The trimeric Fenna-Matthews-Olson (FMO) protein is a bacteriochlorophyll (BChl) a antenna complex, whose X-ray structure is now known at the atomic level for two green sulfur bacterial species (Prosthecochloris aestuarii and Chlorobium tepidum). Its steady-state Q(y) absorption spectrum at low temperature exhibits considerable structure, with at least eight bands attributable to well-defined (groups of) BChl a exciton levels. The low-temperature absorption difference spectra of Cb, tepidum trimers excited at 789 nm are correspondingly multifeatured, and they show rich spectral evolution due to femtosecond and picosecond electronic energy transfers. Global analyses of these time-dependent Delta A spectra lead to a phenomenological scenario for cascading and branching among exciton level groups responsible for specific steady-state absorption bands, The missing link in our understanding of structure-function causality in this protein stems from a lack of an ab initio theory for the effects of known protein environments on BChl a transition energies; this problem still limits our insights into the workings of spectrally heterogeneous antennas with multiple nonequivalent pigment sites. Optical anisotropy studies strongly suggest that FMO excitations are typically localized to the 7 BChl a pigments within one protein subunit, rather than delocalized over the whole trimer, This localization occurs because the resonance couplings between BChls belonging to different subunits (<20 cm(-1)) are several times smaller than the diagonal energy disorder (similar to 70 cm(-1)). Strong oscillations appear in the anisotropies (similar to 220 fs period) for pump-probe wavelengths that simultaneously overlap both of the exciton level groups responsible for the 825 and 815 nm bands in the a, spectrum, These are not vibational coherences, but arise from quantum beating between levels with nearly perpendicular transition moments.