We have extended an activity initiated during the incubation project that focused on building decision maps and calendars to identify entry points for climate information. This effort is viewed as a first step to assess the scope for adaptation. The maps characterize (a) production decisions, (b) their timing, and (c) realistic options and constraints. A paper describing decision maps for maize production in Pergamino has recently been published (Bert et al. 2006).

As an example, we present a map of the decision about the farm area allocated to maize (Figure 2). Some land allocation decisions are initially made in early March, well before sowing of any crop. Nevertheless, these decisions are often revised in response to various factors up to the period preceding planting of summer crops (September–November). How many hectares are sown under the various crops is determined, to a great extent, by the expected gross margin of alternative commodities and by farm (soil moisture at sowing and/or harvest, soil quality, in-farm laborconstraints, etc.) and farmer characteristics (production objectives, economical andfinancial situation, etc.

Information about climate conditions that might affect crop yields plays an important role in the land allocation decision (Figure 2). Expected climate during September (when maize sowing starts), December (maize flowering) and March (maize harvest) may influence the land allocation to maize. For example, farmers may increase maize area if rain during maize flowering, a critical crop period, is expected to be higher than normal, which would allow high yields. This decision also depends on the farmer's perception of the influence of the expected climate on yield (risk aversion). Moreover, this decision may be revised in response to actual weather conditions at sowing (i.e., during September). If September is too rainy, depending on soil attributes (e.g., soil drainage capacity) that determine soil condition at sowing (e.g., soil excessively wet at sowing), a farmer may decide to maintain the amount of land assigned to maize and change production techniques (i.e., sowing is delayed) or, alternatively, if inadequate conditions persist after the end of October, the maize area may be reallocated among other crops such as soybeans. This illustrates the dynamic, iterative nature of decisions, which are subject to continuous revisions in response to updated information.

We are currently completing decision map for full-cycle soybean and wheat-soybean production systems in the Pergamino region. These maps were validated in close interaction with agronomic technical advisors from the Pergamino region. We used questionnaires, a decision exercise and discussions with a group of experts of the region to describe decisions related to soybean production, and the range of viable responses to a predicted climate scenario. Crop model simulations were performed to evaluate the outcomes of the experts’ stated responses to climate forecasts. In early 2006, the decision maps will be produced for the second target area of the project (Pilar, Córdoba) a much more climatically marginal location, and one in which producers have a much shorter history of cropping: most of them entered agricultural production in the last decade or so, and come from industry or business, which gives them an outlook different to more traditional farmers in Pergamino.

Our results show that climate is a factor influencing some of the decisions involved in soybean production, and therefore there are potential entry points to climate information. Decisions about sowing date and genotype selection were identified as the most climate-sensitive. Information about expected precipitation during critical crop stages is considered as very relevant for these decisions. Furthermore, experts are aware of the benefits of changing sowing date and genotypes in response to climate forecasts and, at least in the decision experiment, they seemed willing to adopt changes.

Two key lessons derived from this activity confirm and reinforce previous results in studies of climate forecast applications:

First, expected climate is just one of the factors affecting soybean production decisions, and its relative importance depends on some of the other factors that may reduce or eliminate the ability to respond to climate forecasts. Our results highlight the need to consider the full context influencing a decision.

Second, changes in management made by experts in response to hypothetical climate scenarios resulted in a broad range of results (both positive and negative). We hypothesized that, on one hand, the difference in results may be due to a poor understanding of the regional climatic impacts of predicted scenarios (e.g., a given ENSO phase). On the other hand, it is also possible that advisors have difficulties in anticipating the agronomic outcomes of changing management decisions in response to a climate forecast. Further experimentation is needed to elucidate which of these hypotheses is more likely. In summary, our results suggest that mere dissemination of a climate forecast is not likely to produce useful outcomes. Instead, an iterative process of mutual learning involving information producers and users is needed.