inventory to High-risk Areas. United Nations Environment Programme and the Government of The Netherlands, Draft Report.

WARRICK, R. A. and GIFFORD, R. with PARRY, M. L. (1986) CO2, climate change and agriculture. In: The Greenhouse Effect, Climate Change and Ecosystems. BOLIN, 13., DOOS, B.R., JAGER, J. and WARRICK, R. A. (eds), SCOPE 29, John Wiley & Sons, Chichester, UK.

WARRICK, R. A. and OERLEMANS, J. (199()) Sea-level rise. In: IPPC Scientific Assessment of Climate Change. WMO and UNEP,Geneva and Nairobi.

WILLIAMS, G. D. V., FAUTLE, R. A., JONES, K. H., STEWARD, R. B. and WHEATON, E. E. (1988) Estimating effects of climatic change on agriculture in Saskatchewan. In: The Impact of Climatic Variations on Agriculture, Vol. 1, Assessments in Cool Temperate and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Kluwer, Dordecht, The Netherlands.

WILLIAMS, G. D. V. and OAKES, W. T. (1978) Climatic resources for maturing barley and wheat in Canada. In: &says on Meteorology and Climatology. In honour of Richard W. Longley, Studies in Geography Mono 3. HAYE, K. D. and REINELT, E. R. (eds), University of Alberta, Edmonton, Alberta, USA.

YOSHINO, M. M. (1984) Ecoclimatic systems and agricultural land use in monsoon Asia. In: Climate and Agricultural Land Use in Monsoon Asia. YOSHINO, M. M. (ed), University of Tokyo Press, Tokyo.

YOSHINO, M. M., HORIE, T., CEIN, H. TSUJJI, H., UCHIJIMA, T. and UCHIJIM Z. (1988) The effect of climatic variations on agriculture in Japan. In: The Impact of Climatic Vanations on Agriculture, Vol. 1, Assessments in Cool Temperate and Cold Regions. PARRY, M. L., CARTER, T. R. and KONIJN, N. T. (eds) Kluwer, Dordecht, The Netherlands.

Discussion

The discussion opened with a question on comparison between the level of climatic change predicted and annual variation as well as recent changes in the Sahel. In response the author said that 1976 was a 1 in 800 year event in the UK but with the predicted change it would be a 1 in 20 event. The 200-300 km southward move of the desert in the Sahel in the 1960s and 1970s is the degree of change predicted for a 1°C rise in temperature. No generalisations about changes in climatic variability are possible. It was also agreed that the regional estimates are not yet reliable enough and that standard errors should be added to the estimates. In about 10 years it is expected that sufficiently good models will have been developed to be reasonably sure of the signs of climatic change. It was suggested that climatic change might be an opportunity in that better farmers might respond to the change with output that replaced that of less efficient fanners. It was said that the issue is one of reducing vulnerability and increasing resilience in agricultural systems.