Hemlock Genetic Sampling   Edward Frank
  Jul 05, 2007 20:26 PDT 


Below is a draft document I am working on concerning a genetic sampling
strategy for hemlocks. I would appreciate and corrections, comments, or
elaborations on the subject that could be included in the final
document. These are my ideas....

Some Considerations Concerning a Genetic Sampling Strategy for Eastern
Hemlock (Tsuga canadiensis) and can be applied to Caroline Hemlock as

I have been considering what would be the best way to sample hemlock
populations in order to collect the broadest range of genetic diversity
within the sample set. This will be a general strategy that can be used
when other species are threatened.

What kind of genetic diversity exists in the hemlock population?
Hemlock did undergo a precipitous decline across most of its range
roughly 4500 or so years ago which has generally been attributed to an
insect defoliator. There are many papers that reference this in pollen
records. Other papers have suggested the cause of the decline to be a
widespread drought. Whatever the cause of the decline, unless the
population was reduced to a very small number, to a very limit, or the
hemlock were selectively killed because of some genetic characteristic,
there should have been little shift in the overall diversity of the
species. A drought would affect all populations except those in wet
areas, so it would not selectively remove any specific characteristic.
Likewise an insect defoliator would not selectively kill trees based
upon a specific characteristic, but would opportunistically attack
whatever trees were infected. If it were an insect defoliator, the
surviving trees may have some genetic characteristics that helped them
better survive the infestation and these would be increased in the
overall genetic pool for the species.

Studies of the genetic characteristics of eastern hemlock are few. Dr.
Cathy Zabinski, now at the Big Sky Institute at Montana State
University in Bozeman, studied genetic diversity in hemlock throughout
its range for her Ph.D. project at the University of Minnesota. She
found it to be very narrow like red pine (based on number of alleles
found on foliar samples, published in a paper in Canadian Journal of
Forest Research in December 1992).   Models she ran population genetic
models of the hemlock decline around 5000 ybp and showed that it
probably was not a small enough bottleneck to cause the observed low
levels of diversity. Bottelnecks in trees have to be incredibly small to
change population genetic structure. Hemlock looper parts are found in
the sediments around the time hemlock declined and almost disappeared
simultaneously throughout its range. It is common for a native insect to
cause massive mortality at the time of climate change or an unusual
climate event like an extreme drought, and for recovery to take
several centuries after such events.

Dr. Lee Frelich, University of Minnesota, and Director for the Center
for Hardwood Ecology, goes on to comment: “Hemlock had a glacial refuge
around 20,000 ybp in the Carolinas and spread north from there to New
England and then west to the Lake States, reaching the Porcupine
Mountains about 3000 ybp. I am not sure we know where the species
originated, which would have been millions of years ago with a much less
complete fossil record.”

“It is possible that the species never had a very diverse gene pool. It
depends on how the speciation event occurred. If it was a founder event
where a few trees were separated from the rest of the range of some
other hemlock species, and then underwent selection to adapt to the
environment which came to differ over time from that of the rest of the
species, plus random drift, and that event did not occur long ago, then
the species might have a very narrow genetic base.”

”On the other hand if a species with a large range with millions of
individuals in diverse habitats was split in half by the appearance of a
mountain range, or prairie, or other unsuitable environment, and
selection took the two large pieces of the population in different
directions, it may have a diverse genetic base right from the start of
the species. Both have been observed in the fossil record and in modern
speciation events for various plant species.”

Concerning the origin of hemlock as a species of the southern
Appalachians; an excerpt from E. Lucy Braun’s “Deciduous Forests of
Eastern North America”, published 1950, taken from a 1974 reprint
edition, page 45, chapter 4, The Mixed Mesophytic Forest Region:

   “The category to which hemlock belongs is open to question. Because
it is a dominant in the Hemlock-White Pine-Northern Hardwoods region, it
is usually thought of as northern, although it does not extend far into
the adjacent boreal forest. It is just as conspicuous and abundant in
much of the lower elevation Appalachian forest as it is in the north. To
explain its occurrence in the mixed mesophytic forest by Pleistocene
migration from the north is to overlook the general mixed character of
the Tertiary forest, which by zonal segregation gave rise to the late
Pliocene and Pleistocene banding of formations we recognize today. If
hemlock was not part of the undifferentiated forest of the Tertiary, of
which the mixed mesophytic forest is a persisting remnant, how is the
distribution of its root parasite, Buckleya distichophylla, to be
explained? This rare endemic of the Southern Appalachians (p. 481) has,
as have many other Tertiary genera, two close relations in Asia. Boreal
species do not display this eastern America-eastern Asia relationships.
This feature and the contemporaneous arrival of hemlock and mesophytic
hardwood species in post-glacial northward migration, as shown by pollen
records (p. 465), point so strongly to the long indigenous nature of
hemlock in the area of the Mixed Mesophytic Forest climax that it seems
best to omit mention of it as a “northern” species. If the entire flora
of the Mixed Mesophytic Forest region be considered, developmental and
edaphic communities as well as climax communities, it is apparent that
southern plants far outnumber northern and that the endemic element is
prominent. However, the floristics of the several climaxes is a separate

Dr. David Orwig, or Harvard Forest writes: concerning the topics of
hemlock genetics and the historical hemlock decline. First, in terms of
genetics, there has been very little field testing of hemlock genetic
variation. A study by Zabinski (1992) examined isozyme variation in
disjunct populations of eastern hemlock and found a very low level of
genetic variation. Another study by Schaberg and others (2003) examined
rare alleles in eastern hemlock associated with silvicultural treatment
and results suggest that rare alleles decreased when trees were
selectively cut and increased when diamter-limit cutting was used. I am
no expert when it comes to hemlock and genetics, but a few other related
articles are: Wang et al. 1997 examined chloroplast DNA polymorhpisms in
eastern and Carolina hemlocks and suggested that there was little
evidence of chloroplast DNA differentiation among eastern hemlock
populations. There is current research being done on determining the
genetic structure of hemlock populations by examining the microsattelite
markers by scientists at the USGS Center in Wellsboro PA.

In terms of hybridization between eastern and Carolina hemlock, it
rarely occurs. the only evidence I have is a study in the 2002 HWA
proceedings (Bentz et al. 2002), that a single hybrid of eastern x
Carolina hemlock was identified in a larger breeding program aimed at
resistance to HWA. Lastly, much paleo work and studies have been written
on the historical decline of hemlock across much of its range ~ 5000
years ago. Certainly several factors have been implicated and the
consensus seems to be a climatic trigger that may also have lead to
defoliation of hemlock by a pest similar to our native hemlock looper. A
paper just came out discussing this (Foster et al. 2006).

Zabinski, C. 1992. Isozyme variation in eastern hemlock. Can. J. For.
Res. 22: 1838-1842.

Scaberg, P.G.,G.J. Hawley, D.H. DeHayes, and S.E. Niensohn. 2003.
Silivicultural management and the manipulation of rare alleles, pp67-74.
IN proceedings of the symposium of the North American forest commission,
Forest genetic resources and silviculture working groups, and the
International Union of Forest research organizations (IUFRO).

Wang C. M. H. Perlin, R.R. van Stockum, Jr., S.H. Hamilton, and D.B.
Wagner. 1997. Chloroplast DNA polymorphisms in Tsuga canadensis and
Tsuga caroliniana. Can J. For. Res. 27: 1329-1335.

Bentz, S.E., L.G.H> Riedel, M.R. Pooler, and A.M. Townsend. 2002.
Breeding hemlocks for resistance to the hemlock woolly adelgid. pp.
127-128. in HWA proceedings from the 2002 meeting.

Foster, D.R., W.W. Oswald, E.K. Faison,E.D. Doughty, and E.C.S. Hansen.
2006.   A climatic driver for abrubt mid-Holocene vegetation dynamics
and the hemlock decline in New England. Ecology 87: 2959-2966.


The first step in devising a sampling plan is to research what is know
about the origin and genetic diversity of the species. In the case of
eastern Hemlock, this information was of little help in devising a
sampling plan. The next consideration is determining in a general way
what factors might affect the genetic diversity of hemlocks as
represented in different populations. The goal is to: 1) locate
populations in which the less common genes are concentrated so that they
will be included in a reasonably large sampling; 2) to identify
populations that may have a unique genetic characteristic not present in
the general population, and 3) to obtain samples of unusual meshes of
genetic characteristics that differs from those of the general

A primary characteristic of the genetic of the general
population is homogeneity. Most trees in this population would contain
almost all of the more common genetic traits, with an occasional tree
bearing one of the more unusual genes. A random sample of this
population is likely to not include many of the less common genetic
materials. This common or typical material should be sampled, but
should not make up the entirety of the collection.

There are a number of mechanisms that that in general tend to affect the
genetic diversity of species of plants and animals. These include: 1)
Geographic Isolation or Physical barriers to migration or interbreeding;
2) Chemical Specializations, and 3) Climatic Extremes, 4) Latitudinal

Geographical isolation is a powerful mechanism in the process of
speciation and can enhance the genetic variability between the isolated
population and the general population. An uncommon gene in the general
population may be physically expressed only rarely. In most cases its
contribution to the overall gene pool is lost in the wash of the more
common genetic populations. In an isolated population, if an unusual
genetic trait is present, it will be given a greater weight in the
overall gene pool. And if this trait is somehow advantageous, the
number of trees in the population carrying this trait will increase over
time. It is very difficult to shift the genetic balance of a large
population and this process is very slow, while an area with an isolated
smaller population will shift rapidly to whatever characteristics work
best for that particular environment. An example of this are the two
populations of fox squirrels on either side of the Grand Canyon. When
the plateau was continuous the entire population freely interbred. With
the formation of the Grand Canyon the two populations diverged, each
concentrating a different characteristics gene set, so now they are two
distinctly different populations with different physical
characteristics. (There is some argument about whether they have
speciated or not.)   

Other types of barriers are filter barriers – those are where the
specific conditions, environmental or other, only allow individuals with
specific characteristics to cross the barrier. For example if there was
a dryer stretch of territory between two wetter areas, only those
species or individuals that were able to tolerate the drier climate of
the intervening area would be able to migrate past this filter barrier.
Therefore populations that established themselves on the far side of the
barrier would have a different genetic character than the populations on
the initial side. This may become a more important process as global
warming forces the northward migration of many tree species.

Distance itself is a form of a filter barrier. The pollen or seeds of a
particular plant only travel so far, so the area that is directly
influenced by a particular individual plant is limited. If there is a
population of tree that spreads across a thousand miles, it is clear
that the trees at the far end are not freely interbreeding with trees at
the near end. There often is a gradual change in the characteristics of
a particular population over distance. The populations at either
extreme may be quite different from each other in physical
characteristics and genetic characteristics.

The second category are populations that have differentiated because of
their ability to grow in an environment that is chemically different
from the rest of the overall population. I am not sure if this is
applicable to hemlocks or not. There are a number of areas called
barrens because of the “barrenness” of the land surface where only a
few species of plants have been able to adapt to these chemically harsh
settings. Examples include some of the serpentine barrens, grown on
bedrock primarily derived from serpentine, gypsum outcrops that are only
populated by a few gypsophile plants, there are shale barrens that are
barren because of the weathering mechanism of the bedrock, and pine
barrens, populated mostly by pines. Are there examples of hemlocks that
have been able to adapt to some of these unusual chemical environments?

The third category of mechanisms are those areas with harsh climates.
Examples might include some of the stunted forests atop Appalachian
mountains, populations in the edge of swamps, populations in extensive
scree slopes, populations, often stunted, by dry soils. To the extent
that hemlocks are found in these extreme areas, these likely represent a
concentration of less common genes or other genetic variations.

The fourth category is latitudinal variations. Populations tend to vary
in characteristics from north to south. The trees from southern
climates tend to be more heat tolerant, while those from the north tend
to be more cold tolerant. In deciduous species there are differing
times when the trees flower, leaf out, and drop their leaves that vary
with latitude. That is one reason why it is not good to plant a tree
from too many degree latitude north or south of the planting site.
Southern trees in a northern area will flower or leaf out before the
last frost and be damaged, they may not change color or drop their
leaves before they are killed by the frosts of autumn. Trees from the
north planted southward may not tolerate the heat well. So clearly
there is a genetic variation that is latitude dependant. Latitudinal
variations and heat tolerance will become a more important issue as
global warming occurs. The question is whether or not the remaining
populations can migrate north as fast as the climate warms

I am sure there are other mechanisms that can be identified, but this
gives some basis for developing a sampling strategy.   

One group of populations that need to be sampled are those found in
various outliers and refugia around the edges of the current population
distribution. In the case of hemlock there are many places to consider.
In the southern boundary several outlier populations are defined in the
article: "Disjunct eastern hemlock (Tsuga canadensis) stands at its
southern range boundary" by Justin L. Hart and David Shankman, in
Journal of the Torrey Botanical Society 132(4), 2005, pp. 602–612. The
article deals primarily with Alabama, but some other outliers are
marked. In North Carolina is an outlier east of the general population
range at Hemlock Bluffs Nature Reserve in Cary, NC. In Virginia there
was a population in the southeastern portion of the state, but the
forestry people there report that it was cut. It would be worthwhile to
see if there is any regrowth from this site. There are a number of
smaller pockets in central Ontario and other Canadian Provinces that
mark the northward boundaries of the hemlock populations. In Indiana
there are several outlier pockets. One of these is at Hemlock Cliffs
area of Hoosier National Forest. In Minnesota there are a couple of
pockets that mark the westward extent of the hemlock populations. The
largest of these is at Jay Cooke State Park and the Hemlock Ravines
Natural Area. These should be sampled. Brandon Gallagher Watson a certified
Arborist from Minnesota reported several large (25-40" DBH) E. Hemlocks
in Minneapolis' Theo Wirth Park. It is unclear if these are naturally
occurring or one planted historically. They could be sampled. There
may be other locations of isolated or disjunct populations that should
be sampled.

I have not identified any, but any populations adapted to unusual
chemical environments should be sampled, as well as any adapted to
extreme climatic conditions. Trees from various different latitudes
north to south should be sampled to obtain the range of variation from
north to south. Likewise samples from the eastern, central and western
extent of the range should be sampled because of the filter barrier
nature of distance. It was suggested in another forum that perhaps
samples should be taken from areas whre there is a mixed hemlock and
deciduous forest, rather than focusing on just hemlock dominated sites.
I don’t really see that the mixed forest hemlocks should be any
different genetically from those in the hemlock dominated forests. I am
not opposed to sampling them, but do not believe they will add anything
new to the overall genetic sampling process.

Sampling should focus on older trees and any trees that express a
distinct physical difference from those in the general hemlock
population.   When sampling an old tree, you know what you are getting
in terms of size and longevity. When sampling a younger tree you are
buying a pig in a poke. You do not know if this tree is going to be a
viable long term resident of that forest, whether its genes are “good”
or not. You may be getting a genetic sample just an average tree rather
than an extraordinary tree. Most of the samples collected will consist
primarily of the most common genetic materials, so I don’t see any
reason to specifically sample for those traits. The focus should be on
the exceptional or unusual traits as much as possible.


What should be collected?

There are several ways to collect genetic material. One is to collect
seeds from cones and save them., another would be to collect small
sprouts and replant them somewhere else where they could be monitored
and treated fro infestations, material could be collected fro cloning,
and a final option to consider is to not collect the material at all,
but to treat the trees where they grow with chemicals, such as
imidacloprid, so they are not killed by the adelgid and will continue to
grow and produce seed for as long as they live.

Camcore http://www.camcore.org/projects/hemlock.php is actively
collecting seeds from Carolina hemlock to help preserve the genetics of
the species. Seeds are being collected from various ample sites and
planted, in one of several locations. Details on their efforts are
provided at the website link above.

Clearly the Carolina hemlock is critically threatened with effective
extinction in its native range if something is not done. I applaud the
efforts of Camcore as it is one approach that can be taken. There is
debate about whether the money would be better spent chemically treating
groves of the species, rather than the expensive process of collecting
seeds and growing the trees in a nursery. I question that also, but any
effort that may help preserve the species is a plus from my standpoint.
The methodology used by this group for the collection and growth of
hemlocks from seeds would be a model if that was the option chosen to
preserve the genetic diversity of the species. I see no reason to
reinvent the procedures they have developed. I might however suggest a
independent assessment of where they sampled and try to identify other
potentially worthwhile sample sites they did not sample.

Trees could be transplanted as sprouts or samplings to a single
location, or series of locations. The trees in this nursery grove could
then be treated on a regular basis and monitored for HWA or other pests
that might kill them. This would allow trees to be replanted into the
wild once a viable permanent solution to the HWA was found to serve as
seed trees for future generations.    Multiple sites would be better as
it would prevent a single event, such as a fire from wiping out the
entire genetic repository.

Samples of living tissue could be frozen for potential cloning at some
future time. There generally is a limited time in which a tissue sample
will survive before it dies and is useless for cloning purposes. Recent
attempts to clone the Methusala Bristlecone pine failed because the
samples could not be processed in time to be successfully cloned. A
better methodology might allow you to flash freeze the samples in the
field using liquid nitrogen for future cloning. The disadvantage of
this methodology is the cost involved in keeping the specimens froze for
an extended period of time. The benefit would be that the specimens
could last for hundreds of years and still be viable to work with at
some future time. They would be in a laboratory setting and not subject
to natural disasters that could destroy other growing populations.

Even if samples are collected, I would strongly recommend that a
treatment program be set up with long term funding, an endowment for
example, to allow chemical treatment of at least a representative number
of the trees in any given population. These would then continue to
produce seeds for their lifetime and serve as seed trees for future
reestablishment of the hemlock populations.

Thank you,

Edward Frank

RE: Hemlock Genetic Sampling   Joshua Kelly
  Jul 05, 2007 22:06 PDT 

Ed, Ents:

Lucy Braun did not have the whole story on Buckleya disticophylla.

"If hemlock was not part of the undifferentiated forest of the Tertiary, of
which the mixed mesophytic forest is a persisting remnant, how is the
distribution of its root parasite, Buckleya distichophylla, to be

Buckleya has since been observed growing without hemlocks, perhaps
parasitizing oaks (like Pyrularia). Camp Branch Falls in the Nantahala Gorge
is one site where this has been documented (Kauffman, pers. communication).

As far as sites with unusual soil chemistry, there are hemlocks at the Buck
Creek Serpentine Barrens on the Nantahala NF. Those may have some
interesting physiology. There is at least one narrowly endemic plant at the

Re: Hemlock Genetic Sampling   Edward Frank
  Jul 05, 2007 22:50 PDT 


Thanks for the comments. I will make notes to revise the document.

Re: Hemlock Genetic Sampling   Edward Frank
  Jul 06, 2007 22:21 PDT 


I would like to add an addendum to my original post adding a couple of areas
worth sampling. Those of you who have been following Will Blozan's Tsuga
Search project in the Smokies should know that the tallest hemlocks and the
hemlocks with the greatest volume are found in the Park. The graph of the
tree heights Will has measured forms a beautiful curve, lacking any plateaus
or jumps that scream out something different. However Will has seen all of
these trees and has climbed many of them while I have not. Will having had
these up close and personal encounters with the trees feel there is
something different about them. Even if this difference can not be
articulated yet or measured, that is good enough for me to think they
warrant genetic sampling. A patch of trees of any species that is
significantly bigger than others of their species, may represent a genetic
variation. It is a question of nature or nurture. Are they genetically
different or are the growing conditions just that much better here for some
reason? Since we have only one chance to get it right, these trees should
be sampled before it is too late (It is already too late for most of the
biggest trees).

Along the same lines the oldest hemlock yet documented is a a remnant patch
of Old growth in the Tionesta Scenic and Research are in northern PA at 555
years when sampled in 1991. In fact two of the three oldest are found
there. Is there anything genetically different about them? Probably not,
but these specimens should be sampled just in case, before they too are lost
to the Hemlock Wooly Adelgid.

Ed Frank