During two seasons, ABA concentrations were monitored in roots, leaves and xylem sap of field‐grown maize. The water status of soil and plant was also measured. Plants were grown on plots with compacted or non‐compacted soil, which were irrigated or remained unwatered. ABA concentration in the xylem sap before dawn and in the roots increases 25‐fold and five‐fold, respectively, as the soil dried, with a close correlation with the soil water status, but with no clear effect of the soil structure. In contrast to the results of several laboratory experiments, no appreciable increase in xylem [ABA] and reduction in stomatal conductance were observed with dehydration of the part of the root system located in soil upper layers. These responses only occurred when the water reserve of the whole soil profile was close to depletion and the transpiration declined. Xylem [ABA] measured during the day was appreciably higher in the compacted treatment than in non‐compacted treatment, unlike that measured before dawn. Since a mechanical message is unlikely to undergo such day‐night alterations, we suggest that this was due to a faster decrease in root water potential and water flux in the compacted treatment, linked to the root spatial arrangement. These results raise the possibility that ABA concentration in the xylem sap could be controlled by two coexisting mechanisms: (1) the rate of ABA synthesis in the roots linked to the soil or root water status, as shown in laboratory experiments; (2) the dilution of ABA in the water flow from roots, which could be an overriding mechanism in field conditions. This second mechanism would allow the plant to sense the water flux through the root system.

Abstract. Stomatal conductance, leaf water potential, soil water potential and concentration of abscisic acid (ABA) in the xylem sap were measured on maize plants growing in the field, in two treatments with contrasting soil structures. Soil compaction affected the stomatal conductance, but this effect was no longer observed if the soil water potential was increased by irrigation. Differences in leaf water potential did not account for the differences in conductance between treatments. Conversely, the relationship between stomatal conductance and concentration of ABA in the xylem sap was consistent during the experiment. The proposed interpretation is that stomatal conductance was controlled by the root water potential via an ABA message. Control of the stomatal conductance by the leaf water potential or by an effect of mechanical stress on the roots is unlikely.

The ability of water balance models based on the concept of Transpirable Soil Water to predict the occurrence of water stress and the need for irrigation was tested for several environmental conditions of the root system, to determine in which conditions errors are likely to be appreciable. The response of evapotranspiration, stomatal conductance and leaf water potential to soil water reserve was studied under three conditions: (i) in pots with maize plants, (ii) in the field with deep soil and the root system placed in favourable conditions, with wheat during a dry year and with maize during four years with contrasting climate, (iii) in the field, with soil compaction which disturbed the maize root system, decreasing its efficiency for water uptake, during four years. (i) In the pot experiment, where the volume of the Transpirable Soil Water (TSW) is well defined, the responses followed the hypothesis of water balance models. (ii) The soil depletion was higher than the calculated TSW during two dry years in the field, because of an appreciable contribution of the non-rooted soil layers to the water balance. As a consequence, evapotranspiration, stomatal conductance and predawn water potential did not decrease over the whole range of soil water reserve. Grain yield was no lower in those years than in the wet years, in spite of the fact that the soil water reserve was depleted. Thus, a water balance based on the TSW would have underestimated in these conditions the ability of plants to withdraw soil water, overestimating the necessity of irrigation. Predawn water potential gave, on the contrary, indications consistent with the responses of the stomatal conductance and the net CO2 assimilation. (iii) The water uptake by plants would have been overestimated in the case of compacted soil. Stomatal conductance was low even for high levels of the soil water reserve, except if the densely rooted top 0.