[Don Bertolette, March 12, 2009]

Ed-
I suppose it might vary from species to species, region to region, but in general it has to do with disturbance and a trees response.  The re-iterations you mention are most typical in conifers, as they typically exhibit 'apical dominance'. Which is to say, the tree wants to grow into the sunlight.  When some manner of disturbance causes damage to the apical tip, like a significant ice storm, a wind event that breaks the top out, and/or other such single tree calamaties.
The tree's response, somewhat dependent on genetic predisposition, often times is to route the trees energy to the next best branch.  I have seen a number of species do this several times (tree lives long enough, emerges high enough above the forest canopy, it gets taken back, sometimes, time after time).  I don't think I have a photo anymore, but there are a number of "candelabra trees" in the redwoods of northern California that may be the most extreme example of this, in part due to their longevity (number of opportunities for coastal exposure to Pacific storms), and in part due to their energy allocation strategies.
-Don  

Ed,

Here is an excerpt from the Tsuga Search Project that should help:

Wood distribution

In total, this project sampled 3,340.1 m (10,955 ft) of eastern hemlock
trunks (N=53) with a cumulative wood volume of 1,672.3 m3 (59,053 ft3)[1].
The modeled trees did not always conform to a simplistic, single trunk form
however, and required measuring some challenging structures (See
Reiterations and bifurcations below). In all trees sampled, "conventional"
main trunk wood composed anywhere from 63.7% to 100% of total tree volume.
Bifurcations (forks of vertical leaders) accounted for a maximum of 36.3% of
total wood volume, considerably more than a maximum of 10.05% for
reiterations (limbs upturned into secondary trunks). Bifurcations were much
larger; even though a total of 633.1 m (2,076.6 ft) of reiterations were
measured, total bifurcation volume (42.9 m3; 1514.2 ft3) was 240% more with
only two-thirds of the cumulative path length. Path lengths (length of all
measured wood structures) of individual trees ranged from simply the length
of the trunk to a maximum of 200.7 m (658.4 ft). See Appendix 4: Wood
distribution of subject trees for more detail.

Reiterations and bifurcations

Reiterations are replicated trunks formed by the tree to exploit a light
gap. Structure and growth of reiterations are markedly different from normal
descendent or ascendant branches. Reiterations are included in trunk volume
calculations as they are part of permanent, branch bearing structures of the
tree. Their origin arises from the sprouting of normal branches and can be
initiated by a light gap from the side or from crown damage above. The
apically dominant portion of these structures are often much younger and
model-conforming as in a young tree. The cross-section changes to a more
oval shape as reaction wood is built up to support the added weight from
increased growth and sap demands as the reiteration gains vigor. The
buttressing allows a more permanent attachment than a conventional branch
due to support on the lower portion being less likely to fail from injury to
the upper surface. This also allows wounds with associated decay to persist
and form a suitable substrate for vascular epiphytes (Sillett, Van Pelt
2007)

Studies in redwood forests indicate that structural diversity in the canopy
is strongly linked with aerial biological diversity. Reiterations generate
niches not found anywhere else in the forest (Sillett, Van Pelt 2007). Our
observations indicated moss mats, epiphytic plants and lichen gardens to be
associated with these large structures. Large reiterations and breaks in the
trunks were hosts to shade intolerant shrubs and wind disseminated trees
such as Rubus and Betula. Complex, reiterated crowns could be refugia for
those species lying in wait for a light gap or breach in the rhododendron.

Of the thirty trees discussed in this project, eleven trees had large
reiterations. Four of these were tall trees and seven were large trees (two
trees are replicated). A total of thirty-seven reiterations were measured,
ten of which grew on the Usis Hemlock, the most on a single tree in the
project. Total reiterated trunk length measured on the Usis Hemlock totaled
over 147.8 m (485 ft) for a tree total of 200.7 m (658.5 ft) of trunk.
Likewise, the Cheoah Hemlock had a total of 185.6 m (608.7 ft) of measured
trunk length in reiterations and bifurcations. These structures are likely
to greatly increase the crown area of the tree, and bolster wood production.

Reiterations on the subject trees (no repeated trees) totaled 14.7 m3 (518.8
ft3) in volume, 527.0 m (1,728.6 ft) in cumulative length, and were found to
compose anywhere from 0.5% to 10% of total tree volume, averaging 4.2% for
all trees. Volume of individual reiterations reached a maximum of 1.78 m3
(62.9 ft3) and a maximum structure length of 18.96 m (62.2 ft). On average,
the reiterations were 31% larger and 12% longer on the large subject trees
than on the tall subjects. Fusions between reiterations were documented in
two trees, forming a strong interlocked crown with braced bifurcations that
likely reduce mechanical failure of non-conventional crowns. "Transplanted"
branches were noted in the Cheoah Hemlock. This occurs when a grafted branch
severs its connection from its origin. This phenomenon is well documented in
the coast redwoods of California, and is encountered in dense hemlock hedges
as well. The Cheoah Hemlock and the Usis Hemlock both exhibited reiteration
fusions; one of which started at 14.3 m (47 ft) and fused to another
reiteration 9.7 m (31.9 ft) higher in the tree. The Cheoah Hemlock had a
branch that crossed the inner canopy while grafting to three separate
structures.

Height of origin was noted for every measured reiteration. The lowest
occurred at 12.59 m (41.3 ft) and the highest at 38.35 m (125.8 ft) above
mid-slope. The highest recorded reiteration was over 25 cm (12 in) in
diameter and was encountered in the Usis Hemlock. This tree also contained
the highest cumulative and relative volume of reiterations; 4.36 m3 (154.3
ft----------3) and 10.05% respectively. The most massive reiteration was
found in the Cheoah Hemlock and scaled 1.78 m3 (62.9 ft3). This individual
structure alone contained a path length of 34.33 m (112.6 ft) of measured
trunks. A graphing of the origin heights below likely reveals a random
height of origin as the events initiating the formation of a reiteration
would likewise be random. Data gathered from a control population of
non-superlative trees are needed for comparison.

Of the thirty subject trees, three had large bifurcations that warranted
frame mapping along the section(s) of trunk fusion. These massive, fused
areas were up to 1.6 m (5.25 ft) across. Two other subject trees were
bifurcated but did not exhibit reiterations. Of the subject trees, only the
Cheoah Hemlock was both reiterated and bifurcated. This tree, supporting a
bifurcation more than 83 feet long, was also nearly entirely cloaked in live
crown from base to top. The longest and largest bifurcation occurred on the
Yonaguska Hemlock; 28.75 m (94.3 ft) long and 5.37m3 (189.6 ft3)
respectively. See example of a frame-mapped bifurcation fusion below:

Frame mapping example from the Yonaguska Hemlock bifurcation fusion

All the remaining trees in the project had a single trunk; one of which was
broken at 31.74 m (104.1 ft). This tree (the Headless Giant) - as well as
several others - had the vast majority of the crown originating on
reiterations with very few non-reiterated branches. Several reiterations
reiterated again (second order) with one fourth order reiteration noted in
the Usis Hemlock. This immense tree had a fusion of two giant reiterations
that supported a complex limb system supporting multiple reiterated sprouts
- some originating from the same limb like a row of small trees extending
upwards of ~12 m (~40 ft) from the trunk origin.

Will F. Blozan

President, Eastern Native Tree Society
President, Appalachian Arborists, Inc.