Tropical Forest Mammals and Soil Fertility 357the density and patch size of keystone plant
resources (Peres 1997a, 1999b, 2000c, Chapman
and Chapman 1999, Stevenson 2001). Some of
these variables may be partly nested within the
effects of large-scale edaphic gradients considered
here but a more robust multivariate model could
explain a larger proportion of the variation in
primate abundance over such a vast region.
Primary forest productivity in terms of total
litterfall is a strong predictor of primate species
richness in neotropical forests (Kayet al. 1997),
both increasing with rainfall up to approximately
2500 mm year−^1. This relationship can now be
extended to total primate biomass in that soil
fertility is likely to be correlated with the total
above-ground turnover of the forest biomass. In
general, both primate species richness (Peres and
Janson 1999) and forest biomass turnover (Malhi
et al. 2004) increase from eastern to western
Amazonian forests, and this geographic pattern
also holds for total primate biomass and primate
biomass per species. Above-ground coarse wood
productivity in Amazonian forests increases with
soil fertility, particularly towards the eastern
flanksof theAndes(Malhiet al.2004),whichmir-
rors the geographic pattern of primate biomass. I
predict that the relationship between forest pro-
ductivity (e.g., as indexed by litterfall) and the
biomass of a group of primary consumers such
as primates would be even tighter if flower and
fruit production were considered separately, but
few studies distinguish the vegetative and repro-
ductive fractions of litterfall. I also predict that
mammal biomass in tropical forests is a strong
positive correlate of the above- to below-ground
ratio in forest phytomass, and that this relation-
ship will hold at most meso- to large scales,
depending on the extent to which wide-ranging
mammals integrate local edaphic constraints at
the landscape scale.
Fittkau (1973, 1974) was one of the first
to show a severe deficiency of some nutri-
ents essential to plant growth in central
Amazonia, particularly calcium, phosphorus,
nitrogen, potassium, and a number of trace
elements. He attributed the paucity of snails
and mussels to the notorious calcium deficiency
of Amazonian forest streams. Both terrestrial
and aquatic food webs are affected by regional
differences in geochemistry and soil fertility.
The net productivity of Amazonian nutrient-rich
white-water lakes can be 15- to 19-fold greater
than that of nutrient-poor black-water lakes,
where fish can show signs of severe nutrient defi-
ciency in their vertebrae (Geisler and Schneider
1976, Smith 1979). For example, the fish produc-
tionof sediment-richriversof Andeanoriginsuch
as the Madeira or the Purús (52 kgha−^1 year−^1 )
is much greater than that of rivers drain-
ing primarily nutrient-poor podzols and spo-
dosols such as the Negro–Casiquiare–Guainia
(6.6–13.2 kgha−^1 year−^1 ; Goulding 1979, Clark
and Uhl 1987, Gouldinget al. 1988).
Previousstudieshadalreadyshownlargediffer-
ences in total primate biomass between eutrophic
soils in seasonally inundated Amazonian várzea
forests and mesotrophic or oligotrophic soils in
upland terra firme forests (Peres 1997a,b, 1999b,
Peres and Dolman 2000). These patterns are
consistentwiththosefoundforAmazonianassem-
blages of small canopy mammals (Malcolmet al.
2005), large terrestrial and arboreal mammals
(Emmons 1984, Haugaasen and Peres 2005),
and mid-sized to large-bodied vertebrates (Peres
2000a,b). Primate communities in seasonally
flooded and terra firme forests consistently show
a reverse abundance–diversity relationship with
highbiomass,species-poorassemblagesoccurring
inthemostnutrient-richsoils(Peres1997a).Vari-
ation in primate biomass throughout the western
Amazon can also be explained by regional dif-
ferences in soil types and geochemistry. Primate
densities in southeastern Colombia, eastern and
southern Peru, western Brazilian Amazonia, and
northern Bolivia are consistently higher in white-
water than black-water drainages (Freeseet al.
1982, Peres 1997a, Palacios and Peres 2005),
despite sediment overflow from white-water rivers
to adjacent black-water drainages in exception-
ally high inundation years. For example, primate
biomass along the black-water Rio Nanay, upriver
of Iquitos, Peru, tends to be particularly low
(Freeseet al. 1982), reflecting the nutrient-poor
status of the soils in this region (Kauffmanet al.
1998). On the basis of 300 km of census effort
conducted at seven Amazonian forest sites of
varying productivity, Emmons (1984) suggested
that mammal abundance in Amazonian forests