Patterns of plant functional traits in the biogeography of West African grasses (Poaceae) Marco Schmidt1,2,3, Adjima Thiombiano4, Alexander Zizka1, Konstantin König5, Ulrike Brunken6 and Georg Zizka1,2,3 1 Department of Botany and molecular Evolution, Senckenberg Research Institute, Senckenberganlage 25, 60325 Frankfurt am Main, Germany, Institute for Ecology, Evolution and Diversity, Goethe University, Siesmayerstr 70, 60323 Frankfurt am Main, Germany, 3Biodiversity and Climate Research Centre, Senckenberganlage 25, 60325 Frankfurt am Main, Germany, 4Laboratoire de Biologie et Écologie Végétales, UFR ⁄ SVT, University of Ouagadougou, 03 BP 7021 Ouagadougou 03, Burkina Faso, 5World Agroforestry Centre, Trav. Enéas Pinheiro s ⁄ n, 66.095-100 Belem, PA, Brazil and 6Palmengarten, Abt. Garten, Wissenschaft & Pädagogik, Siesmayerstr 61, 60323 Frankfurt am Main, Germany 2 Abstract Résumé Grasses (Poaceae) are the largest family of vascular plants in Burkina Faso with 254 species. In the savannahs they are the most important family in terms of abundance and species richness, in other habitats, such as gallery forests, there are only few species. On the country scale there is a change in growth form: while in the Sahelian north most grasses are small therophytes, the Sudanian south is characterized by tall, often perennial grasses. To analyse these patterns in detail, we compiled a database on grass occurrences and used it in an ecological niche modelling approach with the programme Maxent to obtain countrywide distribution models. Secondly we used data on photosynthetic type, height, leaf width and growth form to aggregate the species distributions and quantified the relative importance of functional groups per grid cell. Pronounced latitudinal differences could be shown for life forms, photosynthesis and size: the drier north is characterized by smaller, mainly therophytic grasses with a high share of C4 NAD-ME photosynthesis, while the more humid south is characterized by tall, often hemicryptophytic grasses with C4 NADP-ME photosynthesis. For leaf width, no clear country-wide patterns could be observed, but local differences with more broad-leaved grasses in humid areas. Les graminées (Poaceae) sont la plus grande famille de plantes vasculaires du Burkina Faso avec 254 espèces. Dans la savane, elles constituent la famille la plus importante en termes d’abondance et de richesse en espèces; dans d’autres habitats, tels que les galeries forestières, il n’y en a que quelques espèces. À l’échelle du pays, il y a un changement dans la croissance: alors que dans le nord sahélien, la plupart des graminées sont de petits thérophytes, le sud soudanien se caractérise par de hautes graminées, souvent pérennes. Pour analyser ces schémas en détail, nous avons compilé une base de données sur l’occurrence des graminées et nous l’avons utilisée dans une approche de modélisation d’une niche écologique avec le programme Maxent, pour obtenir des modèles de distribution à l’échelle du pays. Deuxièmement, nous avons utilisé des données sur le type photosynthétique, la hauteur, la largeur des feuilles et la forme de croissance pour regrouper la distribution d’espèces et nous avons quantifié l’importance relative de groupes fonctionnels par cellule de grille. On pouvait montrer des différences latitudinales prononcées pour les formes, pour la photosynthèse et la taille observées; le nord, plus aride, se caractérise par des formes plus petites, surtout thérophytes, avec une grande proportion de photosynthèse de plantes C4 NAD-ME, alors que le sud, plus humide, se caractérise par de grandes graminées souvent hémicryptophytes avec une photosynthèse C4 NADP-ME. Pour la largeur des feuilles, on n’a pas pu observer de schémas clairs à l’échelle du pays, mais bien des différences locales, avec des graminées à feuilles plus larges dans les zones humides. Key words: ecological niche models, macroecology, plant functional traits, Poaceae, Sahel, savannah *Correspondence: E-mail: mschmidt@senckenberg.de 490  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Biogeography of West African grasses 491 continuum, with woody cover increasing towards the south, the Sahelian Acacia spp. and Balanites gradually being replaced by Combretaceae, Caesalpiniaceae and other broad-leaved species, and grasses becoming taller and denser, with Andropogoneae savannahs of the Sudanian zone often being 4–5 m tall. Grasses are the dominant family among the savannah herbs, both in terms of species richness and abundance (Schmidt, 2006; Schmidt et al., 2010a). Their predominance is closely linked to the evolution of the C4 pathway (Christin et al., 2008), the grass–fire cycle (Beerling & Osborne, 2006; Higgins, Bond & Trollope, 2000) and influenced by co-evolution with herbivores (Stebbins, 1981). Wild grasses are also of high economical importance: like in many other West African countries, Burkina Faso’s economy is partly based on cattle and small ruminants, which heavily depend on savannah grasses. Some species including Echinochloa spp., Panicum laetum and Cenchrus biflorus are also collected as wild cereals (Brink & Belay, 2006; Pedersen & Benjaminsen, 2008). While broad geographical trends concerning life cycle and size are known, our aim is to uncover more details on the functional biogeography of grasses using a country-wide approach. Original data on local grass floras are sparsely and unevenly distributed with high sampling intensities in focus research areas such as the provinces of Gourma and Oudalan and low sampling intensities in the Subsahel and parts of Grasses are the most important group of herbaceous plants in the African savannahs in terms of species richness, abundance, and economic importance. Our study focuses on the biogeography of grasses in Burkina Faso, a country well representing the Sahelo-sudanian savannah belt of West Africa. Burkina Faso is a landlocked, mostly flat country in West Africa, most of it belonging to a peneplain at 300– 400 m altitude, except for some remains of a Precambrian massif, the Gobnangou Mountains near the border to Benin and the sandstone massif culminating in the Mt. Tenakourou (749 m) in the west of the country (Fig. 1). The low topodiversity makes it an ideal place for analysing changes in species composition along the climatic gradient from the Sahara desert to the Upper Guinean rainforests. This gradient is characterized by lower and more irregular rainfall and a higher seasonal variability of temperatures in the north. The single factors (annual precipitation, annual mean temperature, but also extreme values and seasonality) are not collinear but all roughly in latitudinal direction (Worldclim, Hijmans et al., 2005). Climate is shaping the predominant vegetation types from the Sahelian tiger bush and thornbush savannahs to the South Sudanian woodlands and dry forests. There are no abrupt changes in the savannah types, these rather form a 15° 5° 4° 3° 2° 1° 750 m 1° 2° 360 mm 0m m Gorom-Gorom 14° 14° 40 0° 15° Introduction 140 m 500 mm 13° 13° Sourou 6 00 m m Nakam be 700 mm Na z in Comoe 10° Fig 1 Study area Burkina Faso: altitude, isohyetes, major rivers and cities, and location within Africa 5°  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 900 m m 11° 0m on m 1100 4° mm 3° 2° 1° 0 50 100 0° 1° 200 km 2° 10° 11° Bobo-Dioulasso 100 Fada N'Gourma mm 8 00 12° Mo uh ou 12° n Ouagadougou 492 Marco Schmidt et al. the south-west (Schmidt et al., 2010a). Therefore, we decided to use species distribution models to obtain a good coverage of grass distribution data throughout the country. Material and methods Botanical occurrence data A comprehensive coverage of occurrence records is a precondition for good distribution models. Digitization of the herbarium collections in Frankfurt (FR) and Ouagadougou (OUA) and relevé data of the past 15 years from the West African Vegetation Database (http://www.westafricanvegetation.org; Janssen et al., 2011) have led to a database of c. 25, 000 species occurrence records of grasses in Burkina Faso. The majority of these have been georeferenced at the time of collection by GPS readings, so a high spatial accuracy is guaranteed. Previous sampling gaps in the south-western part of Burkina Faso (Schmidt et al., 2005) have been largely closed by now. Taxonomic accuracy is high for the collection records, as both collections are highly frequented and thereby steadily improved, but certainly lower for the observation records, where determinations can only be proved for the voucher specimens. On the other hand, certain functional and taxonomic groups, rare or frequent species, are represented differently in collection and observation data (Guralnick & Van Cleve, 2005; Schmidt et al., 2010b). The inherent bias of both collection and observation data is partly taken care of by a combination of both data types. Grass specimens have been identified in the field and in both herbaria with Poilecot (1995, 1999) and Scholz & Scholz (1983), the most recent collections also with Akoègninou et al. (2006). The nomenclature follows the African Plants Database (Klopper et al., 2007), and synonyms from the primary data have been assigned to the names accepted there for tropical Africa. Plant functional trait data Data on plant functional traits have been assembled for all species with enough records to be integrated in our models. Data on height and leaf width have been taken from Grass Base (Clayton, Harman & Williamson, 2006), Poilecot (1995, 1999) and Scholz & Scholz (1983), photosynthetic types from Ellis (1977), Downton (1975) and own observations on leaf anatomy following Ellis, Vogel & Fuls (1980). The subtypes of C4 photosynthesis are named after their decarboxylating enzymes, the NADP-malic enzyme (NADP-ME), NAD-malic enzyme (NAD-ME) or phosphoenolpyruvate carboxykinase (PCK). Information on life forms has been taken from Poilecot (1995, 1999), Grass Base (Clayton, Harman & Williamson, 2006) or own field and herbarium observations. For our analysis, we have taken the mean values of plant height and leaf width (if not stated as such, calculated from the range given in the literature). Mean values were then assigned to four classes of plant height and five classes of leaf width. All values used in our analysis are documented in Table 1. Predictor variables For model predictions, we used two standard spectral vegetation indices, the Normalized Differenced Vegetation Index (NDVI) and the Enhanced Vegetation Index (EVI) derived from MODIS satellite data (the MOD13Q1 16 days composite at 250-metre spatial resolution). The two vegetation indices are spectral transformations of red and near-infrared bands of the satellite sensor and are designed to provide spatially and temporally consistent observations of vegetation conditions (Huete et al., 2002) over long time periods. Using the method of Chavez & Kwarteng (1989), we calculated the first principal component of the biweekly NDVI respectively EVI images for each month across 3 years (2001–2003), resulting in one corrected image per month and vegetation index. By this approach, spectral noise and phenological vegetation differences were reduced considerably. Modelling approach We used the programme Maxent (version 3.0.6) with the occurrences and environmental data as described earlier. We allowed all feature types (linear, product, quadratic, hinge, threshold and categorical) and set the regularization factor to 1. The output resolution of the distribution models equals the resolution of the environmental data (grid cells of 250 · 250 m). Following the recommendations of Stockwell & Peterson (2002), only species with >10 spatially unique occurrence points have been modelled. Of the 254 grass species of Burkina Faso, 122 species met our threshold value of >10 occurrence points. (These are listed with the number of occurrence points and functional trait data in Table 1). The reduced set of species meeting this threshold did not significantly change the  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Biogeography of West African grasses 493 Table 1 List of all modelled species with number of occurrences and functional traits Names Occurences Photosynthetic_type Acroceras amplectens Stapf Andropogon chinensis (Nees) Merr. Andropogon fastigiatus Sw. Andropogon gayanus Kunth Andropogon pseudapricus Stapf Andropogon schirensis Hochst. ex A.Rich. Andropogon tectorum Schumach. & Thonn. Aristida adscensionis L. Aristida funiculata Trin. & Rupr. Aristida hordeacea Kunth Aristida kerstingii Pilg. Aristida mutabilis Trin. & Rupr. Aristida sieberiana Trin. ex Spreng. Aristida stipoides Lam. Brachiaria lata (Schumach.) C.E.Hubb. Brachiaria orthostachys (Mez) Clayton Brachiaria ramosa (L.) Stapf Brachiaria stigmatisata (Mez) Stapf Brachiaria villosa (Lam.) A.Camus Brachiaria xantholeuca (Hack.) Stapf Cenchrus biflorus Roxb. Cenchrus ciliaris L. Chasmopodium caudatum (Hack.) Stapf Chloris pilosa Schumach. Chrysochloa hindsii C.E.Hubb. Chrysopogon nigritanus (Benth.) Veldkamp Ctenium elegans Kunth Ctenium newtonii Hack. Cymbopogon caesius (Nees ex Hook. & Arn.) Stapf Cymbopogon schoenanthus (L.) Spreng. Cynodon dactylon (L.) Pers. Dactyloctenium aegyptium (L.) Willd. Digitaria argillacea (Hitchc. & Chase) Fernald Digitaria ciliaris (Retz.) Koeler Digitaria gayana (Kunth) A.Chev. ex Stapf Digitaria horizontalis Willd. Diheteropogon amplectens (Nees) Clayton Diheteropogon hagerupii Hitchc. Echinochloa colona (L.) Link Echinochloa pyramidalis (Lam.) Hitchc. & Chase Echinochloa stagnina (Retz.) P.Beauv. Eleusine indica (L.) Gaertn. Elionurus elegans Kunth Elymandra androphila (Stapf) Stapf Elytrophorus spicatus (Willd.) A.Camus Enteropogon prieurii (Kunth) Clayton Eragrostis aspera (Jacq.) Nees Eragrostis atrovirens (Desf.) Trin. ex Steud. Eragrostis cilianensis (All.) Vignolo ex Janch. 17 85 99 257 243 43 37 315 193 58 84 209 68 16 175 141 25 35 91 130 508 105 57 188 30 19 21 30 21 45 20 650 70 520 35 61 35 15 175 17 47 17 18 18 19 269 12 41 38  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 C3 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C3 C4 C4 C4 C4 Life form raunkiaer Therophyte Hemicryptophyte Therophyte Hemicryptophyte Therophyte Hemicryptophyte Hemicryptophyte Therophyte Therophyte Therophyte Therophyte Therophyte Hemicryptophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Therophyte Hemicryptophyte Therophyte Hemicryptophyte Hemicryptophyte Hemicryptophyte Geophyte Therophyte Therophyte Therophyte Therophyte Therophyte Hemicryptophyte Therophyte Therophyte Geophyte Geophyte Therophyte Therophyte Hemicryptophyte Therophyte MS (NADP-ME) Therophyte PS-NAD (NAD-ME) Therophyte PS-PCK (PCK) Hemicryptophyte PS-NAD (NAD-ME) Therophyte MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) PS-PCK (PCK) PS-PCK (PCK) PS-PCK (PCK) PS-PCK (PCK) PS-PCK (PCK) PS-PCK (PCK) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) PS-PCK (PCK) PS-NAD (NAD-ME) MS (NADP-ME) PS-NAD (NAD-ME) PS-NAD (NAD-ME) MS (NADP-ME) MS (NADP-ME) PS-NAD (NAD-ME) PS-PCK (PCK) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) PS-NAD (NAD-ME) MS (NADP-ME) MS (NADP-ME) Height (cm) Width (mm) 50–100 100–200 100–200 200–400 100–200 100–200 200–400 50–100 0–50 50–100 50–100 50–100 50–100 100–200 0–50 0–50 0–50 50–100 0–50 0–50 50–100 50–100 200–400 50–100 0–50 200–400 100–200 100–200 50–100 50–100 0–50 0–50 0–50 50–100 50–100 50–100 100–200 100–200 0–50 200–400 100–200 50–100 0–50 100–200 0–50 50–100 50–100 50–100 50–100 5–10 0–5 0–5 10–20 0–5 5–10 10–20 0–5 0–5 5–10 0–5 0–5 0–5 0–5 10–20 5–10 5–10 5–10 0–5 5–10 0–5 5–10 10–20 0–5 0–5 5–10 0–5 0–5 5–10 0–5 0–5 5–10 0–5 5–10 5–10 5–10 10–20 0–5 5–10 10–20 10–20 0–5 0–5 5–10 0–5 0–5 5–10 0–5 5–10 494 Marco Schmidt et al. Table 1 (Continued) Names Occurences Photosynthetic_type Life form raunkiaer Eragrostis ciliaris (L.) R.Br. Eragrostis gangetica (Roxb.) Steud. Eragrostis japonica (Thunb.) Trin. Eragrostis pilosa (L.) P.Beauv. Eragrostis tremula Hochst. ex Steud. Eragrostis turgida (Schumach.) De Wild. Euclasta condylotricha (Hochst. ex Steud.) Stapf Hackelochloa granularis (L.) Kuntze Heteropogon contortus (L.) P. Beauv. ex Roem. & Schult. Hyparrhenia barteri (Hack.) Stapf Hyparrhenia involucrata Stapf Hyparrhenia rufa (Nees) Stapf Hyparrhenia smithiana (Hook.f.) Stapf Hyparrhenia subplumosa Stapf Hyparrhenia welwitschii (Rendle) Stapf Hyperthelia dissoluta (Nees ex Steud.) Clayton Imperata cylindrica (L.) Raeusch. Leersia hexandra Sw. Loudetia annua (Stapf) C.E.Hubb. Loudetia arundinacea (Hochst. ex A.Rich.) Steud. Loudetia flavida (Stapf) C.E.Hubb. Loudetia hordeiformis (Stapf) C.E.Hubb. Loudetia simplex (Nees) C.E.Hubb. Loudetia togoensis (Pilg.) C.E.Hubb. Loudetiopsis kerstingii (Pilg.) Conert Microchloa indica (L.f.) P.Beauv. Monocymbium ceresiiforme (Nees) Stapf Oryza barthii A.Chev. Oryza longistaminata A.Chev. & Roehr. Panicum fluviicola Steud. Panicum humile Nees ex Steud. Panicum laetum Kunth Panicum pansum Rendle Panicum phragmitoides Stapf Panicum subalbidum Kunth Paspalum scrobiculatum L. Pennisetum glaucum (L.) R.Br. Pennisetum pedicellatum Trin. Pennisetum polystachion (L.) Schult. Pennisetum unisetum (Nees) Benth. Pennisetum violaceum (Lam.) Rich. Rhytachne gracilis Stapf Rhytachne triaristata (Steud.) Stapf Rottboellia cochinchinensis (Lour.) Clayton Sacciolepis africana C.E.Hubb. & Snowden Sacciolepis cymbiandra Stapf Schizachyrium brevifolium (Sw.) Nees ex Büse Schizachyrium exile (Hochst.) Pilg. 18 19 19 229 261 80 61 103 32 Therophyte 0–50 Therophyte 0–50 Therophyte 100–200 Therophyte 0–50 Therophyte 50–100 Therophyte 0–50 Therophyte 100–200 Therophyte 50–100 Hemicryptophyte 50–100 16 91 63 42 69 27 13 21 13 16 28 10 11 47 149 12 142 25 25 43 22 14 413 125 14 61 89 13 411 139 27 18 22 30 56 11 14 42 143 C4 PS-PCK (PCK) C4 C4 C4 C4 C4 C4 PS-NAD (NAD-ME) PS-PCK (PCK) PS-NAD (NAD-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) C4 C4 C4 C4 C4 C4 C4 C4 C3 C4 MS MS MS MS MS MS MS MS C4 C4 C4 C4 C4 C4 C4 C3 C3 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C3 C3 C4 C4 (NADP-ME) (NADP-ME) (NADP-ME) (NADP-ME) (NADP-ME) (NADP-ME) (NADP-ME) (NADP-ME) Therophyte Therophyte Hemicryptophyte Hemicryptophyte Hemicryptophyte Therophyte Hemicryptophyte Geophyte Geophyte MS (NADP-ME) Therophyte Hemicryptophyte MS (NADP-ME) Hemicryptophyte MS (NADP-ME) Therophyte MS (NADP-ME) Hemicryptophyte MS (NADP-ME) Therophyte MS (NADP-ME) Therophyte PS-PCK (PCK) Therophyte MS (NADP-ME) Hemicryptophyte Therophyte Geophyte PS (NAD-ME or PCK) Hemicryptophyte Therophyte PS-NAD (NAD-ME) Therophyte PS (NAD-ME or PCK) Therophyte Hemicryptophyte PS-NAD (NAD-ME) Therophyte MS (NADP-ME) Hemicryptophyte MS (NADP-ME) Therophyte Therophyte MS (NADP-ME) Therophyte Hemicryptophyte MS (NADP-ME) Therophyte MS (NADP-ME) Therophyte MS (NADP-ME) Therophyte MS (NADP-ME) Therophyte Geophyte Geophyte MS (NADP-ME) Therophyte MS (NADP-ME) Therophyte Height (cm) Width (mm) 100–200 100–200 100–200 100–200 200–400 100–200 200–400 50–100 50–100 50–100 100–200 50–100 50–100 50–100 50–100 50–100 0–50 50–100 50–100 50–100 100–200 0–50 0–50 50–100 100–200 100–200 50–100 200–400 50–100 100–200 200–400 50–100 0–50 50–100 100–200 100–200 50–100 0–50 50–100 0–5 0–5 5–10 0–5 0–5 5–10 5–10 5–10 5–10 0–5 5–10 5–10 5–10 5–10 5–10 0–5 10–20 5–10 0–5 5–10 0–5 5–10 0–5 5–10 0–5 0–5 0–5 5–10 10–20 5–10 0–5 5–10 5–10 10–20 5–10 5–10 20–40 5–10 0–5 20–40 5–10 0–5 0–5 20–40 5–10 0–5 0–5 0–5  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Biogeography of West African grasses 495 Table 1 (Continued) Names Occurences Photosynthetic_type Life form raunkiaer Height (cm) Width (mm) Schizachyrium nodulosum (Hack.) Stapf Schizachyrium platyphyllum (Franch.) Stapf Schizachyrium ruderale Clayton Schizachyrium sanguineum (Retz.) Alston Schizachyrium urceolatum (Hack.) Stapf Schoenefeldia gracilis Kunth Setaria barbata (Lam.) Kunth Setaria pumila (Poir.) Roem. & Schult. Setaria sphacelata (Schumach.) Stapf & C.E.Hubb. ex M.B.Moss Setaria verticillata (L.) P.Beauv. Sorghastrum bipennatum (Hack.) Pilg. Sporobolus festivus Hochst. ex A.Rich. Sporobolus microprotus Stapf Sporobolus paniculatus (Trin.) T.Durand & Schinz Sporobolus pectinellus Mez Sporobolus pyramidalis P.Beauv. Tetrapogon cenchriformis (A.Rich.) Clayton Thelepogon elegans Roth Tragus berteronianus Schult. Tragus racemosus (L.) All. Tripogon minimus (A.Rich.) Hochst. ex Steud. Urelytrum muricatum C.E.Hubb. Urochloa jubata (Fig. & De Not.) Sosef Urochloa mutica (Forssk.) T.Q.Nguyen Urochloa trichopus (Hochst.) Stapf 30 22 31 50 17 517 36 221 14 C4 C4 C4 C4 C4 C4 C4 C4 C4 MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) PS-NAD (NAD-ME) MS (NADP-ME) MS (NADP-ME) MS (NADP-ME) Therophyte Hemicryptophyte Therophyte Hemicryptophyte Therophyte Therophyte Therophyte Therophyte Geophyte 0–50 100–200 100–200 100–200 0–50 50–100 50–100 0–50 100–200 0–5 5–10 0–5 5–10 0–5 0–5 10–20 5–10 0–5 12 50 30 52 11 83 97 108 12 150 90 37 15 51 25 27 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 C4 MS (NADP-ME) MS (NADP-ME) PS-PCK (PCK) PS-NAD (NAD-ME) PS-PCK (PCK) PS (NAD-ME or PCK) PS-PCK (PCK) PS-NAD (NAD-ME) MS (NADP-ME) PS-NAD (NAD-ME) PS-NAD (NAD-ME) PS-NAD (NAD-ME) MS (NADP-ME) PS-PCK (PCK) PS-PCK (PCK) PS-PCK (PCK) Therophyte 50–100 Therophyte 100–200 Hemicryptophyte 0–50 Therophyte 0–50 Therophyte 0–50 Therophyte 0–50 Hemicryptophyte 50–100 Therophyte 0–50 Therophyte 50–100 Therophyte 0–50 Therophyte 0–50 Hemicryptophyte 0–50 Hemicryptophyte 100–200 Hemicryptophyte 50–100 Hemicryptophyte 50–100 Therophyte 50–100 10–20 5–10 0–5 5–10 0–5 0–5 5–10 0–5 10–20 0–5 0–5 0–5 5–10 10–20 5–10 10–20 PCK, phosphoenolpyruvate carboxykinase. functional group composition of the whole grass flora, but excluded some extremes in plant height (the bamboo species Oxythenanthera abyssinica and Bambusa vulgaris) and leaf width (Olyra latifolia). Aggregation of distribution models The modelled species distributions obtained by Maxent were summed up (i) for all species and (ii) for all species belonging to a particular group (a life form, photosynthetic type or one of the classes of height or leaf width). This was done by a VBA routine in MS Access, counting for each grid cell the species with a probability of presence above a particular threshold. For this purpose, we used the ‘Maximum training sensitivity plus specificity threshold’ following the recommendations of Liu et al. (2005). Using the Spatial Analyst of ArcGIS, we subsequently calculated for each grid cell and each functional group the share of  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 species belonging to a particular group within the total species set of the grid cell. Results The species richness of grasses in Burkina Faso (Fig. 2a) is generally increasing towards the south with highest values of nearly 90 species per grid cell in the North Sudanian zone. The geographical pattern is very fragmented. Agricultural areas, such as the southern Mossi plateau, have a richer grass flora than the protected areas of the WAP complex (a complex of protected areas including the national parks of W, Arly and Pendjari), the sandstone mountains west of Bobo-Dioulasso and the valleys of the Nazinon and the Comoé. The life form spectrum of Burkina Faso’s grasses (Fig. 2b–d) is dominated by therophytes, especially in the north, but to a lesser extent also in the south and even some 496 Marco Schmidt et al. (a) (b) 1 spp. 89 spp. (c) 0% 100% 0% 100% (d) 0% 100% Fig 2 Species richness and life forms: (a) species richness per grid cell, (b) geophytes, (c) hemicryptophytes and (d) therophytes, all expressed as the percentage of the local grass flora predicted from the distribution models representatives of the tall Andropogoneae of Sudanian savannahs, such as Hyparrhenia involucrata, are therophytic. There is a therophyte-dominated spot around the capital Ouagadougou and areas with fewer therophytes and more hemicryptophytes in the WAP complex. The proportion of hemicryptophytes increases from north to south. Only the share of geophytes does not follow any clear latitudinal (a) patterns, and there are, however, slightly higher values along rivers, most pronounced in the Sourou valley. Owing to the exclusion of the bamboos, phanerophytes are not represented. The share of the two main photosynthetic types of grasses, C3 and C4, does not change much across the country (Fig. 3). C3 grasses only have a share of about 5% (b) 0% 100% (c) 0% 100% 0% 100% (d) 0% 100% Fig 3 Photosynthetic types and subtypes: species with (a) C3, (b) C4 NAD-ME, (c) C4 NADP-ME and (d) C4 phosphoenolpyruvate carboxykinase metabolism, all expressed as the percentage of the local grass flora predicted from the distribution models  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Biogeography of West African grasses 497 (a) (b) 0% 100% (c) 0% 100% 0% 100% (d) 0% 100% Fig 4 Plant size: species reaching (a) 0–50 cm, (b) 50–100 cm, (c) 100–200 cm and (d) 200–400 cm of height, all expressed as the percentage of the local grass flora predicted from the distribution models of the grass flora. There is, however, a latitudinal trend of C4 subtypes with NADP-ME species increasing and NADME species (and to a lesser extent also PCK species) decreasing towards the south. NADP-ME species are the largest group with 50–60% of all species, the share of PCK species is c. 15% and NAD-ME species 10–15%. Locally, the WAP complex stands out with a higher share of NADP-ME species, especially Andropogoneae. Most species of our study area belong to the lower size classes of 0–50 and 50–100 cm (Fig. 4). Only c. 20% of the northernmost and 35% of the southernmost local grass flora belong to the size classes of 100–200 and 200–400 cm. Plant size generally increased towards the south. While the proportion of the smallest size class of 0–50 cm decreases from north to south, the proportion of the larger size class of 100–200 cm increases. The intermediate class of 50–100 cm is more evenly distributed, with a gap around Ouagadougou and locally higher importance in the Comoé valley in the south-west. The largest species of 200–400 cm are evenly distributed. The even larger bamboo species did not pass our threshold of occurrence points and were therefore not modelled, but are only occurring in the southern parts of Burkina Faso. The spectrum of leaf width hardly changes throughout the study area (Fig. 5). Nearly half of the species belong to  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 the narrow-leaved species with a leaf width of 0–5 mm, another 40% to the leaf width class of 5–10 mm, roughly 10% to the class of 10–20 mm and <1% to the most broad-leaved species with 20–40 mm (Fig. 2). The very broad-leaved (>40 mm) species Andropogon pteropholis, Olyra latifolia, Schizachyrium platyphyllum and Stenotaphrum diminiatum did not pass our threshold of occurrence points and are therefore not included. Although large scale trends are not detectable, some local patterns are noteworthy: large standing water bodies with seasonal inundations such as the dams of Kompienga and Bagré or the Sourou valley, but also the sandstone highlands in the south-west, have a lower share of narrow-leaved species, while they are most important in the western part of the WAP complex. Discussion Our decision to use modelled distributions instead of original occurrence data enabled us to produce results for the whole area of Burkina Faso. The only alternative data source that would have included complete local grass floras are inventories of protected areas, such as Mbayngone et al. (2008), Ouédraogo et al. (2011) or Guinko (2005). These, however, would not have been available in sufficient quantity for a spatial analysis. 498 Marco Schmidt et al. (a) (b) 0% 100% (c) 0% 100% 0% 100% (d) 0% 100% Fig 5 Leaf width: species with leaves reaching (a) 0–5 mm, (b) 5–10 mm, (c) 10–20 mm and (d) 20–40 mm of width, all expressed as the percentage of the local grass flora predicted from the distribution models The use of satellite data as environmental data for species distribution models has not yet been realized often (e.g. Saatchi et al., 2008; König, Schmidt & Müller, 2009), usually modellers use climate grids such as the Bioclim data set of Hijmans et al. (2005). We believe that satellite data are especially suitable for grasses because it take land use and land cover into account, which, in a changing environment, is important for grasses as organisms with high dispersal capabilities and short life span. The collection and observation data used are from the whole gradient covered by the vegetation indices, including savannahs and cultivated lands, special habitats such as the tiger bush in the Sahel, the sparsely vegetated sandstone cliffs of the Chaı̂ne de Gobnangou, and the Chaı̂ne de Banfora, some of the last remaining gallery forests and the large protected areas with dense Sudanian savannahs of southeastern Burkina Faso, despite the restricted accessibility in the rainy season. The gradient expressed by the vegetation indices, although not systematically sampled, has therefore been widely covered. The distribution of grass species richness (Fig. 2a) is similar to the one obtained in a previous study (Schmidt et al., 2005, fig. 6) where climatic variables at a 10¢ resolution have been used as predictors, but is far more detailed and allows the identification of landscape elements such as the fixed dunes of the Sahel. Highest species richness is located in the North Sudanian zone around the 800 mm isohyets, which is an area of high agricultural intensity with mosaics of fields, fallows and near-natural savannah remains. The lower species richness in the protected areas of the WAP complex contributes to the impression that the high species richness might be due to high habitat diversity. The life form spectrum (Fig. 2b–d) is dominated by therophytes in the north and hemicryptophytes in the south. The higher importance of therophytes around the capital Ouagadougou is probably because of land use intensity and degradation, and higher importance along the Chaı̂ne de Banfora may be linked to often shallow soils upon the sandstone with extreme water conditions. There are only a few geophytes in our study, and many of these (Echinochloa, Oryza, Sacciolepis) prefer humid conditions close to water, which makes them more independent from the precipitation gradient and explains the lack of latitudinal patterns and locally higher values for geophytes in the valleys of the Sourou and the Mouhoun. The distribution of photosynthetic types (Fig. 3) is in accordance with the findings of Taub (2000), Vogel, Fuls & Danin (1986), Ellis, Vogel & Fuls (1980) and Hattersley (1983) from other areas of the world: C4 NAD-ME grasses usually prefer drier habitats than the C4 NADP-ME grasses, with C4 PCK species in between. The high importance of C4 NADP-ME Andropogoneae in natural  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Biogeography of West African grasses 499 Sudanian savannahs can be seen in Fig. 3c which clearly reflects the borders of the protected areas of the WAP complex. The few C3 grasses are very evenly distributed. They largely coincide with the group of geophytes, and likewise, the even distribution may be due to the preference of humid habitats, making them more independent from precipitation patterns. The patterns in plant size with smaller species being replaced by taller ones towards the south (Fig. 4) are in congruence with abiotic factors such as water availability, temperature and length of growing season. On the other hand, grazing is also known as a factor influencing the ratio of tall grasses (Diaz, Noy-Meir & Cabido, 2001) and is more pronounced in the Sahelian pastures than in the agricultural mosaics of the Sudanian zone. The area dominated by smaller species extends southwards into the area between Tenkodogo and the border with Ghana and Togo, an area of high population density with widespread cattle breeding and intensive agriculture. From experimental studies on grass traits (Oyarzabal et al., 2008) and observed abundances in main vegetation types in the study area, similarly, strong trends would also have been expected for leaf width. While narrow-leaved Aristida and Eragrostis species are dominating wide areas of the Sahel (Müller, 2003), the dominating Andropogoneae of the Sudanian savannahs (Mbayngone, 2008; HahnHadjali, Schmidt & Thiombiano, 2006) are generally more broad leaved. In our study, we could not detect such large scale patterns. Only locally, there is a lower proportion of the most narrow-leaved species at large waterbodies such as the Bagré and Kompienga dams or the Sourou valley and in the sandstone mountains west of Bobo-Dioulasso. As, in this study, we count species regardless of their abundance, less frequent species and smaller habitats such as tiger bush and forest patches contribute equally to the species pool. Devineau & Fournier (2005) showed a strong response of leaf width to woody cover, so these dense vegetation patches, smaller than the grid resolution of our remote sensing data, might have obscured the pattern prevailing in the main vegetation types. An interesting approach for future investigations would be to integrate abundance into the models, because they will reflect better the structural patterns than mere presences. Higher resolution satellite data have the potential to better reflect habitat conditions in fragmented landscapes as in Schmidt, König & Müller (2008); however, computer power and difficulties with tile-edge effects still limit their use to local studies.  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500 Acknowledgements This study was financially supported by the German Ministry of Education and Research (BMBF) in the frame of the BIOTA project (01 LC 0617D1), by the EU in the frame of the SUN project (INCO 031685) and UNDESERT project (243906) and the German federal state of Hesse by funding the Biodiversity and Climate Research Institute (BiK-F). References Akoègninou, A., Van Der Burg, W.J., Van Der Maesen, L.J.G., Adjakidje, V., Essou, J.-P., Sinsin, B. & Yédomonhan, H. (2006) Flore Analytique du Bénin. Backhuys Publishers ⁄ Wageningen University Papers, Cotonou & Wageningen. Beerling, D.J. & Osborne, C.P. (2006) The origin of the savanna biome. Glob. Change Biol. 12, 2023–2031. Brink, M. & Belay, G. (2006) Plant Resources of Tropical Africa 1–Cereals and Pulses. PROTA Foundation ⁄ Backhuys Publishers ⁄ CTA, Wageningen. Chavez, P.S. & Kwarteng, A.Y. (1989) Extracting spectral contrast in Landsat Thematic Mapper image data using selective principle component analysis. Photogramm. Eng. Remote Sensing 55, 339–348. Christin, P.A., Besnard, G., Samaritani, E., Duvall, M.R., Hodkinson, T.R., Savolainen, V. & Salamin, N. (2008) Oligocene CO2 decline promoted C-4 photosynthesis in grasses. Curr. Biol. 18, 37–43. Clayton, W.D., Harman, K.T. & Williamson, H. (2006) GrassBase– The online world grass flora. Available at: http://www.kew.org/ data/grassbase/index.html. (accessed 30 April 2008). Devineau, J.L. & Fournier, A. (2005) To what extent can simple plant biological traits account for the response of the herbaceous layer to environmental changes in fallow-savanna vegetation (West Burkina Faso, West Africa)? Flora 200, 361–375. Diaz, S., Noy-Meir, I. & Cabido, M. (2001) Can grazing response of herbaceous plants be predicted from simple vegetative traits? J. Appl. Ecol. 38, 497–508. Downton, W.J.S. (1975) The occurence of C4 photosynthesis among plants. Photosynthetica 9, 96–105. Ellis, R.P. (1977) Distribution of the Kranz Syndrom in the Southern African Eragrostoideae and Panicoideae according to bundle sheath anatomy and cytology. Agroplantae 9, 73–110. Ellis, R.P., Vogel, J.C. & Fuls, A. (1980) Photosynthetic pathways and the geographical distribution of grasses in south west Namibia. S. Afr. J. Sci. 76, 307–314. Guinko, S. (2005) Florule de la Forêt Classée du Kou (Burkina Faso). Université de Ouagadougou, Ouagadougou. Guralnick, R. & Van Cleve, J. (2005) Strengths and weaknesses of museum and national survey data sets for predicting regional species richness: comparative and combined approaches. Divers. Distrib. 11, 349–359. 500 Marco Schmidt et al. Hahn-Hadjali, K., Schmidt, M. & Thiombiano, A. (2006) Phytodiversity dynamics in pastured and protected West African savannas. In: Taxonomy and Ecology of African Plants: Their Conservation and Sustainable use–Proceedings of the 17th AETFAT Congress Addis Abeba 21–26.09.2003 (Eds S.A. Ghazanfar and H.J. Beentje). Royal Botanic Gardens, Kew. Hattersley, P.W. (1983) The distribution of C3 and C4 Grasses in Australia in relation to climate. Oecologia 57, 113–128. Higgins, S.I., Bond, W.J. & Trollope, W.S.W. (2000) Fire, resprouting and variability: a recipe for grass-tree coexistence in savanna. J. Ecol. 88, 213–229. Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G. & Jarvis, A. (2005) Very high resolution interpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978. Huete, A., Didan, K., Miura, T., Rodriguez, E.P., Gao, X. & Ferreira, L.G. (2002) Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sens. Environ. 83, 195–213. Janssen, T., Schmidt, M., Dressler, S., Hahn, K., Hien, M., Konaté, S., Lykke, A.M., Mahamane, A., Sambou, B., Sinsin, B., Thiombiano, A., Wittig, R. & Zizka, G. (2011) Addressing data property rights concerns and providing incentives for collaborative data pooling: the West African Vegetation Database approach. J. Veg. Sci. 22, 614–620. Klopper, R.R., Gautier, L., Chatelain, C., Smith, G.F. & Spichiger, R. (2007) Floristics of the angiosperm flora of sub-Saharan Africa: an analysis of the African Plant Checklist and Database. Taxon 56, 201–208. König, K., Schmidt, M. & Müller, J.V. (2009) Modelling species distributions with high resolution remote sensing data to delineate patterns of plant diversity in the Sahel zone of Burkina Faso. In: Recent Advances in Remote Sensing and Geoinformation Processing for Land Degradation Assessment (Eds T. Hill and A. Röder). Taylor & Francis, London. Liu, C.R., Berry, P.M., Dawson, T.P. & Pearson, R.G. (2005) Selecting thresholds of occurrence in the prediction of species distributions. Ecography 28, 385–393. Mbayngone, E. (2008) Flore et végétation de la réserve partielle de faune de Pama, Sud-Est du Burkina Faso. PhD thesis, University of Ouagadougou, Ouagadougou, Burkina Faso. Mbayngone, E., Schmidt, M., Hahn-Hadjali, K., Thiombiano, A. & Guinko, S. (2008) Magnoliophyta of the partial faunal reserve of Pama, Burkina Faso. Checklist 4, 251–266. Müller, J.V. (2003) Zur Vegetationsökologie Der Savannenlandschaften Im Sahel Burkina Fasos. PhD thesis, Goethe University, Frankfurt am Main. Ouédraogo, O., Schmidt, M., Thiombiano, A., Hahn, K., Guinko, S. & Zizka, G. (2011) Magnoliophyta, Arly National park, Tapoa, Burkina Faso. Checklist 7, 85–100. Oyarzabal, M., Paruelo, J.M., Federico, P., Oesterheld, M. & Lauenroth, W.K. (2008) Trait differences between grass species along a climatic gradient in South and North America. J. Veg. Sci. 19, 183–1U1. Pedersen, J. & Benjaminsen, T.A. (2008) One leg or two? Food security and pastoralism in the northern Sahel Hum. Ecol. 36, 43–57. Poilecot, P. (1995) Les Poaceae de Côte-d’Ivoire. Conservatoire et Jardin Botaniques, Geneva. Poilecot, P. (1999) Les Poaceae du Niger. Conservatoire et Jardin Botaniques, Geneva. Saatchi, S., Buermann, W., Ter Steege, H., Mori, S. & Smith, T.B. (2008) Modeling distribution of Amazonian tree species and diversity using remote sensing measurements. Remote Sens. Environ. 112, 2000–2017. Schmidt, M. (2006) Pflanzenvielfalt in Burkina Faso–Analyse, Modellierung und Dokumentation. PhD thesis, Goethe University, Frankfurt am Main. Schmidt, M., König, K. & Müller, J.V. (2008) Modelling species richness and life form composition in Sahelian Burkina Faso with remote sensing data. J. Arid Environ. 72, 1506–1517. Schmidt, M., Kreft, H., Thiombiano, A. & Zizka, G. (2005) Herbarium collections and field data-based plant diversity maps for Burkina Faso. Divers. Distrib. 11, 509–516. Schmidt, M., Ouédraogo, A., Thiombiano, A., Dressler, S. & Zizka, G. (2010a) Assessment of the flora of Burkina Faso. In: Systematics and Conservation of African Plants (Eds X. Van der burgt, J. Van der maesen and J.-M. Onana). Kew Publishing, Royal Botanic Gardens, Kew. Schmidt, M., Thiombiano, A., Dressler, S., Hahn-Hadjali, K., Guinko, S. & Zizka, G. (2010b) Phytodiversity data–strengths and weaknesses. A comparison of collection and relevé data from Burkina Faso. In: Systematics and Conservation of African Plants (Eds X. Van der burgt, J. Van der maesen and J.-M. Onana). Kew Publishing, Royal Botanic Gardens, Kew. Scholz, H. & Scholz, U. (1983) Flore Descriptive des Cypéracées et Graminées du Togo. Cramer, Vaduz. Stebbins, G.L. (1981) Coevolution of grasses and herbivores. Ann. Mo. Bot. Gard. 68, 75–86. Stockwell, D.R.B. & Peterson, A.T. (2002) Effects of sample size on accuracy of species distribution models. Ecol. Model. 148, 1–13. Taub, D.R. (2000) Climate and the US distribution of C-4 grass subfamilies and decarboxylation variants of C-4 photosynthesis. Am. J. Bot. 87, 1211–1215. Vogel, J.C., Fuls, A. & Danin, A. (1986) Geographical and environmental distribution of C3 und C4 grasses in the Sinai, Negev, and Judean deserts. Oecologia 70, 258–265. (Manuscript accepted 5 July 2011) doi: 10.1111/j.1365-2028.2011.01283.x  2011 Blackwell Publishing Ltd, Afr. J. Ecol., 49, 490–500