Discussion of the overall hypotheses

(1) In tropical montane forests, root distribution is more superficial at higher than at lower altitudes. This is due to (i) very low pH in mineral soil at higher altitudes and thus, increased risk of Al toxicity and due to (ii) slow nutrient release in deeper soil layers from litter decay and weathering and thus, a higher importance of input of readily available nutrients into soil from the canopy by precipitation.

It could be shown that the distribution of RLD (Figure 2.2), the N uptake potential (Figure 2.3), the biomass of fine and coarse roots (Figure 4.1), and the origin of roots at the stem base (Figure 5.2) were more superficial at higher than at lower altitudes. Furthermore, the vertical extension of tap roots was more limited at 3000 m than at 1900 m (Figure 5.5) and lateral roots of G. emarginata frequently penetrated into the mineral soil at 1900 m, but hardly did so at 2400 m (Figure 6.3). Thus, distribution of fine and coarse roots was always more superficial at higher than at lower altitudes.

In literature, vertical root distribution is mostly presented for fine root biomass. At 1900 m, distribution of fine root biomass (62 % in the upper 0.25 m) was similar to temperate and tropical lowland forests, where 45-69 % and 42-57 % of total fine root biomass occurred in the upper 0.3 m (Jackson et al., 1997; Moreno Chacón and Lusk, 2004). It was in the range of several temperate broadleaved and coniferous forests where fine root biomass above 0.1 m depth of mineral soil contributed between 42 and 72 % of all fine roots to a depth of 0.3 m (Claus and George, 2005). The pattern of N uptake at 1900 m (Figure 2.3) that resembled the pattern of fine root distribution in the present study was similar to that of a well drained temperate forest where the amount of ¹⁵N acquired from 0.50 m soil depth by mature Quercus robur trees was about 25 to 40 % of the amount acquired from the upper 0.15 m depth (Göransson et al., 2006). Thus, distribution of fine root biomass resembled fine root distribution in other forest biomes.

In contrast, distribution of fine root biomass at 2400 m (94 % in upper 0.25 m) and at 3000 m (83 % in upper 0.30 m) was more superficial than in most other studies. Distribution at these altitudes was similar to boreal forests (83% in upper 0.30 m) (Jackson et al., 1997) which are dominated by soil types, e.g. Podzols and Histosols (Zech and Hintermaier-Erhard, 2002), that occurred frequently at 3000 m. Sporadically, similar shallow rooting pattern were observed in tropical lowland forests (Klinge, 1973; Pavlis and Jeník, 2000). At 3000 m, the pattern of ¹⁵N uptake was similar to the pattern of ⁴⁵Ca uptake by Betula ssp. and Picea abies

trees growing on frequently water saturated and thus, oxygen deficient soils (Brandtberg et al., 2004).

The results of the present study illustrate that rooting in tropical montane forests is not always very superficial. The high abundance of fine roots in the acid mineral soils at 1900 m support the assumption that species growing in tropical montane forests are adapted to low soil pH (Chapter 2.4). It has to be considered that the methodological approach of the present study (one stand per altitude) enhanced the probability that factors unrelated to altitude affected the observed pattern of root growth. Different to the other altitudes, the forest at the 1900 m site had developed on an old landslide profile. In soils developed on landslides nutrient availability (Wilcke et al., 2003) as well as soil pH may be increased (Schrumpf et al, 2001).

However, both parameters had little impact on root growth in the present study (Figure 2.7).

In contrast, oxygen deficiency in soil was likely an important factor for superficial rooting at high altitudes (Chapter 2.4). The frequencies of oxygen deficiency were particularly increased at high altitudes because of increased precipitation and decreased evapotranspiration due to lower temperatures.

(2) Higher proportions of plant biomass are allocated belowground at higher altitudes than at lower altitudes. High root to shoot ratios at higher altitudes are caused by (i) increased requirements for nutrient acquisition because low temperatures at high altitudes are associated with low mineralization of nutrients in soil, and by (ii) increased requirements for anchorage because of high wind speeds.

Despite the decrease in aboveground biomass with increasing altitude, the increasing allocation of plant biomass to the root system was accompanied by an increase in absolute root biomass (Table 4.2), supporting the assumption that plants at high altitudes were exposed to high environmental stress related to root system functioning. This is further supported by the finding that the proportion of root biomass in total biomass at 3000 m was at the upper range (5 - 41 %) observed in many tropical, temperate and boreal forests (Vogt et al., 1996;

Cairns et al., 1997).

Classically, the root to shoot ratio reflects the relative abundance of different resources, being larger when a resource utilized by roots is low (Farrar and Jones, 2003). Understanding of biomass allocation to the root systems by trees is very low (Vogt et al., 1996; Cairns et al., 1997). Fine roots are often a minor fraction of total belowground biomass in trees (Table 4.2, Vance and Nadkarni, 1992), but their contribution to total biomass may be more affected by nutrient availability in soil than root to shoot ratios, since resource acquisition is generally

restricted to fine roots. Concurrent to the decrease in nutrient supply from the organic layer with increasing altitude (Chapter 3.4), the proportion of fine root biomass in total biomass increased from 4.4 % at 1900 m to 5.2 % at 2400 m and 8.4 % at 3000 m. Increased allocation of biomass to fine roots also resulted in an absolute increase in fine root biomass from 1900 and 2400 m to 3000 m (Table 4.2). This is in accordance with the findings from other studies from tropical montane forests where fine root biomass was negatively correlated with nutrient contents in soil (Gower, 1987; Coomes and Grubb, 2000; Powers et al., 2005). Low foliar nutrient concentrations (Table 3.1) at high altitudes support the hypothesis that increased biomass allocation to the fine root system is the result of low nutrient supply but illustrate at the same time that decreased nutrient availability at high altitudes was not offset by increased fine root biomass.

The significant increase in coarse root biomass, associated with an increase in coarse root to shoot ratios from 0.11:1 at 1900 m to 0.33:1 at 2400 m and 0.48:1 at 3000 m suggests that increased belowground biomass at high altitudes is at least partly a response to increased mechanical stress. This assumption stands to reason since wind speeds increased with increasing altitude (Chapter 5.2), rooting depth and thus the depth of a potential root soil plate decreased with altitude (Figure 5.2, Figure 5.5,

Figure 6.3), the proportion of coarse root biomass in total root biomass increased with increasing altitude (Table 4.2), and coarse root systems at 3000 m exhibited a large range of traits that were supposed to improve anchorage (Chapter 5.4). In tree saplings and annual species, root to shoot ratios were increased by exposure to wind (Cordero, 1999; Henry and Thomas, 2002). The root to shoot ratio of mature Picea sitchensis (Bong.) Carr. trees was negatively related to the depth of the root-soil plate (Nicoll and Ray, 1996). Both, high wind speeds and decreased rooting depth may contribute to high allocation of biomass to coarse roots at high altitude.

Summarized, enhanced allocation of biomass to the root systems at high altitudes and the resulting high absolute root biomass may be attributed to low nutrient availability and enhanced mechanical stress.

(3) Coarse root architecture is modified by the prevailing soil conditions at each altitude.

These modifications are adaptive traits to improve anchorage under given environmental conditions.

Coarse root architecture varied considerably between altitudes. As discussed in the chapters 5.4 and 6.4, high requirements to anchorage at high altitudes may contribute to the observed

modifications in coarse root architecture. Modifications were attributed to phenotypic plasticity (Table 6.2) as well as to changes in the composition of species showing specific traits related to anchorage (Chapter 5.3).

It could be shown that phenotypic plasticity of coarse root architecture between altitudes was pronounced in the factor q that describes the topology of a root system. Variations in the branching parameter p, Nsub and l of one tree species (G. emarginata) growing at different altitudes (Table 6.2) were much smaller than variations in average p (Table 7.1), Nsub (2.16-2.35) and l (103-319 mm) between six different tree species growing in the area of the present study, highlighting the genotypic impact on these parameters. Since these parameters affect tapering along root axes and branching intensity they might also influence the species ability to adapt to increased mechanical stress. However, it remains unclear whether plasticity in these parameters was not detected in the present study because of their genotypic determination or because the differences in environmental conditions between 1900 and 2400 m were relatively small.

As discussed in chapter 5.4, stilt roots and the connection of stems by coarse rhizomes were supposed to improve tree stability at 3000 m, where deep rooting was hampered and wind speed was increased in comparison to deeper altitudes. Both traits occurred only on one species, respectively. It is reported from literature that the ability to propagate by rhizomes is at least partly genetically governed and is a frequent trait of woody species occurring in humid and cold environments (Kutschera and Lichtenegger, 2002). The ability to develop stilt roots, is also genetically governed (Jeník, 1978). Thus, species composition at different altitudes may affect the resistance of the forest stands to enhanced mechanical stress.

(RU1) “The high plant diversity in tropical montane forests is maintained by low nutrient availability.”

The reason for the high numbers of plant species that coexist on small spatial scales in tropical forests is poorly understood (Wright, 2002). Possible explanations for high tree diversity in tropical forests include (i) large heterogeneity in habitats influencing the development of tree seedlings and thus, the spatial distribution of tropical trees (Palmiotto et al. 2004; Dalitz et al., 2004; Hood et al, 2004), (ii) negative density dependence, that constrains locally abundant species, e.g. by allelopathy, intraspecific competition or pest facilitation (Wright 2002), and (iii) maintenance of the coexistence of competing species by disturbance (Molino and Sabatier, 2001; Kelly and Bowler, 2002; Sheil and Burslem, 2003).

Several studies summarized in Givnish (1999) show that tree species diversity in tropical

forests is positively correlated with the rates of rainfall and tree turnover, the time following catastrophic disturbances, and forest stature across lowland sites and negatively correlated with diameter at breast height and latitude. An increase in soil fertility along gradients in tropical forests was accompanied with an increase (Gentry, 1988; Aiba and Kitayama, 1999) or decrease (Huston, 1994) in tree species diversity. Ecological theory suggests that intermediate fertility should allow greatest biodiversity, because on fertile sites competitive exclusion of species may be enhanced by one or a few species with characteristics that improve light interception while on infertile sites competitive exclusion may be caused by species with superior nutrient acquisition characteristics (Herbert et al., 2004).

Tree species diversity in montane forests of the tropics is usually lower than in lowland forests (Haber, 2000; Homeier, 2004). On an altitudinal gradient in the RSF the numbers of tree species > 0.1 m in diameter at breast height decreased significantly from 17 species per 400 m^-2 at 1900 m to 2 species per 400 m^-2 at 2500 m. Also total numbers of tree species declined significantly with increasing altitude (Homeier, 2004). As suggested by foliar nutrient analysis, soil fertility in the RSF was high at 1900 m, but lower at higher altitudes (Chapter 3). These results do not support the hypothesis that high tree diversity at lower altitudes in this tropical montane forest is maintained by low nutrient availability.