Ecological Applications, 10(1), 2000, pp.

145–154
q 2000 by the Ecological Society of America

SEED AND SEEDLING ECOLOGY OF TREE SPECIES IN NEOTROPICAL

SECONDARY FORESTS: MANAGEMENT IMPLICATIONS
MANUEL R. GUARIGUATA1
Center for International Forestry Research (CIFOR), Box 6596 JKPWB, Jakarta 10065, Indonesia, and
Unidad de Manejo de Bosques Naturales, Centro Agronómico Tropical de Investigación y Enseñanza (CATIE),
Turrialba 7170, Costa Rica

Abstract. In spite of the growing importance of neotropical secondary forests as

sources of timber and environmental services, the baseline information needed to develop
silvicultural options is still limited. In this paper I describe interspecific patterns of seed
longevity in the soil, germination, and survival and growth of transplanted seedlings under
closed canopy of nine tree species that are common in secondary forest stands in wet,
lowland Costa Rica and most of which are timber species in the region: Cordia alliodora,
Hampea appendiculata, Jacaranda copaia, Laetia procera, Rollinia microsepala, Simarou-
ba amara, Stryphnodendron microstachyum, Trichospermum grewiifolium, and Vochysia
ferruginea. Many of these species also occur throughout the lowland neotropics. Experi-
ments were carried out in three replicate secondary-forest stands (20–30 yr old after pasture
abandonment) located at La Selva Biological Station in northeastern Costa Rica. Longevity
of experimental seed cohorts differed markedly among species, from ,3 mo (Cordia,
Hampea, Simarouba, Vochysia), to .1 yr (Stryphnodendron). Similarly, germination of
recently dispersed seeds in the understory ranged from 0% in Laetia to .75% in Cordia
and Vochysia. In contrast, seedling survival was uniformly low ( ,10% survival one year
after transplanting except for Stryphnodendron, which showed ;20% survival).
The implications of these findings for the management of secondary forest stands for
timber production are varied and depend on the species of interest. First, all study species
appear to require nearly complete canopy opening to regenerate as they show limited
capacity either to germinate or to survive as seedlings in the understory. Second, some
species that can germinate at high levels in the shade can be managed at the seedling stage
by opening up the canopy a few months after germination (e.g., Cordia, Simarouba, Voch-
ysia). Third, species that show little or no germination under closed canopy (e.g., Jacaranda,
Laetia, Rollinia) will need canopy removal simply to germinate in adequate amounts. Due
to rapid declines in seed viability and seedling survivorship, however, any canopy manip-
ulation must be performed not beyond 6 mo in order to guarantee adequate levels of soil-
stored seeds or seedlings for future stand development. Site-preparation techniques may
need implementation, given the potential of competing vegetation to interfere with seedlings
after canopy opening, as suggested by the high abundance of herbs and shrubs present in
the soil seed bank in the study stands. This ecological scenario is likely to occur in secondary
stands elsewhere in the region as secondary stands are usually located within agricultural
land. Finally, the results of this study suggest that ecological classifications of trees solely
based on light preferences for stem growth may fail to account for important differences
among species in their regeneration mode. This is of particular importance for refining
silvicultural guidelines in neotropical secondary forests.
Key words: Costa Rica; forest management; neotropical secondary forests; secondary succession;
seed germination, interspecific variation; seedling survival; soil seed bank; stand dynamics; timber
species, seed ecology; tree seed biology, silvicultural implications; tropical forest trees, life-history.

INTRODUCTION dant information exists on the pattern and process of

secondary forest succession (see review by Brown and
Recent interest in the role that neotropical secondary
Lugo [1990]), this knowledge is largely confined to the
forests can play as sources of timber and environmental
early stages of forest development while less is known
services (e.g., Wadsworth 1987, Finegan 1992, Fearn-
of ecological processes in older successional stands
side and Guimares 1996) calls for strengthening the
(Finegan 1996). Generally, the tree species that dom-
ecological basis for their management. Although abun-
inate secondary forests are apparently unable to re-
generate under their own shade, as suggested by the
Manuscript received 14 September 1998; revised 8 March
1999; accepted 11 March 1999. absence of small size classes in stem-diameter distri-
1 Present address: CATIE 7170, Turrialba 33102, Costa butions (Knight 1975), by examining changes in tree
Rica. E-mail: [email protected] species composition across a forest chronosequence
145
146 MANUEL R. GUARIGUATA Ecological Applications
Vol. 10, No. 1

(Saldarriaga et al. 1988), and by monitoring tree re- soil vary? (2) Are there differences in the capacity of
cruitment over long periods (Lang and Knight 1983). their seeds to germinate under closed canopy and of
Thus, timber-management guidelines targeted at sec- their seedlings to tolerate shade? and (3) What is the
ondary forests should include recommendations for numerical contribution of their seeds to the seed bank?
‘‘drastic’’ canopy opening to facilitate regeneration Because many of the study species are currently being
from seed. However, further research on the seed ecol- extracted selectively from primary forest in the region,
ogy of timber species is warranted for assuring long- their abundance is likely to remain low under this har-
term forest productivity in these successional forests vesting scheme. Thus, their management appears both
and for adding flexibility to management prescriptions. economically and ecologically more attractive in high-
Early studies on tropical-forest succession opened density, secondary stands (Wadsworth 1997).
the way for classifying tree species based on their light
preferences for germination, establishment, and growth METHODS
(e.g., Budowski 1965), a topic that still attracts forest
Study site
ecologists (e.g., Swaine and Whitmore 1988, Welden
et al. 1991, Clark and Clark 1992). Management of The study was conducted between late 1996 and mid-
these ecological groups (as opposed to targeting indi- 1998 at La Selva Biological Station, owned by the Or-
vidual species) has been proposed as a practical option ganization for Tropical Studies (OTS) and located in
in species-rich tropical forests (Panayotou and Ashton the Sarapiquı́ region, Heredia Province, Northeastern
1992). Yet, even among those tree species that domi- Costa Rica (108269 N, 848009 W). The natural vege-
nate the canopies of secondary forests (which usually tation is classified as ‘‘tropical wet forest’’ (sensu Hold-
include species classified as either ‘‘pioneer’’ or as ridge et al. 1975; see Hartshorn and Hammel [1994]
‘‘late successional’’ or ‘‘long-lived pioneer’’), no con- for vegetation descriptions), and annual rainfall and
sistent patterns in terms of seed dormancy and disper- temperature average 3800 mm and 268C, respectively
sibility seem to emerge based on their seed biology (Sanford et al. 1994). A mild dry season occurs during
(e.g., Dalling et al. 1997, 1998). Discerning further January–April, but no month receives on average less
differences in the regeneration from seed within this than 100 mm of precipitation. The landscape is of vol-
ecological group should help in designing sound sil- canic origin, and the topography varies from flat, al-
vicultural systems. luvial terraces to moderately undulating hills. Approx-
In Costa Rica, high deforestation rates for cattle rais- imately 63% of the total area of La Selva (;1500 ha)
ing during the 1970s, followed by pasture abandonment is covered by primary and lightly logged forest, while
due to a drop in export meat prices one decade later secondary forests account for about 11% of the total
(Butterfield 1994), have resulted in the development of area (Hartshorn and Hammel 1994).
secondary forests, particularly in the wet, Caribbean I selected three replicate secondary forest stands,
lowlands. As the agricultural frontier closes in this re- which are ;4–5 ha in size, separated from each other
gion, secondary forests have the potential to serve as ;2 km. These stands regrew after the abandonment of
a low-cost source of timber, at least at the farm level pasture that was used for not more than 10 yr after
(e.g., Schelhas 1996), especially because high-value forest conversion to agriculture. One secondary stand
timber from primary forests is diminishing (Kishor and (‘‘Holdridge,’’ ;30-yr-old at the time of study and lo-
Constantino 1993, Sánchez-Azofeifa and Quesada-Ma- cated on the easternmost part of La Selva) rests on
teo 1995). As is the case elsewhere throughout the low- alluvial soils. The other two secondary stands (‘‘Lin-
land neotropics, secondary forests in northeastern Cos- dero’’ and ‘‘Peje,’’ ;18–20-yr-old at the time of study
ta Rica are characterized by a low number of canopy- and located on central and western La Selva, respec-
tree species but in high relative dominance (Finegan tively) are underlain by residual soils derived from
and Sabogal 1988, Guariguata et al. 1997), two attri- Pleistocene lava flows (Sollins et al. 1994). The can-
butes that are assumed to facilitate their management opies of the three study stands are floristically and
(Ewel 1979). Thus the purpose of this study was to structurally similar to others in the region with com-
develop baseline information on regeneration from parable age and land-use history. Stem density and bas-
seed for nine coexisting tree species that commonly al area of individuals $10 cm in diameter at breast
occur in secondary forests in northeastern Costa Rica, height d.b.h. range between 500 and 600 stems/ha and
many of which have local commercial value for con- 20–25 m2/ha, respectively (Finegan and Sabogal 1988,
struction and furniture (Carpio 1992). Because most of Guillén 1993, Guariguata et al. 1997, Herrera and Fine-
these species range throughout Central and South gan 1997), while those of old-growth forest range be-
American moist and wet lowlands (Croat 1978, Maury- tween 350 and 550 stems/ha and 25–30 m2/ha, respec-
Lechon 1982, Faber-Langendoen 1992, Gentry 1993, tively (Lieberman and Lieberman 1994, Guariguata et
Killeen et al. 1993) the outcome of this research has al. 1997). The three study stands are surrounded at
the potential to be generalizable to similar ecological some point either by pasture or open, successional veg-
regions elsewhere. Specifically, I asked the following etation. All experiments described below were carried
questions: (1) How do patterns of seed longevity in the out at least 300 m away from any of these habitats.
February 2000 TREE REGENERATION IN SECONDARY FORESTS 147

TABLE 1. Characteristics of the nine study species at La Selva Biological Station, Costa Rica.
Nomenclature follows Wilbur (1994).

Seed
Commercial mass Dispersal
Species (Family) status† (mg) mode Fruiting peak
Cordia alliodora (Ruiz & Pavón) Oken C 40 Wind Feb.–Mar.
(Boraginaceae)
Hampea appendiculata (J. D. Sm.) NC 68 Vertebrates Dec.–Jan.
Standley
(Malvaceae)
Jacaranda copaia (Aubl.) D. Don C 9 Wind Aug.–Sep.
(Bignoniaceae)
Laetia procera (Poeppig) Eichl. C 7 Vertebrates Jul.–Aug.
(Flacourtiaceae)
Rollinia microsepala Standley C 40 Vertebrates June–July
(Annonaceae)
Simarouba amara Aubl. C 370 Vertebrates Mar.–Apr.
(Simaroubaceae)
Stryphnodendron microstachyum C 80 Vertebrates Jan.–Feb.
Poeppig & Endl.
(Mimosoideae)
Trichospermum grewiifolium (A. Rich.) NC 3 Wind Feb.–Mar.
Kosterm.
(Tiliaceae)
Vochysia ferruginea Martius C 32 Wind Aug.–Sep.
(Vochysiaceae)
† C 5 commercial; NC 5 noncommercial. Data are from Camacho and Finegan (1997).

Study species a known amount of seeds in nylon mesh bags (1 3 1

I selected the study species (see Table 1) on the basis mm mesh size) and retrieving them at fixed time in-
of their observed canopy ocurrence in the area and their tervals. For each species at each stand, a total of 25
availability of seeds at the time the study was initiated. bags, each containing 50 seeds, were buried 0–3 cm
Only Trichospermum grewiifolium is uncommon in deep at random locations within one of the 20 3 20 m
plots. A total of 3750 seeds per species were used (50
secondary stands at La Selva but it was included be-
seeds per bag 3 25 bags per stand 3 3 stands). For
cause it is abundant elsewhere in the area ( personal
each species at the time of burial, initial seed viability
observation). In contrast, the widespread tree species
was determined by placing one batch of 250 seeds in
Goethalsia meiantha (J.D.Sm.) Burret (Tiliaceae) was
shallow plastic trays filled with alluvial soil. Each batch
excluded because its freshly collected seeds were very
was left to germinate on an open bench that was pro-
difficult to germinate under controlled conditions (cf.
tected from the rain but received direct sunlight for at
González 1991), and therefore estimates of initial seed
least 3–4 h/d. They were watered every other day, and
viability proved unreliable.
observations were terminated if germination (defined
For each species (hereafter referred to by genus),
as emergence of a 2-mm-long radicle) did not occur
seeds were collected from at least four parental indi- over four consecutive weeks. Not more than 2 wk
viduals. Seed collections started in September 1996 and passed between the time of seed collection and burial
finished in July 1997 due to phenological differences of the mesh bags for any species. Values of initial seed
among species (Table 1); therefore, seeds of each of viability for each species were also used as a compar-
the study species were not sown simultaneously in the ison for assessing their ability to germinate under
field (see next section). After collection, seeds were closed canopy (see below).
pooled and either unfilled or damaged seeds were dis- Every 3 mo after the date of burial, five replicate
carded. The two following experiments described be- bags per species were retrieved from each stand. If at
low were performed within each replicate stand in two the time of retrieval any bag showed obvious signs of
parallel 20 3 20 m plots, each subdivided into a 1 3 decomposition or holes, it was discarded and another
1 m grid. Plots were separated by ;10 m, and both was retrieved (this happened only in a very few cases
were haphazardly laid out under closed canopy. The and it was independent of the species involved). Ger-
exact locations of these plots in each stand are refer- minated, rotten, or otherwise-missing seeds from with-
enced on the La Selva Biological Station 50 3 100 m in the bags were not differentiated. Seeds inside the
reserve-wide grid (data not reported). retrieved bags were germinated under the same con-
ditions as described above. No pregermination treat-
Seed longevity and seed germination in the ment was applied to any species, with the sole excep-
understory tion of the hard-coated seeds of Stryphnodendron,
This experiment was designed to assess seed lon- where a slight incision was made at the seed tip to
gevity in the soil for a period of one year by burying facilitate water entrance.
148 MANUEL R. GUARIGUATA Ecological Applications
Vol. 10, No. 1

To determine the capacity of seeds of the study spe- Natural seed banks
cies to germinate in the forest understory, I placed (con- I determined the density and composition of the seed
currently with the burial of the mesh bags) seeds inside bank by taking core samples (10-cm diameter 3 3 cm
pots under the canopy. For each species two replicates deep) at 5-m intervals along a transect that started from
per stand (50 seeds per replicate) were prepared, each the northernmost corner of the plot used for the ex-
consisting of 10-cm-diameter plastic pots filled with perimental seed banks (see previous section). To detect
the same substrate (alluvial soil) as above. Pots were potential seasonal variations in seed input to the soil,
attached to a wooden bench that was leveled 0.5 m sampling was spaced over 6 mo: May 1997 and No-
above the forest floor to avoid seed predation. Each vember 1997, roughly corresponding to the beginning
pot was placed inside a larger one (20-cm diameter), and end of the wet season. The second sampling was
in order to retain any seed that might have been carried done in a parallel fashion, 1 m away from the first. Soil
away due to splashing raindrops. Germination was fol- samples were thinly spread (#5 mm, following Dalling
lowed weekly for each species over nine consecutive et al. [1995]) in plastic trays over a 2-cm-deep heat-
months. Falling leaf litter was removed at each census. sterilized sand layer, and watered every 2–3 d in a shade
house located in a clearing (;20% full sunlight). Seed-
Seedling transplant experiment ling emergence was followed for 6 wk. Three control
trays with sterilized soil were used to check for poten-
Additional batches of seeds of six of the study spe-
tial airborne-seed contamination. Species morphotypes
cies—Cordia, Hampea, Jacaranda, Simarouba, were transplanted and allowed to grow (2–3 mo) for
Stryphnodendron, and Vochysia—were collected to further identification. Species identification of tree
raise seedlings in order to evaluate among-species seedlings was assisted by the local expert, Orlando Var-
seedling performance (growth and survival) in under- gas (OTS). All other emergents were identified at the
story conditions. Species selection was based on pre- life-form level only (sedge/herb, shrub, liana).
vious observations in the study area that suggested
some capacity to persist under closed canopy following Data analysis
germination. However, how long these seedlings are Longevity of experimental cohorts (percentage ger-
able to tolerate shade under natural conditions had re- mination) and seedling transplants (percentage surviv-
mained unexplored. Recently germinated seeds were al) was analyzed for each species with two-way ANO-
transferred to plastic containers (3-cm diameter, 15 cm VA with time and forest stand as fixed main effects.
deep) filled with alluvial soil and grown under shad- Among-species comparisons in percentage seedling
ecloth for at least 3–4 mo before transplanting. For survival were performed with one-way ANOVA only
each species in each stand, 100 seedlings were trans- at 12 mo following transplanting. Interspecific com-
planted in 10 random locations witin the adjacent 20 parisons were not possible for both the seed longevity
3 20 m plot (300 seedlings total per species). Removal and germination experiments because of species-spe-
of existing vegetation was kept to a minimum. At each cific differential seed availability over time (see above,
random location, I demarcated a 1 3 1 m subplot and Study species). Percentages were subjected to arcsine
transplanted 10 seedlings of the same species, spaced transformations to improve normality, and graphic
15–20 cm from each other. Seedlings were transplanted analyses of residuals were assessed to check for ho-
during 3 d during mid-June 1997 (one day for each mogeneity-of-error variances. To test for seasonal dif-
stand) and censused for survival at 2, 4, 7, and 12 mo. ferences within sites in germinant density from the seed
The height growth of surviving seedlings was measured bank, Mann-Whitney U tests were performed. Repli-
only at 12 mo. At each stand, seedling survival was cated goodness-of-fit G tests (Sokal and Rohlf 1981)
calculated from each of the 10 subplots as the number were performed to test both for seasonal differences in
of seedlings that were alive at census time divided by the absolute abundance of germinants of different life-
the initial number (10) of transplants. Thus, each spe- forms across stands, and for within-species differences
cies had a total of 10 observations per census time per in seed germination in the understory across stands.
stand. I assessed the understory light environment at Statistical significance was fixed at a 5 0.05. Post-hoc
the transplant plots through black-and-white (ASA comparisons follow the Tukey procedure (Neter et al.
400) hemispherical photographs (with a Nikkor 8-mm 1985). All analyses were run on SYSTAT (Wilkinson
lens) taken 1 m above the ground in forty randomly et al. 1992).
selected points 1 wk before transplanting. The photo- RESULTS
graphic negatives were analyzed with the CANOPY
software (Rich 1990), and weighted canopy openness Seed longevity and seed germination in the
(indirect site factor) was averaged at each stand. These understory
values were similar across the three stands, and aver- Over a 1-yr period, experimental seed cohorts of the
ages ranged from 5.8 to 7.7 (coefficients of variation study species showed three distinctive longevity pat-
ranged from 24% to 30%). terns in the soil (Fig. 1). Seeds of Cordia, Hampea,
February 2000 TREE REGENERATION IN SECONDARY FORESTS 149

FIG. 1. Changes in seed viability over one

year of artificial cohorts of nine tree species
typical of secondary forest (see Table 1), sown
in mesh bags in three replicate secondary stands
at La Selva Biological Station, Costa Rica. Data
are means 6 1 SE, averaged across stands. Seed
cohorts were sown at different time periods due
to phenological differences among species.

Simarouba, and Vochysia showed no capacity to remain cies, no germination was recorded in Laetia, and seeds
viable for more than 3 mo following burial; seeds were of Rollinia and Hampea germinated poorly (,25% on
found either decomposed or germinated within the average). The highest percentages were recorded in
bags. Two other species, Jacaranda and Rollinia, Cordia, Simarouba, and Vochysia, whose seeds also
reached 0% germination at 9 mo after burial. Of the germinated quickly (completed in ,1 mo). The hard-
remaining species whose artificial seed cohorts did not coated seeds of Stryphnodendron were the only ones
decline to zero 1 yr after burial, Stryphnodendron that showed prolonged germination, recorded up to 8
seemed to retain the highest germination percentage mo after sowing. The remaining species did not show
(;40%) compared to Laetia and Trichospermum (;9% any germination beyond 4 mo after sowing.
and ;5%, respectively). Only three species showed a
statistically significant effect of forest stand on per- Seedling transplant experiment
centage germination of the retrieved seeds: Jacaranda Overall, few seedlings survived under shade one year
(ANOVA, P , 0.05), Laetia (ANOVA, P , 0.01), and post-transplanting (Fig. 3); Stryphnodendron survivor-
Trichospermum (ANOVA, P , 0.05). However, no sta- ship was highest (Tukey, P , 0.01). In contrast, seed-
tistical interaction between forest stand and census time lings of Jacaranda and Simarouba reached nearly zero
was significant for any species (ANOVA, P . 0.2), survival at 7 mo after transplanting. A statistical effect
suggesting that seed decay over time was similar for of forest stand on seedling survival was detected for
all species in all three stands. Simarouba, Stryphnodendron, and Vochysia (ANOVA,
Germination percentages under closed canopy were P , 0.01) but as with the seed-longevity experiment,
also very different, ranging from 0 to .75% (Fig. 2). no statistical interaction between forest stand and cen-
Nine months after sowing seeds in pots for each spe- sus time was evident for any species (ANOVA, P .

FIG. 2. Percentage germination of the nine

study species (see Table 1) in pots under the
canopy of three replicate secondary stands at
La Selva Biological Station, Costa Rica. Values
are averaged across stands; data are means and
1 SE. Circles on top of each bar represent per-
centage germination under open conditions (see
Methods: Seed longevity and seed germination
in the understory).
150 MANUEL R. GUARIGUATA Ecological Applications
Vol. 10, No. 1

TABLE 2. Number and composition of tree seedlings that

emerged over a period of 6 wk from twenty 0.0079-m2
paired soil samples taken at 3 cm depth from three replicate
secondary stands at La Selva Biological Station, Costa
Rica. Nomenclature follows Wilbur (1994).

Sampling date
May November
Species (Family) 1997 1997
Apeiba membranacea Spruce ex Benth 1 0
(Tiliaceae)
Casearia arborea (L.C. Rich) Urban 1 0
(Flacourtiaceae)
Cecropia spp. 40 33
(Cecropiaceae)
Goethalsia meiantha (J. D. Sm.) Burret. 18 6
(Tiliaceae)
Laetia procera 16 2
Miconia multispicata Naudin 24 6
(Melastomataceae)
Rollinia microsepala — 1
Trema sp. 22 0
(Ulmaceae)
Vitex cooperi Standley 1 0
FIG. 3. The percentage of seedlings of six of the nine (Verbenaceae)
study species (see Table 1) that survived over one year at the Total 123 48
forest understory of three replicate secondary stands at La
Selva Biological Station, Costa Rica. Data are means and 1
SE , averaged across stands. Survival curves for Cordia, Si-
marouba, and Vochysia fell between Hampea and Jacaranda observed in much lower numbers. Only at the ‘‘Lin-
after 7 mo post-transplanting. Seedlings were sown during dero’’ stand was seed density higher in May 1997 (ini-
May 1997. The black line on the top right roughly corre- tial census) than 6 mo later (Mann-Whitney U, P ,
sponds to the dry season.
0.01); seed density did not differ within this time in-
terval at the ‘‘Holdridge’’ and ‘‘Peje’’ stands (Mann-
0.4). Annual height growth was also higher for Stryph- Whitney U, P 5 0.36 and 0.11, respectively). However,
nodendron seedlings compared to the other study spe- when plant life-form is taken into account, absolute
cies (Tukey, P , 0.01; Fig. 4). abundance of shrub germinants as well as of sedge/
herb germinants was not consistent across stands on
Natural seed banks both census dates (replicated goodness-of-fit G, P ,
In all three stands, mean densities of seeds that ger- 0.001 in both cases), suggesting seasonal changes in
minated from the seed bank ranged from ;1000 to seed output of these life-forms. In contrast, no seasonal
4000 seedlings/m2, but the relative abundance of tree shifts in the absolute abundance of tree and liana ger-
germinants was remarkably low, not exceeding 14% minants were statistically significant across stands (rep-
(Fig. 5). Of all tree species recorded, Cecropia spp. licated goodness-of-fit G, P 5 0.9 and P 5 0.5, re-
dominated (Table 2). Emerging seedlings from only spectively). Relative abundance of shrub germinants
two of the study species, Laetia and Rollinia, were appeared slightly lower in May than in November (45%

FIG. 4. Average seedling height of those in-

dividuals that survived one year after trans-
planting, in five of the nine study species (see
Table 1) in the forest understory of three rep-
licate secondary stands at La Selva Biological
Station, Costa Rica. Data are means 6 1 SE.
Post hoc comparisons among species in 1-yr
height growth were: Stryphnodendron . Cordia
5 Hampea 5 Simarouba 5 Vochysia. The num-
ber of surviving individuals after one year for
each species were: Cordia, n 5 8; Hampea, n
5 27; Simarouba, n 5 6; Stryphnodendron, n
5 42; Vochysia, n 5 18.
February 2000 TREE REGENERATION IN SECONDARY FORESTS 151

study between seed longevity in the soil and the ca-

pacity to germinate in the forest understory. High di-
versity in tree life histories was also recently reported
at the seed stage for a suite of pioneer tree species that
coexist in seasonally moist tropical forest in Central
Panamá (Dalling et al. 1997). The available evidence
points out that ecological classifications of trees based
solely on light preferences for growth and survival may
fail to account for important differences among species
in their early regeneration requirements (cf. Dalling et
al. 1998, see also Metcalfe and Grubb [1995]).
My findings also shed some light on factors that
either determine initial floristic composition after pas-
ture abandonment in the area, or influence tree replace-
ment patterns in secondary forest stands. On one hand,
for those species whose seeds possess very little dor-
mancy, such as Cordia, Hampea, Simarouba, and
Vochysia, site colonization may be possible only a few
weeks after seed dispersal. Species such as Jacaranda,
Laetia, and Stryphnodendron appear less constrained
by timing of seed dispersal to colonize open areas due
to their greater seed longevity. On the other hand, the
FIG. 5. Average number of seedlings from different life- ability of some of the study species to germinate and
forms that emerged from soil-stored seeds during 6 wk from grow for limited periods in the understory makes them
twenty 0.0079-m2 surface soil samples taken at 3 cm depth capable of self-replacement within their stands perhaps
in three replicate secondary stands at La Selva Biological only if canopy gaps occur. This is expected to be par-
Station, Costa Rica, during middle and late 1997.
ticularly important only beyond intermediate (20–50
yr) successional stages, however, as localized canopy
vs. 55%, respectively), while it showed an opposite gaps appear to be infrequent in secondary stands at this
trend for sedges and herbs (82% vs. 18%, respectively). age class (e.g., Saldarriaga et al. 1988). At La Selva
Biological Station, 18–20 yr old secondary stands
DISCUSSION showed smaller and more evenly distributed canopy
Seeds of the nine study species showed wide vari- openings (based on sensor- and hemispherical photo-
ation in seed longevity and their capacity to germinate graphs-based techniques) when compared to neighbor-
under closed canopy. Less pronounced interspecific dif- ing old-growth stands (Nicotra et al. 1999).
ferences, however, were found in the ability of their This study also bears important practical implica-
seedlings to survive in the understory (there was ,10% tions for the management of secondary stands for sus-
average survival 12 mo post-transplanting for five spe- tained timber production. For the study species as a
cies), perhaps with the exception of Stryphnodendron, whole, almost complete canopy removal may be nec-
which had the highest survival (;20%, 12 mo post- essary either for stimulating seed germination or for
transplanting). Although the study species can be ro- sustaining seedling growth and survival. A ‘‘mono-
bustly classified as ‘‘light demanders’’ based on qual- cyclic’’ system (i.e., where tree cutting takes place only
itative observations throughout the study area (e.g., once during the stand rotation period) that relies on
Guariguata et al. 1997), diameter growth rates (e.g., creating a future, even-aged stand by opening the mid-
Camacho and Finegan 1997), and the results presented dle and upper canopy shortly before tree harvesting
here, this study demonstrates that nevertheless they (i.e., clearcutting) is perhaps the most amenable (see
show contrasting patterns in their early regeneration Wadsworth [1997] for a detailed account of this sil-
requirements. Seed longevity in Simarouba, Cordia, vicultural system). Managers should keep in mind,
and Vochysia declined to zero in ,3 mo, but these however, that timing of this management prescription
species showed moderate-to-high capacity to germinate is critical for all species studied, and that the impor-
under closed canopy. In contrast, seeds of Laetia and tance of phenology-based knowledge to the manage-
Rollinia showed zero or minimal capacity to germinate ment of secondary forests should not be overlooked.
under closed canopy, and their germinability declined First, due to phenological constraints, the time at which
to ,10% one year after burial. Moreover, seeds of any canopy manipulation is performed will inevitably
Styphnodendron showed moderate capacity to germi- favor the regeneration of some species at the expense
nate under closed canopy, and retained the highest of others, especially if seeds are not produced annually.
(;40%) germinability of all species 1 yr after burial. Second, for those species that show little or no capacity
Thus no obvious correlation seems to emerge in this to germinate under shade (e.g., Laetia, Jacaranda, Rol-
152 MANUEL R. GUARIGUATA Ecological Applications
Vol. 10, No. 1

linia), canopy removal should be performed within 6 seeds in the study stands were sedges, herbs, and
mo post-dispersal in order to guarantee adequate shrubs. This dominance of ‘‘weedy’’ seeds is a common
amounts of viable seeds in the soil for germination. phenomenon in forest stands surrounded by agricultural
Third, for species that are able to germinate and es- or early successional vegetation (e.g., Cheke et al.
tablish as seedlings in the understory (e.g., Cordia, 1979, Young et al. 1987, Quintana-Asencio et al. 1996,
Simarouba, Vochysia), canopy removal should also not Dupuy and Chazdon 1998). Again for Cordia allio-
occur later than a few months after germination in order dora, competition by weeds reduces seedling growth
to guarantee adequate densities for management pur- and survival, even when substrate conditions in the
poses. In contrast to the above-mentioned species, how- open are suitable for seed germination (Tschinkel
ever, features of the seed ecology of Stryphnodendron 1965). Because the large majority of secondary stands
make this species not very attractive for management in the region (and across neotropical lowlands) are sur-
based on natural regeneration. The delayed germination rounded by agricultural or pasture land, research on
of the hard-coated seeds may preclude the build up of site-preparation techniques that aim to control the ef-
substantial amounts of seedlings either before or after fect of competing vegetation with tree establishment
canopy manipulation. This is supported to some extent seems justified. Interestingly, 20–100 yr old secondary
by the fact that although locally common, Stryphno- forests in Central Panamá that have been historically
dendron is rarely abundant in secondary stands in the surrounded by primary forests or spatially isolated
area, as opposed to other study species such as Cordia, from agricultural fields showed a very low proportion
Laetia, Simarouba, and Vochysia (Guillén 1993, Guari- of forbs and grasses in their seed banks (Dalling and
guata et al. 1997, Herrera and Finegan 1997). Denslow 1998).
In addition to the above recommendations, formal Decades ago, the Trinidad Tropical Shelterwood Sys-
experiments are further warranted to test directly how tem (Baur 1964) opened up the way in the neotropics
different management techniques may affect early re- for experimental forestry on secondary stands based
generation of the study species. Small-seeded, shade- on a monocyclic approach. Yet silvicultural knowledge
intolerant species such as those in this study are gen- at the seed level of many tree species still needs to be
erally negatively affected during germination and es- refined if we are to fully utilize the potential that sec-
tablishment by the presence of leaf litter (e.g., Vázques- ondary forests offer as timber sources. In particular,
Yanes et al. 1990, Guzmán-Grajales and Walker 1991). knowledge of seed biology has been largely overlooked
However, the magnitude of the affect varies with the from a forest-management perspective (at least in the
species considered and their developmental stage. In neotropics), except for developing strategies for ex situ
tropical moist forest in Central Panamá, Molofsky and conservation of germplasm. To date, most silvicultural
Augspurger (1992) found that for one of my study spe- guidelines in moist and wet neotropical forests narrow-
cies, Cordia alliodora, surface leaf litter had an effect ly focus on increasing stem-growth rates, and few take
only at postgermination emergence compared to other into account the ecological requirements at the seed
light-demanding tree species for which seed germi- and seedling stage (but see Stanley and Gretzinger
nation was more affected. Removal of litter by soil [1996]). Because in Costa Rica in particular (and even-
scarification and controlled burning enhance germi- tually in other neotropical countries) many tree species
nation and seedling survival of some timber species in that occur in secondary forests are becoming usable,
the understory of secondary forests with respect to un- the time is ripe to encourage research on monocyclic
manipulated areas (e.g., Baur 1964, Granados 1995). timber management with a solid knowledge base on
This practice may be suitable for species that can es- the seed and seedling ecology of their component spe-
tablish in the shade such as Cordia, Simarouba, and cies.
Vochysia. Extensive seedling ‘‘carpets’’ of these spe- ACKNOWLEDGMENTS
cies usually appear in the understory a few weeks after I thank Bernal Paniagua for his assistance in both seed and
data collection throughout this study and Leo Campos and
dispersal around fruiting trees (personal observation), William Miranda for assistance in hemispherical photo anal-
and their densities could be further enhanced by ap- ysis. Giselle Arroyo and Carolina González also helped dur-
plying these substrate-preparation techniques at the on- ing various stages of this work. I am grateful to the Orga-
set of seed dispersal. In contrast, species whose seeds nization for Tropical Studies, and especially to Bruce Young
for permission to carry out this study and for facilitating my
show limited germination in the shade, such as Laetia,
work at La Selva. Partial support was provided by the Swiss
Jacaranda, and Rollinia, may not be suitable for these International Cooperation (COSUDE) through CATIE, Costa
substrate-preparation techniques unless light levels are Rica, where thanks are also due to William Vázquez at the
concurrently enhanced, a prescription that may become seed laboratory. Robin Chazdon, James Dalling, Kathy Ewel,
impractical. Bryan Finegan, Rebecca Ostertag, Michelle Pinard, and two
anonymous reviewers provided helpful comments and sug-
In any case, regardless of the species targeted for gestions on an earlier version of the manuscript.
management, I hypothesize that their seedlings are LITERATURE CITED
prone to experiencing competition once the canopy is Baur, G. N. 1964. The ecological basis of rain forest man-
opened. Irrespective of time of sampling, more than agement. Forestry Commission of New South Wales, Syd-
90% of all germinants that emerged from soil-stored ney, Australia.
February 2000 TREE REGENERATION IN SECONDARY FORESTS 153

Brown, S., and A. E. Lugo. 1990. Tropical secondary forests. González, E. 1991. Recolección y germinación de semillas
Journal of Tropical Ecology 6:1–32. de 26 especies arbóreas del bosque húmedo tropical. Re-
Budowski, G. 1965. Distribution of tropical American rain vista de Biologı́a Tropical 39:47–51.
forest species in the light of successional processes. Tur- Granados, J. 1995. Efecto de los tratamientos del suelo sobre
rialba 15:40–42. la regeneración natural en un bosque secundario. Thesis.
Butterfield, R. P. 1994. The regional context: land coloni- Centro Agronómico Tropical de Investigación y Enseñanza
zation and conservation in Sarapiquı́. Pages 299–306 in L. (CATIE), Turrialba, Costa Rica.
A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Guariguata, M. R., R. L. Chazdon, J. S. Denslow, J. M. Du-
Hartshorn, editors. La Selva: ecology and natural history puy, and L. Anderson. 1997. Structure and floristics of
of a neotropical rain forest. University of Chicago Press, secondary and old-growth forest stands in lowland Costa
Chicago, Illinois, USA. Rica. Plant Ecology 132:107–120.
Camacho, M., and B. Finegan. 1997. Efectos del aprove- Guillén, L. 1993. Inventario comercial y análisis silvicultural
chamiento forestal y el tratamiento silvicultural en un de bósques húmedos secundarios en la región Huétar Norte
bosque húmedo del noreste de Costa Rica. Informe técnico de Costa Rica. Thesis. Instituto Tecnológico de Costa Rica,
295. Centro Agronómico Tropical de Investigación y En- Cartago, Costa Rica.
señanza (CATIE), Costa Rica. Guzmán-Grajales, S. M., and L. R. Walker. 1991. Differ-
Carpio, I. M. 1992. Maderas de Costa Rica. Editorial de la ential seedling responses to litter after Hurricane Hugo in
Universidad de Costa Rica, San José, Costa Rica. the Luquillo Experimental Forest, Puerto Rico. Biotropica
Cheke, A. S., W. Nanakorn, and C. Yankoses. 1979. Dor- 23:407–414.
mancy and dispersal of seeds of secondary forest species Hartshorn, G. S., and B. Hammel. 1994. Vegetation types
under the canopy of a primary tropical rain forest in North- and floristic patterns. Pages 73–89 in L. A. McDade, K. S.
ern Thailand. Biotropica 11:88–95. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors.
Clark, D. A., and D. B. Clark. 1992. Life history diversity La Selva: ecology and natural history of a neotropical rain
of canopy and emergent trees in a neotropical rain forest. forest. University of Chicago Press, Chicago, Illinois, USA.
Ecological Monographs 62:315–344. Herrera, B., and B. Finegan. 1997. Substrate conditions, fo-
Croat, T. B. 1978. Flora of Barro Colorado Island. Stanford liar nutrients and the distributions of two canopy tree spe-
University Press, Stanford, California, USA. cies in a Costa Rican secondary rain forest. Plant and Soil
Dalling, J. W., and J. S. Denslow. 1998. Soil seed bank 191:259–267.
composition along a forest chronosequence in seasonally Holdridge, L. R., W. G. Grenke, W. H. Hatheway, T. Liang,
moist tropical forest, Panamá. Journal of Vegetation Sci- and J. A. Tosi, Jr. 1975. Forest environments in tropical
ence 9:669–678. life zones. Pergamon Press, New York, New York, USA.
Dalling, J. W., S. P. Hubbell, and K. Silvera. 1998. Seed Killeen, T. J., E. Garcı́a, and S. G. Beck. 1993. Guı́a de
dispersal, seedling establishment, and gap partitioning árboles de Bolivia. Editorial Quipus, La Paz, Bolivia.
among tropical pioneer trees. Journal of Ecology 86:674– Kishor, N. M., and L. F. Constantino. 1993. Forest manage-
689. ment and competing land uses: an economic analysis from
Dalling, J. W., M. D. Swaine, and N. C. Garwood. 1995. Costa Rica. LATEN Dissemination Note 7. The World
Effect of soil depth on seedling emergence in tropical soil Bank, Washington D.C., USA.
seed-bank investigations. Functional Ecology 9:119–121. Knight, D. H. 1975. A phytosociological analysis of species-
Dalling, J. W., M. D. Swaine, and N. C. Garwood. 1997. rich tropical forest on Barro Colorado Island, Panamá. Eco-
Soil seed bank community dynamics in seasonally moist logical Monographs 45:259–284.
lowland tropical forest, Panamá. Journal of Tropical Ecol- Lang, G. E., and D. H. Knight. 1983. Tree growth, mortality,
ogy 13:659–680. recruitment, and canopy gap formation during a 10-year
Dupuy, J. M., and R. L. Chazdon. 1998. Long-term effects period in a tropical moist forest. Ecology 64:1075–1080.
of forest regrowth and selective logging on the seed bank Lieberman, M., and D. Lieberman. 1994. Patterns of density
of tropical forests in NE Costa Rica. Biotropica 30:223– and dispersion of forest trees. Pages 106–119 in L. A.
237. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hart-
Ewel, J. J. 1979. Secondary forests: the tropical resource of shorn, editors. La Selva: ecology and natural history of a
the future. Pages 53–60 in M. Chavarria, editor. Simposio neotropical rain forest. University of Chicago Press, Chi-
internacional sobre las ciencias forestales y su contribución cago, Illinois, USA.
al desarrollo de la América Tropical. CONICIT/Intercien- Maury-Lechon, G. 1982. Régenération forestière en Guyane
cia/SCITEC, San José, Costa Rica. Française: recrû sur 25 ha de coupe papetière en forêt dense
Faber-Langendoen, D. 1992. Ecological constraints on rain humide (Arbocel). Bois et Forêts des Tropiques 197:3–21.
forest management at Bajo Calima, Western Colombia. For- Metcalfe, D. J., and P. J. Grubb. 1995. Seed mass and light
est Ecology and Management 53:213–244. requirements for regeneration in Southeast Asian rain for-
Fearnside, P. M., and W. M. Guimaraes. 1996. Carbon uptake est. Canadian Journal of Botany 73:817–826.
by secondary forests in Brazilian Amazonia. Forest Ecol- Molofsky, J., and C. K. Augspurger. 1992. The effect of leaf
ogy and Management 80:35–46. litter on early seedling establishment in a tropical forest.
Finegan, B. 1992. The management potential of neotropical Ecology 73:68–77.
secondary lowland rain forest. Forest Ecology and Man- Neter, J., W. Wasserman, and M. H. Kutner. 1985. Applied
agement 47:295–321. linear statistical models. Richard D. Irwin, Homewood, Il-
Finegan, B. 1996. Pattern and process in neotropical sec- linois, USA.
ondary rain forests: the first 100 years of succession. Trends Nicotra, A. B., R. L. Chazdon, and S. V. B. Iriarte. 1999.
in Ecology and Evolution 11:119–124. Spatial heterogeneity of light and woody seedling regen-
Finegan, B., and C. Sabogal. 1988. El desarrollo de sistemas eration in tropical wet forests. Ecology 80:1908–1926.
de producción sostenible en bosques tropicales húmedos Panayotou, T., and P. S. Ashton. 1992. Not by timber alone:
de bajura: un estudio de caso en Costa Rica. El Chasquı́ economics and ecology for sustaining tropical forests. Is-
17:3–24. land Press, Washington D.C., USA.
Gentry, A. H. 1993. A field guide to the families and genera Quintana-Asencio, P. F., M. González-Espinosa, N. Ramı́rez-
of woody plants of Northwest South America. Conserva- Marcial, G. Domı́nguez-Vázquez, and M. Martı́nez-Icó.
tion International, Washington, D.C., USA. 1996. Soil seed banks and regeneration of tropical rain
154 MANUEL R. GUARIGUATA Ecological Applications
Vol. 10, No. 1

forest from milpa fields at the Selva Lacandona, Chiapas, Swaine, M. D., and T. C. Whitmore. 1988. On the definition
México. Biotropica 28:192–209. of ecological species groups in tropical rain forests. Ve-
Rich, P. M. 1990. Characterizing plant canopies with hemi- getatio 75:81–86.
spherical photographs. Remote Sensing Reviews 5:13–29. Tschinkel, H. M. 1965. Algunos factores que influyen en la
Saldarriaga, J. G., D. C. West, M. L. Tharp, and C. Uhl. 1988. regeneración natural de Cordia alliodora (Ruiz & Pav.)
Long-term chronosequence of forest succession in the up- Cham. Turrialba 15:317–324.
per Rio Negro of Colombia and Venezuela. Journal of Ecol- Vásquez-Yanes, C., A. Orozco-Segovia, E. Rincón, M. E.
ogy 76:938–958. Sánchez-Coronado, P. Huante, J. R. Toledo, and V. L. Ba-
Sánchez-Azofeifa, A. G., and C. Quesada-Mateo. 1995. De- rradas. 1990. Light beneath the litter in a tropical forest:
forestation, carbon dynamics, and sustainable mitigation effect on seed germination. Ecology 71:1952–1958.
measures in Costa Rica. Interciencia 20:396–400. Wadsworth, F. H. 1987. A time for secondary forestry in
Sanford, R. L., Jr., P. Paaby, J. C. Luvall, and E. Phillips. tropical America. Pages 189–198 in J. Figueroa, F. H.
1994. Climate, geomorphology, and aquatic systems. Pages Wadsworth, and S. Branham, editors. Management of the
19–33 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, forests of tropical America: prospects and technologies.
and G. S. Hartshorn, editors. La Selva: ecology and natural USDA Forest Service, Institute of Tropical Forestry, Puerto
history of a neotropical rain forest. University of Chicago Rico, USA.
Press, Chicago, Illinois, USA. Wadsworth, F. H. 1997. Forest production for Tropical Amer-
Schelhas, J. 1996. Land-use choice and forest patches in ica. Agriculture Handbook 710. USDA Forest Service,
Costa Rica. Pages 258–284 in J. Schelhas and R. Green- Washington, D.C., USA.
berg, editors. Forest patches in tropical landscapes. Island Welden, C. W., S. W. Hewett, S. P. Hubbell, and R. B. Foster.
1991. Sapling survival, growth, and recruitment: relation-
Press, Washington D.C., USA.
ship to canopy height in a neotropical forest. Ecology 72:
Sokal, R. R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman,
35–50.
New York, New York, USA. Wilbur, R. L. 1994. Vascular plants: an interim checklist.
Sollins, P., F. Sancho, R. Mata, and R. L. Sanford. 1994. Pages 350–378 in L. A. McDade, K. S. Bawa, H. A. Hes-
Soils and soil process research. Pages 34–53 in L. A. penheide, and G. S. Hartshorn, editors. La Selva: ecology
McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hart- and natural history of a neotropical rain forest. University
shorn, editors. La Selva: ecology and natural history of a of Chicago Press, Chicago, Illinois, USA.
neotropical rain forest. University of Chicago Press, Chi- Wilkinson, L., M. Hill, S. Miceli, P. Howe, and E. Vang.
cago, Illinois, USA. 1992. SYSTAT for Macintosh. Version 5.2. SYSTAT, Ev-
Stanley, S. A., and S. P. Gretzinger. 1996. Timber manage- anston, Illinois, USA.
ment of forest patches in Guatemala. Pages 343–365 in J. Young, K. R., J. J. Ewel, and B. J. Brown. 1987. Seed dy-
Schelhas and R. Greenberg, editors. Forest patches in trop- namics during forest succession in Costa Rica. Vegetatio
ical landscapes. Island Press, Washington, D.C., USA. 71:157–173.