Colombia: Dry Montane Dwarf Tropical Forest Restoration


This project evaluated a restoration experiment of late-successional dwarf forest in the Checua valley near Bogota, and was based on the idea of vegetation succession accelerated by canopy closure of planted woody species. In order to execute this restoration project, the project first evaluated habitat requirements of desired species, growing site characteristics of the desired vegetation type, and propagule dispersal abilities. The principal research questions that were addressed in the course of the project were: 1) Which types of vegetation was present in the study area, and what pathways of (secondary) succession were likely to occur?; 2) Was seed dispersal a limiting factor for development of secondary scrub and dwarf forest? 3) How do planted pioneer- and late-successional species perform in open secondary vegetation types? 4) Can natural succession in pastures and scrubs be accelerated by plantation of native woody species? 5) How do planted pioneer- and late-successional species perform in exotic Acacia decurrens forest? 6) Do plantations of native species leave prospects for conversion of exotic forests into native vegetation?

Quick Facts

Project Location:
Checua, Sutatausa, Cundinamarca, Colombia, 5.200050000000001, -73.86666000000002

Geographic Region:
Latin America

Country or Territory:

Tropical Forest

Tropical Forest - Coniferous, Tropical Forest - Seasonal Broadleaf, Other/Mixed

Project Lead:
Universiteit van Amsterdam

Organization Type:
University / Academic Institution


Project Stage:

Start Date:

End Date:

Primary Causes of Degradation

Agriculture & Livestock, Fragmentation

Degradation Description

The high plain of Bogotá is one of the most populated and intensely cultivated regions in the Andes of Colombia. The increase of the human population, and hence of resource demands, have experienced a strong acceleration during the last decades. Bogotá is heavily dependent on the surrounding mountain ranges for its drinking water supply, especially on the northern part, where the Bogotá River originates. This is one of the important water sources for the high plain and the city of Bogotá. Due to poor land use and vegetation cover there is little water retention and almost all water is lost to superficial runoff. This has had a strongly degrading effect on the unprotected soils, and extensive badlands are a common sight. The effects on the high plains water supply are severe as well, as the water flow is not evenly spread over time, but peaks tremendously after a heavy shower after which it drops to unsubstantial levels. The sediment load of the Checua River, frequently reaches 60% of the volume. Plantations with exotics – mostly Acacia decurrens and Pinus radiata – had been established in the 1980s for soil conservation purposes; however, the understory vegetation has hardly developed, and ecosystem function recovery is poor when compared to the native dwarf forest. The picture of poor understory development was confirmed for other sites in the Andes of Colombia where different Pinus species were planted.

Reference Ecosystem Description

The study area is a mosaic landscape of grassland, scrubs, planted exotic forest, and fragments of dwarf forest. Disturbances have been common and severe, and exotic grasses have large cover values. Most of the area is covered by pastures and planted Acacia decurrens forests, the steeper and more rocky slopes are covered with Baccharis macrantha – Dodonaea viscosa scrubs in different states of recovery. The dry andean dwarf forest was described as Xylosmo – Condalietum, or “Condalia forest.” The Quaternary history of the dry vegetation in the study area is not known, and hence it is very difficult to define an “original” ecosystem for the area. However, the Condalia dwarf forest zone can be regarded as an enclave in both a climatical and a botanical point of view. One of the characteristic treelets, Condalia thomasiana (Rhamnaceae), is a recently discovered endemic of this particular zone, and relatively many species, both herbaceous and woody, are not found elsewhere on the high plain of Bogotá and the surrounding mountains, and are typical of areas at a lower altitude above sea level. Hence, I believe that many species typical of Condalia dwarf forest have a long history in the study area, and because of that can be regarded “native” or “original” to the area. Combining this with the assumedly high degree of ecological functioning, I think that the Condalia forest deserves the status of “original ecosystem”, and hence serves as the “desired end point” for restoration efforts.

Project Goals

The project was critical because the semi-arid valley of the Checua River of the Bogotá River is so important for the water supply of the high plain of Bogotá and because it has few remnants of natural little disturbed vegetation. Many areas have been abandoned due to a declining agricultural productivity caused by drought, the loss of soil fertility and erosion. Moreover, parts of the area have been grown over with Acacia decurrens, which shows a very poor recovery in terms of biodiversity and ecological function. The restoration of the original endemic and nearly extinct Condalia dwarf forest is a goal of the natural resource authority Corporación Autónoma Regional de Cundinamarca (CAR).


The project does not have a monitoring plan.


The Corporación Autónoma Regional de Cundinamarca (CAR) manages the (renewable) natural resources in the region and is interested in recovering natural forests in the semi-arid parts of the high plain of Bogotá. The succession-based restoration experiment was executed in cooperation with the CAR, in which mixtures of nine woody native species were inserted in different vegetation types. The choice of these species – important in terms of local abundance – and the estimation of their successional affinity, was based on preliminary observations of the local flora.

Description of Project Activities:
The project has several parts, the first two were to infer the successional patterns for the nine woody species ultimately used in the project first by evaluating data that concerned vegetation structure, vegetation diversity, soil parameters, and disturbance history. A total of 101 relevees were made between October 1999 and April 2001, following Braun-Blanquet. The relevees aimed at including a maximal variation in vegetation diversity and structure, and at attaining a fair spatial spread. Severely eroded sites, rocky substrates and planted vegetation were avoided. Plot size was 10 x 10 m (85 plots), or 10 x 15 m (16 plots). In each relevee, the cover percentage of all vascular plant species was estimated, and the vegetation structure was described by estimating the average height and cover of low herbs (< 10 cm), high herbs (≥ 10 cm), low shrubs, (< 1.5 m), high shrubs (≥ 1.5 m), and treelets. For each of the restoration species, cover was estimated separately of saplings, low shrubs, high shrubs, and treelets. A bulk sample was taken of the upper five centimeters of mineral soil, consisting of five subsamples (from the four corners and the center) of the relevee. Each relevee was located on a series of black-and-white, panchromatic aerial photographs of the region taken in 1941, 1958, 1974, and 1991. The time period covered is about 60 years, including the data gathered during the field work in 1999-2001. The scales of the photographs varied from 1:10,000 to 1:20,000. On the photographs, woody canopy cover and height were assessed, presence of erosion and signs of disturbance (e.g., cutting or grazing) in an area of about 30 x 30 m, directly surrounding each of the vegetation relevees. This led to eight units, presented here in order of increasing structural development: 1) grassland; 2) grassland with scattered low shrubs (<5% woody cover); 3) shrub patches on eroded land; 4) low shrubland with (grazed) gaps; 5) low shrubland; 6) high shrubland with (grazed) gaps; 7) high shrubland; 8) dwarf forest. We attempted to express the trend in unit transitions of each relevee site by creating the nominal dummy variables development (increased woody cover and height), degradation (decreased woody cover and height), instability (both development and degradation took place), and stability (no changes). The second part of the assessment was to evaluate the potential role of soil seed banks in contributing to the regeneration of early and late secondary woody vegetation in dry Andean forests. Because of the heavy fragmentation and reduction of the dwarf forest area, and hence scarcity of seed sources, it was that expected woody plant taxa would be poorly represented in secondary vegetation. The following questions were addressed: To what degree are early and late secondary woody species represented in the seed bank of young secondary vegetation types? Is the richness of woody taxa in the seed bank at a given site related to the composition of the surrounding or locally standing vegetation? To answer these questions, the soil seed bank was sampled in abandoned pastures, secondary scrubs, dwarf forests and Acacia forests, and compared these with the local woody standing vegetation. In November 2001, in each of the five vegetation types four plots of 30 x 30 m were laid out widely distributed over the area. The sides of the plots were situated parallel to the contour lines. In each plot five random points were chosen under the restriction that all points were at least 4 m apart from each other. Around each point, three soil samples of 9 x 9 cm were taken at 0-5 cm and 5-10 cm depth, and pooled for each depth. The litter layer (if present) was included in the upper soil sample. The plot altitude ranged between 2635 and 2912 m above sea level. A vegetation plot of 70 x 70 m was laid out over each seed bank plot, in such a way that the center of vegetation and seed bank plots coincided and their sides were parallel. In the vegetation plot, the aboveground cover of woody plant species was recorded. The cover of scrub (low and high shrub vegetation types combined) and dwarf forest surrounding the seed bank plots was estimated on aerial photographs from 1991 (scale 1:10,000, black & white, panchromatic). For this, circles with 200 and 500 m diameters were projected on each seed bank plot. The vegetation cover was then estimated by applying a grid overlay of square cells, each corresponding to 25 x 25 m in the field. Germination trials were conducted in a ventilated greenhouse, located in a nearby village (Nemocón). The total of 408 trays (20 plots x 5 samples x 2 depths x 2 treatments + 8 control) were placed randomly in the greenhouse, and were watered daily. The seeds were allowed to germinate for a period of 12 weeks (December 2001 - March 2002). During this period the number of vascular plant seedlings in each tray was counted weekly. Most seedlings had emerged after 6 weeks. The third part of the project was the actual planting of the nine species of shrubs and trees that were selected for the project. All of these species are native to the Condalia dwarf forest zone and commonly found there in woody pioneer stands and/or in late-successional dwarf forest patches. Dodonaea viscosa was categorized as a mid-successional species, but was included in the pioneer mixture because it is mostly co-dominant with Baccharis macrantha in pioneer scrubs. Similarly, Hesperomeles goudotiana, Myrsine guianensis and Xylosma spiculifera were classified as mid-successional, but were included in the late-successional plantation mixture because of their frequent co-occurrence with the three other species of that mixture in dwarf forest patches. We chose to plant greenhouse-grown seedlings instead of sowing seeds, thus skipping the germination stage in which most plant species are vulnerable. Seeds of all species were collected from the vegetation present at the study site. The plant material for the initial planting was produced in two different nurseries owned by the Corporación Autónoma Regional de Cundinamarca (CAR). Plants received applications of fertilizers and fungicides on a regular basis during the production and maintenance period in the nurseries. At the time of planting in the field, the seedlings were in 707 cubic cm nursery bags. D. coerulea was grown in bags with a volume of 2389 cubic cm. Three vegetation types were selected for the project: two types of abandoned pasture, and a scrub type representing the least developed, low-stature woody stage. The plantation was established in May - June 2000. In each of the three vegetation types, five plots of 10 x 15 m were established and planted with seedlings of the pioneer species. Another five plots were planted with seedlings of the six late-successional species. Repeats of both the pioneer series and the late-successional series were carried out with two separate additional treatments. The stone treatment consisted of three or four stones - about 10 cm diameter - placed around each individual plant in the plot, about a week after planting. The stone treatment was not applied in shrublands, because it was thought those plots already had a microclimate affected by shrubs and tussock grasses. Additionally, the series of plots planted with late-successional species was repeated with fertilizer treatment. A quantity of 20 g of "triple-15" fertilizer (15% of both N, P, and K, with additional trace elements) was applied superficially in the soil, at some 10 cm distance in the upslope direction of each plant. This was done in July 2000, about a month after planting, when rain was still frequent. All the experimental plots were subsequently put in barbed wire enclosed clusters. Each cluster contained one plot of the different combinations of planted species mixture and additional treatment, and a control plot for the particular vegetation type. The minimum distance between plots within a cluster was 10 m. The clusters of both dry and sub-humid pasture each contained six plots, and those of shrubland four plots. A total of 80 experimental plots were laid out for the experimental setup. Every plantation plot had 189 individuals planted at 1 meter's distance, in a regular triangular pattern. As three different combinations of species were applied in the experimental setup, each had its own basic species density and spatial distribution pattern. The patterns were designed in such a way that the distribution of each species was more or less regular, without severe clustering of a certain species at a certain place in the plots. The first pioneer mix, consisting of Baccharis bogotensis and Dodonea viscosa only, was simply planted in an alternating way, resulting in either 95 B. bogotensis and 94 D. viscosa per plot, or the reverse. The second pioneer mix had 63 individuals of B. bogotensis, D. viscosa, and Dalea coerulea planted in a simple alternating way. This mixture was only applied in dry pastures, because D. coerulea is common around these areas, and does not occur in sub-humid pastures nor in B. macrantha - D. viscosa scrubs. The late-successional species mixture consisted of different quantities of each of the six species, due to differences in availability. The seedlings were planted in such a way the species were regularly spread over the plot. A hole of about 20 cm depth was made and the plant was inserted along with all the earth of the nursery bag it was grown in. In November 2000, dead seedlings were replaced by living ones of the same species. Individuals that showed high mortality of leaves or were in otherwise bad condition were also replaced with vigorous ones. Survival, length and cover of the planted seedlings were monitored in October-November 2000, October 2001, and August 2004. The stem length of every individual was measured from the ground up to the upper living bud or leaf base. Cover of each planted species, and the total planted canopy cover, was estimated visually as percentage of the plot area. At the same time, the composition and abundance of all non-planted vascular plant species (the matrix vegetation) was recorded in all of the experimenal plots, including the control plots. Vegetation structural data - bare ground, herb cover, shrub cover - were also recorded.

Ecological Outcomes Achieved

Eliminate existing threats to the ecosystem:
While the average survival percentages at the plot-level of Dalea coerulea were below 40%, the other two pioneer species performed better, with over 90% survival for D. viscosa seedlings in pasture plots. Myrsine guianensis performed the worst in terms of survival, the percentage surviving seedlings mostly remaining below 35%. The remaining five late-successional species had survival percentages around 50%. Survival on plot-level of seedlings of the pioneer species B. macrantha and D. viscosa was higher in at least one of the two pasture types than in shrublands, while no differences were detected for any of the late-successional species. Stone treatment had a significant survival-enhancing effect for D. viscosa, C. thomasiana, and M. guianensis. Fertilizer addition enhanced survival of C. bogotanus, but only in dry pastures and scrubs. Individual growth of planted seedlings was generally poor: the average length of the planted individuals after four years was around or well below 50 cm for all late-successional species. The pioneer species mostly reached a height between 50 and 100 cm, Dalea coerulea often above 120 cm. Average growth rates were around zero or negative for all late-successional species, except for Myrsine guianensis which attained a rate of 0 - 10 cm growth over the four years period. Planted seedlings of X. spiculifera, D. mutisii, and C. bogotanus appeared to have "shrunk", i.e., the upper living leaves or buds died. Some species, mainly Xylosma spiculifera and Croton bogotanus, showed reiteration from the base after mortality of the main above-ground stem. Four out of six late-successional species showed a higher growth in shrubland plots (without stones or fertilizer) than in those of at least one of the pasture types, and D. mutisii and X. spiculifera more than in those of both pasture types. On the contrary, the pioneer shrubs B. macrantha and D. viscosa showed the highest growth in dry pastures. The pioneers B. macrantha and D. coerulea showed a higher overall growth rate for stone-treated seedlings than non-treated ones, although this does not hold for B. macrantha in dry pasture plots. This was also the case for the late-successional species D. mutisii, M. guianensis, and X. spiculifera. All late-successional species except H. goudotiana showed enhanced growth as a result of fertilizer addition. Although we could not test for interaction effects between vegetation type and treatment, the benefit of fertilizer appeared greater in shrubland than in pastures for C. thomasiana and C. bogotanus. Total cover of the planted seedlings reached average values between 10 - 20% only in pasture plots planted with pioneer species. The three pioneer species showed an increase up to - on average - at least 5% in the pasture plots, B. macrantha and D. viscosa remaining sparse in shrubland plots. In the dry pasture pioneer-planted plots, total cover values of about 20% were reached in both untreated and stones-treated plots, while in the sub-humid ones total cover was around 15%. The six late-successional species C. thomasiana, C. bogotanus, D. mutisii, H. goudotiana, M. guianensis and X. spiculifera did not increase their cover in any of the plots, nor did the total cover of late-successional planted seedlings. In the dry pasture plots, a total of 75 species was found in 2000, and 83 in 2004. In sub-humid pastures, these numbers were 98 and 100 for 2000 and 2004, respectively, and in shrubland plots 137 and 143. There appeared to be no discernable patterns in species increase per plot over time. The number of annual species, on average, slightly decreased in most of plantation series, but again there were no patterns. Average turnover rates varied between 0.20 and 0.30 for dry pasture and shrubland plots, and between 0.30 and 0.40 for sub-humid plots. No differences in turnover rates were found between planted species mixtures or between additional treatments. Moreover, control plots did not have turnover rates different from those of planted plots. Since pasture-plots planted with pioneer species were the only ones yielding a total planted canopy cover between 10 - 20%, which is a substantial change from the initial situation, the analyses of vegetation development is restricted to only these plots and the control plots. Changes in the remaining plots did not appear to be an effect of the plantation. In pasture plots, it was found that there were few herb species normally found in shrublands. Woody species invading pasture plots were mostly pioneers. Dalea coerulea was found in pioneer-planted dry pasture plots, most probably offspring from the planted seedlings. The commonly planted exotic Acacia decurrens invaded five shrubland plots. In dry pasture plots, among the species mostly increasing their cover over time were the grasses Botriochloa saccharoides, Aristida laxa, Sporobolus indicus, and Pennisetum clandestinum. The shrub Baccharis macrantha also increased cover. In sub-humid pastures, the common grass Anthoxanthum odoratum decreased in cover, while Rhynchospora nervosa and Cyperus aggregatus became more abundant. In shrublands, the most common woody species became more abundant (B. macrantha, Dodonaea viscosa), together with the spreading herbs Cuphea serpyllifolia and Rhynchospora nervosa. The grass Aristida laxa declined its cover. Most of the pioneer-planted plots clearly showed an increase in herbaceous cover and a decrease in bare ground. In dry pastures, there was no apparent differences between control and planted plots were regard to herb cover, while the decrease in bare ground was higher in planted plots than in control plots. The high overall rates of mortality, mostly for late-successional species, could be explained by night frosts, which have been exceptionally severe during the dry periods of January to March 2001 and January to March 2004. In addition, drought has probably been another major cause of mortality. The positive effect of the stone treatment on survival of nearly all species suggests desiccation stress for naturally germinated seedlings in both dry and sub-humid grasslands. The positive effect of stones on stem growth of several species planted in sub-humid pastures was mostly not found in dry pastures, or less strongly so. It even appeared negative for B. macrantha in dry pastures. Fertilizer addition generally increased growth but not better survival, of the late successional species. The degree of increase in nutrient uptake by seedlings - causing enhanced performance - resulting from fertilizer addition, depends on local soil physical and chemical characteristics. In shrublands, low growth rates and cover could be explained by competition for water with the abundant bunchgrasses and shrubs. Additionally, the shallow soils of these sites could also explain low overall growth rates. The late-successional species considered here are favoured by a certain minimum level of soil humidity, which could be conserved if (woody) vegetation cover is sufficient. Probably, the cover of the non-planted shrub layer in the shrubland plots was not high enough to have such a facilitating effect, and the observed performance of late-successional species was even less than in pastures.

Factors limiting recovery of the ecosystem:
The only effect measured in planted plots that did not occur in control plots was an increase in herb cover, obviously inversely related to a decrease in bare ground. Hence, instead of a mere effect of grazer exclusion, this is an effect of planted canopy establishment. This might be regarded as the first onset to accelerated succession. Shrub crowns at least facilitated horizontal spread of already present herb species. However, the Baccharis-Dodonaea (-Dalea) canopy has only recently established, and hence microclimatic changes, if any, have not yet had the chance to measurably influence establishment of species. Growth of shrubs and trees was slow, especially of the late-successional ones, indicating establishment of canopy that will take place over a scale of decades, instead of years. Establishment of late-successional species not only depends on a suitable habitat eventually formed by a planted canopy, but also on propagule availability. Seed banks of most late-successional species are very sparse, especially in pastures. Turnover rates were higher in sub-humid pastures than in dry pastures, in spite of less initial bare ground surface and hence less available sites for germination. Species turnover is generally believed to be higher in habitats with high nutrient availablity and soil moisture, although extreme conditions have lower turnover rates. Shrublands were found to have a lower turnover rate as compared to grasslands, which is consistent with results from many other authors. This has been attributed to longer life spans of late-successional invading species, and more competition for resources. Control plots showed changes in species composition similar to planted ones, irrespective of the planted species mixture or the treatment. This was hardly surprising, since survival and growth of the planted seedlings during the four year period was not enough to establish a woody canopy in many plots. Moreover, even after closure of a woody canopy, processes of colonization still take time. The different direction of change in species composition, as revealed by PCA, should probably be interpreted as little change at all; the picture is focused on small changes. This is supported by the low fraction of explained variance by the principal components: mostly below 10%. The changes that did occur, i.e., increased cover of graminoids, appears to most likely be attributed to exclusion of grazing cattle by the fences.

Socio-Economic & Community Outcomes Achieved

Economic vitality and local livelihoods:
The methods used in restoration or rehabilitation practices such as this project should be affordable for developing countries. Although the cost of plantations is relatively high, it is due partly because of the nursery procedures. However, since the cost of labor is relatively low in most tropical countries, plantation, even with survival-enhancing application of stones, could be an affordable way for many institutions to rehabilitate endangered sites, at the same time as providing sustainable income for local communities.

Long-Term Management

The pioneer species used in the project are all useful for plantation experiments in abandoned pastures, given that their survival rates are high and growth is sufficient as to form a significant canopy. Further trials and restoration attempts might include combination of pioneers and late-successional species. During canopy formation of the pioneers, the late-successional species could at least survive, and growth conditions for the latter might improve after some lateral crown growth of B. macrantha and D. viscosa. Especially if the seedlings would be treated with (a combination of) stones and fertilizer, chances of success might be further enhanced. Although growth is very slow, the rates are in tens of centimeters stem length per year, the effects of this type of restoration treatment is clear. Since both growth and propagule availability of the desired species are low, successional changes will probably become evident after a relatively long period. While there was no data to speculate on the length of the time span needed for total regeneration of Condalia dwarf forest, the project illustrated that using plantations could at least overcome a part of the seed dispersal limitation. If collected seeds would be sown out as an alternative for seedling plantation, the loss of individuals would be much higher, and the time span for establishment would be even longer. This would not be desirable, in view of the urgent need for restoration in many sites.

Sources and Amounts of Funding

The project was carried out at the Institute for Biodiversity and Ecosystem Dynamics (IBED), research group Palaeo-ecology and Landscape-ecology, Universiteit van Amsterdam. Financial support came from the Netherlands Organisation for the Advancement of Tropical Research (NWO-WOTRO, grant W 85-325). The project was executed in collaboration with the Corporación Autónoma Regional de Cundinamarca (CAR), Bogotá, Colombia.

Other Resources

Groenendijk, Jeroen Pascal. 2005. Towards Recovery of Native Dry Forest in the Columbian Andes. University of Amsterdam, Dissertation.

Primary Contact

Organizational Contact