Evolution under domestication: contrasting ... - Wiley Online Library

Research

Evolution under domestication: contrasting functional morphology of seedlings in domesticated cassava and its closest wild relatives Blackwell Publishing, Ltd.

Benoît Pujol1, Gilda Mühlen2, Nancy Garwood3, Yael Horoszowski1, Emmanuel J. P. Douzery4 and Doyle McKey1 1

Department of Population Biology, Centre for Functional and Evolutionary Ecology (CEFE, UMR 5175 CNRS), 1919, route de Mende, 34293 Montpellier

Cedex 05, France; 2Federal University of Rondônia, Department of Agronomy, Avenue Norte-Sul, 7300, CEP 78 987–000 Rolim de Moura, Rondônia, Brazil; 3

Department of Botany, the Natural History Museum, Cromwell Road, London SW7 5BD, UK; 4Institute of Evolution Sciences of Montpellier, UMR 5554

CNRS, University Montpellier II, Place Eugène Bataillon 34090 Montpellier, France

Summary Author for correspondence: Benoît Pujol Tel: +33 (0)4 67613299 Fax: +33 (0)4 67412138 Email: [email protected] Received: 14 September 2004 Accepted: 25 October 2004

• Although cassava (Manihot esculenta ssp. esculenta) is asexually propagated, farmers incorporate plants from seedlings into planting stocks. These products of sex are exposed to selection, which in agricultural environments should favour rapid growth. • To examine whether seedling morphology has evolved under domestication, we compared domesticated cassava, its wild progenitor (M. esculenta ssp. flabellifolia) and their sister species (M. pruinosa) under controlled conditions. Field observations complemented laboratory study. • In both wild taxa, the hypocotyl did not elongate (hypogeal germination) and cotyledons remained enclosed in the testa. In domesticated cassava, the hypocotyl elongated (epigeal germination), and cotyledons emerged and became foliaceous. The difference in hypocotyl elongation was fixed, whereas cotyledon morphology varied with environmental conditions in M. pruinosa. • Comparative analysis suggests that epigeal germination is primitive in Manihot, that the lineage including wild ancestors of cassava evolved hypogeal germination – which confers greater tolerance to risks in their savanna environment – and that with domestication, there was a reversion to epigeal germination and photosynthetic cotyledons, traits conferring high initial growth rates in agricultural habitats. Key words: cassava, domestication, Euphorbiaceae, seedling functional morphology, wild Manihot. New Phytologist (2005) 166: 305–318 © New Phytologist (2005) doi: 10.1111/j.1469-8137.2004.01295.x

Introduction Early seedling morphological traits are important in the regeneration strategies of plants. Traits related to the function of cotyledons, which are correlated with many other ecophysiological and life-history traits (Garwood, 1996; Kitajima & Fenner, 2000), appear particularly important. These include whether the hypocotyl elongates (epigeal germination) or not (hypogeal germination); whether the cotyledons remain enclosed within the testa (cryptocotylar) or emerge from it (phanerocotylar); and whether the cotyledons are reserve

www.newphytologist.org

organs or foliaceous, photosynthetic organs. Combinations of these traits constitute seedling functional types that vary among species with different regeneration niches, and these patterns show evolutionary convergence at the community level worldwide (Ibarra-Manriquez et al., 2001). Six seedling types based on cotyledonar traits have been established (Ng, 1978, 1991; Hladik & Miquel, 1990; Garwood, 1996; Ibarra-Manriquez et al., 2001), cotyledons being always reserve organs when cryptocotylar. Seedling traits are evolutionarily conservative, reflecting phylogenetic niche conservatism (Ibarra-Manriquez et al., 2001). Within a genus generally only one of these

305

306 Research

combinations is found (Garwood, 1996). In some genera, epigeal and hypogeal germination distinguish different subgenera (Essig, 1987). This structured diversity reflects the strong selective pressures acting on seedling traits. In general, seedlings with reserve cotyledons have inherently slow relative growth rates (RGR) because initial leaf area relative to whole plant mass is small, while seedlings with foliaceous, photosynthetic cotyledons have inherently rapid RGR because initial photosynthetic surface relative to whole plant mass is large (Kitajima & Fenner, 2000). Seedlings of this latter group also often require higher rates of resource supply (light, water, nitrogen, mineral nutrients) to achieve their full potential for rapid growth. However, seedlings with storage cotyledons have more modest requirements and appear better able to tolerate stress, not only shade (Kitajima & Fenner, 2000), but also physical damage to shoots, via rapid resprouting (Hoshizaki et al., 1997) from dormant axillary buds borne by hypogeal cotyledons and by cataphylls (Burtt, 1972). It is sometimes difficult to separate the effects of cotyledonar function from those of size of seeds and cotyledons, because seeds of species with reserve cotyledons are usually larger than those of species with foliaceous cotyledons (Kitajima & Fenner, 2000). Epigeal germination predominates in most habitats. Although sample sizes are small and geographic coverage limited, hypogeal germination appears to reach relatively high frequency in fire-prone savanna habitats (Rizzini, 1965; Burtt, 1972; Jackson, 1974), perhaps because it confers greater tolerance of seedlings to fire or drought, especially when combined with seed burial. In Brazilian cerrado woody vegetation, the frequency of hypogeal germination was significantly higher in shrub (70%) than in tree species (20%) (Rizzini, 1965, association with life form analysed by Garwood (1996)). Holding stored reserves underground (in reserve cotyledons, or transferred to the hypocotyl) could reduce the risk of lethal damage to aerial parts of seedlings by fire or by drought stress (Rizzini, 1965; Burtt, 1972; Jackson, 1974). This could be of greater importance to shrubs, which colonize early successional vegetation in which fires are more frequent and seedlings more exposed to drying conditions, than to trees. Domestication is often accompanied by dramatic changes in the regeneration environment of the domesticate compared to wild ancestors. Natural selection should have acted on traits of seedlings that resulted in the plant’s being better adapted to agricultural environments. We examined this question in cassava (Manihot esculenta Crantz ssp. esculenta, Euphorbiaceae). We compared seedling morphology in cassava, its wild progenitor (Manihot esculenta Crantz ssp. flabellifolia (Pohl) Ciferri) and their closest wild relative, Manihot pruinosa Pohl, considered the sister species of M. esculenta (Allem et al., 2001), in an attempt to answer the following questions: (i) Was domestication accompanied by evolutionary change in seedling morphology? (ii) Can contrasting seedling morphologies be understood in terms of divergent adaptation to contrasting environments? Although

New Phytologist (2005) 166: 305–318

cassava is actively propagated by farmers in strictly clonal fashion (by stem cuttings), recent work has shown that in the ecological context in which the crop evolved – cultivation by Amerindians in slash and burn agricultural systems – cassava has a mixed reproductive system that includes sex. The plant has retained sexual fertility, and Amerindian farmers regularly incorporate ‘volunteer’ plants issued from seeds into their stock of cuttings (Elias et al., 2000a,b). The regular inclusion of these products of unmanaged sex means that selection can still act on traits related to sexual reproduction in this ‘clonally propagated’ plant. Considerable evidence now exists for the importance of sexual reproduction in the demography and genetics of populations of cassava managed by Amerindians (Elias et al., 2000a,b, 2001a,b; Sambatti et al., 2001; Pujol et al. 2004). However, there is little information on the ecology of sexual reproduction in these populations (Elias & McKey, 2000; Pujol et al., 2002). Information on this topic could provide insights into the evolution of cassava under domestication. To place the comparative study of these three taxa of Manihot into a broader phylogenetic perspective, we compiled observations of seedling morphology in the family Euphorbiaceae, from our own work and from the literature, to examine the following questions: (i) What seedling morphologies characterize genera closely related to Manihot and thus might be ancestral? (ii) What seedling morphologies characterize other species of Manihot less closely related to the domesticate than the one examined here? (iii) In the Euphorbiaceae, what is the diversity of seedling types observed, and is seedling morphology phylogenetically conservative?

Materials and Methods Study species The first taxon we examined is Manihot esculenta ssp. esculenta, domesticated cassava. Despite its importance in the diets of half a billion people in tropical regions, relatively little attention has been focused on the evolution of this crop. Many aspects of its domestication are poorly understood. The second taxon we examined is M. esculenta ssp. flabellifolia, found along the southern rim of Amazonia. It is widely considered to be the ancestor of domesticated cassava (Allem et al., 2001) and considerable molecular evidence supports this hypothesis (Roa et al., 1997, 2000; Olsen & Schaal, 1999, 2001; Colombo et al., 2000). It is a shrub, sometimes scandent, of ecotone between cerrado and tropical moist forest (Piperno & Pearsall, 1998; Olsen, 2002). The typical habitat of M. esculenta ssp. flabellifolia is savanna-forest ecotone that is characterized by a continuum in the frequency and intensity of disturbance, ranging from small forest gaps, sometimes with partial vegetation cover, to large clearings created by disturbances such as fire. Here we will refer to this wild taxon interchangeably as ‘wild progenitor’ and M. esculenta ssp. flabellifolia.

www.newphytologist.org © New Phytologist (2005)

Research

The third taxon we examined is Manihot pruinosa, a shrub found along the southern, eastern (and northern) rim of Amazonia. Manihot includes just under 100 currently recognized species (Rogers & Appan, 1973). Among these, molecular phylogenetic studies indicate that M. pruinosa is the sister group to M. esculenta ssp. flabellifolia plus M. esculenta ssp. esculenta (Olsen & Schaal, 1999, 2001; G. Léotard, unpublished). We studied populations of M. pruinosa from the Guianas, where they have been previously known as M. surinamensis, M. tristis, or M. sprucei (Hoock, 1971; Rogers & Appan, 1973) or included in M. esculenta ssp. flabellifolia (Allem et al., 2001). Molecular work on these populations using the nuclear gene G3pdh indicates that they should be included in M. pruinosa (G. Léotard, unpublished). The typical habitat of M. pruinosa is cerrado vegetation and savanna-forest ecotones. These relatively dry, seasonal habitats are fire-prone (Rizzini, 1965; Oliveira & Marquis, 2002). Plant material We collected mature capsules from two populations of M. pruinosa in French Guiana, two populations of M. esculenta ssp. flabellifolia from Rolim de Moura, Rondônia State (Brazil) and three populations of domesticated cassava, two in French Guiana and one in Rondônia. Within each population, capsules were collected opportunistically and mixed without regard to identity of the maternal parent. In French Guiana, seeds of M. pruinosa were collected in two sites in savannas of the coastal region (site 1 (04°58.124′N, 52°26.314′W), between Cayenne and Macouria; site 8 (05°07.473′N, 52°39.730′W), further to the west near Kourou). In Rondônia, seeds of M. esculenta ssp. flabellifolia were collected in two sites (site 1, Linha 176a (11°44.573′ S, 61°51.091′ W ); site 2, Linha 176b (11°44.573′ S, 61°51.091′ W)). Seeds of domesticated cassava were collected from polyvarietal populations managed in traditional slashand-burn agroecosystems, by Palikur Amerindians in French Guiana (St. Georges de l’Oyapock (03°54′ N, 51°48′ W) and Macouria (05°01′ N, 52°28′ W)), and by small farmers (colonos) in Rondônia (Sidcley 11°44.580′ S, 61°51.091′ W ). Capsules were collected in December 2001 in St. Georges and in January 2002 in Macouria (domesticated cassava), March 2001 in savannas of coastal French Guiana (M. pruinosa) and May 2004 in Rondônia (domesticated cassava and M. esculenta ssp. flabellifolia). Capsules were allowed to dry and dehisce. Seeds were stored in darkness at room temperature until they were used in experiments. Seed mass Seeds were weighed before being set out to germinate. We tested for an effect of taxon on seed mass within each area of origin, and for an effect of population (nested within taxon) for the French Guiana seeds. We then tested for differences between the two areas of origin for each category, wild

© New Phytologist (2005) www.newphytologist.org

(M. pruinosa vs M. esculenta ssp. flabellifolia) and domesticated. Analysis included 392 seeds of M. pruinosa and 135 seeds of domesticated cassava from French Guiana, 95 seeds of domesticated cassava and 79 seeds of its wild progenitor from Rondônia. Because no simple transformation permitted obtaining normality of residuals, these effects were tested using the Kruskal–Wallis test on Wilcoxon rank sums (PROC NPAR1WAY (SAS, 1996)). Experimental design and estimators of morphological variation Seeds were set out to germinate in two experiments (each carried out in the country of origin of seeds) in March 2002 in France (CNRS campus, Montpellier) and in May 2004 in Brazil (Instituto Agronômico de Campinas) using the methods of Pujol et al. (2002) and applying a combination of treatments chosen to maximize germination rate (scarification, or dry heat (55–65°C) for 3 d, followed by incubation in darkness at 35°C). A seed was counted as having germinated when the radicle was visible between the two halves of the split testa. Date of germination was noted for each seedling. As in the previous study (Pujol et al., 2002), seeds of domesticated cassava began to germinate more rapidly than did those of wild cassava. Morphological traits were measured on a subset of seedlings that germinated at about the same time. These seedlings were measured once, at 8 d after germination. On each seedling we measured the length of the hypocotyl (from the collet to the cotyledonar node), the epicotyl (from the cotyledonar node to the apical meristem), the entire shoot (hypocotyl plus epicotyl), and the cotyledon petioles. From these measures we calculated two ratios, the hypocotylar index (HCI; length of the hypocotyl as a proportion of total shoot length (hypocotyl + epicotyl)) and the cotyledon petiole index (CPI; mean length of the two cotyledon petioles in relation to total shoot length). The sum of the two indices (CEI, for cotyledon exposure index) thus indicates the height at which the cotyledons are placed relative to the total height of the seedling’s shoot. We also noted whether cotyledons were cryptocotylar (remaining within the split testa) or phanerocotylar (emerged from the testa) and recorded their colour (green or unpigmented) and form (reserve-type or foliaceous). While cotyledons are always expected to be reserve-type in the case of cryptocotylar germination, all other combinations of traits are independent. We measured a total of 49 seedlings of M. pruinosa, five seedlings of M. esculenta ssp. flabellifolia and 53 seedlings of domesticated cassava (45 from French Guiana and eight from Brazil). Statistical comparison of laboratory-grown seedlings We analysed variation in total shoot length of seedlings and in the indices presented above, analysing Brazilian and French Guiana samples separately. In the former group, we tested for


307

308 Research

an effect of taxon using the Kruskal–Wallis Test on Wilcoxon rank sums (PROC NPAR1WAY (SAS, 1996)). We used the same test to compare representatives of the same category (wild or domesticated) from the two areas of origin. In the French Guiana sample, we tested for the effects of taxon (and population nested within taxon) using a generalized linear model (PROC GLM, unequal sample sizes (SAS, 1996)). We verified the normality of residuals by the Shapiro-Wilk and the Kolmogorov-Smirnov tests for normality (PROC UNIVARIATE (SAS, 1996)). For seedlings from both French Guiana and Brazil, lengths of the hypocotyl and the cotyledon petiole were transformed before the analysis to the logarithm of the hypocotylar index (HCI) and the logarithm of the cotyledon petiole index (CPI), respectively, to remove the effects of variation in overall size among seedlings of each category. Field observations We also examined very young seedlings in two natural populations of M. pruinosa to confirm laboratory observations. One of these populations (5°06.668′ N, 52°37.085′ W ) was near Kourou, along the road leading from RN1 to the town of Guatemala, where a recent small fire had led to the germination of a few seedlings near a single mature plant, in April 2003. The second population (4°58.553′ N, 52°26.424′ W), was located near a country inn (Grand Blanc), where ploughing of an area of savanna containing a small, dense population of wild cassava had led to the germination of a large number of seedlings, of which 70 were dug up and examined in August 2003. We compared depth of burial and length of cotyledon petioles of two groups of seedlings (see below, near the end of the section ‘Development of shoot parts in seedlings’, in the Results) whose cotyledons were exposed (phanerocotylar) or enclosed within the testa (cryptocotylar), using Kruskal–Wallis tests on Wilcoxon rank sums (PROC NPAR1WAY (SAS, 1996)). We also conducted opportunistic observations of seedlings of domesticated cassava in fields of Palikur Amerindian farmers. Survey of seedling morphology in the Euphorbiaceae To permit comparative analysis, we compiled data on seedling morphology in Euphorbiaceae primarily from published information, supplemented with data from seedling collections from Panama and Ecuador (N. Garwood, unpublished; see Garwood, 1995) for methods used to identify this material). For each taxon we noted the state of three cotyledon characters (cotyledons phanerocotylar or cryptocotylar, germination epigeal or hypogeal, cotyledons reserve-type or foliaceous) and other seedling traits. Generic and subfamilial nomenclature of Euphorbiaceae s.l. follows Radcliff-Smith (2001). Subfamilies or tribes recently elevated to family status by APG (2003), Judd et al. (2002), and Stevens (2001) are noted.


Phylogeny reconstruction To reconstruct the phylogeny of cassava and its wild relatives whose seedling morphology has been characterized, three datasets were available: (i) the 304 AFLP characters from Roa et al. (1997); (ii) the 124 microsatellite characters from Roa et al. (2000); and (iii) the G3pdh sequences of Olsen & Schaal (1999) and G. Léotard (unpublished). Because the sets of taxa studied by these authors were partially overlapping, we adopted a supertree approach (Bininda-Emonds, 2004) to depict the phylogenetic relationships among Jatropha ( JAT), Cnidoscolus (CNI), and Manihot aesculifolia (AES), M. carthaginensis (CAR), M. esculenta ssp. esculenta (ESC), M. esculenta ssp. flabellifolia (FLA), and M. pruinosa (PRU). First, we reanalysed the AFLP and microsatellite data under the maximum parsimony criterion, using * (Swofford, 1998), version 4.0b10. Binary matrices with a presence/absence coding of AFLP bands and microsatellite alleles were submitted to 100 replicates of random taxa addition followed by heuristic searches, and tree bissection-reconnection branch swapping. Here, Manihot aesculifolia was considered as the outgroup relative to M. carthaginensis, M. flabellifolia, and M. esculenta individuals on the basis of its cpDNA and rDNA variability (Fregene et al., 1994), and divergent AFLP patterns (Roa et al., 1997). The strict consensus of the resulting best trees – 356 equally parsimonious ones for AFLP (length = 960 steps), and 3921 equally parsimonious ones for microsatellites (length = 442 steps) – were taken as source (initial) topologies. All published trees that show the close relationship of domesticated cassava to M. esculenta ssp. flabellifolia depict the latter taxon as paraphyletic, with domesticated cassava nested within it as either a monophyletic (Roa et al., 1997, 2000; Olsen & Schaal, 2001) or polyphyletic (Olsen & Schaal, 1999) group. To depict this paraphyly of M. esculenta ssp. flabellifolia due to the inclusion of domesticated cassava, two different M. esculenta ssp. flabellifolia taxa were included, ESC-FLA 423 and ESC-FLA 430 (Roa et al., 1997, 2000). We also considered the maximum likelihood topology reconstructed from the G3pdh sequences of Olsen & Schaal (1999) and G. Léotard (unpublished) as another source topology. To depict the paraphyly of M. esculenta ssp. flabellifolia due to M. pruinosa (see also the microsatellite study of Olsen & Schaal, 2001), we incorporated two different M. pruinosa taxa, haplotypes L (Olsen & Schaal, 1999) and S-1 (G. Léotard, unpublished). Similarly, two M. esculenta ssp. flabellifolia were considered, namely haplotypes E and U (Olsen & Schaal, 1999). For these three source topologies, we only selected taxa for which seedling morphology has been scored. Second, we adopted the matrix representation with parsimony (MRP) procedure (Ragan, 1992) for the binary recoding of nodes of the three source trees, while restricting the topological information at the species level only. For M. esculenta ssp. flabellifolia, we merged the AFLP information brought by ESC-FLA 423 and ESC-FLA 430 with one


Research

of the G3pdh haplotypes, E and U, respectively. A total of 10 MRP characters was finally scored for nine taxa. Characters 1–9 arise from the source topologies. Character 10 is from Miller & Webster (1962) who suggest a separation between Cnidoscolus and Jatropha relative to Manihot. A branch and bound maximum parsimony search was then performed under *, considering all binary characters as irreversible. A bootstrap analysis (Felsenstein, 1985) with 1000 replicates was also conducted. This analysis should be viewed only as an attempt to put reliability indices on the supertree, keeping in mind that MRP data are not made of the initial molecular characters but of the recoded nodes of the corresponding source topologies.

Results Seed mass For French Guiana samples, seed mass was 6% higher in domesticated cassava (96 ± 26 mg) than in M. pruinosa (80 ± 15 mg), and for Brazilian samples, seed mass was similar in domesticated cassava (89 ± 28 mg) and in the crop’s wild progenitor (87 ± 19 mg) (Table 1). The difference was highly significant for French Guiana samples (Kruskal–Wallis test on Wilcoxon rank sums, χ12 = 51.98, P < 0.0001), and nonsignificant for the sample from Brazil χ12 = 0.08, P = 0.77). Seed mass also differed significantly between the two populations within each taxon in French Guiana but this variation was lower than that between taxa; seed mass was not different between the two populations of M. esculenta ssp. flabellifolia (results not shown). Development of shoot parts of seedlings Table 1 presents dimensions of different organs of seedlings 8 d after germination in both experiments. Total length (collet to apical meristem) of the main axis gives an idea of the overall size of seedlings. Total length of the shoot was significantly

greater in seedlings of domesticated cassava from French Guiana than in those of M. pruinosa but not significantly different between M. esculenta ssp. flabellifolia and domesticated cassava from Brazil (Table 1), reflecting the respective differences in seed mass. For the French Guiana sample, there were significant effects of population within taxon on some quantitative traits (length of hypocotyl, log-linearized HCI, length of cotyledon petioles, and CEI index (results not shown)). Nevertheless, taxon (wild or domesticated) had a far greater effect on seedling morphology. All the seedling dimensions measured and indices calculated showed highly significant differences, and often strong contrasts, between the seedlings of domesticated cassava and its wild relatives (Table 1). The epicotyl was longer in seedlings of the two wild taxa (this difference was only a trend in the small Brazilian sample), whereas both the hypocotyl and the cotyledon petioles were much longer in domesticated cassava in both regions sampled. Lengths of hypocotyl and cotyledon petioles as proportions of total shoot length (HCI and CPI indices, respectively) were also much greater in domesticated cassava than in its wild relatives, in both regions. In both wild taxa, in the laboratory studies (conducted in darkness), the cotyledons remained more or less enclosed in the split testa, did not expand, and were unpigmented. Cotyledons bore short petioles. The hypocotyl exhibited little elongation, whereas the epicotyl elongated considerably. Cataphylls, with axillary buds, were present on the epicotyl. The first true leaf was the first photosynthetic organ of the plant. Thus in both M. pruinosa and M. esculenta ssp. flabellifolia, cotyledons are cryptocotylar, hypogeal and nonfoliaceous (Fig. 1a,b, respectively). The two wild relatives, originating from the northern and southern rim of Amazonia and separated by over 2100 km, showed no significant differences in any of the seedling dimensions we examined. Seedlings of domesticated cassava showed traits opposite to those of its wild relatives (Table 1). The hypocotyl grew about six to seven times longer than in wild seedlings. The cotyledons always emerged from the split testa, expanded to become thin and foliaceous, even in darkness, and became green when

Table 1 Morphological variables for seedlings of wild and domesticated cassava measured 8 d after germination

Variable

French Guiana Domesticate M. pruinosa

Difference

Domesticate

Wild progenitor

Difference

Sample size Total shoot Epicotyl Hypocotyl Cotyledon petioles HCI CPI CEI

45 128.07 ± 28.06 75.00 ± 28.36 53.07 ± 22.02 37.78 ± 20.06 0.42 ± 0.17 0.30 ± 0.14 0.72 ± 0.23

P = 0.03 P = 0.0002 P < 0.0001 P < 0.0001 P < 0.0001 P < 0.0001 P < 0.0001

8 125.75 ± 27.25 83.25 ± 33.28 41.50 ± 13.08 72 ± 29.95 0.33 ± 0.13 0.59 ± 0.24 0.91 ± 0.24

5 110.80 ± 30.60 105.00 ± 29.76 5.80 ± 1.30 5.60 ± 1.52 0.05 ± 0.01 0.06 ± 0.03 0.11 ± 0.02

P = 0.34 P = 0.14 P = 0.003 P = 0.003 P = 0.003 P = 0.003 P = 0.003

49 109.02 ± 39.16 100.02 ± 38.46 9.00 ± 3.98 6.41 ± 2.57 0.09 ± 0.05 0.07 ± 0.04 0.16 ± 0.08

Brazil

Length of the total shoot, the epicotyl, the hypocotyl, and cotyledon petioles (all expressed in mm) and indices of development (expressed as a fraction of the total shoot length) of the hypocotyl (HCI), cotyledon petioles (CPI), and the cotyledonar exposure (CEI, sum of HCI and CPI). All values are means ± SD. P values of differences between taxa in each experiment (wild or domesticated cassava) are presented.



309

310 Research

Fig. 1 Seedlings of the three Manihot spp. studied, showing the characteristic differences in morphology of the cotyledons. (a) Manihot pruinosa Pohl, from a population alongside the road from Kourou to Guatemala (see text). c., Cataphyll; co, collet; cp, cotyledon petiole; e, epicotyl; h, hypocotyl; l, first true leaves; t, testa (in which cotyledons are enclosed). (b) Manihot esculenta Crantz ssp. flabellifolia, from a population in Rondônia state, Brazil (see text); symbols as in (a). (c) Domesticated cassava, M. esculenta Crantz ssp. esculenta, from Palikur farms near Macouria, co, Collet; cp, cotyledon petiole; ct, cotyledon; e, epicotyl; h, hypocotyl; l, first true leaves. (d) M. pruinosa, set out to germinate from very shallow depth in the laboratory, with PHF morphology; symbols as in (c) In each photograph, the arrow indicates the level of the soil surface (except for the seedling shown in (b), which originated from the laboratory germination experiment). Bar, 1 cm.

exposed to light. Although their areas were not measured, they were clearly much larger in surface area than those of wild seedlings, which remained enclosed in the testa. Cotyledon petioles were much longer in seedlings of domesticated cassava. The epicotyl elongated, but less than in seedlings of its wild relatives, and lacked cataphylls. As a result of all these traits, cotyledons of domesticated cassava are lifted much higher above the ground (higher CEI) and are the seedling’s first photosynthetic organs. Thus in domesticated cassava, germination is epigeal; cotyledons are phanerocotylar and foliaceous (Fig. 1c). Field observations show that they are persistent, often remaining functional on seedlings that have produced 10 or more true leaves. Observations of very young seedlings of M. pruinosa in two natural populations near Kourou and Grand Blanc confirmed that the hypocotyl does not elongate. However, the behaviour of the cotyledons was more variable under field conditions than in the laboratory. Cotyledons of all seedlings at Kourou and most seedlings from Grand Blanc remained buried in the soil and were cryptocotylar and nonfoliaceous (Fig. 1a). These cotyledons had in some cases been shed, and were almost always absent on seedlings with >10 leaves. However, six of the 70 seedlings at Grand Blanc had exposed, green, foliaceous cotyledons. Excavation of these seedlings showed that they had all germinated from a very shallow depth in the soil (mean ± , 10.0 ± 4.4 mm, n = 6), compared to those in which cotyledons had remained enclosed in the testa in the soil (40.9 ± 24.3 mm, n = 64; χ12 = 13.1, P = 0.0003). Their cotyledons had been lifted above the soil surface solely by their petioles, which were longer (mean ± , 20.2 ± 8.7 mm, n = 6)


than in the seedlings whose cotyledons remained buried (9.0 ± 4.5 mm, n = 28 seedlings which still had cotyledons attached; χ12 = 8.2, P = 0.004), as the hypocotyl did not elongate. In these seedlings, the epicotyl did not bear cataphylls, true leaves being borne at the first and all successive nodes. Thus, in M. pruinosa, germination was always hypogeal. Cotyledons were usually cryptocotylar and nonfoliaceous, but in a small fraction of seedlings germinating from shallow depth, cotyledons were phanerocotylar and foliaceous. We were able to reproduce this morphology by setting seeds to germinate at very shallow depth in the laboratory (Fig. 1d). In contrast, field observations of domesticated cassava at a total of six sites found only seedlings with epigeal germination and phanerocotylar, foliaceous cotyledons (Fig. 1c). In seedlings observed in the field, the cotyledonar node was either above the soil surface (as indicated in Fig. 1c) or ≤ 5 mm beneath the soil surface. Comparative survey of seedling morphology in the Euphorbiaceae The exhaustive census of available data on seedling morphology in the five subfamilies of Euphorbiaceae s.l. (sensu RadcliffSmith, 2001) permitted us to characterize 335 species in 100 genera, using three pairs of major cotyledon traits: phanerocotylar (P)/cryptocotylar (C), epigeal (E)/hypogeal (H), and foliaceous (F)/reserve function (R) (Table 2). Four trait combinations were found in Euphorbiaceae s.l. PEF, CER, PER, and PHF, with PEF the most frequent. PHR, found in other tropical floras (Garwood, 1996), was absent.


Research Table 2 Subfamilial and generic classification of seedling morphology in Euphorbiaceae

Genus1

Seedling traits2,3

n of spp. studied

Subfamily Acalyphoideae (31 of 116 genera) Acalypha [450] P-E-F Adelia [13] P-E-F Alchornea [60] P-E-F Alchorneopsis [3] P-E-F Blumeodendron [6] C-E-R (h) Chaetocarpus [13] P-E-F P-E-? Cheilosa [1] P-E-F Claoxylon [80] P-E-F Cleidion [25] P-E-F Conceveiba [12] P-E-F Dalechampia [110] P-E-F Discoglypremna [1] P-E-F Erythrococca [50] P-E-F Galearia [6] P-E-F Homonoia [2] P-E-F Koilodepas [12] P?-E-? Macaranga [280] P-E-F P?-E-? Mallotus [150] P-E-F P?-E-? Mareya [3] P-E-F Mareyopsis [2] C-E-R (h) Melanolepis [2] P-E-F Mercurialis [8] P-E-F P-E-F to C-H-R (h) Microdesmis [10] P-E-F Neoscortechinia [6] P-E-R (s) Omphalea [20] C-E-R (h) C-H-R (h) P-E-F Panda6 [1] Ptychopixis [12] C-E-R (h) Ricinus [1] P-E-F Rockinghamia [2] P-E-F Plukenetia [15] (as Tetracarpidium) C-E-R (h) Trewia [2] P-E-F

6 1 9 1 2 1 1 1 2 1 1 4 1 1 1 1 2 15 7 12 1 1 1 1 1 1 2 1 1 1 1 2 1 1 1 1

Subfamily Crotonoideae (20 of 67 genera) Aleurites [2] P-E-F Baliospermum [12] P-E-F Baloghia [15] P-E-F Cnidoscolus [50] P-E-F Croton [800] P-E-F Dimorphocalyx [20] P-E-F Elateriospermum [1] P-E-R (s) Endospermum [12] P-E-F Paracroton [4] (as Fahrenheitia) P-E-F Fontainea [6] P-E-F Hevea [17] C-H-R (h) Hylandia [3] C-E-R (h) Jatropha [175] P-E-F C-E-R (h) (φ) Klaineanthus [1] P-E-F Manihot [100] P-E-F P-H-F Ricinodendron [1] P-E-F Sagotia [2] P-E-F Tetrorchidium [20] P-E-F P-E-F/R? (φ) Trigonostemon [60] P-E-F Vernica [3] (as Aleurites) C-E-R (h) C/P-E-R (h)

1 1 2 1 17 1 1 3 1 1 3 1 4 1 1 5 1 2 1 2 1 1 1 1


Other notes4

cat-

variable

cat+ cat+ cat-

cat-

– cathypcpet++

cat-

Seedling references5

12, 21, 22, 32 12 2, 8, 12, 16, 17, 18, 24, 31, 32, 40 8 29, 38 14 29 29 16, 29 16 42 12, 22 15, 18, 25, 34 18 29 29 4 4, 15, 16, 18, 24, 25, 29, 38 4 4, 15, 21, 29, 35, 38 4 25 15, 24 5 7, 22, 27 7, 27, 28 18, 25, 29 29 1, 17 12 18, 25 29 7, 8, 10, 22, 30 16 31 35, 38 5, 8, 16 5 16 17 2, 4, 8, 12, 15, 16, 17, 21, 24, 25, 29, 32, 38 16 29, 38 4, 16, 29 4, 29 16 3, 8, 12, 14, 22, 40, 42 16 8, 19, 22 19, 22 15, 24 40 20 14, 15, 25, 34 13 17, 25 32 29 36, 41 14


311

312 Research Table 2 continued

Genus1

Seedling traits2,3

Subfamily Euphorbioideae (17 of 39 genera) Anthostema [3] P-E-F Balakata [2] (as Sapium) P-E-F/R? (φ) Chamaesyce [250] P-E-F Dichostemma [1] C-E-R (h) Elaeophorbia [4] P-E-F/R? (φ) Euphorbia [2000] P-E-F C/P-E-R (h) C-E-R (h) Euphorbia (succulent spp.) P-E-F (w) C-E-R (h, φ) Excoecaria [35] P-E-F Glycydendron [2] P-E-F Hippomane [3] P-E-F Homalanthus [25] (incl. Omalanthus) P-E-F P?-E-? Hura [3] C-E-R (h) or P-E-F (φ) Mabea [39] P-E-F P-E-F/R? (φ) or C-?-? (h) Pimelodendron [8] C-E-R (h) Plagiostyles [1] C-E-R (h) Sapium [25] P-E-F Shirakiaopsis [6] (as Sapium) P-E-F P-E-F/R? (φ) Triadica [4] (as Sapium) P-E-F

n of spp. studied

1 1 2 1 1 15 2 1 16 1 1 1 1 1 1 1 – 1 1 – 1 1 4 1 1 2

Other notes4

cat-

cat-


18, 25 29 12 15, 24 25 6, 7, 22, 28 22 14 37 37 35 42 9 16 4 8, 12, 14, 40 32 42 12 23 4, 29 15, 24 8, 12, 13, 17 25 29 21, 41

Subfamily Phyllanthoideae (23 of 56 genera), excluding Tribe Drypeteae [ = Phyllanthaceae (Stevens, 2001; Judd et al., 2002; APG, 2003)] Actephila [20] P-E-F/R? (s) 1 16 Andrachne [15] P-E-F 1 7 Antidesma [170] P-E-F 8 5, 16, 18, 29 Aporosa [80] P-E-F 8 29, 38 Baccaurea [43] P-E-F 12 29 Bischofia [2] P-E-F 2 5, 16, 21, 29, 35, 41 Breynia [15] P-E-F 2 5, 16 Bridelia [60] P-E-F 10 5, 14, 15, 16, 21, 25, 31, 35, 39 P-E-F/R? (+f) 1 29 Cleistanthus [140] [cot. either C-E-R (?) 4 16 (h) or (s) – genus probably not P?-E-? 3 4 monophyletic] P-E-F/R? (+f) 1 4 P-E-F 3 14, 16 Glochidion [300] P-E-F 11 16, 29 P?-E-? 1 4 Hyeronima [40] P-E-F 1 12, 40 ?-E-? (+f) 1 26 Keayodendron [1] P-E-F 1 25 Maesobotrya [20] P-E-F 1 18 Margaritaria [14] P-E-F 4 12, 13, 15, 16, 24, 32 Martretia [1] P-E-F/R? (+f) 1 18 Phyllanthus [800] P-E-F 14 cat5, 12, 15, 16, 18, 22, 24, 25, 29, 31, 32, 35 P-E-F 1 cat+ 29 Protomegabaria [3] P-E-F 1 25 Richeria [2] P-E-F 1 11 Sauropus [80] P-E-F/R? (+f) 1 38 Savia [25] P-E-F 1 8 Flueggea [6] P-E-F 1 5 (as Securinega) ?-E-? 1 36 Spondianthus [1] P-E-F 1 18, 25 Uapaca [61] P-E-F 5 18, 25, 34, 39 P-H-F 1 cpet++ 40



Research Table 2 continued

Genus1

Seedling traits2,3

n of spp. studied

Other notes4


Subfamily Phyllanthoideae, Tribe Drypeteae (2 of 3 genera) [ = Putranjivaceae (Stevens, 2001; Judd et al., 2002; APG, 2003)] Drypetes [200] P-E-F 19 4, 8, 12, 15, 16, 18, 21, 25, 29, 31, 38, 42 Putranjiva [3] P-E-F 1 33, 35 (incl. Liodendron) C-E-R (h) 1 cat+ 21 Subfamily Oldfieldioideae (7 of 27 genera) [ = Picrodendraceae (Stevens, 2001; Judd et al., 2002; APG, 2003)] Austrobuxus [20] P-E-F 1 16 Choriceras [2] P-E-F 1 16 Dissiliaria [3] P-E-F 2 16 Oldfieldia [4] P-E-F 1 25, 39 Paradrypetes [2] C-H-R (h) 1 hyp13 Petalostigma [6] P-E-F 1 22 Whyanbeelia [1] P-E-F 1 16 1 Accepted generic names [and number of species in genus] from Radcliff-Smith (2001). (Generic names used in original reference.) Generic synonyms from Radcliff-Smith (2001), W3TROPICOS (http://mobot.mobot.org/W3T/Search/vast.html), and P.C. van Welzen (ed.), Euphorbiaceae of Malesia (http://www.nationaalherbarium.nl/euphorbs/). 2 Cotyledon traits (Exposure-Position-Form): Exposure: C = cryptocotylar, P = phanerocotylar; Position: E = epigeal, H = hypogeal; Form: F = foliaceous, R = reserve; ? = unknown or questionable; C/P = semicryptocotylar. Data from seedling references or illustrations within. 3 Nature of reserve cotyledons (excluding P-E-F seedlings): (h) = haustorial (thin cotyledons absorb reserves from endosperm and translocate them to germinating seedling; endosperm in seed abundant to copious, embryo with thin cotyledons); (s) = storage (thick fleshy cotyledons store reserves and translate them directly to germinating seedling; endosperm in seed absent or scant, embryo with thick cotyledons); (+f) = somewhat fleshy (cotyledons thin in seed becoming somewhat thick and fleshy as reserves translocated from endospem to seedling during germination); (w) = exposed cotyledons succulent (thick and fleshy because of high water content). Data from seedling references or deduced from generic seed descriptions (Radcliff-Smith, 2001). 4 Other notes: Cataphylls (for C-H-R and C-E-R seedlings only): cat +, present on epicotyl; cat-, absent on epicotyl. Hypocotyl (C-H-R and P-H-F seedlings only): hyp +, distinct hypocotyl present but very short (often buried) compared to epicotyl; hyp-, hypocotyl absent or scarcely developed. Cotyledonary petioles (P-H-F-seedlings only): cpet + +, petioles long and push cotyledons out of soil. 5 References: 1, del Amo (1979); 2, Barrera Torres (1985– 86); 3, Bobilioff (1923); 4, Bodegom et al. (1999); 5, Burger (1972); 6, Chancellor (1959); 7, Csapody (1968); 8, Duke (1965); 9, Duke (1969); 10, Foster and Gifford (1959); 11, Flores (1992); 12, N. Garwood, unpubished, Panama; 13, N. Garwood, unpublished, Ecuador; 14, Gilbert (1939); 15, Hladik and Miquel (1990); 16, Hyland and Whiffin (1993); 17, IbarraManriquez et al. (2001); 18, de Koning (1983); 19, Kamilya and Paria (1994); 20, Labouriau et al. (1964); 21, Li and Hsieh (1997); 22, Lubbock (1892); 23, Macedo & Prance (1978); 24, Miquel (1987); 25, de la Mensbruge (1966); 26, Moreira and Arnaez (1994); 27, Mukerji (1936); 28, Muller (1978); 29, Ng (1991); 30, National Herbarium Nederland: http://www.nationaalherbarium.nl/euphorbs/images; 31, Onyeachusim (1977); 32, Ricardi et al. (1987); 33, Saha et al. (1998); 34, Taylor (1960); 35, Troup (1921); 36, Vasil’chenko (1960); 37, Verdus (1979); 38, de Vogel (1980); 39, Voorhoeve (1965); 40, Vozzo (2002); 41, Zhuyao (1978); 42, Gazel (1990). 6 Panda is included in subfamily Acalyphoideae by Radcliff-Smith (2001), but segregated into Pandaceae by APG (2003).

In Euphorbiaceae, PEF morphology was present in 87 of the 100 documented genera (including those that are incompletely known but probably PEF, e.g. P?-E-?, P-E-F/R?) of Euphorbiaceae s.l., in 55 of the 68 genera of Euphorbiaceae s.s. (subfamilies Acalyphoideae, Crotonoideae, and Euphorbioideae), and in 16 of the 20 genera in subfamily Crotonoideae, which includes Manihot. Cnidoscolus, the sister genus of Manihot (Miller & Webster, 1962, G. Léotard, unpublished) also has PEF seedlings. The trait combination CER was present in 15 genera of Euphorbiaceae s.l., CHR morphology in four genera, and PER and PHF in two genera each. In only 45 of the 100 genera has seedling morphology been described in at least two species. Two trait combinations were found in only 10 genera (accounting for 72 species): PEF and CER in Jatropha, Euphorbia, Hura, Mabea, Cleistanthus, and Putranjiva; CER and CHR in Omphalea; PEF and CHR in Mercurialis; and PEF and PHF in Uapaca and Manihot. Intraspecific variation has been reported in Hura crepitans


L. (Gilbert, 1939; Duke, 1965; Ricardi et al., 1987, Garwood, unpublished), Mabea occidentalis Benth. (Macedo & Prance, 1978; Garwood, unpublished), and Mercurialis perennis L. (Mukerji, 1936). For the genus Manihot, we found information on seedling morphology of five species additional to those we studied. Four of these, M. aesculifolia (Humboldt, Bonpland, & Kunth) Pohl, M. carthaginensis ( Jacq.) Müll. Arg., M. chlorosticta Standl. & Goldman, and M. pseudoglaziovii Pax & K. Hoffm., have PEF seedlings (Vozzo, 2002). Information is incomplete for the fifth species, M. gracilis Pohl, but suggests the rare trait combination PHF. A poor-quality photograph in Labouriau et al. (1964) shows a hypocotyl that has apparently not elongated (hypogeal germination), but foliaceous cotyledons that have been lifted above the soil surface by elongated cotyledon petioles. None of the five wild species reported on in these studies is as closely related to M. esculenta as the two wild taxa we studied, which include cassava’s closest wild relatives.


313

314 Research

Discussion

Fig. 2 Supertree of different crotonoid taxa inferred from nonoverlapping taxonomic data for AFLP, microsatellites, and G3pdh sequence comparisons. The bootstrap support is indicated on nodes. The distribution of PEF/CHR character states among crotonoids is given on the right. This topology implies a PEF to CHR transformation (black rectangle), followed by a reversal to PEF (white rectangle) in cassava (Manihot esculenta ssp. esculenta). The U/L/S-1/E haplotype nomenclature for M. esculenta ssp. flabellifolia and pruinosa refers to Olsen & Schaal (1999), except for S-1, a haplotype found in French Guiana populations by G. Léotard (unpublished).

Our finding of CHR morphology in M. pruinosa and M. esculenta ssp. flabellifolia is the first report of this seedling type in the genus and the second in subfamily Crotonoideae. Thus seedling morphology of domesticated cassava (PEF) contrasts strongly with that of its closest wild relatives. Such a striking contrast in seedling morphology between closely related species is rare in the Euphorbiaceae. Only seven of the 68 genera in Euphorbiaceae s.s. studied have two seedling types (Table 2): Mercurialis and Omphalea (Acalyphoideae), Manihot and Jatropha (Crotonoideae), and Euphorbia, Hura and Mabea (Euphorbioideae). Only in Mercurialis, Omphalea and Manihot do epigeal and hypogeal germination occur in a single genus. Apart from Manihot, only in Mercurialis do all three defining traits of seedling type show intrageneric variation. Phylogeny reconstruction The species-level source topologies were, respectively (AES, (CAR,(FLA423,(FLA430,ESC)))) for AFLP and microsatellites, and ( JAT,CNI,(FLA_E,(PRU_L,PRU_S,(FLA_U,ESC)))) for G3pdh. The MRP supertree obtained from them is reported in Fig. 2. The genus Manihot is monophyletic, with M. aesculifolia and M. carthaginensis being the first two offshoots. Domesticated cassava, M. esculenta ssp. esculenta, is nested within the flabellifolia–pruinosa complex. The moderate support for most nodes likely reflects the small number of characters scored here. However, both the clear internal position of domesticated cassava in the branching pattern – bootstrap support of 84% – and the character state distribution of seedlings among crotonoid taxa suggest a PEF to CHR transition along the branch leading to flabellifolia–pruinosa– esculenta, followed by a reversal to PEF in domesticated cassava (Fig. 2).


The comparison of domesticated cassava (Manihot esculenta ssp. esculenta) and its two closest wild relatives (Olsen & Schaal, 1999, 2001; Allem et al., 2001) – its wild progenitor, considered subspecifically distinct (M. esculenta ssp. flabellifolia), and the sister species of M. esculenta (M. pruinosa) – showed strong contrasts in seedling morphology. Furthermore, seedling morphology in the wild members of the lineage leading to cassava was in turn highly divergent from that of other wild species of Manihot. Evolutionary pressures acting on seedling traits have been sufficiently strong, and sufficiently divergent, to overcome the phylogenetic inertia that usually characterizes these traits. Intrageneric variation is uncommon, and intraspecific variation like that reported here is even rarer, known for only two other of the 335 studied species of Euphorbiaceae s.l. We argue that the evolutionary lability of seedling functional morphology in Manihot is the result of strong selection. The following sections address the reconstruction of evolutionary transformations of seedling morphology based on phylogeny, and the selective pressures that could have driven such modifications. Phylogeny and the evolution of divergent seedling morphologies in Manihot The mapping of seedling traits onto phylogeny reveals an intriguing pattern of distribution (Fig. 2). First, PEF morphology, the most frequent combination of traits censused in Euphorbiaceae, also characterizes Cnidoscolus, the sister genus of Manihot (Miller & Webster, 1962, G. Léotard, unpublished), other related genera such as Jatropha, and two wild Manihot species occupying basal positions in the genus (with respect to the few species we were able to include). PEF morphology thus appears to be plesiomorphic in the genus. However, in the lineage including cassava’s closest wild relatives, there appears to have been a transformation from PEF to the very different CHR morphology. Domestication of M. esculenta (divergence of subspecies esculenta from within the complex flabellifolia–pruinosa) was accompanied by an apparent reversion to PEF morphology. Thus during the radiation of Manihot, seedling morphology has undergone not one, but two dramatic transformations, the second one very recent in the scale of evolutionary time. Several hypotheses alternative to that of reversion must be considered. One alternative is that current phylogenetic hypotheses are in error, and that domesticated cassava is most closely related to wild species with PEF morphology. However, substantial molecular evidence supports the close relationship of cassava and the two known CHR species (summarized in Allem et al., 2001). This includes the studies used to construct the tree of Fig. 2 (Roa et al., 1997, 2000; Olsen & Schaal, 1999; G. Léotard, unpublished) as well as other studies which could not be used in this analysis because


Research

numbers of markers and/or taxa represented were too low (Olsen & Schaal, 2001) or because technical problems precluded obtaining data matrices for tree recoding (Colombo et al., 2000). Trees published in each of these studies are all consistent with the supertree shown in Fig. 2. Similarly, molecular evidence argues against a close relationship of cassava to M. aesculifolia (Roa et al., 1997, 2000), the only known PEF species once thought to be related to cassava (Rogers & Appan, 1973). Two of the other three known PEF species, M. chlorosticta and M. carthaginensis, also appear from molecular evidence to be less closely related to cassava than is M. esculenta ssp. flabellifolia (Roa et al., 1997, 2000; Colombo et al., 2000). A second alternative is that domesticated cassava is the product of reticulate evolution, and that its seedling morphology (and even that of its wild relatives) may have been affected by hybridization. While the tree shown in Fig. 2 is supported by molecular data and reflects current opinion, the longstanding hypothesis that cassava is a compilospecies resulting from complex interspecificiation hybridization events (Rogers & Appan, 1973) cannot yet be discarded. Manihot bears many marks of a relatively recent radiation, species limits are unclear, and hybridization could be frequent. Hybridization could produce novel variation in many traits, including those we examined. Understanding the genetic architecture underlying traits like those studied here could contribute to resolving the question of the origin of domesticated cassava. A third alternative is that there exists greater variation in seedling morphology among populations of cassava’s closest wild relatives than was revealed by our study. We have shown that some seedling traits show phenotypic plasticity. Perhaps there also exists genetic variation for these traits, within or between populations. In the large geographic range of these close wild relatives of cassava, perhaps there exist some populations with PEF morphology. Instead of evolving under domestication as a result of the selective pressures we postulate, seedling morphology of the domesticate could thus be the result of a founder effect. This hypothesis can only be evaluated once observations of seedlings have been conducted in many parts of the rim of Amazonia. We hope our study will stimulate such efforts. We do have observations (B. Pujol and D. McKey, unpublished) from one additional area, the North Rupununi savannas of Guyana, and seedlings of the wild Manihot common there (M. pruinosa) all show CHR morphology. How widespread is CHR morphology in Manihot? The absence of data for all but a few of the hundred-odd species of Manihot makes this question currently impossible to answer. Manihot reaches its greatest species richness in the cerrados of Brazil (Rogers & Appan, 1973), a habitat where hypogeal germination should often confer important advantages (Rizzini, 1965), and it would not be surprising if CHR morphology were widespread in cerrado Manihot species additional to the two studied here. Incomplete information on one other


cerrado species of the genus, M. gracilis Pohl, indicates that it is characterized by hypogeal germination (Labouriau et al., 1964). The seedling illustrated in that article has leafy cotyledons pushed through at least a few mm of soil by long petioles, giving it a PHF morphology like that of seedlings of M. pruinosa that germinate from shallow depth (Fig. 1d). Is seedling morphology associated with habitat and life form in Manihot? Whereas both taxa with CHR morphology are savanna-forest ecotone taxa, none of the four wild species of Manihot characterized by PEF morphology are savanna species. M. carthaginensis and M. pseudoglaziovii are shrubs to medium-sized trees of dry forest types in north-western South America and north-eastern Brazil, respectively; M. aesculifolia and M. chlorosticta are shrubs and vine-like shrubs, respectively, of dry forests in western Mexico (Rogers & Appan, 1973). A phylogenetically based survey of seedling functional morphology in the genus, in which habitat ranges from savanna shrubs to rainforest vines, would be of obvious interest. Adaptive significance of seedling morphology in Manihot The information we present on the seedling morphology of M. pruinosa and M. esculenta ssp. flabellifolia is the first welldocumented data available for any euphorbiaceous species from savanna habitats. We postulate that their divergent seedling morphology is adapted to the constraints imposed by these environments. Several arguments are consistent with the hypothesis that CHR morphology is adaptive for Manihot spp. in South American savanna habitats. First, fire and water stress are both frequent in savanna and savanna-forest ecotone environments around the rim of Amazonia (Rizzini, 1965; Oliveira & Marquis, 2002). Combined with seed burial by ants (Elias & McKey, 2000), hypogeal germination could reduce the risk of seedling mortality upon passage of fire by protecting buds underground, axillary to the cotyledons and to cataphylls on the below-ground portion of the epicotyl. Hypogeal and ‘cryptogeal’ germination are similar pyrophytic adaptations frequent in African savannas (Burtt, 1972; Jackson, 1974). Second, by reducing the leaf area of very young seedlings and thereby water loss due to transpiration, possession of cotyledons that remain enclosed in a buried seed could also reduce mortality due to drought stress. Finally, soil mineral nutrients are also frequently limiting in South American savannas (see several chapters in Furley et al., 1992). All these ecological conditions should favour traits that enable seedlings to tolerate stress, and limit their ability to exploit the potential for rapid initial growth that would be conferred by epigeal germination and foliaceous cotyledons. During domestication, some ancestral reproductive-ecology adaptations were maintained. Thus temperature-dependent physiological seed dormancy continued to be favoured in slash-and-burn agriculture, because it allowed seeds to survive fallow periods and to germinate synchronously in the favourable


315

316 Research

conditions of newly burnt fields (Pujol et al., 2002). However, domestication led to important changes in the environment experienced by seedlings. First, in savannas, one burn may be followed by another. In contrast, in slash-and-burn agroecosystems there is only a single human-managed burn, which is predictably not followed by any further fire. Seedlings that germinated in response to the burn thus face no further risk that their above-ground parts will be destroyed by fire. Second, in savannas, fires may occur at various seasons. In slash-and-burn systems, in contrast, the burn is carried out just before the start of the rainy season, so that seedlings predictably encounter favourable water conditions. Third, slashand-burn farming of cassava is typically done in forested areas. Slashed and burned secondary forest produces much larger amounts of ash than does burned savanna, creating not only temporarily nutrient-rich conditions but also reducing soil acidity and thereby increasing nutrient availability. Higher availability of water and nutrients, and lower risk of losing aerial parts to fire and drought, should combine to favour traits that allow rapid growth. Finally, farmers also exert selection for rapid growth. Growth rate strongly influences the probability that volunteer seedlings in Amerindian cassava fields survive weeding (Pujol et al. 2004) and competition (B. Pujol, unpublished), thereby becoming candidates for incorporation into the stock of clonal propagules. Epigeal germination and foliaceous cotyledons lead to rapid growth from the start of seedling development by ensuring the precocious deployment of a photosynthetic surface. All these advantages should exert strong selection for PEF seedling morphology in seedlings of domesticated cassava. The divergent morphologies of seedlings of domesticated cassava and its wild relatives thus appear to be adaptively matched to their respective environments. Varying expression of cotyledon traits in M. pruinosa Almost all the seedlings of M. pruinosa observed in natural populations showed CHR morphology, but a few seedlings germinating from shallow depth in the soil showed the PHF morphology, i.e. hypogeal germination with foliaceous cotyledons on long petioles. This morphology was also reproduced in the laboratory by setting seeds to germinate from shallow depth (Fig. 1d). The morphology of these seedlings was highly reminiscent of the seedling of M. gracilis illustrated in Labouriau et al. (1964) and referred to near the end of the section ‘Phylogeny and the evolution of divergent seedling morphologies in Manihot’ in the Discussion. As in the other seedlings, the hypocotyl of these shallowly buried seeds showed no elongation. This observation suggests that the genetic or developmental basis of the difference between the two types of seedlings, PHF and CHR, may in this case be surprisingly simple. Cotyledonar form and function in M. pruinosa may be simply phenoplastic consequences of the environment experienced by the germinating seed (e.g. whether or not cotyledons, apical buds, or other parts are


exposed to light). Whereas cotyledons of the wild relative had long petioles and were foliaceous when exposed to light, and had short petioles and were nonfoliaceous when kept in darkness, cotyledons of domesticated cassava always developed long petioles and became foliaceous, independent of the light environment experienced by the germinating seed. This observation suggests that at least two important differences exist in seedling development between domesticated cassava and this close wild relative: (i) a constitutive difference in the degree of elongation of the hypocotyl; and (ii) a reduction in the plasticity of cotyledonar form and function in domesticated cassava. It is tempting to speculate that the plasticity of cotyledon traits in wild progenitors may have facilitated the evolutionary reversion to PEF morphology suggested by phylogenetic analysis. Loss of plasticity with domestication has also been observed in other traits and plants (Doebley et al., 1990). From domestication syndrome to the evolution of development Although domesticated cassava grown by Amazonian farmers is vegetatively propagated, the important role of volunteer seedlings in populations of cassava has been demonstrated by genetic studies (Elias et al., 2000a, 2001a; Sambatti et al., 2001). They show the regular incorporation of recombinant genotypes into the pool of vegetative propagules. Regular incorporation means that selection can continue to act on the traits of sexual reproduction. We postulate that the morphological traits of seedlings of domesticated cassava are the results of natural selection, including that exerted by humans, acting on volunteer seedlings. PEF traits, a component of the plant’s sexual reproductive ecology, thus form part of the domestication syndrome of cassava, a ‘vegetatively propagated’ crop plant. Whatever the selection pressures responsible for the developmental differences we document, their highly divergent expression in wild and domesticated taxa that are highly interfertile (Jennings, 1995) provides a unique opportunity to study the genetic and developmental basis of traits important in seedling functional morphology that have so far been accessible only to comparative, interspecific studies. Furthermore, the plasticity of cotyledonar form and function shown in M. pruinosa opens avenues for studying environmental influences on development (‘eco-devo’), a rapidly growing field (Sultan, 2004).

Acknowledgements We thank Gérard Laurent for technical help; Marina Pinheiro-Kluppel and Carole Reiff for their participation in the French experiment at the CEFE; Adriane Cristina Servignani Santos, Amós Lemos dos Santos, Fábio Régis de Souza and Thatiane Camargo Leite for their participation in


Research

the Brazilian experiment. We thank Instituto Agronômico de Campinas (Roots and Tubers Section) for technical support. This research was funded by grants to Doyle McKey from several agencies of the French government. These included the programme ‘Impact des Biotechnologies dans les Agroécosystèmes’ (CNRS, Ministry of Research and Technology), the Bureau des Ressources Génétiques, the Ministry of Ecology and Sustainable Development (through the programme ‘Ecosystèmes Tropicaux’), and the CPER (Contrat Plan Etat-Région Guyane). We thank Guillaume Léotard for sharing unpublished data on the molecular phylogeny of Manihot. Carolina Roa, Ken Olsen, Barbara Schaal, Carlos Colombo, and Marianne Elias are thanked for their willingness to let us use their original data matrices in calculating the supertree shown in Fig. 2. Some matrices were not included due to their limited taxon coverage, and that of C. Colombo could not be recovered from a crashed computer. This is contribution ISEM no. 2004-63 of the Institute of Evolution Sciences of Montpellier (UMR 5554 CNRS). Comments by two anonymous reviewers and by Mark Rausher greatly improved the manuscript.

References Allem AC, Mendes RA, Salomão AN, Burle ML. 2001. The primary gene pool of cassava (Manihot esculenta Crantz subspecies esculenta, Euphorbiaceae). Euphytica 120: 127–132. del Amo S. 1979. Clave para plántulas y estados juveniles de especies primarias de una selva alta perennifolia en Veracruz, México. Biótica 4: 59–108. APG (Angiosperm Phylogeny Group). 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnean Society 141: 399–436. Barrera Torres E. 1985–86. Identificación de plántulas de algunas especies arbóreas del bosque de niebla. Perez-Arbelaezia 1–2: 165–209. Bininda-Emonds ORP. 2004. The evolution of supertrees. Trends in Ecology and Evolution 19: 315 –322. Bobilioff W. 1923. Anatomy and Physiology of Hevea Brasiliensis. Zürich, Switzerland: Art. Institut. Orell. Fηssli. Bodegom S, Pelser PB, Keßler PJA. 1999. Seedlings of secondary forest tree species of East Kalimantan, Indonesia. Tropenbos-Kalimantan Series 1. Wageningen, The Netherlands: Tropenbos International. Burger HD. 1972. Seedlings of Some Tropical Trees and Shrubs, Mainly of South East Asia. Wageningen, The Netherlands: Centre for Agricultural Publishing and Documentation. (PUDOC). Burtt BL. 1972. Plumular protection and some related aspects of seedling behaviour. Transactions of the Botanical Society of Edinburgh 41: 393–400. Chancellor RJ. 1959. Identification of Seedlings of Common Weeds. Bull, no. 179. London, UK: Ministry of Agriculture, Fisheries and Food (MAFF). Colombo C, Second G, Charrier A. 2000. Genetic relatedness between cassava (Manihot esculenta Crantz) and M. flabellifolia and M. peruviana based on both RAPD and AFLP markers. Genetics and Molecular Biology 23: 417–423. Csapody V. 1968. Keimlingsbestimmungsbuch der Dikotyledonen. Budapest, Hungary: Akadémiai Kiadó. Doebley J, Stec A, Wendel J, Edwards M. 1990. Genetic and morphological analysis of a maize-teosinte F2 population: implications for the origin of maize. Proceedings of the National Academy of Sciences, USA 87: 9888– 9892.


Duke JA. 1965. Keys for the identification of seedlings of some prominent woody species in eight forest types in Puerto Rico. Annals of the Missouri Botanical Garden 52: 314–350. Duke JA. 1969. On tropical tree seedlings. I. Seeds, seedlings, systems, and systematics. Annals of the Missouri Botanical Garden 56: 125–161. Elias M, McKey D. 2000. The unmanaged reproductive ecology of domesticated plants in traditional agroecosystems: an example involving cassava and a call for data. Acta Oecologica 21: 223–230. Elias M, McKey D, Panaud O, Anstett MC, Robert T. 2001b. Traditional management of cassava morphological and genetic diversity by the Makushi Amerindians (Guyana, South America): perspectives for on-farm conservation of crop genetic resources. Euphytica 120: 143–157. Elias M, Panaud O, Robert T. 2000a. Assessment of genetic variability in a traditional cassava (Manihot esculenta Crantz) farming system, using AFLP markers. Heredity 85: 219–230. Elias M, Penet L, Vindry P, McKey D, Panaud O, Robert T. 2001a. Unmanaged sexual reproduction and the dynamics of genetic diversity of a vegetatively propagated crop plant, cassava (Manihot esculenta Crantz), in a traditional farming system. Molecular Ecology 10: 1895–1907. Elias M, Rival L, McKey D. 2000b. Perception and management of cassava (Manihot esculenta Crantz) diversity among Makushi Amerindians of Guyana (South America). Journal of Ethnobiology 20: 239–265. Essig FB. 1987. Seedling morphology and subgeneric classification in Clematis (Ranunculaceae). American Journal of Botany 74: 733–733. Felsenstein J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. Flores S. 1992. Growth and seasonality of seedling and juveniles of primary species of a cloud forest in northern Venezuela. Journal of Tropical Ecology 8: 299–305. Foster AS, Gifford EMJ. 1959. Comparative Morphology of Vascular Plants. San Francisco, USA: W.H. Freeman. Fregene MA, Vargas J, Ikea J, Angle F, Tohme J, Asiedu RA, Akoroda MO, Roca WM. 1994. Variability of chloroplast DNA and nuclear ribosomal DNA in cassava (Manihot esculenta Crantz) and its wild relatives. Theoretical and Applied Genetics 89: 719–727. Furley PA, Proctor J, Ratter JA. 1992. Nature and Dynamics of ForestSavanna Boundaries. London, UK: Chapman & Hall. Garwood NC. 1996. Functional morphology of tropical tree seedlings. In: Swaine MD, ed. The Ecology of Tropical Forest Tree Seedlings. Man and the Biosphere Series 17. Paris/Carnforth, France/ UK: UNESCO/Parthenon, 59–129. Garwood NC, Tebbs M (illustrator). 1995. Studies in Annonaceae. XX. Morphology and ecology of seedlings, fruits and seeds of selected Panamanian species. Botanische Jahrbücher 117: 1–152. Gazel M. 1990. Flore Forestière de Guyane. Les Euphorbiacées. Kourou, French Guiana: ENGREF-Silvolab-Guyane. Gilbert G. 1939. Observations préliminaires sur la morphologie des plantules forestières du Congo Belge. Institut National Pour l’Etude Agronomique Du Congo Belge. Série Scientifique. 17: 1–28. Hladik A, Miquel S. 1990. Seedling types and plant establishment in an African rain forest. In: Bawa KS, Hadley M, eds. Reproductive Ecology of Tropical Forest Plants. Man and the Biosphere Series 7. Paris/Carnforth, France/UK: UNESCO/Parthenon, 261–282. Hoock J. 1971. Les Savanes Guyanaises: Kourou. Essai de Phytoécologie Numérique. Mémoires ORSTOM. Paris, France: ORSTOM. Hoshizaki K, Suzuki W, Sasaki S. 1997. Impacts of secondary seed dispersal and herbivory on seedling survival in Aesculus turbinata. Journal of Vegetation Science 8: 735–742. Hyland BPM, Whiffin T. 1993. Australian Tropical Rain Forest Trees: an Interactive Identification System. East Melbourne, Australia: CSIRO Publishing. Ibarra-Manriquez G, Martinez Ramos M, Oyama K. 2001. Seedling functional types in a lowland rain forest in Mexico. American Journal of Botany 88: 1801–1812.


317

318 Research Jackson G. 1974. Cryptogeal germination and other seedling adaptations to the burning of vegetation in savanna regions: the origin of the pyrophytic habit. New Phytologist 73: 771–780. Jennings DL. 1995. Cassava, Manihot esculenta (Euphorbiaceae). In: Smartt J, Simmonds NW, eds. Evolution of Crop Plants, 2nd edn. Harlow, UK: Longman Scientific and Technical, 128 –132. Judd WS, Campbell CS, Kellogg EA, Stevens PF, Donoghue MJ. 2002. Plant Systematics – a Phylogenetic Approach, 2nd edn. Sunderland MA, USA: Sinauer. Kamilya P, Paria N. 1994. Seedling morphology of some Indian species of Jatropha and its implications in taxonomy. Acta Botanica Indica 22: 251–256. Kitajima K, Fenner M. 2000. Seedling regeneration ecology. In: Fenner M, ed. Seeds: Ecology of Regeneration in Plant Communities, 2nd edn. Wallingford, UK: CAB International. 331–360. de Koning J. 1983. La Forêt Du Banco / II: les Taxons. Wageningen, The Netherlands: Landbouwhogeschool. Labouriau LG, Valio IFM, Heringer EP. 1964. Sôbre o sistema reproductive de plantas dos Cerrados – I. Anais Da Academia Brasileira de Ciências 36: 449–464. Li S-P, Hsieh C-F. 1997. Seedling morphology of some woody species in a subtropical rainforest of southern Taiwan. Taiwania 42: 207–238. Lubbock J. 1892. A Contributions to Our Knowledge of Seedlings, Vol. 1 & 2. London, UK: Kegan Paul, Trench, and Trüber. Macedo M, Prance GT. 1978. Notes on the vegetation of Amazonia II. The dispersal of plants in Amazonian white sand campinas: the campinas as functional islands. Brittonia 30: 203 –215 de la Mensbruge G. 1966: La Germination et les Plantules Des Essences Arborées de la Forêt Dense Humide de la Côte d’Ivoire. Publication 26. Nogent-Sur-Marne, France. Centre Technique Forestier Tropical. Miller KI, Webster GL. 1962. Systematic position of Cnidoscolus and Jatropha. Brittonia 14: 174 –180. Miquel S. 1987. Morphologie fonctionnelle de plantules d’espèces forestières du Gabon. Bulletin Du Muséum National d’Histoire Naturelle, 4e Série, Section B, Adansonia 9: 101–121. Moreira GI, Arnáez SE. 1994. Morfología de las estructuras reproductoras y germinación de nueve especies forestales nativas de Costa Rica. Revista de Biologia Tropical 42: 73 – 82. Mukerji SK. 1936. Contributions to the autecology of Mercurialis perennis L. Journal of Ecology 24: 38 – 81. Muller FM. 1978. Seedlings of the North-Western European Lowland: a Flora of Seedlings. The Hague/Wageningen, The Netherlands: Dr. Dr W. Junk/ Centre for: Agricultural Publishing and Documentation (PUDOC). Ng FSP. 1978. Strategies of establishment in Malayan forest trees. In: Tomlinson PB, Zimmermann MH, eds. Tropical Trees as Living Systems. Cambridge, UK: Cambridge University Press, 129–162. Ng FSP. 1991. Manual of Forest Fruits, Seeds and Seedlings. Malayan Forest Record No. 34, Vol. 1 & 2. Kuala Lumpur, Malaysia: Forest Research Institute Malaysia. Oliveira PS, Marquis RJ. 2002. The Cerrados of Brazil. Ecology and Natural History of a Neotropical Savanna. New York, USA: Columbia University Press. Olsen KM. 2002. Population history of Manihot esculenta (Euphorbiaceae) inferred from nuclear DNA sequences. Molecular Ecology 11: 901–911. Olsen KM, Schaal BA. 1999. Evidence on the origin of cassava: phylogeography of Manihot esculenta. Proceedings of the National Academy of Sciences, USA 96: 5586 –5591. Olsen KM, Schaal BA. 2001. Microsatellite variation in cassava (Manihot esculenta, Euphorbiaceae) and its wild relatives: further evidence for a southern Amazonian origin of domestication. American Journal of Botany 88: 131–142. Onyeachusim HD. 1977. A provisional identification key for seedlings of woody species in Ijaiye Forest Reserve. BSc Thesis. University of Ibadan, Ibadan, Nigeria.


Piperno DR, Pearsall DM. 1998. Origins of Agriculture in the Lowland Neotropics. San Diego, USA: Academic Press. Pujol B, David P, McKey D. 2004. Microevolution in agricultural environments: how a traditional Amerindian farming practice favors heterozygosity in cassava (Manihot esculenta Crantz, Euphorbiaceae). Ecology Letters in press. Pujol B, Gigot G, Laurent G, Pinheiro-Kluppel M, Elias M, HossaertMcKey M, McKey D. 2002. Germination ecology of cassava (Manihot esculenta Crantz, Euphorbiaceae) in traditional agroecosystems: seed and seedling biology of a vegetatively propagated domesticated plant. Economic Botany 56: 366–379. Radcliff-Smith A. 2001. Genera Euphorbiacearum. London, UK: Royal Botanic Gardens, Kew. Ragan MA. 1992. Phylogenetic inference based on matrix representation of trees. Molecular Phylogenetics and Evolution 1: 53–58. Ricardi M, Hernandez C, Torres F. 1987. Morfología de Plántulas de Árboles de Los Bosques Del Estado Mérida. Mérida, Venezuela: Talleres Graficos Universitarios. Rizzini CT. 1965. Experimental studies on seedling development of cerrado woody plants. Annals of the Missouri Botanical Garden 52: 410–426. Roa AC, Chavarriaga-Aguirre P, Duque MC, Maya MM, Bonierbale MW, Iglesias C, Tohme J. 2000. Cross-species amplification of cassava (Manihot esculenta) (Euphorbiaceae) microsatellites: allelic polymorphism and degree of relationship. American Journal of Botany 87: 1647–1655. Roa AC, Maya MM, Duque MC, Tohme J, Allem AC, Bonierbale MW. 1997. AFLP analysis of relationships among cassava and other Manihot species. Theoretical and Applied Genetics 95: 741–750. Rogers D, Appan SG. 1973. Manihot and Manihotoides (Euphorbiaceae): a Computer Assisted Study. Flora Neotropica 13. New York, USA: Hafner. Saha (Das) B, Sarkar S, Paria N. 1998. Seedling morphology of some important timber yielding plants and its taxonomic implications. Himalayan Ecology and Development 6: 12–15. Sambatti JBM, Martins PS, Ando A. 2001. Folk taxonomy and evolutionary dynamics of cassava: a case study in Ubatuba, Brazil. Economic Botany 55: 93–105. SAS. 1996. SAS/STAT User’s Guide, Release 6.12. Cary, NC. USA: SAS Institute. Stevens PF. 2001. Angiosperm Phylogeny Website. Version 5, May 2004 (and more or less continually updated since). URL http://www.mobot.org/ MOBOT/research/APWeb/ Sultan SE. 2004. Promising directions in plant phenotypic plasticity. Perspectives in Plant Ecology, Evolution and Systematics 6: 227–233. Swofford DL. 1998. PAUP*. Phylogenetic Analysis Using Parsimony ( * and Other Methods), Version 4. Sunderland, MA, USA: Sinauer Associates. Taylor CJ. 1960. Synecology and Silviculture in Ghana., Edinburgh, UK: Thomas Nelson and Sons. Troup RS. 1921. The Silviculture of Indian Trees, Vol. 1–3. Oxford, UK: Clarendon Press. Vasil’chenko IT. 1960. Seedlings of Trees and Shrubs: Systematic Key [in Russian]. Moskva/ Leningrad, SSSR: Izdatel’stvo Akademii Nauk. Verdus M-C. 1979. Convergence intragénérique chez les Euphorbes succulentes. Bulletin de la Société Botanique de France-Actualités Botaniques 126: 99–108. de Vogel EF. 1980. Seedlings of Dicotyledons: Structure, Development, Types; Descriptions of 150 Woody Malesian Taxa. Wageningen, The Netherlands: Centre for Agricultural Publishing and Documentation (PUDOC. Voorhoeve AG. 1979. Liberian High Forest Trees, 2nd edn. Wageningen, The Netherlands: Centre for. Agricultural Publishing and Documentation (PUDOC). Vozzo JA. 2002. Tropical Tree Seed Manual. Agriculture Handbook 721. Washington DC, USA: USDA Forest Service. Zhuyao SZT. 1978. Atlas of Seeds and Young Plants of Trees of Importance [in Chinese]. Beijing, China: Nanking College of Forest Products Industries Agricultural Press.
