Succession in the Brazilian Amazon depends on prior land-use history. Abandoned clearcuts become dominated by Cecropia trees and exhibit species replacements characteristic of natural succession in forest lightgaps. In contrast, abandoned pastures are dominated by Vismia trees that inhibit natural succession for a decade or more. Here we explore how advance regeneration and limited seed dispersal may contribute to the arrested succession in Vismia-dominated stands. Vegetation surveys showed that every Vismia stem in 3–8 year old Vismia stands originated as a re-sprout. In Cecropia stands, all tree species, including Vismia, originated mostly from seeds, after deforestation and abandonment. The 100% re-sprouts of Vismia in the abandoned pastures confirms that Vismia dominance results from re-sprouting following pasture fires. Seed rain in both Vismia and Cecropia dominated stands was limited almost exclusively to second growth species already reproducing in those stands, suggesting that the bats and birds foraging there were not bringing mature forest seeds into the second growth, but simply feeding and depositing local second growth species. As dispersal was similar in both stand types, dispersal differences cannot account for the ongoing dominance of Vismia relative to the ongoing successional transitions in Cecropia stands. Overall, advance regeneration in the form of Vismia re-sprouts is much more likely to be the driver of Vismia dominated succession than differential dispersal of mature forest seeds. In order to avoid extensive forest conversion into unproductive Vismia wastelands in the Amazon Basin, forestry permits for harvesting timber should include restrictions on subsequent anthropogenic degradation, such as conversion to pasture and prescribed burning.
Introduction
In anthropogenically disturbed landscapes in the Neotropics, forest succession may be retarded due to competitive dominance by initial colonizers [1234–5] [or to limited dispersal of mature forest seeds [6, 7]. In the Brazilian Amazon, arrested succession on abandoned pastures is associated with numeric dominance by the genus Vismia [8]. In contrast, abandoned clearcuts are characterized by Cecropia along with a rich mix of other arboreal genera that exhibit species transitions characteristic of forest succession [8].
Vismia's initial abundance is a result of pasture fires prior to abandonment. Vismia can differentiate shoots from root tissue after stem loss, whereas Cecropia and other genera rarely survive repeated burns [9]. However, these monogeneric Vismia stands are often maintained for several decades, although the mechanism of ongoing dominance is not known [10]. In fact, it is not known whether the same initial re-sprouted Vismia trees last for several decades or whether Vismia replaces itself through several generations. In addition, Vismia may inhibit establishment of other mature forest species. In French Guiana, soil conditions and a well-shaded habitat, which assist in the growth of primary forest species, are found under Cecropia canopies but not under Vismia canopies [11]. In the upper Rio Negro basin, the regeneration of primary forest species is restricted under Vismia crowns, perhaps due to a change in soil conditions caused by the leaf litter [12, 13].
Alternatively, seed rain by mature forest species may be limited in Vismia stands due to a lack of avian dispersers. Vismia fruits are consumed mainly by bats, which may only bring a chiropterochore subset of forest seeds into Vismia second growth [14]. In contrast, most Cecropia species attract a broad set of disperser agents, such as birds, bats and primates, which may carry a more diverse array of mature forest seeds into Cecropia stands. Although Vismia stands are impoverished relative to Cecropia stands, species richness of both stand types declines with distance from primary forest, so seed dispersal may be limiting [8].
Here, we explore how aspects of two mechanisms–advance regeneration via re-sprouting and limited seed dispersal–may contribute to Vismia dominance of secondary forest on abandoned pastures in central Amazonia. Specifically, we determined the proportion of stems originating as re-sprouts for Vismia versus other genera in Vismia-dominated and Cecropia-dominated stands 3–15 years after abandonment. If re-sprouting ability is important beyond the first few years after pastures are abandoned, then a high proportion of Vismia stems originating from re-sprouts should still be evident in the older stands surveyed here. Second, in order to determine seed dispersal differences by bats and birds, we monitored the quantity and diversity of seeds that were dispersed into Vismia-dominated and Cecropia-dominated stands. Seed rain into Vismia stands may be depauperate compared to dispersal into Cecropia stands if bird-dispersed species are absent from Vismia stands.
Methods
Study Stands
The study area, located 80 km north of Manaus in the state of Amazonas, Brazil (2º 29′ S, 59º 41′ W, 54 m – 132 m elevation), is roughly 6 km east of the Biological Dynamics of Forest Fragments Project (BDFFP), a research collaboration operated by the Smithsonian Tropical Research Institute (STRI) and Brazil's National Institute for Amazonian Research (INPA). The region is dominated by evergreen terra firme forest, and data collected at the BDFFP show a mean annual temperature of 26° C [15]. The climate is type Am in the Köppen [16] system: tropical humid with excessive rain in some months and an occasional month of less than 100 mm precipitation. The average annual rainfall recorded at the BDFFP is between 2,200 and 2,700 mm [17] with a dry season from June to October. The predominant soils are nutrient poor, clay-rich oxisols with yellow latosols and red-yellow podzols [18].
Study stands were established in young secondary forests on private lands along Zona Franca 7 (ZF7), a two-lane unpaved road northeast of the small town of Rio Preto da Eva, Amazonas state (Fig. 1). The lots on ZF7 are small, 250 m frontage by 1,000 m deep, and are surrounded by extensive old growth forest (Fig. 1). In the past, landowners deforested portions of their land to plant cassava and fruit trees and to harvest wood for charcoal production.
Eight second growth stands between 3 and 15 years after abandonment with different land-use histories were selected on the basis of subjective dominance by Vismia or Cecropia (Table 1). Distances between stands were at least 2 km to ensure independence of species composition. In each stand, all woody stems ≥ 3 cm diameter at breast height (dbh) were recorded in a 10 m by 100 m transect.
Re-sprouting
Inside the established 10 m by 100 m transects, we examined each woody stem ≥ 3 cm dbh to determine if it originated from seed or as a re-sprout. Stems were considered re-sprouts when they were visibly growing out of a remnant stump or when they shared remnants of other stems that had died. The latter condition indicated that there had been multiple re-sprouts, but that by the time of canopy closure usually only one stem had survived. The numbers and proportions of Vismia, Cecropia, and all other stems that originated as seedlings or as re-sprouts were tested for differences in stand type, Cecropia or Vismia (one factor ANOVAs with stands as replicates).
Table 1.
Second growth stand characteristics and their land-use.
Diurnal/Nocturnal Seed Traps
In order to monitor bird versus bat seed dispersal in Cecropia and Vismia stands, eleven 1-m2 seed traps were spaced at 10 m intervals along each established 100-m transect. All traps were at least 10 m from the edge of the stand. Seed traps (Fig. 2) were constructed of 1-m tall PVC pipe frames which supported 1-mm nylon mesh suspended concavely to prevent seeds from bouncing or being washed out of the trap [192021–22]. There were 55 traps in Vismia stands and 33 in Cecropia stands.
Fig. 2.
Seeds were collected twice a day in seed-traps assembled in Cecropia-dominated second growth (shown) as well as in Vismia-dominated second growth
Seeds removed from traps at dawn (6:00 am) were assumed to have been dispersed by bats and those at dusk (6:00 pm) to have been dispersed by birds. To ensure that the seeds counted were those that had been handled by frugivores and not simply fruit fallen directly from the tree, only seeds with attached fecal matter were counted. Because it was impossible to be at all stands at 6:00 am or 6:00 pm on the same day, we sampled stands on a rotating basis for an average of seven days a month from June through August 2010.
Individual species seed abundances were log10 transformed and then tested for differences by stand type and by disperser type in 2-way ANOVAs with stands as replicates. In addition, we calculated species richness, Simpson's diversity index, and Shannon-Wiener diversity and evenness indices for seed collected from each stand. For bird-dispersed seeds, bat-dispersed seeds, as well as all seeds combined, we performed ANOVAs on the diversity metrics to determine differences between stand types. Finally, estimate S was used to determine separate individual-based species accumulation curves with 95% confidence intervals for all seeds collected in both Vismia stands and Cecropia stands [23].
Monthly Seed Traps
A second round of trapping was conducted monthly to compare seed rain by all dispersal modes in mature forest and in second growth stands. Seed traps, emptied monthly, not twice daily, were placed in the five second growth Vismia stands, the three second growth Cecropia stands, and also in six mature forest stands. Five mature forest stands were located within 100 m of adjacent Vismia second growth stands and the other mature forest stand was within 100 m of an adjacent Cecropia stand (Fig. 1). Eleven seed traps were established along 100-m transects within each of the six mature forest sites for a total of 66 traps in mature forest. In addition, the 11 seed traps per transect, used previously in the daily seed trapping, were opened monthly, in the five second growth Vismia stands and the three second growth Cecropia stands. Thus, overall there were 14 stands sampled monthly from January through May of 2010—six in mature forest, five in Vismia second growth and three in Cecropia second growth.
Monthly seed collections contained a mixture of seeds, fruits and litter. Due to limited time during the study, only half of the collections were sorted for analysis. In order to include data from all five months, January through May, seeds were sorted and measured from half the stands in January (14 days), February (28 days) and April (30 days), and the other stands in March (31 days) and May (31 days). All seeds, including those from fruits and feces, were dried, counted and identified to lowest practical taxon, usually to genus [24]. Uncommon seeds that could not be identified were grouped by morphotaxa. Also, the length of each seed was measured, and seeds were grouped into classes used by Cornejo and Janovec [24], whose book exhibits photos of many of the species' seeds. Size classes were: “Tiny”: <0.5 cm at greatest length, “Small”: 0.5–0.99 cm, “Medium”: 1.0–1.99 cm, and “Large”: >2.0 cm.
All seeds from all months in a stand were combined to calculate the number of seeds stand−1 month−1. This variable was log10 transformed and tested for differences by stand type–Cecropia, Vismia and mature forest (ANOVA), for all seeds and for seed size classes separately. Where the ANOVAs indicated significant differences, treatments were compared pairwise by least squares means with Tukey's adjustment for multiple comparisons.
General Data Analysis
All ANOVAs were performed in PROC GLM of SAS 9.0 [25]. As seeds collected per trap were few, seeds were combined from daily or monthly collections and from multiple traps at each site, such that replicates were all the seeds collected at each stand; therefore, there was no provision for repeated measures in our analyses. The ANOVAS reflect differences across the eight stands for daily seed collections, and across the 14 stands for monthly seed collections. However, the values presented in tables and in the text are labeled to show seed numbers per sampling unit, such as seeds/trap/day or seeds/stand/month.
Results
Study Stands
Stands were originally characterized as Cecropia or Vismia dominated, based on our subjective assessment of vegetation (Fig. 3). Sørensen's modified index, based on species basal area, supported the classification of stands as Vismia-dominated and Cecropia-dominated. The pairwise similarities were not different between Cecropia-Cecropia (0.46 ± 0.06) and Vismia-Vismia (0.42 ± 0.08) comparisons (Mann-Whitney U=6.8, P=0.81), whereas Cecropia-Vismia (0.20 ± 0.11) similarities were significantly less than Cecropia-Cecropia (Mann-Whitney U=8.0, P=0.002) and Vismia-Vismia (Mann-Whitney U=8.5, P<0.0001) similarities.
Table 2.
Density and basal area/transect, based on all stems ≥ 3 cm dbh, for five Vismia and three Cecropia second growth stands. “Non-VC” refers to non-Vismia and non-Cecropia stems. Abundances are per 0.10 ha transect. Relative abundance is calculated per transect.
Despite some difference in age of the stands, the mean number of stems was nearly identical in Vismia-dominated (246 ± 20; mean ± SD) and Cecropia-dominated (248 ± 4) stands (F1, 6=0.02, P=0.89; Table 2). However, basal area (m) was greater in the Cecropia (30 ± 1.6) stands than in Vismia stands (19 ± 2.7; F1, 6=36.71, P=0.009; Table 2). Both Cecropia and Vismia were much more common in their respective dominated stands than in the other genera's stands, for both absolute and relative abundances.
Species richness in Cecropia stands was double that in Vismia stands (F1, 6=18.94, P=0.005; Table 2). The number and proportion of stems, not Vismia, nor Cecropia (“non-VC”), found in Cecropia stands were double the number and proportion in Vismia stands (total number: F1,6=16.07, P=0.0070; proportion: F1,6=22.41, P=0.0032; Table 2). In general, Vismia stands were heavily dominated by Vismia, whereas Cecropia stands exhibited much greater evenness among genera.
Re-sprouting
Every Vismia stem in Vismia stands was a re-sprout (N = 635). The proportion of Vismia stems that were re-sprouts in Vismia stands (1.00 ± 0.00, mean ± SD) was higher than in Cecropia stands (0.18 ± 0.12; F1,6=286.16, P<0.0001), while the proportion of Cecropia stems that were re-sprouts in Cecropia (0.01 ± 0.01) and in Vismia (0.11 ± 0.21) were not different (F1,6=0.67, P=0.44). For non-VC stems, the proportion of re-sprouts in Cecropia (0.22 ± 0.30) and in Vismia (0.24 ± 0.14) were not different (F1,6=0.03, P=0.88).
The numbers of Vismia, Cecropia and non-VC stems that were re-sprouts in each stand type followed a pattern similar to the re-sprout proportions. The mean number of Vismia stems per stand that were re-sprouts in Vismia (127 ± 32) and in Cecropia (2 ± 1) were different (F1,6=42.5, P=0.0006). The mean number of Cecropia stems per stand that were re-sprouts in Cecropia (0.33 ± 0.58) and in Vismia (2.2 ± 3.5) did not differ significantly (F1, 6=0.79, P=0.41), and likewise, non-VC stems per stand that were re-sprouts in Cecropia (36.3 ± 47.9) and in Vismia (22.4 ± 10.8) did not differ significantly (F1,6= 0.43, P=0.54).
Fig. 3.
Comparison of dominance (≥3 cm d.b.h.) of five most common second-growth genera based on mean density (A) and basal area (B) in Cecropia (N=3) and Vismia (N=5) dominated second growth stands (Mean ± SD). Vismia includes: V. cayennensis, V. brasiliensis, V. guianensis, V. japurensis and V. sandwithii; Cecropia includes: C. concolor and C. sciadophylla; Bellucia includes: B. grossularioides and B. imperialis; Laetia includes: L. procera; Miconia includes: M. argyrophylla, M. burchellii, M. dispar, M. phanerostila, M. gratissima, M. lepidota, M. pyrifolia, M. regellii, M. tretraspermoides and M. tomentosa.
Diurnal/Nocturnal Seed Traps
Seed traps were open in both Cecropia and Vismia stands for 169 nights and 154 days. Fecal samples collected in traps contained a total of 6,443 seeds. In Vismia stands, seed traps, open for a total of 102 days and 112 nights, collected 4,665 seeds, while traps in Cecropia stands, open for 52 days and 57 nights, collected 1,778 seeds. There was no significant difference (F1,6=0.08, P=0.79) in the number of fecal samples collected per trap-period (12hr) in Vismia (2.91 ± 1.21, mean ± SD, seeds/transect/trap-period) and in Cecropia stands (3.15 ± 1.15, seeds/transect/trap-period), and no difference (F1,6=0.28, P=0.62) in seeds collected per trap period in Vismia (42.8 ± 39.8) and in Cecropia (29.4 ± 20.8) stands.
Seeds dispersed by birds were mainly in the genus Miconia (88.9%) regardless of second growth stand type (Fig. 4). Bat-dispersed seeds were predominantly in the genera Vismia (72.0%) and Miconia (15.6%) regardless of stand type (Fig. 4). Bats and birds deposited significantly different (F1,12=8.53, P=0.013) numbers of Vismia seeds (0.019 per night and 0.00062 seeds per day). Bats and birds deposited significantly different (F1,12=4.89, P=0.047) numbers of Miconia seeds (0.0027 per night and 0.0073 seeds per day). There were no significant differences in the number of Cecropia seeds or any non-VC seed species deposited by birds and bats in the two stand types, although numbers were generally too small to expect statistically significant differences.
Vismia stands and Cecropia stands showed no significant difference (F1,6=1.93, P=0.21) in species richness of seeds dispersed by birds (3.2 ± 1.5 and 1.7 ± 1.5, respectively) nor those dispersed by bats (3.0 ± 1.6 and 2.0 ± 0.0, per stand, respectively; F1,6=1.12, P=0.33). There was a significant difference (F1,6=6.70, P=0.04) in the species richness of all seeds dispersed in Vismia and in Cecropia stands (4.0 ± 1.0 and 2.3 ± 0.6, respectively) although this was a result of more individual seeds in Vismia stands. Species accumulation curves based on the number of individual seeds showed no significant difference in species richness between Cecropia and Vismia stands, as their 95% confidence intervals overlapped extensively. The number of species accumulated was only five in Cecropia stands and seven in Vismia stands, and the combined total was eight, with four species in common. For Simpson's diversity and Shannon-Wiener's diversity and evenness, there were no significant differences by stand type.
Monthly seed traps
More seeds were trapped per month in Vismia stands (13,926 seeds month−1 stand−1) than in Cecropia stands (304 seeds month−1 stand−1) or in mature forest (17 seeds month−1 stand−1). By species, significantly more Vismia seeds month−1 stand−1 were collected in Vismia stands than Cecropia stands (t10=−6.71, Tukey adj. P=0.0001; Table 3), more Cecropia seeds month−1 stand−1 were collected in Cecropia stands (t10=−0.79, Tukey adj. P=0.01), and more unknown seeds month−1 stand−1 collected in mature forest stands (t10=−3.72, Tukey adj. P=0.01).
There were many, many tiny seeds and very few larger seeds, especially in the second growth stands (Table 3). In order to test for differences in sizes, the two classes “tiny and small” were combined into one class, “little”, and the two classes “medium and large” were combined into one class, “big”. There were significant differences in the number of little and big seeds collected: more big seeds month−1 stand−1 collected in mature forest than in Vismia stands (t10=6.77, Tukey adj. P<0.0001; table 3) and more in mature forest than in Cecropia stands (t10=−6.18, Tukey adj. P=0.0001). Conversely, there were more little seeds month−1 stand−1 collected in Vismia stands than in mature forest (t10=−7.39, Tukey adj. P<0.0001); more in Cecropia stands than in mature forest (t10=3.73, Tukey adj. P=0.004); and more in Vismia stands than Cecropia stands (t10=−2.67, Tukey adj. P=0.023). Data were log10-transformed prior to testing.
The diversity of seeds (Shannon-Wiener Index, H), species richness (S) and evenness (E) of monthly seed collections for all stands of a type combined were highest in mature forest (H: 1.4; S: 30; E: 0.75) and lowest in Vismia stands (H: 0.08; S: 9; E: 0.29) with Cecropia intermediate between the two (H: 0.59; S: 7; E: 0.59).
Table 3.
Abundance of Vismia, Cecropia and unknown seeds collected in Cecropia stands, Vismia stands, and mature forest, month−1 stand−1 (Mean ± SD), in the four original size classes (Tiny: <0.5 cm, Small: 0.5–0.99 cm, Medium: 1.0–1.99 cm, Large: >2.0 cm), and in the combined classes (Little: 0.01–0.99 cm and Big: ≥1.0 cm) for statistical tests.
Discussion
Forest regeneration following anthropogenic disturbances such as clearcutting and conversion to agriculture or pasture often deviates from the pathway of secondary succession that follows natural perturbation of mature forest. Anthropogenic land use is believed to alter soil conditions [26, 27] and to affect initial florisitics through changes in the advance regeneration, the soil seed bank, and seed rain [28293031–32]. In such landscapes, succession may be retarded due to soil compaction, nutrient depletion, dispersal limitation of mature forest propagules [6, 7, 26, 29, 32, 33] or competitive dominance by the species initially present [1234–5, 8, 282930–31]. Lianas, commonly found during gap-phase regeneration, may take advantage of high light levels and relatively low canopy heights to impede secondary succession [3, 4]. Similarly, grasses, sometimes in association with fire, have been shown to delay succession [1, 2, 5]. Several of these factors may operate jointly–for example, a depleted seed bank, limited seed rain, and abundant seed predation–to restrict recruitment into secondary succession [32].
In a fragmented landscape that includes anthropogenic and natural elements, how succession will proceed often depends on prior land use history and proximity to mature forest. Throughout much of the Amazon Basin, abandoned pastures are often dominated by species of Vismia because it is the only tree genus capable of regenerating shoots from below ground tissues [8, 9, 34]. Repeated burning of pastures kills other advance regeneration and seedlings stimulated to germinate from the seed bank. Where recruitment limits succession, distance to mature forest becomes critical [8, 32].
Here, we explored questions related to two factors, advance regeneration and seed dispersal, in young Vismia-dominated stands, relative to natural succession in Cecropia-dominated stands [8]. First, how important was Vismia re-sprouting and was the initial abundance resulting from the re-sprouting still evident a decade after pasture abandonment? Second, was seed rain into Vismia stands different from seed rain into Cecropia stands and were there differences in bat and bird dispersed seeds associated with the two stand types?
Composition of the stands studied here reflected the land-use histories in a manner similar to that described by Mesquita et al. [8]; areas that were used for short-term agriculture or that were abandoned shortly after being cleared had a diverse species composition and were dominated by individuals in the genus Cecropia, whereas areas used for crops or for charcoal production, and thus repeatedly burned prior to abandonment, were dominated by Vismia (Fig. 5).
Re-sprouting
In regard to the importance of re-sprouting, our results were unanimous: 100% of the Vismia stems in the 3–8 year old abandoned pastures were re-sprouts. Thus, pasture maintenance, which requires periodic burning, produces an initial composition of re-sprouts of Vismia that remain dominant for at least eight years after abandonment. Given that Vismia is known to maintain its dominance in abandoned pastures for at least two decades [8, 10], and here we have shown that the stems in 8-year old stands were all derived as re-sprouts, then it is reasonable to assume that the same re-sprouts will continue their dominance for one or two decades. Vismia trunks in 20-yr old abandoned pastures no longer have remnant stumps visible, but multiple-stemmed individuals attest to a re-sprouting origin [34].
In contrast, Vismia stems in Cecropia stands were overwhelmingly (82%) from seed, not from re-sprouts. Vismia is a component of natural regeneration in forest light gaps and clearcuts. In gaps the stems originate from seed, not re-sprouts [34]. Therefore, it is no surprise to find individuals derived from seed in unburned secondary forests. However, the number of Vismia individuals in Cecropia stands was only one-tenth the number in Vismia stands.
Vismia in the upper Rio Negro colonizes abandoned slash-and-burn parcels, but apparently as much from seed as from sprouts [35]. In contrast to abandoned pastures, dominance on such agricultural land wanes more quickly, within the first five years [36]. However, in the eastern Amazon, dominance by Vismia apparently varies with the intensity of land use, but lasts at least a decade following moderate to heavy land use [28].
In our study, Cecropia stems were nearly all from seed, 99% in Cecropia stands and 89% in Vismia stands, as expected because the two species here, C. sciadophylla and C. concolor, do not re-sprout when the main stem is killed by fire. A few Cecropia stems sometimes survive pasture fires, originating as re-sprouts where the basal stems of burned trees were protected from the flame front by rocks or fallen logs. Other Cecropia species in the Central Amazon are also killed by fires, although in Brazil's coastal rain forests C. glazioui re-sprouts from lateral roots after fire much like Vismia [9, 33].
Stems of species other than Vismia or Cecropia in our study were about 75% from seed in both stand types. Similar re-sprouting rates in the two stand types imply that the stems originated from seeds dispersed after land abandonment and are not related to fires. Re-sprouting at a 25% rate commonly results from partial stem damage by herbivores and falling litter. Despite equal rates of re-sprouts among the stems, these other species' stems were twice as abundant in Cecropia stands as in Vismia stands.
Seed Dispersal
The majority of seeds collected in seed traps were pioneer species commonly found as adults in the second growth in which the seeds were collected, indicating that dispersers were consuming and depositing seeds in second growth rather than moving between mature forest and second growth stands. Quantitative differences in seeds collected in the two stand types reflected the composition differences of adults in the stands – most notably, more Vismia in Vismia stands. Outside of five unknown seeds, all the seed rain was from the same pioneer species present in the stands where the traps were located. Monthly seed rain did yield larger, unidentified (cf. mature forest) seeds in traps in the mature forest, but not in the secondary stands, so even though seeds were being produced in mature forest, they were not transported into the second growth stands. The same conclusion was reached in an earlier study comparing seed rain in a 6-year old Vismia stand and in nearby primary forest; over 14 months 96 seed morphotaxa were identified from traps in primary forest but only 9 morphotaxa from traps in the Vismia stand [37]. Thus, even though seed diversity is high in primary forest, seed dispersal into nearby second growth is low.
There were some differences in the day and night seed collections, reflecting dispersal by birds versus bats. In both stand types, birds dispersed mostly Miconia seeds and bats dispersed mostly Vismia seeds; however, seeds from both genera were found in fecal samples from both disperser types. Although birds and bats are known to specialize in different genera of seeds [38, 39], bats are more specialized in fruit selection, rarely sampling ornithochore fruits, whereas birds sometimes consume chiropterochore fruits [404142–43]. However, neither group in our study dispersed a diverse array of seed species. This result suggests that fruit consumption in early secondary vegetation involves species whose activities are restricted to secondary vegetation, for both birds [44] and bats [45]. Capture rates of bats have been shown to be roughly equivalent in young Vismia and young Cecropia stands with the three most abundant species, Carollia perspicillata, Rhinophylla pumilio, and C. brevicauda, accounting for 90% of the captures [45]. The three species are small and unlikely to carry primary forest seeds.
We had expected that seeds dispersed in Cecropia-dominated stands would be more diverse than those in Vismia-dominated stands, on the assumption that Cecropia fruits would attract a more diverse set of dispersal agents [38394041–42], notably birds. This was not the case, at least for the months studied here. There were only five species of seeds dispersed in Cecropia and seven in Vismia–clearly unrepresentative of regeneration in nearby mature forest that contains over 1000 tree species [46]. Seed rain in both stand types was relatively species poor, and species accumulation curves confirmed no significant difference in seed species richness between stand types. Birds that foraged and deposited seeds in both stands were small and unlikely to disperse larger primary forest seeds. Birds large enough to consume large-seeded fruits are generally uncommon in human-modified landscapes [47]. In our stands, toucans and guans were never seen.
Where, then, are the seeds of the 1000+ tree species that occupy nearby, old growth forests? One might argue that our seed trap sampling, performed only during June to August, missed the fruiting season of many tree species. The incidence of fruiting is highly variable among primary forest species in the Central Amazon north of Manaus [4849505152–53]. Among pioneers, the highest number of species fruit from January to March, although many species fruit continuously throughout the year [53]. Neither our monthly seed collections from January to May nor our daily collections in June, July and August demonstrated seed rain from nearby mature forest into adjacent second growth stands. Therefore, the most reasonable explanation is that dispersal agents are not moving seeds from mature forest into the second growth stands. Apparently, the birds and bats visiting second growth are mostly restricted to second growth, and more specialized dispersers such as primates, terrestrial mammals, as well as larger birds and bats that inhabit mature forest, are rare visitors to early secondary vegetation. Second-growth birds and bats may well be a distinct group from the primary forest species [44, 45].
Conclusions
In the Central Amazon, the slow turnover rate in Vismia second growth stands, originating in abandoned pastures, is an exemplary case of arrested succession [89–10]. Mesquita et al. (2001) showed that as distance from primary forest increases, species richness declines, suggesting that seed dispersal limits plant recruitment to some degree in young second growth. Our results support limited dispersal of seeds of mature forest species into second growth. Seed rain in both Vismia and Cecropia stands overwhelmingly reflected the dominant secondary species fruiting within each stand type. But neither birds nor bats brought new species into the second growth stands. Therefore, differences in rates of succession in Vismia and Cecropia stands are probably not a result of differences in seed dispersal by the bird and bat species visiting these young secondary forests. Consequently, some other process is operating to determine the alternative successional pathways.
Vismia stems in Vismia stands were 100% re-sprouts, demonstrating that they originated from stems burned in pasture fires. Their continued dominance for several decades may result simply from longevity and perhaps as well from inhibition of other species seedlings, although our seed rain data indicate that few mature forest seeds are dispersed into second growth stands. Without input of mature forest seeds, numeric dominance by the initial flora continues to provide dominance by default.
Overall, it appears that the alternative successional trajectories that characterize abandoned clearcuts versus abandoned pastures [8, 10] are a result of initial species composition which results in species replacements in the Cecropia stands and few replacements in Vismia stands. For the first decade or two of secondary succession, Vismia appears to maintain its dominance. This pattern of diverted or arrested succession on abandoned pastures is widespread, but not universal. Vismia dominance characterizes the ancient and nutrient poor soils of the Central and Eastern Amazon Basin, defined geomorphologically by the Guiana Shield and the Brazilian Highlands. Likewise, the soils of the Atlantic coastal forests are extremely old. However, in the Western Amazon and much of Central America, where soils are richer and geomorphologically younger, patterns of succession are different from those in the Central Amazon, and arrested successions by Vismia are uncommon. Therefore, Vismia dominance of abandoned pastures though re-sprouting after fires may be limited to ancient soils that are nutrient poor [54].
Implications for Conservation
High intensity land-use has been linked to Vismia-dominated successional forests in Brazil [1, 9, 43, 55], French Guiana [11], and Colombia [36]. Secondary succession on these abandoned lands depends on land use history, dispersal from the nearest seed sources [8] and the ability of forest species to re-sprout. Succession in abandoned pastures is arrested because the Vismia re-sprouts maintain dominance for at least one or two decades. Seed dispersal of mature forest species into Vismia-dominated stands is close to nil, but this is no different from dispersal into Cecropia-dominated stands where succession is not arrested. Therefore, the arresting mechanism lies in the early years following abandonment when Vismia, surviving pasture burns, becomes dominant by default.
The elimination of other woody species results from repeated prescribed fires, usually with anthropogenic intent of reducing woody regeneration to the benefit of forage grasses. Repeated burning reduces seed and seedling banks and kills remnant re-sprouts of mature forest species. Meanwhile, re-sprouting by the pioneer genus Vismia is prolific, resulting in monogeneric second growth within which other species are extremely slow to recruit. By contrast, in Cecropia stands, a diverse seed bank, seedling bank and remnant forest re-sprouts produce an enriched flora that regenerates under maturing Cecropia trees.
In order to avoid depauperate wastelands dominated by Vismia, we recommend that clear cutting of primary or older secondary forests should be accompanied by a post deforestation program that effectively limits or taxes land that is burned and converted to pasture. Deforestations occur first and foremost for timber, but deforestation is often followed by less lucrative charcoaling of stumps and slash, and conversion to cattle pastures. Forest conversion to pasture is a form of “pasteurization” of the regeneration resources. As most of the forest value lies in the timber extracted, clearcuts should be abandoned without conversion to pasture.
Enrichment plantings in abandoned pastures have been recommended for some tropical secondary forests in the tropics [56], but studies suggest otherwise in Vismia stands. Experimental transplants have exhibited extremely slow growth short term, suggesting that young Vismia stands inhibit mature forest seedlings [57 and Jakovac pers. com.], although long term studies are needed. Enrichment with seeds of pioneers or mature forest species has had limited success [58] although growth improves when Vismia is removed [59]. Abandoned pastures in the Amazon Basin are overwhelmed by Vismia re-sprouts, so enrichments that function in tropical grasslands [55, 6061–62] may prove unsuccessful in Vismia stands. Likewise, attempts to attract dispersal agents with roosts [63] or tree islands [64, 65] hold little promise where Vismia has already assumed dominance. Planting saplings in artificial gaps may be more successful, but at greater effort and expense.
Clear cutting of Vismia stems is time and cost intensive. As traditional, ecological restoration schemes offer little hope of regenerating old growth Amazonian forests, effective control mechanisms for Vismia stems need to be developed. Given the extensive monogeneric stands now available, finding uses for Vismia stems could result in harvests that might facilitate reversion to natural succession. Otherwise, natural regeneration in abandoned pastures will have to wait for senescence of a generation of Vismia trees.
Acknowledgements
Funding for this research was provided by the Conservation, Food & Health Foundation, the National Science Foundation (DEB-0639114), and Louisiana State University Biograds. Special thanks to A. M. dos Reis and C. Jackovac for assistance in the field and to the staff at the Biological Dynamics of Forest Fragments Project for logistical and technical support. Finally, thanks to P. C. Stouffer and R. Stevens for their thoughtful comments on various stages of this manuscript. Two anonymous reviewers provided insightful comments. This is publication # 583 in the Technical Series of the Biological Dynamics of Forest Fragments Project BDFFP-INPA-SI.
