Agriculture, Ecosystems and Environment 199 (2015) 43–51

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

journal homepage: www.elsevier.com/locate/agee

Effects of management and landscape composition on the diversity and

structure of tree species assemblages in coffee agroforests
Achim Häger a, *, Mauricio Fernández Otárola b , Michelle Faye Stuhlmacher c,
Rafael Acuña Castillo b , Agustín Contreras Arias d
a
The School for Field Studies, Center of Sustainable Development Studies, Apartado 150-4013, Atenas, Alajuela, Costa Rica
b
Universidad de Costa Rica, Escuela de Biología, San Pedro de Montes de Oca, 11501-2060 San José, Costa Rica
c
George Washington University, 1922 F Street, NW, Washington DC 20052, USA
d
Universidad Estatal a Distancia, Mercedes de Montes de Oca, San José 11501-2060, Costa Rica

A R T I C L E I N F O A B S T R A C T

Article history: Understanding the processes that influence tree species composition in agricultural landscapes is
Received 22 February 2014 essential for conservation of tropical biodiversity outside of protected areas. We analyzed the effects of
Received in revised form 9 August 2014 landscape composition (amount of surrounding forest cover) and farm management (conventional vs.
Accepted 22 August 2014
organic) on the diversity and structure of woody plant species assemblages in Costa Rican coffee
Available online 7 September 2014
agroforestry systems. We utilized information from a GIS land-use database, surveys of 1-ha plots located
in 14 coffee farms and 4 forest fragments, and farmer interviews on management practices. The coffee
Keywords:
farms harbored over 100 tree species, including 19% of the native tree species found in the surrounding
Agroforestry
Biodiversity conservation
forests. The majority of tree species on the farms were native (82%) to the study area and originated from
Ecosystem services natural regeneration (73%). Among the tree species that regenerated naturally, 71% were dispersed by
Forest fragmentation animals. On the other hand more than half of the individuals were non-natives (55%) and originated from
Organic agriculture planting, which resulted in low species similarity between farms and forests and a low density for most
Seed dispersal native species on the farms. Forest cover within a 1000 m radius around the farms varied between 4 and
38%. Increasing forest cover around the farms had a significant, positive effect on species richness;
especially on tree species dispersed by animals, and on species similarity between farms and forests. This
suggests that the connection to natural forests increases seed dispersal into adjacent farms. The number
of regenerated species was higher on the organic farms, but tree species richness was not affected by
management type. Although species assemblages on the coffee farms are strongly determined by natural
regeneration, the number of individuals contributed by these processes is low. Tree species conservation
in agricultural landscapes would greatly benefit from protecting remnant forests, from facilitating
natural regeneration processes and promoting native trees on farms, with particular attention to rare
species.
ã 2014 Elsevier B.V. All rights reserved.

1. Introduction
Agriculture is a main driver of habitat destruction, fragmentation,
and loss of biodiversity in the tropics (Harvey et al., 2008; Gibson
et al., 2011). Conversely, it has been argued that the future of tropical
biodiversity depends on the successful integration of rural live-
lihoods with conservation efforts in human-modified landscapes
* Corresponding author. Tel.: +506 2446 5558.
E-mail addresses: ahaeger@fieldstudies.org (A. Häger), maufero@gmail.com
(M. Fernández Otárola), mstuhlma@gwmail.gwu.edu (M.F. Stuhlmacher),
rafael.asurbanipal@gmail.com (R. Acuña Castillo), etnoarias@gmail.com
(A. Contreras Arias).
(Perfecto and Vandermeer, 2008; Philpott et al., 2008; Chazdon et al.,
2009). Agroforestry can act as an intermediary between the
dichotomy of managed ecosystems and natural forests by providing
habitat for forest-dependent species and by increasing connectivity
across altered tropical landscapes (Schroth et al., 2004). It is well
established that agroforestry systems with low management
intensities can increase the diversity of plants, vertebrates and
arthropods in managed landscapes, and that agroforests may share a
large proportion of species with adjacent natural forests (e.g.,
Baghwat et al., 2008; Perfecto and Vandermeer, 2008; Philpott et al.,
2008; Clough et al., 2010; Dawson et al., 2013).
Understanding the processes that mediate diversity patterns in
human-modified landscapes is essential for conservation outside
the protected areas. However, very little is known about how

http://dx.doi.org/10.1016/j.agee.2014.08.022
0167-8809/ ã 2014 Elsevier B.V. All rights reserved.
44 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
species assembly is determined by processes such as dispersal,
species sorting across different habitat types, or management
practices in tropical agroforestry systems (Chazdon et al., 2009;
Livingston et al., 2013).
One of the factors that affect species composition in
agroecosystems is the surrounding landscape mosaic. Forest
fragments may represent sources for biological diversity in
agricultural landscapes. In addition, the proximity of natural
forests determines the abundance and diversity of important
functional groups including mammals, birds and
arthropods – which act as predators, seed dispersers or pollinators
(Ricketts, 2004; Tscharntke et al., 2008; Clough et al., 2010;
Livingston et al., 2013). Few studies have analyzed the relation-
ships between landscape composition and tree species present in
agroforests (e.g., Faria et al., 2007; Sonwa et al., 2007; Dawson
et al., 2013), although such relationships are well established for
more mobile organisms. For instance, Tscharntke et al. (2008) and
Clough et al. (2010) reported that the diversity and composition of
birds and different insect taxa in tropical agroforestry systems
were significantly altered by increasing distance to surrounding
forests and by management intensification. Despite their interde-
pendence, plant species similarity between agroforests and natural
forests is often lower, compared to animals (Baghwat et al., 2008).
This is partially due to the fact that trees are more directly
subjected to farmers’ management decisions, such as planting,
pruning or removal of individuals.
Organic management may further enhance ecological benefits
of agroforestry systems such as shade grown coffee farms
(Philpott et al., 2007). Organic farming limits the use of
agrochemicals like herbicides that can disrupt the natural
regeneration of native tree species in conventional agricultural
systems in the tropics (Esquivel et al., 2008). In addition, organic
farmers rely more directly on services provided by trees – such as
erosion prevention, watershed protection, pest control, or
nitrogen fixation – benefits that may encourage organic farmers
to plant or maintain certain tree species along with their crops
(Muschler, 2000; Perfecto et al., 2007).
This study aimed to understand the effects of landscape
composition (the amount of forest cover within a 1000 m radius
around each farm) and management practices (conventional vs.
organic) on tree species assemblages in Costa Rican coffee
agroforestry systems. We utilized information on landscape
composition from a GIS land-use database, field data on
tree species diversity from 14 coffee farms and 4 nearby forest
fragments, farmer interviews on farm history and management, as
well as information on seed dispersal mechanisms. It was expected
that tree species diversity and species similarity between farms
and forests will increase with increasing forest cover around the
farm and that organic farms will show higher diversity and
similarity with forests than conventional farms. It was also
expected that farms surrounded by forest will show a higher
number of naturally regenerated species and a higher proportion
of animal dispersed tree species in comparison with more isolated
farms. Finally it was hypothesized that the utility of a given tree
species would determine its frequency across the farms, as farmers
should prefer tree species with higher perceived utility.
2. Methods
2.1. Study site
This study was conducted between 2008 and 2013 in the
Western Central Valley of Costa Rica. Data were collected on
14 coffee farms (7 conventional and 7 certified organic) and 4
forest fragments located within the Rio Grande watershed (9.98 N,
84.40 W, Fig. 1). The farms were assessed between November
2008 and April 2011. Forest fragments were assessed between
November 2011 and April 2013. Elevation at the study sites ranged
from 800 to 1250 m a.s.l. Farms varied in size from 1.4 to 24.5 ha.
Forest fragments varied between 10 and 80 ha in size and consisted
of mature secondary forest with patches of old growth forest.
Annual rainfall ranges between 2000 and 2500 mm, with a distinct
dry season from December to May. All study sites were located
either in premontane wet forest or in the tropical moist (transition
to premontane) forest according to the Holdridge Life Zone
classification (ITCR, 2008).
Organic farms were chosen first, due to their limited number
within the study area, and then 7 topographically similar
conventional farms were selected. All farms had coffee crop areas
>1 ha with a cover of shade trees. Farms had been cultivating coffee
for at least 15 years and previous land uses included forests, annual
crops (maize, beans, plantains, tobacco) or pastures. All organic
farms were certified for at least 7 years prior to the time of the
study. Forest fragments were selected based on local land use maps
(scale 1:10,000) from 1991 and 2008 (IGNCR, 1991; IGNCR, 2008).
These maps were based on imagery from 1989 and 2005,
respectively. We selected forest fragments located within the
Rio Grande watershed that were found within the same elevation
range as the farm sites, and had a minimum area of at least 10 ha on
both the historic and current land use maps.
2.2. Data collection
Forest cover was determined within a 1000 m radius around
each farm based on the assumption that most trees have the ability
to disperse within this range (Corlett, 2009). The amount of forest
cover around the farms was extracted from a revised land-use
database, provided by the Costa Rican Geographic Institute
(IGNCR, 2008) using ArcGIS 10.0 (ESRI, 2011) (Fig. 1).
Structured interviews with farmers on farm history and
management were conducted between 2008 and 2013. Interview
questions focused on previous land use, time since land conversion
to coffee, time since organic certification, and means of tree
establishment on each farm. The owners of all 14 farms were
interviewed, with one exception. In the case of the remaining farm,
information was obtained from the owner of the neighboring
coffee farm who indicated that he was thoroughly familiar with the
management and history of the property. A list of all tree species
found within 1 ha on each farm (see details on methods below) was
presented to the respective land owner. To compile these lists, local
common species names were used. Common names for less known
tree species were taken from León and Poveda (2000). If a common
name applied to several species, as for instance in the case of Inga
spp., or in the case of less known species, images from plant
identification books (Zamora-Villalobos and Pennignton, 2001;
Zuchowski, 2005; Condit et al., 2011) were used to clarify
uncertainties. Farmers were asked to indicate if a given species
was remnant, planted or regenerated naturally on their coffee
farm. All farmers were readily able to determine which species
were planted or large remnant trees from previous land uses. In the
vast majority of the cases they were also able to indicate which tree
species had regenerated naturally on their coffee farm. Across all
14 agroforests, farmers classified the origin of tree species in
291 out of 307 cases (each case being the presence of a given
species on a specific farm, coming from a total of 100 species).
There were only 16 cases (14 species) in which farmers indicated
that they were not aware of the presence of a species that was
previously identified during data collection in their coffee agro-
forests. These species were assumed to have regenerated naturally.
A 1-ha plot was established at the approximate center of each
farm. Farm plots were usually 100 m  100 m. In some cases plot
shape varied in order to fit a 1-ha study area within the perimeter
 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51 45

Fig. 1. Map of the study area showing the location of the 14 sampled coffee farms and land-use within a 1000 m radius around each farm. The smaller white circles indicate the
location of the 4 sampled forest fragments. All study sites are located within the Rio Grande Watershed in the Western Central Valley of Costa Rica.

of the farm. For the forests, the 1-ha study area was established
with transect lines that were 10 m wide. Transect lines varied
between 50 and 250 m in length and were evenly
spaced throughout the forest, according to the shape of the
fragment. ArcGIS 10.0 (ESRI, 2011) was used to create a 50 m buffer
zone around the inner perimeter of each forest fragment. No
transects were placed within the buffer to avoid edge effects when
sampling.
Within the 1-ha plots, each woody plant with a diameter at breast
height (DBH, at 1.3 m from the ground) 5 cm was identified to the
lowest taxonomic level possible. Woody monocots such as Dracaena
fragrans,Yucca guatemalensis or palms were included. If a plant was
unidentifiable, it was treated as a morpho-species. Unidentified
plants that could not be assigned an unequivocal morpho-species
(i.e., no plant material was accessible for identification), were
omitted from the analysis. On the farms 9 trees were excluded
(0.2% of the total individuals found) and in the forest fragments
92 trees were omitted (2.2% of the total individuals). Data on species
abundance from the 4 forest fragments were pooled into a single
dataset, for comparison with species composition on each farm.
Dispersal mechanisms for species that regenerated naturally
across farms were compiled from the literature when available or
based on personal observations of the authors. References were
Janzen (1982), Hartshorn (1991), Athiê and Dias, (2011), Kunz et al.
(2011). Tree species were usually found to be dispersed either by
more than one animal group (e.g., birds or mammals), or by wind
and gravity. Consequently, all tree species were classified in two
broader groups, as dispersed by animals or dispersed by
wind/gravity. Of the 73 species that were found to regenerate
naturally, one species was omitted, because dispersal mechanisms
could not be determined.
In addition, growth forms and potential uses for each woody
plant species were compiled from the following sources:
Poveda-Alvarez and Sanchez-Vindas (1999), Zamora-Villalobos
and Pennington (2001), Cordero and Boshier (2003), Flores-Vindas
and Obando-Vargas (2003), Zuchowski (2005), OTS (2009), Condit
et al. (2011), INBio (2013), STRI (2013). The following uses were
taken into account: shade (excludes shrubs), nitrogen fixing,
timber/firewood, fruits, living fence/wind breaks, medicinal uses,
biological control, apiculture, ornamental, and animal forage.
46 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
2.3. Data analysis
Randomized species accumulation curves were calculated for
farms and forests to assess if species composition was represented
exhaustively by the sampled plots.
Following Magurran (2004), the Sørensen Index (CS) was used
for pairwise calculation of presence/absence similarity in species
composition between each coffee farm and the combined dataset
from the 4 forest fragments:
CS = (2a)/2a + b + c,where a = number of species occurring in
both samples, b = number of species occurring only in sample 1 and
c = number of species occurring only in sample 2.
An ANCOVA was used to analyze the effect of management
(conventional, organic) and amount of surrounding forest area on
the floristic similarity between farms and forests (Sørensen Index).
ANCOVA were also used to determine the effects of management
and surrounding forest area on species assemblage on each farm
(number of tree species present, number of tree species
regenerated naturally, number of naturally regenerated tree
species that are dispersed by animals or by abiotic means). In
all cases the covariate forest area was log transformed – to achieve
a linear relationship with the dependent variables. Furthermore,
the relationship between the number of species regenerated
naturally on a given farm and its similarity with the surrounding
forests was analyzed by a linear regression.
The utility of each tree species identified on the farms was
divided in 2 categories, based on the information found in the
literature. ‘Low’ utility was assigned to species with 0–2 known
uses. Utility of species with 3 known uses was classified ‘high’.
We used ordinal logistic models to assess if farm management
(conventional, organic) and utility level affect the frequency of a
given tree species to be found planted, regenerated or remnant
across the 14 farms. For this analysis the total number of species
found on the farms was reduced from 104 to 98, because
4 morpho-species were removed and 3 species from the family
Fabaceae were combined.
In order to explore potential effects of spatial autocorrelation on
tree species composition and diversity in the agroforests, Mantel
tests were performed between pairwise Sørensen Indices (CS) and
geographic distances (see Legendre and Legendre, 1998), as well as
between the pairwise differences in species richness and
geographic distances among all coffee farms. For the Mantel test,
CS was converted into a distance measure, by using 1–CS.
Randomized species accumulation curves, with 1000 iterations of
sample aggregation, were created with EcoSim 7 (Gotelli and
Fig. 2. Species accumulation curves for woody plants (DBH > 5 cm) sampled within
1-ha plots on 14 coffee farms and 4 forest fragments. Error bars show 95%
confidence interval; ORG: organic farms, CON: conventional farms.
Entsminger, 2001). Pairwise Sørensen Indices were calculated using
PAST 2.10 (Hammer et al., 2001). ANCOVA, linear regression analysis
and ordinal regressions were performed using JMP 10 (SAS, 2012).
Mantel tests, with 10,000 matrix permutations, were performed in R
(R Core Team, 2012), using the ‘vegan’ community ecology package
(Oksanen et al., 2013), for the detection of spatial autocorrelation.
3. Results
In total we identified 5091 woody plants on the 14 coffee farms
and 4126 woody plants in the 4 forest fragments (see also
Appendix A and B). A total of 310 different species were identified,
255 in the forest and 104 on the farms. On average 1032  145
(average  standard deviation) woody plants were found per ha in
the forests, belonging to 96  13 different species. On the
conventional farms stem density was 249  152 trees and
non-coffee shrubs per ha, with 19  7 species. Organic farms
had 478  130 woody plants per ha, belonging to 25  9 species.
Species accumulation curves indicate that species composition for
both forests and farms are well presented by the samples (Fig. 2).
In total, 63 different species were identified on all conventional
farms and 86 species on the organic farms (Fig. 2, Table 1). There
were 19 exotic species, of those 10 were present on the
conventional farms and 17 on the organic farms. The exotic
species accounted for a substantial proportion of the total
individuals; on average they contributed 60  21% to the trees
found on conventional farms and 55  18% on the organic farms.
The most common exotic woody plant species included
Dracaena fragrans, Erythrina poeppigiana, Juglans olanchana, Citrus
spp. and Yucca guatemalensis, together they comprised 49% of all
individuals on the 14 farms. Five exotic species were found in the
forests, including Syzygium jambos, Inga edulis, Dracaena fragrans,
Carica papaya and Coffea arabica. Of the 251 native species present
in the forests, the conventional farms shared 32 (13%) and the
organic farms 44 (18%). In total all coffee farms harbored 48 native
species (19%) that were also present in the forests.
Across all farms, most species (73%) originated from natural
regeneration and less than half were planted (46%). Remnant trees
only contributed 16% of all species found (Table 1). However, tree
species could belong to more than one category on a given farm
and in these cases farmers could not reliably estimate proportions.
Thus, it is not possible to indicate the exact number of individuals
planted, regenerated or remnant on each farm. In general, it can be
stated that the majority of trees on the coffee farms originated
from planting. The number of trees that can unambiguously be
classified as planted was 3248 across all farms, accounting for 64%
of the total number of individuals. Some of the most abundant
species, such as Dracaena fragrans, Juglans olanchana or Citrus spp.
were exclusively planted. However, the number of planted
individuals is greatly underestimated, because other highly
abundant trees (e.g., Erythrina spp. and Inga spp.) originated from
both planting and natural regeneration on many farms. Erythrina
spp. was planted on 13 farms, among those Erythrina also
regenerated naturally on 8 farms. Erythrina spp., Dracaena fragrans,
Inga spp., Juglans olanchana and Citrus spp. alone accounted for 72%
of all trees (see also Appendix A).
Table 1
Number of tree species found within 1-ha plots on 14 coffee farms (7 conventional
and 7 certified organic), according to their origin. Tree species may belong to more
than one category on a given farm.
Farm type Planted Remnant Regenerated Total speciesa
Conventional farms 27 10 40 59
Organic farms 40 9 61 86
All farms 46 16 73 100
a
Excluding 4 morpho-species.
 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51 47

Table 2
Number of naturally regenerated tree species, found within 1-ha plots on 14 coffee farms, according to their dispersal mechanisms.

Farm type Dispersal mechanism

Animals Wind, gravity

Conventional farms 24 15
Organic farms 42 18
All farms 50 22

Table 3
ANCOVA results for analyzing the effects of surrounding forest cover and management type (conventional, organic) on different aspects of the tree species assemblages found
within 1-ha plots on 14 coffee farms. Significant P-values are in bold.

Y–variable F2,11 R2 adj. Whole model P Farm management P Surrounding forest area P
Number of species/ha 7.302 0.492 0.0096 0.0594 0.0070
Number of species regenerated 8.140 0.523 0.0068 0.0441 0.0054
Number of species regenerated, dispersed by animals 10.982 0.606 0.0024 0.0155 0.0026
Number of species regenerated, dispersed by wind or gravity 2.005 0.134 0.1810 0.4501 0.0863

Among the tree species that regenerated naturally on the farms,
over two thirds were dispersed by animals (Table 2). The most
common of the naturally regenerated species, however, were all
dispersed by wind. These species included Cordia alliodora
(regenerated on 12 farms), Cedrela odorata (11 farms), Diphysa
americana (10 farms) and Tecoma stans (8 farms), which together
comprised 13% of all trees found on the farms.
The amount of forest within a 1000 m radius around each farm
varied between approximately 12 and 120 ha, representing 4% to
38% of the total area within that radius (Fig. 1). Both, farm
management (conventional, organic) and surrounding forest area
showed significant relationships with aspects of tree species
assemblage across the farms: the combined effect of farm type and
surrounding forest area explained nearly 50% in the variation of
species richness among farms. Within this model the surrounding
forest area had a positive effect on the species richness within the
coffee farms, but no significant differences were found between
organic and conventional farms (Table 3). Farm management and
surrounding forest area both had a significant effect on the number
of species regenerated naturally on the farms (Table 3). The highest
significance was found for the relationship between farm
management, surrounding forest area, and the number of
regenerated species dispersed by animals (Table 3, Fig. 3). There
was no significant relationship between farm management,
surrounding forest area and the number of regenerated species
dispersed by wind or gravity (Table 3).
The floristic similarity between farms and forests was low
overall, for both organic (average CS = 0.10  0.04) and conventional
farms (0.08  0.03). Farm management and surrounding forest
area explained over 40% of the variation in similarity between
farms and forests (whole model R2adj. = 0.412, F2,11 = 5.560,
P = 0.0215). Within the model, farm management did not influence
the similarity between tree species in the farm and forest
(P = 0.1355), but surrounding forest area did have a positive effect
on the floristic similarity (P = 0.0120) (Fig. 4). Furthermore, there
was a highly significant relationship between the number of
species regenerated on the farms and their floristic similarity to the
surrounding forests (R2 = 0.897, d.f. = 13, P < 0.0001) (Fig. 5).
Of the 98 species that were present on the farms and could be
assigned a utility level, 65 were classified as ‘high’ (see Appendix A).
Utility level was an important explanatory variable for the frequency
at which a given species was planted across the 14 farms (x2 = 19.84,
d.f. = 1, P < 0.0001), whereas management type did not have a
significant effect (x2 = 0.65, d.f. = 1, P = 0.4191). On the other hand,
neither utility level nor farm management type showed a significant
relationship with the frequency that a tree species regenerated
naturally across farms (utility: x2 = 0.15, d.f. = 1, P = 0.6995; farm
type: x2 = 0.02, d.f. = 1, P = 0.8851) or the frequency a tree species

Fig. 3. Effects of the amount of surrounding forest cover and farm management on
the number of regenerated tree species, which are dispersed by animals on each Fig. 4. Effects of the amount of surrounding forest cover and farm management on
farm. Full circles represent conventional farms; hollow circles represent certified the floristic similarity between farms and forests. Full circles represent
organic farms. conventional farms, hollow circles represent certified organic farms.
48 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
Fig. 5. Relationship between the number of species that regenerated naturally on
14 coffee farms and the floristic similarity of each farm with the surrounding forests
(Sørensen Index).
remained on the farm from previous land uses (utility: x2 = 0.39,
d.f. = 1, P = 0.5330; farm type: x2 = 1.13, d.f. = 1, P = 0.2888).
Regarding the potential patterns of spatial autocorrelation in
species diversity and composition, there was no relationship
between differences in species richness and geographic distances
among coffee farms (r = 0.029, P = 0.399). On the other hand, a weak
positive correlation was detected between species dissimilarity
(1–CS) and geographic distance (r = 0.253, P = 0.041).
4. Discussion
4.1. Tree species assemblage
Both, anthropogenic factors and natural processes can influ-
ence the vegetation composition in tropical farmlands. We found
that tree species assemblages in coffee agroforests were deter-
mined by interacting factors: the amount of forest cover within
the surrounding landscape, management type of the farm
(conventional or organic) and seed dispersal mechanisms. In
addition, the utility of a given tree species affected the frequency
at which it was planted across the farms. Tree species richness on
farms was 40% that found in the surrounding forests. This is less
than the average proportion of 64%, reported by Baghwat et al.
(2008) for tropical agroforestry systems around the world.
Floristic similarity between coffee farms and forests was very
low, as most individuals on the farms were planted (>60%) and
exotic (55%). The coffee farms only shared 19% of the native tree
species with surrounding forests, which is similar to the
proportion found by Pinard et al. (2014) for coffee agroforestry
systems in Kenya.
On the other hand, most of the species present on farms were
native to the study area (82%) and originated from natural
regeneration (73%). Albertin and Nair (2004) reported a similar
proportion (69%) of shade tree species that were established
naturally on coffee farms in North-Western Costa Rica. Among the
naturally regenerated trees on the farms, most individuals were
dispersed by wind, but the majority of species were dispersed by
animals, most importantly birds. The results show that species
diversity on the coffee farms is strongly determined by natural
regeneration; however, the number of individuals contributed by
these processes is low.
The most important variable for shaping tree species assemb-
lages was the amount of forest cover within a 1000 m radius
around each farm. Increasing forest cover around the farms had a
positive effect on tree species diversity and floristic similarity
between farms and forests. The logarithmic relationships between
forest cover and species composition suggest that tree diversity, as
well as floristic similarity between farms and forests, declines
rapidly in landscapes with <15% forest cover in the study area. To
date there is not much quantitative information available on the
relationships between landscape composition and tree species
diversity in agroforestry systems; although Sonwa et al. (2007)
reported that higher tree diversity of cocoa agroforests in
Cameroon was paralleled by increasing forest cover around farms,
with the highest diversity found in areas with roughly 20–60%
primary forest.
In addition, the spatial structure of tree species diversity and
composition across the coffee farms was taken into account to
better understand species assemblages within the landscape. Most
ecological variables show a certain degree of spatial autocorrela-
tion as a result of interdependent spatial patterns in the physical
environment and biological processes (Legendre and Legendre,
1998). Differences in tree species richness on the coffee farms did
not show a significant relationship with geographic distances
between these sites. At the same time there were strong
relationships between tree species diversity, the amount of forest
cover around the farms, and management practices. These findings
suggest that the variation in species richness, which is not
explained by the amount of surrounding forest cover and
management practices, may not be due to unaccounted spatial
factors. On the other hand it was found that floristic similarity
among farms increased with geographic proximity. Spatial
autocorrelation of species composition was considerably weak
and could at least be partially due to biological processes, such as
seed dispersal.
4.2. Seed dispersal mechanisms
Our results are in agreement with theoretical considerations
about migration dynamics and dispersal limitation as well
as with findings for other organisms, such as arthropods or
birds (Tscharntke et al., 2008; Clough et al., 2010). Forest cover
around the farms had a positive effect on both the number of
naturally regenerated species present on farms, and on the
floristic similarity between farms and forests. This implies that
the surrounding forests are an important source for seed
dispersal into the farms, i.e., that mass effects play a role in
shaping species composition at the coffee farms. This is further
supported by the strong linear relationship between the
number of species regenerated on each farm and the similarity
between farms and forests. Furthermore, increasing forest cover
around the farms had a positive effect on the number of
naturally regenerated species that were dispersed by animals.
These findings confirm that the connection to natural
habitats increases the activity of important functional groups,
such as seed dispersers in agricultural landscapes (Tscharntke
et al., 2008; Clough et al., 2010). In turn, the number of
naturally regenerated species that are dispersed by wind or
gravity on a given farm was not affected by surrounding forest
cover. Similarly, Mayfield et al. (2005) argued that plant
communities in deforested areas have more abiotic dispersal
mechanisms in comparison to forests.
4.3. Farm management
In agroecosystems plants are directly affected by management
decisions (e.g., Lopez-Gomez et al., 2007; Clough et al., 2010), thus
farm management represents an obvious filter for the establish-
ment of tree species either by planting, natural regeneration or
from remnant individuals. It is well established that increasing
management intensity (for instance higher crop plant density,
 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
simplified vegetation structure or agrochemical use) reduces the
diversity of different taxa, such as trees, birds, amphibians, reptiles
or invertebrates in tropical agroforestry systems (e.g.,
Lopez-Gomez et al., 2007; Philpott et al., 2007, 2008; Perfecto
and Vandermeer, 2008; Clough et al., 2010; Wanger et al., 2010).
Our study found only limited evidence for the effect of farm
management type on tree species composition. Whereas the
number of naturally regenerated species (especially those dis-
persed by animals) was significantly higher on the organic farms,
tree species diversity and the similarity between farms and forests
were not affected by management type. Our results suggest that
organic management may facilitate natural regeneration, for
instance by eliminating herbicides. Furthermore, the average stem
density of non-coffee woody plants on organic farms was almost
twice as high as on the conventional farms. Thus, the vegetation
structure may provide more suitable habitats or corridors for seed
dispersing animals, such as birds.
4.4. Species utility
The utility of a given tree species clearly influenced the
frequency at which it was planted across the farms. Accordingly,
Dawson et al. (2013) argued that a tree species’ usefulness to
farmers should at least partially determine the density of this
species in an agricultural landscape. The composition of planted
trees found by our study reflects similar preferences as those
reported by Albertin and Nair (2004), who studied farmers’
preferences for shade trees in Costa Rica. They reported that
farmers predominantly select fruit trees, such as Citrus spp., timber
species (e.g., Cedrela odorata or Cordia alliodora) or shade trees that
can be pruned and fix nitrogen (e.g., Erythrina spp. and Inga spp.).
At the same time the majority of these farmers were indifferent
towards the origin of a given tree species (native vs. exotic).
The utility of a given tree species did not seem to interfere
significantly with the number of farms on which a species
regenerated naturally, indicating that farmers do not entirely
eliminate or tolerate these trees based on their potential uses. It
can be concluded that farmers will likely maintain a tree
species, at least at low densities, as long as it does not directly
compete with the coffee crop. Similarly, remnant trees were
maintained regardless of their utility level. The studied farms
did not include rustic systems with a canopy of remnant forest
trees (sensu Moguel and Toledo, 1999), for this reason large
remnant trees (e.g., Ficus spp., Anacardium excelsum, Enter-
olobium cyclocarpum) were very rare. It appears that most of
them were retained on the farms due to their large size.
4.5. Coffee agroforestry and tree diversity conservation
While most of the species on the farms were established by
natural regeneration, the number of individuals contributed by
these natural processes was very low, which lead to a high
number of rare species across the 14 farms. Within the
104 species found, 67% were represented by less than
10 individuals and almost half of the species (45%) occurred
only once or twice, indicating a density <0.15 individuals per
ha. Similarly, Pinard et al. (2014) reported that planted, exotic
tree species were dominant on Kenyan coffee farms, whereas
most native species occurred at low abundances. Dawson et al.
(2013) emphasized the consequences of rare species for the
long-term conservation value of agroforestry systems. The
authors argued that low density implies restrictions for
regeneration, especially cross-pollination and an increased
vulnerability to management interventions. According to
Dawson et al. (2013) native tree species with a density of less
49
than 0.1 mature individuals per ha may face a high risk of local
extinction from agricultural landscapes. In addition Livingston
et al. (2013) pointed out that the static character of most
studies on diversity in agroecosystems is problematic, as
species richness and similarity with surrounding forests tend
to decline over time after conversion to agriculture. For this
reason tree species conservation in tropical agricultural land-
scapes needs to focus on facilitating natural processes, such as
pollination and seed dispersal from remaining forests across
agroforestry systems. Thus, the protection of existing forest
fragments and the promotion of native trees on the farms, with
specific attention to species present at low densities should
have priority. Dawson et al. (2013) suggested that more
research is needed on the ideal configuration of planting
different tree species to ensure gene-flow across the landscape.
The successful maintenance of tree species diversity across
agroforests offers important synergies for conservation, as high
plant diversity is directly associated to higher diversity of
animals by increasing the permeability and suitability of the
agricultural matrix for many taxa, which in turn ensure the
maintenance of natural processes such as pollination, seed
dispersal and regulation of biotic interactions (Schroth and
Harvey, 2007; Perfecto and Vandermeer, 2008; Philpott et al.,
2008; Tscharntke et al., 2008; Koh and Ghazoul, 2010).
Although there is an agreement on the importance of
maintaining existing forests and increasing plant diversity for
the conservation of overall biodiversity in tropical agricultural
landscapes, farmers may be reluctant to accept management
prescriptions that could interfere with crop production or imply
additional expenses (Schroth and Harvey, 2007; Philpott et al.,
2008; Tscharntke et al., 2011; Kumaraswamy and Kunte, 2013). In
this context, more research and education on the utility of native
tree species is needed. We found over 60 species in the forests that
had high utility levels (3 or more known uses such as timber,
fire wood, soil improvement, medicinal uses, among others), but
were absent from the farms, indicating a potential underuse of the
tree resources present in local forests.
Kumaraswamy and Kunte (2013) proposed a framework of
incentives and policy measures that target the implementation of
sustainable agricultural practices which simultaneously secure
livelihoods, conserve biodiversity and provide ecosystem services,
such as carbon storage. In the specific case of Costa Rica, current
forestry laws protect remaining forests and riparian tree cover, and
establish a payment for ecosystem services program for land-
owners. These laws are applicable to forest management and
conservation and incentivize planting trees in agroforestry
systems. The application of these instruments can be optimized
by stringent enforcement of existing regulations and by linkage of
local incentive programs to international financial mechanisms for
greenhouse gas mitigation and forest conservation.
5. Conclusions
This study shows that species richness on the studied coffee
farms is mostly maintained by natural regeneration. We present
evidence for the importance of surrounding forest cover as a
source of seed dispersal into coffee agroforestry systems. Shade
grown coffee farms, particularly under organic management,
provide corridors that facilitate the reproduction, dispersal, and
survival of native tree species across tropical altered landscapes.
On the other hand, most individuals in coffee farms belonged to
planted, exotic tree species and the number of trees contributed
by natural processes was relatively low. Natural regeneration
processes on farms are restricted by a sequence of filters:
availability of seed sources (determined by amount surrounding
forest cover and tree species present on the farms), limitations
50 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
to seed dispersal (dependent on overall tree cover in the
landscape and permeability of the coffee farms for animals) and
effects from farm management (such as vegetation structure,
applications of herbicides or the active removal of individuals).
Tree species conservation is directly associated to the diversity
of other taxa in agricultural landscapes and would benefit greatly
from protecting remnant forests, from facilitating natural regen-
eration processes and promoting native trees on farms, with
particular attention to rare species that otherwise face the risk of
local extinction.
Acknowledgements
We would like to thank CoopeAtenas and APROCAFE for their
friendly cooperation throughout this study. John DeLeo provided
critical technical support and Lyndon State College provided funding
for building the GIS land-use database. Luis Poveda Alvarez helped us
with the identification of numerous plant samples from the forest
fragments and José Esteban Jiménez revised the final plant list. We
would further like to thank Gerardo Avalos and two anonymous
reviewers for their comments on earlier versions of the manuscript.
I am especially grateful for the invaluable assistance from Clair Fox,
Rosy Cohane-Mann and the SFS students who helped in the field. We
gratefully acknowledge the key financial and logistical support
provided by the School for Field Studies (SFS) Center for Sustainable
Development Studies in Atenas, Costa Rica
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.agee.2014.08.022.
References
Albertin, A., Nair, P.K.R., 2004. Farmers’ perspectives on the role of shade trees in
coffee production systems An assessment from the Nicoya Peninsula, Costa
Rica. Hum. Ecol. 32, 443–463.
Athiê, S., Dias, M.M., 2011. Frugivoria e dispersão de sementes por aves em Casearia
sylvestris Sw. (Salicaceae) na região centro-leste do Estado de São Paulo. Revista
Brasileira de Zoociências 13, 79–86.
Baghwat, S.A., Willis, K.J., Birks, H.J.B., Whittaker, R.J., 2008. Agroforestry: a refuge
for tropical biodiversity? Trends Ecol. Evol. 23, 261–267.
Chazdon, R.L., Harvey, C.A., Komar, O., Griffith, D.M., Ferguson, B.G.,
Martines-Ramos, M., Morales, H., Nigh, R., Soto-Pinto, L., van Breugel, M.,
Philpott, S.M., 2009. Beyond reserves: a research agenda for conserving
biodiversity in human-modified tropical landscapes. Biotropica 41, 142–153.
Clough, Y., Abrahamczyk, S., Adams, M.O., Anshary, A., Ariyanti, N., Betz, L., Buchori,
D., Cicuzza, D., Darras, K., Putra, D.D., Fiala, B., Gradstein, S.R., Kessler, M., Klein,
A.M., Pitopang, R., Sahari, B., Scherber, C., Schulze, C.H., Shahabuddin, Sporn, S.,
Stenchly, K., Tjitrosoedirdjo, S.S., Wanger, T.C., Weist, M., Wielgoss, A.,
Tscharntke, T., 2010. Biodiversity patterns and trophic interactions in
human-dominated tropical landscapes in Sulawesi (Indonesia): plants,
arthropods and vertebrates. In: Tscharntke, T., Leuschner, C., Veldkamp, E.,
Faust, H., Guhardja, E., Bidin, A. (Eds.), Tropical Rainforests and Agroforests
Under Global Change, Environmental Science and Engineering. Springer-Verlag,
Berlin Heidelberg, pp. 15–71.
Condit, R., Perez, R., Daguerre, N., 2011. Trees of Panama and Costa Rica. Princeton
University Press, Princeton, NJ.
Cordero, J., Boshier, D.H. (Eds.), 2003. Arboles de Centroamérica. Un Manual Para
extensionistas. Oxford Forestry Institute, Oxford.
Corlett, R.T., 2009. Seed dispersal distances and plant migration potential in tropical
East Asia. Biotropica 41, 592–598.
Dawson, I.K., Guariguata, M.R., Loo, J., Weber, J.C., Lengkeek, A., Bush, D., Cornelius, J.,
Guarino, L., Kindt, R., Orwa, C., Russell, J., Jamnadass, R., 2013. What is the
relevance of smallholders’ agroforestry systems for conserving tropical tree
species and genetic diversity in circa situm, in situ and ex situ settings?
Biodivers. Conserv. 22, 301–324.
Esquivel, M.J., Harvey, C.A., Finegan, B., Casanoves, F., Skarpe, C., 2008. Effects of
pasture management on the natural regeneration of neotropical trees. J. Appl.
Ecol. 45, 371–380.
ESRI, 2011. ArcGIS Desktop: Release 10. Environmental Systems Research Institute,
Redlands, CA.
Faria, D., Barradas-Paciencia, M.L., Dixo, M., Laps, R.R., Baumgarten, J., 2007. Ferns,
frogs, lizards, birds and bats in forest fragments and shade cacao plantations in
two contrasting landscapes in the Atlantic forest, Brazil. Biodivers. Conserv. 16,
2335–2357.
Flores-Vindas, E., Obando-Vargas, G., 2003. Árboles del Trópico Húmedo.
Importancia socio-económica. Editorial Tecnológica de Costa Rica, Cartago.
Gibson, L., Lee, T.M., Koh, L.P., Brook, B.W., Gardner, T.A., Barlow, J., Peres, C.A.,
Bradshaw, C.J.A., Laurance, W.F., Lovejoy, T.E., Sodhi, N.S., 2011. Primary forests
are irreplaceable for sustaining tropical biodiversity. Nature 478, 378–381.
Gotelli, N.J., Entsminger, G.L., 2001. EcoSim: null models software for ecology.
Version 7.0. Acquired Intelligence Inc., & Kesey-Bear. http://homepages.
together.net/*gentsmin/ecosim.htm.
Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics
software package for education and data analysis. Palaeontol Electron 4. http://
palaeo-electronica.org/2001_1/past/issue1_01.htm.
Hartshorn, G.S., 1991. Plantas. In: Janzen, D.H. (Ed.), Historia Natural de Costa Rica.
Editorial de la Universidad de Costa Rica, pp. 119–353.
Harvey, C.A., Komar, O., Chazdon, R., Ferguson, B.G., Finegan, B., Griffith, D.M.,
Martínez-Ramos, M., Morales, H., Nigh, R., Soto-Pinto, L., Van Breugel, M.,
Wishnie, M., 2008. Integrating agricultural landscapes with biodiversity
conservation in the Mesoamerican hotspot. Conserv. Bio. 22, 8–15.
IGNCR, 1991. Hoja 3345-IV-2Cuajiniquil, Hoja 3346-III-18Hornos, Hoja 3346-III-22.
Morazan, Hoja 3346-III-23 Eulalia. Instituto Geográfico Nacional de Costa Rica, San
José.
IGNCR, 2008. Hoja 3345-IV-2Cuajiniquil, Hoja 3346-III-18Hornos, Hoja 3346-III-22.
Morazan, Hoja 3346-III-23 Eulalia (digital versión, ESRI shape file). Instituto
Geográfico Nacional de Costa Rica, San José.
INBio 2013. Especies de Costa Rica. Instituto Nacional de Biodiversidad. http://
darnis.inbio.ac.cr.
ITCR, 2008. Costa Rica Atlas 2008. Escuela de Ingeniera Forestal, Laboratorio de
Sistemas de Información Geográfica, Instituto Tecnológico de Costa Rica,
Cartago.
Janzen, D.H.,1982. Natural history of guacimo fruits (Sterculiaceae: Guazuma ulmifolia)
with respect to consumption by large mammals. Am. J. Bot 69, 1240–1250.
Koh, L.P., Ghazoul, J., 2010. A matrix-calibrated species-area model for predicting
biodiversity losses due to land-use change. Conserv. Bio 24, 994–1001.
Kumaraswamy, S., Kunte, K., 2013. Integrating biodiversity and conservation with
modern agricultural landscapes. Biodivers. Conserv. 22, 2735–2750.
Kunz, T.H., Braun de Torrez, E., Bauer, D., Lobova, T., Fleming, T.H., 2011. Ecosystem
services provided by bats. Annals NY Acad. Sci. 1223, 1–38.
Legendre, P., Legendre, L., 1998. Numerical Ecology. Elsevier, Amsterdam, Boston,
London, New York, Oxford, Paris, San Diego, San Fransisco, Singapore, Sydney,
Tokyo.
León, J., Poveda, L., 2000. Nombres comunes de las plantas en Costa Rica. Editorial
Guyacán, San José.
Livingston, G., Philpott, S.M., Rodriguez, A.M., 2013. Do species sorting and mass
effects drive assembly in tropical agroecological landscape mosaics? Biotropica
45, 10–17.
Lopez-Gomez, A.M., Williams-Linera, G., Manson, R.H., 2007. Tree species diversity
and vegetation structure in shade coffee farms in Veracruz, Mexico. Agric.
Ecosyst. Environ. 124, 160–172.
Magurran, A.E., 2004. Measuring Biological Diversity. Blackwell Publishing, Oxford.
Mayfield, M.M., Boni, M.F., Daily, G.C., Ackerly, D., 2005. Species and functional
diversity of native and human dominated plant communities. Ecology 86,
2365–2372.
Moguel, P., Toledo, V.M., 1999. Biodiversity conservation in traditional coffee
systems of Mexico. Conserv. Bio. 13, 11–21.
Muschler, R.G., 2000. Árboles en cafetales. Módulo de enseñanza Agroforestal No. 5.
Centro Agronómico Tropical de Investigación y Enseñanza, Turrialba.
Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B.,
Simpson, G.L., Solymos, P., Stevens, M.H.H., Wagner, H., 2013. Vegan:
Community Ecology Package. R package versión 2. 0–10. http://CRAN.R-
project.org/package=vegan.
OTS, 2009. Flora digital de La Selva. Organization for Tropical Studies. http://sura.
ots.ac.cr/local/florula3/index.htm.
Perfecto, I., Vandermeer, J., 2008. Biodiversity conservation in tropical agro-
ecosystems. Annals NY Acad. Sci. 1134, 173–200.
Perfecto, I., Armbrecht, I., Philpott, S.M., Soto-Pinto, L., Dietsch, T.V., 2007. Shade
coffee and the stability of forest margins in Northern Latin America. In:
Tscharntke, T., Zeller, M., Leuschner, C., Guhardja, E., Bidin, A. (Eds.), Stability of
Tropical Rainforest Margins: Linking Ecological, Economic and Social
Constraints of Land use and Conservation. Springer-Verlag, Berlin, pp. 227–
263.
Philpott, S.M., Bichier, P., Rice, R.A., Greenburg, R., 2008. Biodiversity conservation,
yield, and alternative products in coffee agroecosystems in Sumantra,
Indonesia. Biodivers. Conserv. 17, 1805–1820.
Philpott, S.M., Bichier, P., Rice, R., Greenburg, R., 2007. Field testing ecological and
economic benefits of coffee certification programs. Conserv. Bio. 21, 975–985.
Pinard, F., Joetzjer, E., Kindt, R., Kehlenbeck, K., 2014. Are coffee agroforestry systems
suitable for circa situm conservation of indigenous trees? A case study from
Central Kenya. Biodivers. Conserv. 23, 467–495.
Poveda-Alvarez, L.J., Sanchez-Vindas, P.E., 1999. Árboles y Palmas del Pacifico Norte
de Costa Rica. Editorial Guayacan, San José.
R Core Team, 2012. R: a language and environment for statistical computing. R
foundation for statistical computing, Vienna. http://www.R-project.org/.
Ricketts, T.H., 2004. Tropical forest fragments enhance pollinator activity in nearby
coffee crops. Conserv. Bio. 18, 1262–1271.
SAS, 2012. JMP 10. SAS Institute Cary.
 A. Häger et al. / Agriculture, Ecosystems and Environment 199 (2015) 43–51
Schroth, G., Harvey, C.A., 2007. Biodiversity conservation in cocoa production
landscapes: an overview. Biodivers. Conserv. 16, 2237–2244.
Schroth, G., da Fonseca, G.A.B., Harvey, C.A., Gascon, C., Vasconcelos, H.L., Izac, A.N.,
Angelsen, A., Finegan, B., Kaimowitz, D., Krauss, U., Laurance, S.G.W., Laurance, W.
F., Nasi, R., Naughton-Treves, L., Niesten, E., Richardson, D.M., Somarriba, E., Tucker,
N.I.J., Vincent, G., Wilkie, D.S., 2004. Conclusion: agroforestry and biodiversity
conservation in tropical landscapes. In: Schroth, G., da Fonseca, G.A.B., Harvey, C.A.,
Gascon, C., Vasconcelos, H.L., Izac, A.N. (Eds.), Agroforestry and Biodiversity
Conservation in Tropical Landscapes. Island Press, Washington DC, pp. 487–501.
Sonwa, D.J., Nkongmeneck, B.A., Weise, S.F., Tchatat, M., Adesina, A.A., Janssens, M.J.
J., 2007. Diversity of plants in cocoa agroforests in the humid forest zone of
Southern Cameroon. Biodivers. Conserv. 16, 2384–2400.
STRI, 2013. STRI Herbarium. The Smithonian Tropical Research Institute. http://
biogeodb.stri.si.edu/herbarium/.
51
Tscharntke, T., Sekercioglu, C.H., Dietsch, T.V., Sodhi, N.S., Hoehen, P., Tylianakis, J.M.,
2008. Landscape constraints on functional diversity of birds and insects in
tropical agroecosystems. Ecology 89, 944–951.
Tscharntke, T., Clough, Y., Bhagwat, S.A., Buchori, D., Faust, H., Hertel, D., Hölscher, D.,
Juhrbandt, J., Kessler, M., Perfecto, I., Scherber, C., Schroth, G., Veldkamp, E.,
Wanger, T.C., 2011. Multifunctional shade-tree management in tropical
agroforestry landscapes – a review. J. Appl. Eco. 48, 619–629.
Wanger, T.C., Iskandar, D.T., Motzke, I., Brook, B.W., Sodhi, N.S., Clough, Y.,
Scharntke, T., 2010. Effects of land-use change on community composition of
tropical amphibians and reptiles in Sulawesi, Indonesia. Conserv. Bio. 24,
795–802.
Zamora-Villalobos, N., Pennington, T.D., 2001. Guabas y Cuajiniquiles de Costa Rica
(Inga spp.). Instituto Nacional de Biodiversidad, Santo Domingo, Heredia.
Zuchowski, W., 2005. A Guide toTropical Plants of Costa Rica. Zona Tropical, Miami, FL.