TREE 2845 No.

of Pages 9

Trends in
Ecology & Evolution
Opinion

Uniting niche differentiation and dispersal

limitation predicts tropical forest succession
Daisy H. Dent1,2,3,* and Sergio Estrada-Villegas2,4

Tropical secondary forests are increasingly important for carbon sequestration Highlights
and biodiversity conservation worldwide; yet, we still cannot accurately predict Current theories of tropical forest succes-
community turnover during secondary succession. We propose that integrating sion have focused on niche differentiation
among plants but have not adequately
niche differentiation and dispersal limitation will generate an improved theoretical
addressed how dispersal limitation
explanation of tropical forest succession. The interaction between seed sources shapes succession.
and dispersers regulates seed movement throughout succession, and recent
technological advances in animal tracking and molecular analyses enable us to We describe dispersal limitation in terms
of seed source and disperser limitation;
accurately monitor seed movement as never before. We propose a framework
these two factors determine the produc-
to bridge the gap between niche differentiation and dispersal limitation. The tion of seeds throughout succession and
Source-Disperser Limitation Framework (SDLF) provides a way to better predict the arrival of seeds to suitable recruit-
secondary tropical forest succession across gradients of landscape disturbance ment sites.

by integrating seed sources and frugivore behavior. We present a framework to better under-
stand how seed sources interact with
animal abundance, behavior, and diet
Classic theories of succession do not incorporate seed dispersers preferences across the landscape to
In tropical forests, where 65–95% of tree species depend on animals for seed dispersal [1,2], determine seed movement and deposi-
the assembly of plant and animal communities is intrinsically linked during the first decades of tion during forest regeneration.

succession [3]. However, classical theories of tropical succession are founded in temperate Using technological advances in animal
ecology and focus on plant community assembly while neglecting biotic drivers that modulate tracking and molecular tools, the pro-
species turnover (e.g., seed dispersal) [4–6]. Contemporary succession theories emphasize posed framework will improve models
of tropical forest succession beyond
plant functional traits and resource availability as the main determinants of turnover and focus
niche differentiation.
on niche-based tradeoffs (see Glossary) in functional strategies (Box 1). For example, the
growth–survival tradeoff predicts that fast-growing plant species with short life spans are
replaced by slower-growing species with high survival during succession [7–9]. Importantly, most
theories propose that differences in traits among plant species explain turnover and infer that gap-
phase dynamics scale up to describe succession following large-scale human disturbances.

Contemporary succession theories may be insufficient to predict plant turnover in current human-
modified landscapes, where secondary forests now comprise over 50% of remaining tropical
forests [10], and usually have impoverished seed banks, depauperate communities of seed dis-
1
Biological and Environmental Sciences,
persers, and are often isolated from seed sources [3,11,12]. Under this scenario, we lack a unified
University of Stirling, Stirling, Scotland,
theory that incorporates plant tradeoffs and species interactions, such as seed dispersal, to explain UK
2
succession in tropical secondary forests. An improved theory of tropical forest succession will in- Smithsonian Tropical Research
Institute, Balboa, Panama
crease our capacity to predict rates of species accumulation and carbon sequestration across gra- 3
Max Planck Institute for Animal
dients of forest age and landscape connectivity, and will enable us to better quantify the role that Behavior, Konstanz, Germany
4
secondary forests can play in biodiversity conservation. Yale School of the Environment, Yale
University, New Haven, CT, USA

Uniting niche-based models with dispersal limitation

We propose that by combining niche-based models with concepts of dispersal limitation we can
provide a better theoretical explanation of tropical forest succession [9]. Niche differentiation *Correspondence:
explains how plant species exploit shifting resource availability over succession via tradeoffs daisy.h.dent@gmail.com (D.H. Dent).

Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx https://doi.org/10.1016/j.tree.2021.04.001 1
© 2021 Elsevier Ltd. All rights reserved.
 Trends in Ecology & Evolution

Box 1. Niche-based theories of tropical forest succession

Glossary
Current theories of tropical forest succession are based on niche differentiation, where resource availability changes during
Disperser limitation: insufficient
succession, thus filtering plant species according to different functional strategies [8]. In wet and moist forests (>1600 mm
numbers or types of dispersers that
rainfall yr−1), light, space, water, and nutrient availability tend to decline over succession, with light being the strongest
preclude the arrival of seeds at
driver of species turnover. A tradeoff between growth and survival can partially explain turnover; while some species focus
recruitment sites. Disperser limitation
on resource acquisition early in succession, others focus on resource conservation later in succession [67]. Acquisitive
can also occur due to changes in
species tend to have traits [e.g., high specific leaf area (SLA), low wood density] that allow them to grow fast and reproduce
behavior (e.g., animals actively avoid
rapidly (fast species [68]). Conservative species tend to have traits (e.g., low SLA, high wood density) that allow them to
areas with low forest cover).
survive in low-resource environments (slow species [68]). Recently, an additional orthogonal stature–recruitment tradeoff
Endozoochory: seed dispersal via
has shown to also shape moist secondary forests [8]. On one side, species grow to a large size, live long , and recruit in
ingestion by vertebrate animals.
high-light conditions early in succession (long-lived pioneers [8]). On the other side, species produce large numbers of off-
Extirpations: local extinctions of
spring and recruit into low-light conditions after canopy closure (short-lived breeders [8]). Changes in the relative abun-
dispersers, potentially due to predation,
dance of these four strategies (fast, slow, long-lived pioneers, and short-lived breeders) over succession are largely
hunting, or habitat loss, among other
explained by light availability [8,13]. Contrary to moist or wet forests, succession in dry forests may be fundamentally different
factors.
because water availability, not light, is the key limiting resource [69]. In dry forests (<1600 mm rainfall yr−1), water availability
Metabarcoding: using short
increases over succession, so early successional species tend to have high wood density and low SLA to tolerate drought
sequences of DNA from specified genes
and heat [67]. Changes in abiotic resources can explain shifts in species composition across ecosystems; however, turnover
of preidentified species to identify
can be unpredictable, since recruitment of individual species is dependent on species-specific reproduction events and their
multiple unknown individuals.
dispersal into suitable sites [15,34,70]. Thus, to effectively understand changes in composition during succession, it is nec-
Metabarcoding relies on high-
essary to integrate niche-based models with factors that affect the diversity and abundance of seed sources and how ani-
throughput DNA sequencing.
mals contribute to seed dispersal into regenerating forests.
Niche differentiation: due to natural
selection, species require unique
combinations of conditions and
[5,13]. However, community reassembly may be thwarted at the onset or during succession
resources to establish, survive,
because seeds are unable to reach a site and establish [9,13,14]. Thus, dispersal limitation reproduce, and coexist with other
may be a stronger determinant of species composition than niche differentiation in landscapes species. As the environment changes
that lack forest cover and suitable dispersers. In these cases, community turnover can be partially through space or time, species
composition changes because species
explained by life history tradeoffs across species, but a considerable proportion of taxonomic
track the conditions and resources that
variation can only be explained after accounting for seed dispersal [15]. maximize their fitness while avoiding
competitive exclusion.
Recent reviews cover the foundations and consequences of seed dispersal from the perspective Seed banks: viable seeds (sometimes
dormant) naturally occurring in the soil
of the plant [16,17] and describe how dispersal is affected by animal traits and behaviors [18]. profile. Seeds that are dispersed and do
However, we lack an integrated understanding of how niche differentiation and seed dispersal not germinate may become part of the
drive tropical forest succession and, importantly, how different aspects of seed dispersal initiate soil seed bank.
forest regeneration and modify successional trajectories over time. Here we propose a theoretical Seed rain: seeds that fall to the ground
either by gravity or by the action of
framework that unifies the drivers of tropical forest succession, and highlight novel technological dispersal vectors (e.g., animals or wind).
and analytical routes to bridge gaps between seed dispersal and niche differentiation as a way Seed sources: individual plants (trees,
to predict plant turnover during succession. palms, or lianas) or forest remnants that
can produce and broadcast seeds.
Source limitation: insufficiency in the
Components of dispersal limitation: seed and dispersal vector number of seeds produced to saturate
Dispersal limitation can be divided into two core components: the reproductive unit that is dis- available recruitment sites or
persed (seed or diaspore) and the vector of dispersal (animal or wind). Typically, seed limitation is insufficiency in the number of seeds that
are dispersed, even if sufficient
assessed as source limitation and dispersal limitation [16,19]. We suggest a nuanced way to ad-
recruitment sites exist. Such
dress the same phenomena by focusing on ‘source limitation’ and ‘disperser limitation’ insufficiencies can occur because of lack
because it is the interaction between the seed and the vector that results in the movement and of trees or poor pollination.
eventual establishment of a propagule [16,18,19]. Synzoochory: seed dispersal by
vertebrate animals when the seed is not
ingested.
Source limitation Tradeoffs: when one trait or strategy
Source limitation proposes that insufficient seeds are produced to saturate available recruitment cannot increase without the decrease of
sites [19]. Seed availability in secondary forests is dictated by the abundance, location, and spe- another trait or strategy, a negative
relationship between pairs of traits or
cies composition of adult trees in the surrounding landscapes [3]. Seed source limitation is higher strategies is thus created.
in landscapes with lower tree cover, increased distance to old-growth forest, and higher levels of
ongoing disturbance [13,14]. Farmed tropical landscapes often retain considerable tree cover; a
recent global analysis found that agricultural lands in Southeast Asia, Central America, eastern
South America, and West Africa retained, on average, >45% tree cover [20]. In these landscapes,

2 Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx

Trends in Ecology & Evolution

however, trees are often isolated in small patches. For example, in southern Costa Rica, nearly
50% of tree cover is composed of clusters of trees 0.05–100 hectares [21]. These trees act as
seed sources to initiate forest succession, but the composition and abundance of seed rain pro-
duced will depend on the composition of the adult community and their fecundity. Moreover,
fecundity may be reduced in areas with low forest cover due to increased distances between
conspecific mates and altered composition, abundance, and behavior of pollinators [22,23].

Proximity to remnant forests and the presence of remnant trees tend to increase the likelihood that
abandoned pastures and secondary forests recieve a diverse seed rain [24–26]. As the decision to
fell or spare trees is determined by local farmers, there is substantial variation in the number, size, and
distribution of remnant forest and trees across a landscape [27]. Critically, in fragmented landscapes,
the richness and abundance of tree seeds dispersed into abandoned pastures decrease with dis-
tance from mature forest or remnant trees [28]. In Puerto Rico, 35 tree species produced fruits,
but only 14 species were detected in the seed rain, and <1% of seeds were dispersed >4 m
from the forest edge [25]. In addition, as distance from the forest increases, the relative abundance
of vertebrate-dispersed seeds decreases more rapidly than wind-dispersed seeds [24,29]. However,
vertebrate-dispersed seeds contribute disproportionally to the diversity of seed rain [30]. Thus, sites
far from remnant forest tend to receive low numbers of seeds, with low species diversity and a high
proportion of wind-dispersed species. Combined, these factors reduce the rate of successional
change at isolated regeneration sites in the long term [26,29,31].

Although many patches in postagricultural mosaics contain remnant older trees, these patches
also include trees that have recruited after forest clearance [21,30]. Some recent studies have
found no effect of forest cover on patterns of seed rain in agricultural landscapes, potentially
because young regenerating trees and shrubs outside forest patches also contribute to seed
rain [32,33]. The life-history traits of newly recruited trees interact with landscape structure to
determine the abundance and composition of seed rain [14]. For example, early-successional
species are generally more tolerant of disturbance [34], have faster growth rates [35], produce
more seeds per tree [7], and reach reproductive maturity at smaller sizes than later successional
species [36]. Therefore, seed rain can be dominated by early-successional tree species from the
genera Ficus, Cecropia, Miconia (Neotropics), and Macaranga, Mallotus, and Alphitonia
(Paleotropics), among others [12,15,37,38]. Even if diverse communities of adult trees persist
across abandoned farmland, the seed rain produced may be biased toward a subset of pioneer
species that can rapidly establish and reproduce in degraded habitats.

Very few studies have assessed the fecundity and population viability of trees in disturbed land-
scapes, but this is essential to understand if these landscapes can support sustainable populations
of forest trees. A notable exception measured the reproductive success of 27 tree species over the
first 7 years of forest restoration in Mexico [39]. Early-successional species were most likely to
reproduce, while none of the six late-successional species reproduced during the study [39].
These results confirm the selection for fast-growing, disturbance-tolerant species, which limits
the diversity of seed rain in postagricultural mosaics. Limited regeneration of late-successional
tree species will constrain the biodiversity of regenerating forests in degraded landscapes [31].
However, the population dynamics of late-successional trees in degraded and fragmented land-
scapes are highly variable [12,40]; some studies report reduced population density [22], while
others report increases or no change in population growth compared with undisturbed forests
[23]. One common pattern is a shift in population structure toward smaller size classes in degraded
forest, which may be driven by increased mortality and fewer individuals reaching adulthood com-
pared to undisturbed forests [23]. The long-term implications of these demographic changes are
not understood but may result in reduced reproduction of late-successional species over time.

Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx 3

 Trends in Ecology & Evolution

In summary, source limitation early in succession is dictated by landscape composition and dynamics
and by the traits and demography of the tree species present [3,14]. These factors interact to create
steep gradients in the diversity of the seed rain at the landscape scale and may result in the erosion of
seed diversity over time with proliferation of rapidly reproducing pioneer species [24,39].

Disperser limitation
Seed dispersal initiates secondary forest succession in deforested tropical landscapes, with
dispersers transporting seeds from sources and depositing them in areas where seed banks are
depauperate [33]. Wind, frugivorous bats, and small birds are the key seed dispersers in early
succession (Figure 1) because they transport many seeds to regeneration sites [1,24,31]. In
degraded landscapes, wind-dispersed species tend to dominate seed rain in pastures and can
be important to initiate succession, but animal dispersers are essential for development of diverse
forests [12,31,38]. Few animal-dispersed seeds may arrive at isolated regeneration sites because
of shifts in disperser behavior, and low abundance or complete absence of dispersers [41,42].

Disperser behavior and diet preference interact with seed sources across the landscape to deter-
mine the distribution and composition of seed rain [43,44]. While dietary preference of dispersers
determines the composition of seed rain, movement behavior, roosting preferences, and gut-
passage times determine seed dispersal distances [43,45,46]. For example, frugivorous birds can
forage extensively in areas where fruits are aggregated, creating clumped patterns of seed dispersal
[47]. However, the seed rain is more diverse than expected, given the available sources, demon-
strating that birds actively select rare species, thus increasing the diversity of seed rain [47–49].

The composition of animal communities throughout succession can determine the composition
of the seed rain. For example, the abundance of bird species with large body-size and gape-
width increases with forest age, suggesting that lower numbers of large-seeded tree species
will be dispersed into young and isolated secondary forests [44,50]. In the Neotropics, bat

Trends in Ecology & Evolution

Figure 1. Potential changes in seed abundance and richness over secondary tropical forest succession. The
importance of different dispersal modes over succession are represented in the bottom panel, including wind (yellow), bat
(purple), bird (green), and nonvolant mammals (orange). The importance of each dispersal mode will vary, depending on
regional climate, landscape structure, disturbance regime, and hunting pressure; unbroken and broken lines illustrate
different potential trends [24,30,31,47,52–54,56]. Illustration by Damond Kyllo.

4 Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx

Trends in Ecology & Evolution

abundance and richness change during succession, potentially due to different maneuverability
and diet; thus, seed rain from bats also shifts throughout succession [37]. Other animal groups
show similar directional changes in composition during forest regeneration, with larger-bodied,
habitat and dietary specialists increasing in abundance over time [11]. Changes in animal compo-
sition have important implications for the diversity of seed rain during succession [51].

Large seed dispersers (primates and large birds) are often limited in young regenerating forests [41],
whereas small birds and bats are key dispersers in these habitats. In the Neotropics, frugivorous
bats are the primary dispersers of many small-seeded early-successional plants [37,52]. Neotropical
bats often fly across open areas [53,54] while defecating seeds in flight or carrying fruits back to
roosts to eat [21,55,56]. These behaviors make bats key dispersers in early-successional forests
in the Neotropics, whereas birds are more important dispersers later in succession [52]. In the
Paleotropics, bat-dispersed plants rarely dominate early-successional forests, although bats can
be important in long-distance dispersal (Box 2). The most important seed dispersers in deforested
landscapes in tropical Asia are small birds such as Bulbuls (Pycnonotidae) [41]. Across the tropics,
birds disperse a greater diversity of tree species than bats and tend to increase in abundance
and diversity as forests mature; thus, seed dispersal by birds is key to accelerating successional
change (Figure 1).

Local species extirpations can have profound consequences for seed dispersal during succes-
sion. Hunting is pervasive in tropical forests, and coupled with habitat loss, it can drastically diminish
the abundance of dispersers [57,58]. A recent meta-analysis showed that defaunation slowed forest
regeneration, primarily due to local extirpation of primates and birds [57]. For instance, extirpation of
frugivorous birds in Guam has created monodominant degraded secondary forests [59], decreased
seedling recruitment up to 92% [60], and reduced the richness of seed rain into regeneration sites
[61]. Although a wide diversity of frugivores may feed on a single tree species [1], a small proportion
of these animal species may be responsible for a disproportionate share of fruit consumption and

Box 2. Technologies for monitoring seed movement

Frugivore tracking technology: to predict seed dispersal into regenerating forests, we need to understand the movement
of frugivores in degraded landscapes. Recent technological advances in radio and global positioning system (GPS) track-
ing enable us to observe directly the movement patterns of individuals and how these behaviors change over space and
time. While early tracking studies in tropical regions focused on larger frugivores in undisturbed forests [71], recent ad-
vanced tracking devices have allowed estimation of high-resolution movement trajectories across a wide range of taxa
in diverse habitats [72,73]. Tracking data can be paired with landscape composition data and seed-handling/gut-passage
times to estimate species-specific seed dispersal distances [43,45,46]. For example, Neotropical fruit bats can functionally
connect fragmented tropical landscapes by linking undisturbed and degraded forest patches via long-distance foraging,
ranging from 0.2 to 2.5 km [56]; while in Africa, the straw-colored fruit bat, Eidolon helvum, flies, on average, 67.1 km be-
tween colonies and foraging trees and disperses seeds up to 95 km [45,46]. These studies demonstrate how new tracking
technologies can yield nuanced seed dispersal kernels for a wide range of animals [18].

Molecular techniques: modern molecular techniques can inform seed dispersal distances, confirm or discover new
dispersal vectors, and establish gene flow among forest stands [74]. Microsatellites and SNPs via next-generation
sequencing can be used to calculate dispersal distances by establishing the paternity among seedlings and potential
parents [75]. For example, in secondary forests in Costa Rica, parentage analysis of the palm Iriartea deltoidea confirmed
that >70% of seedlings were dispersed >50 m and parent trees were often not within a 0.5-hectare radius of seedlings
[76]. Microsatellites can determine whether genetic diversity of dispersed seeds increases during succession [77]. Other mo-
lecular techniques, such as metabarcoding, can be used to describe frugivore diets and to identify dispersal vectors.
Metabarcoding scats can describe endozoochory and has shown that animals consume more species than previously
recorded and move seeds far from parental sources [78]. Moreover, using smaller primers can help determine how seed rain
is affected by differential habitat use [79]. Synzoochory can go unnoticed when seed dispersal is quantified only from scats,
but it can be assessed by extracting animal DNA from dispersed seeds. Synzoochory is common in birds and bats, where
seeds are transported to temporary roosts and regurgitated or defecated. Traditional and novel molecular techniques, inte-
grated with information of gut-passage time and movement data, will determine how dispersers influence forest
regeneration by linking sources to establishment sites.

Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx 5

 Trends in Ecology & Evolution

associated seed-dispersal service [62]. Therefore, the loss of any single animal species can have
serious implications for the dispersal of their preferred food plants. Compared to undisturbed
forest, forest patches within human-dominated landscapes have reduced animal densities due to
hunting and habitat loss, but, in addition, the remaining dispersers may have altered activity patterns
due to fear of humans and domesticated animals (e.g., dogs) with knock-on effects for seed
dispersal.

In sum, the diversity, community composition, and behavior of dispersers can limit forest regenera-
tion in postagricultural landscapes [29,63]. Seed rain in secondary forests is also affected by the de-
velopment of the forest itself due to increasing structural complexity providing more habitat for larger
seed dispersers and trees within the stands reaching reproductive maturity [34].

Establishment limitation
Seedling establishment may limit forest regeneration if the site is unsuitable for seed germination
and early seedling survival [64]. Establishment is dictated by the biotic and abiotic conditions of
the microsite and the seed and seedling traits of tree species [19]. For example, we see selection
for fast-growing species dispersed into open pastures because they are more likely than slower-
growing species to outcompete grasses and shrubs for available light [29,52]. The high light and
low moisture found in open areas also restricts establishment, particularly for species adapted to
the low-light conditions of forest understory (Box 1). Environmental filtering of dispersed seeds
during seedling establishment is accounted for by current niche-based models of succession
(Box 1), while seed dispersal is not properly integrated in successional theory.

Moving forward using the Source-Disperser Limitation Framework

Given that tropical forest succession hinges on overcoming limitations in seed sources and dis-
persers, we propose the Source-Disperser Limitation Framework (SDLF) to better understand
succession in tropical agricultural landscapes. The central tenet of the SDLF is that the interaction
between seed sources and dispersers controls seed movement to sites starting or undergoing
regeneration (Figure 2A). Seed movement from sources to microsites by dispersers is the primary
limitation, since seedling establishment is contingent on this initial dispersal event. Seed sources
can be limited by landscape structure and dynamics, and by plant traits and pollination success.
Simultaneously, dispersers can be limited by their population dynamics, and their behavior can
preclude movement of seeds where forest cover is low. The interaction between seed sources
and dispersers is what regulates seed movement throughout succession. The SDLF can also
be used to explain plant turnover across gradients of landscape structure (Figure 2B). The
SDLF is ideal for longitudinal studies at the patch scale, where movement of dispersers can be
traced, seed input from dispersal monitored, and turnover of plants and dispersers correlated
(see Outstanding questions).

Observational and experimental data can inform models that use the SDLF to explain and predict
succession. Plot data, remote sensing, animal monitoring, and data on fruit availability and frugi-
vores can help determine limitation of sources and dispersers. Seed rain exclusions (night versus
day) across chronosequences will determine the relative roles of different vectors (birds versus
bats versus terrestrial mammals). Data on animal movement and deposition into deforested
areas, coupled with vector identification with molecular tools, is essential to understand how land-
scape structure affects seed movement and probable establishment (Box 2). Experiments using
artificial perches coupled with controls where seed rain is simulated via planting seeds from the
landscape can show the contribution of dispersers to establishment and plant turnover, relative
to expected seed rain [48,65]. Path analyses with fixed and random effects [66] can then be
used to test how nested factors (traits nested within seed sources) affect traveled distances,

6 Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx

Trends in Ecology & Evolution

Outstanding questions
How does tree fertility change over
forest succession? Do slow species
have particularly low fertility compared
with fast species early in succession?

What is the relative importance of different

dispersal modes over forest succession?
How does such importance change with
other variables (such as climate, hunting
pressure, and landscape connectivity)?

Are parentage analyses across multiple

successional forest stands able to
identify the parent tree? Are analyses
of genetic structure and diversity
a suitable proxy to evaluate seed
dispersal in cases where parentage
analysis is unfeasible?

Can we effectively characterize the diets

Trends in Ecology & Evolution of frugivores using metabarcoding of
Figure 2. Source-Disperser Limitation Framework (SDLF) to study succession in secondary tropical forests. feces?
(A) Colored shapes represent organisms; white boxes represent processes where seeds are involved; and curly brackets
Can descriptions of diet be integrated
list factors that determine sources and dispersers. Movement of seeds by animals into abandoned farmland becomes a
with tracking data to describe dispersal
primary constraint on forest succession and may explain changes in species composition beyond niche differentiation.
kernels for multiple plant species?
Movement depends on seed sources and disperser limitation. Seed sources are contingent on plant life-history strategies,
traits, abundance of pollinators, and tree fecundity. Dispersers are contingent on behavior, shifts in abundance and
To what extent are total dispersal kernels
composition, or local extirpation. Interactions between sources and dispersers are mediated by fruit selection via fruit traits
able to predict species composition after
and by seasonal dietary breadth of dispersers. Once seeds have been moved and deposited, seedlings establish, and
accounting for differential plant functional
turnover is mostly explained by niche differentiation (i.e., microsite conditions, seed traits, and life-history strategies; see
traits in the context of tropical forest
Box 1 in the main text). (B) The SDLF may explain community turnover if seed source and disperser limitations are
succession?
integrated. (i) If seed source limitation is higher relative to disperser limitation (sizes of colored circles), changes in plant
composition may be rapid because available seeds are moved into abandoned sites [59]. (ii) If disperser limitation is higher
How does the movement of different
relative to seed source limitation, changes in plant composition may be slow because seeds do not reach abandoned
frugivores shift across gradients of
sites. Established plants feedback as seed sources through succession. In nature, trajectories are more complex because
forest connectivity?
both source and disperser limitations change simultaneously. Overall, the interplay between seed sources and dispersers
shapes plant composition as sites gain complexity during succession. How does fear of humans and
domesticated animals affect behavior
and movement of animal dispersers in
deposition probability, or establishment. To determine how animal dispersal contributes to degraded landscapes? Does this
demography, integral projection models [23] could be used that incorporate establishment landscape of fear reduce movement
of seeds from areas of high tree cover
data where animal dispersal has and has not occurred. Finally, using probability distributions to areas of low tree cover?
with seed rain data, or seed rain data coupled with animal movement and paternity analyses
(Box 2), it would be possible to establish total dispersal kernels across chronosequences [17]. Are there distinct biogeographical
differences in patterns of seed dispersal
Overall, the SDLF enables the integration of observational data, novel experiments, and sophisti-
and succession between the Neo- and
cated statistical models to describe and predict succession in postagricultural landscapes. Paleotropics? How do regional dif-
ferences in tree community compo-
Concluding remarks sition and plant–animal interactions
influence rates and trajectories of
The unification of niche differentiation and dispersal limitation can provide a mechanistic explana-
secondary forest succession?
tion of succession in the tropics. The study of seed dispersal has a long history in the context of
tropical secondary forests [1,2,16,29,31]; here we are proposing an integration of seed dispersal What are the best sampling strategies
limitation with niche differentiation to better predict tropical forest succession. Specifically, we and experimental designs to test
the seed Source-Disperser Limitation
need a mechanistic approach to understand what limits seed arrival (seed sources and dispersers) Framework?
and how those limitations shift over time (Figure 1). Integrating niche-based models with source
and disperser limitation will allow us to better distinguish stochastic from deterministic processes. What are the most suitable modeling
approaches to evaluate the combined
Collaborative projects that bring together functional plant ecology, animal behavior, movement
effect of seed source and disperser
ecology, quantitative spatial ecology, and theoretical ecology are ideal to update our theories of

Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx 7

 Trends in Ecology & Evolution

tropical forest succession (see Outstanding questions). The SDLF unites key aspects of these five limitation on plant community turnover
during succession?
fields of ecology and is ideal to use recent technological and methodological advances (Box 2) to
fully understand and predict tropical forest succession. Does local plant species diversity
increase as a function of increasing
seed dispersal through time? Is there
Acknowledgments
a positive or negative relationship
We thank Luke Browne, Liza Comita, Meg Crofoot, Dina Dechmann, Ana Cristina Palma, and Pablo Stevenson for com-
between the traits that promote
ments on earlier drafts of this article. We also thank three anonymous reviewers and Andrea Stephens; their comments dispersal and other components of
helped us improve the ideas proposed in this opinion article. D.D. was supported by a Resesearch Fellowship from the Al- fitness (such as competitive ability)?
exander von Humboldt Foundation, Germany.
Are differences in the trajectories of plant
community composition in wet and dry
Declaration of interests
forests explained by shifting limitations in
The authors have no interests to declare.
seed sources and dispersers through
time?
References
1. Muller-Landau, H.C. and Hardesty, B.D. (2005) Seed dispersal 20. Zomer, R.J. et al. (2016) Global tree cover and biomass carbon
of woody plants in tropical forests: concepts, examples and on agricultural land: the contribution of agroforestry to global and
future directions. In Biotic Interactions in the Tropics: Their national carbon budgets. Sci. Rep. 6, 29987
Role in the Maintenance of Species Diversity (Burslem, 21. Mendenhall, C.D. et al. (2016) Quantifying and sustaining biodi-
D.F.R.P. et al., eds), pp. 267–309, Cambridge University Press versity in tropical agricultural landscapes. Proc. Natl. Acad. Sci.
2. Howe, H.F. and Smallwood, J. (1982) Ecology of seed dispersal. U. S. A. 113, 14544–14551
Annu. Rev. Ecol. Syst. 13, 201–228 22. Ghazoul, J. (2005) Pollen and seed dispersal among dispersed
3. Arroyo-Rodríguez, V. et al. (2017) Multiple successional path- plants. Biol. Rev. Camb. Philos. Soc. 80, 413–443
ways in human-modified tropical landscapes: new insights 23. Zambrano, J. and Salguero-Gómez, R. (2014) Forest fragmenta-
from forest succession, forest fragmentation and landscape tion alters the population dynamics of a late-successional tropical
ecology research. Biol. Rev. Camb. Philos. Soc. 92, 326–340 tree. Biotropica 46, 556–564
4. Clements, F.E. (1916) Plant Succession: An Analysis of the 24. Ingle, N.R. (2003) Seed dispersal by wind, birds, and bats between
Development of Vegetation, Carnegie Institution of Washington Philippine montane rainforest and successional vegetation.
5. Tilman, D. (1990) Constraints and tradeoffs: toward a predictive Oecologia 134, 251–261
theory of competition and succession. Oikos 58, 3–15 25. Cubiña, A. and Aide, T.M. (2001) The effect of distance from forest
6. Bazzaz, F.A. and Pickett, S.T.A. (1980) Physiological ecology of edge on seed rain and soil seed bank in a tropical pasture.
tropical succession: a comparative review. Annu. Rev. Ecol. Biotropica 33, 260–267
Syst. 11, 287–310 26. Crouzeilles, R. and Curran, M. (2016) Which landscape size best
7. Coomes, D.A. and Grubb, P.J. (2003) Colonization, tolerance, predicts the influence of forest cover on restoration success? A
competition and seed-size variation within functional groups. global meta-analysis on the scale of effect. J. Appl. Ecol. 53,
Trends Ecol. Evol. 18, 283–291 440–448
8. Rüger, N. et al. (2020) Demographic trade-offs predict tropical 27. Tarbox, B.C. et al. (2018) Divergent rates of change between
forest dynamics. Science 368, 165–168 tree cover types in a tropical pastoral region. Landsc. Ecol. 33,
9. Vellend, M. (2016) The Theory of Ecological Communities, 2153–2167
Princeton University Press 28. Blackham, G.V. et al. (2013) Seed rain into a degraded tropical
10. Food and Agriculture Organization. of the United Nations (FAO) peatland in Central Kalimantan, Indonesia. Biol. Conserv. 167,
(2015) Global Forest Resources Assessment 2015: How Are 215–223
the World’s Forests Changing? FAO 29. Holl, K.D. (1999) Factors limiting tropical rain forest regeneration
11. Parry, L. et al. (2007) Large-vertebrate assemblages of primary in abandoned pasture: seed rain, seed germination, microcli-
and secondary forests in the Brazilian Amazon. J. Trop. Ecol. mate, and soil. Biotropica 31, 229–242
23, 653–662 30. Piotto, D. et al. (2019) Successional, spatial, and seasonal
12. Palma, A.C. et al. (2021) Dispersal and recruitment limitations in changes in seed rain in the Atlantic forest of southern Bahia,
secondary forests. J. Veg. Sci. 32, e12975 Brazil. PLoS One 14, e0226474
13. Chazdon, R.L. (2014) Second Growth: The Promise of Tropical 31. Martínez-Garza, C. et al. (2009) Seed rain in a tropical agricultural
Forest Regeneration in an Age of Deforestation, University of landscape. J. Trop. Ecol. 25, 541–550
Chicago Press 32. Reid, J.L. et al. (2015) Seed dispersal limitations shift over time in
14. Werden, L.K. et al. (2020) Effects of dispersal- and niche-based tropical forest restoration. Ecol. Appl. 25, 1072–1082
factors on tree recruitment in tropical wet forest restoration. Ecol. 33. Howe, H.F. et al. (2010) Early seed fall and seedling
Appl. 30, e02139 emergence: precursors to tropical restoration. Oecologia
15. Dent, D.H. et al. (2013) Secondary forests of central Panama 164, 731–740
increase in similarity to old-growth forest over time in shade 34. Norden, N. et al. (2009) Resilience of tropical rain forests: tree
tolerance but not species composition. J. Veg. Sci. 24, community reassembly in secondary forests. Ecol. Lett. 12,
530–542 385–394
16. Schupp, E.W. et al. (2019) Intrinsic and extrinsic drivers of intra- 35. Martínez-Garza, C. et al. (2013) Are functional traits good predic-
specific variation in seed dispersal are diverse and pervasive. tors of species performance in restoration plantings in tropical
AoB Plants 11, plz067 abandoned pastures? For. Ecol. Manag. 303, 35–45
17. Rogers, H.S. et al. (2019) The total dispersal kernel: a review and 36. Wright, S.J. et al. (2005) Reproductive size thresholds in tropical
future directions. AoB Plants 11, plz042 trees: variation among individuals, species and forests. J. Trop.
18. Zwolak, R. (2018) How intraspecific variation in seed- Ecol. 21, 307–315
dispersing animals matters for plants. Biol. Rev. Camb. Philos. 37. Muscarella, R. and Fleming, T.H. (2007) The role of frugivorous
Soc. 93, 897–913 bats in tropical forest succession. Biol. Rev. 82, 573–590
19. Clark, C.J. et al. (2007) Are plant populations seed limited? A cri- 38. Galindo-González, J. et al. (2000) Bat- and bird-generated seed
tique and meta-analysis of seed addition experiments. Am. Nat. rains at isolated trees in pastures in a tropical rainforest.
170, 128–142 Conserv. Biol. 14, 1693–1703

8 Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx

Trends in Ecology & Evolution

39. Caughlin, T.T. et al. (2019) Demographic costs and benefits of 60. Rogers, H.S. et al. (2017) Effects of an invasive predator
natural regeneration during tropical forest restoration. Ecol. cascade to plants via mutualism disruption. Nat. Commun. 8,
Lett. 22, 34–44 14557
40. Zambrano, J. et al. (2014) Effects of forest fragmentation on the 61. Wandrag, E.M. et al. (2017) Seed dispersal increases local
recruitment success of the tropical tree Poulsenia armata at Los species richness and reduces spatial turnover of tropical tree
Tuxtlas, Veracruz, Mexico. J. Trop. Ecol. 30, 209–218 seedlings. Proc. Natl. Acad. Sci. U. S. A. 114, 10689–10694
41. Corlett, R.T. (2017) Frugivory and seed dispersal by vertebrates 62. McConkey, K.R. et al. (2018) Different megafauna vary in their
in tropical and subtropical Asia: an update. Glob. Ecol. Conserv. seed dispersal effectiveness of the megafaunal fruit Platymitra
11, 1–22 macrocarpa (Annonaceae). PLoS One 13, e0198960
42. Beltrán, L.C. and Howe, H.F. (2020) The frailty of tropical resto- 63. Caughlin, T.T. et al. (2016) When does seed limitation matter for
ration plantings. Restor. Ecol. 28, 16–21 scaling up reforestation from patches to landscapes? Ecol. Appl.
43. Rehm, E. et al. (2019) Animal movement drives variation in seed 26, 2439–2450
dispersal distance in a plant-animal network. Proc. Biol. Sci. 64. Reid, J.L. and Holl, K.D. (2013) Arrival ≠ survival. Restor. Ecol.
286, 20182007 21, 153–155
44. Velho, N. et al. (2012) Shifts in community structure of tropical trees 65. Freeman, A.N.D. et al. (2021) Enhancing bird-mediated seed
and avian frugivores in forests recovering from past logging. dispersal to increase rainforest regeneration in disused pasture –
Biol. Conserv. 153, 32–40 a restoration experiment. For. Ecol. Manag. 479, 118536
45. Abedi-Lartey, M. et al. (2016) Long-distance seed dispersal by 66. Lefcheck, J.S. (2016) piecewiseSEM: Piecewise structural equa-
straw-coloured fruit bats varies by season and landscape. tion modelling in R for ecology, evolution, and systematics.
Glob. Ecol. Conserv. 7, 12–24 Methods Ecol. Evol. 7, 573–579
46. van Toor, M.L. et al. (2019) Linking colony size with quantitative 67. Lohbeck, M. et al. (2013) Successional changes in functional
estimates of ecosystem services of African fruit bats. Curr. Biol. composition contrast for dry and wet tropical forest. Ecology
29, R237–R238 94, 1211–1216
47. Carlo, T.A. and Morales, J.M. (2016) Generalist birds promote 68. Reich, P.B. (2014) The world-wide ‘fast–slow’ plant economics
tropical forest regeneration and increase plant diversity via rare- spectrum: a traits manifesto. J. Ecol. 102, 275–301
biased seed dispersal. Ecology 97, 1819–1831 69. Quesada, M. et al. (2009) Succession and management of tropical
48. González-Castro, A. et al. (2019) How does avian seed dispersal dry forests in the Americas: review and new perspectives. For.
shape the structure of early successional tropical forests? Funct. Ecol. Manag. 258, 1014–1024
Ecol. 33, 229–238 70. Rozendaal, D.M.A. et al. (2019) Biodiversity recovery of Neotropical
49. Gopal, A. et al. (2020) Forest cover and fruit crop size differentially secondary forests. Sci. Adv. 5, eaau3114
influence frugivory of select rainforest tree species in Western 71. Poonswad, P. and Tsuji, A. (1994) Ranges of males of the Great
Ghats, India. Biotropica 52, 871–883 Hornbill Buceros bicornis, Brown Hornbill Ptilolaemus tickelli and
50. Galetti, M. et al. (2013) Functional extinction of birds drives rapid Wreathed Hornbill Rhyticeros undulatus in Khao Yai National
evolutionary changes in seed size. Science 340, 1086–1090 Park, Thailand. Ibis 136, 79–86
51. Dent, D.H. and Wright, J.S. (2009) The future of tropical species 72. Wikelski, M. et al. (2010) Large-range movements of neo-
in secondary forests: a quantitative review. Biol. Conserv. 142, tropical orchid bees observed via radio telemetry. PLoS
2833–2843 One 5, e10738
52. de la Peña-Domene, M. et al. (2014) Roles of birds and bats in 73. Cárdenas, S. et al. (2020) Movement patterns and habitat
early tropical-forest restoration. PLoS One 9, e104656 preference of Oilbirds (Steatornis caripensis) in the southern
53. Gorchov, D.L. et al. (1993) The role of seed dispersal in the Andes of Colombia. Avian Conserv. Ecol. 15, 5
natural regeneration of rain forest after strip-cutting in the 74. Godoy, J.A. and Jordano, P. (2001) Seed dispersal by animals:
Peruvian Amazon. Vegetatio 107, 339–349 exact identification of source trees with endocarp DNA
54. Estrada-Villegas, S. et al. (2007) Seed dispersal by bats in a microsatellites. Mol. Ecol. 10, 2275–2283
mountain forest edge. Ecotropicos 20, 1–14 75. Ashley, M.V. (2010) Plant parentage, pollination, and dispersal:
55. Baldwin, J.W. et al. (2020) Defensive fruit metabolites obstruct how DNA microsatellites have altered the landscape. Crit. Rev.
seed dispersal by altering bat behavior and physiology at multiple Plant Sci. 29, 148–161
temporal scales. Ecology 101, e02937 76. Sezen, U.U. et al. (2009) Proximity is not a proxy for parentage in
56. Ripperger, S.P. et al. (2015) Frugivorous bats maintain functional an animal-dispersed Neotropical canopy palm. Proc. Biol. Sci.
habitat connectivity in agricultural landscapes but rely strongly 276, 2037–2044
on natural forest fragments. PLoS One 10, e0120535 77. Browne, L. and Karubian, J. (2018) Habitat loss and fragmenta-
57. Gardner, C.J. et al. (2019) Quantifying the impacts of tion reduce effective gene flow by disrupting seed dispersal in a
defaunation on natural forest regeneration in a global meta- neotropical palm. Mol. Ecol. 27, 3055–3069
analysis. Nat. Commun. 10, 4590 78. Aziz, S.A. et al. (2017) Elucidating the diet of the island flying fox
58. McConkey, K.R. and O’Farrill, G. (2016) Loss of seed dispersal (Pteropus hypomelanus) in Peninsular Malaysia through Illumina
before the loss of seed dispersers. Biol. Conserv. 201, 38–49 next-generation sequencing. PeerJ 5, e3176
59. Caves, E.M. et al. (2013) Natural experiment demonstrates that 79. González-Varo, J.P. et al. (2014) Who dispersed the seeds? The
bird loss leads to cessation of dispersal of native seeds from use of DNA barcoding in frugivory and seed dispersal studies.
intact to degraded forests. PLoS One 8, e65618 Methods Ecol. Evol. 5, 806–814

Trends in Ecology & Evolution, Month 2021, Vol. xx, No. xx 9