Jökull - 01.12.1993, Page 23

Abies sp. Pinus silvestris Pinus cembra Picea sp. Larix sp.
Resin canals absent present present present present
Transversal tracheid absent walls tooth- shaped walls smooth walls smooth or tooth- shaped walls smooth or tooth-shaped
Ray Parenchyma cells with large open pits with large open pits with small pits With small pits
Transition from earlywood to latewood gradual gradual gradual gradual sharp
Table 1. The main anatom-
ical characteristics of some
tree species and genera occ-
urring as driftwood. Thick
letters indicate the most sig-
nificant characteristics.
— Helstu einkenni nokkurra
algengra viðartegunda sem
fmnast í reka.
beach on Svalbard (Salvigsen and Mangerud, 1991).
However, as these species can not be anatomically
differentiated (Schweingruber, 1978), they must have
been differentiated by some other method unknown to
the present author.
the dendrochronological network
To be able to identify and date the Arctic driftwood
by the dendrochronological method it is necessary to
have access to tree-ring chronologies from the circum-
polar forest regions. Such chronologies are available
from some parts of the boreal forest regions. The
tnost complete networks are from the White Sea region
(Bitvinskas and Kairaitus, 1979) along with networks
from Alaska (Cropper and Fritts, 1981) and the lower
Mackenzie river drainage area in Canada (Eggerts-
son 1994a) (Figure 2). In addition, some unpublished
chronologies (Fritz Schweingruber, pers. comm. 1993
and Eugene Vaganov, pers. comm. 1994) from Rus-
sia (including Siberia) have proven vital for dating the
driftwood.
STATISTICAL METHODS
Sign test (Gleichláufigkeit) is a measure of the sim-
ilarity between two tree-ring curves. The intervals be-
tween successive points in time are examined for an
upward or downward trend. The total Sign test over
all the intervals is a measure of the agreement between
the interval trends of two curves, and is expressed as a
percentage. Where the intervals for annual ring curves
run parallel for many years it can be assumed that the
factors influencing growth were similar in both cases
(Schweingruber, 1988).
By comparing all the curves from one site for a
given time span it is probable to see the homogeneity
of the site by comparing the Sign test values (Table 2).
It is often possible to synchronise samples of unknown
age with dated samples by calculating the Sign test
between the two samples.
Correlation coejficient (r) and t-test. The correla-
tion coefficient is a measure of the linear relationship
between pairs of values from two series and indicates
the similarity between the curves. In contrast to the
Sign test, correlation is based on the actual values
rather than changes from one value or interval to the
next.
The t-test indicates whether two curves are related
and generally gives the synchronous position of two
curves when the level of statistical significance is suf-
ficiently high (Table 2). The t-value is calculated as
follows:
r = correlation coefficient
JÖKULL,No. 43, 1993 21