The URL of this frame;  

  Homepage of Akihiro SUMIDA  Japanese
                              Goto Page 2   


 Dr. Akihiro SUMIDA , forest ecologist
   Associate Professor
Cold Region Plant Ecology and Physiology Group
  Institute of Low Temperature Science, Hokkaido University,

        Publications in English

  Contents may be updated occasionally.l(Last minor update: May 2nd, 2018!)l            

My main interest is to understand the relationship between the development of structural components of a forest (annual shoots, twigs, branches, trunks) and the whole structure of a forest (e.g., population structure, 3-D spatial structure of foliage distribution) which would be formed as a result of inter-individual interaction such as competition.

Studies of tree/forest development have long been done by many researchers. My approach is focused on the relationship between small-scale structure development (i.e., at branch scale) and a large-scale structure (at individual and stand scales).

Changes with time in the relationship between stem diamter (DBH) and tree height (H) of individual trees in accrodance with their crown dynamics
Sumida et al. 2013
The picture above is one of my recent studies. Current studies include climatic factors relating to these these changes, and crown dynamics as related to branch-level dynamics 

What's so interesting?
During my Ph.D. study, I often climbed trees (for research), reached the canopy layer and saw that the foliage of the trees composing a forest formed layered structure. The heights of the layers were undulating from place to place (see below).

A profile of a hardwood forest canopy which is 'sliced' with a thickness of 20 cm. Gray part shows the space occupied by foliage. Isoclines of relative light intensity (%) fluctuates spatially from place to place depending on the presence of foliage, but it also decreased in the space without foliage, mainly because of the influence of foliage that existed around this profile. Original data; Sumida (1993), Sumida (1995).

While the whole layered structure would develop upwards with time as trees composing the layers grow, clearly the layered structure is maintained. This is strange if you know how individual-tree structure develops; growth of trees accompanies branching of current-year shoots.


If an annual shoot (left) is produced in a tree crown, it becomes a mother shoot the next year and plural number of 0-yr-old daughter shoots (red: leaves are not shown) originating from the mother shoot are produced each year.

Counting the number of annual shoots of each year (see above), and plotting the number of annual shoots on y-axis and age of annual shoots on x-axis, the relationship can be approximated by exponential equation (y = a exp (-b x); see below). The slope of the equation suggests that the number of annual shoots of each age increases about 2-3 times as age of a shoot becomes one-year younger. Hence it appears (but actually not) that the number of current-year shoots increases exponentially with time (year).

               Figure 1 in Sumida & Takai (2003)
However, it is impossible for the amount of leaves in a forest stand to increase exponentially with time if the forest canopy is closed; there is very little room for trees to secure the space for expanding crown with exponential branch development.
It is well known that the leaf amount of a forest tends to be more or less constant (stable) after canopy closure. How can the seemingly incompatible phenomena (constant stand leaf amount vs. exponentially increasing number of branches) go together?

  Seemingly exponentially growing structure,
      acturally in a steady state

One of the biggest answers may be simple; when branching structure develops from a single mother shoot, only the shoots originating from the topmost daughter shoot ('Apical' in the left-side figure below) on the original mother shoot continues developing, whereas the shoots originating from the lateral shoots tend to develop less and less each year, and finally stop producing buds for the next year and die.
Simulation of leaf area developping with time on each of the daughter shoots on a mother shoot. Twigs were sampled at the topmost part of crowns of adult Konara oak (Quercus serrata) trees in a forest. Using their morphological data, we obtaiend parameters for representing developping patterns of forking structure and leaf area on each annual shoot. Then twig-structure development was simulated (Sumida & Takai (2003)).

A constant structure of clustered twig resulted at 5-6 years after the onset of branch development (Sumida & Takai (2003), Sumida & Umeki (2007)). We found that a steady state of twig size and constant structure (which I call "foliage cluster") resulted owing to the fate of death in lateral shoots. After the constant structrue is achieved, the foliage cluster just rises upward (see figures below).

Development of branching structure with formation of "foliage cluster", simulated for the topmost canopy of Quercus serrata. Twig structure of each year is shown from left to right, and the red segments show current-year (0-year-old) shoots with leaves. Lower branches (gray shoots) without any current-year shoots in their distal ends indicate that they are dead.  In the first year five 0-yr-old shoots were produced on a mother shoot. In the fifth year, all four lateral shoots of the first year died and only the shoot at the top in the first year sruvived. After that, the size of foliage cluster is stable and the number of 0-yr-old shoots in the cluster does not increase. So, in the figure above, the number of the oldest shoots in the fifth year, or that of 4-yr-old shoots, is only one. Now you know the trick in Fig.1; when the size of the foliage cluster is stable, the number of oldest shoots in the cluster is always one. So the seemingly "exponentially increasing" relationship in Fig 1 holds only within a foliage cluster with a stable size.

So, there is a trick in the 'expornentially growing structure' in Fig. 1. After a foliage cluster acheved a steady state structure, the number of the oldest shoots within the cluster is only one. This oldest shoot is located at the base of the main stem of a foliage cluster . Hence the number of living oldest shoots within a steady-state foliage cluster cannot decrease below one. (If it dies, the whole shoots of a cluster die.) That is, the seemingly "exponentially growing forking structure"in Fig. 1 applies only within a foliage cluster with a steady state.

It may suggest how the economy of a society can be maintained constant; some big leading companies keep increasing exponentially while other "lateral" small companies failures, so that the total economy of the society can be constant (joking..).

  How does a tree compete with neighboring trees?

Of course, competition occurs among trees/tree species composing a forest canopy. Hence the component species/individuals of the canopy may be replaced by others, even if the whole canopy structure is more or less stable. Here "competition" is used to express the interaction among individual branches/trees for spreading leaves in lighter space.

It is without saying that leaves composing forest canopy are spatially (and of course physiologically) supported by branches. To understand how the forest canopy develops spatially, we need to know how branches develop in a stand.

In a hardwood forest in Japan, we observed that crowns of Japanese chestnut trees (Castanea crenata, Fagaceae) in the canopy layer tended to be narrower than other species surrounding the chestnut trees. The trend was not very clear (i.e., not statistically significant) at individual-crown scale, but when we analyzed the trend for individual first-order branches (= those growing from a main trunk) we found that extension of Japanese chestnut branches were less than that of neighboring branches of other species when inter-branch distance (horizontal distance between branch bases of two nearest neighboring branches) was less than about 5 m (Sumida et al., 2002).
In other words, the Japanese chestnut trees are not good at "infighting" with other species. How can this happen?

Measuring branching structure with a total station (for details see Sumida et al (2002))

 Actually measured 3D trees: wear a blue lens on your right eye and a red one on your left eye.
Branch structure of Japanese chestnut (Castanea crenata) trees (yellow) surrounded by other tree species (red) in a deciduous hardwood forest in Shokawa, Gifu prefecture, Japan. Reconstructed from measured data. (original paper: Sumida et al. (2002)). Thanks to for producing the figure above.
One of my co-authors (Prof. Dr. Komiyama of Gifu University, Japan, and his students in those days) had found that the leaf flush of Japanese chestnut in spring was about a few weeks to a month later than other tree species, because Japanese chestnut is a ring-porous species, which are known (in general) to have later leaf-flush phenology in spring.

This suggests that, when there are branches of other tree species near a branch of Japanese chestnut, the latter has to open leaves after other deciduous had already occupied the space for spreading leaves; the space is "first-come first served".
This explains why Japanese chestnut trees are not good at infighting.

If so, another question arises; how can a tree species with less ability of spreading crowns can survive in a forest? Actually, Japanese chestnut is a very common species in deciduous forests in Japan and somtimes becomes dominant. My guess (which is not written in my paper) is that it is related to dispersion patterns of plants.

Even if a plant species is prone to invasion by other neighboring plant species, aggreagtion of plants can retard extinction of "weaker" species (e.g., Silvertown et al. (1992) Journal of Ecology).

Seed dispersal of Japanese chestnuts are known to be made by small animals such as rodents (e.g.. Seiwa et al (2002) For. Ecol. Manage.,164, 149-156 ), and because rodents tend to bury prural seeds within their nest (and if they are stupid enough to forget eating them), prural seedlngs would germinate at the same place. Moreover, because the chestnut tree seeds are relatively heavy, seeds may be dispersed around a mother tree. Such a dispersal patter may lead to aggregated dispersion pattern of chstnut individuals within a relatively narrow area on the forest floor.

To conclude, spatial structure again matters for competition and survival of trees. This is also why I am interested in spatial structure of trees and forest.

Publications in English

Publications in English and Japanese

back to the top