Can Homegardens Conserve Biodiversity in Bangladesh?

Authors

  • Md. Enamul Kabir,

    1. Natural Resources Management, School of Environment, Resources & Development, The Asian Institute of Technology, P.O. Box 4 Klong Luang, Pathum Thani, 12120, Thailand
    Search for more papers by this author
  • Edward L. Webb

    Corresponding author
    1. Natural Resources Management, School of Environment, Resources & Development, The Asian Institute of Technology, P.O. Box 4 Klong Luang, Pathum Thani, 12120, Thailand
    Search for more papers by this author

Corresponding author; e-mail: ewebb@ait.ac.th

ABSTRACT

In managed landscapes, tree-dominated habitats often show promise for biodiversity conservation. In Bangladesh where natural forest cover is less than 10 percent, homegardens, which are maintained by at least 20 millions households, represent one possible strategy for biodiversity conservation. This study investigated the floristic and structural diversity of 402 homegardens from six regions across southwestern Bangladesh. All plants were censused, totaling 419 species (59% native), including six IUCN Red Listed. The median homegarden (800 m2) contained a mean of 34 species. Each region contained a mean of 293 species in a mean of 67 homegardens. A total of 49,478 individuals (107 per homegarden and 1003 per hectare) of trees and shrubs were counted from 45.2 ha total sampled area. Thus, significant botanical richness was exhibited in the homegardens across southwestern Bangladesh. However, most species were rare: 82 percent of all species including 189 native were found in 50 or fewer homegardens, and 63 species (36 native) were found in only one or two homegardens. Sixty percent of all tree and shrub species had 50 or fewer individuals each. Thus, whereas richness across the landscape was high, serious effort must be made to increase the populations of most species. We propose three main conservation activities: (1) awareness building; (2) protection of existing individuals of rare species; and (3) propagation. Overlaying all of these activities is the inclusion of local communities in the process, who were the ones to retain these species in homegardens in the first place, and the stakeholders who will determine whether homegardens indeed act as long-term repositories to biodiversity conservation.

The forests of Bangladesh were home to more than 5000 species of flowering plants (FAO 2000), providing an enormous array of consumptive and nonconsumptive products and services. Forests also provided many ecological and environmental services and possessed an array of social, cultural, religious, aesthetic, and recreational values. Once largely forested, Bangladesh now has a forest cover of less than 10 percent (FAO 2000). Deforestation and fragmentation have resulted in greatly reduced biodiversity and have created major challenges for conservation. Despite governmental interest to increase the protected area system to 10 percent of the territory (FAO 2000), conservation of biodiversity in Bangladesh will require major contributions from private, managed patches outside natural and protected area systems.

Yet very little attention has been given to what extent managed landscapes could hold biodiversity value (Peyre et al. 2006). In managed landscapes, tree-dominated habitats such as natural forest patches (Schelhas & Greenberg 1996) or agroforests such as homegardens (Schroth et al. 2004, Mendez & Bacon 2005) often show promise for biodiversity conservation, so in Bangladesh, tree-dominated patches may also be important for conservation. People have been cultivating, managing, and conserving diverse plants in and around their houses for alternative sources of forest products and services, supporting the idea that these managed patches could be of significant conservation value.

Homegardening is a traditional agroforestry system where a clearly bounded piece of land immediately surrounding the dwelling house is cultivated with a mixture of perennials and annuals (Fernandes & Nair 1986, Wojtkowski 1993, Wiersum 2004). Homegardens are a common feature in many tropical countries, and often exhibit remarkable variability in composition and structure (Price & Campbell 1998, Mendez et al. 2001, Ceccolini 2002, Kehlenbeck & Maass 2004, Albuquerque et al. 2005, Peyre et al. 2006). Homegardens are maintained by at least 20 million households and cover an area of 270,000 ha (2% of the country's total land area and 10% of the total primary forest area, FAO 2000) across Bangladesh (Salam et al. 2000) and therefore represent one possible strategy for biodiversity conservation. Indeed, the limited research on Bangladeshi homegardens has revealed diverse floristic composition and complex garden structure (Millat-e-Mustafa et al. 1996, Uddin et al. 2002, Ahmed & Rahman 2004, Ali 2005).

Species diversity in a homegarden can range from less than five (e.g., Ahmed & Rahman 2004, Coomes & Ban 2004, Withrow-Robinson & Hibbs 2005, Abdoellah et al. 2006) to more than 100 (e.g., Mendez et al. 2001, Vogl & Vogl-Lukasser 2003, Hemp 2006), and therefore can be important islands of diverse plants. Yet systematic study tends to suffer from low sample sizes (i.e., few homegardens sampled), or limited taxonomic treatment (e.g., surveying only trees), so the capacity of homegardens to contribute to biodiversity conservation in Bangladesh remains understudied.

Can homegardens conserve biodiversity in Bangladesh? This is an important question, given the low forest cover in the country and peoples' high dependency on the natural production systems. The first step in assessing the conservation value of homegardens is to undertake a thorough botanical and structural survey. Total plant diversity measures the direct conservation value of homegardens, while structural features may indirectly conserve other taxa, such as frugivorous birds, reptiles, amphibians, small mammals, or arthropods (Institute of Ecology 1979, Soemarwoto & Conway 1992, Grifith 2000, Montagnini 2006). This research therefore explored and quantitatively assessed the total botanical diversity and structural characteristics of 402 homegardens in southwestern Bangladesh. This large sample size allows us to report that homegardens in Bangladesh are excellent repositories of diverse plants, but that significant—but not insurmountable—obstacles remain in making them effective conservation tools.

METHODS

Study site.— Southwetsern Bangladesh is a low (<10 m asl), flat, and fertile deltaic plain predominated by calcareous to noncalcareous alluvium soils (BBS 2004). The coastal plain is sandy and saline. A tropical to subtropical monsoon climate characterizes the region. Three distinct seasons are summer (March–May), rainy (June–October), and winter (November–February). The annual average rainfall is 1800 ± 268 mm. The mean annual temperature is 26°C (range: 19–32°C). In some places, temperatures drop to 10°C during the winter and reach 40°C or more during the summer.

Population density in the region is 650 persons/km2 with a median family size of six. About 50 percent of the population lives on <1 US$/day. Agriculture is the main occupation for the vast majority of people in the region. Rice, wheat, jute, sugarcane, pulses, and potatoes are the principal agricultural crops. Southwestern Bangladesh is also important for production of various types of vegetables, spices, fruits, and nuts along with the principal agricultural crops. Intensive shrimp (tiger prawn) culture is a newly emerging economic activity along the coastal belt of the region. No public forests are accessible to households in the study area.

Data collection.— A botanical inventory was conducted in the homegardens of 402 randomly selected sample households across six major regions in southwestern Bangladesh (Fig. 1) from April 2005 to January 2006—Khulna, Bagerhat, Satkhira, Jessore, Chuadanga, and Faridpur districts. All species present in each sampled homegarden were identified and recorded by the botanical name, or by local name that was later confirmed from the Bangladesh National Herbarium if the botanical name was not immediately known. All individuals of trees and shrubs were counted and recorded except the individuals in hedgerows. No herbs or climbers were counted due to the difficulty in differentiating stems. The location and altitude of each sample household was recorded by a global positioning system (GPS). A botanical inventory was conducted only once in each selected homegarden. Thus, the seasonal variation in floristics and structure was not assessed.

Figure 1.

Study area, southwestern Bangladesh. Black dots show the location of sample homegardens from 402 randomly selected households.

Data analysis.— Each species recorded was classified by family, habit based on the morphology of the plant when it was full grown (tree, shrub, herb, and climber), origin (native or introduced to the Indian subcontinent), and conservation status (on the Bangladesh IUCN ‘Red Listed’ or not; Khan et al. 2001).

Abundance (number of individuals per species) and frequency (fraction of homegardens containing the species; Cox 1990) for each species of tree and shrub (except those planted in hedgerows) were calculated. The importance value of each tree and shrub species was then estimated by totaling each species relative abundance and frequency (Kaya et al. 2002, Das & Das 2005). Relative importance was used to rank species per life form. For herbs and climbers, relative frequency was used to rank species. Species overlap in homegardens across the six major study regions was assessed using Jaccard's Similarity Index (van Tongeren 1995).

RESULTS

Floristic composition.— Homegarden size ranged from 100 m2 to 1.75 ha, with a median (used for non-normal data) size of 800 m2. A median of 7.1 ha (range: 3.2–11.1 ha) area was surveyed per region. The total sample area was 45.2 ha from a total of 402 surveyed homegardens (Table 1), making this inventory the most area-intensive homegarden inventory in South and Southeast Asia, and one of the most intensive globally.

Table 1. Plant species composition and structure of the homegardens of 402 randomly selected households totaling 45.2 ha area across six major regions in southwestern Bangladesh. Native plants were defined according to Indian Subcontinent Floristics (Hooker 1875). Plants that have been actively cultivated by the gardeners were defined as planted, and plants that had established naturally were natural. Plants that had been declared under any level of threat by the IUCN were defined as red listed (Khan et al. 2001). The numbers in parentheses represent the percent of total recorded species. *indicates median values, were used for non-normal data.
 RegionMeanTotal
KhulnaBagerhatSatkhiraJessoreChuadangaFaridpur
No. of homegardens surveyed69697165636567402
Total homestead area surveyed (ha)   11.10    8.17   10.79    6.17    5.73    3.23    7.17*   45.2
Average homestead area (ha)    0.16    0.12    0.15    0.09    0.09    0.05    0.08* 
Homestead area range (ha)0.01–0.80.01–1.00.01–1.750.01–0.670.01–0.400.01–0.200.01–0.74* 
No. of species320 277 321 277 237 326 293 419
 Native179155180153129175162246 (59)
 Exotic141122141124108151131173 (41)
 Planted239209240225194267229314 (75)
 Natural81688152435964105 (25)
 Tree1181051219885109106146 (35)
 Shrub5845504642585067 (16)
 Herb10488107927811397150 (36)
 Climber4039434132464056 (13)
IUCN Red Listed 4 3 4 3 3 5 46 (0.01)
No. of species/homegarden41333129294234 
 Native22181615152118 
 Exotic19151514142116 
 Planted34272624233428 
 Natural7655686 
 Tree22171614152218 
 Shrub 6554465 
 Herb119898119 
 Climber2222232 
No. of individuals12,811    13,800    7630  4962  3316  6959  7295*  49,478
 Tree11,814    13,211    7000  4528   2920  6292  6646*  45,765
 Shrub997589630434396667609* 3713
No. of individuals/homegarden186 200 107 7653107 107*  
 Tree1691829769489697* 
 Shrub171810751110* 
No. of individuals/ha1163  1667  713 844 589 2140  1003*   
 Tree1058  1517  6497685361947  913*  
 Shrub10515064765319390* 

A total of 419 plant species in 109 families were recorded from southwestern Bangladesh homegardens (Table S1). The mean of 293 species (range: 237–326) per region was represented by 106 trees, 50 shrubs, 97 herbs, and 40 woody and nonwoody climbers (Table 1). There were more native species than exotic species across all six regions (Table 1). Fifty-nine percent of all recorded species were native to the Indian subcontinent. Of the 419 species, 146 were trees, 67 shrubs, 150 herbs, and 56 woody and nonwoody climbers (Table 1). Trees and herbs predominated in homegardens across all six regions.

Six species, Schleichera oleosa (kosum), Mangifera sylvatica (forest mango), Rauvolfia serpentina (snake root), Andrographis paniculata (creat), Amomum aromaticum (Bengal cardamom), and Calamus guruba (rattan) appear on the IUCN Red List with a mean of four species (range: 3–5) per region. All recorded red listed species were planted except for M. sylvatica.

Floristic similarity.— The Jaccard Index across all pairwise comparisons of regions averaged 0.68 (range: 0.64–0.74; Table 2). The results of the Jaccard Index demonstrated moderately high (68%) floristic similarity in the homegardens across regions.

Table 2. Jaccard similarity index across six major regions in southwestern Bangladesh. Index used species presence–absence data in each region.
RegionBagerhatSatkhiraJessoreChuadangaFaridpur
Khulna0.680.680.690.640.65
Bagerhat 0.720.700.680.73
Sathkhira 0.680.690.66
Jessore 0.740.72
Chuadanga 0.64

Structural characteristics.— The number of species per homegarden was consistent across all six regions. Each homegarden contained two to 107 plant species, with a mean of 34 species (Fig. 2) represented by 18 trees, five shrubs, nine herbs, and two climbers. The number of species per homegarden would increase if surveys were conducted across seasons. The number of tree individuals was substantially higher than that of shrubs within and among homegardens. A total of 49,478 individuals were accounted by 92.5 percent trees and 7.5 percent shrubs. The median 7295 individuals per region (range: 3316–12,811) was represented by 91 percent trees (range: 88–96) and nine percent shrubs (range: 4–12). The median 107 individuals per homegarden (1003 per hectare) was represented by 97 trees (913 per hectare) and 10 shrubs (90 per hectare).

Figure 2.

Frequency distribution of plant species per homegarden of 402 randomly selected households totaling 45.2-ha area in southwestern Bangladesh. A total of 419 plant species were identified with a mean of 34 species per homegarden of 800-m2 median area.

Approximately half of the ten most important trees and climbers were native, but exotic shrubs and herbs, particularly Citrus limon and Musa spp. dominated the shrub and herb synusae, respectively (Table 3). All species recorded from the homegardens were useful for nine different purposes (Table 3). Most species were used for food (36% of all species) followed by medicine (27%), fuelwood (22%), ornamental (19%), timber (11%), and fodder (8%). Forty-five percent of all species were multipurpose. Eighty percent of the ten most important tree (90%), herb (70%), and climber (80%) species were multipurpose (Table 3).

Table 3. The ten most important species of trees, shrubs, herbs, and climbers (woody and nonwoody) in southwestern Bangladesh homegardens.
Botanical nameLocal nameFamilyOriginUsesRF
  1. Native species are denoted with ‘N’ and exotic species with ‘E.’ For uses, 1 = food, 2 = commercial (used to earn cash from the sale of surplus products after subsistence consumption, not an end use), 3 = medicinal, 4 = fuelwood, 5 = ornamental, 6 = timber, 7 = fodder, 8 = fiber, and 9 = religious/ceremonial. RF is relative frequency of the species that represent the relative importance of the species in their respective life form.

Trees 
 Cocos nucifera L.NarikelPalmae (Arecaceae)N1,474.13
 Mangifera indica L.AmAnacardiaceaeN1, 2, 4, 6,769.15
 Areca catechu L.SuperiPalmae (Arecaceae)E2,665.17
 Psidium guajava L.PearaMyrtaceaeE1, 2,464.93
 Swietenia macrophylla KingMahoganyMeliaceaeE2,655.72
 Artocarpus heterophyllus Lam.KanthalMoraceaeE1, 2,655.22
 Phoenix sylvestris Roxb.KhejurPalmae (Arecaceae)N1, 2, 4,953.23
 Lannea coromandelica (Houtt.) Merr.JigaAnacardiaceaeN451.99
 Samanea saman (Jacq.) Merr.Rain TreeMimosoideaeE4,646.52
 Spondias pinnata (L.f.) KurzAmraAnacardiaceaeN1, 2,445.02
Shrubs 
 Citrus limon (L.) Burm.f.Kagojee LebuRutaceaeE1,236.82
 Eupatorium odoratum L.Assam LataAsteraceaeE320.15
 Clerodendrum inerme (L.) Gaertn.Bon JuiVerbenaceaeN318.66
 Capsicum annuum L.MorichSolanaceaeE2,817.41
 Bougainvillea brachycarpa HeimerlBagan BilashNyctaginaceaeE516.67
 Justicia adhatoda L. NeesBashokAcanthaceaeN316.67
 Lawsonia inermis L.MendiLythraceaeE916.67
 Codiaeum variegatum (L.) BlumePata BaharEuphorbiaceaeE516.17
 Glycosmis pentaphylla (Retz.) DC.Daton GachRutaceaeN415.92
 Cestrum nocturnum L.HasnahenaSolanaceaeE515.67
Herbs 
 Musa spp.KolaMusaceaeE1, 2,766.42
 Alocasia indica (Lour) KochMan KachuAraceaeN1, 251.24
 Xanthosoma nigrum (Vell.) Mans.Dosta KachuAraceaeE1, 233.33
 Colocasia nymphaeifolia KunthBuno KachuAraceaeN132.84
 Bambusa balcooa Roxb.Balco BansGramineaeN2, 4, 6,7, 931.59
 Curcuma longa L.HoludZingiberaceaeN2, 3, 8,1226.87
 Amorphophallus paeoniifolius (Denn.) Nicol.Ol KachuAraceaeN1,224.13
 Tagetes patula L.Boro GadaAsteraceaeE3,523.13
 Heliotropium indicum L.HatisurBoraginaceaeN318.16
 Gomphrena globosa L.Time PhulAmaranthaceaeE515.92
Climbers (woody and nonwoody) 
 Basella alba L.Pui ShakBasellaceaeE1,235.07
 Dioscorea alata L.Mete AluDioscoreaceaeN1,230.35
 Cucurbita maxima Duchesne ex Lmk.Mishti KumraCucurbitaceaeE1,224.88
 Benincasa hispida (Thunb.) Cogn.Chal KumraCucurbitaceaeE1,221.64
 Lagenaria siceraria (Md.) Standl.LaoCucurbitaceaeE1,219.65
 Hoya verticillata (Vahl.) G. DonPargachaAsclepiadaceaeN317.16
 Cucumis sativus L.ShoshaCucurbitaceaeE1,215.67
 Cissus quadrangularis L.Harjora LataVitaceaeN314.93
 Lablab perennans DC.ShimPapilionoideaeE1,214.43
 Piper abbreviatum OpizChuiPiperaceaeN2,812.94

Floristic rarity.— Although species diversity and similarity across regions were high, the majority of species were rare. Among all recorded species, 56 percent were found in 20 or fewer homegardens and 82 percent in 50 or fewer homegardens (Fig. 3). Only 31 species (7%) were found in 100 or more homegardens. Sixty-three species (15%) were found only in one or two homegardens. In terms of native plant species, only 17 (7%) were found in 100 or more homegardens. In addition, 189 native species (77%) were found in 50 or fewer homegardens, and 36 in only one or two homegardens.

Figure 3.

Number of homegardens containing the proportion of total recorded 419 plant species from 402 randomly selected households in southwestern Bangladesh.

All six IUCN Red Listed species were planted in a total of 95 homegardens. Of these, 91 planted only one species and four planted two species. Andrographis paniculata and C. guruba were planted in 53 and 37 homegardens, respectively. Thus, households focused on only two threatened species; otherwise the prevalence of threatened species in homegardens was low.

About 60 percent of all tree (55%) and shrub (72%) species had 50 or fewer individuals each and 74 percent of all tree (66%) and shrub (90%) had 100 or fewer individuals each. Only eight of all 146 tree species had 1000 or more individuals each (max: 9039) and two of all 67 shrub species had 350 or more individuals each (max: 710) (Fig. 4).

Figure 4.

Number of individuals of tree and shrub species from the homegardens of 402 randomly selected households totaling 45.2-ha area in southwestern Bangladesh. A total of 49,478 individuals of tree (45,765) and shrub (3713) were recorded from all sampled homegardens. All recorded individuals were represented by 213 identified species of tree (146) and shrub (67).

DISCUSSION

Homegardens as biodiversity islands? Primary forests are the richest reservoir of biodiversity, but Bangladesh's low primary forest cover and the degraded status of remaining tracts indicate that major efforts are required to maintain viable populations of species across the heavily impacted landscape of Bangladesh. Agroforestry systems, such as mixed shade coffee production (Perfecto et al. 1996) or Indonesian agroforests (Thiollay 1995), can contain significant levels of both plant and animal biodiversity (Grifith 2000, Montagnini 2006). In Bangladesh, homegardens are the most widespread agroforestry system and therefore may represent the only large-scale managed system with potential for biodiversity conservation in the country.

Compared with other published studies across the world, homegardens in southwestern Bangladesh exhibited high species richness. Globally, eight sites (see Karyono 1981, Padoch & de Jong 1991, Soemarwoto & Conway 1992, Herrera-Castro et al. 1993, House & Ochoa 1998, Lok et al. 1998, Mendez et al. 2001, Hemp 2006) have been documented to have higher species richness than this study, accounting for both sample size and total area sampled (Fig. 5). Only eight sites globally and four sites in South and Southeast Asia have been documented to have a per homegarden species richness significantly higher than the present study (Perera & Rajapaksae 1991, Soemarwoto & Conway 1992, Jensen 1993, House & Ochoa 1998, Mendez et al. 2001, Trinh et al. 2003, Vogl & Vogl-Lukasser 2003, Hemp 2006). This trend is also valid for the diversity of native species: ca 59 percent of the species in the homegardens of this study were native. Thus, southwestern Bangladesh homegardens represent important repositories of diverse plants. These homegardens appear to be of greater relative importance to biodiversity conservation in Bangladesh than most other parts of Asia, and possibly the tropical region generally.

Figure 5.

The relationship of species richness with sampling intensity: (A) number of homegardens surveyed and (B) total area surveyed between this study and studies from other regions of the world. The comparison included only those studies that have documented total diversity of plants of all life forms. The line of this study was computed based on the best-fit nonlinear regression model developed from the randomly subsampled data.

Several studies have observed that the floristic composition of tropical homegardens may be similar to natural ecosystems (De Clerck & Negreros-Castillo 2000, Shastri et al. 2002, Albuquerque et al. 2005). Unfortunately, we did not investigate the level of floristic similarity between homegardens and natural forests because there have been no quantitative surveys comparable to this with natural forest. Floristic similarities between Bangladesh homegardens and natural forests should be investigated in future studies.

Structurally and functionally, homegardens may be similar to fragmented forest islands. A small fragmented island can support native wildlife such as frugivorous birds (Bierregaard & Stouffer 1997), small mammals (Malcolm 1997), and butterflies (Brown & Hutchings 1997). For example, 78 species of birds were recorded from homegardens in West Java, Indonesia (Institute of Ecology 1979). During the botanical inventory, the field team observed many frugivorous birds and bats, reptiles, amphibians, insects, and small mammals in homegardens. The diversity of wildlife in homegardens will be a reflection on the structural complexity of the vegetation. For example, a complex forest structure provides different niches for wildlife than simple-structured vegetation (Hanowski et al. 1997, Lindberg et al. 1998). The typical homegarden in this study was dominated by trees but contained plants of all other synusae in several strata. This structure could make homegardens attractive to, and serve as important refuge for, wildlife in Bangladesh. Further study on the use of homegardens by wildlife, and the role of structure in wildlife use, should be undertaken to understand how to maximize the wildlife value of homegardens.

Rare species, farmer awareness and EX SITU conservation in homegardens.— The moderate level of species overlap we found across the six regions suggests some similarity in the total array of species planted by farmers. However, most species were rare (i.e., in low abundance). This was particularly true for the red listed plants, which were found in ca 25 percent of all surveyed homegardens. Besides the obvious explanation that low natural abundance would account for low utilization by farmers, farmers might be unaware of the potential uses or benefits from native species. For example, R. serpentina (snake root) is used as antivenom in rural communities; perhaps it is no longer used by rural households given the modern medical system. Alternatively, indigenous knowledge on this species may have declined despite a potentially important role in contemporary rural society. Another example is timber. Although 35 of the 47 timber tree species in homegardens were native, exotic species accounted for 72 percent of all timber tree individuals. Farmers might focus on exotic timber tree species because of promotion by external agencies, such as the World Bank or the Asian Development Bank, that encourage high-yielding varieties across rural Asia (in contrast, medicinal plants, which are not promoted by external agencies at the rate timber is, were largely native).

Five of six recorded red listed species were deliberately planted by the homegardeners, but the rate of integration for A. paniculata and C. guruba were substantially higher than others. Andrographis paniculata is an important herb for the treatment of liver and spleen complaints, colic, constipation, dysentery, diarrhea, dyspepsia, and phthisis. Calamus guruba is an excellent climber for making furniture, artifacts, and sport goods in household-scale industries. Although most red listed species were deliberately planted by the homegardens, the interviews revealed that those farmers were not aware of their critically threatened status and conservation potential; the farmer on whose land M. sylvatica naturally occurred was also unaware of the threatened status of this species.

The challenge of ex situ conservation of plants in homegardens can be tackled on three fronts. First, homegardeners need to be made aware of the status and rarity of the species they may have on their property; such awareness could result in localized efforts to conserve rare native species by promoting more widespread use. Awareness building campaigns, publications, and educational programs are methods to increase public support for using native species in homegardens (Trewhella et al. 2005). This is a viable proposition because all native species (and indeed, all species encountered) had livelihood uses, and therefore built-in incentives exist for participation. Thus, the ultimate goal is to encourage the use of native species when they provide a legitimate alternative to exotic species. Second, individuals of rare and critically endangered species, and especially those on the IUCN Red List, must be protected through a collaborative effort. The homegardeners on whose land those species reside should be contacted and brought into the conservation process itself. Whereas the initiative of the Bangladesh Forest Department and NGOs to promote native species use by homegardeners is vital for awareness and consensus building, local collaboration is essential for the conservation process, whether it is in managed or natural ecosystems (Steinmetz et al. 2006). In this case, farmers are the keystone actors in the short-term persistence of those species and their importance cannot be overemphasized. Finally, there must be major efforts to propagate species for distribution to homegardeners. The rarity of many of the species in this study is such that artificial propagation methods may be required. The low abundance and frequency values of more than half of all recorded species including native imply that there is limited probability of pollination or viable fruit set for those individuals. Intervention methods such as hand pollination could be crucial in maintaining genetic diversity and regeneration potential of those species.

CONCLUSION

Can homegardens conserve biodiversity in Bangladesh? Tropical homegardens can support the conservation of biodiversity (Coomes & Ban 2004, Kumar & Nair 2004, Schroth et al. 2004, Wiersum 2004, Mendez & Bacon 2005), but it requires, at a minimum, high species diversity in a region, sufficient numbers of individuals for genetic diversity, and successful natural reproduction. The results of this study suggest that indeed, species richness of homegardens in southwestern Bangladesh is high. However, the majority of species were rare, with the implication that natural propagation may not occur. Thus, we identified three key steps to make homegardens more important for long-term maintenance of biodiversity in Bangladesh: awareness building, protection of rare species, and propagation of rare species. Clear scope exists to develop homegardening systems as an important strategy to conserve biodiversity outside the natural and/or protected area systems in Bangladesh. As the process of plant domestication and crop evolution is ongoing, homegardens in Bangladesh may act as refuges for native and rare plants. This is of particular interest to conservationists within the country as well as internationally.

ACKNOWLEDGMENTS

M.E.K. was supported by a grant from the Danish International Development Agency and The Asian Institute of Technology Fellowship administered by the Asian Institute of Technology. D. Kamal provided the GARMIN® GPS. Forestry and Wood Technology Discipline, Khulna University, Bangladesh provided the SUUNTO® clinometer for plant's height measurement. S. Rahman arranged accommodation for M.E.K. during his field survey. Transportation during the field survey was arranged by H. Khatun and Md. Fazlul Karim (Kabir's beloved parents). Our sincere gratitude to those villagers in the study area who consented to provide information and helped in data collection, without which it would have not been possible to conduct this research. Field assistance and on site accommodation were provided by Yunus (Tala); Mithu, Moidul (Dumuria); Nazrul, Badsha, Mithu (Paikgacha); Rashed, Chonchol, Arif, Moon, (Jibonnagar); Palash (Boalmari); Tutul, Borhan (Chowgacha); Mintu (Keshabpur); Bablu, (Kaligonj); Mizan (Rampal). J. F. Maxwell and M. van de Bult provided technical assistance on species nomenclature, habit, and habitat classification. Study area map was prepared with the assistance from Mr. Thiha. The manuscript benefited from the comments of two anonymous reviewers.

Ancillary