Are stored carbohydrates necessary for seed production in temperate deciduous trees?


Correspondence author. E-mail:


  1. Many tree species undergo large fluctuations from year to year in seed production, a phenomenon known as masting. The resource budget model, based on the assumption that abundant seeding in a masting year depends on the abundance of resources stored over several years, is a key hypothesis in explaining the mechanism of masting. But do masting species really need such long-term storage to produce a large seed crop?
  2. To test this hypothesis, we studied the relationship between the carbon accumulation period for seed production, as estimated by radiocarbon (14C) analyses, and the coefficient of variation of annual seed production in 10 canopy tree species in a temperate deciduous forest. These species differ widely in their reproductive intervals.
  3. In all the species studied, the accumulation period was < 1.4 years before seed maturation. Moreover, without taking species or reproductive intervals into account, there was no significant correlation between the carbon accumulation period and the fluctuation of annual seed production; both remained at an even level.
  4. Synthesis. Our results suggest that temperate canopy trees used photosynthates produced in the current and/or the previous year for seed production, regardless of reproductive intervals. It might therefore be necessary to reconsider the importance of stored carbohydrate resources for masting.


The reproductive tactics of plants vary widely among taxa. Some species exhibit a more or less constant flowering effort with time, but others vary (Satake & Biørnstad 2008). Many forest tree species reproduce in synchrony, but with a large interannual variation in the population; this phenomenon is known as masting or mast seeding (Kelly 1994). Masting is a common phenomenon that has been reported in diverse taxonomic groups and on most continents (Janzen 1971; Sork, Bramble & Sexton 1993; Koenig et al. 1994; Herrera et al. 1998; Sakai et al. 1999; Shibata et al. 2002). The masting phenomenon has many ecological consequences; masting strongly influences the populations of animals that feed on seeds as their main food, as well as the recruitment of the plant populations themselves (Janzen 1971; Curran & Leighton 2000; Oka et al. 2004). Resource allocation to reproduction may also have effects on plant growth and on interactions between plants and their herbivores (Kaitaniemi, Neuvonen & Nyyssönen 1999; Selås et al. 2001; but see Klemola et al. 2003). Furthermore, the pollen of masting trees may cause allergic diseases, which today are a major component of the increasing cost of health care in industrialized countries (Ranta et al. 2005).

The resource level of a plant is regarded as an important proximate factor in masting, together with weather cues (Kelly 1994; Kelly & Sork 2002). Because trees require abundant resources to generate large seed crops, they must wait several years until the amount of resources exceeds a threshold level. This model of masting is known as a resource budget model (Isagi et al. 1997; Satake & Iwasa 2000, 2002a,b). In particular, most resource-driven models of masting assume that carbohydrates are the main limiting internal resource for seed production (Isagi et al. 1997; Satake & Iwasa 2000). Although the physiological mechanism behind masting is not fully understood, it is generally believed to be related to temporal variations in the individual resource budget and the associated cost of reproduction, which is strongly influenced by climatic conditions (Hilton & Packham 2003; Richardson et al. 2005). The question arises; do masting trees really need to utilize the resources accumulated over a long time? Recent studies have found various results: for a masting event, trees consume stored carbohydrate or current photosynthates or both (Newell 1991; Obeso 1998; Miyazaki et al. 2002; Hasegawa et al. 2003; Hoch 2005; Ichie et al. 2005b; Hoch & Keel 2006; Yasumura, Hikosaka & Hirose 2006; Han, Kabeya & Hoch 2011). Those studies were all restricted to a limited number of species and/or individuals, because of the difficulty of accessing tree crowns. No comprehensive analyses have been published of the relationship between masting and stored resources.

Radiocarbon (14C) techniques are capable of settling whether species that require long intervals for masting must wait several years until the amount of resources exceeds a threshold level. The 14C content of atmospheric CO2 increased dramatically in the 1950s and early 1960s due to nuclear weapons testing (Levin & Hesshaimer 2000). Since the Nuclear Test Ban Treaty of 1963, the ‘bomb’-14C component of atmospheric CO2 has fallen exponentially, by exchange with oceans and terrestrial biota (Levin & Hesshaimer 2000; Hua & Barbetti 2004; Fig. 1). Because the 14C content of photosynthates in plant leaves is the same as that of atmospheric CO2 at any given time (Burchuladze et al. 1989; Druffel & Griffin 1995; Hua et al. 2000), the falling trend in 14CO2 allows us to estimate the time at which the carbohydrate resources were fixed by trees (Tayasu et al. 2002). It follows that we can estimate the age of the carbon contained in the seed by measuring its 14C content and comparing it with the known trend of 14CO2. Recent yearly changes have been comparable with analytical errors of 14C measurement made by accelerator mass spectrometry (AMS), so that we can estimate the age of carbon to high accuracy (Tayasu et al. 2002; Hyodo, Tayasu & Wada 2006; Hyodo et al. 2008; Tayasu & Hyodo 2010). If species with long reproductive intervals have to store carbon resources for a long time, then the carbon in productive seeds should be considerably older than the reproductive year, whereas species with short reproductive intervals could use mainly current carbon resources deriving from photosynthesis.

Figure 1.

Δ14C values of early leaves of Betula grossa (open circle) collected in OFR, and an exponential curve estimated to be Δ14C values of background atmospheric CO2 at the study site (solid line). Atmospheric CO2 in the Northern Hemisphere is shown in a broken line (Hua & Barbetti 2004). Black circles are Δ14C values of studied seed in each species in collection years. Vertical bars indicate SD.

This study was designed to determine whether there is a correlation between the reproductive intervals, which vary between species, and the carbon accumulation period for seed production. To this end, we studied the 14C content of seeds in 10 deciduous broad-leaved species having different reproductive intervals in a temperate deciduous forest in Ogawa Forest Reserve, Japan. We estimated the period needed to store enough carbohydrate resources for seed production, from the relationship between the age of the carbon contained in the seed and the reproductive year in each species. We then examined the relationship between the accumulation period and reproductive intervals.

Materials and methods

Study site and sample collection

The study was conducted at a 6-ha permanent plot in Ogawa Forest Reserve, which is an old-growth deciduous forest in central Japan (37ºN, 140ºE, 620–670 m in elevation). The mean annual temperature and precipitation from 1986 to 1995 were 10.7°C and 1910 mm (Mizoguchi, Morisawa & Ohtani 2002). The dominant tree species are Quercus serrata, Fagus crenata and F. japonica. The canopy height is ca. 20 m, and tall trees reach 30 m (Table 1; Tanaka & Nakashizuka 1997; Aiba & Nakashizuka 2009). A detailed description of the forest composition is available in Masaki et al. (1992) and of the forest dynamics in Nakashizuka et al. (1992).

Table 1. The tree species studied and their maximum heights in Ogawa Forest Reserve (OFR), Japan
Species, by familyCodeMaximum height (m)a
  1. a

    Data from Aiba & Nakashizuka (2009).

 Acer mono ACM23.3
 Acer amoenum ACA18.9
 Carpinus cordata CRC14.6
 Carpinus japonica CRJ14.0
 Carpinus tschonoskii CRT20.9
 Castanea crenata CSC22.9
 Fagus crenata FGC24.6
 Fagus japonica FGJ24.0
 Quercus serrata QRS27.0
 Styrax obassia STO15.9

Ten of the principal canopy tree species were chosen for this study (Table 1). We looked only at canopy trees, because it is possible that lower trees absorbed not only atmospheric 14C but also 14C of older origin, from contamination stagnant at the forest floor arising originally from plant and soil respiration (Bazzaz & Williams 1991; Kumagai et al. 2006; Sun et al. 2007). Seeds of the studied species were collected by seed and litter traps having a mouth area of 0.5 m2 set at each point of a 7 × 7 m grid in a 1-ha subplot (100 m × 100 m) located at the centre of the 6-ha plot (221 traps in all) from 1987 to 1988; in a 1.2-ha subplot (100 × 120 m) (263 traps) from 1989 to 1991; and at every 10 × 10 m grid point in a 1.2-ha subplot from 1992 (143 traps) (Tanaka & Nakashizuka 2002). Seeds that had fallen into the traps were collected every two or four weeks and were identified and classified into sound seeds (attaining a mature size and having a sound embryo) and the rest (Shibata, Tanaka & Nakashizuka 1998). The mature seeds were air-dried and kept in the warehouse of the Forestry and Forest Products Research Institute, Tsukuba, Japan.

Fluctuation in annual seed production

To examine the intensity of the masting habit in each species, we used the CV of the number of sound seeds per unit area (m−2) falling into the traps in the plot from 1987 to 2000; this was taken as an index of the annual fluctuation of seed production in a population (Table 2; Shibata et al. 2002).

Table 2. Annual variations of the number of sound seeds (m−2) in the studied species in OFR
Species codea19871988198919901991199219931994199519961997199819992000MeanSDCV
  1. a

    See Table 1 for species codes.


Some species in this study, such as Q. serrata, continually produce a relatively constant number of seeds (Table 2; Shibata et al. 2002). In other species, including two Fagus and three Carpinus species, there is a large annual fluctuation in seed production (Table 2). The minimum (Q. serrata) and maximum (F. crenata) species differed by a factor of more than 4.5. In general, the intervals between mast seeding years in F. crenata are 5–7 years (Maeda 1988; Hiroki & Matsubara 1995; Suzuki, Osumi & Masaki 2005; Masaki et al. 2008).

Radioisotope analyses

We used the seeds, including the seed coat and the endosperm, collected from the traps set for the 10 species at the study site from 1990 to 1996 (Table 3). The studied seeds were collected in different years in each species, because there were different masting years in these species and because most of the collected seeds had been used in other studies (Tanaka 1995; Shibata, Tanaka & Nakashizuka 1998; Shibata et al. 2002). All seed samples were powdered to finer than 40 mesh, and about 2 mg of the resulting powder (to produce about 1 mg C) underwent combustion for 2 h in evacuated and sealed Vycor tubes with CuO, Cu and Ag wire at 850°C. After cooling, the tubes were cracked on a vacuum line, and the CO2 was cryogenically purified and then graphitized by using Fe catalysis for 6 h at 650°C (Kitagawa et al. 1993). For measurement of radiocarbon by AMS, the graphite samples were sent to Rafter Radiocarbon Laboratory, Institute of Geological and Nuclear Sciences, New Zealand. The radiocarbon value is shown as Δ14C (‰), which is the deviation in parts per thousand from the activity of nineteenth-century wood, and has been corrected for fractionation by using the stable C isotope ratio of the sample (Stuiver & Polach 1977). The Δ14C signatures were reported for each analysis together with an analytical error (‰), which was 2.3‰ on average (2.5‰ at the maximum). These values were more accurate than those in previous studies (5.5‰ on average in Hyodo, Tayasu & Wada 2006; Hyodo et al. 2008). Because the ‘bomb’-14C of atmospheric CO2 has exponentially decreased since the Nuclear Test Ban Treaty of 1963 (Levin & Kromer 1997), analytical accuracy can be greater when, after the treaty, the Δ14C value is higher and the year is closer to 1963. Therefore, it is suggested that we can estimate the age of carbon with greater accuracy.

Table 3. The number of seeds analysed, seed collection dates, Δ14C and average and total carbon accumulation periods in the species studied
Species N Seed collection date (SCD)Δ14C (‰)SCD-t seed (year)Total carbon accumulation period (year)a
  1. a

    Mean values of Δ14C and seed collection dates were used for calculation.

Acer mono 3November 1995112.1 ± 1.60.35 ± 0.231.08
Acer amoenum 3November 1995111.1 ± 0.50.20 ± 0.070.35
Carpinus cordata 3October 1990149.4 ± 1.90.04 ± 0.210.07
Carpinus japonica 3October 1990149.8 ± 2.40.09 ± 0.260.18
Carpinus tschonoskii 3October 1990152.2 ± 0.70.35 ± 0.081.06
Castanea crenata 3October 1995114.1 ± 0.40.53 ± 0.061.18
Fagus crenata 3October–November 1993127.6 ± 4.50.43 ± 0.621.13
Fagus japonica 3October–November 1993128.5 ± 2.10.56 ± 0.251.25
Quercus serrata 3November 1996107.8 ± 1.30.67 ± 0.201.38
Styrax obassia 3September 1993127.1 ± 0.60.30 ± 0.080.97

To determine the Δ14C of background atmospheric CO2 at the study site during the study periods, we also analysed the leaf litter of Betula grossa (Betulaceae), which is a deciduous and canopy species at the site. Litter was collected by using litter traps in 1992, 1994, 1998 and 2001. We did this because the Δ14C of atmospheric CO2 is known to vary with the latitude, region and surrounding environment (Pataki et al. 2010), mainly due to anthropogenic CO2 emission from fossil fuels (Levin & Kromer 1997; Pataki et al. 2010). Recent studies have shown that deciduous temperate tree species use current photosynthates mostly for leaf formation, because developing leaves become C-autonomous very soon after bud break (Keel & Schädel 2010; Landhäusser 2011). We can therefore use current leaves of deciduous tree species as an indicator of Δ14C of atmospheric CO2 at the study site. Betula grossa has two leaf types, early leaves and late leaves (Miyazawa & Kikuzawa 2004). Early leaves appear following spring budding, whereas late leaves appear successively and continuously after expansion of the early leaves (Jones & Harper 1987a,b; Wilson 1991; Miyazawa & Kikuzawa 2004). It is easy to separate early leaves from late leaves in this species, because the former are of significantly larger leaf size and LMA than the latter. Furthermore, the two leaf types fall at different times: early leaves do not abscise until near the end of the growing season, whereas late leaves begin to abscise around the middle of it (Miyazawa & Kikuzawa 2004). In this study, we used early leaves of B. grossa as an indicator of Δ14C of background atmospheric CO2 at the study site, as only this species met the conditions necessary for the indicator. These conditions include reaching the canopy height so as to prevent contamination of older 14C from plant and soil respiration, being pioneer species which have earlier autotrophic photosynthesis than late successional species (Lambers, Chapin & Pons 2008), and simple separation of leaves from collected litter.

We estimated the carbon accumulation period for seed production by the following method. First, Δ14C of atmospheric CO2 was taken to decrease exponentially (Eqn 1; Levin & Kromer 1997; Hyodo, Tayasu & Wada 2006):

display math(eqn1)

where t represents the dominical year. We assumed that the early leaves of B. grossa had been produced by using current photosynthates from May to June in the leafing year, because they started to develop from the beginning of May and matured in approximately 2 months (Miyazawa & Kikuzawa 2004). Next,

display math(eqn2)

where x1 and x2 are the start and end dates, respectively, for the supply of carbohydrate resources used to produce the early leaves, namely May 1 and July 1 in each year, and l is the period of carbon accumulation, namely 2 months. We then numerically estimated the values of a and b using the function nls in R software, version 2.15.1 (R Development Core Team 2012). It was found that = 2073.7 and = 16.56 (Fig. 1). We found that Δ14C of leaves of B. grossa was slightly higher than that of atmospheric CO2 recorded in the Northern Hemisphere (estimated Δ14C values of atmospheric CO2; Hua & Barbetti 2004) (Fig. 1).

Finally, mean and total carbon accumulation periods for seed production were calculated by using the following formula:

display math(eqn3)

where tseed was calculated by Eqn 1.

The total carbon accumulation period was roughly estimated by the following expression:

display math(eqn4)

where l is the carbon accumulation period and x is the last date of the growing season (i.e. the photosynthetic period for seed maturation) in the seed-producing year. If the estimated date x-l was during the growing season, the total carbon accumulation period was represented by l. But in contrast, if the estimated date x-l was earlier than the beginning of the growing season, the total carbon accumulation period was estimated by the following equation, which assumes the total carbon accumulation to cover over 2 years:

display math(eqn5)

where d1 is the date of the beginning of the growing season in the seed-producing year, d2 is that of its end in the previous year and l2 is the carbon accumulation period in the previous year. In this case, the total carbon accumulation period for seed maturation is given by calculation, x – (d2l2). If this estimated date was earlier than the beginning of growing season in the previous year, this equation was expanded into an accumulation period of 3 years or more.

We consider the growing season to be a photosynthetic production period, the duration of leaf emergence. At the study site, almost all deciduous tree species begin leaf development at the beginning of May and then shed leaves at the beginning of November. We therefore assumed d1 and d2 to be May 1 and October 31 of the focal year, respectively. When seed was collected before October 31, x was assumed to be the date of the seed collection, otherwise x was assumed to be October 31.

Some individuals of the tree species indicated lower values of Δ14C in their seed than those in the atmospheric CO2 of the seed collection date (i.e. negative estimated values of l), probably due to an analytical error. To eliminate this error, we used the mean values of Δ14C in seed (‰) and the seed collection dates for the calculation in each species. Microsoft Excel Goal Seek was used for estimating the values of l.


For all 10 species in this study, there was a difference of not more than 1.4 years in the total carbon accumulation periods, which means a period between the end of June in the previous year and the seed maturation time in the reproductive year (Table 3 and Fig. 2). The species in which the period was the longest, implying that it used the oldest carbohydrate resources for seed production during a single reproductive period, was in Q. serrata (1.38 years), and the shortest was in Carpinus cordata (0.07 year). Fagus crenata underwent relatively large intraspecific variations in the Δ14C value (‰) of the seed and the average carbon accumulation period, which means between the dates of Δ14C and seed production, whereas Castanea crenata, Acer amoenum, Carpinus tschonoskii and Styrax obassia exhibited only small intraspecific variations (Table 3). No significant difference was observed between the species (> 0.05) in the average carbon accumulation period.

Figure 2.

Relation between the carbon accumulation period and the CV of annual mature seed production in each species; r2 = 0.02, > 0.05.

Without reference to species or reproductive intervals, there was no significant correlation between the total carbon accumulation period and the fluctuation in annual seed production, both of which remained at an even level (Fig. 2). In Fagus crenata, with the largest CV among the species studied, the total carbon accumulation period was 1.13 years, whereas Q. serrata, with the smallest CV, had a period of 1.38 years (Tables 2 and 3).


Our results clearly show that temperate trees use mainly current photosynthates for seed production, regardless of reproductive intervals. The carbon accumulation period for seed production did not exceed 1.4 years, regardless of fluctuations in annual seed production (Table 3 and Fig. 2). This implies that temperate deciduous tree species use mostly the photosynthates produced in the current and/or the previous year. Even species that undergo masting at long intervals use the carbon resources produced relatively recently in their masting years. Some studies have also found carbon autonomy in fruit development at the single branch level in tree species that undergo masting (Hasegawa et al. 2003; Ichie et al. 2005b; Hoch & Keel 2006; Han, Kabeya & Hoch 2011). All of the species studied commonly bloom simultaneously with leaf development in spring and then ripen their seed over several months. Because deciduous species shed their leaves in winter, they are forced to rely on the carbohydrate resources stored in the tree to produce new leaves and flowers for a short period. Conversely, they may use mainly current photosynthates in reproductive shoots for their seed production, which is a relatively long-term event lasting several months after leaf development. Switching of the principal carbon source has been reported in masting dipterocarp species in the tropical rain forests of South-East Asia (Ichie et al. 2005b). The present study also suggests that assimilation of accumulated resources is less important for seed production than is commonly believed, regardless of species or degree of masting.

The size of stored carbohydrate resources is not necessarily a limiting factor in seed production in the bright canopy environment of a forest, because all the canopy species studied used mainly current photosynthates for their seed production. A recent study has found that the cost of reproduction varies among trees with different crops of acorns, and also between individuals even within the same crown, in the masting species Quercus lobata (Sánchez-Humanes, Sork & Espelta 2011). Trees and/or branches with large acorn crops were able to allocate resources to both growth and acorn production, whereas those with few acorn crops experienced reduced growth in branches with acorns (Sánchez-Humanes, Sork & Espelta 2011). These results suggest that, in good light conditions, trees and branches produce sufficient photosynthates for both growth and reproduction, whereas trees and branches under limited light conditions must reduce their growth, as there are insufficient photosynthates for both growth and reproduction; but in a masting year, they maximize their acorn production (Sánchez-Humanes, Sork & Espelta 2011). Moreover, Kamoi et al. (2008) report that shrubs grown in sunny conditions can cover their reproductive costs from their current photosynthates, whereas under low light conditions on the forest floor, stored carbohydrates appear to be critical resources for reproduction (Marquis, Newell & Villegas 1997). The main carbohydrate source for seed production of trees may change, depending on the light environment of species, trees and/or branches.

It is possible that the limiting factor in masting in temperate canopy species is not carbohydrate resources, but other mineral nutrients. Some researchers have asserted the importance of stored nitrogen for masting in Fagus crenata (Han et al. 2008) and of stored phosphorus for a mast seed crop in certain dipterocarp species (Ichie et al. 2005a; Ichie & Nakagawa in press). Many masting species are ectomycorrhizal (EM) (Turner, Brown & Newton 1993; Newbery, Songwe & Chuyong 1998), and this could be an important factor in acquiring nutrients such as nitrogen and phosphorus (Newbery 2005). It is possible that, prior to any masting event, the accumulation of nitrogen and/or phosphorus supplied by EM in the tree is important. Our results indicate that the resource budget model, which is based on the utilization and accumulation of carbohydrate resources for masting, should not be taken up with high confidence. To provide a better understanding of the importance of stored resources for masting, further study is needed of the relation between the stored resources of carbon and mineral nutrients and the developmental process of flower bud initiation (Miyazaki 2011). And also necessary would be studies of the relation between stored resources and climate cues that induce synchronous flowering over a wide area, in each species (Masaki et al. 2008) and among various taxa (Shibata et al. 2002), in relation to the phenomenon of masting.


We thank N.F. Ishikawa, T.F. Haraguchi, K. Takayama and Y. Kuzume for their help with the laboratory work. This research was partly supported by the Grant-in-Aid for Scientific Research (no. 23657022, 22370011, 23657021) from the Ministry of Education, Science and Culture, Japan, by the Joint Usage/Research Grant of Center for Ecological Research (2011jurc-cer02), Kyoto University, and by the Environment Research and Technology Development Fund (S-9) of the Ministry of the Environment, Japan. Two anonymous reviewers of this article kindly provided constructive comments for us.