Research Findings

Variability in Annual Acorn Production

All studies of acorn production to date have found considerable variability from year to year. For example, using the Hastings data, the mean annual number of acorns counted in 30 seconds per tree (transformed or not) has ranged over nearly two orders of magnitude or more, with coefficients of variation (CV) on the order of 84-112% (untransformed values) or 55-87% (transformed values) (see Table 9.1). Whether acorn crops ever fail entirely is probably unanswerable, but crops can clearly

Table 9.1

Variability in mean annual and overall mean individual acorn productivity in five species

Table 9.1

Variability in mean annual and overall mean individual acorn productivity in five species

Annual mean

Individual mean

Species individuals

Minimum

Maximum

Coefficient of variation

Minimum Maximum

Coefficient of variation

N

Acorns per 30 seconds (N30)

Q. lobata

0.69

75.69

93.4

0.00

71.84

83.7

85

Q. douglasii

0.93

70.29

108.8

0.00

65.05

91.7

56

Q. agrifolia

0.38

47.79

91.6

0.42

49.68

85.7

61

Q. chrysolepis

0.00

55.76

83.7

0.42

49.68

85.7

21

Q. kelloggii

0.00

39.29

112.0

0.47

39.58

63.6

18

Log-transformed acorns

per 30 seconds (LN30)

Q. lobata

0.20

3.84

56.6

0.00

3.89

52.3

85

Q. douglasii

0.39

3.69

60.2

0.00

3.56

56.7

56

Q. agrifolia

0.13

2.98

68.5

0.04

3.43

53.4

61

Q. chrysolepis

0.00

3.49

54.8

0.19

3.15

48.9

21

Q. kelloggii

0.00

3.36

87.2

0.18

2.54

35.7

18

Notes: Data from Hastings Reservation measured from 1980 through 1998 (19 years) as the mean and log-transformed mean number of acorns counted in 30 seconds. Only individuals with complete data are included.

Notes: Data from Hastings Reservation measured from 1980 through 1998 (19 years) as the mean and log-transformed mean number of acorns counted in 30 seconds. Only individuals with complete data are included.

range from being very poor (only a small proportion of trees containing any acorns whatsoever) to being extremely good (nearly all trees containing a heavy crop of acorns).

Equally dramatic are differences in productivity by individual oaks. For example, in our study at Hastings Reservation, some trees have produced acorns (usually a very good crop) in all or nearly all 19 years of the study, whereas other individuals have not produced one acorn in 19 years; mean variability across individuals again ranges over nearly two orders of magnitude or more, with CVs in the range of 64-92% (untransformed values) or 36-57% (transformed values).

Within the population, however, within-year acorn production is relatively synchronous despite differences among individual trees (Koenig, Mumme, et al. 1994). Thus, even in the best of years there are trees with relatively few acorns, but for those individuals a few acorns constitute a relatively large crop; in most years, such trees do not produce any acorns at all.

The distribution of the CVs in annual acorn production for 1-year and 2-year species from the complete oak data set are summarized in Figure 9.1. Variability among the untransformed data on masting behavior is generally similar to other published compilations, including 42 studies of plants summarized by Kelly (1994), a wide variety of woody species by Herrera et al. (1998), and various northern hemisphere conifers and broad-leafed genera (including oaks) by Koenig and Knops (2000). The majority of CVs fall between 41% and 160% using the untransformed data and almost uniformly below 80% using the log-transformed data. There was no statistical difference in CVs of 1-year and 2-year species using either the untransformed or log-transformed data (Mann-Whitney U-tests, both z < 1.0, P > 0.3), suggesting that there is no clear difference between these groups in the extent to which they exhibit masting behavior. Using the untransformed data, only 5 of the 40 populations yielded CVs greater than 160%, with the maximum of 225% being achieved in a 12-year study of Q. prinus in the Appalachians by Beck (1977) in which 8 of the 12 years were near or total failures. Such extreme variability is the exception rather than the rule.

For both the 1-year and 2-year species there were negative correlations between latitude and CV which were significant or nearly so for the log-transformed data (untransformed data: rs = -0.30, n = 21, P = 0.19 [1-year species], rs = -0.31, n = 19, P = 0.19 [2-year species]); log-transformed data: rs = -0.48, n = 21, P = 0.03 [1-year species]), rs = -0.43, n = 19, P = 0.07 [2-year species]). However, the statistical sig-

Figure 9.1. Distribution of the coefficients of variation in annual acorn production of oaks requiring one and two years to mature acorns; only data sets presenting interval or ratio-level data at least six years in length were included. (a) untransformed data (mean ± SD = 108.6 ± 49.4% [1-year species] and 110.3 ± 35.8% [2-year species]); (b) log-transformed data (mean ± SD = 37.8 ± 30.9% [1-year species] and 40.5 ± 23.2% [2-year species]).

Figure 9.1. Distribution of the coefficients of variation in annual acorn production of oaks requiring one and two years to mature acorns; only data sets presenting interval or ratio-level data at least six years in length were included. (a) untransformed data (mean ± SD = 108.6 ± 49.4% [1-year species] and 110.3 ± 35.8% [2-year species]); (b) log-transformed data (mean ± SD = 37.8 ± 30.9% [1-year species] and 40.5 ± 23.2% [2-year species]).

nificance of these latter relationships was lost in multiple regressions controlling for the number of years of data.

Distribution of Annual Acorn Crop Size

Prior analyses of the oak data set revealed that normality could be rejected for only one of the untransformed and none of the log-transformed sets of data using Kolmogorov-Smirnov one-sample tests (Koenig and Knops 2000). Thus, it is generally not possible to reject the null hypothesis that acorn crops are normally (or log-normally) distributed. Dividing the data sets (including those that are categorical) into thirds, approximately one-third (18 of 49) of the untransformed and one-fourth (13 of 49) of the log-transformed data sets are bimodal in that the frequency of years yielding intermediate acorn crops was less than the frequency of years with larger and smaller crops. Neither of these frequencies is significantly different from the number expected by chance alone (X2 tests, both P > 0.2). Using the untransformed data, there was no difference between 1-year and 2-year species in the frequency of bimodality (1-year: 9 of 26 [34.6%] data sets; 2-year: 9 of 23 [39.1%] data sets). Only 11 of the 49 (22.4%) data sets contained no years of intermediate acorn crops. These 11 data sets were relatively short (mean ± SD length = 8.1 ± 2.9 years) compared to 38 data sets with at least some intermediate years (13.1 ± 6.8 years; Mann-Whitney U-test, z = 2.7, P < 0.01). Thus, failure to record intermediate crops was likely an artifact of small sample size. Identical conclusions are reached using the log-transformed data.

An alternative way to investigate bimodality in these data is to combine the standardized data sets and then compare the total number of years of data falling into the middle third of values against those in the upper and lower thirds. If we do this with the untransformed oak data sets, restricting the analysis to sites with at least six years of information, the number of years in the lower, middle, and upper thirds of the distribution are 350, 122, and 115. Using the log-transformed data sets, comparable values are 148, 164, and 275 years. Thus, neither of these combined data sets is bimodal, even with the conservative definition of bimodality used here.

Using the Hastings data, bimodality in acorn production, again defined as fewer years in the middle third of the overall range of annual acorn production values, is significant for individuals of four of the five species (Q. lobata, Q. douglasii, Q. agrifolia, and Q. kelloggii; Koenig,

Mumme, et al. 1994). However, overlap in the "tails" of the two modes is considerable for all species.

Intermast Interval

The intermast interval is the average time between good acorn crops (mast years). Although it is commonly held that most populations of oaks exhibit some sort of significantly cyclic behavior in acorn production, leading to a regular intermast interval (e.g., Schopmeyer 1974), it is by no means certain that intermast intervals are of regular length, or indeed that they exist at all, except in a few populations where masting intervals are unambiguous, such as in the alternate-bearing population of Q. robur studied by Crawley and Long (1995). Such clear cycles in acorn production are generally not evident for oaks, because, in the absence of a bimodal distribution or some other discontinuity in the distribution of annual seed crops, it is not possible to objectively distinguish between mast and nonmast years (Kelly 1994, Herrera et al. 1998). An alternative way to detect intermast intervals or the presence of cycles of acorn production is to perform time-series analysis. We performed such analyses on the oak data set using only sites providing at least 15 years of continuous data—still a marginal sample size for time-series analyses. Nine 1-year species and five 2-year species met this criterion.

Of the 14 data sets, 4 (29%; two 1-year and two 2-year species) exhibited no significant periodicity as tested by a Kolmogorov-Smirnov one-sample test of the periodogram values against that expected from a uniform distribution; the remaining 10 populations exhibited significant periodicity. Peak spectral density of all three 2-year species falling into this latter category corresponded to periodicities of between 4.6 and 6.3 years (mean 5.7 ± 1.0 years), whereas peak spectral density of five of the seven 1-year species corresponded to relatively short periodicities of 2.0 to 2.4 years. However, two populations of Q. mongolica were significantly periodic and had peak spectral densities at long periods of 10 years. Consequently, there was no significant difference between the peak spectral densities of 1-year and 2-year species (Mann-Whitney U-test, z = 1.7, P = 0.09). Excluding these two Q. mongolica populations, 1-year species cycled at significantly shorter periodicities than 2-year species (1-year: 2.1 ± 0.1 yrs, 2-year: 5.7 ± 1.0 yrs; Mann-Whitney U-test, z = 2.9, P < 0.005). Examples of spectral density plots for a typical 1-year species (Q. alba) and 2-year species (Q. velutina) are presented in Figure 9.2.

These results support the hypothesis that most populations of oaks ex-

Figure 9.2. Representative spectral densities of Q. alba (1-year species) and Q. velutina (2-year species) based on 32 years of data (1949-1980) from Missouri. (Reported by Christisen and Kearby [1984].)

hibit some sort of significantly cyclic behavior in acorn production. However, the fact that peak spectral densities generally were at fractional (noninteger) values indicates that cycles are often of irregular length rather than occurring at the same interval of time. They also suggest that periodicity in 1-year species is often, but not always, shorter than in 2-year species, in contrast with prior results (Sork 1993) based on the intermast intervals reported by Schopmeyer (1974).

Environmental Correlates of Acorn Production

Freezing of flowers in the spring can clearly result in poor acorn crops or apparent total failures the following autumn for 1-year species and in the subsequent year for 2-year species (Uhlig and Wilson 1952, Goodrum et al. 1971, Neilson and Wullstein 1980). Beyond this extreme, attempts to correlate annual acorn production with meterological correlates have met with mixed success (Table 9.2). The only notable generalization that emerges is the finding in five populations of 1-year species in the white oak subgenus (three populations of Q. alba

Table 9.2

Reported relationships between annual acorn production and weather

Table 9.2

Species

Duration of study (years)

Relationship

Reference

1-year species

Q. agrifolia

19

More rainfall year x - 2 and more

This study

rainfall year x = larger crop'

Q. alba

14

Warmer April temperatures =

Sharp and Sprague

larger crop

1967

Q. alba

8

Warmer spring temperatures =

Sork et al. 1993

larger crop

Q. alba

6

Maximum temperature and

Cecich 1997

increased days of hail during

pollination = smaller crop

Q. douglasii

19

Warmer April temperatures, less

This study

rainfall year x and warmer summer

temperatures = larger crop

Q. lobata

19

Same as for Q. douglasii

This study

Q. robur

15

No clear significant correlation found

Crawley and Long 1995

2-year species

Q. chrysolepis

19

More rainfall year x - 1 = larger crop

This study

Q. ilicifolia

Low humidity = larger crop

Wolgast and Stout 1977

(among individuals)

Q. kelloggii

19

None found

This study

Q. rubra

8

Dry summer year x - 2, warmer spring

Sork et al. 1993

temperatures, absence of spring

frost year x - 1 = larger crop

Q. velutina

8

Warmer spring temperatures =

Sork et al. 1993

larger crop

Q. velutina

6

Maximum temperature during

Cecich 1997

pollination = smaller crop

Note: Data from this study update Koenig et al. 1996. Multiple relationships are in decreasing order of significance.

s"Year x" refers to the year of the acorn crop; rainfall is from September of year x - 1 through August of year x.

and one each of Q. douglasii, Q. lobata) of a significant relationship between conditions during the spring flowering and fertilization period and acorn production the following autumn, the only caveat being that one of these (Cecich 1997) found a negative effect of warmer temperatures on the acorn crop rather than the positive effect found by the remaining studies. No clear significant relationship was found between any environmental variable tested and acorn production in Q. robur, while acorn production in the live oak Q. agrifolia is correlated with rainfall rather than spring weather conditions. For the white oaks, at least, this suggests that acorn crops of many species are proximately determined either by pollen dispersal or factors otherwise influencing fertilization in the spring.

For 2-year species results are more variable. No two species or studies demonstrated similar correlations, with the exception of a positive relationship once again between warm spring temperatures and the subsequent acorn crop in both Q. rubra and Q. velutina (see Table 9.2). Otherwise, each study yielded a slightly different set of correlations between weather and acorn production.

There has been considerably less investigation of the effects of weather conditions specifically during flowering on acorn production by individual trees. Among a subset of the Hastings oaks, the mean amount of solar radiation during the time period when 23 Q. douglasii flowered was significantly positively correlated with the size of the subsequent acorn crop after controlling for annual differences, suggesting that conditions favorable for pollination may influence not only annual acorn crop size but also variation among individuals. However, no such relationship was detected for either Q. lobata or Q. agrifolia (Koenig et al. 1996).

Geographic Synchrony in Acorn Production

Until recently, most studies of annual acorn production emphasized site-to-site variability rather than the possibility of geographic synchrony (Neilson and Wullstein 1980, Crawley and Long 1995). However, masting is a population phenomenon (Kelly 1994), and the size of the population involved determines, among other things, the extent to which predators may be affected by crop failures. Consequently, the extent of geographic synchrony in acorn production bears on the issue of the ultimate functional consequences of masting behavior.

We have approached this issue in several ways. First, comparisons of annual acorn crops for Q. lobata, Q. douglasii, and Q. agrifolia at three sites in central coastal California over a 10-year period indicate statistically significant synchrony within, and even between, these species over distances of nearly 300 kms (Koenig et al. 1996, Koenig, Knops, et al. 1999). Second, comparisons over a 16-year period between acorn production by Q. lobata, Q. douglasii, and Q. kelloggii at two sites in coastal California 320 km apart confirms a high degree of synchrony for the 1-

year Q. lobata and Q. douglasii; no significant correlation in the acorn crops between the two sites was detected for the 2-year Q. kelloggii (Koenig, McCullough, et al. 1999). Preliminary indications suggest that geographic synchrony is maintained by the same environmental factors that correlate with acorn productivity within a site. For example, April temperatures, which correlate strongly with annual acorn production of both Q. lobata and Q. douglasii at Hastings Reservation (see Table 9.2), exhibit high spatial autocorrelation between sites as well, more or less matching the degree of synchrony observed in the acorn crops of these two species in coastal California (Koenig et al. 1996).

We are currently expanding these studies by means of a survey of acorn production at 14 sites throughout California, encompassing 34 populations of six different species spanning a distance of nearly 1,000 km. Preliminary results based on the first five years of data (1994 through 1998) for five species, including three 1-year species (Q. lobata, Q. douglasii, and Q. agrifolia) and two 2-year species (Q. chrysolepis and Q. kelloggii), support the hypothesis that 1-year species generally show extensive geographic synchrony whereas 2-year species may not. Of the three 1-year species surveyed at multiple sites, mean (± SD) correlations between pairwise combinations of sites are all high (Q. lobata: 0.76 ±

Figure 9.3. Pairwise Pearson correlation coefficients of mean annual acorn production between sites of two species (Q. douglasii, a 1-year species, and Q. chrysolepis, a 2-year species) measured at nine (Q. douglasii) and six (Q. chrysolepis) sites in California surveyed between 1994 and 1998. (Koenig and Knops, unpublished data.)

Figure 9.3. Pairwise Pearson correlation coefficients of mean annual acorn production between sites of two species (Q. douglasii, a 1-year species, and Q. chrysolepis, a 2-year species) measured at nine (Q. douglasii) and six (Q. chrysolepis) sites in California surveyed between 1994 and 1998. (Koenig and Knops, unpublished data.)

0.17, n = 15; Q. douglasii: 0.80 ± 0.14, n = 36; Q. agrifolia: 0.71 ± 0.29, n = 15), whereas those for the two 2-year species are relatively low and more variable (Q. chrysolepis: 0.29 ± 0.44, n = 15; Q. kelloggii: 0.42 ± 0.38, n = 15). Correlation coefficients for one species in each group are presented in Figure 9.3. None of the species exhibits any significant decline in synchrony with increasing distance between sites, based on Mantel tests, in contrast to the pattern generally found in ecological phenomena (Koenig 1999). In fact, synchrony between sites for all species except Q. lobata actually increases with distance, albeit not significantly. However, this is most likely an artifact of the relatively short time span (5 years) over which we currently have data.

Finally, we can analyze the larger oak database for patterns of geographic synchrony, using recently developed methods of testing for spatial autocorrelation (Koenig and Knops 1998b). Results (Figure 9.4) in-

Figure 9.4. Mean correlations between annual acorn crops of 1-year and 2-year oaks depending on the distance separating sites. Correlations that are significantly greater than zero (based on randomization tests; Koenig and Knops 1998b) for both 1- and 2-year species up to sites 500 km apart are indicated with an asterisk.

Figure 9.4. Mean correlations between annual acorn crops of 1-year and 2-year oaks depending on the distance separating sites. Correlations that are significantly greater than zero (based on randomization tests; Koenig and Knops 1998b) for both 1- and 2-year species up to sites 500 km apart are indicated with an asterisk.

dicate significant synchrony among both 1 -year and 2-year species over distances of up to 500 km and generally support the hypothesis that synchrony is greater, at least between sites up to 500 km apart, for 1-year rather than 2-year species.

Synchrony within Communities of Oaks

At least four studies have reported within-site synchrony of oak species requiring the same number of years to mature acorns but asynchrony between species requiring different numbers of years to mature acorns (Mohler 1990 [three studies cited], Koenig, Mumme, et al. 1994). Updated results from the Hastings data are summarized in Table 9.3. All four comparisons between species requiring the same number of years to mature acorns are positive and all but one is statistically significant. in contrast, all six comparisons between species requiring different numbers of years to mature acorns are negative; none, however, is statistically different from zero.

Given that their phenologies do not overlap except with the final growth of acorns (Sork et al. 1993), this pattern of synchrony within types and asynchrony between types of oaks requiring different numbers of years to mature acorns is expected if acorn production by all species is, to at least some extent, influenced by the same suite of environmental factors. For example, a freeze that kills flowers in year x will result in a crop failure of 1 -year species in autumn of year x, whereas 2-year species will remain unaffected until year x + 1. Such asynchrony appears to have important consequences for several aspects of oak ecology, from reducing the frequency of total acorn crop failures within a site to facilitating co-occurrence of oak species (Mohler 1990).

Table 9.3

Spearman rank correlations between the annual acorn crops of five species

Table 9.3

Spearman rank correlations between the annual acorn crops of five species

Q. douglasii (1)

Q. agrifolia (1)

Q. chrysolepis (2)

Q. kelloggii (2)

Q. lobata (1)

0.85*

0.35

-0.26

-0.07

Q. douglasii (1)

NA

0.58*

-0.35

-0.27

Q. agrifolia (1)

NA

-0.01

-0.21

Q. chrysolepis (2)

NA

0.56*

Source: Data from Hastings Reservation, central coastal California, 1980-1998. Note: Number of years species requires to mature acorns is shown in parentheses. * = P < 0.01; other P > 0.05. N = 19 years.

Source: Data from Hastings Reservation, central coastal California, 1980-1998. Note: Number of years species requires to mature acorns is shown in parentheses. * = P < 0.01; other P > 0.05. N = 19 years.

The surprising feature of such interspecific asynchrony is that, by reducing the annual variability in total acorn abundance and the probability of total crop failures, it also facilitates the persistence of generalist species dependent on acorns. For example, acorn woodpeckers (Melanerpes formicivorus) are virtually restricted to sites containing at least two species of oaks, presumably because of the lower probability of crop failure and reduced annual variability as oak species diversity increases (Bock and Bock 1974, Koenig and Haydock 1999). Such facilitation is not predicted by the hypothesis of predator satiation, as discussed below.

Switching of Resources into Acorn Production

Two sorts of evidence have been used to test for evidence that oaks divert resources from elsewhere to produce large seed crops. The first is a negative autocorrelation between seed crops in successive years. Such negative autocorrelations have been documented within individual trees with a 1-year time lag for the 1-year species Q. lobata and Q. douglasii (Koenig, Mumme, et al. 1994) and Q. alba (Sork et al. 1993) and at longer (2- or 3-year) time lags for the 2-year species Q. rubra and Q. velutina (Sork et al. 1993) and Q. kelloggii (Koenig, Mumme, et al. 1994). Using the combined oak data set, Koenig and Knops (2000) documented a significant negative temporal autocorrelation with a 1-year time lag for 1-year species and both 2- and 3-year time lags for 2-year species.

There have unfortunately been few direct tests of the potential tradeoff between vegetative growth and reproduction in oaks. Crawley and Long (1995) indicate a failure to find such a tradeoff in Q. robur, while Sork and Bramble (1993) found suggestive evidence only for a tradeoff based on five years of data in Q. velutina and Q. rubra. Our own work at Hastings Reservation currently involves comparing acorn production between 1980 and 1994 with tree-ring growth for the deciduous species Q. lobata, Q. douglasii, and Q. kelloggii. Comparisons of annual acorn counts with standardized tree-ring widths within individuals yields no significant correlation between growth and reproduction during the current year for any of the species (see "within individuals" analyses in Table 9.4); in fact, if anything there is a positive correlation in Q. kelloggii. Among individuals in the population, there are significant inverse correlations for Q. lobata and Q. douglasii, but values are still very small. Lagged effects are similarly mixed. For Q. lobata, for example, comparison of acorn production with growth the prior year yields a significant

Table 9.4

Analyses of a tradeoff between growth and reproduction in three species of deciduous oaks

Table 9.4

Analyses of a tradeoff between growth and reproduction in three species of deciduous oaks

Q. lobata

Q. douglasii

Q. kelloggii

No time lag

Within individuals

40 vs. 39"

19 vs. 29

14 vs. 4**

Between individuals

-0.10*b

-0.11**

0.13

Growth in year x vs. acorns in year x + 1

Within individuals

42 vs. 37*

28 vs. 21

7 vs. 10

Between individuals

-0.09**

0.01

-0.00

Acorns in year x vs. growth in year x + 1

Within individuals

51 vs. 28**

25 vs. 24

4 vs. 14

Between individuals

0.06

0.08

-0.19**

Source: Data from Hastings Reservation (Knops and Koenig, unpublished data).

Tree-ring width vs. acorn production by tree i. Numbers are trees exhibiting positive and negative correlations. Tests are by Wilcoxon sign-ranks tests.

bPairwise comparisons of tree-ring width of tree i vs. acorn production by all other trees in sample except tree i. Numbers are mean correlation coefficients. Tests are by randomization. * = P < 0.05; ** = P < 0.01.

Source: Data from Hastings Reservation (Knops and Koenig, unpublished data).

Tree-ring width vs. acorn production by tree i. Numbers are trees exhibiting positive and negative correlations. Tests are by Wilcoxon sign-ranks tests.

bPairwise comparisons of tree-ring width of tree i vs. acorn production by all other trees in sample except tree i. Numbers are mean correlation coefficients. Tests are by randomization. * = P < 0.05; ** = P < 0.01.

inverse correlation between individuals but a positive correlation within individuals, while acorn production compared to growth the following year is significantly positive within individuals. Q. kelloggH demonstrates no significant relationship between acorn production and growth the prior year, but a significantly negative correlation between acorns and growth the following year, at least among individuals in the population.

These results support the possibility of resource switching in oaks, but they apparently accomplish this more by diverting resources from energy reserves than through direct tradeoffs between growth and vegetation. This contrasts with several other major genera of temperate trees for which switching of resources between growth and reproduction have been documented both within populations (Eis et al. 1965, Tappeiner 1969, Woodward et al. 1994, Norton and Kelly 1988) and between populations (Koenig and Knops 1998a).

Was this article helpful?

0 0

Post a comment