Tree Physiology 33, 409–424doi:10.1093/treephys/tpt015
Effect of source/sink ratios on yield components, growth dynamics and structural characteristics of oil palm (Elaeis guineensis) bunches
Benoît Pallas1,5, Isabelle Mialet-Serra2, Lauriane Rouan3, Anne Clément-Vidal3, Jean-Pierre Caliman4 and Michael Dingkuhn3
1Montpellier SupAgro, UMR AGAP, Avenue d’Agropolis, F-34398 Montpellier cedex 5, France; 2CIRAD, DG, Avenue d’Agropolis, F-34398 Montpellier cedex 5, France; 3CIRAD, UMR AGAP, Avenue d’Agropolis, F-34398 Montpellier cedex 5, France; 4SMART Research Institute, Pekanbaru 28112, Indonesia; 5Corresponding author ([email protected])
Received August 29, 2012; accepted January 31, 2013; published online March 26, 2013; handling Editor Annikki Mäkelä
Source/sink ratios are known to be one of the main determinants of oil palm growth and development. A long-term experi-ment (9 years) was conducted in Indonesia on mature oil palms subjected to continuous bunch ablation and partial defolia-tion treatments to artificially modify source/sink ratios. During the experiment, all harvested bunches were dissected and phenological measurements were carried out to analyse the effect of source/sink ratios on yield components explaining variations in bunch number, the number of fruits per bunch and oil dry weight per fruit. An integrative variable (supply/demand ratio) describing the ratio between the assimilate supply from sources and the growing organ demand for carbohy-drate was computed for each plant on a daily basis from observations of the number of developing organs and their sink strength, and of climate variables. Defoliation and bunch ablation affected the bunch number and the fruit number per bunch. Variations in bunch number per month were mainly due to variations in the fraction of aborted inflorescence and in the ratio between female and male inflorescences. Under fluctuating trophic conditions, variations in fruit number per bunch resulted both from changes in fruit-set and in the number of branches (rachillae) per inflorescence. For defoliated plants, the decrease in the number of developing reproductive sinks appeared to be sufficient to maintain fruit weight and oil concentration at the control level, without any major decrease in the concentration of non-structural carbohydrate reserves. Computation of the supply/demand ratio revealed that each yield component had a specific phase of sensitivity to supply/demand ratios during inflorescence development. Establishing quantitative relationships between supply/demand ratios, competition and yield components is the first step towards a functional model for oil palm.
Keywords: fruit filling, fruit-set, inflorescence abortion, oil concentration, sex determination, source/sink ratio, yield components.
Introduction
The ratio between source and sink activities results from the balance between the organ assimilate demand for growth and maintenance and the whole-plant assimilate supply through photosynthesis or reserve mobilization (Sadras and Denison 2009). The ratio between organ assimilate demand and assim-ilate supply at the plant scale has been found to be one of the main factors affecting plant growth and development (Marcelis
et al. 2004, Mathieu et al. 2009). When the assimilate supply/demand ratio is low at the plant scale due to abiotic constraints or low incident radiation, vegetative growth decreases (Tardieu et al. 1999 in maize and sunflower, Lafarge et al. 2010 in rice, Pallas et al. 2011 in grapevine). A lower (or higher) supply/demand ratio leads to a larger (smaller) seed number or seed size depending on the species (see Sadras and Denison 2009 for a review). When considering plants as a meta-population of organs, changes in plant growth and development in response
to variable supply/demand ratios have been interpreted as resulting from competition-like mechanisms between organs. To limit the impact of competition between sinks on yield, thin-ning and pruning are carried out to modify source and sink activities at the whole-plant level (Seehuber et al. 2011).
Oil palm (Elaeis guineensis Jacq.) is a perennial, monoecious species displaying indeterminate development, producing suc-cessive phytomers on a mono-axial shoot. Throughout its pro-ductive period of ~30 years, each phytomer can bear a female or male inflorescence, or none at all if abortion occurs. For oil palm, yield varies considerably from 1 month or 1 year to the next when compared with variability in vegetative growth (Legros et al. 2009a). A large share of that variability depends on source/sink relationships. Several studies have shown that oil palms bear fewer female inflorescences, due to changes in the sex ratio or abortion rates, in response to low assimilate demand (defoliated plants) (e.g., Durand-Gasselin et al. 1999). Based on the high variation in the sex ratio and abortion fre-quencies, it was suggested that yield variations mainly result from changes in the proportion of female inflorescence, as opposed to bunch yield components, or the duration of inflo-rescence development (Corley and Tinker 2003). However, variability in each yield component in response to different ratios between organ assimilate demand and assimilate supply at the plant scale is yet to be characterized quantitatively.
The oil palm female inflorescence has a single branched peduncle and a large number of rachillae (branches) (see
Figure 1 for the yield component breakdown). Each branch (rachilla) bears many flowers (see the figure available as Supplementary Data). Each flower can produce a normal fruit with a high oil concentration in the mesocarp, or a small fruit with a low oil concentration if pollination does not succeed (called small parthenocarpic fruits by the industry, Corley and Tinker 2003, figure 1; also see the figure available as Supplementary Data). As mentioned above, the bunch number for any given period depends on the proportion of phytomers bearing a female inflorescence, but it also depends on the rate at which new phytomers are produced and the time taken for an inflorescence to develop and grow. A recent study showed that the source/sink ratio affected the rate of phytomer appear-ance (Legros et al. 2009b), but nothing is known about the fruit growth rate in oil palm.
Each yield component on any given phytomer is determined over a long period from initiation of the inflorescence meristem to bunch maturity (~3 years for mature palms, Adam et al. 2011). The long time lapse between the determination of yield components and their impact on assimilate demand at the whole-plant level and the fact that photosynthesis activity is not modified in response to different fruit loads in the oil palm leads to major source/sink imbalances (Legros et al. 2009a, 2000b, 2000c). These source/sink imbalances are mostly buffered by a large carbohydrate reserve pool (mostly starch) in the stem (Legros et al. 2009b, 2000c). Although many studies (Breure and Menendez 1990, Corley and Tinker 2003, Adam et al.
410 Pallas et al.
Figure 1. Description of yield components for the oil palm.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
2011) have investigated the timing of yield component determi-nation during inflorescence development, no consensus exists on the existence and timing of specific phases of sensitivity to supply/demand ratios at the whole-plant level.
Trade-offs between yield components have been observed in many species (Gambin and Borrás 2010). In oil palm, a high fruit number per bunch can cause smaller individual fruit weights (e.g., Breure and Menendez 1990). This negative cor-relation may be a result of source limitation during fruit filling for bunches with a large number of fruits, but spatial limitation for fruit growth within a dense inflorescence might also explain this phenomenon (López-Pereira et al. 1999). Confusion about the different factors involved in bunch weight variability can theoretically be solved if the co-variances between yield com-ponents are taken into account and if quantitative terms describing supply/demand ratios at the plant scale during bunch growth are included in the analysis.
Quantifying the ratio between assimilate supply from sources and the demand from growing organs during plant develop-ment is a scientific challenge taken up by various authors. In many studies, sink demand has been derived from fruit load (Kelner et al. 2000) and assimilate supply has been estimated using the plant leaf area (Vaast et al. 2005, Etchebarne et al. 2010) or solar radiation intercepted by the canopy (Monteith 1994). These approaches are not totally satisfactory for oil palm because they do not take into account variability in fruit number per bunch, causing changes in bunch sink strength, or environmental effects on the assimilation rate. In this study we calculated supply/demand ratios at the plant scale, at a daily time step for each plant. Estimation of assimilate demand con-siders the number of growing sinks and their sink strength. Organ sink strength depends on the organ type and on its developmental stage. Assimilate supply estimation is based on canopy structure and climatic variables.
The aims of this work were to (i) evaluate the impact of source/sink ratios on each yield component, (ii) estimate trade-offs between yield components and (iii) build quantitative rela-tionships between the computed supply/demand ratios and yield components during specific phases of inflorescence development. Analyses of non-structural carbohydrate content in stems served to analyse reserve dynamics and their poten-tial buffer function. Nine years of continuous observations on control plants, and partially defoliated and bunch-ablated plants, were used to assess these issues.
Materials and methods
Plant material and growing conditions
The study was carried out between June 2002 and November 2010 in an experimental plantation of the SMART Research Institute (SMARTRI, Smart Tbk.) located in the Kandista estate (Riau province, Indonesia, 0°55′0′N, 101°21′0′E). The hybrid
studied was a commercial high-yielding tenera hybrid (Elaeis guineensis Jacq.), called 64, resulting from the cross between a plant of Deli dura origin and a plant of Avros pisifera origin. The plants were 8 years old at the start of the study.
The site is characterized by a tropical humid climate. The high-est rainfall generally occurs between November and January (average precipitations equal to 220 mm month−1 from 1994 to 2011). The lowest rainfall generally occurs between June and August (120 mm month−1 from 1994 to 2011). The mean daily temperature is 27.2 °C and the mean daily solar radiation is 18.0 MJ m−2 day−1. Temperature and solar radiation vary slightly from one month to another. The mean annual climatic water bal-ance (precipitation − reference evapotranspiration; Allen et al. 1998) is 720 mm year−1. The climatic characteristics of this site are considered as ‘optimum’ for oil palm cultivation.
Experimental design
The experimental plot was part of a larger, long-term genetic experiment covering ~30 ha. The planting density was 136 plants ha−1 in a 9.5 m equilateral triangular pattern. Fertilization was based on leaflet nutrient contents (average values during the experiment: 790 g N plant−1 year−1, 350 g P2O5 plant−1 year−1, 1350 g K2O plant−1 year−1, 270 g MgO plant−1 year−1). Three treat-ments were applied at the beginning of the experiment (June 2002) (four plants per treatment). The first treatment (Control) corresponded to plants without architectural modification or inflo-rescence ablation. For the second treatment (called hereafter FPT, for Fruit Pruning Treatment), from June 2002 to January 2010, all new inflorescences were cut just before anthesis. As for the Control plants, the number of expanded and photosynthetically active leaves for FPT plants was maintained at 35 throughout the experiment. For the third treatment (called hereafter LPT, for Leaf Pruning Treatment) at the start of the treatments, all leaves except the 15 youngest leaves were excised at the petiole extremity. Subsequently, for LPT plants, from June 2002 to January 2010, when a new leaf appeared the oldest leaf of the crown was cut to maintain a total of 15 expanded leaves in the crown. With FPT, the source/sink ratios within the plant were increased by decreasing sink assimilate demand and LPT decreased this ratio by decreas-ing source activity. In January 2010 the pruning treatments were stopped for FPT and LPT plants. As a consequence, harvesting of new bunches started in June 2010 for the FPT plants.
Microclimate, thermal time and water balance calculation
Daily rainfall and solar radiation, daily minimum and maximum relative humidity and temperature, and the mean daily wind-speed (at a height of 2 m) were recorded during the experi-ment using a weather station located near the experimental zone (<1 km).
The thermal time between two dates was computed using a trapezoidal response curve between the effective temperature and daily minimum and maximum temperatures (White et al.
Effect of source/sink ratios on oil palm yield components 411
2005). The four cardinal temperatures (Tbase, Topt1, Topt2 and Tlim) used to compute thermal time were estimated to be 11, 26, 30 and 48 °C, respectively, based on previous obser-vations at this experimental site (Legros et al. 2009a).
A water balance was used to estimate the amount of avail-able water in the soil during the experiment. The water balance was an adaptation of the generic Food and Agriculture Organization model (Allen et al. 1998). This balance was first used by Legros et al. (2009a) and Combres et al. (2013), and was calibrated using previous observations on root depth and stomatal response to soil water depletion (Legros et al. 2009c). The soil water balance was used to calculate the daily FTSW (fraction of transpirable soil water; Sinclair and Ludlow 1986).
Plant measurements and estimated variables
Phenological measurements were taken on each plant approxi-mately every 12–14 days from November 2002 to November 2010 (180 measurement dates per plant during the experi-ment). Each phytomer was labeled, monitoring the leaves appearing on the plant since the beginning of the experiment (~200 phytomers per plant were monitored during the experi-ment). Three types of events were recorded for each phytomer: (i) the date of the leaf opening out from the spear, (ii) the date of spathe opening and (iii) and the date of bunch harvesting. The sex of each inflorescence was also recorded as soon as it was known (male, female or aborted if no inflorescence was produced on the phytomer). Bunches were harvested at matu-rity, when fruits began to fall. For the aborted inflorescence, a hypothetical spathe opening date was estimated using a linear regression between the spathe opening date for the previous and following phytomers bearing an inflorescence. For each month, these hypothetical dates were estimated to be able to compute the proportion of the different types of inflorescences for which spathe opening occurred. Indeed, each month the proportion of female inflorescences was estimated as the ratio between the number of female inflorescences and the total number of inflorescences (female + male + aborted) for which spathe opening occurred during the month in question. The proportion of aborted and male inflorescences was estimated in the same way. Note that it was possible to know the sex of all the inflorescences for FPT plants because inflorescence ablation occurred after spathe opening (inflorescence sex is known at that stage). For each phytomer, the time lapse between the appearance of two successive leaves (i.e., phyl-lochron) and the time lapse between leaf appearance, spathe opening and harvesting were computed in degree-days.
All bunches harvested between November 2002 and November 2010 (828 bunches) were dissected to assess the different yield components. Bunches were split into four parts: rachis, rachillae, small unfertilized fruits with a low oil concen-tration (called hereafter small parthenocarpic fruits) and fertilized fruits with oil. For each sample, the water content was
estimated on a sub-sample placed in a drying oven for 48 h at 104 °C to assess its dry weight. Note that, in this study, the fruit dry weight per bunch (Figure 1) only refers to the total weight of fertilized fruits per bunch. The number of fertilized fruits with oil and small parthenocarpic fruits was counted. The number of flowers per bunch was estimated as the sum of the number of fertilized and small parthenocarpic fruits, as flowers can give rise to either small parthenocarpic fruits or fertilized fruits after fertilization. The ratio between the number of fertil-ized fruits and the number of flowers (called hereafter the frac-tion of fertilized fruits) was then calculated. The ratio between the number of flowers and the rachilla dry weight was deter-mined (called hereafter the flower number/rachillaDW). The mean individual fruit dry weight was computed for each bunch as the ratio between the fruit dry weight per bunch and the fruit number per bunch.
Individual fruit measurements were made on a randomized sample of fertilized fruits (30 fruits). The mesocarp to fruit dry weight ratio was then estimated on the sample. On these sam-ples, from October 2009 to November 2010, kernel and meso-carp oil was extracted with hexane in a Soxhlet extractor to assess oil concentration in the mesocarp and kernel.
The petiole, rachis and leaflet dry weights were measured for each collected leaf. For each leaf, the specific leaf area was estimated on a leaflet sample. Each leaf area was estimated as the ratio between the total leaflet weight and the specific leaf area.
Every 2 months, from September 2002 to December 2005, starch concentration in the stem was measured. The samples were collected as previously described by Legros et al. (2009c) and analysed by high-performance liquid chromatog-raphy. Samples were collected at three heights in the stem (at each third of the stem from the ground to the base of the low-est leaf in the crown).
Computation of a supply/demand ratio
For each plant, a supply/demand ratio was computed based on an estimation of assimilate demand for the growing sinks (D) and an estimation of assimilate supply (S).
Each day, the plant assimilate supply (S(t), kg CH2O day−1 plant−1) was computed according to the Monteith formalism (Monteith 1977) and the Beer–Lambert law (Vose et al. 1995), as follows :
S t
PR t k t K t( ) . ( )( exp( ( ))) ( )= − −10
0 48 1dens
s s potLAI ε
(1)
where Rs(t) is the daily amount of solar radiation (MJ m−2); LAI(t), the daily leaf area index; εpot, the radiation use efficiency in the absence of a water deficit; Ks(t), the daily water stress coefficient according to soil water status (Allen et al. 1998); Pdens, the planting density; and k, the Beer–Lamber extinction coefficient
Tree Physiology Online at http://www.treephys.oxfordjournals.org
(Table 1). εpot was assumed to be constant because no change in maximum photosynthesis activity for mature palms in response to sink demand (or fruit load) has been observed (Legros et al. 2009b). LAI was computed using the number of expanded leaves in the crown and the observed mean leaf area. The leaf area was considered to be constant because of its low variability (Cv = 0.09) compared with that of the plant leaf number. Ks was estimated using a broken-stick function as proposed by Allen et al. (1998). When FTSW is greater than the break point of the function (aFTSW, Table 1) ks = 1. For FTSW between aFTSW and 0, Ks linearly decreases from 1 to 0.
Assimilate demand for growth was computed according to the carbohydrate demand for the vegetative and reproductive parts of the plant. Each day, the number of growing leaves (Nl(t)) was estimated based on their appearance date (opening out from the spear) and considering that leaf growth occurred in the 5 months before their appearance (assumption based on observations). Assimilate demand was considered to be con-stant for each leaf over the growth period, and daily assimilate demand for the vegetative parts of the plant (Dv(t)) was thus determined as follows:
D t C W
T N tv vl
ll( ) . ( )=
10 52
(2)
where Cv is the chemical cost of the vegetative part, Tl is the leaf growth duration, Wl is the mean dry weight of the leaf (peti-ole + leaflets + rachis) and 0.52 is a coefficient to compute the total vegetative assimilate demand from leaf demand (based on observations and Dufrêne 1989 and Combres et al. 2013).
The reproductive demand per plant was computed as the sum of (i) the demand for the structural parts of bunches (Ds) (rachis and rachillae), (ii) the demand for the part of the fruits without oil (Dno) and (iii) the demand for the part of the fruits composed of oil (Do). To compute the carbohydrate demand for the structural part of bunches, (Ds(t)), we considered that the growth of this compartment began 4 months before spathe opening and stopped at maturity (based on observations and Corley and Tinker 2003). To compute demand for the part of the fruits with and without oil (Do, Dno), we assumed that growth of the part of the fruit without oil began at spathe opening and ended at harvesting and that growth of the part of the fruit with oil began 2 months before harvest and ended at harvest (based on observations and Corley and Tinker 2003). For these three compartments, we assumed that daily demand was constant throughout the growth period and was equal to the ratio between their observed mean dry weights during the experi-ment (Ws and Ifr, respectively, for the mean dry weight of the structural part of the bunch and the mean individual fruit dry weights) and their growth durations (Ts, Tno and To, respectively,
Effect of source/sink ratios on oil palm yield components 413
Table 1. Parameters used for computing the supply/demand ratio, the method of estimation and the values.
Parameter name
Definition Unit Value Method of estimation
Carbohydrate supply computationεpot Potential radiation use efficiency (before
implementation of water deficit effect)gCH2O MJ−1 4.6 Legros et al. (2009b); Combres et al. (2013)
k Extinction coefficient for the Beer–Lambert law _ Dufrêne (1989)aFTSW FTSW threshold for water deficit effect on
assimilation (Allen et al. 1998)_ Legros et al. (2009c)
Pdens Planting density Plant ha−1 Field observations_ Mean leaf area m²Carbohydrate demand computation (vegetative part)Wl Mean leaf dry weight g 4980 Field observationsTl Leaf growth duration days 150Cv Chemical cost of the vegetative part gCH2O g DM−1 1.4 Penning de Vries et al. (1989)Carbohydrate demand computation (reproductive part)Ws Mean dry weight of the structural part of bunches
(rachis + rachillae)g 1510 Field observations
Wfr Mean individual fruit dry weight g 6.4Ts Growth duration of the structural part of bunches days 300 Field observations; Corley and Tinker (2003)Tno Growth duration of the fruit compartment without oil days 180To Growth duration of the fruit compartment with oil days 60Cs Chemical cost of the structural part of bunches gCH2O g DM−1 1.4 Penning de Vries et al. (1989)Cno Chemical cost of the fruit compartment without oil 1.4Co Chemical cost of the fruit compartment with oil 3.2OMe Mean oil content in mesocarp g g DM−1 0.80 Field observationsOKe Mean oil content in kernel 0.52RMe/Fr Mean mesocarp/fruit ratio 0.77_ Daily maintenance respiration gCH2O day−1 plant−1 560 Dufrêne (1989); Combres et al. (2013)
Tree Physiology Volume 33, 2013
for the structural part of the bunches, for the part of the fruit without oil and for the part of the fruit composed of oil). We estimated biomass partitioning between the part of the fruit composed of oil and the part of the fruit without oil using the mean observed ratio between mesocarp and total fruit dry weight (RMe/Fr) and oil content in the kernel (OKe) and mesocarp (OMe). The daily reproductive demand of each reproductive compartment was thus computed as follows:
Ns refers to the number of bunches at the stage when struc-tural compartment growth occurred. Nfr,no is the total number of fruits between spathe opening and maturity. Nfr,0(t) is the number of fruits in the oleosynthesis phase (Table 1). Cs, Co and Cno are, respectively, the chemical cost for the structural part of bunches, for the part of the fruit with oil and for the part of the fruit without oil (Table 1).
Maintenance respiration (Mr) was also taken into account as proposed by Dufrêne (1989) assuming constant respiration for a given mature plant. The monthly value of the supply/demand ratio was calculated as the ratio between monthly assimilate supply and monthly assimilate demand.
Correlations between variables and statistical analysis
Statistical analyses were performed using R-software (R Development Core Team 2007). Pearson coefficients between variables were calculated and their significances were tested. Pearson coefficients between the supply/demand ratio and yield components were computed considering different time lapses between the supply/demand ratio variations and their impact on yield components. The slope and intercept of the fitting lines were calculated using linear regression for the rela-tionships between the supply/demand ratio and yield compo-nents, or using standardized major axis regressions for the relationships between yield components (Warton et al. 2006). Analyses were performed on transformed variables (the ratio between the value of the variable and its mean value during the experiment xi/x) to allow comparisons between the slopes of the different regressions. Treatment (FPT, LPT, Control) effects were tested using a one-way analysis of variance (ANOVA) considering plants as replicates and Tukey tests were used as post-hoc tests for multiple mean comparisons.
Results
Variation in total harvested fruit dry weight, the number of harvested bunches and fruit dry weight per bunch
For both the LPT and Control treatments, large inter- and intra-annual fluctuations in total harvested dry weight per month were observed throughout the experiment, with values ranging from 0 to 22.2 kg month−1 plant−1 (Figure 2). If we considered the data collected during the leaf pruning period (November 2002 to January 2010), LPT plants had significantly (P < 0.05) lower total harvested fruit dry weight compared with the Control plants (−36%, mean values = 5.6 for LPT and 9.2 kg for the Control). However, the impact of this pruning practice varied during the leaf pruning period. The LPT effect was par-ticularly strong and significant in 2004–05 and 2008–09, whereas no significant effect (P > 0.05) was observed between November 2002 and March 2004. Total harvested fruit dry weight for the LPT plants took 3 months to reach Control levels after the end of the leaf pruning treatment (January 2010). It remained similar to the Control thereafter. Bunch ablation (FPT) was stopped at the same time as LPT. In the following 6 months no bunches were produced on the FPT plants, but thereafter the total harvested fruit dry weight was >35 kg plant−1 month−1, four times as much as the Control.
The correlations between the number of harvested bunches and the total harvested fruit dry weight were highly significant, whether we considered each treatment separately, or the
414 Pallas et al.
Figure 2. Total harvested fruit dry weight per plant and per month dur-ing the experiment for the three treatments. Each point represents the mean values for total harvested fruit dry weight observed over 3 months. The arrow indicates the end of leaf pruning and bunch ablation for the LPT and FPT plants. A one-way ANOVA was performed to test for significant differences between treatments considering plants as replicates. ‘·’ significant at 0.05 ≤ P < 0.1, *significant at 0.01 ≤ P < 0.05, **significant at 0.001 ≤ P < 0.01 and ***significant at P < 0.001. Treatments with different letters were significantly different at P < 0.05. Long-dash dark and grey lines indicate the mean total harvested fruit dry weight during the experiment for the Control and LPT, respectively.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
whole dataset (r = 0.87 for the whole dataset, Table 2). The correlations between the total harvested fruit dry weight and fruit dry weight per bunch were also significant but with lower coefficients of correlation. Correlations between fruit dry weight per bunch and the number of harvested bunches were not significant. Throughout the experiment the number of har-vested bunches displayed a larger coefficient of variation than fruit dry weight per bunch (Figure 3a and b). Figure 3a and b present the treatment effects on yield components averaged for three periods: (i) the initial year of differential pruning treat-ments, (ii) the subsequent 8 years of treatment and (iii) the 6 months when bunch production for the FPT plants had resumed after bunch ablation treatments. The impact of LPT on the num-ber of harvested bunches was significant between November 2003 and June 2010 (−13%, P < 0.05). After June 2010, the number of harvested bunches for the LPT plants remained sim-ilar to the Control. This number was more than twice the Control level for the FPT plants at the end of the pruning treat-ments (Figure 3a). The fruit dry weight per bunch was signifi-cantly reduced for the LPT plants between November 2002
and July 2010 (−26%) and returned to the Control level there-after (Figure 3b). Halting fruit pruning for the FPT plants led to an approximately two-thirds increase in fruit dry weight per bunch compared with the Control.
Effect of the source/sink ratio on the number of harvested bunches
Variability in the number of harvested bunches results from changes in the fraction of female inflorescences and in the potential number of harvested bunches (i.e., the number of bunches if all inflorescences are female and non-aborted). The potential number of inflorescences depends on the rate of phy-tomer production which, on average, is equal to the reciprocal of the phyllochron. The phyllochron was not affected by LPT but it was greatly reduced by FPT throughout the experiment (P < 0.001, Figure 4a). For the FPT plants, the rate of inflores-cence production was further accelerated by a shortened period between leaf appearance and the opening of the corre-sponding spathe (Figure 4b). Conversely, this period was slightly but significantly increased for the LPT plants between
Effect of source/sink ratios on oil palm yield components 415
Figure 3. Effect of treatments on variables explaining the variability in total harvested fruit dry weight: (a) the number of harvested bunches and (b) fruit dry weight per bunch. For each period and variable, a one-way ANOVA was performed to test for significant differences between treat-ments considering plants as replicates. ns, not significant, *significant at 0.01 ≤ P < 0.05, **significant at 0.001 ≤ P < 0.01 and ***significant at P < 0.001. Treatments with different letters were significantly different at P < 0.05. Cv is the coefficient of variation computed for each yield component.
Table 2. Coefficients of correlation, slopes of the regression lines and standard deviation of the slopes (mentioned below the coefficient of correlation) between total harvested fruit dry weight, the number of harvested bunches and fruit dry weight per bunch.
Control LPT FPT All treatments
Total harvested fruit dry weight vs. the number of harvested bunches 0.87*** 1.01 ± 0.02
0.82*** 0.83 ± 0.02
0.88*** 1.69 ± 0.10
0.87*** 0.96 ± 0.02
Total harvested fruit dry weight vs. fruit dry weight per bunch 0.41*** 0.69 ± 0.10
0.48*** 0.91 ± 0.12
0.49*** 1.71 ± 0.31
0.57*** 1.02 ± 0.07
Number of harvested bunches vs. fruit dry weight per bunch −0.08 ns −0.00 ns −0.26 ns −0.00 ns
Slopes and standard deviation were calculated using standard major axis regressions on transformed values for yield components (xi/ x− ).ns, not significant.*** and bold, Significant at P < 0.001.
Tree Physiology Volume 33, 2013
November 2003 and July 2010. The duration between spathe opening and maturity (the duration of bunch maturation) was not affected by the treatments throughout the experiment (Figure 4c).
Sex determination and abortion frequency determine the fraction of inflorescences that develops into harvestable bunches. From November 2002 to July 2010 a significant decrease in the fraction of female inflorescences was observed for the LPT plants compared with the Control plants (37% for LPT and 46% for the Control, Figure 4d). The LPT effect on the fraction of female inflorescences was probably caused by an increase in abortion frequency because the fraction of male inflorescences was not significantly modified (Figure 4e and f). After July 2010, the fraction of female inflo-rescences for the LPT plants returned to the Control level. Fruit Pruning Treatment involved an increase in the fraction of female inflorescences (Figure 4d, mean value = 0.79) throughout the experiment that could be explained by a sig-nificant decrease in the fraction of male or aborted inflores-cences (Figure 4e and f).
Variation in the number of fertilized fruits per bunch and mean individual fruit dry weight
Although no significant LPT effect on the number of fertilized fruits was observed between November 2002 and October 2003, that number was significantly reduced from November 2003 to November 2010 (−26%, Figure 5a). Conversely, no
significant change was observed for the mean individual dry fruit weight per bunch throughout the experiment for LPT (Figure 5b). The harvested bunches of the FPT plants between July and November 2010 had a larger number of fertilized fruits per bunch (+75%, Figure 5a) and lighter fruits (−12%, Figure 5b) compared with the Control. For bunches harvested throughout the experiment, the number of fertilized fruits per bunch displayed a higher coefficient of variation than the mean individual fruit dry weight.
Variations in the dry weight per bunch were more strongly correlated to variations in the number of fertilized fruits per bunch than to variations in the mean individual fruit dry weight (Table 3). Indeed, the correlations between fruit dry weight per bunch and the number of fertilized fruits per bunch were highly significant for the three treatments, whereas the correlations between fruit dry weight per bunch and the mean individual dry fruit weight were not significant for the Control or were significant with lower coefficients of correlation for LPT and FPT. The overall correlations between treatments highlight these results because a robust relationship between the num-ber of fertilized fruits and fruit dry weight per bunch, and a lack of correlation between the mean individual fruit dry weight and fruit dry weight per bunch, were observed. Moreover, a signifi-cant negative correlation between the number of fertilized fruits and the mean individual fruit dry weights was observed, whether we considered the three treatments separately, or the whole dataset.
416 Pallas et al.
Figure 4. Effect of treatments on variables explaining the variability in the number of harvested bunches: (a) phyllochron (duration expressed in thermal time between the appearance of two successive leaves, (b) duration expressed in thermal time between leaf appearance and opening of the female inflorescence spathe, (c) duration expressed in thermal time between inflorescence spathe opening and maturity, (d) fraction of female inflorescences (ratio between female and (female + male + aborted) inflorescences), (e) fraction of male inflorescences, (f) fraction of aborted inflorescences. See Figure 3 legend for explanations of the statistical analyses. Control, black bars; LPT, grey bars; FPT, dark grey bars.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Variation in yield components determining the number of fruits per bunch
Leaf Pruning Treatment and FPT involved changes in all com-ponents, explaining the variation in the number of fertilized fruits per bunch (Figure 6a, b and c). Between November 2002 and June 2010, LPT led to a significant decrease in the fraction of fertilized fruits (−13%) and rachillaDW (−25%) (Figure 6a and b). However, these decreases were no longer observed for the LPT plants between July and November 2010 (6 months after the end of leaf pruning). Conversely, FPT led to an increase in rachillaDW (+156%) for bunches that were har-vested after the end of bunch ablation. No significant change for the fraction of fertilized fruits was observed between the Control and FPT plants between July and November 2010. Flower number/rachillaDW increased for the LPT plants throughout the experiment (significant between November 2003 and June 2010), whereas it decreased for the FPT plants after the end of bunch ablation (Figure 6c). Among the vari-ables explaining variations in the number of fertilized fruits per bunch, rachillaDW had the greatest coefficient of variation.
The number of fertilized fruits per bunch was significantly correlated with rachillaDW and the fraction of fertilized fruits, whether we considered the three treatments separately, or the whole dataset (Table 4). The number of fertilized fruits was poorly correlated with flower number/rachillaDW if we consid-ered the whole dataset. Significant negative correlations were observed for LPT and the Control, and for the whole dataset between rachillaDW and flower number/rachillaDW.
Effect of the source/sink ratio on oil concentration in the fruit and on the mesocarp/fruit ratio
Oil dry weight in the fruit depends on both the mesocarp dry weight and the oil concentration in the mesocarp. Mesocarp dry weight is commonly considered as the product of individ-ual fruit dry weights and the mesocarp/fruit ratio. During the experiment, the three treatments did not display significantly different mesocarp/fruit ratios (0.76, 0.77 and 0.77, respec-tively, for the Control, LPT and FPT) (data not shown). Similarly, oil concentration in mesocarp was not affected by the treat-ments (0.78, 0.80 and 0.79, respectively, for the Control, LPT
Effect of source/sink ratios on oil palm yield components 417
Figure 5. Effect of treatments on variables explaining variability in the fruit dry weight per bunch: (a) the number of fertilized fruits, (b) the mean individual fruit dry weight per bunch. See Figure 3 legend for explanations of the statistical analyses. Control, black bars; LPT, grey bars; FPT, dark grey bars.
Table 3. Coefficients of correlation, slopes of the regression lines and standard deviation of the slopes (mentioned below the coefficient of cor-relation) between fruit dry weight per bunch, the number of fertilized fruits per bunch and the mean individual fruit dry weight.
Control LPT FPT All treatments
Fruit dry weight per bunch vs. the number of fertilized fruits per bunch
0.77*** 0.87 ± 0.03
0.82*** 0.94 ± 0.03
0.60*** 0.79 ± 0.09
0.85*** 0.86 ± 0.02
Fruit dry weight per bunch vs. the mean individual fruit dry weight
0.08 ns 0.24*** 1.33 ± 0.07
0.38*** 0.31 ± 0.27
0.04 ns
Number of fertilized fruits per bunch vs. the mean individual fruit dry weight
−0.46*** −0.87 ± 0.03
−0.32*** −1.46 ± 0.08
−0.56*** −2.55 ± 0.28
−0.43*** −2.10 ± 0.21
Slopes and standard deviations were calculated using standard major axis regressions on transformed values for yield components (xi/x−).
ns, not significant.*** and bold, significant at P < 0.001.
Tree Physiology Volume 33, 2013
and FPT; data collected between October 2009 and November 2010). Both variables displayed low variations between bunches, with coefficients of variation of 0.03 and 0.05, respectively, for oil concentration in mesocarp and the meso-carp/fruit ratio (data not shown).
Dynamics of the whole-plant supply/demand ratio
Figure 7 presents an example of the calculated plant assimilate demand and supply for one individual Control plant and one LPT plant, along with key determinants of demand (the number of growing and oil accumulating fruits, Figure 7a and b) and supply (drought level, solar radiation, active leaf number, Figure 7c and d). Changes in these variables for these exam-ples were representative of the changes found for the other Control and LPT plants studied. Assimilate demand ranged from ~1.2 to 4.0 kg CH2O day−1 for the Control plants, from 1.3 to 3.2 kg CH2O day−1 for the LPT plants and from 1.2 to 1.8 for the FPT plants (data not shown for FPT) (Figure 7a and b). Assimilate supply ranged between 1.1 and 2.2 kg CH2O day−1
for the Control and FPT plants. Leaf Pruning Treatment led to a 28% decrease in assimilate supply (Figure 7c and d). Changes in assimilate demand were mainly linked to the number of growing fruits in the oleosynthesis phase (R2 = 0.87). If we considered each treatment separately, changes in assimilate supply were mainly linked to solar radiation (R2 = 0.55, 0.52 and 0.43, respectively, for the Control, LPT and FPT). This high impact of solar radiation on assimilate supply resulted from a low water deficit incidence in this environment (except in 2006) and small changes in the active leaf number within each treatment. As shown in Figure 7, the supply/demand ratio pre-sented large intra-month variability. This ratio varied between 0.4 (drought spell in 2006) and 1.8 (Figure 7c) for the Control plants. If we considered all the studied plants, the mean sup-ply/demand ratio during the experiment was 1.08, 1.40 and 0.92, respectively, for the Control, FPT and LPT. If we consid-ered the three treatments together, the supply/demand ratio was significantly correlated to assimilate supply (R2 = 0.32) and to assimilate demand (R2 = 0.54).
418 Pallas et al.
Figure 6. Effect of treatments on variables explaining variability in the number of fertilized fruits per bunch: (a) fraction of fertilized fruits (the num-ber of fertilized fruits/the number of flowers per inflorescence). (b) rachillaDW per inflorescence, (c) the number of flowers per inflorescence divided by the rachillaDW per inflorescence, See Figure 3 legend for explanations of the statistical analyses. Control, black bars; LPT, grey bars; FPT, dark grey bars.
Table 4. Coefficients of correlation, slopes of the regression lines and standard deviation of the slopes (mentioned below the coefficients of correla-tion) between the number of fertilized fruits per bunch and the yield components explaining variability in the number of fertilized fruits per bunch.
Control LPT FPT All treatments
Number of fertilized fruits per bunch vs. fraction of fertilized fruits
0.56*** 1.28 ± 0.11
0.61*** 0.77 ± 0.06
0.19 ns 0.49*** 1.13 ± 0.08
Number of fertilized fruits per bunch vs. rachillaDW
0.54*** 0.53 ± 0.05
0.48*** 0.47 ± 0.05
0.44** 0.32 ± 0.11
0.74*** 0.63 ± 0.02
Number of fertilized fruits per bunch vs. flower number/rachillaDW
0.13* 0.14 ± 0.05
0.01 ns 0.70*** 1.10 ± 0.22
0.04 ns
Fraction of fertilized fruits vs. flower number/rachillaDW
−0.03 ns −0.01 ns −0.18 ns −0.08 ns
Fraction of fertilized fruits vs. rachillaDW
0.10* 0.05 ± 0.02
−0.06 ns −0.42** 0.10 ± 0.02
0.06 ns
Flower number/rachillaDW vs. rachillaDW
−0.57*** −0.53 ± 0.04
−0.53*** −1.13 ± 0.10
−0.17 ns − 0.45*** −0.40 ± 0.03
Slopes and standard deviations were calculated using standard major axis regressions on transformed values for yield components (xi/ x− ).
ns, not significant.* significant at P < 0.05; ** significant at P < 0.01; *** and bold, significant at P < 0.001.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Starch concentration in the stem was analysed every 2 months between September 2002 and November 2005 (Figure 8). Fruit Pruning Treatment significantly increased starch concentration in the top of the stem, except between September 2002 and March 2003. Leaf Pruning Treatment caused a large decrease in starch concentration between January and July 2003 (6–13 months after the beginning of the treatments). After this period, no significant decrease in starch concentration was found for the LPT plants compared with the Control plants. Samples were also collected from the middle and base of the stem (data not presented). The same tenden-cies as those observed for the top of the stem were observed, but starch concentration in these two stem sections was 3–10 times lower than in the top of the stem. For the Control plants, seasonal variations in starch concentration in the stem were observed with values ranging from 30 to 120 mg g DM−1. For the Control and LPT plants, a large decrease in starch
concentration was found from September to November 2003 and May to July 2005. This decrease could be explained by a moderate water deficit over these periods.
Impact of supply/demand ratio during inflorescence growth
Assuming that the supply/demand ratio impacted on yield com-ponents during specific periods of sensitivity during inflorescence/bunch development, we ran yield components vs. supply/demand ratio linear regression analyses for various hypo-thetical time lapses (Figure 9a and b). The supply/demand ratio displayed a significant positive effect on the number of fertilized fruits per bunch, with P < 0.001 at 11–13 months before bunch maturity (coefficients of correlation of ~0.5) and with 0.001 ≤ P < 0.5 at 8–16 months before maturity (Figure 9a). Furthermore, the supply/demand ratio had a significant positive effect on the fraction of fertilized fruits 7–11 months before
Effect of source/sink ratios on oil palm yield components 419
Figure 7. Example of the computation of the carbon balance for one Control plant (a, c, e) and one LPT plant (b, d, f). (a, b) Estimation, for each month, of the mean daily assimilate demand (solid line), the total number of growing fruits in the oleosynthesis phase (dark bars) and the total number of fruits between spathe opening and the beginning of the oleosynthesis phases in each month (grey bars). (c, d) Estimation, for each month, of the daily assimilate supply (dark lines), the number of expanded leaves (grey bars), daily solar radiation (Rs) (dashed grey lines) and water stress coefficient (ks) (grey lines). (e, f) Monthly values of the supply/demand ratio.
Tree Physiology Volume 33, 2013
maturity and on the rachillaDW 11–16 months before maturity. Other weak effects of the supply/demand ratio were observed 25–30 months before maturity on the bunch yield components presented in Figure 9a. The correlation between the supply/demand ratio and the fraction of aborted inflorescences was sig-nificant (P < 0.01) 1–6 months before spathe opening (~6 months before harvesting). This correlation displayed a negative slope and a coefficient of correlation of ~0.50 (Figure 9b). No significant correlation was detected between the other yield components (flower number/rachillaDW, oil concentration, mesocarp/fruit, and the mean individual fruit weight) and the supply/demand ratio, whatever the time lapse.
Discussion
Source–sink relationships greatly modify the number of reproductive organs
This study revealed that source/sink ratios greatly affect oil palm yield (Figure 2). The analysis of yield component variation showed that the yield under fluctuating trophic conditions is mainly determined by changes in the number of growing fruits and not by variations in fruit weight or oil concentration. As previously observed for oil palm (e.g., Corley and Tinker 2003), the number of harvested bunches was affected by source/sink relationships through changes in the sex ratio and abortion rate of inflorescences, and was the yield component that explained most of the yield variability from one month to another (Table 2). However, this study also revealed a
significant impact of source/sink relationships on the number of fruits per bunch. The impact of the source/sink ratio on the number of fruits per bunch was slightly greater than that on the bunch number. Variations (increase for FPT or decrease for LPT) in the fraction of fertilized fruits and rachillaDW were the main determinants for the adjustment of fruit number per bunch (Figure 6). Such variations in the number of seeds/fruits per inflorescence have previously been found to be one of the main determinants of yield for other grasses in response to fluctuating trophic conditions or the environment (e.g., Zhang et al. 2010, Sadras and Slafer 2012). Moreover, as no change in oil palm photosynthesis activity was observed for oil palm in response to a source/sink imbalance (Legros et al. 2009c), as observed for other species (Vaast et al. 2005), an efficient adjustment of the number of growing sinks is necessary to avoid a major source/sink imbalance. Starch in the stem is the main non-structural reserve carbohydrate buffering source/sink imbalances in oil palm (Legros et al. 2009c). The lack of any depletion in starch concentration in the stem for the LPT plants after a time lapse of ~1 year showed that adjusting the number of growing reproductive organs effectively prevents reserve depletion (Figure 8). This absence of any decrease in the carbohydrate reserve is in contradiction with the results showing large decreases in organ carbohydrate concentrations for annual plants in response to a decreasing supply/demand ratio (Kiniry 1993, Yang et al. 2001, Lafarge et al. 2010). Such differences might result from different survival–reproduction trade-offs between annual and perennial plants. Stability in reserve biomass for perennial plants should enhance their abil-ity to increase their life-span and survive unfavourable condi-tions (Vilela et al. 2008). The increase in starch concentration in the stems of the plants undergoing bunch ablation revealed the inability of those plants to increase their vegetative growth rate to use the excess assimilates coming from photosynthesis. However, as the oil palm produces successive phytomers on a monoaxial shoot, the observed increase in the phyllochron for the FPT plants clearly showed that upward adjustments of the potential number of reproductive sinks (i.e., the number of bunches if all inflorescences are female and non-aborted) did occur when resources were abundant.
Fruit characteristics are not affected by the source/sink ratio
In this experiment, the individual fruit weights and oil concentra-tion were found to be less variable when compared with the num-ber of reproductive organs (Figure 5). This lower variability in fruit oil concentration and, to a lesser extent, in the fruit/seed dry weights has been observed in other annual and perennial oilseed crops (Sadras 2007, Trentacoste et al. 2010) under fluctuating source/sink conditions. It reveals the great ability of fruits to cap-ture assimilates resulting from photosynthesis or reserve carbohy-drate during the oleosynthesis phase. Furthermore, no decrease
420 Pallas et al.
Figure 8. Time-course for starch concentration at the top of the stem. Each point represents the mean values of two harvest dates (every 2 months). A one-way ANOVA was performed to test for significant dif-ferences between treatments considering plants as replicates. ns, not significant, **significant at 0.001 ≤ P < 0.01 and ***significant at P < 0.001. A post-hoc analysis was performed and treatments with different letters were significantly different at P < 0.05.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
was found in the time between spathe opening and maturity for the LPT plants (Figure 4). The observed stability of (i) fruit filling duration (Figure 4), (ii) final individual fruit weight (Figure 5) and (iii) oil concentration showed that, whatever the level of the sup-ply/demand ratio during the experiment, no decrease occurred in assimilate accumulation rate during fruit filling, although this pro-cess has a high energetic cost. Some other opposite results have shown a strong decrease in individual fruit weights and, to a lesser extent, in oil concentration under fluctuating trophic conditions for oil seed crops and grasses (Aguirrezábal et al. 2003 on sun-flower, Tanaka and Maddoni 2008 on maize, or Borrás et al. 2004
for a review). These differences might be explained by trade-off between seed number plasticity and seed size plasticity (Sadras 2007). Given this trade-off, the greater plasticity in fruit number and bunches in oil palm which has many growing bunches com-pared with sunflower or maize, which have only one inflorescence, might account for the low plasticity in fruit size. High plasticity in the fruit number might restrict source limitation during fruit filling and enable potential seed or fruit size to be achieved. Differences might also result from the existence of a larger carbohydrate reserve pool in perennial plants, which can be used to fill seeds when assimilates are lacking.
Effect of source/sink ratios on oil palm yield components 421
Figure 9. Analysis of possible time lapses for supply/demand ratio effects on fertilized fruit number, fraction of fertilized fruits and rachillaDW (a) and on the fraction of aborted inflorescences (b) for the Control and LPT plants. The time lapses were calculated from bunch maturity (a) or spathe opening (b). The correlations were performed using transformed variables (ratio between the variable and the mean observed value of the variable during the experiment xi/x
−). Grey, dark grey and black boxes indicate the correlations between the supply/demand ratio and the yield components that are significant at 0.01 ≤ P < 0.05, 0.001 ≤ P < 0.01 and P < 0.001, respectively. White boxes indicate correlations with P ≥ 0.05. The slopes and standard deviations of the regression lines were computed using linear regressions and their values are mentioned for each month when the correlations were significant at P < 0.05.
Tree Physiology Volume 33, 2013
Trade-offs between yield components limit potential yield
The correlation analyses indicated trade-offs between yield components. The most significant trade-off was observed between the number of fertilized fruits per bunch and the mean individual fruit weight (Table 3). Since this trend was observed whatever the trophic situation, it probably reveals assimilate transport restriction or physical space limitation within the bunch as its size increases. This spatial limitation of fruit/seed growth within the inflorescence has been observed in other species (López-Pereira et al. 1999). This trade-off between fruit size and number could also be analysed as proposed by Gambin and Borrás (2010) in a comparison across crop spe-cies. Potential fruit weight and number are established at a simi-lar time between anthesis and fruit-set. According to Smith and Fretwell’s theory (Smith and Fretwell 1974) available resources during this period are proportionally allocated to produce either many small fruits/seeds or fewer larger fruits/seeds. Such a proportional allocation will possibly lead to trade-offs occurring between fruit size and number. However, the impact of such a mechanism remains hypothetical since neutral, positive or neg-ative relationships have been found between seed size and number when considering plants of the same species or geno-type subjected to variations in available resources (Venable 1992). The negative relationship found between flower number per rachilla and rachillaDW (Table 4) probably resulted more from biomass accumulation in the rachillae than from a decrease in the flower number per rachilla. It could be argued that in situa tions of excess assimilates, the surplus cannot be com-pletely buffered by an increase in the number or size of fruits. As a consequence, these plants accumulate biomass in organs displaying a high biomass storage capacity. Such an accumula-tion of biomass in structural parts of the inflorescence has already been observed in many species (Ehdaie et al. 2006 in wheat, Christophe et al. 2008 in Arabidopsis). These studies showed that the carbohydrate stored in structural parts of the inflorescence can also be used as a transitory carbohydrate reserve that can be mobilized to fill seeds.
The negative relationships between yield components revealed the existence of non-linear relationships between each yield component and yield that probably limited potential yield. Quantifying these relationships is fundamental to finding opti-mum ‘target’ values for each yield component that maximizes yield (Loomis and Connor 1996). Establishing these relation-ships is also of major interest to help the modellers to estimate potential yield and to simulate yield component variations (Villalobos et al. 2006, Lecoeur et al. 2011).
Yield components are affected by the plant carbon balance during specific phases
A carbon balance was computed during this study, based on an estimation of both organ assimilate demand and assimilate
supply by leaves (Figure 7). This new algorithm and the result-ing supply/demand ratio helped to explain the variation observed in yield components with trophic constraints that occurred during specific phases of inflorescence development. Similar indicators have been used for annual plants (Kim et al. 2010, during the vegetative phase of sorghum) but, to our knowledge, it is the first time that such a co-variable has been used on tropical perennial plants with an indeterminate devel-opment pattern. Through a regression analysis, this indicator helped to identify phases of sensitivity to the supply/demand ratio during inflorescence development (Figure 9). The tempo-ral organization of these sensitive phases roughly tallied with the literature. Corley and Tinker (2003) found that abortion occurs ~5 months before flowering. In our study, the impact of the supply/demand ratio on abortion was clearly observed from 2 to 6 months before spathe opening (~1 month before flower-ing). The time lapse between the sensitivity to the supply/demand ratio for the different bunch yield components also tal-lied with the elaboration date for each yield component reported by Corley and Tinker (2003) and Adam et al. (2011), except for the fraction of fertilized fruits. The critical period, according to correlation with the supply/demand ratio, was from 7 to 11 months before maturity (~1–5 months before spathe opening), whereas the fraction of fertilized fruits should be logically deter-mined at anthesis (~1 month after spathe opening). However, plant trophic status may already affect flower fertility during flower development or individualization of the floral meristem (occurring from 2 to 8 months before flowering according to Adam et al. 2011).
Conclusions
Under fluctuating trophic conditions, the rate of female inflores-cence production, the rachilla weight per bunch and the frac-tion of fertilized flowers were the main yield components determining the yield. Adjustment of the number of developing reproductive sinks was efficient enough to prevent individual fruit dry weights and oil concentration from being affected by the source/sink ratio and also to avoid a major decrease in carbohydrate reserve concentration, even in situations of decreasing photo-assimilate resources. Potential yield in the oil palm seemed to be limited by compensatory mechanisms at the bunch scale (fruit number vs. fruit weight) whatever the source/sink ratios. Computation of a whole-plant carbon bal-ance is a relevant approach for timing the sensitive phases for each yield component depending on the supply/demand ratio. Integration of the relationships reflecting the impact of the sup-ply/demand ratio on yield components will help to improve simulation models for oil palm. In such models, an auxiliary variable (supply/demand ratio) should incorporate the effect of (i) several types of stress, (ii) the number and potential size of developing sinks and (iii) plant architecture.
422 Pallas et al.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Supplementary data
Supplementary data for this article are available at Tree Physiology online.
Acknowledgments
The SMART Research Institute (SMARTRI), Fahri Arief Siregar and Reni Subawati are acknowledged for their logistic, technical and scientific support. This study was financially supported by PT SMART tbk. This work was also financially supported by the CIRAD through an Action Thématique Programmée (ATP Réserves, project N°11/02, CIRAD) entitled ‘les réserves carbonées chez le cocotier, le palmier à huile, le manguier et l’hévéa: origines, dynamiques et conséquences pour la gestion des plantations’.
Conflict of interest
None declared.
References
Adam H, Collin M, Richaud F, Beulé T, Cros D, Omoré A, Nodichao L, Nouy B, Treager JW (2011) Environmental regulation of sex determi-nation in oil palm: current knowledge and insights from other spe-cies. Ann Bot 108:1529–1537.
Aguirrezábal LAN, Lavaud Y, Dosio GAA, Izquierdo FH, Andrade FH, González LM (2003) Intercepted solar radiation during seed filling determines sunflower weight per seed and oil concentration. Crop Sci 43:152–161.
Allen RG, Pereira LS, Raes D, Smith M (1998) Crop evapotranspira-tion—guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper no. 56. Food and Agriculture Organization of the United Nations, Rome, Italy.
Borrás L, Slafer GA, Otegui ME (2004) Seed dry weight response to source-sink manipulation in wheat, maize and soybean: a quantita-tive reappraisal. Field Crops Res 86:131–146.
Breure CJ, Menendez T (1990) The determination of bunch yield com-ponents in the development of inflorescences in oil palm (Elaeis guineensis Jacq.). Exp Agric 26:99–115.
Christophe A, Letort V, Hummel I, Cournède PH, de Reffye P, Lecoeur J (2008) A model-based analysis of the dynamics of carbon balance at the whole-plant level in Arabidopsis thaliana. Funct Plant Biol 35:1147–1162.
Combres JC, Pallas B, Rouan L, Mialet-Serra I, Caliman JC, Braconnier S, Soulié JC, Dingkuhn M (2013) Simulation of inflorescence dynamics and estimation of environment-sensitive phases: a model based analysis. Funct Plant Biol http://dx.doi.org/10.1071/FP12133.
Dufrêne E (1989) Photosynthèse, consommation en eau et modélisa-tion de la production chez le palmier à huile (Elaeis guineensis Jacq.). Ph.D. thesis, Université Paris Sud Orsay, France.
Durand-Gasselin T, Noiret JM, Kouamé KR, Cochard B, Adon B (1999) Availability of quality of pollen for improved oil palm (Elaeis guinneensis Jacq.) seed production. Plantations, Recherche, Dév 6:264–276.
Ehdaie B, Alloush GA, Madore MA, Waines JG (2006) Genotypic varia-tion for stem reserves and mobilization in wheat: I. Postanthesis changes in internode dry matter. Crop Sci 46:735–746.
Etchebarne F, Ojeda H, Hunter JJ (2010) Leaf:fruit ratio and vine water status effects on Grenache Noir (Vitis vinifera L.) berry composition: water, sugar, organic acids and cations. S Afr J Enol Vitic 31:106–115.
Gambin BL, Borrás L (2010) Resource distribution and the trade-off between seed number and seed weight: a comparison across crop species. Ann Appl Biol 156:91–102.
Kelner JJ, Regnard JL, Lauri PE (2000) Crop load and rootstock effects on maturation rate and harvest quality of cv. Braeburn apples. Fruits 55:73–81.
Kim HK, van Oosterom E, Dingkuhn M, Luquet D (2010) Regulation of tillering in sorghum: environmental effects. Ann Bot 106:57–67.
Kiniry JR (1993) Non structural carbohydrate utilization by wheat shaded during grain growth. Agron J 85:844–849.
Lafarge T, Martin M, Bueno C, Clément-Vidal A, Schreck E, Luquet D (2010) Regulation and recovery of sink strength in rice plants grown under changes in light intensity. Funct Plant Biol 37:413–428.
Lecoeur J, Poiré-Lassus R, Christophe A, Pallas B, Casadebeigh P, Deadaeke P, Vear F, Guilioni L (2011) Quantifying physiological determinants of genetic variation for yield potential in sunflower. SUNFLO: a model-based analysis. Funct Plant Biol 38:246–259.
Legros S, Mialet-Serra I, Caliman JP, Siregar FA, Clement-Vidal A, Dingkuhn M (2009a) Phenology and growth adjustments of oil palm (Elaeis guineensis) to photoperiod and climate variability. Ann Bot 104:1171–1182.
Legros S, Mialet-Serra I, Caliman JP, Siregar FA, Clement-Vidal A, Fabre D, Dingkuhn M (2009b) Phenology, growth and physiological adjustments of oil palm (Elaeis guineensis) to sink limitation induced by fruit pruning. Ann Bot 104:1183–1194.
Legros S, Mialet-Serra I, Clement-Vidal A, Caliman JP, Siregar FA, Fabre D, Dingkuhn M (2009c) Role of transitory carbon reserves during adjustments to climate variability and sink-source imbalances in oil palm (Elaeis guineensis Jacq.). Plant Physiol 29:1189–1211.
Loomis RS, Connor DJ (1996) Crop ecology. Productivity and manage-ment in agricultural systems. Cambridge University Press, Cambridge, UK.
López-Pereira M, Sadras VO, Trápani N (1999) Genetic improvement of sunflower in Argentina between 1930 and 1995. I. Yield and its components. Field Crops Res 62:157–166.
Marcelis LFM, Heuvelink E, Baan Hofman-Eijer LR, Den Bakker J, Xue LB (2004) Flower and fruit abortion in sweet pepper in relation to source and sink strength. J Exp Bot 55:2261–2268.
Mathieu A, Cournède PH, Letort V, Barthélémy D, de Reffye P (2009) A dynamical model of plant growth with interactions between devel-opment and functional mechanisms to study plant structural plastic-ity related to trophic competition. Ann Bot 103:1173–1186.
Monteith JL (1977) Climate and the efficiency of crop production in Britain. Philos Trans R Soc Lond B 281:277–294.
Monteith JL (1994) Principles of resource capture by crop stands. In: Monteith JL, Scott RK, Unsworth MH (eds) Resource capture by crops. Nottingham University Press, Nottingham, UK, pp 1–15.
Pallas B, Loi C, Christophe A, Cournède PH, Lecoeur J (2011) Comparison of three approaches to model grapevine organogenesis in conditions of fluctuating temperature, solar radiation and soil water content. Ann Bot 107:729–745.
Penning de Vries FWT, Jansen DM, ten Berge HFM, Bakema A (1989) Simulation of ecophysiological processes of growth in several annual crops. Pudoc, Wageningen, The Netherlands.
R Development Core Team (2007) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
Sadras VO (2007) Evolutionary aspects of the trade-offs between seed size and number in crops. Field Crops Res 100:125–138.
Sadras VO, Denison RF (2009) Do plants parts compete for resources? An evolutionary viewpoint. New Phytol 183:565–574.
Effect of source/sink ratios on oil palm yield components 423
Tree Physiology Volume 33, 2013
Sadras VO, Slafer GA (2012) Environmental modulation of yield com-ponents in cereals: heritabilities reveal a hierarchy of phenotypic plasticity. Field Crops Res 127:215–224.
Seehuber C, Damerow L, Blank M (2011) Regulation of source: sink relationship, fruit set, fruit growth and fruit quality in European plum (Prunus domestica L.) using thinning for crop load management. Plant Growth Regul 65:335–341.
Sinclair TR, Ludlow MM (1986) Influence of soil water supply on the plant water balance of four tropical grain legumes. Aust J Plant Physiol 13:329–341.
Smith CC, Fretwell SD (1974) The optimal balance between size and number of offspring. Am Nat 108:499–506.
Tanaka W, Maddoni GA (2008) Pollen source and post-flowering source–sink ratio effects on maize kernel weight and oil concentra-tion. Crop Sci 48:666–677.
Tardieu F, Granier C, Muller B (1999) Modelling leaf expansion in a fluc-tuating environment: are changes in specific leaf area a consequence of changes in expansion rate? New Phytol 143:33–43.
Trentacoste ER, Puertas CM, Sadras VO (2010) Effect of fruit load on oil yield components and dynamics of fruit growth and oil accumula-tion in olive (Olea europaea L.). Eur J Agron 32:249–254.
Vaast P, Angrand J, Franck N, Dauzat J, Génard M (2005) Fruit load and branch ring-barking affect carbon allocation and photosynthesis of leaf and fruit of Coffea arabica in the field. Tree Physiol 25:753–760.
Venable DL (1992) Size-number trade-offs and the variation of seed size with plant resources status. Am Nat 140:287–304.
Vilela A, Cariaga R, Gonzalez-Paleao L, Ravetta D (2008) Trade-offs between reproductive allocation and storage in species of Oenothera L. (Onagraceae) native to Argentina. Acta Oecol 33:85–92.
Villalobos FJ, Testi L, Hidalgo J, Pastor M, Orgaz F (2006) Modelling potential growth and yield of olive (Olea europea L.) canopies. Eur J Agron 24:296–303.
Vose JM, Sullivan NH, Clinton BD, Bolstad PV (1995) Vertical leaf area distribution, light transmittance, and application of the Beer–Lambert law in four mature hardwood stands in the southern Appalachians. Can J For Res 25:1036–1043.
Warton DI, Wright IJ, Falster DS, Westorby M (2006) Bivariate line-fit-ting methods for allometry. Biol Rev 81:259–291.
White JW, Hoogenboom G, Hunt LA (2005) A structural procedure for assessing how crop models respond to temperature. Agron J 97:426–439.
Yang J, Zhang J, Wang Z, Zhu Q, Wang W (2001) Remobilization of carbon reserves in response to water deficit during filling of rice. Field Crops Res 71:47–55.
Zhang H, Turner N, Poole M (2010) Source–sink balance and manipu-lating sink–source relations of wheat indicate that the yield potential of wheat is sink-limited in high-rainfall zones RID F-1850–2011. Crop Pasture Sci 61:852–861.
