Global Ecology and Conservation 17 (2019) e00552Contents lists available at ScienceDirectGlobal Ecology and Conservation
journal homepage: http: / /www.elsevier .com/locate/geccoOriginal Research ArticleGathering honey from wild and traditional hives in the
Miombo woodlands of the Niassa National Reserve,
Mozambique: What are the impacts on tree populations?
Natasha Sofia Ribeiro a, *, Laura K. Snook c, Iva Carla Nunes de Carvalho Vaz a, b,
Teresa Alves a
a Faculdade de Agronomia e Engenharia Florestal, Universidade Eduardo Mondlane, Av, Julius Nyerere, 3453, Campus Principal da
Universidade Eduardo Mondlane, P.O, Box 257, Maputo, Mozambique
b Instituto de Investigaça~o Agraria de Moçambique, Av. Das FPLM, 2698, no Bairro de Mavalane Caixa postal 3658, Mozambique
c Bioversity International, Via dei Tre Denari 472/a, 00054, Maccarese, Rome, Italya r t i c l e i n f o
Article history:
Received 11 September 2018
Received in revised form 1 February 2019
Accepted 1 February 2019
Keywords:
Biodiversity conservation
Fire
Julbernardia globiflora
Miombo
Forest management* Corresponding author.
E-mail address: joluci2000@yahoo.com (N.S. Rib
https://doi.org/10.1016/j.gecco.2019.e00552
2351-9894/© 2019 The Authors. Published by Elsev
licenses/by-nc-nd/4.0/).a b s t r a c t
Honey gathering is one of the activities sustaining people's livelihoods in the Miombo
woodlands of southern Africa. Current practices involve killing trees and can cause un-
controlled fires, affecting villages, animals and trees. This study aimed to understand
honey gathering and production and its impacts on the Miombo woodland vegetation, in
Lizongole village, Niassa National Reserve. Data was obtained through semi-structured
interviews with 15 honey gatherers in one village and on 95 plots where all trees

10 cm, dbh (diameter at breast height), alive or dead, were identified at species level, and
their dbh (cm) and heights (m) measured. Forty-seven percent of the respondents pro-
duced and managed beehives and 100% gathered wild honey. Bark beehives are prefer-
entially made from Julbernardia globiflora (‘Ntchenga’) while wild honey is gathered from
natural cavities in a diversity of tree species, typically by felling the tree. Both techniques
are destructive and primarily target larger trees (average dbh of 26.1 ± 1.2 cm for debarked
trees and 30.4 ± 2.4 cm for trees felled). Impacts on tree populations varied among the 12
species killed for honey, depending largely on their relative abundances. For nine species,
felling and debarking were found to have reduced the number of larger trees by between
40% and 100% compared to previous densities. This is diminishing the nectar resource, the
number of bee colonies, and the number of trees suitable for wild hives. However, honey
can be gathered using non-destructive traditional practices based on tree climbing.
Widespread adoption of such practices is needed to sustain honey production.
© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC
BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).1. Introduction
The Miombo woodlands are the most extensive warm dry forest type in Southern Africa, covering ca. 2.7 million km2 and
supporting the livelihoods of more than 150 million people in the region (Frost, 1996; Clarke et al., 1996; Dewees et al., 2010).
The woodlands, stretching across Angola, Zambia, southeastern Democratic Republic of Congo (DRC), western and southerneiro).
ier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/
2 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552Tanzania, Zimbabwe, central and northern Mozambique and northern Malawi, provide ideal ecological conditions for hon-
eybees. The dominant Miombo tree species of the genera Julbernadia, Brachystegia and Isoberlinia provide abundant nectar,
while species of Parinari, Cryptosepalum, Guibourtia, Marquesia and Syzygium ensure availability of nectar between the
flowering seasons of the dominant trees (Alcobia, 1995; Campbell, 1996; Husselman, 2008). Honey and beeswax are the
principal bee products harvested from these woodlands, but others include pollen and brood comb.
The primary honey producer is Apis meillifera scutellata, the African variety of the honeybee raised in Europe, Asia and the
Americas. This subspecies is smaller, more easily provoked, and more prone to swarming behavior than its European
counterpart. It is also more likely to abscond (leave to establish a hive in a new location) when resources are scarce, which
may be an adaptation to frequent pollen and nectar scarcity caused by recurring droughts. Absconding also occurs in reaction
to excessive temperature, predation, fire, and rain entering the nest (Fletcher, 1978). Honey in the Miombo may also be
collected from a variety of stingless bees, Meliponini. Some of these create hives in tree cavities, but others create hives that
hang from tree branches, or in the soil.
The relevance of honey for the subsistence of people inhabiting the Miombo woodlands has been described across the
region. For instance, Syampungani et al. (2009) reported that in Zambia, honey hunting and beekeeping improve diets for an
estimated 250,000 farmers and are an important source of income for over 20,000 rural households. In Tanzania, annual
foreign earnings from both honey and bees wax was estimated at US$ 8 million (Mpuya, 2003). According to Illgner et al.
(1998) and Mudekwe (2017), honey gathering is part of the safety net that communities adopt to reduce vulnerability to
crop failure or to meet exceptional household or individual needs. Flowering times of the different nectar sources spread the
honey flow over much of the year (Lawton,1982). Ntenga andMugongo (1991) reported honey gathering seasons in Tanzania
as extending from April to June and November to January. There are two gathering seasons in Zambia: the first around
OctobereNovember at the start of the rains, and the second at the end of the rainy season (Fischer, 1993).
Traditional practices of honey collection often involve killing trees, either to access the honey in wild hives in natural
cavities or as a result of debarking trees to obtain materials for constructing beehives. These destructive harvesting practices
may compromise the sustainability of honey harvesting. Additionally, fires used to control the bees can spread and have
negative consequences for trees and villages (Ntenga and Mugongo, 1991; Snook et al., 2015, 2016). To better understand this
situation and support both sustainable honey production and appropriate conservation actions in the Niassa Reserve of
Mozambique, this study aimed to 1) determine current practices of honey gathering and production and 2) describe the
Miombo woodland vegetation and analyze how honey-related activities affect its tree populations, particularly those that are
fundamental to sustaining honey yields.
2. Materials and methods
2.1. Study site
The Niassa National Reserve (NNR), located primarily in Mozambique's northern province of Niassa, borders Tanzania
(Fig. 1). With an area of 42,000 km2, it includes the district of Mecula, a major part of Mavago and portions of the districts of
Muembe, Majune, Marrupa e Sanga as well as parts of the districts of Mueda and Montepuez, in the neighboring Province of
Cabo Delgado. The NNR is one of the most important conservation areas in Mozambique, accounting for 36% of the area of the
national conservation system (SGDRN, 2007). The NNR hosts one of the more pristine Miombo woodlands in the country
(Timberlake et al., 2004) as well as a human population most recently estimated at ca. 40,000 people (Cunliffe et al., 2009).
People in the NNR depend on forest resources to sustain their livelihoods, and honey is one of their main nutrition and income
sources (Cunliffe et al., 2009; Snook et al., 2015, 2016). Honey is used as a food or a medicine and a product to be sold, and
represents a safety net during lean times (Alberto Siabu, pers. com. to L. Snook, 2012; Mr. Silas, Pers. com. to N. Ribeiro et al.,
2013). Honey and beeswax are obtained either through gathering from the wild or beekeeping (apiculture), based on the use
of traditional hives made from tree bark (Snook et al., 2015, 2016.) Honey gatherers in Niassa describe five different types of
bee: some produce honey in cavities (both Apis mellifera scutella and the small, stingless sweat bee), some under branches,
some in the soil, and some in rocks (Alberto Siabu, pers. communication to L. Snook, 2012).
This study was carried out in the village of Lizongole, Administrative post of Lugenda, Mecula District, within the NNR.
Lizongole was selected as a case study given its location along the main road through the reserve and the fact that 100% of the
households are involved directly or indirectly in honey gathering and/ormarketing. Lizongole is located between the parallels
of 12 10 42.56 and 12⁰ 24 27.08}S and the meridians 37 29 36.42}and 37 35 26.7}E. It is limited to the north by the
Incalaue River, to the south by the village of Cuchiranga, to the west by the locality of Mecula-Sede and to the East by the
Lugenda River. The total area of the village is estimated at 9000 km2 (Fig. 1; INE, 2010).
The climate is tropical sub-humid, with a mean annual precipitation (MAP) of 800mm and a mean annual temperature
(MAT) of 25 C that ranges from 20 to 26 C during the dry season (MayeOctober). The wet season (NovembereApril) has a
MAP of 900mm and a MAT of 30 C. The topography is characterized by a gently undulating landscape on a plateau with
elevations around 300m above sea level (SGDRN, 2007).
According to White (1983), the ecosystem in this area is classified as “dry Zambezian Miombo woodland”with intrusions
of east African coastal elements along the Lugenda River (located ca. 20 km south of Lizongole). There is no specific
description of the Miombo woodlands in Lizongole, but references about the NNR describe floristically poor woodland,
dominated by Brachystegia spiciformis Benth (‘Mpapa’), Brachystegia boehmii Taub. (‘Ndjombo’) and Julbernardia globiflora
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 3
Fig. 1. Geographic Location of Lizongole Village in Mecula District, Niassa National Reserve, Niassa province, northern Mozambique.
4 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552(Benth) Troupin (‘Ntchenga’). Other ecologically important species in the area are Pseudolachnostylis maprouneifolia Pax (‘
Nsolo’), Pterocarpus angolensis DC (‘Mtumbati’) and Diplorhynchus condylocarpon (Müll. Arg.) Pichon (‘Ntomonha’). The
canopy is generally less than 15m in height and the trees are deciduous for a month or more during the dry season (Ribeiro
et al. 2008, 2013).
The total population in Lizongole is estimated at 457 inhabitants (227 men and 230 women), with a density of 19 people/
km2 (INE, 2010). The population in this village belongs to the Yao ethnic group, whose presence in this area dates back
centuries (Price, 1964; Thorold, 1987; Reader, 1998:178). Among the main characteristics of this group are their historical
trading route toTanzania and their relationship with forest resources and the use of fire as amanagement tool (SGDRN, 2007).
Shifting agriculture (maize and beans), fishing and honey gathering are their main activities, primarily for subsistence
(SGDRN, 2007; Cunliffe et al., 2009). However, given the location of the village on the main road, honey is also sold locally,
along the road.2.2. Data collection
2.2.1. Focus group and interviews
Insights from a pilot focus group discussion in Lizongole were used to develop a questionnaire for honey gatherers and
producers. The focus group included 21 men from the area who were involved in honey-related activities and discussed and
answered 28 questions about honey gathering and production. A translator from the area accompanied the research team,
translating questions from Portuguese to Yao and the answers from Yao to Portuguese. A complementary exercise with cards
and beans was used to obtain specific information about the most important plant species for bees and honey production.
This exercise revealed that J. globiflora and B. boehmii were preferred for producing beehives, but that wild honey was
gathered from a broad spectrum of tree species (Alves, 2011). These results were used to further refine the methods for field
data collection.
Semi-structured interviews (Pijneburg and Cavane, 2000) were conducted in 2011 with 15 honey collectors (100% of the
men in Lizongole who were involved in this activity). Seven out of the 15 honey gatherers were also dedicated to apiculture
using traditional beehives, and were also questioned about these practices. The interviewers used a guide previously pre-
pared and discussed with NNR management authorities and the research team involved, to guarantee that key information
was collected. Interviews were also performed with key informants: the leader of the village, the district director for eco-
nomic activities and the NNR managers. The questions gathered information about the families (number, activities, assets,
etc.), distance to honey collection areas, type and description of collection methods, species used/preferred for collection or
for bee hives, harvesting seasons, production quantities, use of fire and impacts on vegetation.
2.2.2. Ecological assessment
To evaluate the impact of honey gathering we used the method proposed by Filgueiras et al. (1994) and Kent (2012) -
Searching Walks. This involves searching for target trees by walking through the forest with a local guide. This observational
method has been recommended to search for randomly dispersed target trees, like the honey trees in Lizongole. The bush
walks were carried out for 15 days together with two honey collectors, within a radius of ca. 15 km from the village center (a
total sample area of 70,650 ha or 706.5 km2). During these walks, trees of which the bark had been removed for hives or those
with cavities that had been felled were identified and used as a central reference for a 500m 25m transect. Using this
procedure, we identified 19 locations around the village where trees had been affected by honey production, and established
sampling transects including them.Within each transect five plots of 25m 25m, were laid out 75m apart (Fig. 2). This gave
a total sample of 95 plots covering an area of 59,375m2 (5.9 ha or 0.59 km2), equivalent to a 0.008% sample of the area within
the 15 km radius around the village.
In each plot, we measured height (m) and diameter at breast height (dbh; cm) of all living trees 
10 cm dbh, the size at
which trees were well established but still well below the size threshold for use for honey gathering or bee hives. Diameters
were also measured, at the equivalent of 1.3m from the ground, for trees that had been felled or debarked for honeyFig. 2. Schematic representation of a transect for ecological characterization and assessment of the impacts of honey gathering in Lizongole village, Mecula
District, Niassa National Reserve.
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 5production. All trees (living and dead) were identified in the field using local names. Scientific names were identified in the
Herbarium of the Institute for Agricultural Research (IIAM) inMaputo. Dead treeswere identified using crown, bark andwood
characteristics in the field and if leaves, fruits or flowers were present, these were collected for identification in the her-
barium. If leaves, fruits or flowers were not present, we searched for a living tree of the same species to collect these materials
for future identification.
To characterize the damage associated with bark removal wemeasured: (a) the height of the debarked area (m) by using a
meter tape; and (b) thickness of removed bark (mm) by using a paquimeter (Pressler, precision 1mm). On felled trees, we
measured the length and width in cm of the opening of cavity hives (after expansion by honey gatherers), using a meter tape.2.3. Data analysis
Data collected through interviews was analysed by using the qualitativemethod proposed by Pijneburg and Cavane (2010)
in which similar responses were grouped and their frequency and percentage calculated. Ecological data was analysed using
the statistical packages SPSS V.17.0 (Statistical Package for Social Sciences), Excel and Systat 13 (Systat, 2009). Decriptive
statistics (averages and standard deviations) were calculated for hive and cavity sizes. Pearson Correlation analysis was used
to evaluate the relationship among tree dbh and height, cut height and bark thickness (a measure of beehive quality) of trees
that had been debarked for beehives, with a significance level of 5%.
Woodland structure and composition of live trees were analysed by using the Importance Value Index (IVI; Kent, 2012)
and the diametric distribution. IVI was determined using the formula: Importance Value Index ¼ Relative abundance (%
n) þ Relative dominance (% Basal Area) þ Relative Frequency (% of plots where the species was found). By aggregating those
three important ecological parameters, this index provides a good measure of ecosystem tree composition and is an indi-
cation of which species are ecologically more important in the area under local environmental conditions. The diametric
distribution indicates the structure of the ecosystem by documenting the stem frequency per diameter class. To address the
impact of honey gathering we calculated the number and percentage of trees killed per species and diameter class as a result
of beehive production and wild honey gathering.3. Results
3.1. Ways of obtaining honey
3.1.1. Honey gathering
In Lizongole, 15 families gather wild honey, a source of nutrition, medicine and income. Wild honey gathering is typically
carried out by groups of five to seven men, who walk around five (27% of interviewees) to 20 km (73% of interviewees) from
the village to findwild hives in cavities, remaining in the forest for five to 10 days, depending on the amount of honey needed.
The group only leaves the production area after each collector has filled a 20-L plastic jerry can, meaning the group has
collected about 100e140 L. The gathering distance depends on the purpose and needs, being shorter when the needs are
immediate (e.g. medicine for cold or skin burns) and longer when collectors need income.
Before the group moves into the forest, they jointly decide on the gathering areas and once there, they initiate a mutu-
alistic interactionwith the bird called zego or the honey guide (Indicator indicator Step.), which guides men to trees containing
honey in cavities (this interaction also exists between the zego and the honey badger -Mellivora capensis; Spottiswoode et al.,
2016). The interaction starts with a whistle by the honey gatherer, to which the bird responds with a similar sound and starts
to lead the honey gatherer towards a honey tree, landing on one tree after another and calling the gatherer each time by
singing loudly and frequently. The bird only stops singing once the honey gatherer reaches the honey tree.
After finding the tree, the men inspect it to identify the height of the cavity and to decide the best collection method (tree
felling or direct collection). In the next step, the honey gatherers scratch dried sticks of Oxytenanthera abyssinica (A. Rich)
(bamboo) or Cajanus cajan L. (pigeon pea) to initiate fire by using a flint. The burning sticks are then used to burn dried
branches of Diplorhynchus condylocarpon (called locally ntomonha). This smoking branch is then located close to the cavities
andwaved around the tree, to tranquilize the bees. According to the honey collectors, burning branches of ntomonha produces
a hallucinogenic effect upon the bees, reducing the likelihood that theywill sting the collectors. Then they throw the fire stick
away, which may initiate uncontrolled fires. All 15 of the honey gatherers indicated that they fell trees for honey collection,
using a machete (Fig. 3a). After felling the tree, the hunter starts opening the cavity with a machete to widen it to enable
removal of the honeycombs (Fig. 3b). After the honey has been removed, the zego returns to the felled tree to collect the
leftover wax, which it eats, completing the mutualistic interaction.
Twenty percent of the honey gatherers (three) indicated that each man fells about 30 trees, while 80% (twelve) said they
fell about 40 trees per expedition. Given the yearly number of expeditions described by each person (one to two), and the
maximum group size (seven) honey gatherers apparently fell up to 560 trees/year around Lizongole (40 trees x seven men x
two expeditions), or about 0.79 trees/km2/year on the 706.5 km2 within 15 km of Lizongole. Each tree yields approximately
5 L of honey, which may result in close to 200 L per expedition. According to the honey collectors, there is no preference in
6 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552
Fig. 3. Illustration of trees felled for honey gathering in Niassa National Reserve [a) trees felled for honey collection; b) cavity widen up for honeycomb access].terms of tree species fromwhichwild honeymay be gathered. However, they described J. gobliflora as a high valuemelliferous
tree; its nectar is considered to provide high quality honey compared to other species.
3.1.2. Beehive production
Traditional beehives are another method of obtaining honey. They are used by 47% of families (seven men in Lizongole, all
of whom also gather from wild hives). Beehives are made from the bark of three species (J. globiflora, B. boehmii and B.
spiciformis), of which J. globiflora is preferred. This is because its bark is easier to remove and transport (due to reduced
weight) and more durable (five to 10 years). The activity is generally conducted during the wet season due to the greater ease
of removing the bark (100% of the respondents; Mickes-Kokwe, 2006). The process starts with an inspection of the tree for
health conditions and straightness of the trunk. Then, the hive-maker removes the bark (using amachete) bymaking a vertical
incision and two horizontal cuts around the tree, one at the top and one at the bottom of the future hive, to enable bark
removal without damaging its cylindrical shape (Fig. 4a). This process girdles the tree, killing it by interrupting sap conduction
(Franklin et al., 1987). After removing the bark, the hive maker reconnects the vertical edges using bamboo sticks to close theFig. 4. Illustration of beehives production process in Niassa National Reserve [a) standing tree with cylindrical bark removed; b) beehive hung up in a tree].
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 7
Table 1
Average size of Julbernardia globiflora trees and extent of bark removal for beehive production in Lizongole, Mecula District, Niassa National Reserve.
Parameter Number of trees sampled Descriptive statistics
Average Standard deviation Minimum Maximum
Dbh (cm) 22 27.0 5.9 16.0 36.6
Hive length (cm) 1.9 0.8 0.5 1.4
Tree height (m) 10.4 2.6 5.0 14.0
Cut height (m) 1.9 0.7 0.8 3.1
Bark thickeness (mm) 11.7 7.6 0.4 33.3cylinder but maintain enough open space for aeration. Then the cylinder is left to dry for two weeks before being hung in a
tree (Fig. 4b). The seven beekeepers in Lizongole had an average of six beehives each (ranging from four to 31). This would
translate to 28 to 217 adult J. globiflora trees killed every five to 10 years by the beekeepers in the community, equivalent to a
maximum of 0.06 trees/km2/yr on the 706.5 km2 of area within 15 km of Lizongole. Debarked trees found on the transects
ranged from 16.0 to 36.6 cm dbh (see Table 1, Table 5).
All seven of the beekeepers place their hives close to water sources and melliferous plants that flower at different times of
the year (Table 2) to increase the likelihood of having enough honey production throughout the year. The hives are in general
located close to their houses or farms to allow inspection and control and are placed in tall trees of Brachystegia and Jul-
bernadia (about five to 7m high) to avoid interference by honey badgers (Mellivora capensis), the main predators of beehives
in NNR).
In the past, a WWF project introduced conventional beekeeping, but this was not successful, in part because the hives did
not produce the same honey quality as the traditional ones (oral communication from the interviews). The ultimate reason for
their failure, however, was that the hives were set on low planks, where they were destroyed by the fires that frequently
sweep through the bush (oral communication to L. Snook from interviews with Lizongole hive producers). Traditional hives,
set high in trees, are not so vulnerable.
Apiculturists reported that on average, in the wet season (production peak in December) each beehive produces from
three to 20 L, an average of 11.6 L (±7.7 L). In the dry season (production peak in May), they reported yields per hive ranging
from two to 8 L, averaging 5 L (±1.15 L), for a total of 16.6 L/hive/year, on average. The collection method is similar to wild
honey gathering, in that a smoke stick of D. condylocarpon is used to keep the bees from stinging. The beekeeper climbs the
tree to reach the beehive. Once the hive is reached, the apiculturist removes the combs from both open extremities to avoid
destroying the hive.
According to all community interviewees, gathering honey from the wild is preferred over managing bee hives, for several
reasons: (i) difficulty in collecting the bark and making the hives; (ii) comparatively low production from traditional hives;
(iii) the propensity of honey badgers to steal honey from and destroy hives; and (iv) the risk of beehive decolonization.Table 2
Melliferous plant species and blooming season reported by honey gatherers and bee keepers in Lizongole, Mecula District, Niassa National Reserve.
8 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552
Table 3
Importance Value Index of 25 species found on 95 plots totalling 5.9 ha around Lizongole, Mecula District, Niassa Reserve, based on: absolute and percent
abundance (number of trees of each species); dominance (Basal Area of each species, expressed inm2; and frequency (number of plots where the species was
documented). Rows listing species important as nectar sources, for wild hives, or for beehive production, are shaded.2
aOne tree was unidentified and was not included here.3.2. Woodland structure and composition
We found a total of 588 trees (522 living and 67 dead), of 25 species, on 95 plots. The Importance Values of live trees
revealed that J. globiflora (IVI¼ 93.4) and B. boehmii (IVI¼ 63.3) were the most important, representing 52% of the total IVI
(Table 3). The 10 most important tree species accounted for 88% of the total IVI and consisted of typical Miombo indicator
species. Thirteen species occurred five or fewer times on the 5.9 ha of plots.
The shape of the diameter distribution (Fig. 5), with a higher abundance of individuals in smaller size classes, could be
interpreted as revealing continuous regeneration and replacement of trees. The most important honey tree species, J. glo-
biflora, showed a similar pattern, but the lower abundance in the smallest class (10.0e14.9 cm dbh) as compared to the size
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 9
Table 4
Average cavity width (cm) after artificial opening and dbh (cm) per species felled for extraction of honey in Lizongole.
Species Name Local name Number of trees Average cavity width (cm) (STD) Average tree dbh (cm) (STD)
Albizia versicolor Ncalate 3 12 (5) 39.2 (7.4)
Brachystegia boehmii Ndjombo 3 13 (7) 37.9 (33.1)
Brachystegia manga Nanguezo 1 19 (0) 36.3 (0)
Brachystegia spiciformis Mpapa 3 56 (9) 55.8 (30.9)
Burkea africana Nkalati 7 9 (4) 24.3 (4.2)
Julbernadia globiflora Nchenga 10 50 (22) 26.6 (7.3)
Pericopsis angolensis Mbanga 1 11 (0) 37.0 (0)
Pseudalachnostylis maprouneifolia Nsolo 1 17 (0) 27.6 (0)
Pteleopsis anisoptera Nepa 1 17 (0) 23.5 (0)
Pterocarpus angolensis Ntumbati 2 12 (0) 27.2 (3.9)
Tamarindus indica Ncueso 1 20 (0) 60.0 (0)
Terminalia sericea Ntchizo 10 15 (1) 20.3 (0.5)
Table 5
For the 12 species used for obtaining honey in Lizongole, Niassa National Reserve: percent of pre-existing honey trees felled or debarked; number of living
trees over the minimum diameter which were affected by honey-related activities; and percent of pre-existing trees
 the minimum diameter which were
killed as the result of felling or debarking.
Species felled or # # # % of preharvest Range in dbh (cm) of # living % felled or debarked of preharvest
debarked felled debarked living trees felled or felled, debarked trees
minimum dbh trees
minimum dbh felled or
debarked trees felled or debarked debarked
A. versicolor 3 0 9 25% 34.0e47.7 3 50%
B. africana 7 0 12 37% 20.0e32.0 3 70%
B. boehmii 3 1 108 4% 12.5e75.5 89 4%
B. manga 1 0 17 6% 36.3 0 100%
B. spiciformis 3 1 31 11% 29.9e90.0 6 40%
J. globiflora 10 22 189 14% 16.0e36.6 137 19%
Pericopsis angolensis 1 0 1 50% 36.3 0 100%
P. maprounifolia 1 0 36 3% 27.6 0 100%
P. anisoptera 1 0 5 17% 23.5 0 100%
Pterocarpus angolensis 2 0 13 13% 24.4e30.0 7 22%
T. indica 1 0 4 20% 60.0 1 50%
T. sericea 10 0 20 33% 17.0e39.0 10 43%
Fig. 5. Abundance/ha of living J. globiflora and of all other tree species/ha by diameter class derived from 5.9 ha of sample plots around Lizongole, Mecula District,
Niassa National Reserve.
10 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552class of 15.0e19.9 cm dbh may reveal a disturbance dynamic that has impeded regeneration in recent decades. Fig. 3 shows
that 36% of the living trees are concentrated in the smaller classes (10e20 cm), while very few individuals are found in size
classes > 40 cm dbh (fewer than 2 individuals/ha). Several studies carried out in NNR (Ribeiro, 2007; Ribeiro et al., 2013;
Cumbi, 2013) have shown that this distribution is common elsewhere in the reserve.3.3. Impacts of honey production
3.3.1. Impacts of gathering wild honey from natural cavities
Fig. 6 presents the number of individuals of each tree species felled for honey collection on 95 plots around Lizongole
village. With 10 felled trees each, J. globiflora and T. sericeawere the most frequently found, followed by Burkea africana,with
seven individuals found on the ground. A total of 42 trees of the following 12 species were found felled in the forest for honey
gathering: J. globiflora (‘Nchenga’), B. boehmii (‘Ndjombo’), B. spiciformis (‘Mpapa’), B. manga (‘Nanguezo’), Albizia versicolor
Welw. ex Oliv., Burkea africana (‘Nkalati’), Terminalia sericea (‘Ntchizo’), Pterocarpus angolensis (‘Ntumbati’), Pteleopsis ani-
soptera (Welw. ex Laws.) Engl. & Diels. (‘Nepa’), Pericopsis angolensis (Bak.) Meeuwen (‘Mbanga’), Pseudolachnostylis map-
rouneifolia (‘Nsolo’) and Tamarindus indica L. (‘Ncueso’). In total >6% of the total number of trees (living and dead) on our plots
had been felled for honey harvesting, but these represented 8% of the trees of those 12 species. The felled trees ranged from
12.5 cm to 90 cm in diameter, averaging 30.4± 2.4 cm dbh, while the 522 living trees ranged from 10 cm to 89.9 cm dbh,
averaging 20.2± 0.4 cm dbh.
Each felled tree had one cavity. Sizes of cavities after further opening to remove the honey combs are shown in Table 4.
3.3.2. Impact of bark hive production
In our sampled area of 5.9 hawe found a total of 24 trees killed as a result of debarking/girdling for hives (22 of J. globiflora,
one of B. boehmii and one of B. spiciformis) (Fig. 6). This represented 4% of the trees (living and dead) on our plots, but 6.5% of
the individuals of the 3 species from which bark had been removed. The Pearson correlation coefficient between the cut
height, dbh and bark thickness reveals a positive and statistically significant relationship between dbh and cut height
(r¼ 0.73; P< 0.05). Beekeepers prefer larger trees, fromwhich they remove bark from higher up.We also found a positive and
strong correlation between cut height and bark thickness (r¼ 0.5; P< 0.05), indicating that this preference to cut higher on
the tree (average 1.94m; STD¼ 0.62; range: 0.8e3.1m) may reflect a desire to obtain thicker bark, making for more durable
hives. The average bark thickness was 11.7mm (STD¼ 6.96mm; range from 0.35 to 33.26mm). The average diameter of the
24 debarked trees was 26.1± 1.19 cm.
3.3.3. Overall impact of honey production on targeted tree species
In total, among the 12 species of tree affected by honey-related activities, 13% of the individuals that had been living
previously on our sample plots had been killed by felling or debarking. However, the impact of honey gathering and hive
production on the Miombowoodlands varies among tree species. This reflects both their relative abundance in the woodland
and their values to: 1) honey bees as hive trees, whichmakes themvulnerable to felling; and 2) honey producers as sources of
bark for making traditional hives. Based on this sample of 5.9 ha of Miombo woodland affected by honey production, we
found that, depending on the species, between 4% and 50% of the trees larger than the minimum size felled or debarked had
been killed for honey production (Table 5).
Based on our sample, the species most strongly affected was Pericopsis angolensis, because it was so infrequent: we found
only one living tree and one felled tree on our plots. The largest number of trees killed by honey-related activities on our plots
(32) were J. globiflora, but this species was also the most abundant, so honey-related mortality killed only 14% of the total
number of individuals of this species that were standing prior to honey gathering or hive production. Although only seven B.
africana trees had been felled for honey, this represented 37% of the total prior population of trees of this species. Similarly, the
10 individuals of T. sericea that were felled for honey represented 33% of the living population of this species on our transects
before honey gathering (Table 5).
Because felling and debarking for honey focuses on relatively large trees, honey-related mortality disproportionately
affects the largest and oldest trees in the woodland. Large trees are also the most important for nectar, because their
crowns support large numbers of nectar-bearing flowers. They are also most important for wild hives, because it takes a
long time for natural cavities to develop that can be colonized by bees; and for man made hives, because their bark is
thick. When the proportion of large trees felled and debarked was compared to the density of living trees of these sizes
remaining on the transects, we found that, depending on the species, honey production had removed from 4% to 100% of
the trees above the minimum size felled or debarked. Of the 12 species that were felled or debarked, seven had lost 50%
or more of the population of trees of that diameter or larger and four species had lost 100% of the individuals of the sizes
used for honey.
The least affected species was B. boehmii, because of the large number of standing trees of this species, most of which were
over the 12.5 cm minimum diameter felled or debarked. J. globiflora retained 81% of the population of trees 
16 cm dbh, the
minimum size found felled or debarked (Table 5). Even so, that meant that 19% of the trees of this species over that diameter
had been lost from the population.
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 11
Fig. 6. Number of living, debarked and felled trees of the 12 species used for honey collection or traditional hive construction on 5.9 ha of sample plots in Lizongole village, Mecula District, Niassa National Reserve.
12 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e005524. Discussion
4.1. Current honey gathering practices
There are two peak periods of honey gathering in Lizongole: December and May, the latter being the more important.
Although these seasons are dependent on many environmental variables, and may shift from one year to another, these
collection periods are compatible with agriculture (sowing in October/November and harvesting in March/April for most
crops), thus allowing the two activities to be carried out sequentially during the year.
Although honey gathering involves walking long distances (up to 20 km; five to 10 days in the bush; up to 100 km reported
by Fischer, 1993), in this community it was preferred to honey production in traditional beehives (100% of the honey pro-
ducers, a total of 15, were gatherers compared to seven, 47%, who were beekeepers). Snook et al. (2015, 2016) also found a
lower number of beekeepers (only 11% of respondents) compared to honey gatherers (100% of the respondents) in a study
carried out among several communities in the NNR. According to the honey gatherers in Lizongole, honey gathering provides
greater honey quantities (up to 200 L per expedition compared to up to 20 L per beehive) and better honey quality, and
reduces the burden and the risks of managing the hives (including attacks from honey badgers and hive decolonization).
4.2. Are these practices depleting the populations of Miombo tree species?
Clauss (1992) concluded that apiculture was sustainable in the North Western Province of Zambia, where the number of
trees debarked was 3.1 trees/year/km2, from an available resource of about 224 trees/km2. Our estimates, based on the rates
described by honey gatherers and hive makers in Lizongole, were that 0.06 trees/km2/year are debarked for hives, and 0.79
trees/km2/year are felled for honey collection, a total of 0.85 trees/km2/year killed for honey. It would seem that the
woodlands in Lizongole are sufficiently stocked to sustain that rate. Using the calculation made by the reserve management
that the study area has 70% forest cover (SGDRN, 2007), we estimate that the 9000 km2 of land around Lizongole include
6300 km2 of woodlands with potential to be used for honey gathering and production. However, since this study, it is likely
that more land has been cleared for agriculture (Allan et al., 2017) and thus the woodland available has probably been
reduced. The sampling technique used in this study was biased towards finding trees affected by honey hunting, and
therefore would be expected to overestimate the absolute density of these species in the woodlands of the reserve. None-
theless, looking only at one species, the density found on our sample plots would correspond to approximately 1465 trees/
km2 of J. globiflorawhichwere 20 cm dbh or larger (and 1735 tree/km2 between 10 cm and 20 cmdbh). This means that within
a 20 km radius of Lizongole, an area of 1256 km2, there are approximately 879.2 km2 of woodland including up to 1,288,028
trees of J. globiflora which are currently 20 cm dbh or larger. This would seem to imply that honey gathering and hive pro-
duction are sustainable.
However, it is regionally recognized that traditional honey gatheringmethods are destructive (Ntenga andMugongo,1991;
Fischer, 1993; Syampungani et al., 2009; Snook et al., 2015, 2016), and honey gatherers in the Niassa Reserve observed that
areas where honey had been gathered over a period of time no longer yielded honey (Pers. com. Alberto Siabu to L. Snook
2012; Snook et al., 2015; Snook et al., 2016). The reasonwhy was demonstrated by our field study, which found that felling for
honey extraction from wild hives and debarking had killed 13% of the tree population that had existed previously on our
sample plots.While debarking for the production of beehives is selective, targetingmainly J. globiflora (also reported by Snook
et al., 2015, 2016), honey gathering through felling targets a broader variety of tree species, as it is based on the presence of
wild hives in natural cavities on adult trees. In Lizongole, J. globiflora and T. sericeawere the species most frequently felled for
wild honey.
Felling killed almost twice as many adult trees as debarking, probably because it is morewidely practiced. Also, once a hive
is made it may last five to 10 years, whereas each time honey hunters go out they need to harvest from new cavities (i.e. fell
new trees) because they have destroyed the hives and trees from which they previously collected (Snook et al., 2015, 2016).
Our results indicated that although J. globiflora was the species most frequently felled or debarked, its population was not
affected as much as populations of less abundant species.
Less abundant species were more affected by activities associated with obtaining honey. Felling or debarking had killed
between 20% and 50% of the populations of A. versicolor, B. africana, Pericopsis angolensis, T. indica and T. sericea and 40%e100%
of the large trees of nine of the 12 species affected by honey-related activities. These species also face other threats, including
felling for other uses (e.g. J. globiflora is also used for fiber and building material) and fires. Additionally, trees in the NNR
woodlands grow slowly, at a rate of 0.24 cm/year in diameter (Ribeiro et al. in prep) so replacement rates for honey trees that
are killed are slow. It would take 44 years for a 10 cm dbh tree to reach 27 cm dbh (the average to yield a bark beehive), and
even more for it to reach the sizes where it would be likely to provide conditions for establishment of wild bee colonies.
Notwithstanding a few regional examples of honey production's compromising ecosystem sustainability (Towry-Coker,
1995 in Zimbabwe and Ntenga and Mugongo, 1991 in Tanzania), a number of authors have proposed that subsistence
honey production (gathering and apiculture) can contribute to maintaining the ecological dynamic of the ecosystem (Smith,
1962; Fischer, 1993). The killing of larger trees creates small gaps in the tree canopy, which could give room for regeneration
and growth of younger trees.
However, as practiced commonly in the NNR, honey gathering uses fire, and is an important cause of uncontrolled fires (as
was reported also by Guy, 1971). Snook et al. (2015) found higher evidence of fires in areas where honey gathering had been
N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552 13practiced. Although the ecosystem is adapted to fire (Ribeiro et al., 2008, 2017) this accelerated fire regime (fire frequency of
every one to two years) undermines the regeneration of some species (Schrotter, 2014) and may threaten the ecosystem
(Ribeiro et al., 2017). Local community members have reported that fires kill some mellifeorus plants and promote dry
conditions, which further compromises bee colonies as well as future honey production. It is important to consider that there
are other causes of fire, hunting and shifting cultivation being among the most important, and thus defining strategies for fire
management requires a broader approach than simply addressing honey gathering.
5. Conclusion: what are the management options to sustain honey yields and reduce impacts on tree populations?
Honey harvesting and beehive constructionwere found to have substantially reduced the abundance of larger trees of the
12 targeted species, as well as destroying bee colonies when cavity trees are felled. This represents a threat to the sustain-
ability of honey gathering as an activity, and is also likely to reduce the potential of honey production from hives, since bees
depend on the nectar resource provided by large-diameter trees with large crowns. Human populations are also growing in
this area, increasing agricultural clearing and pressure on the woodland resource (Allan et al., 2017).
Alternatives do exist to debarking or felling trees to obtain honey. For example climbing techniques that were traditional in
the past, and are still used to obtain honey from baobab (Adansonia digitata) trees, which are too large to fell, could contribute
to making honey gathering more sustainable. After these techniques were demonstrated to groups of honey gatherers in
several communities in the Niassa National Reserve, most of them reported they had adopted them, finding them easier and
less time-consuming than felling hive trees (Snook et al., 2015, 2016).
Compartmentalizing different areas for continuous use by specific communities, a practice already implemented by some
communities in the NNR, provides an incentive for using sustainable honey gathering practices, like tree climbing, which
allow for repeated honey gathering from the same trees (Snook et al., 2015, 2016). Similarly, the use of beehives made of
materials other than bark obtained from local trees, would reduce the ecological impact of beehive management. Consid-
eration would have to be given to avoiding the risk of destruction by the frequent fires in the area, or by predation by honey
badgers.
Acknowledgements
Funding for this project was provided to Bioversity International by the Austrian Development Agency/Austrian Devel-
opment Cooperation (ADA 2009/01) and by the donors to the CGIAR Research Program on Forests, Trees and Agroforestry. The
funders had no input to the study design, the article, or the decision to publish. The authors thank the former Sociedade Para a
Gesta~o da Reserva Nacional do Niassa (SGDRN), the community and honey gatherers of Lizongole and the NNR managers
(Mrs. Anabela Rodrigues, Mr. Madyo Couto and Mr. Cornelio Miguel), for their support and information. Other project col-
laborators who contributed to this study included Mr. Feliciano Mazuze, Ms. Rosalina Mahanzul,; Ms. Camila Sousa and Dr.
Judy Loo. Mr. Alberto Siabu of Mecula provided valuable insights on honey-related activities and bees in the Niassa National
Reserve and Ms. Julia Fogerite carried out a useful literature review on bees in the Miombo woodland.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.gecco.2019.e00552.
References
Alcobia, J.J.F., 1995. Apicultura: Relatorio de trabalho. Cooperativa de Apicultores de Maputo, Maputo.
Allan, J., Grossman, F., Craig, R., Nelson, A., Maina, J., Flower, K., Bampton, J., Deffontaines, J.-B., Miguel, C., Araquechande, B., Watson, J., 2017. Patterns of
forest loss in one of Afria's last remaining wilderness areas: Niassa Natonal Reserve (Northern Mozambique). Parks 23 (2), 39e50.
Alves, T., 2011. Focus Groups. Unpublished Draft Report, Bioversity International-Led Project, “Sustaining Forest Resources for People and the Environment
in the Niassa National Reserve, Mozambique”, p. 30. Maputo.
Campbell, B., 1996. The Miombo in Transtition: Woodlands and Welfare in Africa. CIFOR, Bogor, p. 263.
Clarke, J., Cavendish, W., Coote, C., 1996. Rural households and miombo woodlands: use,value and management. In: Campell, B. (Ed.), The Miombo in
Transition: Woodlands and Welfare in Africa, first ed. CIFOR, Bogor, pp. 101e135.
Clauss, B., 1992. Bees and Beekeeping in the North-Western Province of Zambia: Report on a Beekeeping Survey. Forest Department and Integrated Rural
Development Programme (IRDP), Ndola.
Cumbi, A., 2013. Avaliaça~o dos conflitos relacionados com o uso do fogo e a conservaç~ao de algumas especies florestais pelas comunidades na Reserva
Nacional do Niassa. Dissertation, Faculdade de Agronomia e Engenharia Florestal. Universidade Eduardo Mondlane.
Cunliffe, R., Mandondo, A., Games, I., Ngariyhume, J., Dore, D., 2009. Reconciling conservation goals with agriculturally based livelihoods: a proposal for
future development of the Niassa national reserve and surrounding areas. In: Niassa Imperial Tobacco Project, Main Report, vol. 1. Centre for Agrarian
and Environmental Research, Harare, Zimbabwe, p. 47.
Dewees, P., Campbell, B., Katerere, Y., Sitoe, A., Cunningham, A., Angelsen, A., Wunder, S., 2010. Managing the Miombo Woodlands of Southern Africa:
Policies, Incentives and Options for Rural Poor. Program on Forests (PROFOR), Wasington DC.
Filgueiras, T.S., Nogueira, P.E., Brochado, A.L., Guala, I.J., 1994. Caminhamento: Um metodo para levantamentos floristicos qualitativos. Cadernos de Geo-
ciências, IBGE nº 12, 39e43 (Piracicaba, Brasil).
Fischer, F.U., 1993. Beekeeping in the Subsistence Economy of He Miombo Savanna Woodlands of South₋Central Africa. Rural Development Forest Network
15c, p. 1₋12.
Franklin, J.F., Shugart, H.H., Harmon, M.E., 1987. Tree death as an ecological process. The causes, consequences, and variability of tree mortality. Bioscience
37 (8), 550e556.
14 N.S. Ribeiro et al. / Global Ecology and Conservation 17 (2019) e00552Fletcher, D.J.C., 1978. The African bee, Apis mellifera adansonii, in Africa. Annu. Rev. Entomol. 23, 151e171.
Frost, P., 1996. The ecology of miombo woodlands. In: Campell, B. (Ed.), The Miombo in Transition: Woodlands and Welfare in Africa, first ed. CIFOR, Bogor,
pp. 11e55.
Guy, R., 1971. Mozambique. South African Bee Journal 43 (I), 11e19.
Husselman, M., November 2008. Generating incomes from dry forest products: case studies from Mwinilunga, Kapiri and Chongwe districts, Zambia. In:
Paper to the Sustainable Forest Management in Africa Symposium. Stellenbosch, South Africa, pp. 3e7.
Illgner, P.M., Nel, E.L., Robertson, M.P., 1998. Beekeeping and local self-reliance in rural Southern Africa. Geogr. Rev. 88 (3), 349e362.
Instituto Nacional de Estatistica (INE), 2010. Indicadores Socio demograficos. Maputo.
Kent, M., 2012. Vegetation Description and Data Analysis: A Practical Approach. Wiley-Blackwell, Oxford, p. p242.
Lawton, R.M., 1982. Natural resources of miombo woodland and recent changes in agricultural and land-use practices. For. Ecol. Manag. 4 (3), 287e297.
Mickels- Kokwe, 2006. Small-Scale Woodland-based Enterprises with Outstanding Economic Potential. The Case of Honey in Zambia. CIFOR, Bogor Barat,
Indonesia, p. 82. ISBN 979-24-4673-7.
Mpuya, P., 2003. An overview of bee industry in Tanzania. In: Production of Bee Products and Marketing Promotion Workshop. Arusha, Tanzania.
Mudekwe, J., 2017. Forest resource use patterns in a protected forest reserve in western Zimbabwe-the case of fuller protected forest. International Journal
of Bio-resource and Stress Management 8 (2), 346e359.
Ntenga, G.M., Mugongo, B.T., 1991. Honey Hunters and Beekeepers: a Study of Traditional Beekeeping in Babbati District, Tanzania. Swedish University
Agricultural Sciences. International Rural Centre, p. 84.
Pijnerburg, S.E., Cavane, E., 2000. Metodos e tecnicas de investigaçaeo socio-economica. FAEF.UEM.Maputo, p. 78.
Price, T., 1964. Yao origins. The Nyasaland Journal 17 (2), 11e16. Retrieved from. http://www.jstor.org/stable/29545968.
Reader, J., 1998. Africa: A Biography of the Continent. Penguin Books, London, p. 803.
Ribeiro, N.S., 2007. Interactions between Fires and Elephants in Relation to Vegetation Composition and Structure of Miombo Woodlands in Northern
Mozambique. Dissertation. University of Virginia.
Ribeiro, N.S., Shugart, H.H., Washington-Allen, R., 2008. The effects of fire and elephants on species composition and structure of the Niassa Reserve,
northern Mozambique. For. Ecol. Manag. 255, 1626e1636.
Ribeiro, N.S., Matos, C.N., Moura, I., Washington-Allen, R.A., Ribeiro, A.I., 2013. Monitoring vegetation dynamics and carbon stock density in miombo
woodlands. Carbon Balance Manag. 8, 1e9.
Ribeiro, N.S., Cangela, A., Chauque, A., Bandeira, R.R., Ribeiro-Barros, A.I., 2017. Characterisation of spatial and temporal distribution of the fire regime in
Niassa National Reserve, northern Mozambique. Int. J. Wildland Fire 26 (12), 1e30.
Schrotter, C., 2014. Influence of Honey Harvesting on Regeneration of Selected Tree Species in the Niassa National Reserve, Mozambique. Masters Thesis.
Institute of Forest Ecology, Department of Forest and Soil Sciences University of Natural Resources and Life Sciences, Vienna, p. 76.
SGDRN (Sociedade de Gestao e Desenvolvimento da Reserva do Niassa), 2007. Plano de Maneio da Reserva Nacional do Niassa, p. 139. Maputo.
Smith, F.G., 1962. Beekeeping as a Forest Industry. Commonwealth Forestry Conference, p. 6.
Snook, L., Alves, T., Sousa, C., Loo, J., Gratzer, G., Duguma, L., Schrotter, C., Ribeiro, N., Mahanzule, R., Mazuze, F., Cuco, E., Elias, M., 2015. Relearning traditional
knowledge to achieve sustainability: honey gathering ́in the mioḿbo woodlands of northern Mozambique. In: XIV World Forestry Congress, Durban,
South Africa, 7-11 September 2015. http://foris.fao.org/wfc2015/api/file/552e8b8e9e00c2f116f8eac2/contents/028c3429-30dd-4b71-856b-
22ea44d4849c.pdf.
Snook, L., Alves, T., Sousa, C., Loo, J., Gratzer, G., Duguma, L., Schrotter, C., Ribeiro, N., Mahanzule, R., Mazuze, F., Cuco, E., Elias, M., 2016. Relearning traditional
knowledge to achieve sustainability: honey gathering in the miombo woodlands of northern Mozambique. Poster and Paper. Tropentag 2016. In:
Conference on International Research on Food Security, Natural Resource Management and Rural Development, Vienna, Austria, Sept 19-21, 2016.
http://www.tropentag.de/2016/abstracts/full/444.pdf.
Spottiswoode, C.N., Begg, K.S., Begg, C.M., 2016. Reciprocal signalling in honeyguide human mutualism. Science 353, 387e389.
Syampungani, S., Chirwa, P.W., Akinnifesi, F.K., Sileshi, G., Ajayi, O.C., 2009. The miombo woodlands at the cross roads: potential threats, sustainable
livelihoods, policy gaps and challenges. Nat. Resour. Forum 33, 150e159.
Systat 13 2009 Systat Software, Inc.,́Chicago, ́IL. ́
Thorold, A., 1987. Yao conversion to islam. Camb. Anthropol. 12 (2), 18e28. Retrieved from. http://www.jstor.org/stable/23817396.
Timberlake, J., Golding, J., Clarke, P., 2004. Niassa Botanical Expedition: June 2003, Prepared for Sociedade para a Gesta~o e Desenvolvimento da Reserva do
Niassa Moçambique. Biodiversity Foundation for Africa, pp. 3e20.
Towry-Coker, S., 1995. Honey: A Practical Manual for African Community Beekeepers. Harare, Zimbabwe. Participatory Development Resource Center for
Africa.
White, A., 1983. The Vegetation in Africa: A Descriptive Memoir to Accompany the UNESCO/AETFAT/UNSO. Africa Natural Resources Research, vol. 20.
UNESCO, Paris.