3.4. Correlation and ordination
The correlation analysis showed a strong and significant positive
correlation between understory biomass and the amounts of litter
(r = 0.86), SOC, and SON (Table 4). In contrast, the amount of litter
was positively but not significantly related to SOC and SON. Bulk
density was negatively correlated with SOC (r = −0.70) and SON
(r = −0.70) as well as percent slope (r = −0.74) (Table 4). Soil pedestal
height and 210Pbex were not significantly correlated with any of
the other variables, although SOC tended to increase with increasing
210Pbex (Figs. 5, 6).
Five principal components explained 87.6% of the observed variation
in the soil and vegetation data (Table 5, Fig. 5). The first component had
high positive loadings for SOC, SON, C/N ratio, and slope, and a strong
negative loading for soil bulk density (Table 5). This component reflects
the soil chemical status and physical degradation due to terracing or
cultivation, and it explained 28% of the total variance. The second component
had positive loadings on understory biomass, mass of litter and
total cover, and a negative loading on soil pedestal height (Table 5). This
component is an indicator of soil cover and short-term soil erosion, and
it accounted for an additional 21% of the total variance. The third
component was dominated by 210Pbex (−0.79), which was indirectly
related to ground cover. The fourth and fifth components were
dominated by clay content and total phosphorus, and each of these
last three components explained between 11 and 16% of the total
variance (Table 5).