only from seed. Among typical genera of short-lived pioneer trees (sensuFine-
gan 1996) of the neotropics, Vismiaspp. may resprout frequently, whereas
Cecropia spp. show this regeneration mechanism much less frequently
(Kammesheidt 1998; Mesquita et al. 2001). These differences probably
explain the dominance of young fallows by Vismiaspp. on Amazonian sites
observed by Uhl (1987) and Vester and Cleef (1998), although the latter
authors did not identify regeneration mechanisms. Work on slash-and-burn
practices in subtropical Australia (Stocker 1981) and Indonesia (Riswan and
Kartawinata 1991; Kiyono and Hastaniah 1997) has also shown variations in
regeneration mechanisms between species. In shifting cultivation landscapes,
the abundance and diversity of tree species with limited powers of regenera-
tion by sprouting depend partially on the presence of seed trees and disper-
sal agents. Most Macarangaspecies, very common short-lived pioneers in
Africa and Asia, seldom resprout and depend essentially on the soil seed bank
and the seed rain for regeneration.
The relative importance of resprouts in the development of fallow vegeta-
tion may vary between agricultural frontiers of different ages or landscapes
with different areas of remnant primary vegetation. Resprouts were the most
important sources of woody regeneration in 2- to 5-year-old vegetation on
low-fertility oxisols in Kammesheidt’s (1998) Paraguayan subtropical moist
forest study site, but their relative importance declined with age, and trees
regenerated from seed were more important in 10- to 15-year-old stands.
Results from 5- to 20-year-old fallows on similar soils in the Bragantina region
of eastern Pará State in the Brazilian Amazon were in marked contrast. There,
resprouts contributed the greatest proportion of both stems and species 5 cm
or more in diameter at breast height (130 cm) throughout the range of stand
ages studied (Vieira and Proctor 1998).
Possible reasons for such intersite differences in the importance of
resprouts as a regeneration mechanism may be suggested. It seems likely that
the dominance of fallows and secondary forest by resprouts over long time
periods in Bragantina (Vieira and Proctor 1998) results from the much greater
time since settlement there than in Paraguay (more than 100 years in compar-
ison with 30 years; Kammesheidt 1998). Especially when farmers are shorten-
ing the length of fallow periods, it is reasonable to hypothesize that on-site
seed production by many tree species must be reduced or nonexistent in shift-
ing cultivation landscapes such as those of Bragantina simply because stems
barely or never reach reproductive status. Possible reductions in the size and
diversity of the seed rain caused by the preceding factors might be exacerbated
by the loss of primary forest habitat and its function as a seed source (Denich
1991; Vieira et al. 1996). Additionally, if most remnant primary forest is ripar-
ian or swamp vegetation, some of the plant species that make it up probably
have limited abilities to colonize dryland sites, especially in competition with
the pioneers typical of such sites. Vieira et al. (1996) nevertheless recorded
162 III. The Biodiversity of Agroforestry Systems