DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu




ŠUMARSKI LIST 1-2/1966 str. 90     <-- 90 -->        PDF

ADAPTING PROVENANCE TRIALS TOWARDS THE MOST
EFFICIENT SELECTION AND PRESERVATION OF DESIRABLE
FOREST POPULATIONS


(Contributed paper)
by


MACIEJ M. GIERTYCH


Institute of Dendrology and Körnik Arboretum, Polish Academy of Science, Poland.


Earlier provenance tests that have yielded so useful information, have
been laid out on fairly large plots, however, they were usually without or with
only few replications. In recent years it has become obvious that in order to
obtain a truly reliable comparison of provenances, replication in essential and
the number of replicates needed is considerably larger than was anticipated
earlier.


When introducing several replicates the experimenters had to reduce the
plot size in order to keep the experiment within manageable limits. A discussion
ensued as to the most efficient ratio between plot size and number of replicates.
Wright and Freeland (1958) have calculated that most information per tree is
obtained when dealing with plots containing only one tree each. The main
purpose of provenance trials is to compare the provenances and to select the
best. Evaluation of the productivity of a provenance may not be possible from
single trees. There are of course also other objections to the use of single
tree plots.


Earlier provenance tests have shown that it is not always possible to locate
the same stand from which the seed was collected for an experiment. In fact
the collection is either made on fellings, or the stand is felled in the period
between the seed collection and the time when the provenance tests have yielded
reliable results.


Thus the information obtained is at best an indication of the expected
value of the principal region in which the provenance was located.


The exact original population may not be available for the utilization of
the experimental results in practice. In this case the population can only be
reconstructed from the trees still remaining on the experimental site. These
trees will constitute an even better population than the original one, since by
means of thinnings it was further selected for the specific conditions of the
new locality on which its progeny is growing. Thus the general requirements
that have to be specified for the provenance trials now established are that
the trials must provide reliable information about the differences between the
provenances in respect of their economically important characters, and also that
they must provide enough trees per provenance at the end of the experiment,
to permit a reconstruction of the original population.




ŠUMARSKI LIST 1-2/1966 str. 91     <-- 91 -->        PDF

More specifically the requirements for provenance experiments can be
summarized under 8 headings:


1. The experiment must provide a reliable estimation of tree quality, growth
vigour, resistance to climatic and pathogenic agents, wood quality etc. for each
provenance separately, and irrespective of the site variability within the
experimental area.
2. Apart from the information about the quality and productivity of
individual trees within a provenance, the trials have to provide information
about the productivity of the stand per unit area, both in terms of wood volume
and money returns.
3. The experiment must permit a reliable comparison of the provenances
with respect to the characters mentioned in par. 1° and 2°. The lay-out must
therefore be such as to enable an elimination of environmental influences as
far as the variability within the experimental site is concerned.
4. In view of the fact that the growth and shape of a tree are affected
by its neighbours, it would be useful if each experimental plot had a surround
of trees that would permit an evaluation of the provenances irrespective of the
neighbour effect.
5. The experiment should be so conceived as to permit a comparison of
early and late results. Any juvenile-mature correlations established will permit
a reduction of the trial period for future studies.
6. In view of the high expense involved in the establishment of provenance
experiments they have to be so laid out as to provide maximum information
for the work performed.
7. As already stated, the provenance experiments must contain a sufficient
number of trees at a time when a reliable comparison of provenances is
available in order to permit a reconstruction of these populations which have
been selected for reproduction in forest practice.
8. Finally, the experimental stand must be established and tended according
to the forestry practices which are standard for the local conditions. It is not
realistic to rely on empirically chosen silvicultural innovations or to integrate
provenance studies with silvicultural experiments.
A brief glance at the eight requirements specified above is sufficient to
realize that no single experiment will fully satisfy them all.


The older experiments with large plots gave a good estimation of productivity
per unit area and permitted elimination of the boundary effects,
yet they were not replicated in sufficient numbers, and therefore the provenances
are not comparable. The very small or even single tree plots suggested
by Wright and Freeland (1958) do not permit an evaluation of productivity
per unit area; they are very complicated in the lay-out, which may lead to
mistakes during the beating up and later during the recording of the data,
and finally some plots will disappear as a result of thinning. The pilot tests
laid out in Waldsieversdorf (East Germany), where rows of 6 trees are used per
plot are useful only to the early part of the rotation, when some information
can already be obtained, but it will not be possible to confirm it later. Nor
will the productivity per unit area be calculable.


The situation would seem to be such that, on one hand, large plots are too
expensive and cover an area too large, thereby introducing an unnecessarily
large site variability that may obscure the genetic differences, wheres, on the




ŠUMARSKI LIST 1-2/1966 str. 92     <-- 92 -->        PDF

other hand, very small plots yield insufficient data on productivity and do not
permit a reconstruction of the population tested.


Obviously a compromise is needed. The requirements specified above place
such extreme demands on the experiment that it is absolutely necessary to
minimize them.


In order to achieve this minimization of requirements, six questions have
been asked, and it is the purpose of this paper to find answers to these
questions on the basis of information available in the literature.


Question 1. What is the minimum number of sample trees that will describe a stand?


Here recourse must be taken to the experience of the forest mensurationist.


According to Tischendorf (1927) for observations of equal weights the mean
value for these observations will change very little when the number of sample
trees is increased from 14 to 18.


A further increase of sample the number of trees will leave the arithmetic
mean almost unchanged.


An even more extreme point of view is advanced by Hummel (1955). He
has calculated that about 8 sample trees may lead to maximum errors of
between five and ten percent in the estimate of the total volume of a plot
when his own tariff method of volume estimation is used. To be on the safe
side, he recommends the use of 20 trees in calculating the tariff (volume basal
area line) for a stand. His method is good only for conifers and young hardwoods.
Its popularity with the British forestry practice is a measure for its
success.


The volume basal area line is a very good indicator of the productivity of
a stand. If the error in volume estimates is only 10!%, the error in height or in
diameter estimates should be considerably less.


For practical purposes smaller differences between provenance are of little
consequence. Thus it can be assumed that Hummel´s minimum size of a sample
(namely 8 trees) corresponds to the minimum number of trees that will give
a fairly reliable estimate of the value of a given provenance.


Both Tischendorf and Hummel refer to uniform stands. In the provenance
experiments the uniformity of site conditions can be assured, provided the
experimental area is not too large.


It can be assumed that the site conditions, age of the trees and the silvicultural
treatments will be more uniform in the experimental plots than is the
case in normal woods. Also the genetic variability within a provenance is likely
to be less than within a seed lot obtained from a commercial seed extraction
plant. Thus it is certain that the uniformity conditions specified by Tischendorf
and Hummel will be fulfilled. For this reason the adoption of the minimal
requirement of 8 trees would appear permissible.


Question 2. At what minimum age will reliable information on productivity per
unit area be available?


The standard volume tables for Central Europe prepared by Schwappach
(1943) start from the respective ages of 25 and 30 years in the better site classes
for pine and spruce. Thus presumably the volume per unit area for the younger
stands cannot be meaningfully estimated.


Tyszkiewicz (1961) claims that at the age of 15—25 years pine has already
passed the period of its mostTntensive growth in height. This would presumably




ŠUMARSKI LIST 1-2/1966 str. 93     <-- 93 -->        PDF

correspond to the time when the trees have reached a period of steady increment
in growth, and therefore the future performance is predictable.


From the juvenile/mature correlations established so far it appears that
at the time corresponding to 1/3 of the rotation age the quality and growth of
trees can be predicted for the rest of the rotation. This is the criterion used
in Sweden for the evaluation of the genetic value of different progenies
(Johnsson 1964).


Thus it would apear that not earlier than at the respective ages of about
30 and 35 years for pine and spruce it is possible to obtain information about
the quality and productivity of the different provenances.


Question 3. What is the minimum number of trees needed to reconstruct a population
for production purposes?


The most obvious way of reconstructing a population is by means of
establishing a seed orchard from the available trees. All seed orchards already
in existance are too young to give reliable information on the optimum number
of clones for commercial seed production. To provide sufficient genetic
variability in the resultant population the number of clones should not be
too small.


Andersson (1963) suggests that there should be 36—60 clones per
plantation.


Matthews (1964) considers 20 or 30 trees as an acceptable minimum
number of trees in a seed orchard composed of plus trees. These are only
estimates and they may be too optimistic. The more clones in a seed orchard,
the greater the variability in the resultant population.


Since seed orchards established from plus trees are likely to yield more
elite seed than those established from trees remaining on the experimental plots
of the best provenance, there must be some scope left for the thinning of clones
in the seed orchard.


Thus I would suggest that 40 is the minimum number of trees that have
to remain per provenance at the time of establishment of a seed orchard from
the provenance trial.


Question 4. What is the maximum forest area that can be assigned to a provenance
experiment?


The two basic criteria that have to be considered before answering such
a question are site uniformity and conformance with sound silvicultural
practices.


In Polish forestry regulations (Ministerstwo Lesnictwa 1961) it is specified


that clear fellings should cover no more than 6 ha in one belt with a maximum


width of 80 m.


Generally, compartments in Polish forestry are of the order of 25 ha.


These are divided into sub-compartments on the basis of site conditions. As far


as possible every difference in site requiring a different silvicultural treatment


or a different species composition is delimited into a sub-compartment. A


provenance experiment should be located within one sub-compartment, and it


should be rectangular in shape.


Thus it is not likely that more than 5 ha. could be found for one trial area,


but usually less.




ŠUMARSKI LIST 1-2/1966 str. 94     <-- 94 -->        PDF

Question 5. What is the minimum number of replicates needed?


Wright (1962) has studied the question of replicate number and comes up
with a table where he presents the number of replicates needed in order to
detect the differences of a certain magnitude when the average range in plot
means is known. The latter value can be calculated from comparing a few
randomly selected plots within a uniform forest population. For the purpose
of provenance studies it seems reasonable to specify that about 1/3 of the range
of variation in plot means for a uniform stand should be detectable. If the
range of variation is very great (low heritability), then the provenance is so
sensitive to environmental differences that its value for practical purposes is
low in view of the risk involved in matching it with the proper environmental
conditions. Small differences between provenances may be of theoretical interest,
but in practice such statistical accuracy is not called for.


If differences equal in magnitude to 1/3 of the range in plot means are
considered as the minimal differences that are of practical interest, then 7
replicates are sufficient to indicate a significance at 5´% level (Wright 1962).
Thus 7 is considered to be the minimal number of replicates.


Question 6. At what age is the neighbour effect severest?


Johnsson (1963) believes that the neighbour effect is relatively unimportant
in a plot of 36 or more trees since, coupled with randomization, there will be
all sorts of neighbours at the different sides of the plot and at different
replicates. He also points out that if a surround is being considered, one should
use the surround of the same progeny or provenance as the plot, since a
standard (hold-over tree) would suppress some plots and improve others.


Thus one could have a larger plot, and either consider the marginal trees
to be a surround, or else include them in the plot, depending on whether this
makes much of a difference.


It is likely that the neighbour effect is severest at the thicket stage when
competition between the trees is at its culminating point (about 15 years).


During that time the need for a surround may be greatest; later, however,
after some improvement thinnings, all the trees in a plot could be used for
the description of a provenance.


To summarize, the six questions posed above lead to the following
conclusions:


1. The minimum number of trees that will describe a stand is 8.
2. The minimum age at which productivity per unit area can be estimated
is 33 years.
3. In order to reconstruct a population, a minimum of 40 trees are needed.
4. The total area of the plantation should not exceed 5 hectares.
5. The minimum number of replicates needed is 7.
6. At the age of 15 years at least one row of surround trees is needed.
To satisfy these minimal conditions, coupled with the need of maintaining
silvicultural treatments that are standard for Central Europe it is proposed to
use the following design for provenance experiments with pine and spruce:
A complete block design, with 8 replicates and 49 pine trees per plot planted
at a 0,60 X 1,20 m spacing and 36 spruce trees per plot planted at a 1,0 X 1,0
m spacing.


92




ŠUMARSKI LIST 1-2/1966 str. 95     <-- 95 -->        PDF

This will give rectangular plots 4,20 X 8,40 m = 35,28 m2 for pine and
square plots 6 X 6 m = 36 m2 for spruce. With the 8 replicates planned each
provenance will cover 282,24 m2 and 288,0 m2 for a pine and spruce plantation
respectively. Thus on a site of 5 ha it will be possible to include about 170
provenances. Smaller number of provenances will require a relatively smaller
area.


In Table 1 the number of trees per plot at different ages is given for pine
and spruce. The number of trees per ha. at different ages shown in Table 1
were taken from Schwappach´s yield tables (1943) for the best site class. On
poorer sites there will be more trees per ha.


Table 1


Number of trees per hectare and per plot at different ages for pine and spruce


Age,
years
Pin e
per ha.
1.20 X 0.60 m.
per plot 35.22 sq. m.
Spruc e
per ha.
1 X 1 m.
per plot 36:
1 14,000 49 10,000 36
25 4,380 15.5 5,110 18.4
30 2,880 10.0 3,702 13.3
35 2,080 7.3 2,800 10.0
40 1,570 5.5 2,210 7.9
5(1 998 3.5 1,468 r>.´i
60 739 2.6 1.037 3.7
70 583 2.1 771 2.8
80 480 1.7 598 2.2
90 403 1.4 479 1.7
100 342 1.2 396 1.4
120 265 0.9 284 1.0


The values per plot have been calculated from Schwappach´s figures
corrected for the actual plot size.
If the site quality is lower than class I, then there will be more trees per
plot at a given age.


As can be seen from Table 1, the proposed arrangement will give about
8 trees per plot at 1/3 of the rotation age assuming a rotation of 100 years for
pine and 120 years for spruce. Thus, as has been discussed above, there will
be enough trees to evaluate the productivity per unit area.


At the end of the rotation there will still be one tree per plot and, with
the 8 replicates, 8 trees of the same provenance in the whole experimental
area. This will be sufficient to compare the provenances in respect of tree
quality and of numerous other features.


Productivity per ha will of course no longer be estimable. However, it
will be possible to corroborate all other characters with those tested earlier,
and to establish juvenile-mature correlations.


For the reconstruction of the population by means of a seed orchard, 40
trees are deemed necessary. These will be available at the age of 40 years in
pine (from all the 8 replicates), and at the age of 50 years for spruce. At these
ages both the productivity per hectare and most of the other qualities of
interest will have been reliably compared for the provenances in the experiment
and the best ones would have been selected.




ŠUMARSKI LIST 1-2/1966 str. 96     <-- 96 -->        PDF

Thus it will be possible to utilize in practice the results of the experiment


without depending on the preservation of the original stands from which the


seeds have been collected.


The 7 X 7 m. or 6 X 6 m. arragement spacings will permit the delimitation


of a surround in case the neighbour effect is severe. For the first period of


the experiment until the trees come into crown contact with each other and a


cleaning is necessary, it will be possible to eliminate from the analysis of data


either one or for pine species even two rows of surround trees, leaving at least


9 trees as the plot proper.


During the ticket stage (15—20 years) there will be about 1/2 of the
original number of trees per plot, as a result of the removal at the cleanings.
The border trees in a plot could still be considered as an isolation strip, leaving
at least 8 trees inside the plot, to be treated as the plot proper. At later stages
of the experiment it will be necessary to consider the whole plots, and it will
not be possible to eliminate the neighbour effect.


From the above discussion it will be seen that by basing the experiment
on minimal conditions it will be possible to satisfy all requirements specified
for provenance experiments at the beginning of this paper. Minimization of
the conditions is not a popular practice with biological research workers, but
it is the only way in which all requirements can be fulfilled.


CONCLUSIONS


A lay-out for provenance experiments is proposed, which will comprise
49 pine trees and 36 spruce trees per plot, i.e. 8 replicates within a complete
block design.


The plots will be rectangular for pine (4,20 X 8,40 m.) with the trees
planted at 0,60 X 1,20 m. spacing, and square for spruce (6X 6 m.) at 1 X 1


m. spacing, both cultivated according to the standard practices for Central
Europe.


This type of lay-out will make possible:


1) treatment of marginal trees as a surround for 15—20 years;


2) measurement of productivity per unit area at an age of 33—40 years;


3) establishment of a seed orchard from 40 clones per provenance at an


age of 40—50 years, thus preserving the genetically valuable populations;


4) maintainance of at least one tree per plot until rotation age, to permit


the establishment of juvenile-mature correlations;


5) inclusion of about 170 provenances in an experimental area of 5 ha.


It is suggested that when the requirements placed before the provenance
experiments are minimized to the extreme limit, they can all be satisfied by the
lay-out proposed.


KRATKI SADRŽAJ


POKUSI PROVENIJENCIJA I NJIHOVO PRILAGOĐIVANJE NAJEFIKASNIJOJ
SELEKCIJI I OČUVANJU POŽELJNIH ŠUMSKIH POPULACIJA


Predlaže se osnova za postavljanje pokusa s provenijencijama koji će obuhvatiti
49 borovih i 36 smrekovih stabala po pokusnoj plohi, tj. 8 replika (ponovljenja) unutar
plana pokusa po shemi potpunog bloka (grupe).


94




ŠUMARSKI LIST 1-2/1966 str. 97     <-- 97 -->        PDF

Pokusne plohe su pravokutnog oblika (za bor 4,20 x 8,40 m) s biljkama sađenim
u poredaju 0,60 x 1,20 m, te kvadraticnog oblika za smreku (6x 6 m) s poređajem
biljaka 1 x 1 m s time da se uzgajaju prema standardnoj praksi srednje Evrope.


Takva osnova za postavljanje pokusa omogućuje:
1) tretiranje rubnih stabala kao zaštitnog pojasa kroz 15—20 godina;
2) mjerenje proizvodnosti po jedinici površine u dobi od 33—44 godina;
3) podizanje sjemenske plantaže od 40 klonova po provenijenciji u dobi cd 40—
50 godina, održavajući na taj način genetski vrijedne populacije;
4) održavanje najmanje jednog stabla po pokusnoj plohi sve do kraja ophodnje


tako da se omogući utvrđivanje korelacija juvenilno-zrelog uzrasta;
5) uključivanje oko 170 provenijencija na pokusnoj plohi od 5 ha.
Ukazuje se na to da ukoliko se zahtjevi koji se postavljaju na pokuse s pro


venijencijama svedu na minimalnu granicu, onda se svima njima može udovoljiti
predloženom shemom pokusa.


LITERATURE CITED


Andersso n E. 1963: Seed stands and seed orchards in the breeding of conifers.


FAO/FORGEN-63, 2, 8/1.
Humme l F. C. 1955: The volume-basal area line. For. Comm. Bull., 24.
Johnsso n H. 1963: Arrangement and design of field experiments in progeny te


sting. FAO/FORGEN-63, 1, 2a/l.
Johnsso n H. 1964: Forest tree breeding by selection. Silvae Genet., 13: 41—49.
Matthew s J.: Seed production and seed certification.
Ministerstwo lesnictwa i przemyslu drzewnego 1961: Zasady ho


dowlane obowiazujace w panstwowym gospodarstwie lesnym. PWRiL, Bydgoszcz.
Schwappac h A. 1943: Ertragstafeln der wichtigeren Holzarten in tabellarischer
und graphischer Form. Prag II, Handelsdruckerei »Merkur«.


Tischendor f W. 1927: Studie zum mittleren Fehler des arithmetischen Mittels.
Ost. Z. Vermessungsw. From T. Gieruszynski: Dendrometria. Warsaw, IBL Se-
ria D, 5, 1499.


Tyszkiewicz S. and Z. Obminski 1961: Hodowla i uprava lasu. Warszawa,


P´VRiL.
Wrigh t J. W. 1962: Genetics of forest tree improvement. FAO Rome.
W r i g h t J. W. and F r e e 1 a n d F. D. 1958: Plot size in forest genetics research. Pap.


Mich. Acad. Sci., 44: 177—182.