Some differences between developing and developed countries
About 40% of the worldwide net plant production (natural vegetation and agriculture) has been utilized and appropriated by human activities (Skole, 1997), but only 15% of the world food is traded (Fischer et al., 1996). Only a small part of agricultural production thus enters commercial channels, which is to say that the majority of farmers grow their own food.

This is the definition of "subsistence farming". It usually entails that the production is also low and precarious.

According to UN (1996) figures, the 2000 rural population will be only 24% in developed countries, and 75% in the least developed ones (see note 1).

In developed countries, only a fraction rural population is actually involved in farming, and the population relies on markets for their food supply.This is to say that the strategies of farmers, and their perception of risk, are vastly different: a subsistence farmer aims at stabilizing his production, while farmers in developed countries aim essentially at maximizing their economic return.

There are many other differences: developing countries usually make a clear difference between food crops and cash crops, while all crops in developed countries are "cash crops". In developing countries, most cash crops are industrial export crops (coffee, sisal...), and some of them are grown as an "insurance against good years" (see note 2).

An attempt of a more systematic treatment of the subject is given in Table 1 below. The listed differences entail large differences in strategies.

Risk management sensu stricto
There is little doubt that one of the main problems facing the individual farmer in developing and developed countries alike is the unpredictability of weather at a seasonal scale.

Even under traditional farming conditions with no inputs other than labour, several management decisions have to be taken: choice of crops and varieties to be planted either in isolation or mixed, choice of land where there id sufficient land to chose, choice of planting dates, etc. Subsistence farmers sometimes make use of bewildering spectrum of traditional varieties, of which they know and understand the eco-physiological response. Diversification is thus one of the most basic risk management approaches used at the subsistence level as well. It should also be stressed that traditional systems can be very robust because of their low water consumption (as compared with improved varieties) and low input requirements (fertilizers too increase water consumption and the risk of agricultural drought!).

It remains that subsistence farmers will be gradually forced out of their traditional farming, willingly or unwillingly, mainly because of land shortage, urban demand and general development.

Climate risk management techniques in agriculture can be categorized as below. All the techniques listed assume that long-term data are actually available to properly assess the risks and statistically most efficient response mechanisms.

Structural measures that reduce the variability of climate resources at plant level
They include irrigation, water harvesting, windbreaks, frost protection, artificial and controlled climates (greenhouses...), microclimate manipulation... Most of them entail significant costs that actually require government participation when they are implemented in developing countries.

Non-structural measures
Again, most non-structural measures include a cash component that is non-existent or difficult to implement at the subsistence level. Credit to farmers, for instance, is still the exception rather than the rule. Crop insurance can be resorted to only when there is sufficient spatial variability of the environmental stress (e.g. with hail), but remain extremely difficult to implement for some of the major risks, such as drought, which typically affect large areas, sometimes whole countries. They are certainly not feasible without government intervention. One of the techniques which have been adopted with credit and insurance is to make them conditional to the adoption by farmers of improved, risk-reducing practices, like early planting.

Improved legislation, if it can actually be implemented is probably one of the measures that has the best potential in the long-run. This includes watershed protection (for instance by preventing grazing in some forest areas), and all the measures aiming at improving farming efficiency under a no-regrets approach. Finally, preparedness plans are mentioned.

Improved use of climate knowledge and technology
This includes the development of monitoring systems and response mechanisms to current weather, both at farm and government level. Under technology we include mainly the modelling of future impacts based on current weather (within season), and decision tools of varying complexity.

In practice, the decision tools are tables/flow-charts or software that assist farm-level management decision-making based three types of inputs:

• the knowledge of local environmental/agricultural conditions (reference data; see note 4);
• the measurement of local "decision parameters" by local extension officers or farmer;
• economic considerations, e.g. Cost of inputs Vs expected output.

The major challenge facing all crop-weather modelling is the incorporation of qualitative changes deriving from complex interactions. For instance, the above-mentioned shortage of land in several countries will entail severe qualitative changes. Whether the resulting chain of interactions (social conflicts) can be modelled is rather dubious.

Notes
Note 1. The corresponding figures projected for 2030 are 17 and 56 %, respectively.

Note 2. Farmers market a small part of their food crops. During good years, prices tend to drop, in which case the government still buys cotton or sisal, thereby ensuring some cash income.

Note 3. This is sometimes referred to as "contemplative herding".

Note 4. A simple example of this could be, for instance, a threshold of air moisture or sunshine duration to decide on pest risk, or a threshold of salt content of water to decide on irrigation-salinity risk. More complex applications require the use of models, which can usually not be run at the village level but require regular communication with a central office

References
Skole, D.L., 1997. Introduction to the proceedings of the open science meeting, pages 11-12 in Fresco, L., R. Leemans, B.L. Turner II, D. Skole, A.G. vanZeijl-Rozema and V. Haarmann, 1997. Land use and cover change. Open Science meeting proceedings, 29-31 January 1996, Royal Academy of Arts and Sciences, Amsterdam. LUCC Report series N.1, Institut Cartogrāfic de catlunya, Barcelona, Spain. 143 pp.

UN, 1996. World population prospects, the 1996 revision. Annex I: Demographic indicators. United nations Secretariat, Population Division, New-York. 253 pp.

Fischer, G., K. Frohberg, M.L. Parry and C. Rozenzweig, 1996. The potential effects of Climate Change on World Food Production and Security. Pages 199-235 in : Bazzaz, F., and W. Sombroek (Eds.), 1996. Global climate change and agricultural production. Direct and indirect effects of changing hydrological, pedological and plant physiological processes. FAO and John Wiley & Sons, 345 pp.