Biological corridors as important habitat

structures for maintaining bees in a tropical

fragmented landscape

Catalina Gutiérrez-Chacón, Carlos

Valderrama-A & Alexandra-Maria Klein

Journal of Insect Conservation

An international journal devoted to
the conservation of insects and related
invertebrates

ISSN 1366-638X
Volume 24
Number 1

J Insect Conserv (2020) 24:187-197

DOI 10.1007/s10841-019-00205-2

1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Nature Switzerland AG. This e-offprint is
for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197
https://doi.org/10.1007/s10841-019-00205-2

ORIGINAL PAPER

Biological corridors as important habitat structures for maintaining

bees in a tropical fragmented landscape
Catalina Gutiérrez‑Chacón1,2   · Carlos Valderrama‑A3,4 · Alexandra‑Maria Klein1

Received: 5 June 2019 / Accepted: 6 December 2019 / Published online: 11 December 2019
© Springer Nature Switzerland AG 2019

Abstract
Biological corridors are an important conservation strategy to increase connectivity between populations—mainly verte-
brates—in fragmented landscapes, which often require habitat restoration to achieve physical connections. Non-target groups
such as bees could benefit from corridors while contributing to the restoration process given their role as pollinators, but
little is known about the use of corridors by bees. Here we assessed the habitat value for bees of four biological corridors
in the Colombian Andes by comparing bee species richness, community composition and functional diversity between cor-
ridors (which had two land-cover sections: riparian forest and restored forest), forest patches being connected by corridors
and surrounding pastures. We found a higher species richness in riparian than in restored sections of corridors, which was
comparable to that in forest and higher than in pasture. Community composition in forest and riparian sections were similar
and differed from that in pasture. In contrast, functional diversity was similar among all land-use types, suggesting a higher
species redundancy in forest and riparian corridors, given the higher species richness, compared to pastures. Our results
show that riparian corridors are holding forest-associated species that could not survive in pastures, and given the higher
redundancy, can significantly contribute to the maintenance of pollination services in fragmented landscapes. Our results also
indicate that 13 years of restoration process have not been sufficient to reach reference levels (i.e. forest/riparian) in terms
of bee species richness, but the recovery of some forest-associated species points to the potential of biological corridors to
functionally connect forest patches.

Keywords  Restoration · Species traits · Tropical forest · Connectivity · Colombia

Introduction

Reestablishing connectivity to enable wildlife movement

and the flux of ecological processes is a central approach to
Electronic supplementary material  The online version of this
mitigate deleterious effects of habitat loss and fragmentation
article (https​://doi.org/10.1007/s1084​1-019-00205​-2) contains
supplementary material, which is available to authorized users. (Crooks and Sanjayan 2006). Biological corridors, which
are in general linear landscape elements serving as physical
* Catalina Gutiérrez‑Chacón connections between otherwise isolated habitat remnants,
cgutierrez@wcs.org have been widely promoted as a strategy to increase land-
1
Faculty of Environment and Natural Resources, Chair scape connectivity (Beier and Noss 1998; Hilty et al. 2012;
of Nature Conservation and Landscape Ecology, University Rudnick et al. 2012). While corridors are established with
of Freiburg, Tennenbacher Straße 4, 79106 Freiburg, the goal of connecting populations of vertebrates (Bennett
Germany 1999), other non-target groups such as insects, which are
2
Wildlife Conservation Society –Colombia Program, Av 5N # important providers of ecosystems services (Kremen and
22N‑11, Cali, Colombia Chaplin-Kramer 2007; Losey and Vaughan 2006), could
3
Department of Biological Sciences, Universidad Icesi, Cl. 18 benefit from corridors (Cranmer et al. 2012; Herrera-Rangel
# 122‑135, Cali, Colombia et al. 2015; Hill 1995). However, evidence is limited across
4
Present Address: Faculty of Natural Sciences ecosystems and biological groups (Hunter 2002; Öckinger
and Mathematics, Universidad del Rosario Calle, 12C No and Smith 2008).
6‑25, Bogotá D.C., Colombia

13
Vol.:(0123456789)
 Author's personal copy

188 Journal of Insect Conservation (2020) 24:187–197

The implementation of biological corridors often requires not provide information on which species compose local
the restoration of disturbed habitats, that have potential to communities and how they contribute to ecosystems func-
become near-natural habitats to achieve physical connections tioning (Cadotte et al. 2011). The responses of species to
(Hilty et al. 2012). As most wild flowering plant species environmental conditions and their effects on ecological pro-
depends to some extent on wild pollinators for their repro- cesses and ecosystem functions are determined by functional
duction (Ollerton et al. 2011), recovery of pollinators is cru- traits, which are morphological, physiological, phenologi-
cial to restore functional plant communities. Nevertheless, cal, or behavioral features that have an effect on the organ-
few studies have assessed the response of key pollinators ism fitness (Carmona et al. 2016; Kaluza et al. 2018). Thus,
such as bees to habitat restoration (but see Williams 2011; quantitative measures of trait variation or multivariate trait
Winsa et al. 2017), particularly in the context of biological differences within a community (referred to as ‘functional
corridors aimed to connect tropical forest patches. diversity) have been increasingly used to understand ecosys-
The tropical Andes reaches its higher complexity in tems functioning (Carmona et al. 2016; Laliberté and Leg-
Colombia, resulting in an outstanding ecosystem and high endre 2010; Petchey and Gaston 2006; Villéger et al. 2008).
biodiversity (Kattan et al. 2004). A substantial portion of Combined measures of species richness, community com-
this biodiversity, however, survive in fragmented landscapes position and functional diversity can thus provide a compre-
as cattle ranching and agriculture have largely diminished hensive understanding on how conservation actions influ-
natural habitats (Etter et al. 2006; Mendoza et al. 2006). ence different aspects of biodiversity (De Arruda Almeida
Since the late 90s, an increasing number of initiatives aimed et al. 2018; Dı́az and Cabido 2001).
at enhancing landscape connectivity through corridors has In this study, we assessed the habitat value of the Bar-
been developed in Colombia (Cracco and Guerrero 2004). bas-Bremen biological corridors combining measures of
The Barbas-Bremen biological corridors, implemented by taxonomical, compositional and functional diversity. Spe-
the Colombian Biodiversity Research Institute Alexander cifically, we asked whether bee species richness, commu-
von Humboldt, has been a national reference given its suc- nity composition, and functional dispersion differ between
cessful implementation through agreements and incentives restored and riparian sections of biological corridors, forest
to landowners (Gutiérrez-Chacón et al. 2013; Lozano-Zam- and pasture. Since the abundance of floral resources and
brano 2009). Four corridors were established to connect the amount of surrounding natural habitat may influence
two large patches of sub-Andean forests, the Barbas canyon bee diversity at a specific site (Winsa et al. 2017; Gutiér-
(790 ha) and the Bremen Forest Reserve (747 ha) (hence- rez-Chacón et al. 2018), we also evaluated these variables
forth ‘forest’), which are isolated mostly by a matrix of pas- to explain the relative importance of habitat succession,
ture for cattle ranching (henceforth ‘pasture’). To achieve resource availability and landscape context for bee diversity
forest corridors, existing riparian forests strips already con- and composition.
nected to the forest patches were used; pasture gaps between
forest strips were subjected to an active restoration process
with native plant species in 2004 (Vargas 2008). Hence, Materials and methods
each corridor is composed of two land-use types, riparian
forest (henceforth ‘riparian section’) and a restored forest Study area and biological corridors
(henceforth ‘restored section’). Target vertebrate species
such as the Cauca Guan (Penelope perspicax Bangs), the The four biological corridors are located in the Munici-
Red-Howler monkey (Alouatta seniculus Linnaeus), Ocelot pality of Filandia, Department of Quindío, in the western
(Leopardus pardalis) and Jaguarundi (Herpailurus yagoua- versant of the Central cordillera of the Colombian Andes
roundi) have been reported using the corridors, including the (4° 41 ′N, 75° 40 ′W), between 1800 and 2100 m a.s.l. The
restored sections (Gutierrez-Chacón; Valderrama-A, data no vegetation is sub-Andean forest sensu (Cuatrecasas 1958);
published), but long-term monitoring studies are lacking. A mean annual rainfall is 2817 mm and mean monthly tem-
recent study focused on hunting ants (Formicidae), however, perature ranges between 16 and 24 °C (Fagua et al. 2013).
suggest that restored sections have not yet reached the nec- Length of corridors connecting the Barbas river canyon
essary conditions to harbor a similar community to that in and the Bremen-La Popa Forest Reserve vary between 800
forests patches (Herrera-Rangel et al. 2015). and 2500 m, and 50–100 m wide. Distance among corri-
Indices based on species richness have commonly been dors range between 500 and 1300 m. Restored sections of
used to describe biodiversity (e.g. Magurran 1988) and how corridors connecting riparian forest strips were previously
it respond to conservation efforts, including habitat restora- covered with grass (Pennisetum clandestinum Hochst. Ex
tion (Wortley et al. 2013). Although species richness is the Chiov.) for cattle grazing. In 2010, restored sections had
simplest way to describe community and regional diversity not reached similar conditions in terms of canopy cover as
(Gotelli and Colwell 2010, 2001), this measure alone does compared to riparian forest strips, showing lowers values

13
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197 189

of canopy cover (Herrera-Rangel et al. 2015).We measured Trait assignment

canopy cover at three random points in each corridor sec-
tion with a spherical densiometer (Lemmon Forest Den- For each species, we assigned information on four traits that
siometers®), held at 1.5 m. We also found a significantly have been extensively used in previous studies, as they are
higher canopy cover in riparian sections (94 ± 4%, n = 12) known to respond to disturbance, extinction sensitivity and
than in restored sections  (67 ± 31%, n = 12) (t = 2.55, ecological interactions: body size, nest location, method
df = 8.18, p = 0.03). of nest construction and sociality (Bommarco et al. 2010;
Jauker et al. 2013; Williams et al. 2010). First, body size was
measured as the distance between the two insertion points of
Bee sampling the wings (i.e. inter-tegular distance (ITD)). ITD is an indi-
cator of the flight musculature (Cane 1987) and it is a strong
Bees were surveyed in four sites within each land-use type, predictor of foraging ranges (Greenleaf et al. 2007). ITD
i.e. forest, pasture, riparian and restored sections of corri- was measured from pinned specimens and averaged by spe-
dors, for a total of 16 sampling sites (Fig. 1). In forest and cies from 10 individuals, or the maximum number available.
corridor sections, bees were sampled along edges to avoid Second, each bee species was classified as below or above-
differences due to edge effects between riparian strips and ground nester, the latter group including species that nest in
continuous forest (Winfree et al. 2007), and because capture pre-existing cavities in trees and those that burrow tunnels
rates of bees inside tropical forest are usually low even com- into wood. Third, in terms of method of nest construction,
bining several sampling methods (Brosi et al. 2008). Sites we use the term ‘rent’ for species that nest in existing holes
in pasture were distanced at least 100 m from any forest or cavities either above or below ground; ‘construct’ for
area. In each site, bees were sampled along a 150 × 4 m plot, species that build the nest exterior using collected materi-
which was surveyed during two periods, June–July 2014 and als, and ‘excavate’ for species that drill the tunnel/cavity
January 2015. In each period, we sampled every site three in which the brood cells are constructed (following Wil-
times combining the following methods: (i) aerial netting liams et al. 2010). Finally, species were classified as social
for 40 min between 8:00 and 14:30; (ii) four pan-trap sets or solitary bees. Only the honeybee Apis mellifera (in our
of three colors each (fluorescent yellow, fluorescent blue sites they are feral) and all stingless bees (tribe Meliponini)
and white), modified from Nuttman et al. (2011) by placing were assigned to ‘social’ to group species with a behavior
a small receptacle in each bowl containing 15 ml of indus- that improves foraging efficiency (Nieh 2004); all other spe-
trial honey (inverted sugar and Carboxymethyl cellulose) as cies were categorized as solitary. Information on sociality,
bait. Bowls were then filled with a soapy water solution (1 nest location and construction method were mostly obtained
tsp TopTerra® soap per 3 l water). Sets of pan traps were from the primary literature and complemented with direct
placed every 50 m, elevated 3–7 m (except in pasture where observations in the field, particularly on nesting behavior.
bowls were set in the floor) and exposed on average 7 h For some morphospecies, information at the genus level
daily between 8:00 and 17:00; (iii) chemical baited traps for was not consistent across all species (e.g. nest location for
orchid bees (Tribe Euglossini) (Vélez and Pulido-Barrios Augochlora spp. or sociality for Chlerogella spp.), there-
2005); two traps were set in each transect, one with cineole fore these morphospecies were excluded from our functional
(4 drops) and the other with methyl salicylate (2 drops), diversity analysis.
located at the transect extremes. These baits are commonly
used for studying orchid bees (Brosi 2009a; Nemésio 2012)
and similar to Brosi (2009a), we used smaller quantities Functional diversity
than have been used in previous studies to avoid deleteri-
ous impacts on local orchid bees’ populations and to have We calculated functional diversity at each sampling plot
a smaller radius of attraction. Baited traps were exposed using the functional dispersion metric (FDis), as it accounts
for the same duration as pan traps. Collected bees were for different trait types, including qualitative and quantita-
pinned for reference collection and voucher specimens of tive traits. FDis estimates the mean distance of each species
each species will be deposited in the Alexander von Hum- from its community centroid in a multivariate space defined
boldt Institute (Colombia) according to the research permit, by all included traits (Laliberté and Legendre 2010). This
and the ICESI University (Colombia). Bees were identified metric can account for species abundances by shifting the
to the lowest possible taxonomic level using available keys position of the centroid toward the more abundant species,
and with the help of taxonomic experts. Still, identifica- and weighting distances of individual species by their rela-
tion to morphospecies (at least genus level) was necessary tive abundances (Laliberté and Legendre 2010). To calculate
because many bee species in the tropics are not yet described FDis, we used the function dbFD in package FD, with the
(Michener 2007). Cailliez correction for non-Euclidean distances generated by

13
 Author's personal copy

190 Journal of Insect Conservation (2020) 24:187–197

Fig. 1  Map (a) and schematic presentation (b) of the study area and location of the sampling sites in the studied land-cover types: forest (cir-
cles), riparian sections of corridors (triangles), restored sections of corridors (squares) and pasture (rhombus)

13
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197 191

inclusion of categorical traits. Traits were given equal weight of similarities (ANOSIM, Clarke and Gorley 2006) to test
and species were weighted by their relative abundance. for significance of differences and a sequential Bonferroni
adjustment of p values as a post hoc. The percentage con-
Local and landscape variables tribution of each taxon to the average dissimilarity between
habitats was assessed with a similarity percentage analysis
Both flower abundance (as a local habitat variable) and pro- (SIMPER). Differences in mean flower abundance and forest
portion of forest (as a landscape variable) have been found proportion (within 500 m radius) between forests, riparian
to influence bee diversity and composition in our study area and restored section of corridors, and pastures were tested
(Gutiérrez-Chacón et al. 2018). Therefore, we tested whether in two separate one-way analysis of variance (ANOVA) and
these two variables differ among the four land-use types (for- post hoc Tukey tests. All analyses were performed in R 3.3.2
est, riparian and restored section of corridors, and pasture) (http://www.r-proje​ct.org) except pairwise test (sequential
to determine whether differences in bee communities among Bonferroni adjustment of p values) which was performed in
land-use types could be explained by differences in the avail- the program PAST (Ryan et al. 1994).
ability of floral resources or by the amount of natural habi-
tats surrounding the sampling sites. We estimated flower
abundance in each site counting individual flowers along Results
150 × 2 m transects. This was carried out once per survey
and always by the same person. For flowers occurring in We collected 658 bees of four families, 27 genera and 41
inflorescences in the family Melastomataceae (e.g. Miconia (morpho) species. Aerial netting and pan traps collected 24
spp. and Graffenrieda sp.), we counted individual functional and 23 species, respectively; 14 species were collected by
(open) flowers because we noticed differences in the number the two methods while nine species were exclusively col-
of functional flowers among inflorescences between sites lected by each method. Eight additional bee species from
and surveys, likely due to the time of the flowering period genus Euglossa were found in chemical baited traps. A total
(beginning/ending vs. peak). For this, and given the large of 32 species were recorded in forest, 24 in corridors (22
amount of inflorescences in some surveys, we averaged the in riparian sections and 14 in restored sections), and eight
number of flowers from six inflorescences and multiplied species in pasture (Table 1).
it by the number of total observed inflorescences (to obtain Mean species richness was significantly different between
total functional flowers). Similarly, we counted individual forest, pasture, riparian, and restored sections of corridors
capitula in Asteraceae, even for those species in which they (F = 23.32, df = 3, p < 0.01). According to the post hoc Tukey
occur in panicle (e.g. Verbesina nudipens). Availability of tests, riparian sections of corridors showed a significantly
natural habitat was estimated as the proportion of forest higher species richness than restored sections and pastures,
within a buffer of a 500 m radius. Land-use information but species richness did not differ between restored sections
was obtained from http://www.sigqu​indio​.com by webmap and pastures (p > 0.05). Bee species richness was higher
service and digitalized with the software ArcGIS® 10.2.1. in forests compared to the other habitat types (p > 0.05)
(Fig. 2a). Similarly, bee community composition was signifi-
Statistical analyses cantly different among habitats (R = 0.41, p < 0.01, Fig. 2b),
but differences were observed between forest and restored
All bees captured by the three methods across the two sur- sections of corridors, forest and pasture, and pasture and
veys were pooled to provide a count of species richness, riparian sections (p < 0.05). Bee species composition was
community composition and FDis per site. We assessed not significantly different between the restored and the ripar-
the bee sample completeness in each habitat to make a fair ian sections (p > 0.05). The stingless bee species Trigona
comparison through the concept of sample coverage (Chao amalthea Olivier and Scaptotrigona barrocoloradensis
and Jost 2012), estimated in the R-package iNEXT (Hsieh Schwarz where highly abundant in forest, driving to a great
et al. 2016). As sample coverage in each land-use type was extent the difference with restored areas in corridors and
over 96%, we used observed rather than extrapolated bee pastures. While the stingless bee Partamona peckolti Friese
species richness. To assess the recovery of bees in restored was highly abundant in restored sections of corridors, Lasio-
sections of corridors and how they differ with the other land- glossum sp4 was dominant in pastures (Online Resource 1,
use types, we compared mean species richness and FDis SIMPER results).
between habitats with a one-way ANOVA. The change in Mean values of FDis were not significantly different
bee community composition between habitats was ana- among land-use types (F = 1.156, df = 3, p = 0.34), even per-
lyzed using Kruskal Non-metric Multidimensional Scal- forming the analysis excluding the Euglossa species given
ing (NMDS), calculated as Bray–Curtis distances and with the uncertainty in their sociality traits (F = 1.555, df = 3,
three-axis solution (k = 3), followed by a one-way analysis p = 0.25). Although FDis in restored and pasture were lower

13
 Author's personal copy

192 Journal of Insect Conservation (2020) 24:187–197

Table 1  List of bee species recorded in the study area, with infor- construction, which could be ‘rent’ for species that nest in existing
mation on functional traits and presence in the four habitat types holes or cavities, ‘construct’ for species that build the nest exterior
evaluated: forest patches being connected by corridors (forest), using collected materials, and ‘excavate’ for species that drill the tun-
riparian sections of corridors (riparian), restored sections of corri- nel/cavity; and sociality, with eusocial species (Apis mellifera and
dors (restored) and surrounding pastures (pasture). Functional traits stingless bees (tribe Meliponini)) categorized as ‘social’ and all other
included inter-tegular distance (ITD); nest location, either above- species as solitary
ground (above) or below-ground (below) nesting bees; method of nest
Bee species ITD Sociality Nest location Nest construction Forest Riparian Restored Pasture

Apidae
 Apis mellifera 2.68 Social Above Rent x x x x
 Centris similis 5.81 Solitary Above Excavate x
 Euglossa cf. variabilis 3.45 NIa Above Rent x
 Euglossa crassipunctata 3.52 NIa Above Rent x x
 Euglossa dissimula 3.30 NIa Above Rent x
 Euglossa maculilabris 3.53 NIa Above Rent x
 Euglossa modestior 3.53 NIa Above Rent x
 Euglossa trinotata 3.32 NIa Above Rent x x
 Eulaema cf. meriana 4.96 Solitary Above Rent x
 Eulaema nigrita 5.40 Solitary Above Rent x
 Exomalopsis sp1 1.81 Solitary Below Excavate x x
 Melipona nigrescens 3.04 Social Above Rent x x x
 Nannotrigona cf. perilampoides 1.27 Social Above Rent x
 Parapartamona caliensis 1.38 Social Above Excavate x x x x
 Paratetrapedia sp1 1.99 Solitary Below Excavate x x
 Paratetrapedia sp2 1.71 Solitary Below Excavate x
 Paratrigona rinconi 1.26 Social Above Construct x x
 Partamona peckolti 1.56 Social Above Construct x x x x
 Plebeia sp1 1.24 Social Above Rent x x x
 Scaptotrigona barrocoloradensis 1.68 Social Above Rent x x x x
 Thygater aethiops 3.19 Solitary Below Excavate x
 Trigona amalthea 2.01 Social Above Construct x x x x
 Trigona fulviventris 1.39 Social Above Rent x
 Xylocopa lachnea 6.51 Solitary Above Excavate x
Colletidae
 Hylaeus sp1 1.17 Solitary Above Rent x
Halictidae
 Augochlora sp1 1.39 Solitary NI NI x x x
 Augochloropsis sp1 1.86 Solitary Below Excavate x
 Caenaugochlora sp1 1.24 Solitary Below Excavate x
 Caenohalictus sp1 1.11 Solitary Below Excavate x
 Chlerogella sp1 1.49 Solitary NI NI x
 Habralictus sp1 1.12 Solitary Below Excavate x
 Habralictus sp2 0.97 Solitary Below Excavate x
 Lasioglossum sp1 1.65 Solitary Below Excavate x x
 Lasioglossum sp2 1.37 Solitary Below Excavate x
 Lasioglossum sp4 0.99 Solitary Below Excavate x x x x
 Neocorynura sp1 1.34 Solitary Below Excavate x x x
 Neocorynura sp5 1.80 Solitary Below Excavate x
 Neocorynura sp6 0.95 Solitary Below Excavate x x
 Neocorynura sp8 1.57 Solitary Below Excavate x x x
 Pseudaugochlora graminea 2.28 Solitary Below Excavate x
Megachilidae
 Megachile sp1 3.28 Solitary Above Rent x

NI no available information
a
 Likely solitary. Thus, we first run the analyses with the Euglossa species as solitary, and then excluding these species

13
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197 193

Fig. 2  Comparison of bee species richness (a), community composi- with median marked as a horizontal line. NMDS with a three axis
tion (b) and functional dispersion (FDis) (c) between forest, corridors solution (k = 3) and resulting stress of 0.09
(riparian and restored sections) and pasture. Box indicates quartiles

than forest and riparian sections of corridors, there was a species are using well preserved sections of corridors, while
high variability in FDis values in each habitat type (Fig. 2c). pastures seems to be hostile habitats for such species. Func-
Looking at the vegetation, mean flower abundance did not tional dispersion, however, did not indicate significant differ-
differ between forest, riparian and restored sections of corri- ences between habitat types. This suggest that communities
dors and pasture (F = 0.948, p = 0.45). However, differences with different species composition may have similar func-
were found to be significant in flower composition between tional diversity, and higher species richness is not neces-
the habitat types (ANOSIM R = 0.40, p ≤ 0.01), specifically sarily coupled with higher functional dispersion (Cadotte
between pasture and the other habitat types (Online Resource et al. 2011).
2). Similarly, the proportion of forest within a 500 m radius Differences in species richness between restored and
around the study sites were also different between habitat riparian sections of corridors were not driven primarily
types (F = 4.94, p = 0.02), being significantly higher around by differences in flowering-plant communities, as floral
forest patches (Mean ± SD = 0.44 ± 0.12) than around pas- resources availability (flower abundance and flower com-
tures (Mean ± SD = 0.24 ± 0.09) (p < 0.05). The proportion position) did not differ between these habitat types. Like-
of surrounding forest was not different between restored and wise, proportion of surrounding forest was similar between
riparian sections of corridors. riparian and restored sections. Thus, differences in bee
community richness likely arose from physical differences
between restored and riparian sites. Canopy cover was
Discussion lower in restored sections than in riparian sections, where
mature trees were more often observed. This point to other
Actively restored sections of corridors supported commu- resources such as nesting sites as limiting factors in restored
nities of wild bees with lower species richness than those areas. Many bee species depend on a wide variety of nest-
found in the riparian sections of corridors, and similar to ing resources such as trees, dead wood and twigs, which
those in pastures. Thus, it appears, on the one hand, that are more likely found in more advanced successional stages
restoration time of 13 years is not long enough to harbor sev- of vegetation. Therefore, the recovery of bee communities
eral forest-associated species. On the other hand, a lack of in the restored areas also depends on the recovery of such
protection of some restored areas from cattle intrusion may resources. In our study, recovery time of the restored sec-
also be delaying the colonization of these areas by plants tions may not be sufficient to successfully recuperate sev-
and bees. However, the similarity in species composition eral bee species and the resources they require. Contrast-
between restored and riparian sections suggest progress ing with our results, Williams et al. (2011) found that after
towards the recovery of some species found in the reference 6 years of restoration of riparian vegetation in California,
systems (i.e. riparian). Likewise, although species richness bee species richness was similar between restored and ref-
was higher in forest than in both sections of corridors, the erence sites, although community composition was differ-
comparable community composition between forest and ent. Although recovery time of bee communities might be
riparian sections, and the significant differences between context-dependent, more research is needed to better under-
forest and pastures, indicate that several forest-associated stand bee responses to habitat restoration. Information from

13
 Author's personal copy

194 Journal of Insect Conservation (2020) 24:187–197

other insect taxa show that recovery of reference communi- and restored sections of corridors suggest that more time
ties in restored areas can be a slow process. For instance, is needed for the restored sites to develop suitable nesting
Audino et al. (2014) found that after 18 years of restoration sites for stingless bees, specifically large trees, which can
of tropical low-land rainforest (from pastures), dung beetle more likely offer proper nesting conditions for the diverse
(Coleoptera: Scarabaeinae) communities were still depau- requirements of such bees. Yet, the similar bee community
perated showing no progress from the starting point. Like- composition between riparian and restored sections of cor-
wise, the only published study conducted in the biological ridors reveals some progress towards forest-like conditions
corridors here evaluated also revealed that restoration time of restored sites. This is important particularly for stingless
has not been enough to recover hunting ant communities bees, since new nests are usually established only after mul-
in restored sections of corridors, as higher species richness tiple trips from the old nest to provide it with stored food
were found in riparian forest (our riparian sections of corri- (Roubik 1992), so genetic flux between isolated and distant
dors) after 5–7 years of restoration process (Herrera-Rangel habitat patches would be restricted by the low flying capac-
et al. 2015). ity of most stingless bees given their small body size (Araújo
Forest patches showed the highest bee species richness et al. 2004). Therefore, the establishment of stingless bees´
compared to corridors and pastures. This pattern is likely the colonies along biological corridors is a positive signal on
result of higher amounts of forest cover in the surrounding the potential role of corridors as landscape linkages for oth-
landscape of forest sampling sites, from the patch itself but erwise isolated populations.
also from other forest areas. Previous research has demon- Despite differences in species richness between forest,
strated that bee species richness is enhanced in landscapes corridors (riparian and restored sections) and pasture, these
with a higher proportion of natural habitats (Brosi 2009b; differences did not translate into differences in functional
Park et al. 2015; Gutiérrez-Chacón et al. 2018; Steffan- dispersion. Our results parallel previous findings in which
Dewenter 2002; Viana et al. 2012). Natural areas are impor- species of pollinators were higher in less intensively man-
tant for the survival of bees in human-dominated landscape aged habitat types, but no differences in functional disper-
as they offer permanent food and nesting resources (Gari- sion were observed (Forrest et al. 2015; Rader et al. 2014).
baldi et al. 2011). Particularly in the Neotropics, the diver- Although positive linear relationships between species rich-
sity of an important group of pollinators of wild and culti- ness and FD have been observed, that trend lacks of univer-
vated plants, the stingless bees (Tribe Meliponini, Roubik sal support (Cadotte et al. 2011). That functional diversity
1992, 1995), has been strongly and positively related to the remain unchanged despite the loss of species richness may
availability of forest areas (Brosi 2009b; Gutiérrez-Chacón indicate the coexistence of functionally redundant species
et al. 2018). Most of the stingless bee species required trees in the community (Cadotte et al. 2011; Petchey and Gas-
for building their nests, which are located either in limbs, ton 2006). As high functional redundancy buffers ecosys-
cavities or between the roots (Roubik 1992). In addition to tem functioning against environmental changes (Laliberté
higher availability of nesting resources in forest, these can and Legendre 2010; Tscharntke et al. 2012), forest (with
offer diverse and abundant floral resources meeting high highest species richness) are less likely to decline in the
food demands of large colonies, as all stingless bees are pollination function if some species are lost. Likewise, as
eusocial (Michener 2007). corridor sections showed intermediate levels of taxonomical
Differences in community composition between pastures diversity between forest and pastures, their functional redun-
and forests may reflect differences in flower composition dancy would be higher than in pastures, which highlight
but also support the idea that the scarcity of above-ground their potential contribution to the maintenance of pollination
nesting sites in intensive-production areas act as a strong services in fragmented landscapes.
environmental filter on bee communities (Forrest et  al. The ability to detect functional differences between spe-
2015; Williams et al. 2010). Lasioglossum sp4 was domi- cies increases with the number of functional traits included
nant in pastures while almost absent from forest samples, in the calculation of functional diversity (Cadotte et  al.
which can be explain by the need of species in this genus 2011). As we only included four traits, the likelihood of clas-
for bare soil to nest, making them common inhabitants of sifying species as functionally redundant could be high. Yet,
productive lands (Basu et al. 2016; Ngo et al. 2013). On the there is no a ‘correct’ number of traits and rather all traits
contrary, the stingless bee species T. amalthea and S. bar- functionally important should be account for (Petchey and
rocoloradensis, which require trees for nesting, dominated Gaston 2006). Although life-history and behavioral traits
forest samples. As those species also dominated riparian here included may be relevant to bee conservation and their
sections of corridors, no differences in species composition role as pollinators (Williams et al. 2010), their number was
were observed with forest, indicating that well conserved limited by the ease to measure them (e.g. body size) and the
riparian forest-strips are suitable habitats for forest-associ- available information in literature, which are a general limi-
ated bees. However, significant differences between forest tations for trait-based approaches (Lichtenberg et al. 2017).

13
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197 195

Moreover, some individuals not identified to species but Compliance with ethical standards 
genus level were discarded from the analyses, as informa-
tion on nest location or sociality was not consistent across Conflict of interest  The authors declare that they have no conflict of
all species in the genus (e.g. Augochlora and Chlerogella). interest.
Thus, a better understanding on how functional diversity Informed consent  Informed consent was obtained from all individual
varies with environmental conditions will require improving participants included in the study.
of our taxonomic knowledge, particularly of tropical bees
Research involving human participants and/or animal  This article does
(Gonzalez et al. 2013), and the characterization of traits that
not contain any studies with human participants or animals performed
most importantly influence fitness and the ecosystem func- by any of the authors.
tions (Carmona et al. 2016), in this case pollination.
In conclusion, restored sections of corridors appear to be
slowly progressing towards the reference riparian sections,
although differences in species richness are still signifi-
cant. Compared to large forest patches, biological corridors References
immersed in pasturelands appear to have limited carrying
Araújo ED, Costa M, Chaud-Netto J, Fowler HG (2004) Body size and
capacity, likely due to less available habitat area. However, flight distance in stingless bees (Hymenoptera: Meliponini): infer-
our results suggest their potential value as habitat for sev- ence of flight range and possible ecological implications. Braz J
eral forest-associated species that otherwise could not find Biol 64:563–568
Audino LD, Louzada J, Comita L (2014) Dung beetles as indicators of
enough resources in open pastures. This is the case of some
tropical forest restoration success: is it possible to recover species
stingless bee species, whose prevalence in forests and ripar- and functional diversity? Biol Conserv 169:248–257
ian sections of corridors drove the difference in terms of Basu P, Parui AK, Chatterjee S, Dutta A, Chakraborty P, Roberts S,
community composition with open pastures. As the restora- Smith B (2016) Scale dependent drivers of wild bee diversity
in tropical heterogeneous agricultural landscapes. Ecol Evol
tion process advance, restored sites can improve their offer
6:6983–6992
of nesting sites, in particular, consolidating the role of cor- Beier P, Noss RF (1998) Do habitat corridors provide connectivity?
ridors as linkages in the landscape for bee populations. Our Conserv Biol 12:1241–1252
results also show that functional diversity was decoupled Bennett AF (1999) Linkages in the landscape: the role of corridors and
connectivity in wildlife conservation. IUCN Publications Services
from species richness and it was rather similar among habitat
Unit, UK
types, although the inclusion of more functional-traits and Bommarco R, Biesmeijer JC, Meyer B, Potts SG, Pöyry J, Roberts SP,
a better taxonomical information in the calculation of func- Steffan-Dewenter I, Öckinger E (2010) Dispersal capacity and diet
tional diversity could influence this result. Still, our findings breadth modify the response of wild bees to habitat loss. Proc R
Soc B 277:2075–2082
point to the higher functional redundancy of forest and cor-
Brosi BJ (2009a) The effects of forest fragmentation on euglossine bee
ridors, and underscore the importance of using complemen- communities (Hymenoptera: Apidae: Euglossini). Biol Conserv
tary biodiversity metrics as one dimension of biodiversity 142:414–423
was not a good surrogate for another (Audino et al. 2014; Brosi BJ (2009b) The complex responses of social stingless bees
(Apidae: Meliponini) to tropical deforestation. For Ecol Manag
Cisneros et al. 2015).
258:1830–1837
Brosi BJ, Daily GC, Shih TM, Oviedo F, Durán G (2008) The effects of
Acknowledgements  We thank J.C. Mejía, M.B. Duque, F. Builes, J. forest fragmentation on bee communities in tropical countryside.
Builes, H.F. Gómez, H. Castro, S. Giraldo, and landowners in Filandia J Appl Ecol 45:773–783
for research permission and valuable help with fieldwork. We further Cadotte MW, Carscadden K, Mirotchnick N (2011) Beyond species:
thank our field assistants, M. Rodriguez, J. Home, L.F. Estrada and F. functional diversity and the maintenance of ecological processes
Gamboa, V.H González and R. Ospina for their help with bee iden- and services. J Appl Ecol 48:1079–1087
tification, and C. Ríos for GIS help. We would like to thank CRQ Cane JH (1987) Estimation of bee size using intertegular span
(Corporación Autónoma Regional del Quindío) for the research permit (Apoidea). J Kansas Entomol Soc 60:145–147
No. 576-2014 and Universidad del Valle for research facilities; C.G.C Carmona CP, de Bello F, Mason NW, Lepš J (2016) Traits without
was funded by COLCIENCIAS (Colombian Administrative Depart- borders: integrating functional diversity across scales. Trends Ecol
ment of Science, Technology and Innovation). This study was also sup- Evol 31:382–394
ported by the University of Freiburg, Universidad Icesi, the Colombian Chao A, Jost L (2012) Coverage-based rarefaction and extrapolation:
Biodiversity Research Institute Alexander von Humboldt and Wildlife standardizing samples by completeness rather than size. Ecology
Conservation Society (WCS). 93:2533–2547
Cisneros LM, Fagan ME, Willig MR (2015) Effects of human-modified
Funding  Corresponding author Catalina Gutiérrez-Chacón received landscapes on taxonomic, functional and phylogenetic dimensions
financial support from the Colombian Administrative Department of of bat biodiversity. Divers Distrib 21:523–533
Science, Technology and Innovation (COLCIENCIAS), through a Clarke KR, Gorley RN (2006) User manual/tutorial. Primer-E Ltd.,
doctorate scholarship (No. 049). Field work for this study was sup- Plymouth, p 93
ported by the Colombian Biodiversity Research Institute Alexander Cracco M, Guerrero E (2004) Aplicación del Enfoque Ecosistemático a
von Humboldt, Universidad Icesi and Wildlife Conservation Society la gestión de corredores en América del Sur. Memorias del Taller
(WCS)—Colombia Program. Regional, 3 al 5 de junio. UICN, Quito, Ecuador

13
 Author's personal copy

196 Journal of Insect Conservation (2020) 24:187–197

Cranmer L, McCollin D, Ollerton J (2012) Landscape structure influ- Jauker B, Krauss J, Jauker F, Steffan-Dewenter I (2013) Linking life
ences pollinator movements and directly affects plant reproductive history traits to pollinator loss in fragmented calcareous grass-
success. Oikos 121:562–568 lands. Landsc Ecol 28:107–120
Crooks KR, Sanjayan M (2006) Connectivity conservation. Cambridge Kaluza BF, Wallace HM, Heard TA, Minden V, Klein A, Leonhardt
University Press, Cambridge SD (2018) Social bees are fitter in more biodiverse environments.
Cuatrecasas J (1958) Aspectos de la vegetación natural en Colombia. Sci Rep 8:12353
Revista Acadademia Colombiana Ciencias Exactas, Fisicas y Kattan GH, Franco P, Rojas V, Morales G (2004) Biological diversi-
Naturales 10:221–268 fication in a complex region: a spatial analysis of faunistic diver-
De Arruda Almeida B, Green AJ, Sebastián-González E, dos Anjos L sity and biogeography of the Andes of Colombia. J Biogeogr
(2018) Comparing species richness, functional diversity and func- 31:1829–1839
tional composition of waterbird communities along environmental Kremen C, Chaplin-Kramer R (2007) Insects as providers of ecosystem
gradients in the Neotropics. PLoS ONE 13:e0200959 services: crop pollination and pest control, in: Insect Conservation
Dı́az S, Cabido M (2001) Vive la difference: plant functional diversity Biology. In: Proceedings of the Royal Entomological Society’s
matters to ecosystem processes. Trends Ecol Evol 16:646–655 23rd Symposium. pp 349–382
Etter A, McAlpine C, Pullar D, Possingham H (2006) Modelling the Laliberté E, Legendre P (2010) A distance-based framework for
conversion of Colombian lowland ecosystems since 1940: drivers, measuring functional diversity from multiple traits. Ecology
patterns and rates. J Environ Manag 79:74–87 91:299–305
Fagua JC, Cabrera E, Gonzalez VH (2013) The effect of highly variable Lichtenberg EM, Mendenhall CD, Brosi B (2017) Foraging traits
topography on the spatial distribution of Aniba perutilis (Lau- modulate stingless bee community disassembly under forest loss.
raceae) in the Colombian Andes. Revista de Biología Tropical J Anim Ecol 86:1404–1416
61:301–309 Losey JE, Vaughan M (2006) The economic value of ecological ser-
Forrest JR, Thorp RW, Kremen C, Williams NM (2015) Contrasting vices provided by insects. Bioscience 56:311–323
patterns in species and functional-trait diversity of bees in an agri- Lozano-Zambrano FH (2009) Herramientas de manejo para la conser-
cultural landscape. J Appl Ecol 52:706–715 vación de biodiversidad en paisajes rurales. Instituto de Investi-
Garibaldi LA, Steffan-Dewenter I, Kremen C, Morales JM, Bommarco gación de Recursos Biológicos Alexander von Humboldt y Cor-
R, Cunningham SA, Carvalheiro LG, Chacoff NP, Dudenhöffer poración Autónoma Regional de Cundinamarca (CAR)
JH, Greenleaf SS et al (2011) Stability of pollination services Magurran AE (1988) Ecological diversity and its measurement. Prince-
decreases with isolation from natural areas despite honey bee vis- ton University Press, Princeton
its. Ecol Lett 14:1062–1072 Mendoza JE, Lozano FH (2006) Composición y estructura de la bio-
Gonzalez VH, Griswold T, Engel MS (2013) Obtaining a better tax- diversidad en paisajes transformados en Colombia (1998-2005).
onomic understanding of native bees: where do we start? Syst In: Chaves ME, Santamaría M (eds) Informe sobre el avance en
Entomol 38:645–653 el conocimiento y la formación de la biodiversidad 1998-2004.
Gotelli NJ, Colwell RK (2001) Quantifying biodiversity: procedures Bogotá, Instituto de Investigación de Recursos Biológicos Alex-
and pitfalls in the measurement and comparison of species rich- ander von Humboldt, pp 67–84
ness. Ecol Lett 4:379–391 Michener CD (2007) The bees of the world. Johns Hopkins University
Gotelli NJ, Colwell RK (2010) Estimating species richness Biologi- Press, Baltimore, p 953
cal diversity: frontiers in measurement and assessment. Oxford Nemésio A (2012) Methodological concerns and challenges in ecologi-
University Press, Oxford cal studies with orchid bees (Hymenoptera: Apidae: Euglossina).
Greenleaf SS, Williams NM, Winfree R, Kremen C (2007) Bee Biosci J 28(1):118–135
foraging ranges and their relationship to body size. Oecologia Ngo HT, Gibbs J, Griswold T, Packer L (2013) Evaluating bee (Hyme-
153:589–596 noptera: Apoidea) diversity using Malaise traps in coffee land-
Gutiérrez-Chacón C, Franco P, Solano C (2013) Iniciativas de conec- scapes of Costa Rica. Can Entomol 145:435–453
tividad en Colombia, in: In: Hurtado A., Santamaría M. y Matal- Nieh J (2004) Recruitment communication in stingless bees (Hyme-
lana-Tobón C.L. Plan de Investigación y Monitoreo Del Sistema noptera, Apidae, Meliponini). Apidologie 35:159–182
Nacional de Áreas Protegidas (Sinap): Avances Construidos Nuttman CV, Otieno M, Kwapong PK, Combey R, Willmer P, Potts SG
Desde La Mesa de Investigación y Monitoreo Entre 2009 y 2012. (2011) The utility of aerial pan-trapping for assessing insect pol-
Instituto de Investigación de Recursos Biológicos Alexander von linators across vertical strata. J Kansas Entomol Soc 84:260–270
Humboldt y Parques Nacionales Naturales, Bogotá D.C, Colom- Öckinger E, Smith HG (2008) Do corridors promote dispersal in grass-
bia, p 200 land butterflies and other insects? Landsc Ecol 23:27–40
Gutiérrez-Chacón C, Dormann CF, Klein A-M (2018) Forest-edge Ollerton J, Winfree R, Tarrant S (2011) How many flowering plants
associated bees benefit from the proportion of tropical forest are pollinated by animals? Oikos 120:321–326
regardless of its edge length. Biol Conserv 220:149–160 Park MG, Blitzer EJ, Gibbs J, Losey JE, Danforth BN (2015) Negative
Herrera-Rangel J, Jiménez-Carmona E, Armbrecht I (2015) Monitor- effects of pesticides on wild bee communities can be buffered by
ing the diversity of hunting ants (Hymenoptera: Formicidae) on landscape context. Proc R Soc B 282(1809):20150299
a fragmented and restored andean landscape. Environ Entomol Petchey OL, Gaston KJ (2006) Functional diversity: back to basics and
44:1287–1298 looking forward. Ecol Lett 9:741–758
Hill CJ (1995) Linear strips of rain forest vegetation as potential dis- Rader R, Bartomeus I, Tylianakis JM, Laliberté E (2014) The win-
persal corridors for rain forest insects. Conserv Biol 9:1559–1566 ners and losers of land use intensification: pollinator community
Hilty JA, Lidicker WZ Jr, Merenlender A (2012) Corridor ecology: disassembly is non-random and alters functional diversity. Divers
the science and practice of linking landscapes for biodiversity Distrib 20:908–917
conservation. Island Press, Washington DC Roubik DW (1992) Ecology and natural history of tropical bees. Cam-
Hsieh TC, Ma KH, Chao A (2016) iNEXT: iNterpolation and EXTrap- bridge University Press, Cambridge
olation for species diversity. R package version 2.0.12. https​:// Roubik DW (1995) Pollination of cultivated plants in the tropics. Food
chao.shiny​apps.io/iNEXT​Onlin​e/. Accessed 1 Dec 2019 & Agriculture Org
Hunter MD (2002) Landscape structure, habitat fragmentation, and the Rudnick D, Ryan SJ, Beier P, Cushman SA, Dieffenbach F, Epps C,
ecology of insects. Agric For Entomol 4:159–166 Gerber LR, Hartter J, Jenness JS, Kintsch J et al (2012) The role

13
 Author's personal copy
Journal of Insect Conservation (2020) 24:187–197 197

of landscape connectivity in planning and implementing conserva- Williams NM (2011) Restoration of nontarget species: bee commu-
tion and restoration priorities. Issues in Ecology nities and pollination function in riparian forests. Restor Ecol
Ryan R, Harper DA, Whalley JS (1994) PALSTAT: user’s manual and 19:450–459
case histories: statistics for palaeontologists and palaeobiologists. Williams NM, Crone EE, Minckley RL, Packer L, Potts SG et al (2010)
Springer, New York Ecological and life-history traits predict bee species responses to
Steffan-Dewenter I (2002) Landscape context affects trap-nesting bees, environmental disturbances. Biol Conserv 143:2280–2291
wasps, and their natural enemies. Ecol Entomol 27:631–637 Winfree R, Griswold T, Kremen C (2007) Effect of human distur-
Tscharntke T, Tylianakis JM, Rand TA, Didham RK, Fahrig L, Batary bance on bee communities in a forested ecosystem. Conserv Biol
P, Bengtsson J, Clough Y, Crist TO, Dormann CF et al (2012) 21:213–223
Landscape moderation of biodiversity patterns and processes- Winsa M, Öckinger E, Bommarco R, Lindborg R, Roberts SP, Wärns-
eight hypotheses. Biol Rev 87:661–685 berg J, Bartomeus I (2017) Sustained functional composition of
Vargas WG (2008) Evaluación de dos estrategias de restauración, su pollinators in restored pastures despite slow functional restoration
aplicación y evaluación en el establecimiento del corredor Bar- of plants. Ecol Evol 7:3836–3846
bas-Bremen, Quindío. Universidad del valle, Programa biología, Wortley L, Hero J-M, Howes M (2013) Evaluating ecological restora-
Maestría en ciencias. Tesis, Santiago de Cali tion success: a review of the literature. Restor Ecol 21:537–543
Vélez D, Pulido-Barrios H (2005) Observations on the vertical strati-
fication of orchid bees (Apidae: Euglossini) in a riparian forest of Publisher’s Note Springer Nature remains neutral with regard to
the Colombian Orinoquia. Caldasia 2:267–270 jurisdictional claims in published maps and institutional affiliations.
Viana BF, Boscolo D, Neto EM, Lopes LE, Lopes AV, Ferreira PA,
Pigozzo C, Primo L (2012) How well do we understand landscape
effects on pollinators and pollination services? J Pollinat Ecol
7:31–41
Villéger S, Mason NW, Mouillot D (2008) New multidimensional func-
tional diversity indices for a multifaceted framework in functional
ecology. Ecology 89:2290–2301

13