Vo. Sadras et Sp. Milroy, SOIL-WATER THRESHOLDS FOR THE RESPONSES OF LEAF EXPANSION AND GAS-EXCHANGE - A REVIEW, Field crops research, 47(2-3), 1996, pp. 253-266
Typical responses of leaf expansion and gas exchange rate to plant ava
ilable soil water (PAW) can be described with two straight Lines that
intersect at PAW,, i.e., the PAW threshold for which the rate of the p
rocess in stressed plants starts to diverge from a reference value. PA
W, is a parameter widely used in modelling plant responses to water de
ficits. It also reflects some apparent physiological mechanism because
plants appear to be able to sense soil water status or related variab
le(s). in this paper comparisons are made between PAW(t) for various s
pecies (monocots and dicots) and plant processes (leaf expansion and g
as exchange) in order to: (i) point out methodological sources of vari
ation in published values of PAW(t); and (ii) analyse variations in PA
W, in relation to plant and environmental factors. Reported values of
PAW(t) vary over almost the whole possible range of PAW (i.e., 0 to 1)
. Average thresholds reflect the greater responsiveness to water defic
its of tissue expansion (average PAW(t) = 0.56) relative to gas exchan
ge (0.40). Average PAW(t) for leaf water potential (0.61) and stomatal
conductance (0.37) are very close to the average thresholds for expan
sion and gas exchange, respectively. Soil water thresholds for leaf ex
pansion are also shown to discriminate between plant types (0.50 for m
onocots vs. 0.66 for dicots) and soils (0.72 for coarse- vs. 0.43 for
fine-textured soils). The simplicity of characterising plant responses
to water stress in terms of PAW is attractive. In agreement with know
n physiological relationships, however, our analysis highlights how, f
or given processes and species, the measured value of PAW, can be affe
cted by evaporative demand, root distribution, soil texture and soil b
ulk density, among other factors, thus making explicit some of the ass
umptions underlying the use of fixed soil-water thresholds in simulati
on models.