Community Ecology Processes, Models, and Applications

dispersal rates among the more closely aligned me-
tacommunities, resulting in a reduced degree of sto-
chasticity and higher similarity in community
composition. In Effects of dispersal rates on local
community structure (below), we discuss how vari-
ation in the proximity of habitats (which is related to
the rates of dispersal) within a metacommunity
might alter patterns ofa-,b-andg-diversity.

5.5 Species–area relationships and metacommunity structure

Although it is conceptually quite useful, partition-
ing diversity intoa,bandgcomponents is not
always straightforward. Theoretically, one can
imagine localities as patches surrounded by inhos-
pitable matrix (e.g. Mouquet and Loreau 2003;
Amarasekareet al. 2004). In some metacommunities
with patchy structure this assumption is largely
fulfilled, and several such systems have been used
as empirical models of metacommunities. Exam-
ples include water-filled inquilines such as pitch-
er-plant leaves (e.g. Kneitel and Miller 2003; Gotelli
and Ellison 2006), ponds and rockpools (e.g.
Bengtsson 1989, 1991; Chase 2003), islands (e.g.
Simberloff and Wilson 1969; Diamond 1975; Scho-
eneret al. 2002), host plants for insects (e.g. Zabel
and Tscharntke 1998; Kreuss and Tscharnkte 2000),
moss patches on rocks (Gilbertet al. 1998) and rocky
outcrops (e.g. Harrison 1997, 1999; Ostmanet al.
2007). However, in many terrestrial (forests, grass-
lands) and aquatic (oceans, large lakes) landscapes,
the distinction between where one ‘locality’ ends
and another begins is ambiguous at best (Loreau
2000). In these continuous landscapes, diversity
might be scaled or partitioned more objectively
using species–area relationships (SARs).
The positive SAR is one of the oldest and best-
supported patterns in ecology (Lomolino 2000).
SARs can be calculated in two ways: (1) using in-
crementally larger areas in which smaller areas are
nested or (2) using data on richness and area from
distinct separate patches or islands (e.g. Rosenz-
weig 1995). The identity of mechanisms driving
variation in the shape of SARs remains at the fore-
front of research on metacommunity ecology (e.g.
Drakareet al. 2006). The function most frequently
used to describe the SAR is a power law:

S¼CAz

whereSis species richness andAis area. When log-

transformed, we get the familiar linearized SAR,

logS¼logCþzlogA

where andCandzare curve-fitting parameters

depicting the intercept (C) and rate of increase (z)

of species with area.

Of primary interest here isz, the slope of the log-

transformed SAR, indicating the steepness of the

increase in the number of species with increasing

area. Although somewhat more complex, varia-

tion in the parameterzis related to variation in

b-diversity (e.g. Connor and McCoy 1979; Rosenz-

weig 1995; Drakareet al. 2006). For example, SARs

are often calculated using a ‘nested’ design (Schei-

ner 2003), by adding together the area of similar

sized smaller patches or sampling units into a

larger area. In this case, if the smaller patches are

highly divergent in their community composition

(higherb-diversity), the increase in species rich-

ness with increasing area – thez-value – will be

higher than if the smaller patches are more similar

to one another (lowerb- diversity). Thus, variation

inzamong sites can provide important informa-

tion regarding the partitioning of diversity across

scales.

Several factors can create variation in the slope

(z)oflog-transformedSARs.Evenwhensampling

methodology is controlled,zvaries among com-

munities (reviewed in Drakareet al. 2006). Some of

this variation can be attributed to factors such as

organism size or trophic position (Holtet al.1999;

Drakareet al. 2006). However, variation inzcan

also emerge from differences in metacommunity

structure. For example,zis generally higher on

islands than on continents (Rosenzweig 1995;

Whittaker and Ferna ́ndez-Palacios 2007). This

might result because islands have lower rates of

dispersal, and thus higher inter-island differentia-

tion than similar sized areas of continents that

have higher rates of dispersal (see Effects of dis-

persal rates on local communities, below). In addi-

tion, in a meta-analysis of SARs, Drakareet al.

(2006) found generally higherz(and thus higher

b-diversity) at lower latitudes (Fig. 5.2). This

indicates the possibility that metacommunity

INCREASING SPATIO-TEMPORAL SCALES 63