Unlike nearly all animals, plants violate Weismann's doctrine for the separation of germline and soma. In animals, the germline - the cell lineage giving rise to gametes - is segregated early in development from the cell lineages which give rise to the soma, that is the rest of the body. These germline cells remain relatively isolated from the body for the remainder of the animal's life. Any mutations that occur outside of the germline cannot be inherited through gametes.

In plants, which have meristematic growth, this is not at all the case. When plants create gametophytes in the form of pollen grains and ovules, the gametes contained within are the descendents of meristematic cell lineages that have ultimately given rise not only to the gametophytes but also all of the airborne tissues of the plant. Thus gametes may be produced from cell lineages that are comparatively active and may have undergone dozens to hundreds to hundreds of thousands of potentially imperfect mitoses which may give rise to errors during DNA replication or genomic damage during interphase. If these somatic mutations persist within the cell ineage, they may be inherited through gametes.

Sources of de novo mutation load due to somatic mutation in a growing plant
Flower locations are indicated in red, and the slope of the line is equal to the somatic mutation rate Um.

I use a combination of fieldwork, labwork, theoretical modelling and meta-analysis to develop a general model explaining stature-based differences in evolutionary constraints on plant mating systems. The Φ model of plant evolution assumes that the per-generation mutation rate has two primary components. The first component is the rate of mutations occurring during meiosis (U_e), including errors during DNA replication, insertions/deletions due to unequal crossing-over, etc. The second component is the rate of mutations occurring during each mitotic division within a cell lineage (U_m). The total contribution of mitotic mutations is a positive function of Φ, the number of mitoses that occur in a plant's lifetime from zygote to gamete production. The association between the Greek letter Φ and mitosis is reflected in its resemblance to a dividing cell.

Some general predictions of the Φ model include: (1) correlations between Φ and any plant traits that depend upon mutation rate; (2) the importance of somatic mutation increases with plant size (and hence Φ); and (3) selection arising from somatic mutation strengthens with plant size.

With respect to mating systems evolution, the Φ model predicts that while small-statured (`low-Φ') plants such as herbs are free to have a mating system that includes selfing, large-statured (`high-Φ') plants such as trees have a per-generation mutation rate that is too high to allow for selfed progeny to reach reproductive maturity under nearly all natural conditions. This occurs because inbreeding depression, the reduced fitness of selfed vs. outcrossed progeny, is maintained at a higher level and is more resistent to being decreased via selection when the mutation rate is higher, as it is expected to be in large-statured plants. For further details, see Scofield and Schultz (2006).

Population genetic data supporting the Φ model of plant evolution
Data are adult inbreeding coefficient (F) vs. progeny selfing rate (S) measured in the same population of a variety of species of (a) small-statured, `low-Φ plants, and (b) large-statured, `high-Φ' plants. All data from published studies. Lines represent model fits as determined by several genetic models for the evolution of inbreeding depression. Figure from Scofield and Schultz (2006).

Because Φ is a critical component of the model, I've developed methods for estimating Φ itself. I started by estimating Φ in the tropical legumeDelonix regia using mature medial pith cells in twigs. For further details see Scofield (2006).

Distribution of medial pith cell sizes (Φ per meter) in Delonix regia

Based on predictions of the Φ model, I developed two techniques for estimating somatic mutation parameters in large-statured plants: the autogamy depression test (first proposed by Klekowski in his 1988 book), which relies upon fitness differences between selfed progeny created from gametes belonging to different cell lineages within the same tree; and the flower position test, which relies upon fitness differences occurring at different flower positions within the same cell lineage. From these fitness differences, and the variance in fitness differences among experimental trees and branches, the rate and selection and dominance coefficients of somatic mutations may be estimated. For further details, see Schultz and Scofield (2009).

Estimating somatic mutation parameters via the autogamy depression test...
Autogamous (within-flower) selfed progeny are created via pollination of a flower with its own pollen, while geitonogamous (between-flower) selfed progeny are created via pollination of a flower with pollen from a different flower borne on a different primary branch. If somatic mutation is non-negligible, then fitness of progeny resulting from autogamy (WA) will be lower than fitness of progeny resulting from geitonogamy (WG) with the fitness difference increasing with the number of cell divisions separating the cell lineages involved in geitonogamous pollination (Φi). This is expected because a higher number of de novo somatic mutations are shared by the ovules and pollen grains from the same flower than ovules and pollen grains from different branches within the same tree.
... and the flower position test
Autogamous progeny are created within flowers at two positions within the same cell lineage. These positions have an ancestor-descendant relationship that is either indirect (left) or direct (right). If the effects of somatic mutation are not negligible, then the difference in fitness between flowers at basal (Wai) and apical (Waj) positions within the same cell lineage is expected to be a function of the number of cell divisions (Δ) separating the basal and apical positions. This fitness difference is expected because flowers at more apical positions within a cell lineage have had a greater opportunity to accumulate somatic mutations than have flowers at more basal positions.