The spectral reflectance of leaves is influenced by both the environment and relatedness between species since it is a product of the absorption of light by important structural and functional components of the leaf and other traits shared among closely related species. Through this work we illustrate that the spectral reflectance of leaves is an effective tool for inferring plant taxonomy, but the primary drivers of variability, including time of the year, need to be included in the spectral classification. We collected spectral reflectance, from the visible to the shortwave infrared (500 – 2400nm), of leaves from 42 woody tropical species belonging to six botanical families growing at the Enid Haupt Conservatory at the NYBG during the spring of 2019, summer of 2020, and winter of 2021. Although, these species are not growing in their natural environments, we know that the spectral reflectance varies with the seasons in the tropics as well. To classify leaves to species we ran a partial least square discriminant analysis (PLS-DA) for the spring leaves, and this yielded high accuracy (>90%) at the family and genus levels. Testing for the effect of time of collection in leaf reflectance in the PLS-DA classification, we first use the spring dataset to train the PLS-DA model and used the summer and winter datasets to test the model. In this approach we found that the classifications at the family, genus, and species levels were significantly lower in accuracy ( 90%). We also calculated the coefficient of variation (CV) to visualize the effects of the collection date on the spectral signature and showed that the areas of highest CV were in the visible and the shortwave infrared (1300-2400nm). We also ran a correlation analysis with functional traits: leaf pigment, water, nitrogen concentrations and leaf mass per area – estimated from spectra by inverting existing PLS models and a Bayesian inversion of the PROSPECT model to see how traits and spectra co-vary over the dates. With these functional traits we run a principal component analysis taking into consideration the phylogenetic relationships of species. We find that accuracy in calculating species richness and the functional space of plants can be maintained if spectral variability is factored in.