Non-linear, unimodal techniques are suitable for estimation of environmental properties in the northeast (NE) Pacific based on fossil diatom taxa (species and/or species groups) found in modern (core-top) sediments. Canonical Correspondence Analysis (CCA), a constrained ordination technique, discerned best-fit relationships among two multivariate datasets (the floral and the environmental) and thus yielded insight into the environmental variables that best explain the species variance within diatom populations. Based on these insights, we developed predictive functions for annual Primary Productivity (PP) and seasonal range of sea Sea-Surface Temperature using unimodal models and cross validation techniques. Estimates of annual PP (r2jack = 0.92; RMSEP = 91.94 gC/m2/y) explained the highest percentage of variance in the core-top diatom record (22.1%).