Dominican Republic: Restoring Tropical Forest at Sites Dominated by Anthropogenic Fern Thickets


In many parts of the tropics, deforestation results in areas dominated by colonizing ferns of the genus Dicranopteris. This study investigated methods for restoring native wet tropical forest on a degraded site in the Ebano Verde Scientific Reserve, located in the central highlands of the Dominican Republic. Fern cover was manually removed from the site, and then a trial planting of 18 nursery-reared native tree and shrub species was conducted. The seedlings were planted in high densities in order to offset the effects of seedling mortality and inhibit the reestablishment of the fern. Monitoring conducted over a period of three years found high rates of survival and positive growth, and showed the plantation of woody species to be a viable option for complementing forest succession and accelerating restoration on these disturbed sites.

Quick Facts

Project Location:
Ebano Verde Scientific Reserve, Dominican Republic, 19.0585018, -70.5443421

Geographic Region:
Latin America

Country or Territory:
Dominican Republic

Tropical Forest

Tropical Forest - Moist Broadleaf

Area being restored:
approx. 1 acre

Organization Type:
University / Academic Institution


Project Stage:

Start Date:

End Date:

Primary Causes of Degradation

Agriculture & Livestock, Deforestation, Invasive Species (native or non-native pests, pathogens or plants)

Degradation Description

At present, the Ebano Verde Scientific Reserve contains species-poor secondary forest concentrated along streams (Slocum et al. 2000), and small, isolated patches of mature forest that contain the majority of the reserve’s woody species (274 species; García et al. 1994). Besides these forests, the reserve contains expansive thickets of Dicranopteris pectinata (Willd.) Underw. (Gleicheniaceae), which colonized in the 1970s after logging, agriculture, burning, and subsequent soil erosion (García et al. 1994). Although precise estimates of cover for the thickets over the entire reserve are not known, 90% of the southwestern part of the reserve is covered by the fern (Slocum et al. 2000). In these areas, the thickets consist of a layer of living fronds, a layer of dead fronds and stems, and a root mat. The few trees and shrubs in the fern thickets are small, occur at low densities, and consist of only approximately 25 species (García et al. 1994; Slocum et al. 2000). These trees presumably arrived and grew to a sufficient height before the fern canopy was fully established, as regeneration under the thickets is extremely limited (Slocum et al. 2004). Similar thickets of D. pectinata, and its congener, D. linearis (Burm. f.) Underw., are found on disturbed sites throughout the tropics and have similar inhibitive effects on native woody vegetation (Walker 1994; Cohen et al. 1995; Walker & Boneta 1995; Mejía & Jiménez 1998; Russell et al. 1998).

Project Goals

To restore anthropogenic fern thickets to a species composition representative of the region’s previously extant montane forest.


The project does not have a monitoring plan.

Description of Project Activities:
Selection of woody plant species was based on several criteria. First, species were sought that already grew commonly in the thickets and therefore showed potential to grow and survive after fern removal. Seven such species were found, six early-successional and one late-successional (Slocum et al. 2004). Secondly, in order to maximize the chances of finding some species that would have good rates of growth and survivorship, roughly equal numbers of early- and late-successional species of both trees and shrubs were collected. Life history assignments were based on the habitats the species colonized (Weaver 1986; García et al. 1994; Slocum et al. 2000). Lastly, only species that were common in the southwestern part of the reserve were selected, in order that their seeds and seedlings could be easily collected both for this study and for future efforts, and to ensure that their collection would not cause further degradation in the reserve. Seedlings were collected from November 1998 to January 1999 and grown until early June 1999. They were grown in nursery bags (4.5 x 10 x 12 cm) in a shade house (50% incident sunlight). They were regularly watered and fertilized with poultry litter (poultry manure mixed with bedding material) bearing the following nutrient composition: 18 ± 0.9 g/kg N, 14 ± 0.5 g/kg P, and 21 ± 0.3 g/kg K (mean ± 1 SD, n 1/4 3). Before out-planting, all plants were placed outside of the shade house for two weeks to adjust them to full sunlight. At the time of planting (mid-June 1999), seedlings selected for out-planting were robust (mean ± 1 SD, 20 ± 11 cm tall, n 1/4 2,280). For out-planting, three blocks were randomly located along an old logging road in the southwest part of the reserve. These blocks (each approximately 45 x 30 m) were located more than 100 m away from the road and approximately 250 m from each other. These blocks all had different aspects and slopes and were representative of fern thickets in this part of the reserve; but they may not be representative of thickets on steeper slopes or at higher elevations. In the blocks, thickets were removed using machetes between January and March 1999. Ferns were cut to the root layer, and the debris was piled up into rows 1 - 2 m apart, leaving cleared rows in between. As much as possible, these rows were placed along contour lines to minimize erosion. Erosion was also reduced by not removing the root mat; however, the root mat was cut to reduce the likelihood of resprouting. Along the cleared rows, seedlings were planted 1 m apart. This high density was used because high mortality was anticipated and because it was anticipated that a large number of seedlings would be needed to shade out recolonizing fern. In each block, approximately 50 seedlings of each species were planted, and half were fertilized with approximately 100 mL of poultry litter placed at the bottom of the planting hole. No additional fertilizer was applied after planting. The position of each seedling along the cleared rows was random, regardless of its species or fertilizer treatment. Plant height was measured at the time of planting, and plant height and mortality were determined 12, 24, and 36 months after planting. To understand how clearing the fern affected soils, cores of soil (12 cm deep x 5 cm diameter) were collected in the cleared blocks and compared to those collected in nearby thickets. For each block, six randomly selected points along its edges were selected. At each point, a core was taken 3 m into the cleared block and 3 m into the thicket. A total of 36 cores were collected (3 blocks x 2 fern treatments x 6 replicates). The effect of fern removal on foliar nutrient concentrations was determined by collecting leaves from adult trees of Myrsine coriacea and Brunellia comocladifolia. It was assumed that these trees were approximately 25 years old because they appeared to have established when agricultural activities ceased, and before the fern thickets became fully established (Slocum et al. 2004). Three years after clearing (June 2002), two recently matured leaves were collected from three trees of each species in each block. In the adjacent thickets, leaves were also collected from three trees of each species, for a total of 36 samples (3 blocks x 2 fern treatments x 2 tree species x 3 replicates). For out-planting, three blocks were randomly located along an old logging road in the southwest part of the reserve. These blocks (each approximately 45 x 30 m) were located more than 100 m away from the road and approximately 250 m from each other. These blocks all had different aspects and slopes and were representative of fern thickets in this part of the reserve; but they may not be representative of thickets on steeper slopes or at higher elevations.

Ecological Outcomes Achieved

Eliminate existing threats to the ecosystem:
The soils in the cleared blocks did not significantly differ from those in the thickets overall, nor did foliar nutrient concentrations for either of the two tested tree species, Myrsine coriacea or Brunellia comocladifolia. Soils in the clearings and thickets both had a pH of 4.6--typical of a highly weathered tropical soil (Motavalli et al. 1995)--and a cation exchange capacity of 15 cmol/kg. In fact, levels of exchangeable cations were high compared to other tropical soils (Silver et al. 1994; Motavalli et al. 1995; Menzies & Gillman 1997; Davidson et al. 1998), particularly for exchangeable Ca. Soils at the study site were rich in carbon (Silver et al. 1994; Tian 1998; Tornquist et al. 1999) but had low to moderate total N (2.6 g/kg) (González & Fisher 1994; Silver et al. 1994; Motavalli et al. 1995; Tornquist et al. 1999), resulting in a high C:N ratio of 16 (Erickson et al. 2001). Available NH4-N and NO3-N also appeared in low to moderate levels (Davidson et al. 1998; Erickson et al. 2001), as did available phosphate. After three years, overall mortality of the woody plants was 21%. Mortality was not significantly affected by block, fertilizer, or any of their interactions. There was, however, a significant species by fertilizer interaction. Only one species, Piper aduncum, had significantly greater mortality (73%) when not fertilized. The strongest effect was that of species, as almost no seedlings of Inga fagifolia died, whereas more than 50% of the seedlings of Cyrilla racemiflora and B. comocladifolia died. Some species appeared to have higher mortality in the first year, including B. comocladifolia, Myrcia deflexa, Prestoea acuminata var. montana, Turpinia occidentalis, and Psychotria berteriana. Others species had constant mortality over all three years, including C. racemiflora, Ocotea leucoxylon, Clidemia umbellata, and P. aduncum. There was no clear effect of life history on woody seedling mortality. Plant height increase after three years was observed in an increment of 0.3 to 3.4 m, and was significantly affected by both species and the interaction of species with fertilizer. Although fertilizer led to an overall 10-cm height increase for all species combined, this was statistically significant for only one species, B. comocladifolia, which had the highest mean height increase of the tested species and which grew 73 cm/year faster when fertilized. Average plant height increase was 126 ± 103 cm, with 28% of the plants growing more than 2 m in height. Early-successional species grew about twice as rapidly as late-successional species (153 ± 103 vs. 81 ± 67 cm), whereas trees grew 25% faster than shrubs (141 ± 105 vs. 111 ± 90 cm). The six species with the most rapid height increase included five early-successional species (B. comocladifolia, P. berteriana, Alchornea latifolia, C. umbellata, and M. coriacea), and one late-successional species (I. fagifolia). On average, these species grew more than 1.5 m in three years, with 5% of their individuals growing more than 4 m. The six slowest-growing species included four late-successional species (C. racemiflora, P. acuminate var. montana, M. deflexa, and Mora abbottii) and two early-successional species (Tabebuia bullata and O. leucoxylon). On average, these species grew between 28 and 67 cm in height over three years, with only 0.4% of their individuals growing more than 2 m in three years. It was expected that M. coriacea and B. comocladifolia would probably have the highest growth rates and survivorship because they are the dominant tree species in the thickets (Slocum et al. 2004). However, once the fern was eliminated, these species did not perform better than species that were rare or that did not occur in the thickets at all. Indeed, the species with the best combination of growth and survivorship included one that was moderately common in the thickets (Alchornea latifolia) and one that did not occur in the thickets (Inga fagifolia). The latter, a late-successional nitrogen-fixer, achieved a mean height increment of 159 cm and had almost no mortality over three years. In the same time period, A. latifolia, an early-successional species, grew an average of 207 cm in height and had only 4% mortality. As mentioned above, this project of sown seedlings was conducted concurrently with a study of natural regeneration in the clearings (Slocum et al. 2004). Natural regeneration was substantial after clearing the thickets, with many individuals growing more than 2 m in height from seed in 3 years. Moreover, three years after clearing, regrowth of D. pectinata reached only 16% cover, and most of this regrowth could be easily controlled because it stemmed from thickets along clearing edges.

Factors limiting recovery of the ecosystem:
The fertilizer treatment (poultry litter) increased height increment of only one species, B. comocladifolia, and the survivorship of another, Piper aduncum. This result may be explained by the use of only a small amount of the poultry litter (a single application of approximately 100 mL), which was decided upon when preliminary experiments (M. G. Slocum, Louisiana State University; T. M. Aide, University of Puerto Rico; J. K. Zimmerman, University of Puerto Rico; and L. Navarro, Universidad de Vigo, Spain, unpublished data) showed that some species exhibited high mortality when fertilization with poultry litter was repeated. A limited response to fertilization may also be explained by the fact that the poultry litter had concentrations of N (1.8%) and P (1.4%) lower than the average found in poultry litter in the U.S.A. (4.9% N and 2.1% P) (Sharpley et al. 1998). Percent K of the litter (2.1%) was the same as the U.S. mean. Alternatively, Davidson et al. (1998) suggest that nutrients released by fertilizer might have been absorbed by the soil and might, thus, have become unavailable for uptake by plants. Two other studies conducted on acidic, highly weathered tropical soils found a similar lack of response of woody plants to fertilizer (Harcombe 1977; Davidson et al. 1998).

Socio-Economic & Community Outcomes Achieved

Key Lessons Learned

Although Dicranopteris pectinata has inhibited natural succession at the EVSR for more than 20 years, this study has demonstrated that by simply clearing large plots and planting a mixture of native woody species, one can successfully initiate forest restoration. The soils of the site were apparently sufficiently fertile to support the growth of sown woody plants and also to foster natural regeneration. In fact, the concurrent study of natural regeneration at the site showed that after clearing the fern thickets, natural forest succession alone could be a cost-effective way to restore forests in the EVSR. However, the process was found to be patchy and species-poor, as only 23 species became established, and only 12 exhibited good growth rates (Slocum et al. 2004). This lack of species diversity might be overcome with time, as the development of secondary forest begins attracting seed-dispersing animals (Holl et al. 2000), but the process is slow and could allow the reestablishment of ferns or other competitors. Therefore, the planting of woody species is recommended as a means of complimenting natural regeneration by increasing overall species diversity and filling areas where vegetative succession is limited.

By using a large number of native species that varied in life history characteristics, this study enabled practitioners to identify promising species for future reforestation efforts. Based on the initial assessment of species used in this study, researchers recommend that open areas be planted with I. fagifolia and A. latifolia in order to shade out competitors and perhaps improve soils. Given their high survivorship, these species could be planted using a wider separation (every 3 – 4 m) than that employed in this study (1-m spacing). Once they have created a closed canopy, the planting of late-successional canopy and understory species (Ashton et al. 2001), such as Cyrilla racemiflora, Prestoea acuminata var. montana, Myrcia deflexa, and Mora abbottii, could help increase site diversity. These treatments are relatively inexpensive and could easily be tested in other montane ecosystems where D. pectinata covers thousands of hectares.

Sources and Amounts of Funding

Funding for this project was provided by NASA-IRA (NAGW-4059). Luis Navarro obtained additional funding from the Ministry for Education and Science of the Spanish government and the Xunta de Galicia.

Other Resources

Slocum, Matthew G. et al., 2006. A strategy for restoration of montane forest in anthropogenic fern thickets in the Dominican Republic. Restoration Ecology 14(4): 526-536.

Profile: Hispaniolan Moist Forests

Primary Contact

Organizational Contact