Austral Ecology (2008) 33, 78–87

Patterns and mechanisms of masting in the large-seeded southern hemisphere conifer Araucaria araucana JAVIER SANGUINETTI1 AND THOMAS KITZBERGER2* 1 Parque Nacional Lanín, Administración de Parques Nacionales, San Martín de los Andes, Argentina, and 2CONICET – Laboratorio Ecotono, CRUB, Universidad Nacional del Comahue, Quintral 1250, 8400, Bariloche, Argentina (Email: [email protected])

Abstract Masting, the intermittent and synchronous production of large seed crops, may result from either of two major processes: resource matching and economy of scale. Components of cone production in Araucaria araucana were partitioned among populations and trees to ascertain the existence of masting and the processes involved. Cone production data from seven populations were obtained during a 9-year period and seed gathering data were available for an 18-year time series from six sites in an area of more than 7600 km2. Araucaria araucana showed environmentally triggered, intermittent, moderately fluctuating, and highly regionally synchronous reproduction.The mean pairwise correlations of cones production among populations and seed gathering sites were 0.89 and 0.74, respectively, suggesting synchrony in reproduction. Among trees we observed a mean correlation of 0.74 with values ranging from 0.66 to 0.81 for the analysed populations. The existence of negative autocorrelation in seed production between year 0 and year -2 at the individual tree level suggests the presence of ‘switching’ or internal resource allocation, thus discarding the Resource Matching hypothesis. Mean coefficient of variation (CVp) among populations was moderate (0.95) and similar to the modal CVp values reported in the published reports. Mean CVi among individual trees was 1.16, suggesting a large number of equally and synchronously fluctuating trees, rather than a few largely fluctuating individuals. These results suggest that pollination efficiency and/or predator satiation hypotheses could be responsible for the masting cycles in this conifer. Ancillary data about limitation of airborne pollen dispersion and temporal variation in the amount of seeds per cone and about seed predator satiation, also support both proposed mechanisms. Key words: Araucaria araucana, masting, pattern, pollination, synchronization.

INTRODUCTION Masting is the intermittent and synchronous production of large seed crops by individuals in populations of long-lived plants (Janzen 1971; Kelly 1994). The fitness benefits accrued by plants and the effects of masting on other ecosystem components are determined by the spatial extent of synchrony (Curran & Leighton 2000). The causes and selective advantages of masting have been explained by several hypotheses proposed in recent decades (Janzen 1971; Waller 1979; Nilsson & Wästljung 1987; Lalonde & Roitberg 1992). Norton and Kelly (1988) and Kelly (1994) categorized these hypotheses into two types: (i) Resource Matching hypothesis, in which plants vary their reproductive effort in response to fluctuations in available resources, usually correlated with weather conditions (e.g. temperature or precipitation); and (ii) Economy of Scale, in which larger reproductive effort is more efficient, favouring an occasional large effort *Corresponding author. Accepted for publication May 2007.

© 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

rather than regular smaller ones. This last category encompasses more specific hypotheses such as the Pollination Efficiency hypothesis (Nilsson & Wästljung 1987) and the Predator Satiation hypothesis (Janzen 1971; Silvertown 1980). The Resource Matching hypothesis proposes that seed production tracks some limited resource, usually correlated with weather, so a positive correlation between growth and reproduction within years is expected (Norton & Kelly 1988; Sork et al. 1993; Kelly & Sork 2002). ‘Switching’, in which plants allocate to and then away from reproduction in successive years, would be key evidence against the Resource Matching hypothesis. The Pollination Efficiency hypothesis proposes that masting has evolved because fertilization efficiency is disproportionately high during mast years (Smith et al. 1990; Kelly et al. 2001). The Predator Satiation hypothesis proposes masting as an antipredator adaptation that increases seed survival by alternately starving and satiating predators. At least three conditions should be met to support this hypothesis: (i) seed production must be enough to guarantee the satiation of predators and the doi:10.1111/j.1442-9993.2007.01792.x

M A S T I N G I N A R AU C A R I A A R AU C A N A

79

survival of some seeds during highly productive years; (ii) the interval between mast years must be long enough to produce a decline in the predator population; and (iii) spatial synchronization of seed production must be proportional to the maximum range of movement of the important predators. Some authors highlight the importance of assessing spatial and temporal variation of seed production at the individual and population levels to understand the ecology of masting (Koenig et al. 2003). Currently, few studies evaluate mechanisms promoting masting in southern hemisphere trees (Herrera et al. 1998; Kelly et al. 2001; Kelly & Sork 2002; Schauber et al. 2002; Monks & Kelly 2006). The long-lived large-seeded southern conifer Araucaria araucana has life-history and ecological characteristics that are plausible in a species with intermittent variable reproduction (Koenig et al. 1994). In this paper we study for the first time the spatiotemporal pattern of cone production of A. araucana and its variation within and among years at tree and population levels in an area of 7600 km2. The objectives of our study were to: (i) quantify the degree of variation among years in seed production within and among sites; and (ii) test the predictions of various hypotheses for the evolution of masting seeding.

METHODS Study area The study was conducted in Neuquén Province, Argentina, at c. 39°-40°S latitude, in northern Lanín National Park. Within this area, covering nearly 1540 km2, we selected seven populations 7–46 km apart, where cone production censuses were conducted. Complementary data of human seed gathering recorded annually by provincial government surveys from six additional locations were also used, extending the study area to more than 7600 km2 (Fig. 1). This region is characterized by three physiographic subregions from west to east: the main Andean cordillera, the precordillera of foothills (glacial lakes and valleys), and the Patagonian steppe. Annual average rainfall decreases abruptly from west to east from more than 3000 mm year-1 on the western side of the Andes to about 800 mm year-1 on the eastern side (Barros et al. 1983). Rainfall occurs mainly during the colder period of the year (April to September), with summer (December to February) often being very dry. Average January air temperatures are 17–19°C and average July temperatures are 7–8°C (De Fina 1972; Huesser et al. 1988). © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

Fig. 1. Map of the study area showing the distribution of Araucaria araucana in Argentina (grey area) and the locations of visual cone count sites (triangle), sites where seed gathering occurs (circles) and weather stations (rectangle: 1. Añihueraqui, 2. Ea. Mamuil Malal, 3. Huechulafquen). Chapelco weather station is off map, 20 km south of the Huechulafquen station. CAL, Calfiquitra; ELA, El Arco; KIL, Kilca; LAN, Lanin; LIT, Litran; MAL, Malalco; MOQ, Moquehue; PHA, Pino Hachado, RUE, Rucachoroy Este; RUO, Rucachoroy Oeste; TAQ, Taquinquin; TRE, Tromen Este; TRO, Tromen Oeste.

Within the study area, A. araucana occurs in pure stands and in mixed forests with Nothofagus spp. From 1200 m to tree line, A. araucana forms mixed forests with N. pumilio. In valleys and lower, drier, north-facing slopes, it is mixed with the tall shrub N. antarctica. Extensive pure stands of A. araucana occur on poor volcanic soils and rocky slopes from 1000 to 1800 m a.s.l. and in the eastern part of its range (Rechene et al. 2003).

Study species Araucaria araucana (family Araucariaceae) is a longlived (>1200 years), large-seeded emergent conifer from the temperate forests of southern Argentina and Chile (Veblen et al. 1995). This conifer is adapted to doi:10.1111/j.1442-9993.2007.01792.x

80

J. S A N G U I N E T T I A N D T. K I T Z B E R G E R

Table 1. Cone production and seed collection data: Temporal variation (CVp); period of time and years of sampling of each population and site Populations

CVp

Years

Period

Sites

CVp

Years

TRE TRO RUE RUO CAL MAL TAQ Mean† Mean‡

0.83 0.85 0.86 1.03 1.12 0.83 1.10 0.95 0.84

9 9 9 9 7 6 5 7.7 4.7

2000–08 2000–08 2000–08 2000–08 2002–08 2003–08 2004–08

PHA LIT KIL MOQ ELA LAN

1.68 1.39 1.19 2.69 1.01 0.81

18 18 18 18 18 13

1.46 1.19

17.2 6

Period 1988–05 1988–05 1988–05 1988–05 1988–05 1993–05

†

Overall mean across populations and sites. ‡Mean during overlap period (2000–05). CAL, Calfiquitra; ELA, El Arco; KIL, Kilca; LAN, lanin; LIT, Litran; MAL, Malalco; MOQ, Moquehue; PHA, Pino Hachado; RUE, Rucachoroy Este; RUO, Rucachoroy Oeste; TAQ, Taquinquin; TRE, Tromen Este; TRO, Tromen Oeste.

stressful conditions, is shade tolerant and grows mainly on poor volcanic soils or rocky places. Its adaptations to fire include a thick bark, epicormic buds that sprout after fire and terminal buds that are protected by modified leaves. Mature trees often lose lower branches, giving the tree an umbrella-like shape. All of these features allow A. araucana to compete successfully with Nothofagus spp. (Veblen 1982; Burns 1993). Trees reach sexual maturity when they have trunks greater than 20 cm d.b.h. and are more than 30 years old (Muñoz Ibañez 1984). A. araucana is dioecious (rarely monoecious) and the sex of the trees can be determined by the cones it makes. Female cones are very large (15–20 cm in diameter), and contain 100– 200 large seeds (3.5 g) that are dispersed by gravity over short distances. Seed production increases with age and hierarchical position within the canopy (Muñoz Ibañez 1984). A. araucana has one of the largest pollen grains (80–100 mm in diameter) among all conifers (Muñoz Ibañez 1984; Huesser et al. 1988). Female cones are wind-pollinated during summer and seed maturation takes between 16 and 18 months. This prolonged development means that two consecutive seed generations are on the tree from summer to autumn, when the mature seeds fall. Short seed viability (less than 6 months), low fertility (fitness) and poor passive dispersal result in poor overall seed regeneration (Muñoz Ibañez 1984; Armesto et al. 1997; Donoso 1998). Most seedlings grow directly under the parent tree but only those seedlings that establish beyond the parent canopy have a good chance of surviving (Finckh & Paulsch 1995). Seeds are valuable edible resources, subject to preand postdispersal predation (Armesto et al. 1997; Shepherd & Ditgen 2005). Historically, indigenous Mapuche people have depended on A. araucana and consider it a sacred tree. They collect the seeds for personal consumption, and to exchange for goods and cash (Aagesen 1998; Herrmann 2006). doi:10.1111/j.1442-9993.2007.01792.x

Cone production Every summer between 2000 and 2007 we counted individual tree cone production from seven populations using binoculars, a technique that has been successfully carried out in previous studies (Muñoz Ibañez 1984). Female ‘seed trees’ over 20 cm d.b.h., were selected randomly while the observer was walking through trails. Depending on the amount of visible crown, the sample was obtained over the whole or only on half of the tree. If half of the tree was sampled, we multiplied the data by 2 to standardize the whole data set. Production for the next year was also estimated by counting immature green cones (Muñoz Ibañez 1984), enabling us to extend the record to 2008. Four populations were sampled from 2000 and three more populations were added to the study from 2002 to 2004 (Table 1). Each year, between 50 and 100 seed trees per population were sampled. A total of 3679 trees were sampled during the study with an average of 613 trees per year (range 163–800). In 2003, 20 trees were tagged in each population to monitor the individual cone production over the last 5 years of the study.These data were used to evaluate the reproductive effort and synchronization among individual trees within and among populations. The annual proportion of non-reproductive trees was estimated by counting mature individuals without cones that had evidence of previous reproductive structures. Additional information about seed production was inferred from national and provincial government estimates of seeds gathered by native people throughout the species distribution from 1988 to 2005 (Table 1).

Data analyses For most of the analyses, number of cones per tree was used without transformation. Where parametric © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

M A S T I N G I N A R AU C A R I A A R AU C A N A

analyses were carried out, the data were logtransformed (ln (cones +1)) to reduce correlation between the mean and variance and to normalize the data (Sokal & Rohlf 1994).

Temporal variability in cone production To estimate the average among-year variation, we calculated the coefficient of variation (CV = standard deviation/mean of untransformed data) of the global mean cone number per tree for each population (CVp). The mean CVp from all populations was used as an index of the among-year cone production variation (Koenig et al. 2003). At tree level, cone production variability was assessed by the coefficient of variation for each tagged tree (CVi) during 5 years (2004–2008). For each population, mean CVi was calculated and the average of these values was used as an index of the among-year variation at tree level (Koenig et al. 2003).

Reproductive effort bimodality among years Among-year bimodality of reproductive effort was tested following Herrera et al. (1998). Annual cone output from four populations measured during 9 years and seed gathering data from six sites were log transformed and then standardized to a mean of 0 and standard deviation of one.With these data (n = 139), a combined frequency distribution was created. We tested for bimodality using a Kolmogorov-Smirnov Test to find departure from a zero-mean normal distribution.

Crop failure We used the Kruskal-Wallis test to find differences in the annual percentage of non-reproductive mature trees in the four populations that were sampled throughout the study. At tree level, a Friedman Test was carried out using the tagged tree data among seven populations and the Kendall’s concordance coefficient was estimated to evaluate non-reproductive trees pattern among populations.

Endogenous factor: Occurrence of consecutive masting years The endogenous aspects of masting were analysed by two means: (i) tree-level lagged regressions between cone production during years t and t-1, and t-2; and (ii) lag-1 and lag-2 autocorrelation functions (ACF) using the 9-year mean cone production data. Because © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

81

individual trees differed in intrinsic production, regressions were carried out using deviations in cone productions from each 5-year individual mean. This relationship is intrinsically negative (i.e. random production patterns lead to negative relationships). We tested for non-random deviations in slope and R2 by comparing observed parameters with a distribution generated from 1000 simulated lagged regressions by randomizing (without replacement) the 5-year cone production of each individual tree.

Identification of climatic cues for masting We conducted a superposed epoch analysis (SEA; Baisan & Swetnam 1990) using years with the highest (1991, 1995, 2000) and the lowest (1989, 1992, 1996 and 2005) seed production. Climatic data from Chapelco (1989–2005), Estancia Mamuil Malal (1997–2005), Seccional Añihuaraqui (1997–2005), and Huechulafquen (2000–2005, Fig. 1) were on mean normalized seasonal data. SEA compared the mean climatic conditions (monthly precipitation and temperature) during and 3 years before the high/low events and built confidence intervals (2.5 and 97.5 percentiles) based on resampling the same time window on the series 1000 times.

Synchronization in cone production Synchronization among trees Pearson correlation coefficients were calculated for each pair of trees in each population based on cone production (2004–2008). Mean correlation in each population, rp, and of all trees was used as an index of synchrony among trees. We also calculated the mean r of 1000 pseudopopulations of 20 tree pairs constructed by randomly assigning to each tree the 5-year cone production of one of the 139 trees. If correlation within populations, rp exceeded the 95% confidence interval for correlation among the randomized pseudopopulations, it would suggest that synchrony among trees within populations was greater than synchrony among trees of different populations.

Synchronization among populations Correlation coefficients were calculated based on mean annual cone production and mean annual seed gathering data. Correlation coefficients were calculated for all pairwise cone production populations, seed gathering sites and their pairwise combinations based on a variable length period (3–18 years) as well doi:10.1111/j.1442-9993.2007.01792.x

J. S A N G U I N E T T I A N D T. K I T Z B E R G E R

Spatial synchronization Each year, the coefficient of variation was calculated for the number of cones per tree in each population. The average of these values was used as an index of the within-year variation to reflect the spatial synchronization among trees. Low CVs would indicate high synchrony among trees (Houle 1999).

(a) KIL MOQ PHA LIT ELA LAN

100 80 60 40 20 0

1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

as for a common period of observation (2000–2005). The significance of each pairwise combination was evaluated after sequential Bonferroni correction (Sokal & Rohlf 1994). The effect of distance between A. araucana populations (cone production and seed collection sites) on synchronization at population level was evaluated by simple linear regression.

Seed collection (tonnes)

82

Year (b) 80 70

doi:10.1111/j.1442-9993.2007.01792.x

TRE TRO RUE RUO CAL MAL TAQ

60 50 40 30 20 10 0

2000 2001 2002 2003 2004 2005 2006 2007 2008

Year

(c) Seed collection Cone production

4 3 2 1 0 -1

90

92 19 94 19 96 19 98 20 00 20 02 20 04 20 06 20 08

19

19

88

-2

19

Seed collection showed high interannual variability in seed crops during the 1988–2005 period (Fig. 2a). Peaks in seed production occurred in 1991, 1995, 1997, 2000 and 2004. Seed crops were low during 1989, 1992, 1993, 1994, 1996, 2001, 2002 and 2005. Mean cone production between 2000 and 2008 ranged from two to 45 cones per tree with higher cone production in 2000, 2004 and 2007, and lower production in 2001, 2002, 2005 and 2008 (Fig. 2b). In all cases, greater variation between years was observed following peaks of production and/or gathering data. Where time series overlapped (2000–2005), mean seed gathering from six sites and mean cone production from seven populations were positively correlated (r = 0.968, P < 0.01, n = 6), thus validating the use of the longer seed gathering records as an indicator of reproductive effort of A. araucana (Fig. 2c). The coefficient of variation in cone production among-years at the population level (CVp) for the seven populations varied between 0.83 and 1.12 and for the seed collection between 0.81 and 2.69 (Table 1). For the full period, mean CVp for cone production and seed collection data were 0.95 and 1.46, respectively, but 0.84 and 1.19 for the period of overlap (Table 1).The mean CVp for the four populations sampled for 9 years was 0.89 and for all populations during the overlap period (2004–2008) was 1.02. The mean individual variation (CVi) for all tagged trees from all populations together, which reflects temporal variability in reproductive effort at tree level, was moderate (1.16 ⫾ 0.06 SE) but varied significantly among populations (d.f. = 6; F = 6.314; P < 0.0001). The trees in Tromen Este (TRE) had higher temporal

Standardized index

Temporal variability in reproductive effort

Mean cones per tree

RESULTS

Years

Fig. 2. Seed collection and cone production data: (a) Seed gathering level (tonnes) in six sites outside Lanín National Park between 1988 and 2005; (b) Mean cone production in seven monitored populations between 2000 and 2008; (c) Standardized mean annual seed gathering (solid dots) and mean number of cones per tree-1 (empty dots). CAL, Calfiquitra; ELA, El Arco; KIL, Kilca; LAN, Lanin; LIT, Litran; MAL, Malalco; MOQ, Moquehue; PHA, Pino Hachado ; RUE, Rucachoroy Este; RUO, Rucachoroy Oeste; TAQ, Taquinquin; TRE, Tromen Este; TRO, Tromen Oeste.

© 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

M A S T I N G I N A R AU C A R I A A R AU C A N A

83

Table 2. Mean individual variation (CVi), population variation (CVp), spatial (CVs) and temporal (rp) synchronization of 139 tagged trees from seven populations and 5 years (2004–08). Results are using untransformed data Temporal variability indices

Spatial variability/synchrony indices

Population

CVi

CVp

CVs

rp

TRE TRO MAL CAL RUE RUO TAQ Mean

1.48 1.15 1.07 1.03 1.14 1.22 1.03 1.16

1.16 0.83 0.91 0.95 0.96 1.06 0.93 0.97

1.56 1.29 0.94 0.75 1.26 1.26 1.16 1.17

0.73 0.66 0.74 0.81 0.79 0.68 0.74 0.74

CAL, Calfiquitra; MAL, Malalco; RUE, Rucachoroy Este; RUO, Rucachoroy Oeste; TAQ, Taquinquin; TRE, Tromen Este; TRO, Tromen Oeste.

Table 3. Matrix intersite correlation coefficients based on cone production populations (codes in italics) and seed gathering data (codes in bold) LAN LAN KIL MOQ PHA LIT ELA TRE TRO RUE RUO CAL

0.241 0.629 0.242 0.233 0.203 0.750 0.629 0.564 0.544

KIL

MOQ

PHA

LIT

ELA

TRE

TRO

RUE

RUO

CAL

MAL

0.228

0.111 0.077

0.209 0.923 -0.010

0.199 0.950 0.003 0.994

0.099 0.882 -0.087 0.873 0.908

0.750 0.322 0.611 0.323 0.332 0.367

0.629 0.798 0.907 0.799 0.786 0.734 0.731

0.564 0.860 0.955 0.860 0.860 0.854 0.876 0.908

0.544 0.943 0.955 0.944 0.941 0.924 0.647 0.903 0.878

0.866 0.541 0.479 0.514 0.533 0.537 0.869 0.919 0.918 0.928

0.936 0.981 0.956 0.972 0.978 0.980 0.981 0.953 0.917 0.963 0.819

0.859 1.00 0.999 0.989 0.322 0.798 0.860 0.943

0.860 0.852 0.818 0.611 0.907 0.955 0.955

1.00 0.989 0.323 0.799 0.860 0.944

0.993 0.332 0.786 0.860 0.941

0.367 0.734 0.854 0.924

0.455 0.717 0.553

0.867 0.909

0.948

Top-right half of the matrix are pairwise correlations coefficients with variable n (3–18). Figures in the lower-left half of the matrix are correlation coefficients based on a common period of observation (2000–05). Bold type is significant after sequential Bonferroni correction with a 0.05. Populations CAL and MAL were monitored since 2002 and therefore could not be correlated during the common 2000–2005 period. CAL, Calfiquitra; ELA, El Arco; KIL, Kilca; LAN, Lanin; LIT, Litran; MAL, Malalco; MOQ, Moquehue; PHA, Pino Hachado ; RUE, Rucachoroy Este; RUO, Rucachoroy Oeste; TRE, Tromen Este; TRO, Tromen Oeste.

variability in cone number than trees located in Taquinquin (TAQ), Calfiquitra (CAL), Malalco (MAL), Rucachoroy Este (RUE) and Tromen Oeste (TRO). The relationship between CVp and CVi was positive and linear (CVp = 0.338 + 0.546*CVi, R2 = 0.64; P < 0.05) suggesting that temporally variable populations (e.g. TRE, RUO, Table 2) are due to a high synchronization among individual trees. Less variable populations (e.g. TAQ, CAL) had lower CVi suggesting that individual trees were less temporally variable but not asynchronous.

Reproductive synchrony among populations Reproductive effort was highly synchronized among populations distributed over thousands of square kilometres (Fig. 1). Twenty-four out of 66 population © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

pairs (36%) had significant pairwise correlation coefficients (period 1988–2005, P < 0.05, Table 3). Based on a common period 2000–2005, 25 (55.6%) out of 45 population pairs had significant correlation coefficients (P < 0.05) and overall mean correlation was r = 0.731 (⫾0.037 SE). Mean correlation between gathering sites was 0.727 ⫾ 0.08, between populations based on cone production 0.89 ⫾ 0.02, and between cone production and gathering 0.732 ⫾ 0.04. When all populations were included, the relationship between intersite correlation (CORR) 2000– 2005 and geographical distance (DIST) was negative but only 10% of the variance was explained by distance (DIST = 0.876–0.003*CORR, R2 = 0.099, P = 0.035). Two populations, Lanin (LAN) and TRE had overall low correlation with all other populations (Table 3). The reason for this may be that at LAN, the human seed gathering pressure increased more than in doi:10.1111/j.1442-9993.2007.01792.x

J. S A N G U I N E T T I A N D T. K I T Z B E R G E R

Deviation in cone production (t)

84

1.0

Correlation

0.9

2

R = 0.57 0.8 0.7 0.6 0.5 0

20

40

60

80

100

120

Distance (km) Fig. 3. Relationship between synchrony in cone observation sites plus seed gathering sites expressed as pairwise correlations between sites, and distance between sites (common period 2000–2005). Only highest correlating sites were considered for regression analysis (lower correlations involving Lanin (LAN) and Tromen Este (TRE) are shown with open circles).

100 80 t-2 b1= -0.74 2 60 R = 0.65

t-2 t-1

40 20 0

t-1 b1= -0.30 2 R = 0.07

-20 -40 -60 -100 -80

-60

-40

-20

0

20

40

60

80

100

Deviation in cone production (t-1 or t-2) Fig. 4. Relationships between cone production (deviations from the 5-year individual mean) during year t and production during year t-1 (empty triangles) and t-2 (solid diamonds).

not vary significantly between populations (c2 = 7.78; d.f. = 6; P = 0.255) (Table 2). the other sites in the last 4 years (Fig. 2a) and that TRE had the poorest soil of all populations, pure volcanic sand, which could explain its higher production variability (Table 2). When these populations were excluded from the analysis, the explanatory value of distance on correlation increased to 57% (Fig. 3). Synchrony with distance followed a relatively slow decline, so that correlation coefficients remained positive and significant (P < 0.05) over distances of <100 km between populations (Fig. 3).

Bimodality of reproductive effort among years The combined frequency distribution of standardized annual cone counts and seed gathering data did not depart significantly from the standardized normal distribution (D = 0.085, P > 0.20; Kolmogorov test), which indicates that reproductive effort was not bimodal among years.

Synchrony among trees Mean temporal synchronization (rp) in cone production among tagged trees ranged from 0.66 to 0.81 (Table 2, mean rp = 0.74). Level of synchronization differed among populations (F6,1336 = 9.168; P < 0.001, Table 2). Only one (CAL) out of seven populations had rp that exceeded expectations of synchrony based on randomly drawing trees from any sampled population. Spatial mean synchronization in cone production among tagged trees during five years (CVs) was 1.17. Mean CVs of trees at each population did doi:10.1111/j.1442-9993.2007.01792.x

Distribution of reproductive effort and crop failure The percentage of trees with non-reproductive trees varied significantly between years (H8,36 = 27.99; P = 0.0005). During 2005 and 2008, 52% and 45% of total trees, respectively, were non-reproductive individuals. In contrast, during 2000, 2004 and 2007, when the highest seed production occurred, less than 0.7% of total trees were non-reproductive.

Endogenous masting pattern At the individual tree scale, deviations in cone production from the 5-year individual mean during year t were significantly negatively related to the production during year t-2 (Fig. 4). Neither slope nor R2 of the 1-year lagged relationship between deviations in cone production deviated significantly from random expectations. In contrast, Monte Carlo analysis showed that none of the 1000 simulations carried out by randomizing individual cone production among years had slopes lower than the observed -0.74 and R2 larger than the observed 0.65 in the lag-2 regression. Similar results were found with ACF analyses. Absence of significant negative 1-year lag autocorrelations in cone production (rs = -0.14, P = 0.76) suggests that while mast years are typically followed by a year of low cone production, a year of few cones was not necessary followed by a mast year. In contrast, a © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

M A S T I N G I N A R AU C A R I A A R AU C A N A

significant negative 2-year lag autocorrelation was detected (rs = -0.83, P = 0.04).

Climatic cues promoting masting Superposed epoch analyses indicated that the only climatic variable with a significant effect on cone production was summer (Jan–Mar) precipitation 2 years before a masting event. Summers 2 years before masting events were significantly drier than expected from randomly generated events (P < 0.05). No other environmental variable within 3 years of a masting or non-masting year had a significant effect.

DISCUSSION Araucaria araucana showed environmentally triggered, intermittent, moderately fluctuating, and regionally synchronous reproduction. This study showed that populations spanning an area of more than 7600 km2 were highly synchronized in reproduction, with rp similar to the higher values reported in the published reports for masting trees (Koenig et al. 2003). The degree of synchrony at tree level was high, regardless of the spatial location of trees. The spatial pattern of this synchronization may have been due to the influence of regional climatic cues (Norton & Kelly 1988) and/or related to selective forces promoting masting at large spatial scales (i.e. tens to hundreds of km). Temporal variation in the population reproduction (CVp) was not exceptionally high and was similar to the modal CVp values reported by other authors for masting trees (Kelly 1994; Koenig et al. 2003). The results showed that most of the temporal variability in cone production resulted from a large number of equally and synchronously fluctuating trees, rather than a few with large fluctuations. The difference in crop size between good producers and poor producers was small and mean CVi and CVp were similar in each of the seven populations studied, as predicted by the theoretical model proposed by Buonaccorsi et al. (2003). Our indirect analytical approach to test the existence of bimodal reproductive effort among years showed that A. araucana had an annual variability in seed production that fluctuated randomly around an average value instead of having distinctly high and low seed production years. Although it was impossible to confirm the existence of bimodality because of small sample size, dividing years into ‘mast’ and ‘non-mast’ years was inappropriate, as the masting concept must be applied cautiously (Kelly 1994). Summer drought 2 years before seedfall may be the climatic cue that triggered region-wide masting, possibly by inducing high cone production and therefore a © 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

85

more efficient pollination. This timing is consistent in the closely related Araucaria angustifolia, in which it takes 20–24 months from male and female cone bud initiation to final seed maturation (Mantovani et al. 2004). Pollen grains of Araucariaceae, unlike that of other conifers, are very large (up to 100 mm in diameter) and lack aeriferous vesicles (Owens et al. 1998). These physical characteristics of pollen may cause short dispersal distances so that large pollen clouds (or more receptive ovules) may be necessary for adequate pollination and gene flow (Sousa & Hattemer 2003).

Mechanisms of masting The existence of a negative autocorrelation in seed production between year 0 and year -2 is strong evidence against the Resource Matching hypothesis. Supra-annual female cone maturation and an extremely large seed size could be selective forces for a resource allocation mechanism. In contrast, the high degree of tree and population synchrony, the existence of years with a high proportion of non-reproductive trees and the low variability of effort within years of high production, all suggest that neither pollination efficiency nor predator satiation can be discarded as mechanisms that selectively promote masting. Araucaria araucana may have developed a strategy to ‘swamp’ the system and satiate seed predators while escaping predation and ‘starve’ predators during intervening poor production years. Alternatively, A. araucana may have developed a strategy of pollen and ovule saturation by sharply increasing pollen and female cone production in order to override the strong inherent pollen dispersal limitation, thereby increasing fertilization and gene flow. The Pollination Efficiency and Predator Satiation hypotheses predict very similar patterns of spatial and temporal variation in reproductive effort (Koenig et al. 2003). Selection may favour both pollination efficiency and predator satiation together in the development of this reproductive strategy. Key ancillary information may help in judging the relative importance of these mechanisms:

Spatial reproductive synchrony and mobility of seed predators In A. Araucana forests, the main native predispersal seed predator is the austral parakeet (Enicognathus ferrugineus) which moves distances greater than 10 km (Díaz & Kitzberger 2006; Valeria Ojeda, pers. comm. 2006) and feeds on large quantities of A. araucana pollen and seeds. Reproductive synchrony among populations located 10–100 km apart, as demonstrated in this study, may facilitate parakeet satiation. doi:10.1111/j.1442-9993.2007.01792.x

86

J. S A N G U I N E T T I A N D T. K I T Z B E R G E R

On the other hand, at least four species of generalist rodents are postdispersal predators with limited mobility and strong microhabitat requirements (Shepherd & Ditgen 2005; Sanguinetti & Kitzberger unpubl. data, 2006). These rodents would mitigate the need for long range spatial synchrony.

Predator satiation during masting Austral parakeets showed a negative relationship between predispersal seed predation rates and the mean cone number per tree during the observation period of this study. However, parakeets seem not to be at starvation points during poor production years, as predation rates were 2% and 20% in peak and low-production years, respectively (Shepherd et al. unpubl. data, 2006). Therefore parakeets may exert moderate selection against asynchronous reproducing trees (Janzen 1971). Preliminary data on postdispersal predation by less mobile rodents suggest a stronger satiation effect. During the masting event of 2004, seed predation rates by rodents increased from 30% in early autumn to 70% in late autumn, possibly due to a fast numerical response. Although seed fall continued throughout early winter, mortality stabilized (seed disappearance rates 6.8% per night) possibly due to satiation and/or lower rodent activity. In contrast, during 2005, a year of extremely low seed production, 98% seed mortality was measured with a fivefold increase in disappearance rates (39.1% per night). During 2006, a year of intermediate production, 95% of seeds were consumed by rodents and disappearance rates were intermediate (15.9% per night) (Sanguinetti & Kitzberger 2004; Sanguinetti & Kitzberger, unpubl. data, 2006).

Seed viability and production relationships and energetic costs of reproductive structures There is some evidence to indicate that seed set in A. araucana tends to be higher when cone density is higher. For instance, during a peak production year the number of seeds per cone was 135 ⫾ 8 (n = 31), with 99. 4% ⫾ 0.4 (n = 31) of seeds viable, whereas during a mean production year (2001), the number of seeds per cone was 90 ⫾ 12 (n = 19), with 68.6% ⫾ 7.9 (n = 21) of seeds viable (Javier Sanguinetti, unpubl. data, 2002). This 33% difference in seed set between the peak and the mean years may indicate that A. araucana increases its pollination efficiency by masting in concordance with the theoretical model proposed by Kelly et al. (2001). The consequences of inefficient pollination could be severe for A. araucana because it is a dioecious species that allocates more than 55% of the female cone weight to woody tissue doi:10.1111/j.1442-9993.2007.01792.x

regardless of the amount of seeds it is producing (Muñoz Ibañez 1984; Smith et al. 1990).

Incidence of successive masting years In this study we did not detect successive years of high productivity so we could not reject the Predation Satiation hypothesis (Koenig et al. 1994, 2003). In summary, this study strongly suggests that A. araucana shows Normal Masting Switching type, according to Kelly’s definitions of the variety of masting forms (Kelly 1994). Although evidence points to Pollination Efficiency and/or Predator Satiation as the main mechanisms involved, further studies will be necessary to better discriminate the selective forces that have led to such an impressively synchronized masting system.

ACKNOWLEDGEMENTS The authors thank Marcelo Gonzalez Peñalba, Luis Chauchard, Leonardo Maresca, Liliana Lozano, Nicolas Katuchin, Santiago Quiroga and the summer students and volunteers from Global Vision International (GVI) for their help during the field work. We especially thank Leonardo Gallo and Fernanda Izquierdo from INTA-Bariloche for their invaluable support during the early stage of this work. We thank John Shepherd and Michael Bull for improvement of the manuscript.This work is part of the PhD thesis of JS supported by the German Government (GMZ Minister) and the International Plant Genetic Resources Institute (IPGRI), by the Russell Train Scholarship (WWF), by the Agencia Española de Coop. Internacional (AECI) with the project CGL2004-0176-Feder (Salvador J. Peris coord.) and by the National Park Administration (Lanín National Park).

REFERENCES Aagesen D. L. (1998) Indigenous resource rights and conservation of the monkey-puzzle tree (Araucaria araucana, Araucariaceae): a case study from southern Chile. Econ. Bot. 52, 146–60. Armesto J. J., Villagran C. & Arroyo M. K. (1997) Ecología de los bosques nativos de Chile. Ed. Universitaria, Santiago de Chile. Baisan C. & Swetnam T. W. (1990) Fire history on a desert mountain range: Rincon Mountain Wilderness, Arizona, USA. Can. J. For. Res. 20, 1559–69. Barros V., Cordon V., Moyano C., Mendez R., Forquera J. & Pizzio O. (1983) Carta de precipitación de la zona oeste de Río Negro y Neuquén. Universidad Nacional del Cornahue, Fac. De Agronomía Cinco Saltos, Neuquén, Buenos Aires. Buonaccorsi J. P., Elkinton J., Koenig W., Duncan R. P., Kelly D. & Sork V. (2003) Measuring mast seeding behavior:

© 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

M A S T I N G I N A R AU C A R I A A R AU C A N A

relationships among population variation, individual variation and synchrony. J. Theol. Biol. 224, 106–14. Burns B. R. (1993) Fire-induced dynamics on Araucaria araucana-Nothofagus antarctica forest in the southern Andes. J. Biogeogr. 20, 669–85. Curran L. M. & Leighton M. (2000) Vertebrate response to spatiotemporal variation in seed production of mast-fruiting Dipterocarpaceae. Ecol. Monogr. 70, 101–28. De Fina A. L. (1972) El clima de la región de los bosques andino-patagónicos argentinos. In: La región de los bosques andino-patagónicos (eds M. J. Dimitri) pp. 35–58. INTA, Buenos Aires. Díaz S. &. Kitzberger T. (2006) High Nothofagus flower consumption and pollen emptying in the southern South American austral parakeet (Enicognathus ferrugineus). Austral Ecol. 31, 759–66. Donoso C. Z. (1998) Bosques templados de Chile y Argentina. Variación, Estructura y Dinámica. Ecología Forestal. Ed. Universitaria, Universidad de Chile, 484 pp. Finckh M. A. & Paulsch A. (1995) The ecological strategy of Araucaria araucana. Flora 190, 365–82. Herrera C. M., Jordano P., Guitian J. & Traveset A. (1998) Annual variability in seed production by woody plants and the mast concept: reassessment of principles and relationship to pollination and seed dispersal. Am. Nat. 152, 576–94. Herrmann T. M. (2006) Indigenous knowledge and management of Araucaria araucana forest in the Chilean Andes: implications for native forest conservation. Biodivers. Conserv. 15, 647–62. Houle G. (1999) Mast seeding in Abies balsamea, Acer saccharum and Betula alleghaniensis in an old growth, cold temperate forest of north-eastern North America. J. Ecol. 87, 413–22. Huesser C. J., Rabassa J., Brandani A. & Stuckenrath R. (1988) Late-Holocene vegetation of the Andean Araucaria region, Province of Neuquén, Argentina. Mt. Res. Dev. 8, 53–63. Janzen D. H. (1971) Seed predation by animals. Annu. Rev. Ecol. Syst. 2, 465–92. Kelly D. (1994) The evolutionary ecology of mast seeding. Trends Ecol. Evol. 9, 465–70. Kelly D. & Sork V. L. (2002) Mast seeding in perennial plants: why, how, where? Annu. Rev. Ecol. Syst. 33, 427–47. Kelly D., Hart D. E. & Allen R. B. (2001) Evaluating the wind pollination benefits of mast seeding. Ecology 82, 117–26. Koenig W. D., Mumme R. L., Carmen W. J. & Stanback M. T. (1994) Acorn production by Oaks in Central Coastal California: variation within and among years. Ecology 75, 99–109. Koenig W. D., Kelly D. & Sork D. V. (2003) Dissecting components of population-level variation in seed production and the evolution of masting behavior. Oikos 102, 581–91. Lalonde R. G. & Roitberg B. D. (1992) On the evolution of masting behavior in trees: predation or weather? Am. Nat. 139, 1293–304. Mantovani A., Morellato P. C. & Dos Reis M. S. (2004) Fenologia reprodutiva e produção de sementes em Araucaria angustifolia (Bert.) O. Kuntze. Revista Brasileira de Botanica 27, 787–96.

© 2008 The Authors Journal compilation © 2008 Ecological Society of Australia

87

Monks A. & Kelly D. (2006) Testing the resource-matching hypothesis in the mast seed tree Nothofagus truncaata (Fagaceae). Austral Ecol. 31, 366–75. Muñoz Ibañez R. (1984) Análisis de la productividad de semillas de Araucaria araucana (Mol.) C. Koch, en el Area de Lonquimay – IX Región. Tesis para optar al Titulo de Ingeniero Forestal. Facultad de Ciencias Agrarias, Veterinarias y Forestales. Escuela de Ciencias Forestales, Universidad de Chile, 98 pp. Nilsson S. G. & Wästljung U. (1987) Seed predation and crosspollination in mast-seeding beech (Fagus sylvatica) patches. Ecology 68, 260–5. Norton D. A. & Kelly D. (1988) Mast seeding over 33 years by Dacrydium cuppressinum Lamb. (rimu) (Podocarpaceae) in New Zealand: the importance of economics of scale. Funct. Ecol. 2, 399–408. Owens J. N., Takaso T. & Runions C. J. (1998) Pollination in conifers. Trends Plant Sci. 3, 479–85. Rechene C., Bava J. & Mujica R. (2003) Los bosques de Araucaria araucana en Chile y Argentina. Programa de Apoyo Ecológico (TOEB), GTZ, Agencia de Cooperación Alemana, Informe TWF-V/40s, 158 pp, Eschborn, Alemania. Sanguinetti J. & Kitzberger T. (2004) Variación espacial y temporal de la predación de semillas de Araucaria araucana por roedores y jabalí (Sus scrofa): influencia de los microhabitats. II. Reunión Binacional de Ecología, Mendoza, Noviembre 2004. Schauber E. M., Kelly D., Turchin P. et al. (2002) Masting by eighteen New Zealand plant species: the role of temperature as a synchronizing cue. Ecology 83, 1214–25. Shepherd J. D. & Ditgen R. S. (2005) Human use and small mammals communities of Araucaria forests in Neuquén, Argentina. Mastozoología Neotrop. 12, 226. Silvertown J. (1980) The evolutionary ecology of the mast seeding in trees. Biol. J. Linn. Soc. 14, 235–50. Smith C. C., Hamrick J. L. & Kramer C. L. (1990) The advantage of mast years for wind pollination. Am. Nat. 136, 154– 66. Sokal R. R. & Rohlf J. (1994) Biometry:The Principles and Practice of Statistics in Biological Research, 3rd edn. W.H. Freeman and Company, NY. Sork V. L., Bramble J. & Sexton O. (1993) Ecology of mastfruiting in three species of Missouri oaks, Quercus alba, Quercus rubra & Quercus velutina (Fagaceae). Ecology 74, 528–41. Sousa V. A. & Hattemer H. H. (2003) Pollen dispersal and gene flow by pollen in Araucaria angustifolia. Aust. J. Bot. 51, 309–17. Veblen T. T. (1982) Regeneration patterns in Araucaria araucana forest in Chile. J. Biogeogr. 9, 11–28. Veblen T. T., Burns R. B., Kitzberger T., Lara A. & Villalba R. (1995) The ecology of the conifers of southern South America. In: Ecology of the Southern Conifers (eds N. Enright & R. Hill) pp. 120–155. Melbourne University Press, Parkville. Waller D. M. (1979) Models of mast fruiting in trees. J. Theor. Biol. 80, 223.

doi:10.1111/j.1442-9993.2007.01792.x