LAI and related parameters

The three-dimensional structure of the forest is of great importance as it determines radiation regimes, spatial competition etc.. There are several methods of describing and

101

quantifying the three-dimensional structure and canopy cover: from above, over large areas with remote sensing, or from underneath the canopy with hemispherical photo techniques, which was the method used in this study.

5.4.4.1 LAI in the studied stands as compared to other tropical forests

Cournac et al. (2002) obtained LAI-values around 4 (ranging from 1.0 to 5.8, obtained with an optical device constructed by the authors) in a tropical lowland forest in French Guiana, which is comparable to the results of the present study. The mean LAI observed in the tropical submontane natural forest in this study was 3.6 m² m^-2 (range: 2.5-6.8), which is also very similar to the 4.0 m² m^-2 reported from an undisturbed natural forest in Brazil by Meir et al. (2001). Trichon et al. measured PAI(plant area index, which includes branch and trunk areas, and thus reaches slightly higher values than LAI) and obtained values between 3.1 and 9.2 m² m^-2. These results should be particularly well comparable with the present study, since both were obtained in malesian natural forests with the same method. With the difference between PAI and LAI in mind, these data are also similar to the results of the present study. Some other literature data show higher LAI-values. Granier et al. (2000), working in the same area as Cournac, reported a mean LAI of 8.6 m² m^-2. These data however were collected with another method, using LiCor LAI 2000-equipment. It is hazardous to compare LAI values achieved with different technical methods and with different estimations and calculative approximations. Thus, it should be assured that the same equipment and the same calculation methods are used for comparing different forest types or different regions (Cournac et al. 2002).

5.4.4.2 The relation between LAI and leaf traits

LAI was negatively correlated with SLA and leaf N concentration if plot averages of several forest types were considered, representing a broad spectrum of growth strategies.

The same correlation has been reported by Pierce et al. (1994) in a comparison of different forest types in Northern America and could be explained by the enhancing effects of an open canopy structure (low LAI) on plant growth rate, favouring trees with large, N-rich leaves.

This kind of leaves typically also has high SLA values. In this study, of course the division between AF-plots with N-rich leaves and low stand level LAI on one hand, and NF-plots with high LAI and lower N on the other hand have influenced the found correlation.

If analysed within a forest type however, the relations between LAI, N and SLA depend on the heterogeneity of the forest stand studied. Natural forest sun canopies, mainly

consisting of late-successional trees in their mature phase, have little variance in N

5 DISCUSSION

concentration and SLA, whereas LAI in this forest type varies with the presence of gaps, but shows only low deviation as well. Therefore, the small variances in N concentration, SLA and LAI in natural forest stand are not likely to cohere. Pierce et al. (1994) also realized this, and concluded that using LAI to estimate canopy average SLA might provide inappropriate estimates of SLA in canopies that have recently been disturbed.

Pierce et al. (1994) further found the closest correlations between leaf N concentration and LAI at their coniferous sites in spring and explained this with the mobilization of N for new needle construction. The positive tendency found within the Sulawesi secondary forest stands, which are assumed to have a high growth rate, could have a similar explanation.

The agroforestry systems studied had great variances in LAI, also within plots, reflecting the heterogeneous structure of these systems. There was 5 - 10 m distance between the single cocoa trees, sparsely covered by shade trees. The small cocoa trees themselves had a very dense canopy (up to LAI = 9 m² m^-2 found on one of the plots studied). The mean leaf N concentration and SLA though, were very homogenous among the plots of this land use type, because of the low species number. Thus, no correlation between SLA or N and LAI existed within the agroforestry systems.

5.4.4.3 The visible-sky parameter for forest dynamic assessment

Trichon et al. (1998) used the hemispherical photo technique to qualitatively

characterize the degree of disturbance and spatial changes in rainforest areas of Sumatra, Indonesia. Their method based on combining quantitative results with the qualitative observation of each picture, giving a valuable link between numeric values and actual disturbance or dynamic in a forest. They classified the pictures into three groups: gap, building and mature forest sites, which applied to terminology used in forest dynamic studies by Watt (1947) and Whitmore (1989). Trichon et al. (1998) state that the plant area index and leaf area index are not suitable for this quantification method, since similar LAI values were obtained at gaps and mature forest. Instead, the visible sky parameter should be used. A high percentage of visible sky suggests a potentially high dynamic. According to Whitmore at al. (1993) canopy openness calculated from hemispherical photos is also highly related to microclimate in gaps, and is thus important for forest regeneration issues.

Trichon’s visible sky values ranged between 7.6 and 28.8% in the gap phases and from 1.4 to 5.1% on pictures classified as building and mature natural forest sites. Compared to this data the natural forest studied in this work would be classified as mature phase, having values ranging from 3.0 to 7.9%, however with some gaps present.

103

5.4.4.4 MLA and visually observed leaf angles

Even after adjusting recalculation, it is not possible to compare the visually observed leaf angle with the mean leaf angle calculated by the HemiView program (MLA) as equals.

As obtained by different methods, the two parameters are differently defined in the three-dimensional space. The visually observed single leaves were classified according to the angle and direction of the midrib, whereas the MLA and the adjusted values of the observed leaf angles do not indicate whether the leaf tips are directed upwards or downwards. Thus, a canopy consisting of 50% upwards (45º) and 50% downwards (135º) orientated leaves would generate a less informative MLA-value of 45º. However, in a light interception context this is of less significance. Further, MLA is an estimate of all leaves in the canopy, independent of age, position etc., whereas the observations of single leaves only considered fully sunlit mature leaves, which usually have a wider leaf angle than newly developed leaves.

Means of the visually observed leaf angles are conclusively more reliable and true for mature sun leaves of the observed species, whereas MLA might be more representative for the whole canopy, including all strata and all age classes. These differences should be kept in mind, when comparing the achieved leaf angle data. Because of the divergences

mentioned above, only the observed leaf angles will be subject to discussion, also in relation with photosynthetic rates etc..

5.4.4.5 Differences in mean leaf angle between land use types

The secondary forest had the steepest mean leaf angle of the three land use types.

Among the secondary forest species studied, there were several large-leaved species with characteristically steep hanging leaves. Typical examples are two Macaranga species and Mallotus mollissimus from the most frequent family in the secondary forest, Euphorbiaceae.

The mean leaf angle of the agroforestry systems was horizontal, as a result of pooling the often slightly upwards directed leaves of Gliricidia sepium with the large, heavier

Theobroma cacao leaves that have wider leaf angles.

The natural forest showed a smaller mean leaf angle (more horizontal) than the secondary forest, but steeper than the agroforestry system. This was revealed by the

observation of eight species, but nonetheless coheres with the overall impression if looking at a NF canopy. A natural forest sun canopy consists in general of many small-leaved trees with quite horizontal leaf angles, with some species with more steep leaves, like Litsea sp. and Meliosma sumatrana.

5 DISCUSSION