348
D. Raffaelli and M. Emmerson
levels of productivity (Grime 1973; Rosenzweig and Abramsky 1993). At small
scales (metres to kilometres) increases in productivity through the addition
of nutrients lead to declines in species richness (DiTommaso and Aarsen
1989; Schindler 1990).
These relationships can be accommodated to some extent within Tilman's
(1987) resource heterogeneity hypothesis (RHH). The RHH states that if the
landscape is uniform with respect to resource availability, at low levels of
resource the average location will sustain few species and productivity will be
low. As a mean quality of the landscape increases, spatial variability and diversity of resources also increase, in turn leading to increases in both diversity
and productivity. At high levels of resource availability, space and resources
become less patchy and diversity is reduced, because when all sites are equally
good, competitively superior species dominate. The RHH thus predicts that
increasing nutrient supplies will lead to increases in species richness at sites
with low initial productivity, but will lead to decreases in species richness at
sites where resources are already abundant.
To test these predictions both we (Emmerson 2000) and Hall (2000) manipulated nutrient levels (and hence primary productions) in replicate smallscale artificial habitat units (HUs). In Hall's study, these were placed near
the sediment surface in a shallow sub-tidal seagrass bed, Boston Bay, Port
Lincoln, South Australia, a region with a particularly low nutrient status,
whilst we placed HUs in the Ythan estuary which is characterised by higher
nutrient levels (Raffaelli 1999). These two locations thus provide contrasting
biogeographies as well as differing in their nutrient status. Although there are
some minor differences in protocol between the Australian and the Ythan
studies, the experiments are essentially identical and so comparisons are
valid. Commercial pan scourers were used to construct artificial HUs into
which plant and animals could assemble. Nutrient supply was manipulated by
placing 60 g of commercial (Osmocote) slow-release fertiliser (nitrate and
phosphate) in a fine mesh bag inside each HU (see Hall 2000 for technical
specifications regarding Osmocote and nutrient release rates). In the Ythan,
HUs were suspended above an intertidal mudflat and left for 20 weeks (MayOctober). Each week, replicate HUs were collected, macrofauna enumerated,
and species richness, species accumulation and Shannon-Weiner indices of
diversity calculated. The RHH here predicts a decrease in species richness
given the addition of nutrients (additional resources) in the Ythan study. The
Ythan has a high nutrient status and hence resources are already widely available. In contrast the RHH predicts an increase in species richness (diversity)
and productivity given the addition of nutrients in the Boston Bay study; this
area has a low nutrient status and increasing resources here serves to increase
the mean quality of the environment.
Analysis of the Ythan pooled data using a two-way ANOVA without replication (treatment and time were considered fixed factors) demonstrated
D. Raffaelli and M. Emmerson
levels of productivity (Grime 1973; Rosenzweig and Abramsky 1993). At small
scales (metres to kilometres) increases in productivity through the addition
of nutrients lead to declines in species richness (DiTommaso and Aarsen
1989; Schindler 1990).
These relationships can be accommodated to some extent within Tilman's
(1987) resource heterogeneity hypothesis (RHH). The RHH states that if the
landscape is uniform with respect to resource availability, at low levels of
resource the average location will sustain few species and productivity will be
low. As a mean quality of the landscape increases, spatial variability and diversity of resources also increase, in turn leading to increases in both diversity
and productivity. At high levels of resource availability, space and resources
become less patchy and diversity is reduced, because when all sites are equally
good, competitively superior species dominate. The RHH thus predicts that
increasing nutrient supplies will lead to increases in species richness at sites
with low initial productivity, but will lead to decreases in species richness at
sites where resources are already abundant.
To test these predictions both we (Emmerson 2000) and Hall (2000) manipulated nutrient levels (and hence primary productions) in replicate smallscale artificial habitat units (HUs). In Hall's study, these were placed near
the sediment surface in a shallow sub-tidal seagrass bed, Boston Bay, Port
Lincoln, South Australia, a region with a particularly low nutrient status,
whilst we placed HUs in the Ythan estuary which is characterised by higher
nutrient levels (Raffaelli 1999). These two locations thus provide contrasting
biogeographies as well as differing in their nutrient status. Although there are
some minor differences in protocol between the Australian and the Ythan
studies, the experiments are essentially identical and so comparisons are
valid. Commercial pan scourers were used to construct artificial HUs into
which plant and animals could assemble. Nutrient supply was manipulated by
placing 60 g of commercial (Osmocote) slow-release fertiliser (nitrate and
phosphate) in a fine mesh bag inside each HU (see Hall 2000 for technical
specifications regarding Osmocote and nutrient release rates). In the Ythan,
HUs were suspended above an intertidal mudflat and left for 20 weeks (MayOctober). Each week, replicate HUs were collected, macrofauna enumerated,
and species richness, species accumulation and Shannon-Weiner indices of
diversity calculated. The RHH here predicts a decrease in species richness
given the addition of nutrients (additional resources) in the Ythan study. The
Ythan has a high nutrient status and hence resources are already widely available. In contrast the RHH predicts an increase in species richness (diversity)
and productivity given the addition of nutrients in the Boston Bay study; this
area has a low nutrient status and increasing resources here serves to increase
the mean quality of the environment.
Analysis of the Ythan pooled data using a two-way ANOVA without replication (treatment and time were considered fixed factors) demonstrated