Cytoplasmic male sterility (cms) is a maternally inherited inability to produce viable pollen. In Texas (T)-cytoplasm maize, the combined action of two restorer of fertility genes, rfl and rfla , can overcome the male sterile phenotype. Previous research established rf2a encodes a mitochondrial aldehyde dehydrogenase (ALDH) with broad substrate specificity. This broad substrate specificity complicates determining the molecular mechanism of rf2a-mediated fertility restoration. In response, a combination of transgenic, biochemical, reverse genetic, and genomic approaches were utilized. Three additional functional maize ALDH genes (rf2b, rf2c, and rf2d) with differing substrate specificities were identified. Each ALDH gene was characterized via in vitro kinetic enzyme analyses and individually tested for its ability to complement T-cytoplasm rf2a-mediated male sterility in planta. Heterologous complementation constructs containing the Rf2a and Rf2c coding regions complemented the male sterile phenotype. Comparison of the kinetic enzyme analyses between RF2A and RF2C identified four aldehyde candidates for being RF2A's physiologically significant substrate, two of which are structurally similar to intermediates of the phenylpropanoid biosynthetic pathway.;In male sterile plants, the tapetal cell layer of the anther degenerates prematurely during microsporogenesis. Spikelets house developing anthers in two compartments, the upper and lower florets. Each floret typically contains three anthers, which are identical at the gross anatomical level and proceed through same developmental stages. However, recent findings led Liu et al. (2001) to hypothesize that anthers in the upper and lower florets exhibit differences in gene expression. To test this hypothesis, a microarray containing 12,000 probes was used to examine processes that differ throughout anther development and to test for differences between anthers from the upper and lower florets. By acquiring data from three laser settings, the number of statistically significant differences detected was increased by 30 to 40% versus the common single scan strategy. Furthermore, approximately 9% of the probes exhibited statistically significant differences in gene expression between anthers in the upper and lower florets, which strongly suggests that the upper and lower florets are physiologically distinct. These experiments also lend support to the hypothesis that in normal anther development tapetal degeneration occurs during microsporogenesis via programmed cell death.