vol. 169, no. 6
the american naturalist

june 2007
Temporal Patterns in Rates of Community
Change during Succession
Kristina J. Anderson*
Biology Department, University of New Mexico, Albuquerque,
New Mexico 87131
Submitted May 22, 2006; Accepted December 13, 2006;
Electronically published April 6, 2007
Online enhancements: appendix, data files.
abstract: While ecological dogma holds that rates of community
change decrease over the course of succession, this idea has yet to
be tested systematically across a wide variety of successional sequences. Here, I review and define several measures of community
change rates for species presence-absence data and test for temporal
patterns therein using data acquired from 16 studies comprising 62
successional sequences. Community types include plant secondary
and primary succession as well as succession of arthropods on defaunated mangrove islands and carcasses. Rates of species gain generally decline through time, whereas rates of species loss display no
systematic temporal trends. As a result, percent community turnover
generally declines while species richness increases—both in a decelerating manner. Although communities with relatively minor abiotic
and dispersal limitations (e.g., plant secondary successional communities) exhibit rapidly declining rates of change, limitations arising
from harsh abiotic conditions or spatial isolation of the community
appear to substantially alter temporal patterns in rates of successional
change.
Keywords: colonization, extinction, turnover, primary succession,
secondary succession, arthropods.
Succession—community development following a disturbance or formation of a new habitat—is traditionally
thought to embody increasing community stability
through time (e.g., Odum 1969; Whittaker 1975); that is,
rates of community change often decrease through time
during succession (e.g., Drury and Nisbet 1973; Jassby and
Goldman 1974; Bornkamm 1981; Schoenly and Reid 1987;
Prach et al. 1993; Myster and Pickett 1994; Foster and
* E-mail: [email protected].
Am. Nat. 2007. Vol. 169, pp. 780–793. 䉷 2007 by The University of Chicago.
0003-0147/2007/16906-41849$15.00. All rights reserved.
Tilman 2000; Sheil et al. 2000). Meanwhile, species richness usually increases initially (e.g., Odum 1969; Swaine
and Hall 1983; Saldarriaga et al. 1988; Whittaker et al.
1989) but then often declines (e.g., AuClair and Goff 1971;
Schoenly and Reid 1987; Lichter 1998). However, a clear
synthesis regarding temporal patterns in rates of community change during succession is currently lacking.
Here, I (1) describe measures of species gain and loss rates
and how these combine to determine turnover rates and
species richness, (2) examine the temporal patterns in
community change rates during succession across a variety
of community types, and (3) discuss the mechanisms that
may underlie predominant temporal patterns in species
colonization rates and richness.
Species gain (colonization) rate. Gain rate (G; time⫺1) is
the rate at which previously absent species appear in the
community. In order to measure the magnitude of gain
relative to the existing community, gain rate may be expressed as a proportion of the average number of species
present during the measurement period (Gp):
Gp p
G
.
(1/2) [S(t 1) ⫹ S(t 2 )]
(1)
Here, S(t1) and S(t2) are species richness at the beginning
and end of the sampling interval, respectively. The reappearance of previously present species that had disappeared may be included (G and Gp) or excluded (G and
G p ); exclusion assumes that absences are an artifact of
sampling and/or population stochasticity rather than a biologically meaningful event.
Several major mechanisms may be expected to influence
temporal patterns in gain rate. First, gain rate will be constrained by the number of species that can establish themselves and simultaneously persist in the community (KS).
Early in succession, when S is far below KS, G will be
limited primarily by dispersal. As S approaches KS and the
intensity of competition increases, gain rate will decrease
(e.g., MacArthur and Wilson 1963; Tilman 2004) until, at
KS, it is approximately balanced by loss rate (Goheen et
al. 2005). Although clearly an oversimplification, this
Temporal Patterns in Succession Rate
schema is useful for making first-order predictions regarding temporal patterns in succession rate. For example,
in the simple case where dispersal is not highly limiting
and where KS remains relatively constant over the course
of succession—as may generally be the case in secondary
succession—gain rate should start high and rapidly decrease as S approaches KS and the intensity of competition
increases (e.g., MacArthur and Wilson 1963; Bazzaz 1979;
Walker and Chapin 1987; Tilman 2004). In the more complex case where KS changes substantially—perhaps as a
result of changing resource availability—gain rate will take
the form of the derivative of KS(t). Thus, for example, a
sigmoidal increase in KS over time—as may be the case in
harsh environments where time and/or facilitation are required to make resources available (e.g., Walker and
Chapin 1987)—would result in a peaked function of G
(fig. 1B), whereas a peak in KS—as may be the case for
succession on ephemeral resources such as corpses—
would imply a roughly linear decrease in G (fig. 1C). In
both cases, the maximum G would be lower than that of
a community that does not face such abiotic limitations
(cf. fig. 1A–1C). Second, G will be controlled in large part
by the rate at which propagules of new species arrive at
781
the site. If the rate at which propagules arrive remains
constant through time, the rate at which new species arrive
necessarily decreases simply because many species are no
longer new. If, at each time step, a constant proportion
of the species pool that is not yet represented arrives, G(t)
will take an exponential form (fig. 1A, 1D, 1E). Sites that
receive large numbers of propagules (e.g., 90% of unrepresented species arrive at each time step) will have rapidly
decreasing G(t) and rapidly plateauing S(t) (fig. 1A). The
lower the rate of propagule arrival, the less rapid the decrease in G(t), the lower the G(0), and the longer the time
until S reaches KS (cf. fig. 1A, 1D, 1E). As a result, successional communities facing strong dispersal limitation
will display relatively nondescript temporal patterns in G
(fig. 1E). Note that, under this scenario, the size of the
regional species pool should affect gain rate but not temporal patterns therein. Thus, gain rates should decrease
less dramatically in isolated locations (MacArthur and Wilson 1963; Walker and del Moral 2003) and for communities composed of poorly dispersing species than in successional communities with high dispersal rates. Third, G
may be affected by herbivory or predation at any stage of
succession (e.g., Walker and Chapin 1987; Fraser and
Figure 1: Schematic diagram showing hypothesized effects of the number of species a community can potentially hold (KS; A–C) and dispersal
rates (A, D, E) on species richness (S) and gain rate (G). A–C, Hypothesized effects of KS being constant (A), sigmoidal (B), or peaked (C) over
time. A, D, E, Consequences of dispersal rates being such that each time step witnesses the arrival of 90% (A), 50% (D), and 10% (E) of the
potential colonists that had not yet arrived. An implicit assumption is that when dispersal limitations do not interfere, S tracks KS.
782
The American Naturalist
Grime 1999; Howe and Brown 1999; Fagan and Bishop
2000). Finally, G(t) may be influenced by loss rate (Bartha
et al. 2003), especially in the later phases of succession
when competition is more intense (e.g., MacArthur and
Wilson 1963; Bazzaz 1979; Lichter 2000). In combination,
these four factors may affect G(t) in a variety of ways.
Generally, G will decrease at any time that KS is not increasing, and the rate of this decrease will depend on dispersal rates. An increase in KS will counteract this tendency
for gain to decrease, sometimes causing it to increase.
Conversely, a decrease in KS will force G to be less than
the loss rate.
Species loss (extinction) rate. Loss rate (L; time⫺1) is the
rate at which species disappear from the community. As
with gain rate, this may be expressed as a proportion of
the species present over the measurement period (Lp):
Lp p
L
.
(1/2) [S(t 1) ⫹ S(t 2 )]
(2)
This measure represents the probability that any given
species will be lost in one unit of time. Again, species that
disappear and later reappear may be included (L and Lp)
or excluded (L and Lp), depending on whether such temporary absence is deemed to be biologically significant.
Several mechanisms may act on temporal trends in loss
rate. For example, L should increase with the number of
species that may potentially be lost (S). Additionally, both
L and Lp may be expected to increase as the intensity of
competition increases (e.g., MacArthur and Wilson 1963;
Bazzaz 1979; Lichter 2000). On the other hand, this may
be counteracted by a decreasing rate of invading species
that could potentially outcompete existing ones. Additionally, if average body size increases significantly over
the course of succession, increasing life spans may result
in decreasing loss rates (Drury and Nisbet 1973). Thus, it
is difficult to predict a priori how L and Lp will change
over successional time. The findings of previous studies
are likewise ambivalent, showing no relationship (Foster
and Tilman 2000), a positive relationship (Facelli et al.
1987), or a peaked relationship (Lichter 1998) between Lp
and time. Species turnover rate and richness can be expressed straightforwardly as functions of G and L.
Species turnover rate. Turnover rate (T; time⫺1) is the
average of gains and losses:
1
T p (G ⫹ L ) .
2
(3)
efficient (CS; Sørensen 1948; Koleff et al. 2003), to express
the rate of percent turnover (Tp):
Tp p
CS
1 ⫺ {2S C /[S(t 1) ⫹ S(t 2 )]}
p
t 2 ⫺ t1
t 2 ⫺ t1
p
G⫹L
T
p
.
S(t 1) ⫹ S(t 2 )
(1/2) [S(t 1) ⫹ S(t 2 )]
Here, SC is the number of species present at both the
beginning and the end of the measurement period. It
should be noted that, as opposed to narrow-sense measures of turnover that focus on changes in species identity
(e.g., Routledge 1977), this measure will also be strongly
influenced by changes in species richness (Koleff et al.
2003). Note also that Tp is the average of Gp and Lp and
relates to T in the same way that G and L relate to Gp and
Lp (eqq. [1], [2], [4]). Just as with gain and loss, turnover
may include (T and Tp) or exclude (T and Tp) species
that disappear temporarily.
Turnover rate, as the average of gain and loss rates, will
be driven by the mechanisms that drive them. As gain
generally substantially exceeds loss during early succession,
it is likely that T and Tp will decrease with time, if such
a trend exists for gain rate. Such a trend may be accentuated in communities where increasing size results in
lengthening life cycles (Drury and Nisbet 1973; Foster and
Tilman 2000). As species richness increases, Tp will also
tend to decrease and possibly to increase toward the end
of succession in communities using ephemeral resources
(e.g., corpses). These patterns have been previously observed in both plant and animal communities (e.g., Bornkamm 1981; Schoenly 1992; Myster and Pickett 1994; Foster and Tilman 2000; Chytrý et al. 2001). However, it
should be noted that studies reporting a decrease in turnover rate based on Shugart and Hett’s (1973) l do so in
error (Myster and Pickett 1994; Blatt et al. 2003). This
measure is flawed in that (1) while purporting to measure
turnover, it actually considers only loss and (2) it is defined
as the fraction of original species remaining (ln transformed) divided by the age of the community, resulting
in a mathematically trivial relationship between rate and
time (i.e., y/x vs. x) that is guaranteed to decrease in a
decelerating manner.
Species richness. Species richness (S) is defined as the
number of species present in a community and may or
may not exclude species that are temporarily absent (S and
S , respectively); S(t) is the cumulative difference between
gains and losses:
冕 冕
t
Percent turnover has been defined in a variety of ways
(Wilson and Shmida 1984; Koleff et al. 2003); I modify a
common measure of community turnover, Sørensen’s co-
(4)
Sp
t
G(t) ⫺
0
0
L(t).
(5)
Temporal Patterns in Succession Rate
Thus, elucidation of temporal patterns in G and L will
allow description of temporal patterns of S.
Here, I analyze temporal patterns in rates of species gain
(G, Gp, G , and G p ), loss (L, Lp, L, and Lp), and turnover
(T, Tp, T , and Tp) over multiple successional sequences
in a variety of community types (table 1). Specifically, I
consider plant secondary succession in worldwide locations; plant primary succession on volcanic substrates, on
sand dunes, and following a receding glacier; terrestrial
arthropod succession on defaunated mangrove islands;
and arthropod succession on corpses. Detailed descriptions of these successional seres are given in the appendix
in the online edition of the American Naturalist. For each
rate measure–successional sequence combination, I consider several mathematical forms that may potentially describe the community change rate, Y(t) (i.e., gain, loss, or
turnover), as a function of time over the course of succession. First, the null hypothesis is that Y(t) is constant:
Y(t) p Y0 .
(6)
Second, if a rate is driven by a process that changes linearly
with time, it may be described by a linear function:
Y(t) p Y0 ⫹ yt.
(7)
Third, a community change rate may display a power relationship with time:
Y(t) p Y0 ⫹ yt a.
(8)
Fourth, if a change rate depends on the number of species
present in the community (e.g., MacArthur and Wilson
1963), an exponential form is to be expected:
Y(t) p Y0 ⫹ ye .
at
(9)
Finally, in the event that a community becomes more conducive to community change as a linear function of time,
the exponential form (eq. [9]) may be modified by adding
a linear component of time:
Y(t) p Y0 ⫹ yte at.
(10)
In equations (6)–(10), Y0 refers to an initial and/or a final
value of Y, y characterizes the magnitude of the rate’s
response to time, and a is an exponent characterizing the
rate at which Y changes over time. While these mathematical forms are by no means the only ones that may be
useful in describing temporal patterns of community
change rates, they are able to describe the range of predicted temporal trends (fig. 1). For example, a negative,
783
decelerating function of community change rate with time
(fig. 1A, 1D, 1E)—as may be expected for gain and turnover rates—could be described by equations (8), (9), or
(10), with a negative a, a positive y, and a Y0 representing
background community change rates equal to those of an
equivalent steady state community. A rate that peaks and
subsequently declines (fig. 1B)—as may be expected if
succession gets a slow start—can be described by equation
(10) under the above conditions; the prominence of the
initial increase before the subsequent decline depends on
the value of a (as FaF increases, the time at which Y0 peaks
decreases).
It is important to note that temporal patterns in succession rate will be influenced by the temporal and spatial
scales of sampling. Regarding timescales, it is to be expected that the relative influence of different mechanisms
will change through time; for example, communities initially limited by dispersal or abiotic conditions will almost
invariably eventually become more strongly shaped by biotic interactions (e.g., Walker and Chapin 1987; Lichter
2000). As a result, temporal patterns in succession rate
depend on the rate of succession relative to the timescale
of measurement, and, therefore, the dynamics of species
turnover during succession must always be viewed in light
of the frequency of sampling and the duration of the study.
In terms of spatial scales, the species-area relationship (e.g.,
Arrhenius 1921) implies that KS should scale with area. In
communities limited by KS, this should result in higher
peak G and/or a more sustained period during which there
is a net accumulation of species in the community (i.e.,
G 1 L). In dispersal-limited communities, an increase in
KS will result in stronger dispersal limitation, as a smaller
proportion of KS would arrive at each time step.
Methods
A search of the literature yielded 62 successional sequences
whose data were published or available from the researcher(s) (table 1; appendix). Studies selected used longterm monitoring rather than chronosequences because
stochastic spatial species turnover and failure of chronosequences to represent identical environmental conditions
may result in artificially high community change rates.
However, I included eight chronosequences for plant primary succession, as the slow pace of this process precludes
effective long-term monitoring. Data for one primary
plant succession sequence (Surtsey, Iceland) were obtained
from long-term monitoring. Presence/absence data
through time for all species detected (as opposed to only
dominants) were required. Alternatively, when a study reported one or more of the variables of interest (G, Gp,
G , G p , L, Lp, L, Lp, T, Tp, T , and/or Tp) without providing
a presence-absence matrix, I used the reported values di-
Table 1: Successional sequences considered in this study
Habitat
Plant secondary succession:
Abandoned field
Disturbed heathland
Abandoned field
Postfire chaparral
Garden with imported soil
Disturbed grassland
Cut and burned forest
Clear-cut forest
Plant primary succession:
Lava flows on Mauna Loa
Sand dunes on lakeshore
Receding glacier
New volcanic island
Location
New Jersey
No.
successional
sequences
10
Brnenský Kraj, Czech
Republic
Kansas
California
Berlin
Niedersachsen,
Germany
Amazonas, Venezuela
Ghana
9
1
1
Hawaii
6
Michigan
Alaska
Surtsey, Iceland
1
1
1
4
2
2
1
Treatment
Last crop; season and
mode of abandonment
Disturbance type; plot
size
Patch size on landscape
Slope aspect
Soil type
Elevation
No.
observations
Final
age
Rates
available
14–23
14–23
All
8–9
8
G
5
4
8
19
6
3.7
8
19
4
4
1.83
5.2
G
All
Uhl et al. 1981
Swaine and Hall 1983
4–5
3,400
All
14
8
39
2,375
1,500
40
All
All
G
G. H. Aplet, personal
communication
Lichter 1998
Reiners et al. 1971
http://www.surtsey.is/
pp_ens/biola_lines.htm
Data source
Buell-Small succession
study
Chytrý et al. 2001
G, Gp, L, Lp, T, Tp Holt et al. 1995
All
Guo 2001
Tp
Bornkamm 1981
Tp
Bornkamm 1981
Arthropod-mangrove succession:
Defaunated mangrove islands Florida
6
Isolation; island size
14–17
295–542
All
Simberloff and Wilson
1969; Wilson and Simberloff 1969
Arthropod-carrion succession:
Rat carcass
Rabbit carcass
Paraná, Brazil
Colorado
6
5
Location; season
Elevation
16–37
9–19
16–37
23–51
G
All
Virginia
England
England
4
1
1
Season (replicated)
8–21
42
14
8–21
104
14
All
All
All
Moura et al. 2005
De Jong and Chadwick
1999
Tabor et al. 2004
Smith 1975
Chapman and Sankey 1955
Pig carcass
Fox carcass
Rabbit carcass
Note: No. observations p number of ages at which species presence-absence data were recorded; final age p age of the last observation. For plant primary and secondary succession, final ages are in
years; for arthropod-mangrove and arthropod-carrion succession, final ages are in days.
Temporal Patterns in Succession Rate
rectly. No studies meeting the above criteria were excluded
from this analysis.
For each successional sequence, I defined each time period (Dt; t 2 ⫺ t 1) as the time from one survey to the next.
For each time step, I counted species richness (S) and the
number of species gained and lost. Gain rate (G and G ;
year⫺1) and loss rate (L and L; year⫺1) were obtained by
dividing gains and losses by elapsed time (Dt). From these
values, I calculated Gp and G p (year⫺1; eq. [1]), Lp and
Lp (year⫺1; eq. [2]), and T, Tp, T , and Tp (year⫺1; eqq.
[3], [4]). Rates were calculated both including (G, Gp, L,
Lp, T, and Tp) and excluding (G , G p , L, Lp, T , and Tp)
species that temporarily disappeared.
Using Matlab 7.0.1, I used least squares regression to
fit equations (6)–(10) to each community change rate for
each successional sequence. To avoid unreasonable fits to
the data, I constrained a (eqq. [8]–[10]) between 2 and
⫺2 for equation (8) and between ⫺100/t and 5/t for equation (9), where t is the time span of the entire successional
sequence. For equation (10), Y0 and y were constrained
to be positive, and a was constrained between ⫺100/t and
0. Calculated P values for equations (7)–(10) reflect the
probability that these explain more variation in the data
(i.e., have a smaller standard deviation) than does a constant rate (eq. [6]).
Results
The summary statistics for all regressions, representing 192
mathematical model–rate measure–community type combinations, are given as both a Microsoft Excel file and a
tab-delimited ASCII file, available in the online edition of
the American Naturalist. Here, I focus on rates calculated
under the assumption that temporary absences of species
from the successional community are an artifact of sampling (e.g., Whittaker et al. 1989) and/or population stochasticity rather than a biologically meaningful event (i.e.,
G , G p , L, Lp, T , and Tp). The results for rates calculated
under the assumption that such disappearances and reappearances are biologically meaningful (i.e., G, Gp, L, Lp, T,
and Tp) are generally very similar (Excel data, tab-delimited ASCII data); I note any important differences. It
should be noted that equations (8), (9), and sometimes
(10) are usually approximately equally successful in describing the observed patterns (table 2); statistically, it
would be unreasonable to favor one of these mathematical
forms over the other(s) (McGill 2003). It must be emphasized that many of the successional sequences considered here come from the same study (see table 1; appendix)
and therefore are not statistically independent. While none
of the results presented here would differ qualitatively in
the absence of this pseudoreplication, it is important to
bear in mind that quantitative values are influenced. I note
785
any cases in which pseudoreplication affects the
conclusions.
Gain Rates
Rates of species gain consistently decline (negative slope
when fitted with a linear function) over the course of
succession (n p 54 of 55 successional sequences), although this decline is sometimes preceded by an initial
increase (n p 10 of 55). In plant secondary successional
seres, G (t) is well described (R 2 ≥ 40%) as a decelerating
decrease (eqq. [8], [9], or [10], with a positive y and a
negative a; fig. 2Ai) in all but the 1.8-year secondary forest
sere in Venezuela (Uhl et al. 1981) and the two 3.7-year
seres in postfire chaparral (Guo 2001), in which cases the
decrease was essentially linear (Excel data, tab-delimited
ASCII data). Generally, these fits are significant at a p
.05 and explain 190% of the variation (table 2). In plant
primary successional seres, G (t) is generally well described
either as a decelerating decrease (n p 4 of 9) or as a
peaked function (eq. [10]; n p 4 of 9; table 2; fig. 2Bi).
With the exception of Surtsey Island, which displays no
detectable temporal pattern, at least 70% of the temporal
variation in G (t) could be described by equations (8), (9),
or (10). For arthropods on mangrove islands, G (t) tends
to peak (eq. [10]) or decrease in an accelerating manner
(eqq. [8], [9], with negative y and positive a; fig. 2Ci),
although no fits are statistically significant (table 2) and
R2 tends to be low (averaging 15%–35%). For arthropods
on carcasses, G (t) generally decreases in a roughly linear
fashion that is alternately best described as a peak (29%),
a decelerating decline (18%), or an accelerating decline
(12%; table 2; fig. 2Di). Again, most fits are not statistically
significant (table 2), and average R2 is less than 50%.
When expressed relative to the existing community
(G p ), gain rate is well described as a decelerating decrease
(eqq. [8], [9], or [10]) for all plant secondary successional
communities, all arthropod-mangrove communities, 72%
of arthropod-carrion communities, and 50% of plant primary seres (table 2). The other 50% of plant primary
seres—the four highest-elevation sites on Mauna Loa, Hawaii—are best described as peaked functions (eq. [10];
table 2).
Loss Rates
With the exception of plant primary seres, where loss rates
often peak (table 2; fig. 2Bi), there are no consistent temporal patterns in any of the measures of species loss rate
(L, Lp, L, and Lp); these measures tend to increase or
decrease with approximately equal frequency and are rarely
significantly at P p .05 (table 2); L, however, displays an
increasing trend in all plant secondary and arthropod-
Table 2: Summary statistics for regressions relating gain rate (G ), percent gain rate (Gp ), loss rate (L), percent loss rate (Lp ), turnover rate (T ), and percent turnover rate
(Tp ) to community age
Peaked
Decelerating decrease
Equation (8)
n
Plant secondary succession:
G
Gp
L
Lp
T
Tp
Plant primary succession:
G
Gp
L
Lp
T
Tp
Arthropods on mangrove
islands:
G
Gp
L
Lp
T
Tp
Arthropods on carcasses:
G
Gp
L
Lp
T
Tp
Decreasing Well describedb by eqq. Well describedb as
(7), (8), (9), or (10)
decelerating decrease P ≤ .05
trenda
Equation (9)
Equation (10)c
R
P ≤ .05
R
P ≤ .05
2
2
23
13
10
10
13
13
100
100
0
10
100
100
100
100
80
50
100
100
(87)
(100)
(30)
(10)
(77)
(100)
91
100
0
10
92
100
83
100
…
0
77
100
92.5 (94.2)
… (98.4)
…
38.2 (…)
76.4 (80.6)
… (97.7)
83
100
…
0
77
100
92.3 (94.0)
… (99.1)
…
45.3 (…)
78.9 (83.7)
… (98.6)
87
100
…
0
77
100
9
8
7
7
8
8
89
100
86
86
100
100
89
100
100
100
100
100
(44)
(50)
(43)
(14)
(50)
(50)
44
50
14
29
50
50
33
50
0
0
38
50
90.9 (98.7)
… (99.4)
44.2 (…)
69.9 (…)
87.3 (94.4)
… (98.8)
44
38
0
0
50
38
… (91.4)
95.6 (95.9)
46.1 (…)
68.8 (…)
… (89.5)
95.3 (95.4)
11
13
0
0
13
13
6
6
6
6
6
6
100
100
0
17
50
100
33
100
50
0
0
100
(0)
(100)
(0)
(0)
(0)
(100)
0
100
0
0
0
100
…
100
…
…
…
100
…
… (91.7)
…
…
…
… (89.0)
…
100
…
…
…
100
…
… (91.6)
…
…
…
… (89.7)
17
7
7
7
7
7
100
100
71
57
100
100
59
86
0
0
43
86
(18)
(50)
(0)
(0)
(0)
(43)
18
72
0
0
0
72
6
43
…
…
…
43
67.3 (83.4)
82.3 (89.2)
…
…
…
71.8 (77.9)
6
43
…
…
…
43
65.9 (82.1)
79.4 (85.0)
…
…
…
70.8 (74.5)
2
R
91.9 (93.8)
… (99.1)
…
45.3 (…)
79.2 (82.8)
… (98.6)
Equation (10)c
Well describedb
as peaked
R
0
0
0
0
8 (0)
0
…
…
…
…
77.3 (…)
…
(0)
(0)
(14)
(0)
(0)
(0)
2
27.05 (86.28)
35.7 (86.2)
44.6 (…)
69.4 (…)
41.4 (80.4)
38.3 (84.7)
44
50
57
57
50
50
…
100
…
…
…
100
…
… (91.4)
…
…
…
… (90.4)
17 (0)
0
0
0
0
0
51.2 (…)
…
…
…
…
…
12
43
…
…
…
29
59.4 (70.0)
72.5 (75.2)
…
…
…
64.9 (82.3)
29
14
0
0
43
14
59.2 (62.8)
48.3 (…)
…
…
49.4 (…)
45.7 (…)
(18)
(0)
(0)
(0)
81.0
83.2
75.6
70.4
80.4
82.9
(…)
(…)
(71.7)
(…)
(…)
(…)
Note: Emphasis is placed on decelerating decreases (eqq. [8]–[10]) and peaked functions (eq. [10]). All values are given as percentages, and those 185% are in bold. Values in parentheses refer to regressions
that are statistically significant at a p .05 . Regressions summarized in this table may be found in a Microsoft Excel data file or a tab-delimited ASCII data file, available in the online edition of the American
Naturalist.
a
“Decreasing trend” refers to a negative slope when fitted with a linear function (eq. [7]).
b
“Well described” refers to a fit with R2 ≥ 40%.
c
Equation (10), as constrained in this study (see “Methods”), always describes a peaked function; however, the peak may occur before the time of the earliest data record (t1), in which case the function
effectively describes a decelerating decrease. If the predicted value of t1 is greater than or equal to that of t2 (the second data record), the relationship is counted as a decelerating decrease. If the predicted
value of t1 is less than or equal to that of t2 and that of the second-to-last datum greater than or equal to that of the last (such that a peak occurs within the range of data values), and if the R2 is greater
than that obtained for an exponential fit (eq. [9]), the relationship is considered peaked.
Temporal Patterns in Succession Rate
787
Figure 2: Representative temporal patterns in gain (G ; Ai, Bi, Ci, Di; black symbols, black lines), loss (L ; Ai, Bi, Ci, Di; gray symbols, gray lines),
percent turnover (Tp ; Aii, Bii, Cii, Dii), and species richness (S ; Aii, Bii, Cii, Dii) for (A) plant secondary succession on an abandoned field in New
Jersey (Buell-Small succession study, field 4), (B) plant primary succession on Michigan sand dunes, (C) arthropod succession on mangrove island
E2 in Florida, and (D) arthropod succession on a rabbit carcass in Colorado (elevation 2,786 m). Temporal patterns in G , L , and Tp are fitted with
power (eq. [8]; solid lines), exponential (eq. [9]; dashed lines), and linear # exponential (eq. [10]; dash-dotted lines) functions (statistics given in a
Microsoft Excel data file or a tab-delimited ASCII data file, available in the online edition of the American Naturalist).
mangrove sequences (n p 16, three significant at P p
.05), partially because this value artificially increases at the
end of the monitoring period when species that disappear
have decreasing time in which to reappear. Similar trends
are displayed by Lp, with a slightly greater tendency toward
decreasing; L and Lp, which do not suffer from artificial
increases at the end of the sequence, increase or decrease
with approximately equal frequency.
788
The American Naturalist
Turnover Rates
Species turnover rates, as the average of gain and loss rates
(eqq. [3], [4]), generally display temporal patterns similar
to those of species gain rates (table 2) because gain tends
to dominate early successional change (fig. 2). Rates of
species turnover consistently decline (negative slope when
fitted with a linear function) over the course of succession
(n p 31 of 34 successional sequences), although this decline is sometimes best described as a peaked rate (n p
8 of 35). In plant secondary seres, T (t) is always well
described (R 2 ≥ 40%) as a decelerating decrease (eqq. [8],
[9], or [10], with a positive y and a negative a; table 2)
in all but one of the 3.7-year seres in postfire chaparral
(Guo 2001), in which case the decrease is essentially linear
(Excel data, tab-delimited ASCII data). These fits are significant at P p .05 more than 75% of the time and explain
more than 90% of the variation (table 2). In all plant
primary seres, more than 65% of the temporal variation
in T (t) can be explained either as a decelerating decrease
(eqq. [8] or [9]; n p 4 of 8) or as a peaked function (eq.
[10]; n p 4 of 8; table 2). For arthropods on mangrove
islands, T (t) increases and decreases with equal frequency
and is never well described by any of the mathematical
functions (eqq. [7]–[10]). For arthropods on carcasses,
T (t) is sometimes well described as a peaked function
(n p 3 of 7); the remainder, which all decrease when fitted
with a linear function, never have more than 40% of their
variation explained by equations (7)–(10).
When expressed relative to the existing community
(Tp), turnover rate is well described as a decelerating decrease (eqq. [8], [9], or [10]) for all plant secondary seres,
all arthropod-mangrove seres, 72% of arthropod-carrion
seres, and 50% of plant primary seres (table 2). The other
50% of plant primary seres—the four highest-elevation
sites on Mauna Loa, Hawaii—are best described as a
peaked function (eq. [10]; table 2).
rate can be very well described as a decelerating decrease
over time (fig. 2; table 2), indicating that by far the greatest
amount of relative change occurs early in succession. Additionally, species richness generally increases rapidly during early succession, plateaus when gain and loss rates
converge, and subsequently decreases when (and if) gain
rate drops below extinction rate (fig. 2). Thus, the results
generally support the idea that community stability increases over the course of succession (e.g., Odum 1969).
These results can be used to address a couple of longstanding hypotheses regarding the nature of successional
community assembly. First, my results universally contradict a strictly Clementsian notion of discrete communities
sequentially replacing one another during succession
(Clements 1916), which would predict low turnover rates
interspersed with spikes at each community transition.
They are far more consistent with Gleason’s (1917) notion
that species appear and disappear as relatively independent
units. Second, gain is generally highest early in succession—especially in plant communities (fig. 2), thereby
lending some support to the “initial floristics” hypothesis
that a large proportion of a community’s taxa arrives early
in succession (Egler 1954; Drury and Nisbet 1973). However, a strict interpretation of the initial floristics hypothesis would require that G be virtually absent after the
earliest stages of succession, that the majority of subsequent community change be in the form of species loss,
and therefore that species richness start high and decline.
My results are not consistent with these predictions (fig.
2), indicating that the initial floristics hypothesis is not
realistic in any strict sense.
Differences in temporal patterns of community change
rates between and within community types point to the
mechanisms that may underlie these patterns (fig. 1). Specifically, my results suggest that three major factors driving
temporal patterns in succession rate are competition, abiotic limitations to the number of species a habitat can
support, and dispersal limitations.
Discussion
Despite fundamental differences in the successional communities considered, some general trends emerge. First,
the rate of species gain generally declines over the course
of succession (fig. 2; table 2; Swaine and Hall 1983; Facelli
et al. 1987; Lichter 1998; Foster and Tilman 2000; Bartha
et al. 2003). This is not surprising, given increasing competitive pressure and a decreasing pool of potential new
colonists (fig. 1; e.g., MacArthur and Wilson 1963). Meanwhile, relative to gain rates, loss rates are generally low
and not strongly temporally patterned (fig. 2). Because
colonization generally dominates early successional change
(fig. 2; Sheil et al. 2000), turnover rates decline in a manner
similar to gain rates (table 2). Generally, percent turnover
Competition
The majority of plant successional sequences display dramatically decelerating decreases in colonization rates (eqq.
[8], [9], or [10]; fig. 2Ai, 2Bi; table 2). Specifically, all but
the three shortest (!4 years) plant secondary successional
sequences display this pattern (fig. 2Ai; table 2; Swaine
and Hall 1983). Somewhat surprisingly, some plant primary seres exhibit the same patterns (table 2); specifically,
rates of community change decrease in a decelerating manner for succession on Michigan sand dunes (fig. 2Bi, 2Bii),
following a receding glacier, and at low elevations on
Mauna Loa (fig. 3A). These patterns match those predicted
for successional communities that are shaped largely by
Temporal Patterns in Succession Rate
789
Figure 3: Rates of primary succession, and temporal patterns therein, on the aa lava flows of Mauna Loa, Hawaii, differ with elevation (A), temporal
patterns in the rate of new species gain (G ) at six elevations. Note that open and filled symbols are plotted on separate Y-axes that differ by an
order of magnitude. B, Average rates of species gain (G ), loss (L ), and turnover (T ) over the first ∼3,400 years of succession at six elevations. All
rates decrease significantly (P ! .008) with increasing altitude.
competition early in succession (fig. 1A). While this study
does not directly test the idea that competition causes this
decline in G(t), such a mechanism is likely in light of the
recurring finding that G depends on the number of species
already present (e.g., MacArthur and Wilson 1963; Tilman
2004; Fargione and Tilman 2005) or, in a broader sense,
on the number of individuals and/or total biomass (e.g.
Peart 1989; Bartha et al. 2003). Competition’s role in determining G is likewise supported by the fact that mature
communities maintain relatively constant S through compensatory colonization and extinction (Goheen et al.
2005). The suggestion that G(t) in successional plant communities is shaped primarily by competition does not conflict with previous observations that early succession—
especially in primary seres—is limited by factors such as
dispersal, harsh abiotic conditions, and herbivory (e.g.,
McClanahan 1986; Wood and del Moral 1987; Tsuyuzaki
1991; Chapin et al. 1994; Fagan and Bishop 2000; Fridriksson 2000; Lichter 2000). Rather, my model (fig. 1A)
assumes that factors other than competition will determine
initial gain rates and that competition’s role in shaping
G(t) will appear when S nears KS. While it may be surprising that secondary and primary seres display similar
patterns in G(t), it must be recalled that the timescales
differ dramatically (table 1). Thus, my findings do not
imply that primary succession progresses at the same rate
as secondary succession—only that relatively early primary
seres are often habitable to many species (e.g., Walker and
Chapin 1986; Chapin et al. 1994; Lichter 2000) and accommodate far higher rates of community change than
do the later seres (fig. 2Bi, 2Bii). Thus, while direct tests
will be required to decisively prove the role of competition
in shaping G(t), my results suggest that competition becomes a strong influence on the assembly of plant communities at relatively early stages of succession.
Abiotic Limitation
Some successional communities appear to be abiotically
constrained by the development of favorable conditions
and/or accumulation of resources, which affects KS. Specifically, many species may be unable to establish themselves in a harsh environment before its modification by
time and/or facilitating species (e.g., Connell and Slatyer
1977; McAuliffe 1988). These communities, which are exemplified by primary succession at Mauna Loa’s higher
elevations, display temporally peaked rates of species gain
(fig. 3A), turnover, and even loss (eq. [10]; table 2), as
790
The American Naturalist
predicted for successional communities whose capacity to
support species (KS) develops relatively slowly (fig. 1B). In
these seres, the resources necessary to support a full set
of species are initially unavailable and appear with time
and/or facilitation, resulting in a temporally peaked rate
of community change (eq. [10]; fig. 1B). Here, colonization rates appear to be inhibited during early succession,
peak when KS is growing most rapidly, and decline again
as competition becomes limiting. The idea that abiotic
conditions affect temporal patterns in succession rate in
this manner is supported by the fact that peak colonization
rates occur progressively later (P p .06) and have progressively lower values (P p .02) at higher elevations on
Mauna Loa (fig. 3A). Note that the average rate of succession also decreases with increasing elevation (Aplet and
Vitousek 1994; Aplet et al. 1998), such that succession rates
at the highest elevation (2,434 m) are approximately an
order of magnitude less than those at the lowest elevation
(1,219 m; fig. 3B). This gradient is likely related to corresponding decreases in rates of biomass, nutrient, and
soil accumulation resulting from the colder and drier conditions at higher elevations (Vitousek et al. 1992; Aplet
and Vitousek 1994; Aplet et al. 1998). The Mauna Loa
gradient clearly exemplifies the principle that the temporal
pattern observed is influenced by the rate of succession
relative to the timescale of measurement and hints that
succession rates at low elevations might likewise show a
peak if measured on a finer timescale. Thus, in addition
to slowing the overall rate of succession (e.g., Walker and
del Moral 2003), harsh abiotic conditions appear to delay
peak rates of community change (fig. 3).
Dispersal Limitation
A number of the successional seres considered here display
temporal patterns expected for dispersal-limited communities (fig. 1D, 1E). First, in the secondary seres with
the shortest records (i.e., the secondary forest in Venezuela
and the postfire chaparral in California; table 1), G decreases in an approximately linear fashion (Excel data, tabdelimited ASCII data), indicating that the limiting effects
of KS are not yet strongly inhibiting the colonization of
new species, as appears to be occurring in other plant
secondary successional seres. Another example of a succession that is likely dispersal limited is that of Surtsey
Island (e.g., Fridriksson 2000), which is located 33 km off
the coast of Iceland. There, G displays no significant temporal patterns (Excel data, tab-delimited ASCII data), indicating that dispersal limitations may obscure temporal
patterns resulting from abiotic and/or competition-driven
trends. As this successional sequence describes only the
first 3 decades of primary succession, the fact that it fails
to display temporal patterns similar to those of other pri-
mary seres is in line with the prediction that observed
temporal patterns and their underlying mechanisms will
depend on the timescale of measurement. Additionally,
most insect successional sequences display relatively nondescript (high variance; Simberloff and Wilson 1969),
roughly linear decreases in colonization rate over the
course of succession (fig. 2Ci, 2Di; table 2) that may be
indicative of dispersal limitation (fig. 1D, 1E). Such limitation is probable, as both mangrove islands and decomposing carcasses are “islands” from the perspective of their
inhabitant species, and succession rates may therefore be
dispersal limited. This possibility is supported by the fact
that arthropod colonization rates on mangrove islands
close to colonization sources tend to decline more rapidly
than do those of distant islands (n p 6, P p .20). Additionally, gain rate decreases more slowly on large islands
(n p 6, P p .06), supporting the prediction that dispersal
limitation should be more pronounced on large islands
(data from Simberloff and Wilson 1969; Wilson and Simberloff 1969; Excel data, tab-delimited ASCII data). In the
case of arthropod succession on carcasses, the often
roughly linear decreases in gain rate may be explained
alternatively by dispersal limitation (fig. 1E) and/or as a
result of a temporally peaked pattern in resource availability (KS; fig. 1C); a combination of the two is not unlikely. Thus, spatially isolated successional communities
and/or those measured over a short timescale may display
less dramatic temporal patterns as a result of dispersal
limitation. The idea that dispersal limits the rate of community assembly agrees well with previous theory indicating (1) that succession rate depends on distance from
seed sources (e.g., McClanahan 1986) and (2) that anthropogenic increases in propagule arrival in unsaturated
communities result in increased species diversity (e.g., Foster 2001; Sax et al. 2002).
Thus, it appears that competition, abiotic limitation,
and dispersal are three major processes influencing temporal patterns in succession rate. No successional community should be assumed to be free of the influence of
any of these; rather, it is likely that all influence succession
and that their relative importance differs through time and
with community type. I propose that successional sequences may be classified according to the relative influence of these three processes and that temporal patterns
in succession rate indicate which of these processes have
the greatest influence over the timescale of interest (fig.
4). My classification of the successional communities analyzed in this study (fig. 4) is done according to the time
period for which data were available; it is to be expected
that consideration of other timescales would indicate that
other mechanisms dominate over longer or shorter time
periods; for example, all communities may be expected to
move toward competition limitation as S approaches KS—
Temporal Patterns in Succession Rate
791
Figure 4: Schematic diagram outlining three main controls on temporal patterns in succession rate: competition, dispersal, and abiotic conditions.
Successional sequences considered in this study are placed in hypothesized locations according to their temporal patterns in succession rate, abiotic
conditions, and isolation. Note that these classifications are largely dependent on the timescale of interest.
as appears to be occurring with the plant secondary seres.
While the triangular framework (fig. 4) is suggested by the
results of this study, further research will be necessary to
rigorously test it.
Two additional factors that probably affect temporal
patterns in succession rate cannot be addressed using this
data set but may form additional axes of variation (fig.
4). First, increasing body sizes may result in decreasing
succession rates (e.g., Drury and Nisbet 1973) such that
the magnitude of change in average body size may affect
the rapidity with which rates of community change decline
during succession. Second, trophic interactions undoubtedly affect successional communities. The rate of species
gain early in succession may be significantly reduced by
primary consumers (Howe and Brown 1999; Fagan and
Bishop 2000), and there is evidence that herbivores’ preference for early successional plants (e.g., Coley 1983; Godfray 1985; Fagan et al. 2004) may accelerate the replacement of early successional species by later ones (e.g., Fraser
and Grime 1999). Likewise, ant predation on sarcosa-
prophagous arthropods slows the rate at which carcasses
decay (Early and Goff 1986), thereby inevitably affecting
succession. Thus, trophic interactions have the potential
to either hinder or accelerate community change at a variety of successional stages; accordingly, their impact on
temporal patterns in succession rate cannot be easily generalized without further research.
It must be borne in mind that the appearance and disappearance of species from successional communities is
only one aspect of successional change. The numbers and
sizes of individuals representing each species may potentially change dramatically, with little concurrent change in
species occurrence. Thus, temporal patterns in rates of
community change that consider relative abundance (e.g.,
percent similarity/dissimilarity, detrended correspondence
analysis) may differ from those observed here; a comparative analysis of temporal patterns in these rates would be
instructive.
In conclusion, there is a general tendency for rates of
community change to decline in a decelerating manner
792
The American Naturalist
over the course of succession (fig. 2; table 2), a trend that
is consistent with the idea that the size of the existing
community affects species gain rates (e.g., MacArthur and
Wilson 1963; Bazzaz 1979; Peart 1989; Van der Putten et
al. 2000; Bartha et al. 2003; Tilman 2004). However, this
tendency may be modified by abiotic or dispersal limitations (fig. 4) such that temporal patterns in community
change rates peak (figs. 1B, 3A) or decline in a more linear
fashion (fig. 1E; fig. 2Ci, 2Di), respectively. While this
analysis identifies some general trends in temporal patterns
of succession rate and identifies some of the potential
mechanisms that may shape them, its more important
contribution may be to provide baseline data and quantitative methods for comparing successional communities
that differ in species composition, isolation, trophic structure, and abiotic setting.
Acknowledgments
Special thanks to J. G. Anderson for Matlab programming
assistance and to J. H. Brown for helpful comments. I am
grateful to the Buell-Small succession study (National Science Foundation [NSF] Long-Term Research in Environmental Biology grant DEB-9726992) and to G. H. Aplet
for providing data and to all researchers whose published
data were included in this analysis. Thanks also to S. Baez,
S. L. Collins, J. P. DeLong, E. P. White, the Brown and
Milne labs, and two anonymous reviewers for helpful comments. I was funded by an NSF biocomplexity grant (DEB0083422).
Literature Cited
Aplet, G. H., and P. M. Vitousek. 1994. An age-altitude matrix analysis of Hawaiian rain-forest succession. Journal of Ecology 82:137–
147.
Aplet, G. H., H. R. Flint, and P. M. Vitousek. 1998. Ecosystem development on Hawaiian lava flows: biomass and species composition. Journal of Vegetation Science 9:17–26.
Arrhenius, O. 1921. Species and area. Journal of Ecology 4:68–73.
AuClair, A., and F. Goff. 1971. Diversity relations of upland forests
in the western Great Lakes area. American Naturalist 105:499–527.
Bartha, S., S. J. Meiners, S. T. A. Pickett, and M. L. Cadenasso. 2003.
Plant colonization windows in a mesic old field succession. Applied
Vegetation Science 6:205–212.
Bazzaz, F. 1979. Physiological ecology of plant succession. Annual
Review of Ecology and Systematics 10:351–371.
Blatt, S., J. Janmaat, and R. Harmsen. 2003. Quantifying secondary
succession: a method for all sites? Community Ecology 4:141–156.
Bornkamm, R. 1981. Rates of change in vegetation during secondary
succession. Vegetatio 46:213–220.
Chapin, F. S., L. R. Walker, C. L. Fastie, and L. C. Sharman. 1994.
Mechanisms of primary succession following deglaciation at Glacier Bay, Alaska. Ecological Monographs 64:149–175.
Chapman, R., and J. Sankey. 1955. The larger invertebrate fauna of
three rabbit carcasses. Journal of Animal Ecology 24:395–402.
Chytrý, M., I. Sedlakova, and L. Tichy. 2001. Species richness and
species turnover in a successional heathland. Applied Vegetation
Science 4:89–96.
Clements, F. 1916. Plant succession: an analysis of the development
of vegetation. Carnegie Institution of Washington, Washington,
DC.
Coley, P. D. 1983. Herbivory and defensive characteristics of tree
species in a lowland tropical forest. Ecological Monographs 53:
209–233.
Connell, J., and R. Slatyer. 1977. Mechanisms of succession in natural
communities and their role in community stability and organization. American Naturalist 111:1119–1144.
De Jong, G. D., and J. W. Chadwick. 1999. Decomposition and arthropod succession on exposed rabbit carrion during summer at
high altitudes in Colorado, USA. Journal of Medical Entomology
36:833–845.
Drury, W., and I. Nisbet. 1973. Succession. Journal of the Arnold
Arboretum Harvard University 54:331–368.
Early, M., and M. L. Goff. 1986. Arthropod succession patterns in
exposed carrion on the Island of Oahu, Hawaiian Islands, USA.
Journal of Medical Entomology 23:520–531.
Egler, F. 1954. Vegetation science concepts. I. Initial floristic composition, a factor in old-field vegetation development. Vegetatio
4:412–417.
Facelli, J. M., E. Dangela, and R. J. C. Leon. 1987. Diversity changes
during pioneer stages in a subhumid pampean grassland succession. American Midland Naturalist 117:17–25.
Fagan, W. F., and J. G. Bishop. 2000. Trophic interactions during
primary succession: herbivores slow a plant reinvasion at Mount
St. Helens. American Naturalist 155:238–251.
Fagan, W. F., J. G. Bishop, and J. D. Schade. 2004. Spatially structured
herbivory and primary succession at Mount St. Helens: field surveys and experimental growth studies suggest a role for nutrients.
Ecological Entomology 29:398–409.
Fargione, J. E., and D. Tilman. 2005. Diversity decreases invasion via
both sampling and complementarity effects. Ecology Letters 8:604–
611.
Foster, B. L. 2001. Constraints on colonization and species richness
along a grassland productivity gradient: the role of propagule availability. Ecology Letters 4:530–535.
Foster, B. L., and D. Tilman. 2000. Dynamic and static views of
succession: testing the descriptive power of the chronosequence
approach. Plant Ecology 146:1–10.
Fraser, L. H., and J. P. Grime. 1999. Interacting effects of herbivory
and fertility on a synthesized plant community. Journal of Ecology
87:514–525.
Fridriksson, S. 2000. Vascular plants on Surtsey 1991–1998. Surtsey
Research 11:21–28.
Gleason, H. 1917. The structure and development of the plant association. Bulletin of the Torrey Botanical Club 44:463–481.
Godfray, H. C. J. 1985. The absolute abundance of leaf miners on
plants of different successional stages. Oikos 45:17–25.
Goheen, J. R., E. P. White, S. K. M. Ernest, and J. H. Brown. 2005.
Intra-guild compensation regulates species richness in desert rodents. Ecology 86:567–573.
Guo, Q. F. 2001. Early post-fire succession in California chaparral:
changes in diversity, density, cover and biomass. Ecological Research 16:471–485.
Holt, R. D., G. R. Robinson, and M. S. Gaines. 1995. Vegetation
dynamics in an experimentally fragmented landscape. Ecology 76:
1610–1624.
Temporal Patterns in Succession Rate
Howe, H. F., and J. S. Brown. 1999. Effects of birds and rodents on
synthetic tallgrass communities. Ecology 80:1776–1781.
Jassby, A., and C. Goldman. 1974. A quantitative measure of succession rate and its application to the phytoplankton of lakes.
American Naturalist 108:688–693.
Koleff, P., K. J. Gaston, and J. J. Lennon. 2003. Measuring beta
diversity for presence-absence data. Journal of Animal Ecology 72:
367–382.
Lichter, J. 1998. Primary succession and forest development on
coastal Lake Michigan sand dunes. Ecological Monographs 68:487–
510.
———. 2000. Colonization constraints during primary succession
on coastal Lake Michigan sand dunes. Journal of Ecology 88:825–
839.
MacArthur, R., and E. Wilson. 1963. An equilibrium theory of insular
zoogeography. Evolution 17:373–387.
McAuliffe, J. R. 1988. Markovian dynamics of simple and complex
desert plant communities. American Naturalist 131:459–490.
McClanahan, T. R. 1986. The effect of a seed source on primary
succession in a forest ecosystem. Vegetatio 65:175–178.
McGill, B. 2003. Strong and weak tests of macroecological theory.
Oikos 102:679–685.
Moura, A. O., E. L. D. Monteiro-Filho, and C. J. B. de Carvalho.
2005. Heterotrophic succession in carrion arthropod assemblages.
Brazilian Archives of Biology and Technology 48:477–486.
Myster, R. W., and S. T. A. Pickett. 1994. A comparison of rate of
succession over 18 yr in 10 contrasting old fields. Ecology 75:387–
392.
Odum, E. 1969. The strategy of ecosystem development. Science 164:
262–270.
Peart, D. R. 1989. Species interactions in a successional grassland.
II. Colonization of vegetated sites. Journal of Ecology 77:252–266.
Prach, K., P. Pysek, and P. Smilauer. 1993. On the rate of succession.
Oikos 66:343–346.
Reiners, W. A., I. A. Worley, and D. B. Lawrence. 1971. Plant diversity
in a chronosequence at Glacier Bay, Alaska. Ecology 52:55–69.
Routledge, R. 1977. On Whittaker’s components of diversity. Ecology
58:1120–1127.
Saldarriaga, J. G., D. C. West, M. L. Tharp, and C. Uhl. 1988. Longterm chronosequence of forest succession in the upper Rio Negro
of Colombia and Venezuela. Journal of Ecology 76:938–958.
Sax, D., S. Gaines, and J. Brown. 2002. Species invasions exceed
extinctions on islands worldwide: a comparative study of plants
and birds. American Naturalist 160:766–783.
Schoenly, K. 1992. A statistical analysis of successional patterns in
carrion-arthropod assemblages: implications for forensic entomology and determination of the postmortem interval. Journal of
Forensic Sciences 37:1489–1513.
Schoenly, K., and W. Reid. 1987. Dynamics of heterotrophic succession in carrion arthropod assemblages: discrete seres or a continuum of change? Oecologia (Berlin) 73:192–202.
Sheil, D., S. Jennings, and P. Savill. 2000. Long-term permanent plot
observations of vegetation dynamics in Budongo, a Ugandan rain
forest. Journal of Tropical Ecology 16:765–800.
Shugart, H., and J. Hett. 1973. Succession: similarities of species
turnover rates. Science 180:1379–1381.
793
Simberloff, D., and E. Wilson. 1969. Experimental zoogeography of
islands: the colonization of empty islands. Ecology 50:278–296.
Smith, K. 1975. The faunal succession of insects and other invertebrates on a dead fox. Entomologist’s Gazette 26:277–287.
Sørensen, T. 1948. A method of establishing groups of equal amplitude in plant sociology based on similarity of species content, and
its application to analyses of the vegetation on Danish commons.
Kongelige Danske Videnskabernes Selskabs Biologiske Skrifter 5:
1–34.
Swaine, M., and J. Hall. 1983. Early succession on cleared forest land
in Ghana. Journal of Ecology 71:601–627.
Tabor, K. L., C. C. Brewster, and R. D. Fell. 2004. Analysis of the
successional patterns of insects on carrion in southwest Virginia.
Journal of Medical Entomology 41:785–795.
Tilman, D. 2004. Niche trade-offs, neutrality, and community structure: a stochastic theory of resource competition, invasion, and
community assembly. Proceedings of the National Academy of
Sciences of the USA 101:10854–10861.
Tsuyuzaki, S. 1991. Species turnover and diversity during early stages
of vegetation recovery on the volcano Usu, northern Japan. Journal
of Vegetation Science 2:301–306.
Uhl, C., K. Clark, H. Clark, and P. Murphy. 1981. Early plant succession after cutting and burning in the upper Rio Negro region
of the Amazon basin. Journal of Ecology 69:631–649.
Van der Putten, W. H., S. R. Mortimer, K. Hedlund, C. Van Dijk,
V. K. Brown, J. Lepš, C. Rodriguez-Barrueco, et al. 2000. Plant
species diversity as a driver of early succession in abandoned fields:
a multi-site approach. Oecologia (Berlin) 124:91–99.
Vitousek, P., G. Aplet, D. Turner, and J. Lockwood. 1992. The Mauna
Loa environmental matrix: foliar and soil nutrients. Oecologia
(Berlin) 89:372–382.
Walker, L. R., and F. S. Chapin. 1986. Physiological controls over
seedling growth in primary succession on an Alaskan floodplain.
Ecology 67:1508–1523.
———. 1987. Interactions among processes controlling successional
change. Oikos 50:131–135.
Walker, L. R., and R. del Moral. 2003. Primary succession and ecosystem rehabilitation. Cambridge University Press, Cambridge.
Whittaker, R. 1975. Communities and ecosystems. Macmillan, New
York.
Whittaker, R., M. Bush, and K. Richards. 1989. Plant recolonization
and vegetation succession on the Krakatau islands, Indonesia. Ecological Monographs 59:59–123.
Wilson, E., and D. Simberloff. 1969. Experimental zoogeography of
islands: defaunation and monitoring techniques. Ecology 50:267–
278.
Wilson, M. V., and A. Shmida. 1984. Measuring beta diversity with
presence absence data. Journal of Ecology 72:1055–1064.
Wood, D. M., and R. del Moral. 1987. Mechanisms of early primary
succession in subalpine habitats on Mount St. Helens. Ecology 68:
780–790.
Associate Editor: Catherine A. Pfister
Editor: Monica A. Geber