Biological Conservation 109 (2003) 123–135

[Link]/locate/biocon

Integrity and isolation of Costa Rica’s national parks and biological

reserves: examining the dynamics of land-cover change
G. Arturo Sánchez-Azofeifaa,*, Gretchen C. Dailyb, Alexander S.P. Pfaffc,
Christopher Buschd
a
Department of Earth and Atmospheric Sciences, Earth Observation Systems Laboratory, University of Alberta, Edmonton, Alberta, Canada T6G 2E3
b
Department of Biological Sciences, Center for Conservation Biology, 371 Serra Mall, Stanford University, Stanford, CA 94305-5020 USA
c
Department of International and Public Affairs, Department of Economics, and Center for Environmental Research and Conservation, Columbia University,
420 W, 118th Streeet, Room 1306, New York, NY 10027 USA
d
Department of Agricultural and Resource Economics, University of California, Berkeley, CA 94720, USA

Received 26 August 2001; received in revised form 11 February 2002; accepted 25 April 2002

Abstract
The transformation and degradation of tropical forest is thought to be the primary driving force in the loss of biodiversity
worldwide. Developing countries are trying to counter act this massive lost of biodiversity by implementing national parks and
biological reserves. Costa Rica is no exception to this rule. National development strategies in Costa Rica, since the early 1970s,
have involved the creation of several National Parks and Biological Reserves. This has led to monitoring the integrity of and
interactions between these protected areas. Key questions include: ‘‘Are these areas’ boundaries respected?’’; ‘‘Do they create a
functioning network?’’; and ‘‘Are they effective conservation tools?’’. This paper quantifies deforestation and secondary growth
trends within and around protected areas between 1960 and 1997. We find that inside of national parks and biological reserves,
deforestation rates were negligible. For areas outside of National Parks and Biological reserves we report that for 1-km buffer zones
around such protected areas, there is a net forest gain for the 1987/1997 time period. Thus, it appears that to this point the
boundaries of protected areas are respected. However, in the 10-km buffer zones we find significant forest loss for all study periods.
This suggests that increasing isolation of protected areas may prevent them from functioning as an effective network. # 2002
Elsevier Science Ltd. All rights reserved.
Keywords: Costa Rica; Tropical deforestation; National parks; Biodiversity conservation

1. Introduction implementation of agreements and assessment of policy

goals (Daily and Ellison, 2002). Monitoring will not
The loss of biodiversity may be the single most only track global and regional deforestation trends, but
important consequence of global environmental change will also permit more spatially disaggregated analysis of
driven by land-use and land cover change. Finding regional landscape transformation and conservation
relatively undisturbed ecosystems, as well as seeking a policies.
place for biological resources within human-dominated Costa Rica’s history exemplifies the tension between
ecosystems, will be increasingly difficult as the size and economic development and environmental conserva-
resource demands of the human population continue to tion. While reported deforestation rates vary depending
increase. Finding a balance between the twin needs for on the assumptions used by different analysts (Sánchez-
production and conservation will require comprehensive Azofeifa, 1996), there is agreement that the country has
and fair policies, along with sound monitoring for had one of the highest tropical deforestation rates in the
world. That rate averaged in the vicinity of 3.7% from
the early 1970s until the early 1990s before dropping to
* Corresponding author. Tel.: +1-780-492-1822; fax: +1-780-492-
2030.
less than 1.5% at the end of the twentieth century
E-mail address: [Link]@[Link] (G. Arturo Sán- (Castro-Salazar and Arias-Murillo, 1998; FAO, 1990;
chez-Azofeifa). Sanchez-Azofeifa et al., 2001). The tension arises from
0006-3207/02/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S0006-3207(02)00145-3
124 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

the fact that many population sectors derived benefits totaled over $115 million dollar over the past 10 years
from this pattern of land use. From the Spanish arrival (Heindrichs, 1997).
until the end of the 1950s and beginning of the 1960s, Given such a significant public investment, there is a
thousands of hectares of forest were converted to crop- clear need for understanding how these areas have
land and pasture (Sader and Joyce, 1988). Official poli- fared, how they interact, and whether they are effective
cies prioritized demographic growth and new conservation tools. Our analyses are not sufficient for
agricultural production systems (Harrison, 1991; Solor- conclusions regarding the latter. However, they do
zano et al., 1991; Rosero-Bixby and Palloni, 1998; Sader indicate that National Parks and Biological Reserves
and Joyce, 1988; Sanchez-Azofeifa, 2000; Sanchez-Azo- may lower local deforestation rates, as to this point
feifa et al., 1999). Thus, as indicated in the national these areas have fared quite well. Further, our analyses
Strategy for Sustainable Development (ECODES), if suggest that ecological services requiring a park net-
50-year land-use/land-cover trends had continued, pri- work appear threatened, and the spatial linkages
mary forest commercial timber would have been deple- between parks are threatened by land-cover change.
ted by 1995 (Quesada-Mateo, 1990). Specifically, this paper addresses the following ques-
However, the trends changed. While precise knowl- tions: (1) what is the status of protection of native
edge of why this occurred will require more research, habitats in Costa Rica? (2) what are the contemporary
there are some clear candidate factors. Econometric (1986–1997) rates of deforestation within protected
analysis of deforestation over space and time suggests areas’ boundaries? (3) what are the long-term (1960–
that both reduction of forest stocks on productive lands 1997) rates of tropical deforestation in areas surround-
and changes in development strategies were significant ing national parks and biological reserves? (4) what are
factors (Kerr et al., 2000). Consider that some land in the long-term deforestation rates in proposed biological
Costa Rica is productive for coffee, while other land is corridors? and (5) given the current level of degradation
not. Clearing may be rapid on good coffee land, but of the proposed corridors, would they allow for con-
when that is cleared the deforestation rate is likely to nectivity between parks?
fall. Additionally, production and employment that
does not involve forest clearing can reduce deforestation
pressure. Urban population and employment have 2. The conservation areas
increased, as have both the attention given to and the
revenue derived from eco-tourism. The former draws In response to rapid land-cover change during the
people away from forest clearing, while the latter creates first part of the twentieth century, over the past 30 years
direct monetary returns from the conservation of intact the Costa Rican government has been creating a com-
forest, and can even lead to reforestation. prehensive protected areas system (Table 1). Between
The government has taken an active role in changing 1974 and 1978 the area covered by National Parks and
development strategies. Steps taken by government Biological Reserves expanded from 3 to 12% of the
bodies include the establishment and consolidation of a national territory (Table 1). Currently, Costa Rica has
national park system, limited programs promoting sus- 25% of the national territory dedicated to conservation,
tainable management of tropical forests, and financial which involves the prohibition of productive activities
compensation of private landowners for environmental that could damage renewable or non-renewable resour-
services derived from their land (Castro-Salazar and ces (Castro-Salazar and Arias-Murillo, 1998).
Arias-Murillo, 1998). Public payments for environ- Though the current conservation network is extra-
mental services in Costa Rica are estimated to have ordinary in extent, it does not cover all of the different

Table 1
Establishment dates and characteristics of 132 protected areas in Costa Rica

Category Number Area (km2) No./decade % Of National

Territory
Before 1960 60s 70s 80s 90s

National Parks 24 5415 1 1 11 1 10 10.6

Biological Preserves 9 396 – – 5 2 2 0.8
National Wildlife Refuges 39 1810 – – – 9 30 3.5
Forestry Reserves 12 2915 – 2 6 1 3 5.7
Protection Zones 31 1786 – – 10 11 10 3.5
Wetlands 14 504 – – 1 1 12 1.0
Special categories 3 16 – 1 1 – 1 <0.1
Total 132 12,842 1 4 34 25 68 25.1
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 125

‘‘life zones’’ (Powell et al., 2000, Table 2). A life zone is either ‘‘Level-1’’ or ‘‘Level-2’’. Level-1 denotes areas
defined by a combination of elevation, relative humidity under absolute protection such as national parks and
and biotemperature; life zones differ in the nature of the biological reserves, where no land-cover change is
climax vegetation they produce [see Holdridge (1967) allowed. Level-2 conservation areas comprise a mix of
for a clear explanation of the concept of life zones and conservation areas with more relaxed regulations con-
application to Costa Rica]. Generally, when considering cerning land cover change. Level-2 conservation areas
the benefits of additional protected areas, it will be include, in general, forest reserves and wildlife refuges.
worth analyzing the gains to covering unique habitats, Though a steady process to consolidate National
given all of the habitats already covered in the network. Parks and Biological reserves has been in place over the
This might include protecting pieces of additional life last 30 years, the integrity of forests within Costa Rica’s
zones. It might also mean protecting forests that con- protected areas cannot be taken for granted (Busch et
nect existing components of the reserve network, as the al., 2000). A 1994 constitutional ruling by Costa Rica’s
remaining forest is significantly fragmented (Sanchez- Supreme Court held that the government could seize
Azofeifa et al., 2001). land from private owners only if the latter are fairly
During the last 20 years, Costa Rica has also seen the compensated. Much of the land within SINAC has not
conceptualization, implementation and enactment of yet been purchased, as the government seized without
three Forestry Laws (1979, 1986 and 1996). These purchase an estimated 76% of the land in protected
changing laws reflect the dynamic nature of the coun- zones; 74% in forest reserves; 15% in national parks;
try’s conservation system, and the adaptation of this 46% in biological reserves, national monuments and
system to the country’s evolving development policies. absolute natural reserves; 59% in national wildlife refu-
In addition to the Forestry Laws, a Biodiversity Law ges and 12% in mangroves (Segnini, 2000). While to this
(1998) was enacted to provide a legal definition of con- point the owners of that land have largely been con-
servation. Furthermore, in the last 10 years, there have vinced not to convert the forest areas to other uses,
been changes in the composition of the government eventually such conversion will take place unless the
agencies in charge of conservation. The National Sys- land is purchased and permanently protected. Several
tem of Conservation Areas (SINAC) was consolidated efforts, at the international level, are currently in place
in 1995 from the three major agencies in charge of the to try to buy these lands. The example of the Guana-
protection of the conservation systems (The Forestry caste Conservation Area is a clear example of the
General Direction, the National Parks Direction and potential success of such an initiative.
the Wildlife General Direction). Although Costa Rica’s success in creating and start-
To date, SINAC has established 11 Conservation ing to fully consolidate a true conservation network of
Areas (different from National Parks and Biological protected areas is unique in Latin America, questions
Reserves) scattered over the country. These are terri- are arising about the ability of the current areas to
torial units, regulated by the same administrative and function effectively as a network. For instance, the
development program that carries out private and gov- Level-1 conservation areas are thought to be becoming
ernmental activities for the management and conserva- isolated in an ever more intensively used landscape, and
tion of natural resources. The territories dedicated to the government and environmental organizations seem
conservation inside of these regions are designated to feel that the intended function of national parks and

Table 2
Percentage of total life zone protected by both national parks and biological reserves in Costa Rica. Life zones ranked as percentage of total area
non-protected

Life zone type Total area (km2) % Total area % Total area % Total area
protected protected (NP) protected (BR)

Moist lower-montane 23,925 0.0 0.0 0.0

Wet montane 1728 0.0 0.0 0.0
Tropical humid 1,068,499 1.3 1.2 0.1
Wet forest premontane 1,198,697 1.8 1.4 0.3
Tropical humid 113,790 2.5 2.5 0.0
Tropical dry 141,544 3.7 3.7 0.0
Wet tropical 1,151,777 12.1 11.3 0.8
Moist premontane 553,550 12.8 12.3 0.5
Wet montane 373,038 25.5 23.2 2.3
Rain lower-montane 348,464 45.1 44.9 0.1
Rain forest montane 127,785 54.5 53.7 0.7
Sub-Alpine paramo 4590 93.1 93.1 0.0

NP, National Park; BR, Biological Reserve.

126 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

biological reserves may not be achieved if connectivity data sets have been explained elsewhere (Chomentowski
between these areas does not exist. et al., 1994; Sanchez-Azofeifa, 1996, 2000; Skole and
Some current efforts are aiming to connect Level-1 Tucker, 1993). This map was validated using a total of
conservation areas. The Costa Rican government and 800 control points collected by the Tropical Science
environmental organizations had recognized since the Center. Overall accuracy of the map, for the forest/non-
mid 1990s that the conservation and integrity of forest classification, is estimated to be 92% for all forest
national parks and biological reserves could not be units with a minimum mapping unit of 0.03-km2 and
achieved if connectivity between then does not exist. In canopy closure equal or higher than 80%.
order to account for integration at the country level, as Cloud cover over the Talamanca region, the Osa
well as at a regional level (Mesoamerican Biological Peninsula and a section of northwest Costa Rica limited
Corridor), a series of biological corridors have been interpretation of landscape characteristics using the
recommended for implementation. This network of FONAFIFO data set. This problem was corrected by
biological corridors is known as GRUAS and represent integrating three new satellite images acquired by the
the result of a multi-disciplinary discussion processes Cotopaxi Landsat TM receiving station. Images were
between the government and private groups in Costa acquired for 1987 and 1998; processed using the same
Rica. Additional information on the nature of the par- algorithms used by FONAFIFO and were later inte-
ticipatory processes followed to design this biological grated into the original Geographic Information System
corridors can be found on Garcia (1995). A key ques- (GIS). Despite the availability of new satellite images,
tion, of course, is whether these proposed corridors can cloud cover still obscured the most north-central part of
be, and in fact will be, effective in integration. Costa Rica, but this did not present a serious problem
for our analyses since there are no national parks or
biological reserves in the region.
3. Materials and methods The following data sets were also digitized and used
to create a GIS database: location of national parks,
In order to assess long-term land-cover change we biological reserves and proposed biological corridors
used three digital land-cover (forest / non-forest) data scale 1:50,000 (Fig. 1), boundaries for conservation
sets. The first was derived from 1:250,000 topographic regions (1:200,000), and Holdridge life zones
maps produced in 1960 from aerial photography (1:200,000). Additionally, a derived data set consisting
acquired by the US Defense Mapping Agency. This of a buffer zone of 0.5-, 1- and 10-km around all
serves as the baseline for our study, since major defor- national parks and biological reserves was generated
estation was initiated in the early 1970s (Sanchez-Azo- using the GIS. The 0.5- and 1-km buffer zones were
feifa, 2000). The second data set was created from selected to study which parks are currently being most
digitizing a total of ten 1:200,000 scale land cover maps. affected by local deforestation while the 10-km seeks to
The 1979 land cover map was produced by the Costa document the potential isolation of the parks and bio-
Rica National Meteorological Institute (IMN, 1992) logical reserves at a regional scale. A long-term analysis
using black and white prints of images acquired from of deforestation rates (1960–1979, 1979–1986 and 1986–
the Landsat Multispectral Scanner Satellite (MSS). This 1997) was performed for the 10-km buffer zone and all
map reveals the height of the deforestation process in Biological Corridors to quantify the land cover change
the country. There are no validation/ground truth stud- dynamics. Fig. 2 presents the 10-km buffer zone around
ies (field verification) available for these two data sets. all Level-1 conservation areas.
Both maps were integrated into a Geographic Informa- Connectivity between national parks was studied con-
tion System using a minimum mapping unit of 0.03- sidering two landscape structure variables: percolation
km2. Additional description of these two data sets can probability (Gardner and O’Neill, 1991; Stauffer, 1985)
be found in Pfaff et al. (2000). and landscape division (Jaeger, 2000). Connectivity,
The third set of digital, forest/non-forest information defined for this study as the possibility of linkage
was provided by Costa Rica’s National Forest Finan- between Level-1 conservation areas by means of con-
cing Fund (FONAFIFO; Castro-Salazar and Arias- nected forest fragments, was studied using two different
Murillo, 1998). FONAFIFO’s data set was produced approaches (a) biological corridors plus National Parks
from the interpretation of Landsat 5 Thematic Mapper and Biological Reserves; and (b) just biological corridors
(TM) scenes (seven spectral bands and 28.5 m spatial without any connection to National Parks or Biological
resolution) acquired during the years 1986 and 1997. Reserves. We implemented this analysis to quantify the
The data set had the following categories: forest, non- relative impacts additional parks could have in promot-
forest, cloud, shadow, secondary growth (1986–1997), ing connectivity between the Level-1 conservation areas.
deforestation (1986–1997), continental water bodies, Although we recognized that the definitions of con-
and mangroves. The minimum mapping unit was 0.03- nectivity and percolation are both in fact species depen-
km2. Methods regarding the production of FONAFIFO dent, in this study we focus our questions on landscape
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 127

Fig. 1. Spatial distribution of Level-1 conservation areas (Black) and proposed level-1 and biological corridors (Gray).

fragmentation and the potential connectivity that may These two life zones account for the most fertile soils in
exist between the different forest islands. the country (Tosi, personal communication). Minimal
protection is also present on the tropical humid (1.3%),
wet forest premontane (1.8%), tropical humid (2.5%)
4. Results and tropical dry (3.7%). These findings are not a sur-
prise since in general this ecosystems are more suitable
4.1. Countrywide analysis for agricultural activities (Ewel, 1999) and more pres-
sure exists against the implementation of conservation
Table 2 presents the results from a countrywide ana- activities. Higher protection exists for life zones with
lysis taking into consideration the 12 Holdridge (1967) less productive soils and those at high elevations (rain
life zones existing in the country. Our results indicate forest montane, rain forest lower-montane), in which
that efforts to promote conservation in different life exploitation of forest and agricultural resources is ham-
zones are not uniform across the national landscape. No pered by both poor accessibility and rough terrain.
Level-1 conservation area currently exists for the moist Most of this protection is based on the presence of
forest lower-montane and wet forest montane life zones. National Parks rather than Biological Reserves.
128 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

Fig. 2. 10-Km buffer zones around Level-1 Conservation Areas, National Parks NP and Biological Reserves BR. Buffer zone are coded as follows:
1—Guanacaste/Santa Rosa NP, 2—Tenorio Volcano NP, 3—Lomas de Barbudal NP, 4—Arenal Volcano NP, 5—Braulio Carrillo NP, 6—Tortu-
guero NP, 7—Cabo Blanco NP, 8—Carara BR, 9—Manuel Antonio NP, 10—Chirripo/La Amistad NP, 11—Cahuita NP, and 12—Corcovado-
Piedras Blancas NP.

4.2. Regional analysis bands) resolution of Landsat TM. These could obscure
the detection of small encroachments, illegal deforesta-
Our results indicate that deforestation rates have been tion, and illegal mining (i.e. Corcovado National Park)
negligible inside of national parks and biological or selective logging within a national forest inventory.
reserves over the last 10-years (Table 3). Thus, while While we believe the basic conclusions of this paper are
again it would require additional analysis to conclude robust, such potential errors are worth noting, and per-
that areas aimed at preventing land cover change were haps suggest the value of having a good on-the-ground
effective, our results suggest that this type of protection understanding of land-use activities.
may have been effective. Further, human impacts may Our results (Table 4) also indicate that as distance
be even lower than indicated in Table 3 as, in many increases from national parks and biological reserves,
cases, observable large changes in forest cover inside total deforestation and deforestation rates also increase.
protected areas are the result of landslides triggered by Deforestation in the 0.5- and 1-km buffer zones is low
earthquakes and storms. For example, the 1991 earth- compared with that in the 10-km buffer area. A total of
quake centered outside the port of Limon (7.5 on 17.02 km2 (0.29%/year assuming a constant rate),
the Ritcher Scale) produced important landslides on the 35.96 km2 (0.32%/year) and 549.91 km2 (1.0%/year)
Atlantic side of Costa Rica specifically on the Tala- were deforested within the 0.5-, 1- and 10-km buffers
manca region covered by several of the National Parks. respectively.
This was verified by using 1:40,000 scale aerial photo- The Level-1 areas most affected by tropical deforesta-
graphy provided by FONAFIFO during 1997/1998. tion close to their boundaries (within the 0.5- and 1.0-km
Also, it is important to note limitations of this buffer zones) are Alberto Manuel Brenes, Corcovado,
approach in terms of spatial (30 m) and spectral (seven Chirripo/Amistad, Braulio Carrillo, Juan Castro Blanco,
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 129

Table 3
Total land cover change in national parks and biological reserves between 1986 and 1997

Name Status Deforestation Secondary Net forest

(km2) growth (km2) change (km2)

Amistad NP 1.92 4.77 6.69

Barbilla NP 3.24 0.00 3.24
Barra Honda NP 0.00 0.00 0.00
Braulio Carrillo NP 1.67 2.31 0.65
Cabo Blanco NP 0.00 0.00 0.00
Cahuita NP 0.00 0.01 0.01
Chirripo NP 1.77 0.33 1.43
Corcovado NP 1.99 1.06 0.93
Guanacaste NP 2.40 4.22 1.83
Juan Castro Blanco NP 2.10 1.93 0.18
Manuel Antonio NP 0.07 0.00 0.07
Palo Verde NP 0.00 0.00 0.00
Piedras Blancas NP 1.08 2.79 1.71
Rincon de la Vieja NP 2.21 2.07 0.13
Santa Rosa NP 0.00 0.00 0.00
Tapanti NP 0.55 0.00 0.55
Tortuguero NP 0.56 0.23 0.33
Volcan Arenal NP 2.80 6.26 3.46
Volcan Irazu NP 0.23 2.28 2.05
Volcan Poas NP 0.14 0.51 0.36
Volcan Tenorio NP 0.62 1.54 0.91
Volcan Turrialba NP 0.19 2.06 1.87
Alberto Manuel Brenes BR 0.00 0.28 0.28
Carara BR 0.43 5.15 4.72
Cerro Vueltas BR 0.00 0.12 0.12
Hitoy Cerere BR 0.18 0.00 0.18
Lomas de Barbudal BR 0.00 0.00 0.00

NP, National Park; BR, Biological Reserve.

Hitoy-Cerere, Tortuguero and Volcan Tenorio Volcan Arenal, Chirripo, Cahuita and Corcovado
(Table 4). At a regional scale (10-km buffer zone) regions present the former tendency. Third, we have
deforestation (Table 5) rates were significantly high those regions that show a tendency toward increased
during the 1979–1986 period when compared against deforestation rates during the last 10-years, specifically
the other two time periods. Average annual deforesta- the buffer zones around the Tortuguero, Braulio Car-
tion rates for the 1960–1979, 1979–1986 and 1986–1997 rillo and Manuel Antonio National Parks. These results
time periods are estimated to be 1.3, 2.9 and 0.8%. The are in contrast to the with current and long-term
high deforestation rate observed during the 1979–1986 investment coordinated by the national government and
time period is comparable with national deforestation conservation non-governmental organizations trying to
rates observed reported by FAO (1970s to 1990s) and control deforestation trends in those areas.
Sanchez-Azofeifa et al. (2001) for 1986–1992. Defor- Our results also indicate that areas surrounding
estation was greatest around Braulio Carrillo, Cahuita, Level-1 conservation areas are highly fragmented
Tortuguero and Volcan Tenorio. Parks on the Atlantic (Fig. 3) and with a domination of non-forest classes
slopes are experiencing higher deforestation on sur- over forest fragments (Fig. 4). Level-1 areas such as
rounding lands than those on the Pacific slopes, specifi- Braulio Carrillo and Cahuita are surrounded by highly
cally the Cahuita, Tortuguero and Braulio Carrillo fragmented landscapes (Fig. 5). Some outcomes can be
National Parks. linked to land use choices. High deforestation and
Three main trends are observed when deforestation habitat fragmentation of Level-1 conservation areas in
rates are compared across time periods. The first trend the Atlantic region is attributable to the intensification
is observed for the Guanacaste, Volcan Tenorio and and expansion of agricultural cash crops such as
Carara 10-km buffers where high deforestation rates banana, and pineapple (Sanchez-Azofeifa et al., 1999,
during the first two periods were followed by a sharp 2001).
reduction during the 1976–1997 period. The second Deforestation rates for all biological corridors are
trend reveals almost no deforestation during the first significantly reduced for the 1986–1997-time period
period, a significant increase during the second one (Table 6). The same trend observed at the regional level
and a dramatic reduction during the third time period. for buffer zones (Table 5) is also observed for the
130 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

Table 4
Total deforestation (1986–1997) and deforestation rates in 0.5- and 1.0-km buffer zones around all national parks and biological reserves, Costa
Rica

Name Status Total Mean annual Total Mean annual

deforestation deforestation deforestation deforestation
(km2) 0.5-km rate (%) 0.5-km (km2) 1-km rate (%) 1-km

Barbilla NP 1.26 0.44 3.00 0.53

Barra Honda NP 0.00 0.00 0.00 0.00
Braulio Carrillo NP 2.26 0.32 5.44 0.40
Cabo Blanco NP 0.00 0.00 0.00 0.00
Cahuita NP 0.00 0.00 0.00 0.00
Chirripo/Amistad NP 2.02 0.10 5.32 0.14
Corcovado NP 0.47 0.14 1.36 0.22
Guanacaste NP 0.51 0.26 1.24 0.36
Juan Castro Blanco NP 2.32 1.40 4.04 1.40
Manuel Antonio NP 0.23 2.11 0.64 3.08
Palo Verde NP 0.00 0.00 0.00 0.00
Piedras Blancas NP 0.72 0.79 1.04 0.56
Rincon de la Vieja NP 0.64 0.34 1.08 0.29
Tapanti NP 0.39 0.24 0.64 0.20
Tortuguero NP 2.19 0.70 3.04 0.53
Volcan Arenal NP 0.72 0.30 1.16 0.25
Volcan Irazu NP 0.43 1.23 0.96 1.25
Volcan Poas NP 0.03 0.02 0.04 0.01
Volcan Tenorio NP 1.52 0.76 4.12 1.09
Volcan Turrialba NP 0.23 0.49 0.40 0.47
Alberto Manuel Brenes BR 0.00 0.00 0.00 0.00
Carara BR 0.55 0.97 0.72 0.77
Cerro Vueltas BR 0.00 0.00 0.04 0.02
Hitoy Cerere BR 0.53 0.23 1.68 0.35

Table 5
Deforestation rates in a 10-km buffer zones around all national parks and biological reserves, Costa Rica (1960–1997)

Buffer zone name Deforestation Deforestation Deforestation

rate (1960–1979) rate (1979–1986) rate (1986–1997)

Guanacaste 2.8 5.0 0.5

Volcan Tenorio 1.8 5.9 2.4
Tortuguero 1.0 1.7 1.9
Lomas de Barbudal 3.1 0.0 0.0
Volcan Arenal 0.4 3.7 0.6
Braulio Carrillo 0.8 0.8 1.2
Chirripo 0.5 1.1 0.4
Carara 1.6 6.7 0.2
Cahuita 0.9 6.1 1.1
Cabo Blanco 0.0 0.0 0.0
Manuel Antonio 3.0 0.0 1.1
Corcovado 0.0 3.8 0.5

biological corridors (Table 6) where a significantly during the last 20 years (1979–1997), been them in
higher-than-average deforestation rate (3.1% per year) average 1.9% per year.
is observed for the 1979–1986 time period versus 1.1%
and 0.8% per year for the 1960–1979 and 1986–1997 4.3. Connectivity between Level-1 conservation areas
time periods. With the exception of the Juan-Northwest
(2% per year), the Volcanoes Rinco-Tenorio-Arenal Our results considering both percolation theory and
(3% per year) and Tortuguero (1.9% per year) corri- landscape division indicate, when carried out at a local
dors, all other connectivity regions present minimum scale eliminating the Level-1 areas, indicate a high
deforestation rates under 1% per year. Of major con- degree of landscape division and low percolation
cern is the Tortuguero proposed biological corridor that (Fig. 6). This suggests that currently proposed biologi-
shows no significant changes on deforestation rates cal corridors are not allowing for connectivity between
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 131

Fig. 3. Distribution of forest patches in 10-km buffer zones around Level-1 conservation areas in Costa Rica. Areas around Braulio Carrillo,
Chirripo/Amistad and Guanacaste/Santa Rosa National Parks present the highest levels of forest fragmentation. Black represents forest and gray
represents non-forest.

Fig. 4. Percentage of the total area of buffer zone (10-km) covered with forest around Level-1 conservation areas in Costa Rica. Black represents
forest cover and gray non-forest cover.

Level-1 areas, due to the high level of landscape division. Costena, Braulio, Carara-Northeast and Palo Verde-
Our analysis indicates that 5 out of 22 biological corri- Rincon). Based on De-Camino-Beck and Sanchez-
dors present high landscape division and low percolation Azofeifa (2001), it seems that re-establishment of con-
(upper left corner of Fig. 6: Rincon-Tenorio-Arenal, nectivity between these corridors and parks would
132 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

Fig. 5. Percentage of forest and non-forest cover as function of patch density in the 10-km buffers around Level-1 areas in Costa Rica. Each dia-
mond signifies a Level-1 conservation area with numbers corresponding the regions outlined on Fig. 2. Areas are becoming isolated as non-forest
area increases and the number of forest fragments decreases.

require significant additional investment in ecosystem Level-1 conservation areas (national parks and biologi-
restoration. cal reserves) was negligible. This result can be see as in
contrast with the high levels of implied deforestation
in some regional land-use-change simulations done by
5. Discussion and conclusion others (e.g. Pontius et al., 2001). Thus, in answer to our
first question of interest, it seems conservation-area
Although Costa Rica has made significant efforts over boundaries are being respected to this point.
the last 30 years to produce one of the most compre- However, as we moved even just to larger buffer
hensive national park systems in the world, our study is zones, the story changed. We found significant clearing
one of the first to try to meet the demand for monitor- in proximate areas. In many cases, such as at the Cor-
ing one aspect of the success of the implementation of covado and Braulio Carrillo National Parks, deforesta-
this conservation network. We find that 1986–1997 tion has closely approached the park, and the
deforestation inside and within small buffers around the Tortuguero National Park is becoming isolated from
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 133

Table 6
Deforestation rates for all proposed biological corridors in Costa Rica

Corridor name Deforestation rate (1960–1979) Deforestation rate (1979–1986) Deforestation rate (1986–1997)

Alberto-Tortuguero 0.6 (72.0) 3.8 (63.4) 1.0 (46.5)

Alberto-Juan-Poas 1.2 (38.6) 1.6 (30.1) 0.4 (26.7)
Braulio 0.6 (93.9) 0.2 (83.3) 0.6 (82.1)
Carara-Chirripo 0.8 (74.4) 2.5 (63.2) 0.5 (52.2)
Carara-Northeast 2.7 (30.6) 5.2 (14.7) 0.8 (20.0)
Corcovado-Piedras 0.2 (89.1) 5.7 (93) 0.5 (56.1)
Costena 0 (80.1) 6.7 (79.5) 0.6 (41.9)
Hitoy-Cahuita 0.8 (98.3) 0.6 (83.6) 0.9 (80.3)
Juan-Northwest 0.7 (43.0) 0.5 (37.6) 2.0 (38.9)
Palo Verde-Rincon 2.7 (38.6) 5.4 (19.0) 0.4 (26.1)
Palo Verde-Southeast 1.3 (41.1) 6.8 (41.1) 0.0 (45.9)
Rincon-Tenorio-Arenal 2.6 (54.8) 1.4 (27.4) 3.0 (30.1)
Tenorio-Northeast 1.1 (56.3) 14.3 (44.1) N/A
Tortuguero 0.1 (85.5) 2.0 (84.4 1.9 (72.7)
Turrialba 0.3 (94.8) 8.5 (100) 0 (40.7)

Deforestation rate defined as percentage per year for the study period, data in parentheses represent percentage of forest cover for the corridor at the
beginning of the analyzed time period. N/A: area affected by cloud cover, negative values indicate secondary growth processes.

Fig. 6. Relationship between percolation probability and landscape division for all biological corridors connecting Level-1 conservation areas in
Costa Rica. Results indicate that 5 (1—Rincon-Tenorio-Arenal, 2—Costena, 3—Braulio, 4—Carara-Northeast and 5—Palo Verde-Rincon) out of
22 of the proposed biological corridors are current beyond recuperation in this generation and the amount of energy necessary to convert them to
functional corridors (increasing or creating the possibility of percolation) is too high.

the national conservation system due to high deforesta- fate accorded to many other Costa Rican parks like
tion rates and an increase in cash crops such as banana Braulio Carrillo, Volcan Rincon de La Vieja and Vol-
and palm heart plantations. Should the current condi- can Tenorio.
tions continue, we would expect that Tortuguero would Unfortunately, for provision of some forest services,
become completely isolated within the Costa Rica land- such as species habitat, there is reason to believe that
scape during this given time span, receiving the same conserved forest areas must be linked in an effective
134 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135

network (Forman, 1995; Noss and Cooperrider, 1994; (Grant No. BCS 9980252), The Tinker Foundation
Soulé and Ternorgh, 1999). For instance, if forest pat- (Institutional Grant). We would like to thank Dr. Suzi
ches are too small and/or too isolated, they are suscep- Kerr for her comments during the early preparation of
tible to loss of both genetic and species diversity, this manuscript. We also thank that guidance and com-
especially during extreme climatic and other disturbance ments from two anonymous reviewers. In addition, this
events. Gene flow is restricted and extinction of local work was conducted as part of the Carbon Sequestra-
plant and animal populations becomes a significant tion Working Group supported by the National Center
possibility. For these reasons, corridors that link con- for Ecological Analysis and Synthesis, a Center funded
servation areas are thought to be important for habitat by NSF (Grant No. DEB-0072909), the University of
services, and thus for the long-term protection of biodi- California, and the Santa Barbara campus.
versity. In light of such value of linkages, it is unfortu-
nate that in answer to our second question, increasingly
Costa Rican National Parks and Biologic Reserves are References
not functioning as a linked conservation network.
Highly fragmented forest corridors, in conjunction with Busch, C., Sathaye, J., Sanchez-Azofeifa, G.A., 2000. Estimating the
forests on areas not suitable for agricultural expansion, Greenhouse Gas Benefits of Forestry Projects: A Costa Rican Case
or in regions on which timber extraction occurred long Study. LBNL-42289, Lawrence Berkeley National Laboratory,
Berkeley, CA.
ago, cannot necessarily be considered a conservation
Castro-Salazar, R., Arias-Murillo, G., 1998. Costa Rica: Toward the
success. Sustainability of its Forest Resources (Technical Report). Fondo
This raises the issue of the effects of past policies Nacional de Financiamiento Forestal, San Jose, Costa Rica.
aimed to stop tropical deforestation, and of the effects Chomentowski, W., Salas, B., Skole, D., 1994. Landsat Pathfinder
of current incentives and mechanisms aimed to promote project advances deforestation mapping. GIS World 7 (4), 34–38.
Daily, G.C., Ehrlich, P.R., Sanchez-Azofeifa, G.A., 1998. Countryside
forest conservation. We cannot simply assert that the
biogeography: Utilization of human-dominated habitats by the avi-
lack of deforestation observed in conservation areas fauna of southern Costa Rica. Ecological Applications 11 (1), 1–13.
identifies those areas as effective conservation tools. Daily, G.C., Ellison, K., 2002. The New Economy of Nature:
These National Parks and Biological reserves are not The Quest Make Conservation Profitable. Island Press, Washing-
randomly located, but may have been located in places ton, DC.
De-Camino-Beck, T., Sanchez-Azofeifa, G.A., 2001. Percolation in
where there is less deforestation pressure. However, neutral landscapes: Landscape metric sensitivity to critical thresh-
controlling for other drivers of deforestation and the olds and a new class division index. National Network of Centers of
process of park location in testing for the effects of these Excellence-Sustainable Forest Management Network. Working
policies is part of our future research agenda. Joining paper no. 2001–4.
the current analyses with econometric analyses of Costa Ewel, J.J., 1999. Natural Systems as models for the design of sustain-
able systems of land use. Agroforestry Systems 45, 1–21.
Rican land use cited earlier will permit precisely such
FAO, 1990. 1990 Forest Resources Assessment. United Nations Food
tests. and Agriculture Organization, FAO, Rome: Italy.
In terms of future policy actions, if it is determined Forman, R., 1995. Land Mosaics the Ecology of Landscapes and
that the forest services that could be provided by a Regions. Cambridge University Press, New York.
linked conservation network would be of sufficient Garcia, R., 1996. Propuesta tecnica de ordenamiento territorial con
fines de conservacion de biodiversidad. Informe de pais: Costa Rica.
value, then Costa Rica needs to act to protect rich bio-
Project Corredor Biological Mesoamericano CCA-PNUD/GEF
logic resources by affecting land use outside of current RLA/95/G41.
national parks and biologic reserves. Policies must be Gardner, R.H., O’Neill, R.V., 1991. Pattern, process, and predict-
aimed to seek a balance between the goals of conserva- ability: the use of neutral models for landscape analysis. In: Turner,
tion and the need to use these lands for agricultural M.G., Gardner, R.H. (Eds.), In Quantitative Methods in Landscape
Ecology. Springer-Verlag, New York, pp. 289–307.
production (Daily et al., 1998). New approaches and
Harrison, S., 1991. Population growth, land use and deforestation in
mechanisms, such as developing regional conservation Costa Rica, 1950–1984. Interciencia 16 (2), 83–93.
networks through greater protection of buffer zones Heindrichs, T., 1997. Innovative Financing Instruments in the For-
around protected areas, creating protected biological estry and Nature Conservation Sector of Costa Rica, Sector Project.
corridors among protected areas, and strategically Deustche Gesellschaft fur Technische Zusammenarbeit, GTZ,
Eschborn, Germany.
locating new areas, would have to rise on the govern-
Holdrighe, L., 1967. Life Zone Ecology. Tropical Science Center, San
ment’s agenda. Jose.
IMN, 1992. Mapa de cobertura de la Tiena para 1992. Instituto
Meteorologico Nacional, Ministerio del Ambiente y Energia, San
Acknowledgements Jose, Costa Rica.
Jaeger, J., 2000. Landscape division, splitting index, and effective mesh
size: new measures of landscape fragmentation. Landscape Ecology
We would like to thank the generous support of the 15, 115–130.
Canada Foundation for Innovation (Grant No. 2041 to Kerr, S., Pfaff, A., Sanchez-Azofeifa, G.A., 2000. The Dynamics of
Sanchez-Azofeifa), the US National Science Foundation Deforestation and the Supply of Carbon Sequestration: Illustrative
 G. Arturo Sánchez-Azofeifa et al. / Biological Conservation 109 (2003) 123–135 135

Results from Costa Rica, Central America Project, Environment: Sanchez-Azofeifa, G.A., 1996. Assessing Land Use/Cover Change in
Conservation and Competitiveness. Harvard Institute for Interna- Costa Rica. Unpublished Doctoral Dissertation, University of New
tional Development, Harvard University, Boston. Hampshire, Durham, New Hampshire.
Noss, R.F., Cooperrider, A.Y., 1994. Saving Nature’s Legacy: Pro- Sanchez-Azofeifa, G.A., 2000. Land use and cover change in Costa
tecting and Restoring Biodiversity. Island Press, Washington, DC. Rica. In: Hall, C. (Ed.), Quantifying Sustainable Development.
Pfaff, A., Kerr, S., Hughes, F., Liu, S., Sanchez-Azofeifa, G.A., Schi- Academic Press, New York, pp. 477–505.
mel, D., Tosi, J., Watson, V., 2000. The Kyoto protocol and pay- Sanchez-Azofeifa, G.A., Harriss, R.C., Skole, D.L., 2001. Deforesta-
ments for tropical forest: an interdisciplinary method for estimating tion in Costa Rica: a quantitative analysis using remote sensing
carbon-offset supply and increasing the feasibility of a carbon mar- imagery. Biotropica 33 (3), 378–384.
ket under the CDM. Ecological Economics 35, 203–221. Sanchez-Azofeifa, G.A., Quesada-Mateo, C.A., Gonzalez-Quesada, P.,
Pontius, R., Cornell, J., Hall, C., 2001. Modeling the spatial pattern Dayanandan, S., Bawa, K., 1999. Protected areas and conservation of
of land-use change with GEOMOD2: application and validation biodiversity in the tropics. Conservation Biology 13 (2), 407–411.
for Costa Rica. Agriculture, Ecosystems and Environment 81, 191– Segnini, G., 2000. Gran deuda por conservacion. La Nacion (10
203. February).
Powell, G., Barborak, J., Rodriguez, M., 2000. Assessing representa- Skole, D., Tucker, C., 1993. Tropical Deforestation and Habital
tiveness of protected natural areas in Costa Rica for conservation Fragmentation in the Amazon: Satellite Data from 1978 to 1988.
biodiversity: a preliminary gap analysis. Biological Conservation 93, Science 260, 1905–1910.
41–53. Solorzano, R., de Camino, R., Woodward, R., Tosi, J., Watson, V.,
Quesada-Mateo, C., 1990. Estrategia de Conservacion para el Desar- Vasquez, A., Villalobos, C., Jimenez, J., Repetto, R., Cruz, W.,
rollo Sostenible de Costa Rica. Ministerio de Recursos Naturales 1991. Accounts Overdue: Natural Resource Depreciation in Costa
Energia y Minas, San Jose, Costa Rica. Rica. World Resources Institute, Washington, DC.
Rosero-Bixby, L., Palloni, A., 1998. Poblacion y deforestacion en Soulé, M.E., Ternorgh, J., 1999. Continental Conservation: Scientific
Costa Rica, Conservacion del Bosque en Costa Rica. Academia Foundations of Regional Reserve Networks. Island Press,
Nacional de Ciencias, San Jose, Costa Rica. Washington, DC.
Sader, S., Joyce, A., 1988. Deforestation Rates and Trends in Costa Stauffer, D., 1985. Introduction to Percolation Theory. Taylor &
Rica, 1940 to 1983. Biotropica 20, 11–19. Francis (Printers) Ltd, London, UK.