CARL OLOF T AMM

35

Table III. Variation in ratio of segment 2 to segment 3, within and between pairs of connected sympodia. Sample 672, collected 15.. VI. 1950 at Os near Bergen, and eonstitoting the entire moss earpet from a 25 X 25. cm plot beneath a spruce, light

exposure fairly good.

Source of variation l ^Degrees

l

^{Sum of}

l

Mean square

l

^standard

of freedom squares deviation

Total variation . ; . .

l

₃₇

l·

^10.7034

l

^0.2893

l

^±0.54

Between pairs ....

·l

¹⁸

l

^7.188o

l

^0.3993

l

-Within pairs ... 19 3-5163 0.1851 ± 0.43

Variance ratio 0.3993 = 2.157 Prohability slightly above o.o5 o.1851

analysisthus fails to show any clear difference in variation between neighbouring plants as compared with that between randomly ehosen plants within the same small plot. This result is probably due to the limited number of cases; very likely a larger sample would have given a significant variance ratio, as the existence of such local differences is very probable.

From another point of view, however, Table III can give valuable information, viz. an idea about the variation within the pairs as compared with that between randomly ehosen plants. In the last column of Table III we find the over-all standard deviation among the 38 individuals, 0.54, and the standard deviation within pairs, 0.43. Even if these figures are not so accurately determined as if the sample had been larger, it is evident that a considerable part of the variation remains when we compare individuals of the same clone growing a few centimetres apart. This result must be kept in mind when we consider the eauses of variation.

Perhaps it ma y be of interest that Table III deals with a less usual type of group difference: components of variation, not fixed relations among means, are respon-sible for the difference in mean squares between and within groups (cf. ErsENHAR1 1947)·

Question z), the correlation between a segment and its grandparent segment, can best be studied by calculating the partial correlation coefficient between these two segments after elimination of the influence of the segment between. Such calculations have been carried out for four different samples, listed in Table IV together with the results of the calculations.

If we first look at the direct correlation coefficients in Table IV, we find "highly significant" values in all cases, though the coefficients are higher for correlation between two directly connected segments than between segments 2 and 4· The bud (segment I) shows less correlation with its parent segment than this segment with segments 3 and 4· One of the possible explanations might be that some buds had started their growth period when the sample in question was collected, while others had not.

The highest correlation coefficients have been obtained in samples from shaded plots, +o.85*** between segments 3 and 4 in sample 492, and +o.87*** between segments 2 and 3 in sample 677. The correlation coefficients between segments 3 and 4 have been lower than those between segments 2 and 3 in the Norwegian samples (Nos. 672 and 677). In this connection we may remember that decom

No.

of sample

rSs

492

672

677

CARL OLOF TAMM

Table IV. Correlation coefficients between weights of different segments from the same Hylocomium · spiendens individuals.

No. Correlation coefficients between segments

Date of Locality of I and 2 2 and 3 3 and 4 2 and 4

sampling Exposure

speci-mens direct dir e et

l

^{partial direct}

l

^{partial dir e et}

l

^partial

r u r,3 r•3·4 r34 r34·• r24 rz4·3

7.VIII-48 Site I. Gren- IOO +0.42*** +0.74*** +o.sr*** +0.79*** +o.62*** +o.63*** +o.II holmen.

Moderately exposed.

I4.VIIL49 Rönninge. 57 not det. +0.77*** +o.68*** +o.Ss*** +0.79*** +o.s6*** -0.30*

Very dark spruce forest

rs.VI.so Os. !30 not det. +0-73*** +o.sr*** +o .6g*** +0.42*** +o.6s***

Moderately exposed spruce forest

rs.VI.so Os. 6g not det. +o.87*** + 0·75*** +o.7o*** +0.32** +o.68***

Very dark spruce forest

position started very early in this samples, sothat segment 4 appeared to be con-siderably affected. This process may of course lower the earrelation coefficients, if different segments are decomposed at a different rate .

. In Table IV the partial earrelation coefficients are also given. They are alllower than the direct coefficients. As far as the earrelation between two directly connected segments is concerned, all partial coefficients bu t one are still "highly significant".

The exception is r34 .² in sample 677, which is "significant" only.

In the last column of Table IV we find the partial earrelation coefficients between a segment and its grandparent segment, after elimination of the segment between.

Of the four values three are positive and the fourth negative. Only one of the values, +o.z8** in sample 672, is "significant". The negative value- 0.30*

is "suggestive". There isthus not much evidence for the dependence of a segment on its grandparent segment, in addition to that resulting from the physiological connection between two directly joined segments. Even if a partial earrelation may exist in some cases, values such as +o.z8 and -0.30 do not imply more than a very loose relation.

The most likely cause of the earrelations between different segment-partial as well as direct correla tions - i s of course direct physiological action of on e segment on the other. A strong positive partial earrelation would for example be expected between segments z and 4, if substances were translocated from dying segments to the youngest segment. Another effect of the parental segment is that it supports the younger ones. A strong negative partial earrelation between granddaughter and grandparent segment might be difficult to understand, but it is possible that a more loose negative earrelation may be brought about by a kind of "position effect". A general tendency of a moss plant which once had reached a

+o.28**

+o. r g

37 favourable position, to maintain this position for at least some years, and a ear-respanding persistent effect of unfavourable positions, would be expected to give rise to such a negative partial correlation. It is not quite clear whether this case can be realized in a plant community in equilibrium.

While it would be dangerous to explain an existing strong earrelation in this rather speculative way, the absence of a partial earrelation between granddaughter and grandparen t segment in two or three cases out of four, and the possible existence of a weak positive earrelation in the fourth case, suggests that the factors described above do not have great importance. It is of course possible but not very likely that both factors-physiological action from segment 3 to segment r, and per-sistence of a favourable position of the individual-are important but balance each other even under the very different conditions in the four sample plats in question.

Conclusions

The results of our studies on the individual growth variation within the Hylocomium splendens community can be summarized as follows:

The size of a Hylocomium segment depends to a certain extent on the size of its parent segment, but it is subject to a strong and irregular variation. The eauses of this variation are almost certainly to be found outside the moss plant, as strong variation occurs even between moss individuals belonging to the same clone. Besides, it is difficult to imagine genotypes with dispositions for weak and strong growth under the prevailing conditions growing side by side for a longtime in a community where propagation is mainly vegetative.

The heterogeneity in moss growth thus leads us to believe that one or more environmental factors (including biotic factors) change from place to place and from year to year in an irregular way within the Hylocomium splendens com-munity.

Another major point of interest in the structure of the Hylocomium earn-munity is the low frequency of small individuals, which shows us that the rate of renewal, except by branching, must be very small.

It would be interesting to campare the structure of the moss community with that of other plant communities, particularly with respect to the growth variations. The quantitative growth of herb and grass specimens during a series of years is difficult to measure. We know, however, that the same indi-vidual may occupy a similar position in the community for many years (TAMM 1948).

I t is easier to measure tree growth than the growth of herbs and grasses.

Records of diameter growth during two subsequent years for pine, birch and spruce give pictures similar to Figs. 7 and 8, with a large variation around a regression line. Diameter growth is, however, not the same as total production, and some of the scatter ma y be due to imperfect earrelation between diameter growth and total production. A similar view ma y be taken in the case of top

CARL OLOF T AMM

·shoot length of beech, where HoLMSGAARD (1950) has found only a loose earrelation between the lengths of two subsequent shoots.

Af present we have to confine ourselves to the moss community, but the . value of measurements

of

growth variation ought to be stressed more often in plant ecology. The competitive activity of a plant is a function of its growth potential, and variations in the growth of the members of a plant community thus indicate variations in the intensity of competition.

Chapter V. Variation in Producdon and Structure of

In document BAND 43 1953 (Page 38-42)