The fitness consequences of self-fertilization are largely determined by ho
w self-pollination occurs. Within-flower self-pollination (autogamy) may be
advantageous, since it can provide reproductive assurance without much see
d or pollen discounting. In contrast, between-flower self-pollination (geit
onogamy) provides no reproductive assurance and can cause severe seed and p
ollen discounting. I used floral emasculations with marker gene analysis to
estimate the components of self-fertilization ina tristylous, self-compati
ble, clonal, mass-flowering plant, Decodon verticillatus. This species prod
uces 30% of progeny through selfing. I assessed the contribution of autogam
y to selling by comparing pollen deposition, seed production, and the selfi
ng rate of flowers emasculated before anther dehiscence with intact flowers
. Emasculation had no effect on pollen deposition, caused a small increase
in seed production, and only reduced self-fertilization by 14%, suggesting
that most selfing occurs through geitonogamy. Geitonogamy can be due to sel
f-pollination between flowers on the same branch, different flowering branc
hes of the same plant, or different ramets of the same clonal genet. I asse
ssed the contribution of within-branch geitonogamy by emasculating all flow
ers on a branch, which reduced selfing from 0.29 +/- 0.03 (mean +/- 1 SE) t
o 0.19 +/- 0.04. Between-branch geitonogamy was estimated by comparing the
selfing rate of plants with only a single flowering branch (0.16 +/- 0.07)
to plants with multiple flowering branches (0.27 +/- 0.04). Selfing rates o
f individual branches also correlated positively with the daily number of f
lowers open on nonfocal branches of the same plant. Between-ramet geitonoga
my was suggested by significant self-fertilization by single-branch plants,
even when all flowers were emasculated (0.10 +/- 0.07). Selfing rates of i
ndividual branches also correlated negatively with measures of local clonal
diversity. Based on these results, autogamy accounts for only 18 +/- 14% o
f self-fertilization, with the remainder (82 +/- 17%) due to geitonogamy, w
hich occurs about equally through pollination within branches (31 +/- 22%),
between branches (38 +/- 32%), and between ramets (31 +/- 28%). Because se
lfing occurs mostly through geitonogamy and is associated with strong inbre
eding depression, it seems disadvantageous. Selfing in D. verticillutus has
probably not been selected directly, but is a by-product of self-compatibi
lity, large plant size, mass-flowering, and clonal propagation.