Seed Dispersal Syndromes in the Madagascan Flora: the Unusual Importance of Primates

Seed dispersal syndromes in the Madagascan flora: the unusual importance of primates

A URÉLIE A LBERT-DAVIAUD,SARAH P ERILLO and W OLFGANG S TUPPY

Abstract Madagascar is one of the most threatened bio- Dewar & Richard, ; Ali & Krause, ), sympatric spe- diversity hotspots, and protection of its biodiversity is be- ciation (Buerki et al., ) and relatively recent climate coming increasingly urgent as deforestation of the island changes (Yoder & Nowak, ; Buerki et al., ; continues. For the long-term success of conservation efforts Mercier & Wilmé, ). Levels of endemism are as high it is essential that key ecological processes, such as seed dis- as % in vascular plants (Callmander et al., ) and persal, are protected and restored. Therefore, the identifica- % in vertebrates (Goodman & Benstead, ). With on- tion of ecological gaps is a vital task. For Madagascar, only going deforestation, Madagascar’s unique biodiversity is little is known about plant–animal interactions, and trad- highly threatened (Brooks et al., ). itional methods of ecological research are too time-consum- Deforestation inevitably causes defaunation. Both pro- ing to provide crucial information about breakdowns in cesses result in imbalanced, disturbed or even dysfunctional these interactions. To identify likely dispersal gaps we there- ecosystems. A critical consequence, which has so far re- fore used a theoretical approach to analyse plant–disperser ceived little attention from conservationists, is the break- interactions in Madagascar. We used data science tools to down of key plant–animal interactions, such as seed impute missing data on relevant plant traits to subsequently dispersal. The extinction of animal dispersers has cata- predict the most likely dispersal agents for each of strophic consequences for the long-term survival of animal- Madagascar’s endemic plant species. We found that % dispersed plant species (Howe & Smallwood, ; Herrera, of the endemic species (N = ,) are endozoochorous, ; Caughlin et al., ). Plant regeneration is limited in and among these –% display a primate syndrome and the absence of seed dispersers, resulting in an increasing ag- –% a bird syndrome (depending on the definition of syn- gregation of individuals and a strongly increasing probabil- dromes). This lower percentage of endozoochorous species ity of extinction (Terborgh et al., ; Caughlin et al., ). and higher percentage of species with a primate syndrome Long-term conservation can therefore be successful only if in Madagascar compared to other tropical areas reflects the gaps in dispersal mutualisms are identified and repaired. unusual disperser guild on the island. Only five bird species For Madagascar, knowledge about plant species and but  lemur species are frugivorous, and  of those lemur plant–animal interactions is still scarce, and deforestation species are currently threatened with extinction. The dis- and defaunation are progressing rapidly. Traditional meth- appearance of frugivorous lemurs would significantly change ods of studying ecological processes such as seed dispersal the vegetation dynamics of Madagascar’s ecosystems, and a (e.g. long-term field observations) are too time-consuming high proportion of Madagascar’s endemic plants would to provide comprehensive information about the breakdown enter an extinction vortex. of vital plant–animal interactions. Dispersal syndromes (i.e. sets of fruit or seed traits) provide a practical concept for pre- Keywords Bird, dispersal spectra, endozoochory, Madagascar, dicting plant–animal interactions and interdependencies in a ornithochory, primate, primatochory, seed dispersal community quickly and accurately (van der Pijl, ;Howe, Supplementary material for this article is available at ; Kuhlmann & Ribeiro, ). In the absence of compre- https://doi.org/./S hensive data to infer dispersal syndromes in Madagascan ecosystems, we used data science tools to impute missing data on relevant plant traits. Using this approach we were able to assign dispersal syndromes to Madagascar’sendemic Introduction plant species and to infer vital plant–animal relationships adagascar is one of the most threatened biodiversity and interdependencies. Mhotspots (Brooks et al., ). Most of the island’s Of the various dispersal strategies plants have evolved, plant and animal species are endemic as a result of its endozoochory is usually the most common in tropical and long isolation from other continents (c.  million years; subtropical rainforests, where at least % but more often –% of tree species bear fleshy fruits (Howe & Smallwood, ; Jordano, ; Willson & Traveset, AURÉLIE ALBERT-DAVIAUD (Corresponding author), SARAH PERILLO and WOLFGANG ; Osuri et al., ). Diaspores of endozoochorous spe- STUPPY Royal Botanic Gardens, Kew, Wakehurst Place, Ardingly, West Sussex RH17 6TN, UK. E-mail [email protected] cies generally offer an edible reward to attract animal disper- Received  July . Revision requested  September . sers (e.g. fruit pulp, elaiosomes or arils). The predominance Accepted  October . First published online  February . of a syndrome is not always proportional to the abundance

Oryx, 2018, 52(3), 418–426 © 2018 Fauna & Flora International doi:10.1017/S0030605317001600 Downloaded from https://www.cambridge.org/core. IP address: 170.106.33.42, on 01 Oct 2021 at 14:25:38, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605317001600 Seed dispersal in the Madagascan flora 419

of the corresponding dispersal agent in a community TABLE 1 Extant frugivores in Madagascar. Only species for which (Hughes et al., ; Herrera, ); however, Madagascar fruits constitute the largest component of their diet during most is unique in that its extant frugivorous guild is strikingly of the year are included; granivores are excluded. Data from   small compared to other tropical areas (Goodman & Goodman & Benstead ( ), IUCN ( ) and Mittermeier et al. (). Ganzhorn, ; Ganzhorn et al., ). Only  lemur, three bat, five bird and one tortoise species on the island Species Family Conservation status are mainly frugivorous (excluding granivores) (Table ). Bats Therefore, an analysis of the frequency of the various seed Eidolon dupreanum Pteropodidae Vulnerable dispersal strategies present in Madagascar’s endemic plant Pteropus rufus Pteropodidae Vulnerable species (i.e. seed dispersal spectra; Hughes et al., ) Rousettus Pteropodidae Near Threatened madagascariensis would be informative. The proportion of frugivorous species Birds in Madagascar has always been small, even before the  Alectroenas Columbidae Least Concern Holocene extinctions (Godfrey et al., ). Hypotheses to madagascariensis explain the unusually low number of frugivores include the Hypsipetes Pycnonotidae Least Concern unpredictable pattern of fruit production on the island madagascariensis (Goodman & Ganzhorn, ; Wright, ) and the Philepitta castanea Philepittidae Least Concern lower absolute abundance of fruiting trees at any given Philepitta schlegeli Philepittidae Near Threatened Treron australis Columbidae Least Concern time compared to other tropical areas (Federman et al., Primates ). The most unusual aspect of Madagascar’s disperser Cheirogaleus major Cheirogaleidae Data Deficient guild is that there are only five species of frugivorous birds Cheirogaleus medius Cheirogaleidae Least Concern on the island, whereas in other tropical areas birds are the Eulemur albifrons Lemuridae Endangered most important and diverse group of frugivores (Fleming Eulemur cinereiceps Lemuridae Critically et al., ). Endangered Our aim was to answer the following questions: ()Is Eulemur collaris Lemuridae Endangered the small size of Madagascar’s disperser guild reflected in Eulemur coronatus Lemuridae Endangered a lower than average proportion of endozoochorous plant Eulemur flavifrons Lemuridae Critically  Endangered species compared to other tropical areas? ( ) Is the low Eulemur fulvus Lemuridae Near Threatened ’ proportion of birds among Madagascar s seed dispersers Eulemur macaco Lemuridae Vulnerable reflected in a lower than average proportion of species Eulemur mongoz Lemuridae Critically with a bird dispersal syndrome compared to other tropical Endangered areas? () Is there a difference in the frequencies of Eulemur rubriventer Lemuridae Vulnerable dispersal syndromes between Madagascar’s various Eulemur rufifrons Lemuridae Near Threatened bioclimates and vegetation formations? Generally, forests Eulemur rufus Lemuridae Vulnerable Eulemur sanfordi Lemuridae Endangered and humid areas harbour a higher proportion of Lemur catta Lemuridae Endangered endozoochorous species than open vegetation (e.g. grass- Microcebus murinus Cheirogaleidae Least Concern lands) and dry areas (Howe & Smallwood, ; Willson Microcebus myoxinus Cheirogaleidae Vulnerable et al., ). Microcebus rufus Cheirogaleidae Vulnerable Varecia rubra Lemuridae Critically Endangered Varecia variegata Lemuridae Critically Methods Endangered Reptiles Data collection Pyxis planicauda Testudinidae Critically Endangered Trait data of endemic Madagascan plant species were com- piled from the literature, public databases (Madagascar Catalogue, ) and examination of herbarium specimens at the Royal Botanic Gardens, Kew (, specimens) and at epicarp (thin: ,  mm vs thick: $  mm), pulp type (i.e. fle- the Muséum National d’Histoire Naturelle, Paris ( speci- shy, non-fleshy (e.g. dry, spongy, sticky, fibrous, farinose) and mens). Traits recorded included life form (tree, shrub, herb, aril), fruit length and width, seed length and width, biocli- liana), fruit type (indehiscent, including achene, camara, mate (humid, subhumid, subarid, dry, montane; Cornet, caryopsis, lomentum, utricle, samara; dehiscent, including ) and vegetation formation (forest/woodland, bush- capsule, follicle, coccum, legume, follicarium; and externally land/thicket, shrubland, grassland/wooded grassland, man- fleshy, including berry, drupe, epispermatium, syncarpium, grove, anthropic). We coded diaspore colours as follows: syconium; Spjut, ), diaspore colour, thickness of fruit black (including dark-bluish and reddish black), blue, purple,

brown (including dull dark red), red (including scarlet and TABLE 2 Percentage of missing data per trait, before imputation, for pink), green, yellow, orange, grey and white. the Madagascan flora. Trait % missing data Imputation of missing data Genus 0 Family 0 ’ Knowledge about Madagascar s dispersal ecology is still in- Life form 2.46 complete. Our dataset of plant traits relevant for the identi- Fruit type 3.36 fication of dispersal syndromes contained many gaps (% Pulp type 6.09 missing data). Given the random occurrence of missing va- Bioclimate 17.26 lues across the dataset, a complete case analysis in which all Vegetation formation 24.72 observations with one or more missing values were omitted Thickness of fruit epicarp 42.75 Fruit length 49.64 would include only  species (i.e. only % of the original   Seed length 61.78 , species). This would lead to non-negligible loss of Fruit width 62.27 power, with misleading results and biased estimates of para- Seed width 76.76 meters (Ambler et al., ; Giorgi et al., ). We there- Diaspore colour 80.85 fore decided to perform multiple imputations of missing values (Rubin, ). The percentage of missing data per trait was –.%(Table ), and the percentage of missing TABLE 3 Definition of seed dispersal syndromes. data per species was –% (mean = . ± SD .%). Syndrome Definition Firstly, we removed the four species (Euphorbia ambatome- Endozoochory Diaspores with pulp or aril (e.g. Garcinia spp.), or nahensis, E. beuginii, E. nicaisei and Venana cauliflora) for  without pulp/aril but with mimetic seeds* (e.g. which we had no data at all (i.e. % missing values). We Erythrina spp., Adenanthera mantaroa, Abrus then used multivariate imputations by chained equations aureus, Abrus diversifoliolatus and Abrus (van Buuren & Groothuis-Oudshoorn, ), assuming a parvifolius) missing at random mechanism (Penone et al., ). Primate Diaspores with green, brown or white colour (e.g. Multivariate imputations by chained equations have proved Adansonia spp.) to perform well for continuous (Penone et al., ) and cat- Bird Diaspores with thin epicarp or without epicarp egorical data (Giorgi et al., ; Akande, ), even for (e.g. seeds), and contrasting purple or blue colour (e.g. Commiphora spp.) high-dimensional data (Deng et al., ). We used the pre- dictive mean matching for continuous data to preserve non- *Seeds that deceptively suggest the presence of a fleshy edible reward (i.e.  linear relationships (van Buuren & Groothuis-Oudshoorn, fruit pulp or aril) (van der Pijl, ). ), the logistic regression for binary variables, the multi- nomial logistic regression for categorical variables with more Definition of dispersal syndromes than two factors, and the proportional odds model for or- dered categorical data. Fruit length and width, and seed We defined endozoochorous and non-endozoochorous spe- length and width were log-transformed to produce a normal cies based on the presence of edible rewards, such as fruit distribution. To generate each imputed dataset, we repeated pulp or aril (Table ) (van der Pijl, ). Subsequently, the algorithm with  iterations to obtain stable predictions where possible, we assigned either the primate dispersal syn- for the missing data (van Buuren & Groothuis-Oudshoorn, drome or the bird dispersal syndrome to each endozoochor- ). We generated  imputed datasets to capture most of ous species (Table ; Plate ). Using a dual approach we the uncertainty underlying the imputation process (Graham distinguished between these dispersal syndromes by applying et al., ; Bodner, ; White et al., ). Given the mod- two alternative definitions based on ()diasporecolouronly, erate rate of missing data (i.e. %) in the original dataset and and () diaspore colour and fruit epicarp thickness. We used the large number of observations (, species), we kept all these two definitions in parallel because they are both com- species with at least one data entry for the multivariate im- monly used in the literature (Voigt et al., ). Diaspore col- putations by chained equations (Hardt et al., ). The algo- our has proved to be an important and reliable selection rithm converged for all traits (Supplementary Fig. S), even factor for primates and birds (Lomáscolo & Schaefer, ). when the rate of missing values was high (e.g. diaspore col- However, as there is no widely accepted definition of colours our). We analysed each imputed dataset individually and characterizing the bird and primate dispersal syndromes, we identically, and pooled the estimates from each dataset to ob- inferred our own definitions from published seed dispersal tain the final result (Rubin, ). We used the packages mice studies based in Madagascar (Supplementary Table S). We and miceadds in Rv... (R Development Core Team, ). checked that birds and primates consumed diaspores with  The corresponding scripts are available from AAD upon different colours by performing a χ test of independence, fol- request. lowed by post-hoc pairwise tests applying a Bonferroni

PLATE 1 Plant species displaying a bird syndrome: (a) Erythrina madagascariensis and (b) Commiphora sp.; and a primate syndrome: (c) Adansonia rubrostipa and (d) Adansonia suarenzensis (showing the inside of the fruit). Photographs by David Du Puy (a) and Wolfgang Stuppy (b, c, & d).

 correction (χ = .,d.f.=,P, .; post-hoc test: plant species to be primarily but not exclusively dispersed bird vs primate: P , .). We used the function by a certain agent, their primary disperser. Some plant species pairwiseNominalIndependence in the package rcompanion may have lost their primary dispersers (e.g. giant lemurs) but (Mangiafico, ). We then performed a multiple corres- they continue to display their dispersal syndromes (e.g. pri- pondence analysis to find the colours most frequently asso- mate syndrome), a circumstance referred to as dispersal an- ciated with each disperser type. We used the function MCA achronism (Janzen & Martin, ; Guimarães et al., ; in the package FactoMineR. We found that birds consumed Andriantsaralaza et al., ). predominantly diaspores with contrasting blue or purple colours, whereas primates consumed mainly brown, green or  white diaspores (Supplementary Fig. S ). Therefore we used Data analysis these colours to define the respective dispersal syndromes (Table ). Fruit epicarp thickness has also often been used We analysed the variation of traits between endozoochorous in the definition of syndromes (Janson, ;Gautier-Hion and non-endozoochorous species, and between species dis- et al., ): primates can eat fruits with thin as well as playing a primate and a bird syndrome (following the defi- thick epicarps, whereas birds cannot eat fruits with thick epi- nitions above). We fitted models using generalized least carps. Seed dispersal syndromes reflect the adaptation of a squares and maximizing the log-likelihood, with traits as

dependent variables and syndrome as an independent vari- able, and compared them to null models using the likelihood-ratio test. We used the function anova combined with the function gls in the nlme package (Pinheiro et al., ). We performed all analyses in Rv.... Data on traits of endemic Madagascan plant species before and after imputation of missing values are available from the Dryad Digital Repository: http://dx.doi.org/./dryad.dt.

Results

Given estimations from the imputed datasets, all life forms are well represented within the Madagascan endemic flora, with shrubs dominating (%), followed by trees (%), herbs (%) and lianas (%). Fruits are mainly dehiscent (%) or externally fleshy (%), with indehiscent fruits ac- counting for only % of the flora. The distribution of Madagascar’s endemic plant species across the various bioclimates is as follows: % of species occur in sub-humid, % in humid, % in dry and % in sub-arid areas. Seventy-five percent of the island’s endemic species occur in closed vegetation (i.e. forest, woodland, bushland or thicket). Almost half of the species (%) occur only in forests, and % occur in two or more vegetation formations. Of the endemic Madagascan flora (N = , species), %of species are endozoochorous. When bird and primate dispersal syndromes are defined based on diaspore colour only, % of the endozoochorous species qualify for primate dis- FIG. 1 Percentage of endemic Madagascan plant species persal, % qualify for bird dispersal and % are ambigu- (N = ,) producing each fruit type, comparing (a) ous (i.e. although they are endozoochorous, they possess non-endozoochorous and endozoochorous syndromes, and (b) fruits that are not specifically adapted to be dispersed by bird and primate syndromes (with syndromes defined according to diaspore colour only). either primates or birds). Focusing on trees of humid forests, the distribution of endozoochorous species shows the same pattern (% primate syndrome vs % bird syndrome). adaptation to endozoochory. Of these, % are trees. When we add fruit epicarp thickness to the syndrome defi- Focusing on humid forest tree species, the percentage of en- nitions, the distribution pattern is even more biased towards dozoochorous species is %, which is low compared to species with a primate syndrome: % of the endozoochor- other tropical areas. In most tropical and subtropical ous plant species display a primate syndrome, % display a moist broadleaf forests the percentage of animal-dispersed bird syndrome and % are ambiguous. Endozoochorous tree species is –% (Osuri et al., ). The only excep- and non-endozoochorous species differ significantly in tion is in South-east Asia, where some forest tree communi- terms of life form, fruit type (Fig. a), bioclimate and vege- ties comprise . % of abiotically dispersed species. This tation formations (Table ). Species with a bird or primate result could be explained by the low number of frugivores syndrome differ significantly in terms of fruit type (Fig. b), in Madagascar (only  vertebrate species) compared to regardless of the definition of syndromes used for the ana- other tropical areas, where .  vertebrate frugivorous lyses. They also differ with a very low level of significance in species may occur (Fleming et al., ). Another explan- terms of bioclimate when diaspore colour only is taken into ation for the high proportion of non-endozoochorous spe- account (Table ). When fruit epicarp thickness is also cies in Madagascar could be the presence of vast areas of included in the definition of syndromes the difference in grasslands where they are predominant. However, the ques- bioclimate is no longer significant. tion as to whether Madagascar’s extensive grasslands are natural or the result of human interference has yet to be re-   Discussion solved (Bond et al., ; Crowley & Samonds, ; Vorontsova et al., ); an anthropogenic and therefore re- Our analyses indicated that, overall, slightly more than a cent origin of grasslands would make this last explanation third of Madagascar’s endemic plant species display irrelevant.

TABLE 4 Characteristics linked to endozoochorous and non-endozoochorous endemic Madagascan plant species.

Endozoochorous Non-endozoochorous F(df1, df2) P

Life form Tree Herb, liana 2,138.71(1, 1478.43) 0 Fruit type Externally fleshy Dehiscent, indehiscent 441.71(1,646.1) 0 Bioclimate Humid 290.11(1,338.34) 0 Subhumid 162.38(1,827.46) 0 Subarid 48.29(1,493.33) 0 Dry 7.65(1,216.57) 0.006 Montane 148.71(1,1030.62) 0 Vegetation formation Forest 191.81(1,39.48) 0 Forest only 404.70(1,120.38) 0 Closed vegetation 204.21(1,49.54) 0 Bushland 30.05(1,222.57) 0 Shrubland 107.67(1,242.25) 0 Grassland 103.45(1,494.51) 0 , Anthropic 20.83(1,472.68) 0.0001 Main families Rubiaceae (541 species) Orchidaceae (751 species) Arecaceae (188 species) Acanthaceae (467 species) Melastomataceae (171 species) Asteraceae (438 species)

TABLE 5 Characteristics linked to endemic Madagascan plant species with a bird or primate syndrome, when syndromes are defined according to diaspore colour only.

Bird Primate F(df1, df2) P , Fruit type Dehiscent Externally fleshy 29.09(1,25.96) 0.0001 Bioclimate Dry 4.37(1,82.85) 0.04 Main families Rubiaceae (85 species) Rubiaceae (88 species) Araliaceae (56 species) Pandanaceae (69 species) Melastomataceae (40 species) Malvaceae (57 species)

By applying the concept of dispersal syndromes to iden- factors (Herrera, ). In particular, it would be interesting tify plant–animal interactions we were able to confirm what to compare the biomass of frugivorous birds and lemurs. has previously been suggested by several studies of Moreover, it is important to note that we did not take Madagascan ecosystems, namely the unusually strong reli- fruit bats into account. Although fruit bats are usually ance on primates as seed dispersers compared to other trop- important dispersers in tropical areas (Muscarella & ical areas (Bleher & Böhning-Gaese, ; Bollen et al., ; Fleming, ), including Madagascar (Bollen & Elsacker, Voigt et al., ). Our results are in line with those of Voigt ; Picot et al., ), we found a strong correlation be- et al. (), who found that the proportion of plant species tween the proportion of each colour in the endozoochorous with a primate syndrome was higher in Madagascar than in flora and the proportion of colours in species eaten by bats South Africa. In other words, in Madagascar primate disper- (r = .), which suggests that bats do not seem to select sal plays a greater role than in other tropical areas, where the fruits according to their colour. It is likely that fruit bats ar- majority of plants have a bird syndrome. This reflects the rived by dispersal on the island much later than lemurs, en- much higher proportion of primates compared to birds in countering plants that were probably already largely the Madagascan frugivore assemblage (Bleher & Böhning- adapted to dispersal by lemurs (Hutcheon, ). Gaese, ; Bollen et al., ). Although Hughes et al. Typically we found the following associations between () stated that ‘the availability of specific dispersal vec- fruit type, life form and vegetation formation: fleshy fruits tors seems rarely to be an important determinant of disper- (endozoochorous) were associated with trees and forests, sal mode’, Madagascar could be the exception, given its and dry fruits (non-endozoochorous) were associated with strikingly small frugivore assemblage and the unusually herbs and grasslands (Armesto & Rozzi, ; Willson et al., strong importance of primates as dispersers. More studies ; Herrera, ; Bolmgren & Eriksson, ; Lorts et al., are necessary to confirm that the dispersal spectra we ). Moreover, endozoochorous (i.e. fleshy) fruits were observed in Madagascar are linked to the proportion of dis- more strongly associated with humid areas than dry fruits. persers and not only to the presence of particular abiotic As other studies have shown (Willson et al., ; Bunney,

), these results follow the metabolic costs hypothesis and many of these species have already disappeared from proposed by Willson et al. (), which assumes that some regions (Godfrey et al., ). The complete dis- plant species require some critical amount of water to pro- appearance of lemurs would change the natural regener- duce fleshy fruits. Another explanation could be the lower ation cycle and recruitment of primate-dispersed plant proportion of arboreal frugivores in open areas. species (Jordano et al., ) and increase the risk of extinc- Plant species with a primate syndrome more often tion of these plants (Caughlin et al., ). Moreover, the produce externally fleshy fruits, whereas species with a fact that .% of the endozoochorous plant species were bird syndrome more often produce dehiscent fruits found to occur only in forests highlights the urgent need (Table ). Arillate seeds are, almost as a logical consequence, not only to prevent deforestation in Madagascar but also associated with dehiscent fruits, as an aril has to be exposed to promote forest restoration. (visually or olfactorily) to attract animals for dispersal. Moreover, arillate seeds are often associated with ornithoch- Acknowledgements ory (van der Pijl, ). Therefore, it is unsurprising to observe that a high proportion of Madagascar’s species We thank Kathy Willis for her support and encouragement; with a bird syndrome produce dehiscent fruits. Gwilym Lewis for his support and advice on all matters re- We found that in Madagascar, similarly to other tropical lated to legumes; Stuart Cable, William Baker, Wolf areas (Turner, ), plants with a bird syndrome belong Eiserhardt, Hélène Ralimanana, Franck Rakotonasolo, mainly to the families Rubiaceae and Melastomataceae. Guy Onjalalaina and Romer Rabarijaona for sharing their However, plants with a primate syndrome belong mainly knowledge about Madagascar and its flora; Ian Willey, to the Rubiaceae, Pandanaceae and Malvaceae, whereas in Janet Terry, Patricia Wood, Aurelie Grall, Charlotte other tropical areas, primate dispersal occurs mostly in Couch, Nina Davies and Joanna Osborne for their assistance members of the Sapindaceae, Meliaceae and Euphorbiaceae with the Millennium Seed Bank and herbarium collections (Turner, ). It would have been interesting to test whether of the Royal Botanic Gardens, Kew; the Muséum National the distribution of dispersal syndromes across plant families d’Histoire Naturelle for permission to work on their collec- in Madagascar is phylogenetically biased; for example, a tions, and particularly Peter Lowry and Thomas phylogenetic signal in dispersal syndromes has already Haevermans; and Martin Fisher and two anonymous re- been found in the plant species of Ranomafana National viewers for their helpful comments. This study was part- Park (Razafindratsima, ). However, a comprehensive funded by The Joseph Jones and Daisy and Graham molecular phylogeny for the entire Madagascan flora is not Rattenbury Charitable Trust. yet available and the most comprehensive phylogenetic tree  published (reconstructed by Hinchliff et al., ) includes Author contributions only , of the , species analysed in this study. AAD and WS designed the research; AAD and SP collected Conclusion data; AAD conducted the research and analysed the data; and all authors wrote the article. Our results confirm that Madagascar’s flora is unique with respect to the unusual dispersal spectra expressed among References endemic plants. Seed dispersal syndromes are a useful tool for predicting plant–animal interactions and interdepend- AKANDE, O.M. () A comparison of multiple imputation methods encies. However, they must be applied cautiously in any for categorical data. MSc thesis. Duke University, Durham, USA. ALI, J.R. & KRAUSE, D.W. () Late Cretaceous bioconnections attempts to conserve and restore seed-dispersal processes  between Indo-Madagascar and Antarctica: refutation of the (Howe, ). Although syndromes probably reflect specific Gunnerus Ridge causeway hypothesis. Journal of Biogeography, , interactions with dispersers, the frugivore assemblage and –. the effectiveness of each disperser species may have changed AMBLER, G., OMAR, R.Z. & ROYSTON,P.() A comparison of over time and across regions (Howe, ). Although we do imputation techniques for handling missing predictor values in a not conclude or suggest that all lemurs provide equally risk model with a binary outcome. Statistical Methods in Medical Research, , –. effective seed dispersal to plant species with a primate syn- ANDRIANTSARALAZA, S., PEDRONO, M., TASSIN, J., EDMOND,R.& drome, it is clear that the protection of lemurs in DANTHU,P.() Baobabs de Madagascar: un anachronisme de la Madagascar is of primary importance for the maintenance dispersion? Bois et Forêts des Tropiques, , –. of plant communities. During the last , years  frugiv- ARMESTO, J.J. & ROZZI,R.() Seed dispersal syndromes in the rain orous lemur species have become extinct (Burney et al., forest of Chiloe: evidence for the importance of biotic dispersal in a temperate rain forest. Journal of Biogeography, , –. ; Crowley, ). Sixteen of the  extant frugivorous BLEHER,B.&BÖHNING-GAESE,K.() Consequences of frugivore lemurs are threatened with extinction (i.e. Vulnerable, diversity for seed dispersal, seedling establishment and the spatial Endangered or Critically Endangered sensu IUCN, ) pattern of seedlings and trees. Oecologia, , –.

BODNER, T.E. () What improves with increased missing data choice and seed dispersal in a tropical forest vertebrate community. imputations? Structural Equation Modeling, , –. Oecologia, , –. BOLLEN,A.&ELSACKER, L.V. () Feeding ecology of Pteropus GIORGI, R., BELOT, A., GAUDART, J., LAUNOY,G.&THE FRENCH rufus (Pteropodidae) in the littoral forest of Sainte Luce, SE NETWORK OF CANCER REGISTRIES FRANCIM () The Madagascar. Acta Chiropterologica, , –. performance of multiple imputation for missing covariate data BOLLEN, A., ELSACKER, L.V. & GANZHORN, J.U. () Tree dispersal within the context of regression relative survival analysis. Statistics strategies in the littoral forest of Sainte Luce (SE-Madagascar). in Medicine, , –. Oecologia, , –. GODFREY, L.R., JUNGERS, W.L., SCHWARTZ, G.T. & IRWIN, M.T. BOLMGREN,K.&ERIKSSON,O.() Fleshy fruits – origins, niche () Ghosts and orphans: Madagascar’s vanishing ecosystems. In shifts, and diversification. Oikos, , –. Elwyn Simons: A Search for Origins (eds J.G. Fleagle & C.C. Gilbert), BOND, W.J., SILANDER,JR, J.A., RANAIVONASY,J.&RATSIRARSON,J. pp. –. Springer, New York, USA. () The antiquity of Madagascar’s grasslands and the rise of C GODFREY, L.R., JUNGERS, W.L., SIMONS, E.L., CHATRATH, P.S. & grassy biomes. Journal of Biogeography, , –. RAKOTOSAMIMANANA,B.() Past and present distributions of BROOKS, T.M., MITTERMEIER, R.A., DA FONSECA, G.A.B., GERLACH, lemurs in Madagascar. In New Directions in Lemur Studies (eds J., HOFFMANN, M., LAMOREUX, J.F. et al. () Global biodiversity B. Rakotosamimanana, H. Rasamimanana, J.U. Ganzhorn & S. conservation priorities. Science, , –. M. Goodman), pp. –. Kluwer Academic/Plenum Publishers, BUERKI, S., DEVEY, D.S., CALLMANDER, M.W., PHILLIPSON, P.B. & New York, USA. FOREST,F.() Spatio-temporal history of the endemic genera GOODMAN, S.M. & BENSTEAD, J.P. () The Natural History of of Madagascar. Botanical Journal of the Linnean Society, , Madagascar. The University of Chicago Press, Chicago, USA, and –. London, UK. BUNNEY,K.() Seed dispersal in South African trees, with a focus on GOODMAN, S.M. & BENSTEAD, J.P. () Updated estimates of biotic the megafaunal fruit and their dispersal agents. MSc thesis. diversity and endemism for Madagascar. Oryx, , –. University of Cape Town, South Africa. GOODMAN, S.M. & GANZHORN, J.U. () Rarity of figs (Ficus)on BURNEY, D.A., BURNEY, L.P., GODFREY, L.R., JUNGERS, W.L., Madagascar and its relationship to a depauperate frugivore GOODMAN, S.M., WRIGHT, H.T. & JULL, A.J.T. ()A community. Revue d’Ecologie (Terre et Vie), , –. chronology for late prehistoric Madagascar. Journal of Human GRAHAM, J.W., OLCHOWSKI, A.E. & GILREATH, T.D. () How Evolution, , –. many imputations are really needed? Some practical clarifications of CALLMANDER, M.W., PHILLIPSON, P.B., SCHATZ, G.E., multiple imputation theory. Prevention Science, , –. ANDRIAMBOLOLONERA, S., RABARIMANARIVO, M., GUIMARÃES,JR, P.R., GALETTI,M.&JORDANO,P.() Seed RAKOTONIRINA, N. et al. () The endemic and non-endemic dispersal anachronisms: rethinking the fruits extinct megafauna ate. vascular flora of Madagascar updated. Plant Ecology and Evolution, PLoS ONE, (), e. , –. HARDT, J., HERKE, M., BRIAN,T.&LAUBACH,W.() Multiple CAUGHLIN, T.T., FERGUSON, J.M., LICHSTEIN, J.W., ZUIDEMA, P.A., imputation of missing data: a simulation study on a binary response. BUNYAVEJCHEWIN,S.&LEVEY, D.J. () Loss of animal seed Open Journal of Statistics, , –. dispersal increases extinction risk in a tropical tree species due to HERRERA, C.M. () Seed dispersal by vertebrates. In Plant−Animal pervasive negative density dependence across life stages. Proceedings Interactions: An Evolutionary Approach (eds C.M. Herrera & of the Royal Society of London B, , . O. Pellmyr), pp. –. Blackwell Science, Oxford, UK. CORNET,A.() Carte bioclimatique de Madagascar. Office de la HINCHLIFF, C.E., SMITH, S.A., ALLMAN, J.F., BURLEIGH, J.G., recherche scientifique et technique outre mer. CHAUDHARY, R., COGHILL, L.M. et al. () Synthesis of phylogeny CROWLEY, B.E. () A refined chronology of prehistoric Madagascar and taxonomy into a comprehensive tree of life. Proceedings of the and the demise of the megafauna. Quaternary Science Reviews, , National Academy of Sciences of the United States of America, , –. –. CROWLEY, B.E. & SAMONDS, K.E. () Stable carbon isotope values HOWE, H.F. () Making dispersal syndromes and networks useful confirm a recent increase in grasslands in northwestern in tropical conservation and restoration. Global Ecology and Madagascar. The Holocene, , –. Conservation, , –. DENG, Y., CHANG, C., IDO, M.S. & LONG,Q.() Multiple HOWE, H.F. & SMALLWOOD,J.() Ecology of seed dispersal. imputation for general missing data patterns in the presence of Annual Review of Ecology and Systematics, , –. high-dimensional data. Scientific Reports, , . HUGHES, L., DUNLOP, M., FRENCH, K., LEISHMAN, M.R., RICE, B., DEWAR, R.E. & RICHARD, A.F. () Evolution in the hypervariable RODGERSON,L.&WESTOBY,M.() Predicting dispersal spectra: environment of Madagascar. Proceedings of the National Academy a minimal set of hypotheses based on plant attributes. Journal of of Sciences of the United States of America, , –. Ecology, , –. FEDERMAN, S., SINNOTT-ARMSTRONG, M., BADEN, A.L., CHAPMAN, HUTCHEON, J.M. () Frugivory by Malagasy bats. In The Natural C.A., DALY, D.C., RICHARD, A.R. et al. () The paucity of History of Madagascar (eds S.M. Goodman & J.P. Benstead), pp. frugivores in Madagascar may not be due to unpredictable –. The University of Chicago Press, Chicago, USA, and temperatures or fruit resources. PLoS ONE, (), e. London, UK. FLEMING, T.H., BREITWISCH,R.&WHITESIDES, G.H. () Patterns IUCN () The IUCN Red List of Threatened Species v. -. of tropical vertebrate frugivore diversity. Annual Review of Ecology Http://www.iucnredlist.org [accessed  November ]. and Systematics, , –. JANSON, C.H. () Adaptation of fruit morphology to dispersal GANZHORN, J.U., ARRIGO-NELSON, S., BOINSKI, S., BOLLEN, A., agents in a Neotropical forest. Science, , –. CARRAI, V., DERBY, A. et al. () Possible fruit protein effects on JANZEN, D.H. & MARTIN, P.S. () Neotropical anachronisms: the primate communities in Madagascar and the Neotropics. PLoS fruits the gomphotheres ate. Science, , –. ONE, (), e. JORDANO,P.() Fruits and frugivory. In Seeds: The Ecology of GAUTIER-HION, A., DUPLANTIER, J.-M., QURIS, R., FEER, F., SOURD, Regeneration in Plant Communities (ed. M. Fenner), pp. –. C., DECOUX, J.-P. et al. () Fruit characters as a basis of fruit CABI, Wallingford, UK.

JORDANO, P., FORGET, P.-M., LAMBERT, J.E., BÖHNING-GAESE, K., TERBORGH, J., NUÑEZ-ITURRI, G., PITMAN, N.C.A., CORNEJO TRAVESET,A.&WRIGHT, S.J. () Frugivores and seed dispersal: VALVERDE, F.H., ALVAREZ, P., SWAMY,V.etal.() Tree mechanisms and consequences for biodiversity of a key ecological recruitment in an empty forest. Ecology, , –. interaction. Biology Letters, , –. TURNER, I.M. () The Ecology of Trees in the Tropical Rain Forest. KUHLMANN,M.&RIBEIRO, J.F. () Fruits and frugivores of the Cambridge University Press, Cambridge, UK. Brazilian Cerrado: ecological and phylogenetic considerations. Acta VAN BUUREN,S.&GROOTHUIS-OUDSHOORN,K.() mice: Botanica Brasilica, , –. Multivariate Imputation by Chained Equations in R. Journal of LOMÁSCOLO, S.B. & SCHAEFER, H.M. () Signal convergence in Statistical Software, , –. fruits: a result of selection by frugivores? Journal of Evolutionary VAN DER PIJL,L.() Principles of Dispersal in Higher Plants. Biology, , –. Springer-Verlag, Berlin, Germany. LORTS, C.M., BRIGGEMAN,T.&SANG,T.() Evolution of fruit VOIGT, F.A., BLEHER, B., FIETZ, J., GANZHORN, J.U., SCHWAB,D.& types and seed dispersal: a phylogenetic and ecological snapshot. BÖHNING-GAESE,K.() A comparison of morphological and Journal of Systematics and Evolution, , –. chemical fruit traits between two sites with different frugivore MADAGASCAR CATALOGUE () Catalogue of the Vascular Plants of assemblages. Oecologia, , –. Madagascar. Missouri Botanical Garden, St. Louis, USA, & VORONTSOVA, M.S., BESNARD, G., FOREST, F., MALAKASI, P., MOAT, Antananarivo, Madagascar. J., CLAYTON, W.D. et al. () Madagascar’s grasses and grasslands: MANGIAFICO,S.() rcompanion: Functions to support extension anthropogenic or natural? Proceedings of the Royal Society B, education program evaluation. R package version ... Https:// . CRAN.R-project.org/package=rcompanion. WHITE, I.R., ROYSTON,P.&WOOD, A.M. () Multiple imputation MERCIER, J.-L. & WILMÉ,L.() The Eco-Geo-Clim model: using chained equations: issues and guidance for practice. Statistics explaining Madagascar’s endemism. Madagascar Conservation & in Medicine, , –. Development, , –. WILLSON,M.&TRAVESET,A.() The ecology of seed dispersal. In MITTERMEIER, C.G., LOUIS, E.E., RICHARDSON, M., SCHWITZER, C., Seeds: The Ecology of Regeneration in Plant Communities (ed. LANGRAND, O., RYLANDS, A.B. et al. () Lemurs of Madagascar. M. Fenner), pp. –. CABI, Wallingford, UK. Conservation International, Arlington, USA. WILLSON,M.F.,IRVINE,A.K.&WALSH,N.G.()Vertebrate MUSCARELLA,R.&FLEMING, T.H. () The role of frugivorous bats dispersal syndromes in some Australian and New Zealand plant in tropical forest succession. Biological Reviews, , –. communities, with geographic comparisons. Biotropica, , OSURI, A.M., RATNAM, J., VARMA, V., ALVAREZ-LOAYZA,P., –. ASTAIZA, J.H., BRADFORD, M. et al. () Contrasting effects of WILLSON, M.F., RICE, B.L. & WESTOBY,M.() Seed dispersal defaunation on aboveground carbon storage across the global spectra: a comparison of temperate plant communities. Journal of tropics. Nature Communications, , . Vegetation Science, , –. PENONE, C., DAVIDSON, A.D., SHOEMAKER, K.T., DIMARCO, M., WRIGHT,P.() Lemur traits and Madagascar ecology: coping RONDININI, C., BROOKS, T.M. et al. () Imputation of missing with an island environment. Yearbook of Physical Anthropology, , data in life-history trait datasets: which approach performs the best? –. Methods in Ecology and Evolution, , –. YODER, A.D. & NOWAK, M.D. () Has vicariance or dispersal been PICOT, M., JENKINS, R.K.B., RAMILIJAONA, O., RACEY, P.A. & the predominant biogeographic force in Madagascar? Only time CARRIÈRE, S.M. () The feeding ecology of Eidolon dupreanum will tell. Annual Review of Ecology, Evolution, and Systematics, , (Pteropodidae) in eastern Madagascar. African Journal of Ecology, –. , –. PINHEIRO, J., BATES, D., DEBROY, S., SARKAR,D.&RCORE TEAM () nlme: Linear and Nonlinear Mixed Effects Models. R package version .-, http://CRAN.R-project.org/package=nlme. Biographical sketches RAZAFINDRATSIMA, O.H.F.E. () Nonrandom seed dispersal by lemur frugivores: mechanism, patterns and impacts. PhD thesis. Rice AURÉLIE ALBERT-DAVIAUD is a specialist in seed dispersal by ani- University, Houston, USA. mals. She studies the impact of dispersers on forest maintenance and RDEVELOPMENT CORE TEAM () R: A Language and Environment regeneration, and the impact of human disturbance on seed dispersal for Statistical Computing. R Foundation for Statistical Computing, interactions and networks. SARAH PERILLO’s research focuses on dis- Vienna, Austria. tribution and dispersal of Arecaceae in Madagascar, with interests in RUBIN, D.B. () Multiple Imputation for Nonresponse in Surveys. policy and governance. WOLFGANG STUPPY is a specialist in seed John Wiley & Sons, New York, USA. and fruit morphology and development. His main research interests SPJUT, R.W. () A Systematic Treatment of Fruit Types. The are the structure and biology of fruits and seeds, especially adaptations New York Botanical Garden Press, New York, USA. to dispersal by animals.