The tropical palm species Iriartea deltoidea is an important resource in Amazonian Ecuador. Population models, based on short-term field measurements, have been used to analyze management scenarios for Iriartea sustainable harvesting. An existing matrix population model for Iriartea in secondary and mature forests could not adequately simulate the population dynamics from clearing to 30 years or more. In order to find a better model of the Iriartea population, we used inverse analysis with two optimization techniques (Golden Section search and genetic algorithm (GA)) to find fecundity and seedling demographic parameters that best matched the observed secondary forest (30 years after clearing) and mature forest size class distributions. Re-estimation of fecundity using the Golden Section optimization resulted in a good match to seedling numbers, but a poor fit for other size classes. Genetic algorithm optimization of fecundity and seedling survival and transition parameters found good fits from clearing to age 30 with an exponential or density-dependent model. In order to find a single model that described the population trajectory from clearing to mature forest, we used the genetic algorithm to best fit the observed size class distributions at age 30 and for the mature forest (assumed ages of 65-150 years). The genetic algorithm produced plausible density-dependent models for each of the assumed mature forest's ages. Simulated seedling demographic parameters 40–60 years after clearing were similar to field estimates from 30-year-old and mature forest populations.