E. Limitations of Using to Select for TE
Carbon partitioning and would not be expected to be stable across all environments and with changes
in plant hormonal balance. For example, cytokinins and abscisic acid (ABA) can affect both leaf gas ex-
change and carbon allocation [106]. Also, there are some problems of assessment of TE through carbon
isotope estimations: (1) it is a ratio and not directly correlated with yield or productivity, (2) the small
sample size may introduce subsampling errors and careful grinding is required, and (3) the technique re-
quires considerable capital investment in equipment and technical expertise [106].
Also, there are a number of potential sources of nongenetic variability in the measurement of .
Some can be readily overcome by technical or sampling precautions, as they are associated with the com-
position of plant dry matter [143] and the size and storage of the dry matter sample used in the measure-
ment [38]. Other sources of variation in among plant organs result from temporal variation in the
growth environment. Increased salinity [144,145], decreased soil water availability [29,66,108], soil com-
paction [136], and a decrease in vpd [146] could all result in lower values of .
Genotypic variation for measured under field conditions could be complicated by inherent differ-
ences in root growth [137]. This would affect the degree of dehydration postponement that could allow
prolonged maintenance of relatively large gs, thus decreasing TE but increasing growth and yield. Posi-
tive correlations between root length density and have been reported in crops such as beans [137,147]
(Fig. 4), and thus selection for low (high TE) may lead to selection of genotypes with poor root at-
tributes, such as shallow rooting and low root densities. Bean genotypes that had a deeper root system had
highvalues compared with the shallow-rooted genotypes [137]. Thus, leaf physiology (as measured by
846 SUBBARAO AND JOHANSENFigure 4 Relationship between leaf carbon isotope discrimination and root length density for rain-fed bean
genotypes at two locations: Palmira (upper graph) and Quilichao (lower graph). (From Ref. 137.)