Abstract

During the last two decades, an increase of the frequency of naturally regenerated beech seedlings has been reported. This may be due to an increased masting. In this investigation in southern Sweden, mast year frequency, mast crop size and the relationships between mast year and climatic variables were studied. Our analysis shows that while the average mast year interval was 4–6 years from the end of the seventeenth century up to the 1960s, the mean interval has decreased to 2.5 years during the most recent 30 years, and there have been two consecutive mast years on two occasions during this later period. Mast years have often followed years in which the temperature in July and September was higher than the 30-year mean. However, there were significant variations in the amounts of beechnuts produced between different mast years, and beechnut production increased with increasing site index. Climatic changes, especially increases in temperature, may have been responsible for the higher frequency of mast years, but increased atmospheric nitrogen deposition may also have been a contributory factor.

Introduction

In Sweden, natural regeneration is the most common way to regenerate beech (Fagus sylvatica) stands. According to this method, beechnuts from one mast year are used. soil scarification and shelter woods are important tools to control the outcome (Agestam et al., 2003). However, beech does not set fruit every year (Peters, 1997) and to promote strong regeneration, it is important to coordinate regeneration operations with rich mast years. During a mast year, mostly occurring over subcontinent- to continent-wide areas (Wachter, 1964; Perrins, 1966), the beechnut production should be large enough to establish a new stand of good quality; Henriksen (1988), for instance, claims that a minimum of 50 viable beechnuts per square metre is required. Small amounts of seed are usually also produced in non-mast years, but there are large differences in the amounts produced between mast years and non-mast years. Therefore, more accurate forecasting of mast years would greatly facilitate silvicultural planning in beech forestry.

Mast years are reported to occur irregularly at intervals of 3–10 years (Matthews, 1955; Maurer, 1964; Bjerregaard and Carbonnier, 1979; Eickhoff et al., 1995; Rydberg, 2001). Information about the size of the crop has often been noted on an arbitrary scale, and the numbers of beechnuts produced have rarely been recorded. Many authors, e.g. Lindquist (1931) and Jenni (1987), have compiled information about mast years from different sources and sometimes by hearsay. This explains the variation sometimes found between different sources for the same year and the same area.

Lindquist (1931) claimed that an individual beech tree cannot flower in two consecutive years and that the flowering therefore has an endogenous periodicity. Holmsgaard and Olsen (1960) provided evidence supporting this claim, indicating that few beechnuts are produced in the year following a mast year. A hypothesis explaining the apparent inability of beech trees to produce large numbers of nuts in the years following mast years was proposed by la Bastide and van Vredenburch (1970), who suggested that vigorous flowering and fructification draw so heavily on carbohydrate reserves that little or no initiation of flower buds can take place in a mast year. Chan and Cain (1967) extended this theoretical framework by proposing that flowering and seed production levels are hormonally regulated.

The general opinion among foresters in northern and central Europe, albeit not based on any rigorous analysis, is that it has been easier to regenerate beech during the last 20 years than previously. This could be due to reduction in the intervals between mast years and increases in mast crops due to increases in temperature (SMHI, 2006) and accumulated nitrogen deposition, as reported (inter alia) by Westling (2001).

Many authors have studied various factors that may trigger flowering in beech, and even though most of them agree on a high temperature during the summer months to be the most important for flower bud formation and seed production, the next year, precipitation, sunshine, drought or combinations of these factors are often mentioned. Hauch and Oppermann (1898–1902), Lindquist (1931), Matthews (1955), Matyas (1965), Henriksen (1988) and Schmidt (2006) all claim that the weather during the summer months must be warm and dry for the following year to be a mast year.

However, Büsgen (1916) found that the first indications of flower predispositions were detectable by the end of May, implying that flower induction takes place at the end of April or beginning of May in the year before the flowering (Röhrig et al., 1978), while Piovesan and Adams (2001) and Gruber (2003) propose that the weather conditions during the 2 years preceding a mast year influence flowering.

In addition to the weather in the preceding year or years, the amount of seed produced may also be affected by climatic events during the year of seed production. Holmsgaard and Olsen (1960), for instance, show that heavy rains in April of the present year have a negative influence. Spring frost events, hailstorms and strong winds may also destroy the flowers (Gruber, 2003).

Since 1989, seed fall has been monitored and mast years have been identified in four different experiments in southern Sweden. Data from these investigations have been compiled together with information on mast years reported in earlier periods, and used in the present study to address the following hypotheses:

  • 1 The mast year frequency has increased during the last three decades.

  • 2 Summer temperature and/or precipitation data can be used to forecast mast years.

  • 3 The amount of seeds produced varies between mast years and is affected by site fertility.

Material and methods

Quantitative mast yield records

Henriksen (1988) claims that a minimum amount of seed to get a new stand of good quality is 50 seeds m−2 (equal to 500 000 seeds ha−1). This threshold is used as a definition of a mast year in this paper. This should also take place over a larger geographical area (Wachter, 1964; Perrins, 1966).

In this study, the mast crop in beech stands was measured in four different regeneration experiments (Figure 1). In all experiments, two similar types of seed traps were used, one constructed from stiff wire, duct tape and a plastic grain bag (Hughes et al., 1987), while the other was made of a nylon basket on an aluminium ring nailed to two wooden poles. The area of each seed trap was calculated by cross-measuring its diameter. The seed traps were set out at the end of August or beginning of September and cleared at the end of December. Seed data from damaged traps were not used in subsequent calculations. All seeds, both filled and empty, were counted.

Figure 1.

Sites of regeneration experiments where the beechnut falls were measured: 1–12, the liming experiment; 9, Skarhult; 12, Ryssberget and 13, Biskopstorp. Climate stations: H, Hörby; K, Kristianstad and L, Lund. Seed orchards: So A, Albjershus and So R, Ramsåsa.

Figure 1.

Sites of regeneration experiments where the beechnut falls were measured: 1–12, the liming experiment; 9, Skarhult; 12, Ryssberget and 13, Biskopstorp. Climate stations: H, Hörby; K, Kristianstad and L, Lund. Seed orchards: So A, Albjershus and So R, Ramsåsa.

Natural regeneration in beech was studied in an experiment in Skarhult (Figure 1, site 9), where seed fall was measured with 60 seed traps in autumn 1989 before regeneration cutting (Agestam et al., 2003).

Natural regeneration in beech without soil preparation was studied at Ryssberget (Figure 1, site 12) in stands in different phases of the regeneration. The seed fall was collected from seven stands in a total of 84 seed traps every year from 1992 to 2001, and after this from five stands in a total of 60 seed traps.

In an experiment where the effect of liming on natural regeneration in beech is being studied (Gemmel and Övergaard, 1995), the sizes of the seed crops have been registered every mast year since 1993 using a total of 88 seed traps. The experiment is done at 12 sites (Figure 1, sites 1–12) and the stand age varies from 75 to 160 years. The site index, i.e. dominant height at 100 years total age (Carbonnier, 1971), varies from F 20 to F 36. No effect of the liming on the seed fall between the treatments was found. Site 12 was established in December 1993 and no records of the seed fall that year are available. At site 11, all seed traps were damaged, probably by ungulates, in 1998, and there is no record of the seed fall that year.

In a natural regeneration experiment in a beech-dominated forest in the forest of Biskopstorp (Figure 1, site 13), seed fall was registered every year from 1998 to 2004 using 13 seed traps.

The study also considers information from a quantitative study during the years 1974–1983 previously published by Simak (1993).

Earlier mast yield records

At the Department of Forest Improvement (currently Skogforsk), descriptive data related to seed fall have been collected from two seed beech orchards, Albjershus and Ramsåsa (Figure 1), every year from 1976 to 1993 and mast years occurred in the same years in both seed orchards (M. Werner, personal communication). The observations of mast years correspond with notes made by G. Almgren (personal communication), former head of the Forest County Board of Scania.

In addition to these records, observations of mast crops made by the state forest district personnel during the years 1895–1929 in the southern Swedish provinces of Scania and Halland (Lindquist 1931) and data presented by Tirén (1931–1934), Perrins (1966), Jenni (1987), Henriksén (1988) and Osbeck (1996) were also used in the analyses presented here (Table 1).

Table 1:

Mast years and mean intervals between them for different time periods and provinces in Sweden from 1687 to 2006

Province Halland Scania Southern Sweden 
Period 1687–1711 1753–1795 1895–1929 1895–1929 1926–1964 1974–2006 
Mast years 1687 1753 1897 1897 1926 1974 1995 
 1690 1758 1906 1904 1933 1976 6, 7 1998 
 1698 1760 1909 1909 1944 1983 6, 7 2000 
 1702 1765 1915 1915 1954 4, 5 1986 2002 
 1707 1767 1918 1918 1956 4, 5 1987 2004 
 1711 1776 1926 1921 1960 4, 5 1989 7, 8 2006 
   1789   1924 1964 1992   
   1795   1926   1993   
Mean interval 4.8 abc 6 ab 5.8 b 4.1 bc 5.7 ab 2.5 c 
P-value 0.0932 0.0396 0.0343 0.2452 0.0187 – 
Province Halland Scania Southern Sweden 
Period 1687–1711 1753–1795 1895–1929 1895–1929 1926–1964 1974–2006 
Mast years 1687 1753 1897 1897 1926 1974 1995 
 1690 1758 1906 1904 1933 1976 6, 7 1998 
 1698 1760 1909 1909 1944 1983 6, 7 2000 
 1702 1765 1915 1915 1954 4, 5 1986 2002 
 1707 1767 1918 1918 1956 4, 5 1987 2004 
 1711 1776 1926 1921 1960 4, 5 1989 7, 8 2006 
   1789   1924 1964 1992   
   1795   1926   1993   
Mean interval 4.8 abc 6 ab 5.8 b 4.1 bc 5.7 ab 2.5 c 
P-value 0.0932 0.0396 0.0343 0.2452 0.0187 – 

Mean intervals with the same letter are not significantly different. The P-value indicates the probability for the mean interval of the period 1974–2006 to be separated from the other time periods. The numbers after the mast years refer to the following sources: 1, Osbeck (1996); 2, Lindquist (1931); 3, Tirén (1931–1934); 4, Jenni (1987); 5, Perrins (1966); 6, Simak (1993); 7, M. Werner (personal communication) and 8, experiments in the present study.

Factors inducing flower bud set

Climate data registered by the Swedish Meteorological and Hydrological Institute at the Hörby meteorological station (Figure 1) during the period 1 August 1995–2006 were used in the analyses of correlations between seed fall and climatic variables. Data from this station were used since it is located in the centre of the range of the study sites examined here. Data from 1973 to August 1995 were missing. To obtain values for this period, a multiple regression function was constructed between existing data from Hörby and data from the meteorological stations at Lund and Kristianstad. This function was thereafter used to fill the gaps in the data series. The potential evapotranspiration was calculated using the Thornthwaite method (Dunne and Leopold, 1978).

The effects of the following climatic variables (for each month from May to September, except where otherwise stated) have been analysed:

  • Average temperature (mean of daily temperature, i.e. mean of three daily readings and maximum and minimum temperature),

  • duration of warmth (number of days with maximum temperature exceeding 20°C),

  • intensity of warmth (the sum of maximum degrees exceeding 20°C),

  • precipitation,

  • average precipitation for the months June–August and

  • the precipitation minus potential evapotranspiration in June and July 2 years and 1 year before each mast year.

For every year between 1973 and 2005, the mean temperature and precipitation values for each of the months from May to September were compared with the corresponding monthly mean values for the entire 33-year period (Table 2). Higher temperature and lower precipitation values than the long-term means are indicated with a ‘1’ in the rows for the following years in Table 3.

Table 2:

Average values of investigated climate variables during the period 1973–2005

 Temperature, °C Monthly precipitation, mm 
 Mean Duration Intensity Individual month June–August 
May 10.9 5.5 12 45  
June 14.2 11 35 65  
July 16.4 18 65 71 65 
August 16.1 16.4 58 60  
September 12.2 3.2 67  
 Temperature, °C Monthly precipitation, mm 
 Mean Duration Intensity Individual month June–August 
May 10.9 5.5 12 45  
June 14.2 11 35 65  
July 16.4 18 65 71 65 
August 16.1 16.4 58 60  
September 12.2 3.2 67  

Monthly mean temperature, a mean value of the daily mean (a mean of three readings and the maximum and minimum temperatures) for the month; duration of warmth, number of days in each month with a maximum temperature exceeding 20°C; intensity of warmth, the sum of maximum degrees in the month exceeding 20°C.

Table 3:

Relationships between climatic variables and masting during the period 1974–2005

 Temperature Precipitation 
 Average (/month) Duration (/month) Intensity (/month) Average (/month) 
Year/Month 6 7 8 
1974*            
1975                  
1976*      
1977        
1978              
1979           
1980            
1981               
1982               
1983*           
1984          
1985                 
1986*                
1987*           
1988                     
1989*           
1990       
1991           
1992*             
1993*      
1994                 
1995*            
1996        
1997                
1998*     
1999                 
2000*          
2001                
2002*         
2003       
2004*    
2005             
2006*        
MY 50 57 86 71 71 43 64 86 43 50 36 64 79 57 43 57 71 79 29 50 57 
Not MY 47 47 26 42 47 42 47 26 47 16 37 42 26 37 21 58 53 42 63 47 53 
 Temperature Precipitation 
 Average (/month) Duration (/month) Intensity (/month) Average (/month) 
Year/Month 6 7 8 
1974*            
1975                  
1976*      
1977        
1978              
1979           
1980            
1981               
1982               
1983*           
1984          
1985                 
1986*                
1987*           
1988                     
1989*           
1990       
1991           
1992*             
1993*      
1994                 
1995*            
1996        
1997                
1998*     
1999                 
2000*          
2001                
2002*         
2003       
2004*    
2005             
2006*        
MY 50 57 86 71 71 43 64 86 43 50 36 64 79 57 43 57 71 79 29 50 57 
Not MY 47 47 26 42 47 42 47 26 47 16 37 42 26 37 21 58 53 42 63 47 53 

The 14 mast years during this period are indicated by an *. Each number one (1) indicates years when the mean temperature was higher, or the precipitation lower, in the preceding year than the corresponding mean for the entire 33-year period (see Material and methods). The numbers of times (in percentages) these criteria were met and the following year was a mast year (MY) and a non-mast year are shown at the bottom of the table.

Calculations

The mean mast year interval for each period considered was calculated and statistical analyses were performed using the Wilcoxon rank test (Wonnacott and Wonnacott, 1985) and pair-wise comparisons by the general linear model (GLM) procedure as implemented in MINITAB (Minitab Inc., State College, PA, USA). Tukey's test was used to identify differences in mean mast year frequencies between the time periods that were significant at the 5 per cent probability level.

To detect and evaluate possible correlations between weather conditions and mast years, the Wilcoxon rank-sum test was used to test the significance of differences in the variables' mean temperature, duration and intensity of warmth, and precipitation in each of the months from May to September from 1973 to 2005 between mast and non-mast years and the years preceding them.

To detect differences in mast crops between the mast years in the liming experiment, statistical analyses were applied using the GLM procedure in SAS version 9.1. The size of the mast crop at each of the sites and the mast years were the dependent variables, while the site index of each site was treated as a covariate.

In order to detect significant differences in mast crops between the period investigated here and the period investigated by Simak, the Wilcoxon rank test (Wonnacott and Wonnacott, 1985) was applied and a pair-wise comparison was made by the GLM procedure (MINITAB, Minitab Inc.). Tukey's test was used to assess whether the differences in seed production were significant at the 5 per cent probability level.

Results

Mast years occurred with mean intervals of ∼5 years from the end of the seventeenth century up to the last 33 years, when mast years became more frequent with a mean interval of 2.5 years. This period is significantly different in this respect from the previous period of the twentieth century (P = 0.0187). Furthermore, although there was a 7-year interval between two successive mast years in the latter period (in 1976 and 1983), the intervals have never been longer than 3 years since 1983 and there were two consecutive mast years on two occasions, 1986–1987 and 1992–1993 (Tables 1).

The occurrence of mast years was strongly related to the weather in July, especially to the temperature variables (Table 3). Significant differences in four July and one September weather variables between years preceding mast years and non-mast years were found: mean temperature (P = 0.0012), duration of warmth (P = 0.0001), intensity of warmth (P = 0.0009) and precipitation (P = 0.0138), all in July, and mean temperature in September (P = 0.0357). Twelve of 14 mast years were preceded by a year with higher values than mean for mean temperature and duration of warmth in July, and precipitation was lower than mean levels in July in years preceding 11 of the 14 mast years. The year that was followed by a mast year, if the year was not a mast year in itself, had in July an average temperature of at least 15.8°C and 16 days or more with a maximum temperature exceeding 20°C. Neither temperature nor precipitation during the months May, June or August were significantly correlated to the occurrence of masting in the following year. The occurrence of masting in a given year reduced the possibilities of it occurring in the following year, even if the weather conditions were conductive for masting. The 5 years in which the average July temperature was higher than the long-term mean but were not followed by mast years were all mast years. The difference between precipitation and potential evapotranspiration was less than the 33-year mean for seven of the 14 mast years.

In the liming experiment, the average mast crop varied from 1.1 to 10.1 million seeds ha−1 (in 2004 and 1995, respectively; Table 4), and there were significant differences in seed crops between many of the mast years. The sizes of the mast crops of Skarhult, Ryssberget and Biskopstorp are shown in Table 5, where also the amount of seed falling in non-mast years can be seen. Site index had a positive influence on the seed production (P = 0.005), which increased by ∼160 000 beechnuts ha−1 for every site index metre (Figure 2). No significant differences in mast crop sizes were found between those reported by Simak (1993) and those in the liming experiment reported here, but the mean mast crop (3.7 million seeds ha−1 per mast year) recorded by Simak (1993) was lower than the corresponding mean in the liming experiment (5.3 million seeds ha−1).

Table 4:

Size of the crops in mast years (millions of beechnuts ha−1) in different years in the liming experiment

 Sites  
 Liming experiment, site number  
Year 10 11 12 Mean 
1993 1.7 1.0 2.5 2.4 2.1 2.1 1.0 1.9 2.5 1.8 1.1 – 1.8 
1995 10.5 11.2 2.8 12.5 11.1 9.2 13.9 11.1 10.0 12.6 11.1 5.4 10.1 
1998 2.5 7.7 4.9 7.7 5.3 4.7 6.5 4.7 4.2 6.0 – 1.5 5.1 cd 
2000 3.3 3.0 1.3 2.0 5.4 3.3 4.8 3.4 3.5 7.2 3.0 0.8 3.4 ce 
2002 6.3 9.8 4.6 5.9 5.3 6.3 8.5 7.2 7.2 8.8 12.9 2.8 7.1 bd 
2004 0.3 0.5 0.4 0.2 1.8 2.4 0.8 0.7 1.4 2.7 1.4 0.4 1.1 
2006 8.0 12.2 8.5 10.9 8.7 8.0 6.8 6.4 3.6 10.2 12.6 5.4 8.4 ab 
Mean 4.6 6.5 3.6 5.9 5.7 5.1 6.0 5.1 4.6 7.1 7.0 2.7 5.3  
Site index F 28 F 22 F 20 F 24 F 26 F 24 F 27 F 30 F 26 F 30 F 36 F 26   
 Sites  
 Liming experiment, site number  
Year 10 11 12 Mean 
1993 1.7 1.0 2.5 2.4 2.1 2.1 1.0 1.9 2.5 1.8 1.1 – 1.8 
1995 10.5 11.2 2.8 12.5 11.1 9.2 13.9 11.1 10.0 12.6 11.1 5.4 10.1 
1998 2.5 7.7 4.9 7.7 5.3 4.7 6.5 4.7 4.2 6.0 – 1.5 5.1 cd 
2000 3.3 3.0 1.3 2.0 5.4 3.3 4.8 3.4 3.5 7.2 3.0 0.8 3.4 ce 
2002 6.3 9.8 4.6 5.9 5.3 6.3 8.5 7.2 7.2 8.8 12.9 2.8 7.1 bd 
2004 0.3 0.5 0.4 0.2 1.8 2.4 0.8 0.7 1.4 2.7 1.4 0.4 1.1 
2006 8.0 12.2 8.5 10.9 8.7 8.0 6.8 6.4 3.6 10.2 12.6 5.4 8.4 ab 
Mean 4.6 6.5 3.6 5.9 5.7 5.1 6.0 5.1 4.6 7.1 7.0 2.7 5.3  
Site index F 28 F 22 F 20 F 24 F 26 F 24 F 27 F 30 F 26 F 30 F 36 F 26   

Means with the same letter are not significantly different.

Table 5:

Size of the crops in mast years and non-mast years in Skarhult, Ryssberget and Biskopstorp (millions of beechnuts ha−1)

 Site 
Year Skarhult Ryssberget Biskopstorp 
1989 4.0 – – 
1990 – – – 
1991 – – – 
1992 – 5.273 – 
1993 – 1.500 – 
1994 – 0.004 – 
1995 – 7.766 – 
1996 – 0.000 – 
1997 – 0.000 – 
1998 – 0.933 1.925 
1999 – 0.002 0.000 
2000 – 2.717 0.490 
2001 – 0.001 0.000 
2002 – 5.677 5.405 
2003 – 0.000 0.000 
2004 – 0.896 0.043 
2005 – 0.002 – 
2006 – 7.435 – 
 Site 
Year Skarhult Ryssberget Biskopstorp 
1989 4.0 – – 
1990 – – – 
1991 – – – 
1992 – 5.273 – 
1993 – 1.500 – 
1994 – 0.004 – 
1995 – 7.766 – 
1996 – 0.000 – 
1997 – 0.000 – 
1998 – 0.933 1.925 
1999 – 0.002 0.000 
2000 – 2.717 0.490 
2001 – 0.001 0.000 
2002 – 5.677 5.405 
2003 – 0.000 0.000 
2004 – 0.896 0.043 
2005 – 0.002 – 
2006 – 7.435 – 

The values of Ryssberget are a mean of seven stands up to 2001, and after this a mean of five stands.

Figure 2.

The recorded seed fall during mast years in relation to site index of the study sites. The lines indicate the least-square mean estimates of the models from the statistical analysis where site index was treated as a covariate.

Figure 2.

The recorded seed fall during mast years in relation to site index of the study sites. The lines indicate the least-square mean estimates of the models from the statistical analysis where site index was treated as a covariate.

In the period, 1895–1929, investigated by Lindquist (1931), there were eight mast years in Scania and six in Halland (three of which did not coincide with mast years in Scania). In the present study, the seed crop during mast years was substantial at all sites in both regions.

Discussion

Our analyses show that the mean frequency of mast years has been higher during the last 33 years in southern Sweden than in earlier periods. The validity of these apparent differences in mast year intervals between the last three decades and previous periods may be questioned because of the varying recording methods used in the earlier periods. However, analyses of studies from other regions show the same trend, although seed fall was not quantified in them. The interval has decreased from 5 to 4.3 years in Denmark in the last 100 years (Hauch and Oppermann, 1898–1902; Jenni, 1987; Holmsgaard and Bang, 1990; H.G. Knudsen, personal communication), from 5.8 years between 1953 and 1976 to 3.2 years between 1990 and 2006 in England (Perrins, 1966; Gurnell, 1993; Packham and Hilton, 2002; Hilton and Packham, 2003; M. Broadmeadow, personal communication), from 3.8 years between 1940 and 1970 to 2.6 years between 1974 and 1995 in the Netherlands (la Bastide and van Vredenburch, 1970; Hilton and Packham, 2003) and from 8 years between 1869 and 1909 to 2.8 years between 1987 and 2004 in northern Germany (Hase, 1985; Jenni, 1987; Bartsch et al., 1993; Lange, 1995; Schmidt, 2006).

Possible explanations for the increased frequencies of mast years and mast yields may be found in changes in weather conditions and nitrogen deposition. The mean annual temperature in Sweden has increased by 0.5°C in the last 20 years (SMHI, 2006) and Schmidt (2006) found seed production in mast years to be higher if they were preceded by a warm summer than if they were preceded by a cooler summer. In addition, Nemec (1956) reported that flowering and masting are promoted by fertilization at sites on nutrient-poor soils, while Le Tacon and Oswald (1977) found that fertilization of a beech stand in the Massif Centrale, France, increased seed production threefold. The amount of nitrogen in the stem parts of the tree decreases after a mast year (Hartig, 1889). This may be compensated to some extent by atmospheric nitrogen deposition, which has resulted in accumulated nitrogen deposits of ∼300–450 kg ha−1 in the soils in southern Sweden since the 1950s according to calculations by Westling (2001).

The analyses show that the duration and intensity of warmth, as well as mean temperature and precipitation, all for July, in Sweden have been good predictors of a masting in the following year. These factors are naturally strongly correlated. Also the mean temperature of September seems to be of importance. The results are, with small variations, consistent with findings of many other studies, but may perhaps contribute to the subject of what factors are of importance, especially since we, during the investigated period, seem to have met a change in climate. Hauch and Oppermann (1898–1902) find high temperature in July, and partly June, August and September, to be essential for flower bud formation. Lindquist (1931) claimed the July temperature to be the most important factor, finding intensity of warmth being more important than duration, while precipitation is not mentioned in his work. Matthews (1955) finds significant correlations between the size of the mast and July temperature and sunshine. Matyas (1965) claims that a dry period with at least 10 h of sunshine daily is needed for fruit setting, while Henriksén (1988) argues a high temperature and moderate precipitation during June and July to be important. Also Schmidt (2006) finds a sunny, warm and dry summer, especially during July, to favour the fruit setting the coming spring.

However, Büsgen (1916) found that the first indications of flower predispositions were detectable by the end of May, implying that flower induction takes place at the end of April or beginning of May in the year before the flowering (Röhrig et al., 1978), while Piovesan and Adams (2001) and Gruber (2003) propose that the weather conditions during the 2 years preceding a mast year influence flowering. The former report that during a summer with sufficient amounts of available water, the trees build up carbohydrate reserves and if the following summer has a dry period during June–July, the setting of flower buds is favoured. Energy for the next years' flowering and masting can then be taken from the carbohydrate reserves. The latter claims that global radiation, sunshine duration and temperature during the months June–October 2 years before the mast year have a strong influence on flower induction, and the same factors strongly influence flower formation during September and October in the year before the mast year. Kon et al. (2005) claim low temperatures in late April to mid-May to induce a mast year the coming year for Fagus crenata.

According to the literature, two consecutive mast years occur rarely, for instance such events have been recorded in northern Germany 1989–1990 (Bartsch et al., 1993), in the Netherlands 1986–1987 (Hilton and Packham, 2003), in England 1906–1907 (Watt, 1925) and in Denmark 1807–1808 (Hauch and Oppermann, 1898–1902). However, consecutive mast years occurred twice during the last 30 years in southern Sweden (i.e. 1986–1987 and 1992–1993). Extremely warm and dry weather seems to override the inhibitory effect of a masting in the preceding year on masting in the current year. The weather during the summer of the mast year of 1992 was close to optimal for inducing beech flower buds. Mean temperatures in May, June and July were several degrees higher than average. Moreover, the precipitation in May–September was low in 1992, but high for the same months in 1991, yielding a high precipitation–potential evapotranspiration ratio (cf. Piovesan and Adams, 2001). This very propitious weather may have promoted the occurrence of a new mast year in 1993. Likewise, the weather in the summer of the mast year of 1986 favoured induction of flower buds and facilitated masting in 1987.

Apart from the experiments presented in this paper, there have been very few records of measured mast quantities. Hence, it is difficult to compare the amount of beechnuts produced in recent decades and earlier periods. However, Simak (1993) measured the mast crop in three stands in southern Sweden from 1974 to 1983, finding a mean of 3.7 million seeds ha−1 per mast year, lower than the corresponding mean in the liming experiment (5.3 million seeds ha−1).

The allocation of nutrients and carbohydrates to flowering and beechnut production reduces diameter growth by up to 40 per cent in a mast year and to lesser degrees in the two following years (Holmsgaard, 1955), and the amount of nitrogen stored in wood and bark of the stem has been shown to decrease after a mast year (Hartig, 1889). Thus, if the seed production levels during mast years in the most recent decades were similar to (or higher than) those in earlier periods, the increased frequency of masting presumably had an inhibitory effect on the growth of the trees.

The positive influence between site index and seed production supports the theory of the importance of resources for seed production. This is also claimed by Nemec (1956) and Le Tacon and Oswald (1977) who increased the seed crops by fertilization. This implies that we have at present bigger seed crops compared to the first part of the twentieth century since Falkengren-Grerup and Eriksson (1990) report a two percent increased site index for beech between 1947 and 1988, interpreted as a fertilization effect caused by the increased nitrogen deposition.

Shorter mast year intervals, accompanied by increasing mast crops, may have several positive effects for beech forestry. Firstly, since most of the valuable timber is cut when conditions favour the regeneration, more frequent mast years result in more even timber flows, thereby reducing price fluctuations in the timber market and increasing the evenness of industrial supplies. Secondly, if an attempt to regenerate a stand naturally fails, the best option is to try again in the next mast year. The prospects for the second attempt will usually be less good than those for the first attempt since the previous stand, which would have provided shelter on the first occasion, has already been cut, thereby increasing the availability of light, water and nutrient resources for field vegetation species, e.g. wavy hair grass and raspberry, that could be severe competitors to new seedlings. With shorter mast year intervals, this competing vegetation will not be so severe, which should mitigate the adverse circumstances for a second regeneration attempt. Thirdly, more frequent mast years should make planning forest activities easier.

The most important factor for a forest manager to consider when attempting to forecast mast years and planning his/her silvicultural activities is that there are seldom two mast years in a row. Secondly, a mast year is most often preceded by a warm and dry July.

Conclusions

The first hypothesis that the mast year frequency has increased during the last three decades was supported by information compiled regarding mast years in southern Sweden.

The second hypothesis that summer temperature and/or precipitation data can be used to forecast mast years was confirmed by the finding that the temperature in July was a good predictor of mast years.

Seed fall varies between mast years and is positively influenced by the site index, supporting the third hypothesis.

Funding

Broadleaf program.

Conflict of Interest Statement

None declared.

Special thanks are also due to M. Werner for kindly providing data from, and knowledge about, seed orchards. G. Almgren most kindly shared his notes with us. We are also indebted to M. Broadmeadow and H. G. Knudsen for data from England and Denmark. Thanks also to Jan-Eric Englund for statistical advice, and to the two reviewers for valuable comments.

References

Agestam
E
Ekö
PM
Nilsson
U
Welander
NT
The effects of shelterwood density and site preparation on natural regeneration of Fagus sylvatica in southern Sweden
For. Ecol. Manage.
 , 
2003
, vol. 
176
 (pg. 
61
-
73
)
Bartsch
N
Röhrig
E
Hesse
R
Langfristige Sukzession in alten Buchenbeständen
Forst Holz
 , 
1993
, vol. 
48
 (pg. 
697
-
699
(in German)
Bjerregaard
J
Carbonnier
C
Beech silviculture
Sver. Skogsvårdförb. Tidskr.
 , 
1979
, vol. 
3
  
(in Swedish)
Büsgen
M
Blütenentwicklung und Zweigwachstum der Rotbuche
Z. Fors. Jagdwes.
 , 
1916
, vol. 
48
 (pg. 
289
-
306
(in German)
Carbonnier
C
Yield of beech in southern Sweden
Stud. For. Suec.
 , 
1971
, vol. 
91
 pg. 
88
  
(in Swedish with English summary)
Chan
BC
Cain
JC
The effect of seed formation on subsequent flowering in apple
J. Am. Soc. Hortic. Sci.
 , 
1967
, vol. 
91
 (pg. 
63
-
68
)
Dunne
T
Leopold
LB
Water in Environmental Planning
 , 
1978
New York
W.H. Freeman & Co
(pg. 
136
-
138
)
Eickhoff
K
Nilsson
SG
Almgren
G
Bjerregaard
J
Ederlöf
E
Fredriksson
G
, et al.  . 
The Beech, a Manual in Beech Forestry. Sydved AB
 , 
1995
 
(in Swedish)
Falkengren-Grerup
U
Eriksson
H
Changes in soil, vegetation and forest yield between 1947 and 1988 in beech and oak sites of southern Sweden
For. Ecol. Manage.
 , 
1990
, vol. 
38
 
1–2
(pg. 
37
-
53
)
Gemmel
P
Övergaard
R
Liming and natural regeneration of beech
1995
 
Report No. 9. The Southern Swedish Forest Research Centre, The Swedish University of Agricultural Sciences (in Swedish)
Gruber
F
2003
 
Control and forecasting of the fructification of European beech (Fagus sylvatica L.) by climate factors. Schriften aus der Forstlichen Fakultät der Universität Göttingen und der Niedersächsischen Forstlichen Versuchsanstalt, Band 136. J. D. Sauerländers Verlag Frankfurt am Main (in German with English summary)
Gurnell
J
Tree seed production and food conditions for rodents in an oak wood in southern England
Forestry
 , 
1993
, vol. 
66
 (pg. 
291
-
315
)
Hartig
R
Ueber den Einfluss der Samenproduktion auf Zuwachsgrösse und Reservestoffvorrathe der Bäume
Allg. Forst Jagdztg.
 , 
1889
, vol. 
65
 (pg. 
13
-
17
(in German)
Hase
W
Die Buche in Schleswig-Holstein
Forstarchiv
 , 
1985
, vol. 
56
 (pg. 
22
-
31
(in German)
Hauch
LA
Oppermann
A
Haandbog I skovbrug
 , 
1898
Det nordiske forlag København
pg. 
149
  
(in Danish)
Henriksen
HA
Skoven og dens dyrkning
 , 
1988
København
Dansk skovforening, Nyt Nordisk Forlag Arnold Busch
(pg. 
158
-
159
(in Danish)
Hilton
GM
Packham
JR
Variation in the masting of common beech (Fagus sylvatica L.) in northern Europe over two centuries (1800–2001)
Forestry
 , 
2003
, vol. 
76
 (pg. 
319
-
328
)
Holmsgaard
E
Tree-ring analyses of Danish forest trees
Forst. Forsøgsvæs. Dan.
 , 
1955
, vol. 
XXII
 (pg. 
76
-
96
(in Danish with English summary)
Holmsgaard
E
Bang
C
Blomstring, frugtsætning og tilvægsttab
Skoven
 , 
1990
, vol. 
3
  
(in Danish)
Holmsgaard
E
Olsen
HC
Experimental induction of flowering in beech
Forst. Forsøgsvœs. Dan.
 , 
1960
, vol. 
30
 (pg. 
1
-
18
)
Hughes
JW
Fahey
TJ
Browne
B
A better seed and litter trap
Can. J. For. Res.
 , 
1987
, vol. 
17
 (pg. 
1623
-
1624
)
Jenni
L
Mass concentrations of bramblings Fringilla montifringilla in Europe 1900–1983: their dependence upon beech mast and the effort of snow cover
Ornis Scand.
 , 
1987
, vol. 
18
 (pg. 
84
-
94
)
Kon
H
Noda
T
Terazawa
K
Koyama
H
Yasaka
M
Proximate factors causing mast seeding in Fagus crenata: the effects of resource level and weather cues
Can. J. Bot.
 , 
2005
, vol. 
83
 (pg. 
1402
-
1409
)
la Bastide
JGA
van Vredenburch
CLH
The influence of weather conditions on seed production of some forest trees in the Netherlands
Mededelingen Landbouwhogeschool Wageningen, Nederland
 , 
1970
, vol. 
102
 (pg. 
1
-
2
)
Lange
V
Die Fruktifikation der Buche in verschiedenen Höhenlagen des Harzes
1995
Göttingen, Germany
Dipl. Arb. Forstwiss. Fachber, University of Göttingen
 
(in German)
Le Tacon
F
Oswald
H
Influence de la fertilization minérale sur la fructification du hêtre (Fagus sylvatica)
Ann. Sci. For.
 , 
1977
, vol. 
34
 (pg. 
89
-
109
(in French)
Lindquist
B
The ecology of Scandinavian beechwoods
Sven. Skogsvårdsfören. Tidskr.
 , 
1931
, vol. 
29
 (pg. 
486
-
520
(English digest)
Matthews
JD
The influence of weather on the frequency of beech mast years in England
Forestry
 , 
1955
, vol. 
28
 (pg. 
107
-
116
)
Matyas
V
Some ecological factors affecting the periodicity of fruiting in oak and beech
Erdesz. Kutatas., Budapest
 , 
1965
, vol. 
61
 (pg. 
99
-
121
(in Hungarian with German summary)
Maurer
E
Buchen- und Eichensamenjahre in Unterfranken wahrend der letzten 100 Jahre
Allg. Forst.
 , 
1964
, vol. 
31
 (pg. 
469
-
470
(in German)
Nemec
A
Improving the seeding of beech by soil improvement
Vedecke Prace Vyskumneho Ustavu Lesniho Hospodarstvi vi CSR
 , 
1956
, vol. 
11
 (pg. 
5
-
25
(in Czech with English summary)
Osbeck
P
Djur och natur i södra Halland under 1700-talet
 , 
1996
Halmstad
Spektra
(pg. 
23
-
24
(in Swedish)
Packham
JR
Hilton
GM
Inter- and intrasite variation in the fruiting of common beech (Fagus sylvatica L.) in England over a twenty-two year period (1980–2001)
Arbor. J.
 , 
2002
, vol. 
26
 (pg. 
1
-
22
)
Perrins
CM
The effect of beech crops on great tit populations and movements
Br. Birds.
 , 
1966
, vol. 
59
 (pg. 
419
-
432
)
Peters
R
Beech Forests
 , 
1997
Dordrecht, The Netherlands
Kluwer Academic Publishers
(pg. 
39
-
40
)
Piovesan
G
Adams
JM
Masting behaviour in beech: linking reproduction and climate variation
Can. J. Bot.
 , 
2001
, vol. 
79
 (pg. 
1039
-
1047
)
Röhrig
E
Bartels
H
Gussone
HA
Ulrich
B
Investigations on natural regeneration in beech (Fagus sylvatica)
Forstwiss. Centralbl.
 , 
1978
, vol. 
97
 (pg. 
121
-
131
(in German with English summary)
Rydberg
D
Ädellövskogen
 , 
2001
(pg. 
17
-
18
(in Swedish)
Schmidt
W
Temporal variation in beech masting (Fagus sylvatica L.) in a limestone beech forest (1981–2004)
Allg. Forst Jagdztg
 , 
2006
, vol. 
177
 (pg. 
9
-
19
(in German with English summary)
Simak
M
Beech Mast in Forestry. Department of Silviculture, The Swedish University of Agricultural Sciences
 (pg. 
22
-
23
(in Swedish)
Swedish Meteorological and Hydrological Institute (SMHI)
Climatic indications and observations: temperature
2006
 
Tirén
L
Set of fruit of forest trees in the years 1930–1933
Statens skogsförsöksanstalt, Flygblad
 , 
1931
(pg. 
41
-
44
(in Swedish)
Wachter
H
The connexions between weather and Beech mast years
Forstarchiv
 , 
1964
, vol. 
35
 (pg. 
69
-
78
(in German)
Watt
AS
On the ecology of British beechwoods with special reference to their regeneration
J. Ecol.
 , 
1925
, vol. 
13
 (pg. 
28
-
29
)
Westling
O
Effects of Air Pollution Exceeding the Critical Limits in Sweden. Annual report of 2001. International and National Abatement Strategies for Transboundary Air Pollution
2001
(pg. 
14
-
15
(in Swedish)
Wonnacott
R
Wonnacott
T
Introductory Statistics
 , 
1985
New York
Wiley