Introduction

Nearly 80% of all new agricultural land in the tropics, including plantations, croplands, and grazing pastures, is currently established through the clearing of preexisting natural forest habitat (Foley et al. 2005; Gibbs et al. 2010; Lambin and Meyfroidt 2011). This extensive forest loss is predicted to increase as human populations, and therefore the demand for arable land, continues to rise rapidly on a global scale (Flohre et al. 2011; Foley et al. 2005; Godfray et al. 2010). This magnitude of habitat loss is a serious concern for many primate species, more than half of which are now threatened with extinction as a direct result of habitat loss (Cowlishaw 1999; Estrada and Coates-Estrada 1988; Johns and Skorupa 1987; Mittermeier et al. 2010). Although 90% of primate taxa worldwide are thought to be completely dependent on forest habitat (Chapman et al. 2006; Mittermeier and Cheney 1987), studies of Neotropical (Clarke et al. 2002; Estrada 2006), African (Cowlishaw 1999; Isabiryse-Basuta and Lwanga 2008), and Asian primates (Johns 1986; Srivastava 2006) suggest a degree of tolerance and adaptability to anthropogenic, agricultural landscapes. However, it is often unclear whether these agricultural landscapes and the introduced plant species in them can support permanent, stable populations of primates (Gérard et al. 2015; Irwin et al. 2010; Perfecto and Vandermeer 2008).

The lemurs of Madagascar inhabit a landscape under ever-increasing agricultural expansion and intensification (Keck et al. 1994). As a result of rapid and widespread decline of their forest habitat, lemurs have become the most imperilled group of mammals on Earth, with 94% of species now threatened with extinction (Schwitzer et al. 2014). Much of this habitat destruction has been caused by human population growth and an increased demand for arable land for the cultivation of food and cash crops (Irwin et al. 2010; Schwitzer et al. 2011). Tolerance of these degraded, anthropogenic habitats varies greatly among even the most closely related lemur species (Lehman et al. 2016; Setash et al. 2017). For example, in the nocturnal Cheirogaleidae, some species have more generalist adaptations to degraded habitat types, such as broad ecological feeding niches, varied diets, and the ability to use matrix habitats as migration corridors, in comparison to other more specialized species (Kappeler and Rasoloarison 2003; Rakotoniaina et al. 2016; Schäffler and Kappeler 2014). Large agricultural areas are now increasingly common throughout Madagascar’s landscape, much of which is dominated by highly anthropogenic grasslands (Vorontsova et al. 2016). Yet there have been surprisingly few studies of the use of these agricultural habitats by lemurs, and we do not know whether they are suitable/sink habitats for lemurs or merely potential travel corridors between natural forest fragments (Irwin et al. 2010). There are reports of lemurs in eucalyptus and exotic tree plantations and anecdotal reports of the black lemur (Eulemur macaco) and blue-eyed black lemur (E. flavifrons) in cash-crop plantations of mango, cashew, coconut, and pepper (Ganzhorn 1987; Petter et al. 1977; Ramanamanjato and Ganzhorn 2001; Schwitzer et al. 2007; Scobie et al. 2017; Simmen et al. 2007). Moreover, researchers have recently observed five species of lemur in cacao plantations in northwest Madagascar (A. Webber pers. comm.). Given the large and increasing expanse of agricultural habitats in Madagascar, there is a pressing need for additional and more in-depth studies to better understand the use of plantations by lemurs.

Bourbon vanilla (Vanilla planifolia) is Madagascar’s most well-known, trademark cash crop (AFB 2016). Approximately 80% of the world’s vanilla is cultivated and processed in northern Malagasy plantations, contributing nearly $200 million in export income (in 2015), making it the country’s second-most valuable export after raw nickel (AFB 2016). Vanilla plantations are now widespread in many areas of northern Madagascar, particularly in the Sava region of the northeast, and are the main source of income for many farmers in the region (Bomgardner 2016; Perfecto and Vandermeer 2008). Vanilla is an orchid that originates from Mexico, where it is pollinated by a few hymenopteran species (Lubinsky et al. 2006). In Madagascar, where these species are absent, pollination has to be done manually, making it a labor-intensive crop. Vanilla plants are grown on the trunks/stems of other types of vegetation for support (known as “tutor” plants), and the seed pods are harvested, dried, and used to flavor many products and foods. The crop is grown using a variety of cultivation methods, ranging from traditional small-scale plantations in and around existing forests (where the vanilla plants grow on preexisting vegetation that act as natural tutors), to large intensively farmed plantations grown on an industrial scale (where vanilla is grown among rows of a few specifically planted and highly modified, introduced tutor species). However, despite their wide-scale distribution, the biodiversity contained in these vanilla plantations remains understudied, and we do not know whether they harbor lemur populations, nor whether they have a role to play in the conservation of lemur populations (Ganzhorn 1987; Hending et al. 2017). In this study, we assessed lemur diversity in natural forest fragments and vanilla plantations of the Sava region, northeastern Madagascar. We compared lemur diversity among plantation types of varying management intensity, with the aim of determining whether lemurs were present within these different agricultural habitats. We hypothesized that some of the species of lemurs, particularly the nocturnal species, that are present within the natural forest fragments would be present in plantations in close proximity to natural habitat, but not in highly anthropogenic, intensively managed plantations. This is because some lemurs, such as the nocturnal species, are known to tolerate and inhabit degraded habitats (Kappeler and Rasoloarison 2003).

Methods

Site Selection

We made all observations between January 10 and April 25, 2017 in the Loky–Manambato Protected Area, the Vohemar district, and the Sambava district of the Sava region, north eastern Madagascar. We surveyed 17 cash-crop vanilla plantations, classified as either 1) ecoplantations, where the preexisting vegetation has not been removed, modified, or managed and vanilla is grown on preexisting vegetation tutors (N = 12) (Fig. 1), or 2) intensive/nontraditional plantations, where the preexisting vegetation has been heavily modified, managed, removed, or planted to make maximal use of space; vanilla is grown on modified or introduced tutors; and natural fertilizers are used to modify soil quality (N = 5) (Fig. 1) (Goldthorpe 1988). The term ecoplantation reflects growing the vanilla crop in a preexisting ecosystem (such as forest or other vegetation), rather than a judgment of its ecological merits or shortcomings. We subclassified ecoplantations based on their proximity to forest fragments as either 1) forest plantations: located ≤100 m of a forest fragment (N = 6), or 2) nonforest plantations: located >100 m from a forest fragment (N = 6); the 100-m threshold for determining whether an ecoplantation was “forest” or “nonforest” was an arbitrary distinction, but the vegetation of forest plantations was always closely connected with the vegetation of a natural forest fragment, whereas nonforest plantations were much further from the nearest forest fragment. We measured the area of each plantation site using GPS waypoints of the site perimeter and ArcGIS (ArcGIS, Esri, Redlands, CA, USA). We deemed all plantations studied to be stable in their position and agricultural intensity; ecoplantations are regarded as the traditional method of vanilla cultivation by members of the vanilla farming community, are family owned, and have frequently been cultivated over several generations; in contrast, intensive plantations tend to be managed by corporate or nongovernmental organizations under long-term programs. We also surveyed nine natural forest fragments in close proximity to the vanilla plantations (Fig. 2). Forest fragments comprised predominantly native floral species in mostly primary growth, with some secondary growth evident in places (Kirby et al. 1984).

Fig. 1
figure 1

Photographs showing the structure of ecoplantations and intensive plantations of the Sava region, studied January 10–April 25, 2017. (a) Structure of preexisting vegetation in an ecoplantation at Brondra. (b) Vanilla growing on preexisting vegetation at Brondra. (c) Rows of planted tutor trees supporting vanilla at the intensive plantation Floribis. (d) Rows of planted tutor trees supporting vanilla at an intensive plantation in Angalandrava.

Full size image
Fig. 2
figure 2

Locations of forest fragment sites and vanilla plantations sampled in the Sava region of northern Madagascar near Vohemar–Daraina (a) and Sambava (b), January 10–April 25, 2017.

Full size image

Transect Sampling

To assess the presence of lemurs, we used one or two parallel line transects (total range of transect lengths in vanilla plantations = 0.2–1.1 km, in forests = 0.7–1.05 km) established on preexisting trails within the study sites. Two researchers walked at a slow and constant pace (1 km/h) along each transect three times: at dawn (05:00 h), midday (12:00 h), and at night (21:00 h), noting any lemur species observed along the transect line. We used head torches (Tikka, Petzl, Crolles, France) and high-lumen flashlights (EC20, Nitecore, Guangzhou, China) to aid nocturnal lemur sightings and identification. We identified lemur species using field keys and published data on geographic ranges (Louis et al. 2008; Mittermeier et al. 2010). We distinguished between the sympatric, morphologically similar dwarf lemurs by their size, pelage color, and geographic range (Blanco and Godfrey 2013; Frasier et al. 2016; Mittermeier et al. 2010).

Data Analysis

We conducted all statistical analysis in IBM SPSS 23.0 (SPSS Inc., Chicago, IL, USA) and set an α level of 0.05 to test for statistical significance. We examined if plantation type (i.e., forest ecoplantation, nonforest ecoplantation, intensive/nontraditional) influenced whether any lemurs were observed (response = lemurs present or absent) using a Fisher’s exact test, and made explicit contrasts between plantation type pairs after adopting post hoc Holm–Bonferroni corrections. This approach is conservative, as it asks only whether lemurs (of any species) were present or absent in a plantation, and does not consider the number of species or individuals observed. Both plantation area and transect length varied between sites, and in some instances, transect lengths were constrained by plantation area. This may have led to differences in survey effort across different plantation types, introducing bias in the likelihood of observing lemurs. We therefore examined whether plantation areas and transect lengths varied systematically between plantation types using a Kruskal–Wallis test and a one-way ANOVA, respectively (in the former, the distribution of residuals was nonparametric and could not be appropriately transformed). Similarly, differences in sightings of lemurs between forest and vanilla plantations may have arisen because of a systematic bias in survey effort between the two contrasting habitat types. We tested for differences in survey effort (transect length) using a t-test.

Data Availability

The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.

Ethical Note

All research complied with UK Home Office policies when working with animals and all research adhered to the legal requirements of Madagascar. Research in the Loky–Manambato Protected Area was permitted by Madagascar’s Ministry of Environment (permit number 295/16/MEEF/SG/DGF/DSAP/SCB.Re).

Results

We observed five lemur species in vanilla plantations (Table I); the diurnal crowned lemur, Eulemur coronatus (N = 2 plantations); and four nocturnal species: the northern rufous mouse lemur, Microcebus tavaratra (N = 2 plantations), Sheth’s dwarf lemur, Cheirogaleus shethi (N = 2 plantations), Crossley’s dwarf lemur, C. crossleyi (N = 2 plantations), and the greater dwarf lemur, C. major (N = 2 plantations). We observed at least one species of lemur in 8 of the 17 sampled sites (47%; Table I). There was a significant difference in the presence of lemurs between plantation types (Fisher’s exact test: P = 0.006), with lemurs (of any species) significantly more likely to be found in forest ecoplantations compared to other plantation types (Fisher’s exact test post hoc comparisons following Holm–Bonferroni corrections; lemurs present in 6/6, 1/6, and 1/5 forest ecoplantations, nonforest ecoplantations, and intensive/nontraditional plantations, respectively; Table I). Transect length covaried with plantation area (r s  = 0.556, N = 17, P = 0.021). However, neither transect length nor plantation area differed significantly between plantation types (transect length, one-way ANOVA: F2,14 = 0.193, P = 0.827; plantation area, Kruskal–Wallis: χ2 = 0.518, df = 2, P = 0.772), suggesting that the greater incidence of lemurs in forest ecoplantations was not the result of biased survey effort in that plantation type.

Table I Location, areas, and total transect lengths walked in vanilla plantations of different types and natural forest fragments in the Sava region of northeast Madagascar, January 10–April 25, 2017
Full size table

We typically observed only one individual of each of the four nocturnal species at each site where present (except for Brondra; Table I). However, we observed 11 Eulemur coronatus individuals in one tree in Antingana, an ecoplantation located at the edge of a forest fragment. We also observed E. coronatus in the Ankalotany plantation, but not during formal transect walks. All individuals of all five species were foraging or resting when first sighted. We did not observe any lemurs using the vanilla plants themselves as a food source, or as a travel route.

We observed four lemur species that were present in vanilla plantations in the natural forest sites: Cheirogaleus shethi, C. crossleyi, Eulemur coronatus, Microcebus tavaratra. We also observed an additional four lemur species that were not present in any vanilla plantations: golden-crowned sifaka (Propithecus tattersalli), Sanford’s lemur (E. sanfordi), Daraina sportive lemur (Lepilemur milanoi), and the amber mountain fork-marked lemur (Phaner electromontis). We observed C. major only in plantations. Mean transect lengths in forests were significantly longer than those in plantations (forest transect length mean ± SE = 916 ± 40.8 m, plantation mean ± SE = 521 ± 58.8 m; two-tailed t-test: t = 4.57, df = 24, P < 0.001), so it is possible that our observations of additional lemur species in forests arose from greater sampling effort.

Discussion

Our observations provide evidence of the presence of lemurs in vanilla plantations. Cheirogaleid lemurs have previously been observed in exotic tree and eucalyptus plantations (Ganzhorn 1987; Ramanamanjato and Ganzhorn 2001; Scobie et al. 2017). Given this adaptability to anthropogenically disturbed habitats, and their tolerance for shrub layer and understory habitat (Kappeler and Rasoloarison 2003), it is perhaps unsurprising that we found the four nocturnal cheirogaleid species (Microcebus tavaratra, Cheirogaleus shethi, C. crossleyi, and C. major) in vanilla plantations. Most of our lemur observations were in vanilla plantations close to or within forest fragments (forest ecoplantations), suggesting that lemurs use these plantations for dispersal or as travel corridors between forest fragments, or as extensions of forest habitat. However, we also observed one C. major individual in a heavily degraded and disturbed intensive/nontraditional vanilla plantation (Biarafa; Table I), suggesting that cheirogaleid lemurs can use even the most anthropogenic plantations.

To sustain permanent populations of these four lemur species, plantations need to contain food sources such as fruiting trees, flowers, nectar, and insects (or large quantities of gum) to meet their dietary requirements (Atsalis 1999; Génin et al. 2010; Lahann 2007). In addition, suitable sleeping trees with adequate insulation properties would be needed to shelter these species during bouts of daily and seasonal torpor (Schmid 2000; Schülke and Ostner 2007). Although we observed many trees in our study sites that can provide both feeding and sleeping opportunities for these species, further research is needed to determine whether permanent populations of cheirogaleid lemurs can be sustained in vanilla plantations.

It is perhaps surprising that we witnessed numerous Eulemur coronatus individuals at two vanilla plantation sites. These individuals were feeding on fruit trees that are known to constitute part of their diet (Wilson et al. 1989), suggesting that flora found in vanilla plantations provide feeding opportunities for native lemur populations. These fruits occurred within the plantation sites and were not part of the vanilla farmer’s crop; thus the lemurs were not regarded as crop pests by the plantation owners. Further research is needed to fully determine vanilla plantation use by lemurs, and investigate how it varies between plantations with different management practices (Irwin et al. 2010; Schwitzer et al. 2011).

We observed an additional four lemur species in natural forest fragments that were not present in any vanilla plantation sites (Propithecus tattersalli, Eulemur sanfordi, Lepilemur milanoi, and Phaner electromontis). These additional species sightings may have arisen as a result of the greater sampling effort in forests compared to plantations. However, we feel that this explanation is unlikely, based on the biology of the different species observed. The apparent absence of P. tattersalli, E. sanfordi, and L. milanoi from vanilla plantations is likely because these species are much less tolerant of human-altered habitats, in comparison to the adaptable, more generalist cheirogaleids that we witnessed in vanilla plantations (Schwitzer et al. 2011; Vargas et al. 2002). Furthermore, Phaner are thought to require large trees with a high canopy for nocturnal travel and a diet preference of tree gum (Charles-Dominique and Petter 1980; Génin et al. 2010), suggesting they would avoid the understory habitats characteristic of the Sava region vanilla plantations where there are fewer potential feeding opportunities.

The remaining natural forest in the Sava region is diminished and highly fragmented. The reduced population sizes and geographical isolation that result from fragmentation may lead to higher genetic differentiation between lemur populations and lower levels of intrapopulation genetic variance for some species (Frankham et al. 2010; Holmes et al. 2013; Keller and Waller 2002). Although genetic diversity has been found to be high in some lemur species within our study area, such as Propithecus tattersalli (Quéméré et al. 2010), migration routes are needed for gene flow to continue (Quéméré et al. 2009). Small, isolated populations have a poor prognosis given the rate of deforestation in Madagascar (Schwitzer et al. 2014), especially given their susceptibility to deleterious stochastic events. Yet they could potentially be rescued if more means of gene flow (migration routes) between them were established. Our findings suggest that appropriately managed vanilla plantations, particularly those in close proximity to natural forest fragments, are suitable extensions of habitat for some lemur species, providing feeding opportunities and marginal travel corridors. Vanilla plantations therefore represent potentially valuable habitat extensions for some lemur species, through which gene flow can occur between forest fragments to improve population connectivity (Baden et al. 2014; Quéméré et al. 2009). This information is important for future assessments of lemur habitat use and population densities and distributions in areas of widespread vanilla cultivation, such as in the Sava region (AFB 2016).

Vanilla is a valuable crop, generating income for more than 200,000 people in Madagascar (AFB, 2016), most of whom live in the Sava region. Given its financial worth, vanilla has provided, and is likely to continue to provide, sufficient incentives for local people to establish, protect, and maintain plantations. Although natural forests are the best ecosystem for maintaining species biodiversity, our results suggest that forest ecoplantations have the highest conservation value of the three studied plantation types (in addition to their high financial worth), and are the most encouraging method of vanilla cultivation from both sustainable agricultural livelihood and lemur conservation perspectives. If vanilla plantations are established at the local (e.g., village or homestead) level then the payoff for exploiting forest via, for example, tavy (slash and burn agriculture), will likely be shifted such that the forest, and the vanilla it contains, is more valuable than converted land, thereby reducing the threat of habitat loss. Many of the vanilla farmers also employ guards to deter theft of vanilla plants and pods (S. Cotton et al. unpubl. report); this may inadvertently reduce the likelihood of hunting in plantations by people from elsewhere, protecting lemurs and other fauna. The existing network of vanilla plantations could therefore be further expanded to provide additional habitat for some species of lemurs, a winning combination in terms of both species conservation and sustainable agroforestry for local farming communities.