Subsistence hunting affects vast tracts of tropical wilderness that otherwi
se remain structurally unaltered, yet distinguishing hunted from nonhunted
tropical forests presents a difficult problem because this diffuse form of
resource extraction leaves few visible signs of its occurrence. I used a st
andardized series of line-transect censuses conducted over a 10-year period
to examine the effects of subsistence game harvest on the structure of ver
tebrate communities in 25 Amazonian forest sites subjected to varying level
s of hunting pressure. Crude vertebrate biomass, which was highly correlate
d with hunting pressure, gradually declined from nearly 1200 kg km(-2) at n
onhunted sites to less than 200 kg km(-2) at heavily hunted sites. Hunting
had a negative effect on the total biomass and relative abundance of verteb
rate species in different size classes at these forest sites, but it did no
t affect their overall density. In particular, persistent hunting markedly
reduced the density of large-bodied game species (>5 kg), which contributed
a large proportion of the overall community biomass at nonhunted sites (65
-78%) and lightly hunted sites (55-71%). Nutrient-rich floodplain forests c
ontained a consistently greater game biomass than nutrient-poor unflooded f
orests, once I controlled for the effects of hunting pressure. Conservative
estimates of game yields indicate that as many as 23.5 million game verteb
rates, equivalent to 89.224 tons of bushmeat with a market value of US$190.
7 million, are consumed each year by the rural population of Brazilian Amaz
onia, which illustrates the enormous socioeconomic value of game resources
in the region. My cross-site comparison documents the staggering effect of
subsistence hunters on tropical forest vertebrate communities and highlight
s the importance of considering forest types and forest productivity in gam
e management programs.