Architectural analysis of root system of sexually vs. vegetatively propagated yam (Dioscorea rotundata Poir.), a tuber monocot

Charles-Dominique T, Mangenet T, Rey H, Jourdan C, Edelin C. 2009 Architectural analysis of root system of sexually vs. vegetatively propagated yam (Dioscorea rotundata Poir.), a tuber monocot. Plant and Soil 317: 61-77

Architectural analysis of root system of sexually vs. vegetatively propagated yam (Dioscorea rotundata Poir.), a tuber monocot T Charles-Dominique1, T Mangenet1, H Rey1, C Jourdan2 and C Edelin1 1Université

Montpellier 2, UMR AMAP, Montpellier, F-34000 France, 2CIRAD, UPR Ecosystèmes de Plantations, Montpellier, F-34000 France

Architectural descriptors were used to understand root system structure and development in white yam (Dioscorea rotundata Poir., Dioscoreaceae), a tuber monocot. Observations were made on seedlings and plant derived from tuber fragments, cultivated in greenhouses over a developmental cycle. This study demonstrated that both seedlings and plants derived from tubers have two distinct root systems that are highly organized. The first (seminal or tubercular) has been called the temporary root system which is small and short lived. The architectural unit here is made up of two root axis categories. The second (adventitious in both cases) has been called the definitive root system. It is larger and has a far longer lifespan than temporary root systems. The architectural unit here is made up of three root axis categories. Adventitious root systems are formed by structural repetitions of their own architectural unit. The temporary and definitive root systems possess the same structural and functional properties and become established and succeed one another in time following an identical developmental sequence. Neo tuber development is coupled with the root system development. Our results highlight to what extent it is important to study simultaneously the different parts of a root system in order to understand its development. This study confirms how architectural tools can be used to understand root system structure and development and prove accurate informations on root system development for use in agricultural management. Architectural analysis - Development - Root - Root functions - Tuber - Dioscorea rotundata poir

INTRODUCTION

paper we attempt to answer these questions by means of an architectural analysis. This approach was initially developed to describe the structural development of shoots (Hallé and Oldeman 1970, Oldeman 1974), and is now also used to describe root systems (Atger and Edelin 1994a, b, Jourdan and Rey 1997a, b, Danjon et al. 2005, Khuder et al. 2007). Since they were first established, architectural concepts have provided powerful tools for studying plant form and development (Barthélémy and Caraglio 2007). In the study of root systems, architectural analysis has been sufficiently efficient to succeed in producing relevant simulations (Jourdan et al. 1995, Jourdan and Rey 1997b, Pagès et al. 1989, 2004). Architectural analysis can be used to make specific agronomic recommendations from modifying planting design to avoiding root competition in young stages, to perform fertilizer applications driven by fine root distribution, or to adapt silvicultural techniques (Tobin et al. 2007, Danjon et al. 2005, Collet et al. 2006).

The root system plays several major roles in plant development including anchorage, storage, hormonal biosynthesis, mineral nutrients and water absorption (Lopez-Bucio et al. 2003, Yang et al. 2004). In this way, root system development and spatial occupation should be considered as means by which to optimize crop soil and fertilizer management, and control pests and rot (Liedgens and Richner 2001, Malamy 2005). Structural knowledge is essential if a plant's function is to be thoroughly understood. In fact, from an agronomic standpoint, information is needed concerning four main aspects of the root system: (i) the qualitative and quantitative composition of the root system both overall and in detail, (ii) checking if there is a hierarchy in root structure and functions, (iii) identifying the level of structural repetitions which can be used to simplify observation methodology in the field, and lastly (iv) what are the developmental phases of the root system. In this

-1-

Yam is a perennial herbaceous vine which is vegetatively propagated by tuber fragmentation (Clairon and Zinsou 1980, Degras 1994, Cornet 2005). It is a plant of important agronomic interest given that its tuber feeds over 60 million people every day (Orkwor et al. 1998). The development of yam tuber and shoots has been thoroughly documented (Trouslot 1985, Degras 1994) unlike the yam root system which has received far less attention (Cornet 2005) despite the fact that the root system plays an important role in tuber development (Degras et al. 1980). Also, knowledge about root system growth and development can be useful for yam breeders and agronomists. No full description has ever been established of the yam root system's structure and architecture. The only data on its root system concern the number of main roots (Martin 1974, Degras et al. 1980, Trouslot 1985), their length (Okezie et al. 1981), their spatial occupation of ground layers (Dumont and Chopart 1992) and total root system fresh weight (Njoku et al. 1973). Taken individually, these elements do not provide sufficient information on its structure and development to understand, represent and manage the root system. Given that yam is generally produced in the field by tuber propagation, agronomists need information about plant development from tubers. Here, we describe a comparative architectural study of the root systems in plants derived from vegetative propagation, i.e. tuber fragment, and from seeds used as a reference. The study of early developmental stages was necessary to establish accurate architectural information on plant structure. Our aim was first to determine the qualitative composition of the root systems and describe their development over time. Data were then acquired for discussions about the variable and stable elements and properties of root systems. Lastly, root system ontogenesis was analysed.

fungicide (Dithane® Duo Capiscol) before being planted in a sand–peat mixture (volume proportions 3:1). Germination pots were placed in an incubator in the dark at 27°C until the first root occurrence (February 2007 for minisett plants and March 2007 for seedling plants). The plants were then transferred into greenhouses and put in pots of various sizes (from 3 l pots for first sampled plants to 260 l pot for last sampled plants). Greenhouse conditions were 26°C with a 12 h day-cycle. The plants were placed on two racks under four 400 W-lights (1 m above the racks). The plants were watered every 2 days. Liquid fertilizer (NFU 42- 001, NPK 7-10-10) was added to the water once in every two waterings as half the quantity of recommended fertilizer (Filloux D, pers. com.). Root sampling Preliminary observations were made on seedlings and minisett plants using an aero-hydroponic system (Rainforest ™) with a nutritive solution (Bakry Frédéric, CIRAD unpublished). This aerohydroponic system simplifies root system observation and provides the determination of the most relevant descriptors used to characterize root system architecture easier. These observations are not presented herein. At the same time, the choice of the various descriptors was validated by observing the root system of plants growing in a sand–peat mixture. The selected descriptors were then used to describe the root systems of the different potted plants. For each sampling, sufficient destructive excavations (420 plants) were made to produce at least 20 roots of each root category described per date. Root system development in the potted plants was monitored for 4 months in minisett plants (200702-06 to 2007-06-06) and 3 months in seedling plants (2007-03-07 to 2007-06-06) with once-a-week observations. Architectural analysis

MATERIALS AND METHODS

The architectural analysis began by separating roots into axis categories on the basis of qualitative and quantitative morphological traits. An architectural unit was then characterized in the structure of the entire branched system. This architectural unit is a hierarchical branched system composed of axis categories that are morphologically and functionally differentiated (Edelin 1977, Barthélémy et al. 1989, 1991). For any given species, the architectural unit is in most cases invariant between individuals (Barthélémy and Caraglio 2007). According to Cornet, yam root systems are qualitatively similar among varieties and species (Cornet 2005). The notion of reiterates (Oldeman 1974) has also been used to describe root systems. Reiteration is a morphogenetic process through which an organism duplicates its own elementary architecture (Hallé et al. 1978, Barthélémy et al. 1989, 1991). The reiteration process may be expressed during sequential development or

Plant material This study concerned the white yam (Dioscorea rotundata Poir., Dioscoreaceae). The seedling plants used were of the ‘Gnidou’ variety. 1,100 seeds were collected in Glazoué (Benin) on December 12th, 2006. Minisett plants were derived from ‘Pouna’ Variety tubers, sourced from Ghana. 130 minifragments or minisetts (Otoo et al. 1985) of 40 g to 100 g were used. Dumont and Chopart (1992) have shown that yam root systems are qualitatively the same among commercial yams. Planting Seeds were stored under dry, dark and constant temperature (22°C) conditions for 3 weeks. Minisetts (tuber fragments) were kept under the same conditions for 3 days. All were then treated with

-2-

after a traumatic event (Barthélémy and Caraglio 2007). The traumatic reiteration process is related to lost (apex necrosis) or lower (vascular bundles squeezing) apical dominance. The reiteration process may duplicate the total architectural unit or only a part of it: we speak of ‘total reiterate’ and ‘partial reiterate’ (Oldeman 1974). Traumatic reiteration processes are not described in the ontogenic stage description because they do not belong to the sequential part of development. The nomenclature used to describe the root axes corresponds to SR1 and SR2 for the first and second branching order root of seminal roots; TR1 and TR2 for the first and second branching order root of the tubercular root system; AR1, AR2 and AR3 for the first, second and third branching order root of the adventitious root system. During ontogenesis of the yam using minisett, a callus-like structure develops which has been called the primary nodal complex (PNC; Ferguson 1972). When describing the ontogenic stage, the root systems of seedling plants and plants derived from minisetts were studied at several stages. Each stage was characterized by a new morphogenetic event. The age of the different root systems was reported in days and considering the first day of seminal or tubercular root occurrence as time zero. The root system was represented, for each stage, using the mean values of the quantitative parameters and this resulted in typical diagrams that did not take account of stochastic events.

confirmed by statistical tests using a non parametric test: Kruskal Wallis test (Null hypothesis, Ho: At least one root type is different from the other). Multiple comparison tests have been performed in order to indicate grouping of the root types. Non-parametric test were used first because homoscedasticity was not verified and because only 20 roots were described per root axis category during the description of the ontogenic stage. This choice was justified by the large number of roots analyzed (around 4,300 roots described for 17 parameters). We used mean values for diameter, IBD and LAUZ because they do not evolve through time for a given root and maximal values for length and number of lateral roots because they vary as a function of plant development. In order to detect global differences and not only differences due to developmental stages, we have chosen to process together variables for all ontogenic stages. Root system growth analysis We calculated total root length for each root system. Total root length took account of both the elongation and branching processes and helps visualize the spatial dimensions of root system components. Applied to each part of the root system, it is used to establish their relative importance in time and space. Total root length (TRL) was been calculated using the formula:

Root descriptors We selected the following diagnostic qualitative descriptors: monopodial or sympodial structure, determinate or indeterminate growth, growth direction, reiteration ability and type. In order to complete and reinforce our morphological observations, quantitative variables were acquired using the following descriptors: length, growth rate, length of apical unbranched zone (LAUZ), number of lateral roots, inter-branch distance (IBD), basal diameter and emission angle. These parameters are frequently used in morphological root system studies (Harada et al. 1993, Morita and Yamazaki 1993, Jourdan and Rey 1997a, b). Concerning length, the maximal lengths observed were reported. For the number of lateral roots and LAUZ: mean maximal values were reported. Concerning shape, conicity corresponds to a basal/distal difference in diameter of 30% or more. Cylindrical shaped roots were defined when this difference was less than 10%. For diameter, absolute encountered minimal and maximal values of basal diameter are reported. Lastly, the results have been summarized in the form of architectural tables.

where nx = number of roots of the xth order on the (x−1)th branching order root, Lx = length of xth order root. RESULTS Ontogenic stage description Each date reported corresponds to the mean occurrence date of the corresponding morphogenetic events though variations of more or less 2 days were observed in the occurrence dates between individuals.

Seedling root system At around the 14th day (Figs. 1a, 2a), the root system was made up of two main axis categories which were not of the same origin. Firstly, a single seminal root emerging from the seed at germination, and secondly a short adventitious root derived from the collar. The seminal root (SR1) had an orthotropic growth direction and showed monopodial development. On average, it measured 5.10 cm in length and bore two lateral roots (SR2s). The SR1 was conical-shaped while SR2s were cylindrical. The SR1 did not show any secondary growth, but proximal root tissue turgidity may explain the conical shape observed (this explanation is also valid for the next observed conical

Statistical analysis The different axis categories were separated on the basis of their morphological traits and this was

-3-

shaped root). SR2s were shorter and thinner than the SR1. They showed monopodial development and an ageotropic growth direction, i.e. they did not adjust their growth direction after their emission. SR2s were emitted at right angles to the SR1. SR1s and SR2s constituted the seminal root system. The first order adventitious root (AR1) showed monopodial development, a plagiotropic growth direction and a

conical shape. It was unbranched and thinner than the SR1. Ten days later (on average on the 24th day after germination; Fig. 1b), SR1 was still present and there was almost no change in its length and diameter. Its apex usually showed necrosis and it bore 2.7 lateral roots on average. There were six AR1s and they had lengthened since the 14th day stage. Every AR1

-4-

possessed the same properties. AR1 emission followed an acropetal sequence. The primary diameter of successive roots increased. On average, each AR1 gave rise to 13 lateral roots (AR2s), which were shorter and thinner than the AR1s. AR2s showed monopodial development, an ageotropic growth direction and were inserted at a right angle on the AR1s. Around the 35th day of development (Fig. 1c), the neo tuber arose just below the AR1 emission zone on the collar. The seminal root system had completely disappeared due to its necrosis. On average 12 AR1s were present and were clearly more branched than in previous stages. We noticed that the AR1 are emitted following a continuous rhythm. There were 40 AR2s on average on each AR1. AR2s were branched and were now bearing a third order category axis (AR3). AR3s were shorter and thinner than the AR2s. They showed monopodial development, a cylindrical shape and an ageotropic growth direction. They were supported by AR2s and inserted at a right angle. In the last studied stage (76th day, Figs. 1d, 2c, 4b), the tuber continued to grow while half of the AR1s had disappeared; on average six AR1s were still alive. Their mean length was shorter that at the previous stage because of necrosis. The number and length of

the other axis types (AR2s and AR3s) had decreased for the same reason, resulting in the presence of numerous necrosed AR3s.

Minisetts root system On average, five tuber roots were observed 7 days after the first tuber root occurrence (TR1, Figs. 2d, 3a). These were located heterogeneously throughout the minisett epidermis. TR1s initiated in the cortical layer under the tuber’s epidermis, showed monopodial development, a plagiotropic growth direction and a conical shape. On average they presented one lateral root (TR2) shorter and thinner than the TR1s. TR2s showed an ageotropic growth direction and right angle insertion on TR1. The TR1s and their lateral roots (TR2s) formed the tuber root system. As long as TR1s remained alive, tuber tissues remained tight, whereas if the TR1s died at this stage, this induced tuber rot. The tuber root system continued to show qualitatively similar properties for an average of 40 days (until the 49th day of development, Fig. 3b). TR1s and TR2s grew in length and TR1s supported more lateral roots (16 lateral roots on average). The major event at this stage was the formation of the

-5-

primary nodal complex (PNC) which is preferentially initiated along the epidermic layer border but can also occur, at a lower frequency, throughout the entire epidermic surface of the minisett. By the 66th day of development (Fig. 3c), the tuber root system was still present but began to show necrosis. At this point the TR1s bore on average 20 TR2s. On average, four AR1s emerged from the rooting zone on the PNC which is located across the entire PNC surface except the upper and lower poles. These AR1s showed monopodial development, a plagiotropic growth direction, a conical shape and a diameter two- to three-fold that of TR1s. AR1s were not branched at this stage. Forty days later (around the 107th day on average, Figs. 2e, 3d, 4d), the tuber root system had completely disappeared. AR1s had increased in length and gave rise to lateral roots (AR2s). AR2s bore AR3s. There were on average 12 AR2s by AR1 and 13 AR3s by AR2. Secondary and tertiary roots showed monopodial development and an ageotropic growth direction. The length and diameter of each order of axis decreased from the AR1s to AR3s. AR2s and

AR3s were both inserted at right angles on their supporting axis. At the last studied stage (119th day on average, Fig. 3e) more than half the AR1s showed necrosis. Of all the AR1s in the previous stage, only two fifths of the roots were still alive. Their mean length had increased to a maximum of 1.62 m. We observed on average 51 AR2s by AR1 and 103 AR3s by AR2. The new tuber was in formation at the lower part of the PNC, under the AR1s insertion zone. We noticed that determinate growth and different lateral root life spans in all ramified roots led to conservation of typical morphology. Every ramified structure composed of a bearing root and its laterals can be subdivided in three parts. The distal part is near to the apex, where lateral roots are short and still growing. In the central part, lateral roots have reached their maximal length. Thirdly, in the basal part, the lateral roots necrosis process results in a shortening of lateral roots (Figs. 1, 3). We noted that lateral roots (SR2s, TR2s, AR2s and AR3s) were not regularly distributed on their bearing roots and did not follow any visible rhizotaxy.

-6-

Analysis of root system development

and IBD (interbranch distance). Mean lengths (for all stages) of AR1s, AR2s and AR3s were 23.50 cm, 6.00 cm and 0.96 cm, respectively. Fig. 5b shows that there is an intersection of their distribution, but they are significatively different (Table 1). For parameters linked to the branching process, differences have been tested between AR1s and AR2s. Their mean LAUZ were respectively 8.37 cm and 3.17 cm. Fig. 5c shows that there is no intersection of their distributions. A significant difference for LAUZ has also been found (Table 1). For the mean number of lateral roots (for all stages), values are 12.67 for AR1s and 32.6 for AR2s. Fig. 5d shows a heavy intersection of their distributions. We found that their numbers of lateral roots are not significantly different (Table 1). Lastly, for AR1s and AR2s’ IBD (for all stages), which are respectively of 1.43 cm and 0.29 cm, there

Root system hierarchy In order to validate the axes types, the discriminations previously made by qualitative analyses needed to be statistically tested (Fig. 5, Table 1). We then verified the effectiveness of differences by statistical tests (Table 1). Concerning diameters, all AR1s were between 2.30 and 3.40 mm, AR2s were between 0.32 and 0.80 mm and AR3s were between 0.28 and 0.35 mm. Fig. 5a shows that there was no intersection between there respective distributions. Statistical analysis therefore confirmed that the three adventitious roots categories were significatively different in diameter (Table 1). The same approach was used for length, LAUZ, number of lateral roots

-7-

is no intersection in their distribution (Fig. 5e). The significant difference observed on graphic is validated by the statistical test (Table 1). The same approach was used to differentiate root axis categories of the tubercular root system and seedling plants root systems. The quantitative measurements and statistical results are outlined in the Table 1 and the qualitative features are outlined in the architectural table (Tables 2 and 3). Axis categories in each root system were discriminated using combinations of their qualitative features. The qualitative character states providing information for the discrimination consist of growth modalities (indeterminate or determinate), growth direction, branching and reiteration abilities, deciduousness and shape. Concerning quantitative features, the axes were differentiated by length, diameter, LAUZ and inter-branch distance. The number of lateral roots is a less powerful discriminative factor; for instance, in minisett root systems, the first and second adventitious roots do not have a significantly different number of lateral roots. All axis types of a root system showed a decrease in their LAUZ, in diameters and in lengths from the first order axis to higher order axes. Adventitious first order roots and seminal or tubercular first order root are clearly different: firstly because of their distinct origin. Secondly, they have significatively different length, diameter (except for SR1 and seedling AR1), number

of lateral roots, inter-branch distance and LAUZ.

Reiteration Some lateral roots behaved differently from the other roots emitted on the same axis (Fig. 6). These roots possessed almost the same qualitative and quantitative properties as their bearing root: determinate or indeterminate growth, growth direction, shape and branching ability. Their length and diameter were intermediate and could not be used as discriminating factors. Their emission angle was different from those of lateral roots: lateral roots were emitted at right angles whereas these particular roots were emitted at around 60° and generally showed a secondary reorientation following the growth direction of the bearing root. These roots were emitted when the root apex suffered necrosis or encountered a strong mechanical constraint. This branching process did not seem to be sequential. It may be concluded from these observations that these roots are indeed reiterates, and more particularly, traumatic reiterates. We noted that AR1s (Fig. 6) in both seedling and minisett plants showed a high incidence of traumatic reiterates, whereas AR3s were far less frequently reiterated. Not all the components in the root systems had the same reiterative properties. Tubercular roots (Fig. 6) were also able to reiterate whereas the first and second order seminal

-8-

roots of both adventitious root systems have a longer lifespan than those of seminal and tuber roots and both adventitious root systems began to show necrosis shortly after neo-tuber formation.

roots did not appear to be able to reiterate.

Dynamic approach The seminal root system of seedling plants (Fig. 7a) developed for approximately 20 days before disappearing through necrosis. From germination to the 19th day, the length of the first order seminal root and the number of its lateral roots increased regularly. First, adventitious roots occurred at around the 13th day; they occurs successively, lengthen and regularly ramify up to the 42nd day. This was followed by the root necrosis process and tuber initiation. The diameter of successive first order roots increased. Lateral roots (SR2s, AR2s and AR3s) showed determinate growth and their diameter did not change from one root to the next emitted. Concerning minisetts root systems (Fig. 7b), all the first order tubercular roots were emitted over a short period of time (4 to 5 days). They then lengthened and regularly ramified up to the 55th day whereas their diameter remained unchanged. The PNC was initiated at around the 49th day. The tubercular root system disappeared at the same time that first order adventitious root branching occurred (around the 72nd day). AR1s lengthened and branched until occurrence of the neo-tuber (119th day). The diameter of successive first order roots increased. Lateral roots (TR2s, AR2s and AR3s) showed determinate growth and their diameter did not change from one root to the next emitted. Lastly,

-9-

Total root length

The first of these two AU is formed by the SR1 and its laterals. The second is composed of an AR1 and its associated AR2s and AR3s. We called these the “seminal architectural unit” and “adventitious architectural unit”. In the same manner as for seedling plant, the minisett plant root system is composed of different root axis categories: the first order tubercular roots (TR1s), the second order tubercular roots (TR2s), the first order adventitious roots (AR1s), the second order adventitious roots (AR2s) and the third order adventitious roots (AR3s). These axis categories, which are organized in a strong hierarchy, form two distinct AUs. The first of these two AUs is formed by a TR1 and its laterals and is called the “tubercular architectural unit”. The second is composed of an AR1 and its associated AR2s and AR3s and forms the “adventitious architectural unit”. Here the axis categories correspond to the axis orders (except for reiterates) but it is important to note that this is not always the case (Atger and Edelin 1994b). Jourdan and Rey working on oil-palm found that the adventitious root system is organized and also found a root architectural unit (Jourdan and Rey 1997a). Atger also highlighted the organization of a root system with an architectural unit in five dicotyledonous plants (Atger 1992). Even if the architectural units for these different taxa are not equivalent, they can be used to describe and understand root system development. It emphasizes that the architectural unit is a good tool to describe the root system ontogenesis. In rice (Harada et al. 1993, Morita and Yamazaki 1993) and banana (Lassoudière 1971, Swennen et al. 1986, Araya 2003, Belalcazar et al. 2003), the adventitious root system is also composed of three root axis orders. These similarities may reflect a conserved developmental pattern among some Monocotyledons but further investigations are needed to determine the degree of similarity between their architectural units.

The results obtained for total root length were used to compare root system dimensions at each developmental stage. Two systems succeeded one another over time in minisett and seedling plant. Adventitious root systems succeeded seminal root system in seedling plants and tubercular root systems in minisett plants. The development of both systems was subdivided into two main phases: first the main axes (SR1s, TR1s and AR1s) became established and lengthened, then underwent ramification. This resulted in a substantial increase in total root length (Fig. 8a,b). We noted, in both cases, that the second component (adventitious) is far more developed than the first (seminal and tubercular): 170 times more for the minisett plants (Fig. 8b) and 1,400 times more for seedling plants (Fig. 8a). The decline of the first root system (seminal or tubercular) coincides with the branching phase of the adventitious root systems. At this moment, root system dimensions are multiplied by a factor 100. DISCUSSION Root system structure and functions

Architectural units Architectural analysis consist in characterizing quantitatively a structure which may appear random (like Fig. 4b), it is the first step when comparing the effects of several treatments on plant architecture or when performing growth simulation (Vercambre et al. 2003). This structural study of the seedling plant root system allowed us to identify root axis categories with distinctive qualitative and quantitative morphological features. The different axis categories were distinguished by means of a set of features in a similar manner to distinguishing between aerial vegetative structures (Edelin 1984, Barthélémy and Caraglio 2007). These axis categories for seedling plant root system consisted of the seminal root (SR1), the second order seminal roots (SR2s), the first order adventitious roots (AR1s), the second order adventitious roots (AR2s) and the third order adventitious roots (AR3s). They were then grouped together into two different architectural units (AUs).

Comparison of seedling and minisett root system The seminal root system is composed of an architectural unit (AU) which is not duplicated, unlike the tubercular and adventitious root systems which are made up of several repetitions of their AU. The numerous repetitions of adventitious AUs give the yam seedling and minisett root system its fasciculate

- 10 -

appearance. Fasciculate rice or oil-palm root systems are also made up of architectural unit repetitions (Morita and Yamazaki 1993, Jourdan et al. 1995) and this may be a frequent process in Monocotyledons. Seedling and minisett root systems were found to show equivalent architectural units. The first architectural units correspond to the seminal and tubercular AUs. They are both composed of two axis categories. The seminal AU and the tubercular AU possess the same qualitative properties (except concerning the growth direction of SR1 and TR1s). They have equivalent relative quantitative properties in comparison to those of adventitious AUs; for example: their length, diameter (except for seedling root system) and growth rate are significatively less than those of adventitious root systems. But they are not emitted from the same structure: from the seed for SR1 and from the tuber’s epidermis for TR1s (Wickam et al. 1980). The adventitious architectural units of seedling and minisett root system are both composed of three root axis categories. They have the same qualitative features, and equivalent relative quantitative properties. Moreover, they are emitted from quite similar structures. Effectively, according to Wickam, the PNC seems to be a structural equivalent of the seedling hypocotyl (Wickam et al. 1980).

hierarchy into account to refine the root system interpretation.

Root functions It may be assumed that the morphological differentiation in axis categories causes a functional differentiation of these categories. Based on this hypothesis, each root axis category can be linked to its main suspected functions. Morphological traits used to discriminate between root functions are length, diameter, shape, growth rate, lifespan and number of lateral roots (Atger and Edelin 1994b). Thus, in seminal and tubercular root systems, their SR1 and TR1s which are branched, conical-shaped (here due to the root turgidity), long and have high growth rate and large diameter should play roles of exploration and colonization. The role of exploitation in these root systems should be played by SR2s and TR2s. Concerning adventitious root systems, their AR1s, which are long, conical-shaped and have large diameter and fast growth rate should play a main role in exploring the environment. Their plagiotropic growth direction drives this exploration into the humus layer. AR2s are shorter and have many lateral roots: they should play a major role in colonisation. Lastly, a large number of AR3s play an exploitation role. As Dioscorea rotundata is a liana, the main function for the root system is water and nutrient uptake and transport. Moreover it may be supposed that the anchorage function is mainly assumed by SR1, TR1s, and the AR1s for both adventitious root systems. These show the greatest diameters and the largest number of lateral roots. The emission of lateral roots at right angles also reinforces their anchorage ability. The emission of lateral roots at right angles is observed in many other Monocotyledon plants (Chopart et al. 1970, Jourdan and Rey 1997a, b, Belalcazar et al. 2003, Draye et al. 2003) as optimum branching angle to resist pull out (Stokes et al. 1996).

Hierarchy Each architectural unit shows a strong hierarchical system. This hierarchy translates as the differential expression of qualitative and quantitative traits throughout root axis categories. With regard to qualitative features, we observed transitions in character states from first order axis category to higher rank axis categories: oriented (orthotropic or plagiotropic) to ageotropic axis, medium or long-term to short term deciduousness, branched to unbranched axis, indeterminate (AR1s) to determinate growth and total to partial reiteration ability (except for seminal root system). With regard to quantitative traits, from first order axis category to higher rank axis categories, the length, diameter, lifespan, growth rate, reiteration rate, number of lateral roots (except for minisett adventitious root system) and LAUZ decrease. Strong hierarchies have also been found in root systems of Monocotyledons such as banana (Swennen et al. 1986), rice (Harada et al. 1993), coconut (Colas 1997) and oil-palm (Jourdan and Rey 1997a). On the contrary, it can be observed in a dicotyledonous plant like Prunus persica L. Batsh, that the root system appears as a broad continuum between the two extreme root types. However, each root type present large variability, especially in its growth (Vercambre et al. 2003), indicating a weaker hierarchy. The pronounced hierarchy within each yam architectural unit sheds light on an important morphological differentiation process. Axes among the branched system do not all have the same properties. Thus, the root system analysis, concerning function, development and growth, should take this

Structural repetitions We have seen that adventitious root systems are formed by multiple duplications of their architectural unit. In line with the definition of reiteration, we can suggest that all adventitious roots are total and sequential reiterates. The same suggestion may be made concerning the tubercular root system. It lends the plant efficient spatial occupation (Atger and Edelin 1994b). These duplications of equivalent structures are particularly interesting when attempting to acquire accurate information concerning the entire root system by observing only one architectural unit. In the field, such a technique will save time when evaluating root system development and growth. Other mechanisms of duplications are expressed in response to environmental stochasticity: i.e. traumatic reiterations. We suspect that this ability is responsible for a large part of root system plasticity in yam as it allows the root system to react to

- 11 -

environmental constraints. In this way, the traumatic reiteration rate in the root system is an indicator of a favourable or unfavourable soil environment. Despite its role in yam root system adaptability, the reiteration process has not yet been taken into account by researchers. Reiteration process is common mechanism in monocotyledon plants (Swennen et al. 1986, Harada et al. 1993).

branching of the first order adventitious roots and the decline of the temporary root system, the root system expanded in size by a factor 100. These observations have implications in matters of crop management. For instance, during temporary root development, fertilizer use could be restricted to an area corresponding to the temporary root distribution. According to our result, the definitive root system reaches maximum efficiency after branching of the first order adventitious roots. The second fertilizer input should take this period into account and the spatial expansion of the adventitious root system. However, given that our observations were made in a greenhouse, they cannot be extrapolated to field management. Determining when and where to apply fertilizer in the field would require the calibration of developmental sequences and root length. The tuber is a key factor in seedling and minisett plant lifecycles. Our results show that rapid development of the neo-tuber coincides with necrosis of the adventitious root system. It may be supposed that the constitutive elements of the adventitious root systems are at this point translocated into the neotuber. This coincidence of these two events has been observed previously by several authors (Njoku et al. 1973, Clairon and Zinsou 1980, Okezie et al. 1981). The observations made by Degras tend to prove that constitutive elements are translocated from the entire plant to the neo-tuber during its formation. This phenomenon results in necrosis of both shoot and root zones (Degras et al. 1980). Here, we noted that the developments of parts of the plants were interrelated. This mechanism illustrates the importance of simultaneously studying the different parts of the plant in order to understand fully its development. The disappearance of shoots and roots does not mean plant death as its perennity is based on the transfer of reserves from the whole plant to a deciduous storage organ, the tuber. If we consider seedling and minisett plants to be two states of a pluriannual lifecycle, intermediate stages ontogenesis, which has not been described in this study, should provide interesting new information concerning plant developmental strategies.

Temporary vs. definitive root system We have previously seen that seedling and tuber plant root systems are composed of two main components: seminal or tubercular root systems and adventitious root systems. Both have similar morphological and functional properties. We have called the first the “temporary root system” because of its short lifespan. The second system has been called the “definitive root system” because it remains alive until total disappearance of the root system which has a far longer lifespan than temporary root system. Studies on total root length and ontogenesis in seedling and minisett root systems have shown that the definitive root system succeeds the temporary root system (Fig. 8a,b). In fact, as we saw with total root lengths, the decline in the temporary root system corresponds to rapid development of the definitive root system. It may be supposed that this phenomenon is accompanied by the translocation of constitutive elements from the temporary root system to the adventitious root system. This notion was further supported by the study of development in the ontogenic stage where events were positioned on a timescale (Fig. 7). This showed a close coincidence in time between temporary root system necrosis and AR1s branching. Similar phenomena of constitutive element translocation have been demonstrated in other Monocotyledons. In Sorghum, its carbon stocks are translocated during the decline of seminal roots to allow the branching of first order adventitious roots (Blum et al. 1977). Also, in sugarcane, stem cuttings emit a temporary root system which behaves in the same manner as yam temporary root systems (Blackburn 1984, Smith et al. 2005). According to Namoro, there would appear to be a physiological link between this temporary root system and the definitive root system in sugarcane (Namoro 1983). If the first order axes of the definitive root system are pruned on their progressive occurrence, the temporary root system remains alive for a period of several additional months. This strategy founded on the succession of a temporary and a definitive system should be a frequent process throughout the vegetal kingdom. In fact, dicotyledonous plants such as Myriophyllum spicatum L. (Patten 1956) or Coussapoa latifolia (Prosperi et al. 1995) and generally all hemi-epiphytic plants also possess a definitive root system replacing a temporary root system. The total root length study was used to determine root system dimensions at each stage of development (Fig. 8a,b). At the stage corresponding to the

CONCLUSION An architectural analysis of the yam root system in seedling plants and minisett plants was used to identify different root axis categories which are morphologically distinct. Their morphological traits are related to their functional features and result in typical behaviours. We were able to distinguish between root axis categories that are more implicated in exploration, in colonization or in exploitation. The root axis structure of yam provides information concerning their function as anchorage, exploration and exploitation should help understand root system development and thus help us better manage crops. This morphological study has to be reinforced by an anatomical analysis in order to refine our

- 12 -

comprehension of root system developmental processes. The different root categories are arranged in conserved structural and temporal patterns and form two different root architectural units. The two fundamental structures are organized according to a strong hierarchy. From the main axis category to the peripheral axis category, this hierarchy is based on axis specialization. This arrangement results in coherent structures called the architectural units which are responsible for all root system functions. The entire root system is formed by several repetitions of these architectural units caused firstly by sequential processes but also by traumatic processes. Considering the whole root system as an arrangement of architectural units has highlighted a succession of two functional systems: a temporary and a definitive root system. These systems are found in seedling as well as in minisett plants. The temporary root system plays an important role in mobilization and reserve storage. Thereafter, these reserves may contribute to the rapid development of the definitive root system. This succession appears to be a frequent strategy in the vegetal kingdom but more in-depth studies need to be performed to confirm this assertion. This result should prompt us to reconsider the root system as an organized structure capable of great plasticity and adaptability, in part due to reiteration processes. Understanding structure is a necessary step to acquire an integrative view of the root system and cannot be ignored. In practice, the architectural approach facilitates root system studies as invariant and repeated structures can be detected by means of just a few measurements. In the field, simple descriptors such as diameter and length coupled with structural considerations such as number of lateral orders and reiterations can be used to characterize root system ontogenic stages and provide information concerning the extent of environmental constraints. It may be concluded, for the yam root system, that this tool is relevant and does not lead to ambiguous results. This is a first architectural analysis of the root system of a tuber monocot which provides a framework for the description of the root systems of similar plants. Lastly, it can be used to obtain accurate information on root systems from an agronomic perspective and for modelling studies.

Fabre and Denis Filloux for helpful suggestions on greenhouse management during this study and to Michaël Gueroult for help with plant handling. REFERENCES Araya M (2003) Stratification and spatial distribution of the banana (Musa AAA, Cavendish subgroup, cvs ‘Valery’ and ‘Grande naine’) root system. In: Turner DW & Rosales E (eds) Banana root system: towards a better understanding for its productive management. San José, Costa Rica, pp 83–103, 2003 Atger C (1992) Essai sur l’architecture racinaire des arbres. Université Montpellier II, France, p 287 PhD Thesis Atger C, Edelin C (1994a) Premières données sur l’architecture comparée des systèmes racinaires et caulinaires. Can J Bot 72:963–975 doi:10.1139/b94-122 Atger C, Edelin C (1994b) Stratégies d'occupation du milieu souterrain par les systèmes racinaires des arbres. Rev Ecol (Terre Vie) 49:343–355 Barthélémy D, Caraglio Y (2007) Plant architecture: a dynamic, multilevel and comprehensive approach to plant form, structure and ontogeny. Ann Bot (Lond) 99:375–407 doi:10.1093/aob/mcl260 Barthélémy D, Edelin C, Hallé F (1989) Architectural concepts for tropical trees. In: Holm-Nielsen LB, Balslev H (eds) Tropical forests: botanical dynamics, speciation and diversity. Academic, London, pp 89–100 Barthélémy D, Edelin C, Hallé F (1991) Canopy architecture. In: Raghavendra AS (ed) Physiology of trees. Wiley and sons, Chichester, pp 1–20 Belalcazar S, Rosales FE, Pocasangre LE (2003) Development and formation of plantain roots (Musa AAB Simmonds). In: Turner DW, Rosales E (eds) Banana root system: towards a better understanding for its productive management. San José, Costa Rica, pp 75–82, 2003 Blackburn F (1984) Sugarcane. Longman, London and New York, p 414 Blum A, Jordan WR, Arkin GF (1977) Sorghum root morphogenesis and growth. I. Effect of maturity genes. Crop Sci 17:149–153

ACKNOWLEDGEMENTS The authors would like to acknowledge with thanks the funding received from the ARPEGE project (Architecture racinaire des plantes: facteurs environnementaux et génétiques) of the French Agricultural Research Centre for International Development (CIRAD). AMAP (Botany and Computational Plant Architecture) is a joint research unit which associates CIRAD (UMR51), CNRS (UMR5120), INRA (UMR931), IRD (R123), and Montpellier 2 University (UM27); http://amap.cirad.fr/ Thanks are extended to Denis

Chopart JL (1970) Morphologie et croissance de l'enracinement du sorgho (Sorghum vulgare) en deux conditions de fertilité. I.S.T.O.M., Bambay, p 97 Clairon M, Zinsou C (1980) Etude de plantations échelonnées d'igname Dioscorea alata cv. Lupias : effet du vieillissement du tubercule sur la croissance et le développement de la plante. In: Institut National de la Recherche Agronomique

- 13 -

(ed) Séminaire international de Pointe-à-Pitre, Pointe-à-Pitre, pp 125–141, 1980

Harada J, Yamazaki K, Morita S (1993) Science of the rice plant. Chapter 2 Roots—Root. In: Matsuo T, Hoshikawa K (eds) Science of the rice plant. Food and Agriculture Policy Research Center, Tokyo, pp 133–161

Colas H (1997) Association de cultures cocotier cacaoyer. Modélisation de leur système racinaire. Etudes préliminaires sur l’interaction racinaire et la consommation en eau des deux plantes. Université Montpellier II, Montpellier, p 260 PhD Thesis

Jourdan C, Rey H, Guédon Y (1995) Architectural analysis and modelling of the branching process of the young oil-palm root system. Plant Soil 177:63–72 doi:10.1007/BF00010338

Collet C, Löf M, Pages L (2006) Root system development of oak seedlings analysed using an architectural model. Effects of competition with grass Plant Soil 279:7–383 doi:07/s11104-0052419-9

Jourdan C, Rey H (1997a) Architecture and development of the oil-palm (Elaeis guineensis Jacq.) root system. Plant Soil 190:217–233 doi:10.1023/A: 1004290024473

Cornet D (2005) Etude du fonctionnement physiologique d'un couvert végétal d'igname (Dioscorea alata L.). In: Sciences Agronomiques (ed.) Faculté Universitaire des Sciences Agronomiques, Gembloux, pp 115

Jourdan C, Rey H (1997b) Modelling and simulation of the architecture and development of the oil-palm (Elaeis guineensis Jacq.) root system. Plant Soil 189:33–48 doi:10.1023/A: 1004218030608

Danjon F, Fourcaud T, Bert D (2005) Root architecture and wind-firmness of mature Pinus pinaster. New Phytol 168 (2):387–400

Khuder H, Stokes A, Danjon F, Gouskou K, Lagane F (2007) Is it possible to manipulate root anchorage in young trees? Plant Soil 294(12):87–102 doi:10.1007/s11104-007-9232-6

Degras L (1994) L'igname. W CTA, Maisonneuve et Larose (ed) Paris, pp 133

Lassoudière A (1971) La croissance des racines du bananier. Fruits 26(7-8):501–512

Degras L, Sanner C, Vautor A, Poitout R, Suard C, Arnolin R (1980) Nouvelles données sur l'évolution morphologique pondérale de l'igname cousse-couche, D. trifida, en culture en plein champ. In: Séminaire international de Pointe-àPitre, Pointe-à-Pitre, pp 233–253, 1980

Liedgens M, Richner W (2001) Minirhizotron observations of the spatial distribution of the maize root system. Agron J 93:1097–1104 López-Bucio J, Cruz-Ramırez A, Herrera-Estrella L (2003) The role of nutrient availability in regulating root architecture. Cur Op Plant Biol 6:208–287

Draye X, Lecompte F, Pages L (2003) Distribution of banana roots in time and space: new tools for an old science. In: Turner DW, Rosales E (ed) Banana root system: towards a better understanding for its productive management. San José, Costa Rica, pp 58–74, 2003

Malamy JE (2005) Intrinsic and environmental response pathways that regulate root systems architecture. Plant Cell Environ 28:67–77 doi:10.1111/j.1365-3040.2005.01306.x

Dumont R, Chopart JL (1992) Amélioration du système de production de l'igname dans la zone de contact FORET/ SAVANE Ivoirienne. CIRAD-IRAT, Bouaké, p 55

Martin FW (1974) Tropical yams and their potential. USDA, Washington, p 18 Morita S, Yamazaki K (1993) Science of the rice plant. Chapter 2 Roots—Root system. In: Matsuo T, Hoshikawa K (eds) Science of the rice plant. Food and Agriculture Policy Research Center, Tokyo, pp 161–186

Edelin C (1977) Images de l'architecture des conifères. Université Montpellier II, Montpellier, p 255 Edelin C (1984) L’architecture monopodiale: l’exemple de quelques arbres d’Asie Tropicale. Université Montpellier II, France, p 259 PhD Thesis

Namoro B (1983) Influence de l'acidité d'un sol sableux sur le développement racinaire et le rendement de la canne à sucre dans un périmètre irrigué de Haute-Volta. Institut National Polytechnique de Lorraine, Vandoeuvre-lèsNancy, France, p 148

Ferguson TU (1972) The propagation of yams by vine cutting: a critical review. Trop root crops 5:4–7 newsletters

Njoku JE, Okonkwo SNC, Nwoke FIO (1973) The pattern of growth and development in Dioscorea rotundata Poir. In: CLA. Leakey. (ed.) 3rd International Symposium of the ISTRC, Ibadan, Nigeria, pp 1–6, 1973

Hallé F, Oldeman RAA (1970) Essai sur l’architecture et la dynamique de croissance des arbres tropicaux. Masson (ed) Paris, pp 178 Hallé F, Oldeman RAA, Tomlinson PB (1978) Tropical trees and forests. Springer-Verlag, Berlin, p 441

Okezie CEA, Okonkwo SNC, Nweke FI (1981) Mode de développement et analyse de la

- 14 -

croissance de l'igname blanche cultivée à partir de semences. In: Terry ER, Oduro KA, Caveness F (eds) Tropical Root Crops: Research Strategies for the 1980s. IDRC, Ottawa, Canada, pp 180– 188

tree root system in an orchard using a quantitative modelling approach. Plant Soil 251:1–11 doi:10.1023/A:1022961513239 Wickam L, Wilson LA, Passam HC (1980) The anatomical origin of the primary nodal complex during vegetative propagation in yam (Dioscorea spp.) tubers. In: Séminaire international, Pointeà-Pitre, pp 179–184, 1980

Oldeman RAA (1974) Architecture de la forêt guyanaise. O.R.S.TO.M., Paris, p 204 Orkwor GC, Asiedu R, Ekanayake IJ (1998) Food Yams as determined by chromosome counting and flow cytometry. Advances in research. IITA and NRCRI, Nigeria, p 249 Otoo JA, Osiru Techniques semenceaux d'Agronomie 57

Yang C, Yang L, Yang Y, Ouyang Z (2004) Rice root growth and nutrient uptake as influenced by organic manure in continuously and alternately flooded paddy soils. Agric Water Manage 70:67–81 doi:10.1016/j.agwat.2004.05.003

DSO, Ng SY, Hahn SK (1985) améliorées de production de d'igname. Institut international Tropicale, Ibadan, Nigeria, pp 11–

Pagès L, Jordan MO, Picard D (1989) A simulation model of the three-dimensional architecture of the maize root system. Plant Soil 119(1):147–154 doi:10.1007/BF02370279 Pagès L, Vercambre G, Drouet J-L, Lecompte F, Collet C, Le Bot J (2004) Root typ: A generic model to depict and analyse the root system. Plant Soil 258(1):103–119 doi:10.1023/B:PLSO.0000016540.47134.03 Patten BC (1956) Notes on the biology of Myriophyllum spicatum L. in a New Jersey Lake. Bull Torrey Bot Club 83:5–18 doi:10.2307/2482818 Prosperi J, Edelin C, Michaloud G (1995) Modalités d'occupation de l'espace aérien par les hémi-épiphytes : le rôle de la réitération caulinaire. Rev Ecol (Terre Vie) 50:167–187 Smith DM, Inman-Bamber NG, Thorburn PJ (2005) Growth and function of the sugarcane root system. Field Crops Res 92:169–183 doi:10.1016/j.fcr.2005.01.017 Stokes A, Ball J, Fitter AH, Brain P, Coutts MP (1996) An experimental investigation of the resistance of model root systems to uprooting. Ann Bot (Lond) 78:415–421 doi:10.1006/anbo.1996.0137 Swennen R, De Langhe E, Janssen J, Decoene D (1986) Study of the root development of some Musa cultivars in hydroponics. Fruits 41:515–524 Tobin B, Čermák J, Chiatante D, Danjon F, Di Iorio A, Dupuy L et al (2007) Towards developmental modelling of tree root systems. Plant Biosyst 141(3):481–501 doi:10.1080/11263500701626283 TrouslotMF (1985) Analyse de la croissance et morphogenèse de l'igname Dioscorea complexe D. cayenensis-D. rotundata. Université ClermontFerrand II, France, p 370 PhD Thesis Vercambre G, Pagès L, Doussan C et al (2003) Architectural analysis and synthesis of the plum

- 15 -