Polyvariant ontogeny (PVO) is expressed visually in the life cycle graphs (LCGs) for Calamagrostis woodreeds as a variety of pathways for individual plants to develop through many of their states distinguishable by the ontogenetic stage and chronological age (in years). PVO is recognized as the basic mechanism of adaptation in local plant populations to their environments, while they find a quantitative measure of the adaptation via developing a matrix model of double-structured population, calibrating the matrix of vital rates from empirical data, and calculating its dominant eigenvalue λ1. This approach encounters an obstacle typical for rhizome grasses: while the rates of aging and ontogenetic transitions can be deterinined from field data mainly by the morphology of aboveground parts of the plant, the rates of vegetative propagation can be reliably determined only from digging up the belowground rhizome system, i.e., by destroying the sample plot ('reproductive uncertainty'). Therefore, the former (non-destroying) calibrations of matrix models were subjective to an extent, resulting in correspondingly subjective estimations. A novel method to overcome the 'reproductive uncertainty' makes use of the maximation hypothesis: the uncertain rates are such that λ1 attains its maximal possible value under the given conditions. To test the hypothesis, we have conducted a field experiment by a new technique with a model species, the woodreed Calamagrostis epigeios (L.). Roth, that reproduces vegetatively in a meadow phytocoenosis and a spruce forest clearance. Excavating the whole system of ramets with their rhizomes and analyzing the maternal-child links in the laboratory have provided for (in addition to the former data on the local population structures and ontogenetic transitions) a new kind of data to calculate the status-specific rates of reproduction. The novel method of calibration has enabled us to find an exact range of λ1 values, i.e., the true quantitative bounds of adaptation for a given local population. Obtained under the reproductive uncertainty and maximation hypothesis, the values of λ1 have turned out close to the upper bounds of the ranges, thus verifying the hypothesis. The experiment has discovered generative subsidiary plants sprouting from the rhizomes of maternal ramets without entering the virginal stage. As a result, the LCG enriches with new reproductive pathways, and the new (not yet published) situations emerge where λ1 fails in its accuracy as a measuring tool of comparative plant demography. We propose a general method to adjust the adaptation measure in this kind of situation.