Fire structures pine Serotiny at different scales

Abstract

Serotiny (delayed seed release with the consequent accumulation of a canopy seedbank) confers fitness benefits in environments with crown-fire regimes. Thus, we predicted that serotiny level should be higher in populations recurrently subjected to crown-fires than in populations where crown-fires are rare. In addition, under a high frequency of fires, space and resources are recurrently available, permitting recruitment around each mother to follow the seed rain shadow. Thus, we also predicted spatial aggregation of serotiny within populations.
We compared serotiny, considering both the proportion and the age of serotinous cones, in populations living in contrasting fire regimes for two iconic Mediterranean pine species (Pinus halepensis, P. pinaster). We framed our results by quantitatively comparing the strength of the fire-serotiny relationship with previous studies worldwide.
For the two species, populations living under high crown-fire recurrence regimes had a higher serotiny level than those populations where the recurrence of crown-fires was low. For P. halepensis (the species with higher serotiny), populations in high fire recurrence regimes had higher fine-scale spatial aggregation of serotiny than those inhabiting low fire recurrence systems. The strength of the observed fire-serotiny relationship in P. halepensis is among the highest in published literature.
Fire regime shapes serotiny level among populations, and in populations with high serotiny, recurrent fires maintain a significant spatial structure for this trait. Consequently, fire has long-term evolutionary implications at different scales, emphasizing its prominent role in shaping the ecology of pines.

Full text

1
American Journal of Botany 100(12): 000–000. 2013.
American Journal of Botany 100(12): 1–8, 2013 ; http://www.amjbot.org/ © 2013 Botanical Society of America
Fire is a major ecological factor in many ecosystems, and
therefore, plants have developed traits to cope with recurrent
fi res ( Pausas et al., 2004 ; Pausas and Keeley, 2009 ). There is
an increasing bulk of information suggesting that different fi re
regimes may act as an evolutionary force shaping plant traits
( Keeley et al., 2011 ; Pausas and Schwilk, 2012 ; He et al.,
2012 ). One of the most apparent traits related to fi re is se-
rotiny, that is, the delayed seed release for more than a year,
requiring an environmental stimulus (a heat shock) for disper-
sal. The consequence of this delay is the retention of seeds in
“cones” (conifer cones or woody fruits) in the canopy for
more than one reproductive cycle (canopy seed bank; Lamont
et al., 1991 ). By delaying dispersal until a fi re occurs, seroti-
nous species recruit in postfi re conditions with high resource
availability, low competition, and low predation (predator
saturation), and thus, serotiny confers fi tness advantages in
ecosystems under crown-fi re regimes ( Lamont et al., 1991 ;
Keeley and Zedler, 1998 ; Lamont and Enright, 2000 ; He et al.,
2012 ). Consequently, the dynamics of serotinous populations
follow postfi re pulses of recruitment, in contrast to nonseroti-
nous populations that may produce recruits in any year. Se-
rotiny can be disaggregated in two components, the proportion
of serotinous cones and the time these cones remain closed on
the plant ( Midgley, 2000 ). A given level of serotiny can be
achieved by accumulating many weak (short-lived) serotinous
cones or few strong (long-lived) serotinous cones across years.
However, the two components have rarely been simultane-
ously considered.
Serotiny is variable not only among closely related species
( Keeley and Zedler, 1998 ; Lamont et al., 1991 ; He et al., 2012 )
but also within and among populations of a single species ( Muir
and Lotan, 1985 ; Schoennagel et al., 2003 ; Tapias et al., 2004 ).
There is evidence that serotiny increases with the frequency of
crown fi res ( Gauthier et al., 1996 ; Radeloff et al., 2004 ), and
even a single fi re may increase the population serotiny level
given enough variability of the trait in the population ( Muir and
Lotan, 1985 ; Goubitz et al., 2004 ). However, variability in the
strength of the fi re–serotiny relationship remains unexplored.
In addition, there is a lack of information on the within-popula-
tion variability in serotiny. Such variability may have profound
implications for dynamic processes. Crown fi res could act as a
selective force favoring the establishment of the offspring of
serotinous individuals ( Givnish, 1981 ). As far as we know, there
are no explicit measures of fi tness advantages of serotinous in-
dividuals. However, the evidence of genetic control of serotiny
( Perry and Lotan, 1979 ; Tapias et al., 2004 ), together with the
higher serotiny in recurrently burnt ecosystems points toward
fi re as a selective force.
Within-population variability in serotiny can be spatially
structured at different scales ( Tinker et al., 1994 ). In fact,
1 Manuscript received 19 May 2013; revision accepted 11 September
2013.
The authors thank Katharina Budde, Mario Zabal-Aguirre, Diana Turrión
and Jordi Chofre for fi eld assistance. This study was fi nanced by the following
projects supported by the Spanish government: VAMPIRO (CGL2008-
05289-C02-01/02), LinkTree (ERAnet-BiodivERsA: EUI2008-03713 and
EUI2008-03721), VIRRA (CGL2009-12048/BOS), SOBACO (CGL2011-
29585-C02), and TREVOL (CGL2012-39938-C02-01). CIDE is a joint
institute of the Spanish Research Council (CSIC), the Generalitat
Valenciana , and the University of Valencia. The authors have no confl ict of
interest to declare.
4 Author for correspondence (e-mail: juli.g.pausas@uv.es)
doi:10.3732/ajb.1300182
F IRE STRUCTURES PINE SEROTINY AT DIFFERENT SCALES 1
A NA H ERNÁNDEZ-SERRANO 2 , M IGUEL V ERDÚ 2 , S ANTIAGO C. GONZÁLEZ-MARTÍNEZ 3 ,
AND J ULI G. PAUSAS 2,4
2 CIDE-CSIC, Ctra. Nàquera Km. 4.5 (IVIA campus) 46113 Montcada, Valencia, Spain; and
3 CIFOR-INIA, Ctra. A Coruña Km
7.5 28040 Madrid, Spain
• Premise of the study: Serotiny (delayed seed release with the consequent accumulation of a canopy seedbank) confers fi tness
benefi ts in environments with crown-fi re regimes. Thus, we predicted that serotiny level should be higher in populations recur-
rently subjected to crown-fi res than in populations where crown-fi res are rare. In addition, under a high frequency of fi res, space
and resources are recurrently available, permitting recruitment around each mother to follow the seed rain shadow. Thus, we
also predicted spatial aggregation of serotiny within populations.
• Methods: We compared serotiny, considering both the proportion and the age of serotinous cones, in populations living in
contrasting fi re regimes for two iconic Mediterranean pine species ( Pinus halepensis , P. pinaster ). We framed our results by
quantitatively comparing the strength of the fi re–serotiny relationship with previous studies worldwide.
• Key results: For the two species, populations living under high crown-fi re recurrence regimes had a higher serotiny level than
those populations where the recurrence of crown-fi res was low. For P. halepensis (the species with higher serotiny), popula-
tions in high fi re recurrence regimes had higher fi ne-scale spatial aggregation of serotiny than those inhabiting low fi re recur-
rence systems. The strength of the observed fi re–serotiny relationship in P. halepensis is among the highest in published
literature.
• Conclusions: Fire regime shapes serotiny level among populations, and in populations with high serotiny, recurrent fi res main-
tain a signifi cant spatial structure for this trait. Consequently, fi re has long-term evolutionary implications at different scales,
emphasizing its prominent role in shaping the ecology of pines.
Key words: fi re ecology; Pinaceae; Pinus halepensis ; Pinus pinaster ; seed bank; serotiny; spatial structure.
http://www.amjbot.org/cgi/doi/10.3732/ajb.1300182The latest version is at
AJB Advance Article published on November 11, 2013, as 10.3732/ajb.1300182.
Copyright 2013 by the Botanical Society of America

2AMERICAN JOURNAL OF BOTANY [Vol. 100
under these conditions, most regeneration events are driven by fi re (HiFi
populations). The remaining six populations were located inland at higher
altitudes and subject to subhumid climate, where crown-fi res are rare and
most regeneration events are independent of fi re (LoFi populations) ( Verdú
and Pausas, 2007 ). In the study area, fi re is strongly linked to climatic condi-
tions, specifically to drought ( Pausas, 2004 ). Furthermore, recent fi re history
information ( Pausas, 2004 ; Pausas and Fernández-Muñoz, 2012 ) shows that
more than 50% of the study area at <800 m a.s.l. (HiFi conditions) burned at
least once during the 1978–2001 period, while for >800 m a.s.l. (LoFi condi-
tions), the proportion was about 15% ( Abdel Malak and Pausas, 2006 ). That
is, despite the lack of long-term fi re statistics for the specifi c study sites, there
is strong evidence that the fi re interval is much shorter in HiFi areas than in
LoFi. In the study area, fertility is strongly related to nature of the bedrock
type (siliceous bedrocks are typically nutrient-poorer than calcareous bed-
rocks; Ojeda et al., 2010 ); all P. pinaster populations were on siliceous soils,
while P. halepensis included sites in siliceous and in calcareous soils in both
HiFi and LoFi conditions (Appendix S1).
Sampling — We selected and geo-referenced 40 to 67 individuals in each
population including a wide range of distances between trees (Appendix S1)
but avoiding subcanopy trees and trees with diameters less than 10 cm. There
were no differences in mean tree diameter between HiFi and LoFi populations
for any of the species ( P. halepensis : p = 0.4; P. pinaster : p = 0.1; linear mixed
models considering population as random factor; Appendix S1). Serotiny was
estimated considering both the proportion of serotinous cones and the cone
age. In each individual, we fi rst searched for the oldest serotinous cone. The
age of each cone was estimated by counting the number of internodes from
the tip to the location of the cone. Then we counted and dated serotinous
(closed) and nonserotinous (open or partially open) cones in two opposite
branches belonging to the upper third of the canopy and in two opposite
branches contained in the second third of the canopy. For each of these
branches, we recorded the number of cones in each cone cohort with the help
of binoculars. For P. pinaster , we also included the main trunk as this species
often retains an important fraction of serotinous cones on the trunk. Because
changes in humidity might close open cones, serotiny was assessed during
dry seasons. Serotiny level for each tree was estimated as the number of
closed cones (those that have remained closed after maturation) with respect
to the total number (open and closed) (see Tapias et al., 2001 for similar pro-
cedure). The most recent cone cohort (brown cones) was omitted because it is
impossible to assess whether they will become serotinous or not. A subset of
the serotiny data obtained in P. pinaster was also used for an association ge-
netics study (see Budde et al., 2013 ).
Statistical analyses — The variability within and among populations in the
proportion of serotinous cones was estimated by analysis of deviance using a
binomial error distribution. We compared the two species and the two fi re
regimes (HiFi vs. LoFi) on the two components of serotiny, the proportion of
closed cones and the age of the oldest cone, by means of generalized linear
mixed models with population as a random factor (GLMM). Given that se-
rotiny could also vary with tree size, we included tree diameter as a covariate
in the model. A binomial distribution of errors was used in the case of the
proportion of closed cones and a Poisson distribution for the age of the oldest
cones. Finally, to have a more integrated model of serotiny, we combined the
two serotiny components in a weighted linear mixed model in which the per-
centage of serotinous cones was used as dependent variable and cone age as
weight. All these analyses were repeated using mean age of serotinous cones
instead of their maximum age; because the results did not change, they are not
reported here. All the models were fi tted using the lme4 package for R ( Bates
et al., 2011 ).
For each population, we tested the presence of fi ne-scale spatial struc-
ture of the proportion of serotinous cones and maximum serotinous cone
age by computing autocorrelation coeffi cients (Moran’s I index) consider-
ing the pairwise geographical distance between all trees using the program
AutocorQ 2.0 ( Hardy, 2009 ). The regression coeffi cient between the matrix
of pairwise autocorrelation coeffi cients and the matrix of pairwise geo-
graphical distances (computed on a logarithmic scale) is an indicator of the
fi ne-scale spatial structure of serotiny: the more negative the regression co-
effi cient the stronger is the fi ne-scale spatial structure. A desirable property
of this method is that it is not dependent on the classes of distance interval.
AutocorQ tests the statistical signifi cance of the spatial structure by means
of resampling methods. For populations showing a signifi cant spatial struc-
ture of serotiny, we also computed correlograms to reveal the spatial scale
of this autocorrelation.
contrasting fi re regimes can produce different spatial pat-
terns of serotiny through processes related to natural regen-
eration dynamics. Under a high frequency of fi res, space and
resources are recurrently available for recruitment (after
each fi re), potentially permitting the recruitment to be spa-
tially aggregated around each mother (refl ecting the seed
rain shadow). In contrast, in the absence of fi re, recruitment
depends on the availability of safe sites and is not necessar-
ily spatially aggregated, but driven by gap dynamics related
to disturbances other than fi re. Consequently, and assuming
that the level of serotiny is heritable ( Perry and Lotan, 1979 ;
Budde et al., 2013 ) and that there is marked intrapopulation
variation in levels of serotiny, we would expect a more aggre-
gated spatial pattern of matched serotiny levels in recurrently
burnt ecosystems than in systems that rarely burn. In other
words, if there are variations in serotiny among individuals
within the population, and if the majority of seed dispersal is
within short distances (postfi re conditions), a spatial aggrega-
tion is likely. In addition, postrecruitment mortality erases ini-
tial spatial aggregated patterns driven by dispersal ( Steinitz et al.,
2011 ); thus, the lower the recurrence of fi re the less the aggre-
gation pattern of serotiny is expected.
Our hypothesis was that contrasted fi re regimes should pro-
duce different spatial patterns of serotiny for both within and
between populations. These differences are predicted to in-
crease with the serotiny level of the species. Specifi cally, we
expected higher serotiny levels and more spatially aggregated
patterns of serotiny in systems with high crown-fi re recur-
rence compared with populations where recruitment is inde-
pendent of fi re. We tested these predictions by comparing
serotiny patterns in populations in contrasting fi re regimes for
two Mediterranean nonresprouting pine species that produce
serotinous cones ( Pinus halepensis , P. pinaster ). In addition,
we quantitatively compare the strength of the fi re–serotiny re-
lationship with previous studies performed in different pine
species worldwide, to better frame our results in the current
knowledge.
MATERIALS AND METHODS
Study species — Pinus halepensis Mill. and Pinus pinaster Aiton (Pinaceae)
are two widespread Mediterranean pines that do not resprout and have seroti-
nous cones. They typically live in crown-fi re ecosystems where all individuals
die after fi re. They both have serotinous cones that are opened by the heat of fi re
(pyriscent cones), although the proportion of these cones is variable depending
on the population (ranging from 2 to 82% in P. pinaster and 40 to 80% in
P. halepensis ; Tapias et al., 2004 ). In the absence of fi re, a proportion of the
serotinous cones opens during dry and hot summer days. Both species are pre-
cocious fl owering trees with age at fi rst fl owering recorded as 4 to 10 yr old in
P. pinaster and 4 to 8 yr old in P. halepensis ( Tapias et al., 2004 ). Pinus pinaster
is considered a longer-lived species than P. halepensis (ca. 300 vs. 250 yr;
Tapias et al., 2004 ). From the biogeographic point of view, P. halepensis occurs
in the Iberian Peninsula as a result of a relatively recent colonization that was
accompanied by depletion of its genetic diversity, while P. pinaster has been
present for much longer and has high levels of genetic diversity ( Bucci et al.,
2007 ; Grivet et al., 2009 , 2011 ).
Study sites — Eight populations of P. halepensis and seven of P. pinaster
were selected in the eastern Iberian Peninsula (Appendix S1, see Supplemen-
tal Data with the online version of this article), an area with typical Mediter-
ranean climate ( Pausas, 2004 ). Straight-line distances between populations
ranged from 11.7 to 149.3 km for P. halepensis , and from 22.5 to 127.8 km
for P. pinaster . Populations of each species fell in two contrasting environ-
ments with different fi re regimes. Nine populations were located in warm and
dry coastal areas (<800 m a.s.l.) where crown-fi res are historically frequent;

3
December 2013] HERNÁNDEZ-SERRANO ET AL.—FIRE STRUCTURES PINE SEROTINY
RESULTS
The variability in the proportion of serotinous cones was
higher within (62% in P. halepensis , and 54% in P. pinaster ) than
among populations (38 and 46%, respectively; deviance analy-
sis). For the two species, HiFi populations had a higher propor-
tion of serotinous cones than LoFi populations did ( Tables 1, 2a ;
Fig. 1 ). Pinus halepensis had a higher proportion of serotinous
cones than P. pinaster did in both HiFi and LoFi ( Tables 1, 2a ;
Fig. 1 ). The maximum age of the serotinous cones was signifi cantly
higher in P. pinaster , and for the two species, it was much higher
in HiFi populations ( Tables 1, 2b ; Fig. 2 ; online Appendix S3).
To compare the strength between fi re and pine serotiny with previous studies,
we searched in the literature for papers addressing this question in which a Pearson
correlation coeffi cient ( r ) between fi re regime and serotiny could be extracted di-
rectly or through the summary statistics provided in the paper. Most of these stud-
ies reported serotiny as the proportion of serotinous cones per tree or as proportion
of serotinous trees per site (Appendix S2, see online Supplemental Data). Follow-
ing traditional meta-analytic procedures, r was standardized as effect size using the
Fisher transformation [ Z ( r )], and the variance associated with the effect size was
calculated as 1/( N − 3), with N the number of plots where serotiny was estimated
( Hedges and Olkin, 1985 ; Rosenthal, 1991 ). To obtain the overall effect size across
studies, we ran a Bayesian meta-analysis by fi tting a GLMM with the help of the
MCMCglmm package for the program R ( Hadfi eld, 2010 ) and using the default
initiation options. To account for pseudoreplication due to the use of the same spe-
cies in different studies, we included species as a random factor.
T ABLE 1. Summary of the variables describing serotiny (proportion of serotinous cones, maximum and mean cone age) and the spatial pattern of serotiny
in each population of the two Pinus species. The spatial pattern of serotiny is shown as the slope and p -value of the relationship between the phenotypic
autocorrelation coeffi cient (Moran’s I ) and the pairwise geographical distance (on logarithmic scale) for the proportion of serotinous cones and the
maximum age of the cones. Details of the populations are provided in Appendix S1.
Spatial pattern
Population
Serotiny Serotinous cones Maximum age
Serotinous cones (%) Maximum cone age (yr) Mean cone age (yr) Slope p Slope p
P. halepensis
HH1 54.21 7.17 3.34 −0.236 <0.0001 −0.004 0.7023
HH2 38.10 8.17 3.78 −0.260 <0.0001 −0.244 <0.0001
HH3 52.94 5.75 2.83 −0.031 0.201 −0.033 0.217
HH4 52.70 6.45 2.91 −0.052 0.027 −0.040 0.067
HH5 61.35 6.12 3.32 −0.099 0.021 −0.059 0.135
HL6 28.93 3.88 2.51 0.025 0.335 0.032 0.147
HL7 26.70 3.96 2.53 −0.020 0.320 0.019 0.323
HL8 23.88 3.77 2.60 0.026 0.511 0.021 0.612
P. pinaster
PH9 58.19 8.34 3.54 −0.008 0.564 0.004 0.940
PH10 28.84 8.34 3.86 0.020 0.652 −0.002 0.748
PH11 29.28 11.22 4.67 0.010 0.726 0.025 0.764
PH12 14.11 5.19 3.22 −0.002 0.369 0.033 0.331
PL13 5.32 3.00 2.55 0.012 0.962 0.019 0.832
PL14 6.41 3.25 3.08 −0.030 0.612 0.021 0.696
PL15 22.07 4.17 2.95 −0.037 0.944 0.001 0.878
T ABLE 2. Summary of the GLMM for the different measures of serotiny including species (Ph: Pinus halepensis ; Pp: Pinus pinaster ), tree diameter, fi re
regime (HiFi and LoFi) and the species-fi re interaction (sequential addition of variables). Population was included as a random factor. Degree of
freedom, Akaike information criterion (AIC), χ 2 and the associated p value of each step in the models are presented. The rightmost column provides
the estimated parameters for the fi xed effects of the fi nal model.
Model df AIC χ 2 p Estimate
a) Proportion of serotinous cones
Null 2 3381.8 0.037 [intercept]
Species 3 3276.0 107.74 <0.0001 0 [Ph], −0.462 [Pp]
+ Diameter 4 3264.1 13.91 0.0001917 −0.010 · diameter
+ Fire 5 3258.1 8.06 0.0045156 0 [HiFi], −1.399 [LoFi]
+ Species × fi re 6 3255.8 4.29 0.0382815 0 [Ph-HiFi], 0.224 [Pp-LoFi]
b) Maximum age of cones
Null 2 874.33 1.711 [intercept]
Species 3 845.85 30.48 <0.0001 0 [Ph], 0.286 [Pp]
+ Diameter 4 844.86 3.00 0.08 0.005 · diameter
+ Fire 5 831.92 14.93 0.0001113 0 [HiFi], −0.587 [LoFi]
+ Species × fi re 6 825.90 8.02 0.0046173 0 [Ph-HiFi], −0.275 [Pp-LoFi]
c) Proportion of serotinous cones weighted by maximum age
Null 2 21 962 0.108 [intercept]
Species 3 21 665 298.77 <0.0001 0 [Ph], −0.318 [Pp]
+ Diameter 4 21 578 89.71 <0.0001 −0.01· diameter
+ Fire 5 21 572 7.41 0.006487 0 [HiFi], −1.058 [LoFi]
+ Species × fi re 6 21 495
Q1 78.70 <0.0001 0 [Ph-HiFi], 0.428 [Pp-LoFi]

4AMERICAN JOURNAL OF BOTANY [Vol. 100
four of fi ve populations (all except HH3, Serra Calderona;
Table 1 ). The autocorrelation for maximum age of the cones
in HiFi populations of P. halepensis was also negative in all
cases but only signifi cant in one population (HH2, Cabanes)
and marginally signifi cant in another (HH4, Eslida; Table 1 ).
In contrast, none of the LoFi populations of P. halepensis
showed any signifi cant autocorrelation pattern either in the
proportion of serotinous cones or in the maximum age of the
cones. That is, serotiny of HiFi populations in P. halepensis
was more spatially structured than in LoFi populations, espe-
cially in relation to the proportion of the serotinous cones
(mean slopes for HiFi = −0.135 and LoFi = 0.011; t = 2.82, df =
4.19, p = 0.04), and to a lesser extent for the maximum age of
When serotiny was considered as the proportion of serotinous
cones weighted by the maximum age of the cones (for each
tree), it was also different between species (higher for P. halepen-
sis ) and between fi re regimes (higher in HiFi populations), with
the differences between fi re regimes greater for P. halepensis
( Table 2c ). Similar results were obtained when considering mean
cone age of each tree instead of the maximum cone age (not
shown; Table 1 ), as the two measures were highly correlated ( r =
0.85, p < 0.0001).
All HiFi populations of P. halepensis showed a decreasing
trend in autocorrelation of the proportion of serotinous cones
with the distance between trees (negative slopes; Table 1 ).
The strength of the relationship was variable and signifi cant in
Fig. 1. Proportion of serotinous cones across age cohorts on trees for each of the studied populations and for the two species ( Pinus halepensis and
P. pinaster ). Gray boxes refer to populations with dominant fi re-dependent recruitment (HiFi populations), white boxes to populations for which the recruit-
ment is not dependent on fi re (LoFi populations). Boxplots indicate the median (horizontal line), the fi rst and third quartiles (box), the range that excludes
outliers (whiskers), and the outliers (circles).

5
December 2013] HERNÁNDEZ-SERRANO ET AL.—FIRE STRUCTURES PINE SEROTINY
The strength of the relationship between fi re and serotiny in
different pine studies ranges from r values of 0.09 to 0.98
( Table 3 ; Appendix S2), with a signifi cant overall effect across
all studies [ r = 0.737; Z ( r ) = 0.945 with 95% credible interval =
0.601−1.351, p = 0.002]. Our study sites show a higher fi re-
serotiny strength than the mean overall effect for P. halepen-
sis ( r = 0.94), and lower for P. pinaster ( r = 0.68).
DISCUSSION
The reviewed studies analyzing the relationship between
pine serotiny and fi re regime showed a strong positive overall
effect. Compared with pine studies worldwide, the strength of
this relationship in our study is intermediate for P. pinaster
and among the highest for P. halepensis ( Table 3 ; Appendix S2).
Indeed, for both species serotiny levels were greater in areas
affected by frequent crown fi res (HiFi) than areas where
crown-fi res were rare (LoFi), suggesting that recurrent crown-
fi res increase serotiny at the population level. This was true
whether we considered serotiny as the proportion of seroti-
nous cones or as the age of the cones stored. These results
suggest that in fi re-prone environments serotiny favors the re-
cruitment and persistence of P. halepensis and P. pinaster
populations ( Gauthier et al., 1996 ; Keeley and Zedler, 1998 ;
Tapias et al., 2004 ). This contrasting pattern of a plant trait
under different fi re regime adds further evidence to the emerg-
ing view that fi re shapes intraspecifi c variability of multiple
traits and generates phenotypic variability between popula-
tions ( Keeley et al., 2011 ; Pausas and Schwilk, 2012 ). Recent
results have also shown a tight link between plant variability
and fi re regime in other fi re-related traits like fl ammability
and postfi re germination ( Gómez-González et al., 2011 ; Moreira
et al., 2012 ; Pausas et al., 2012 ).
The different serotiny level between different fi re regimes
occurs despite the large variability in serotiny within popula-
tion. In fact, all individuals of P. halepensis and most of
P. pinaster (79%) had serotinous cones, even those living in
low fi re frequency environments ( Fig. 1 ). There were no single
P. halepensis tree having either all cones closed or all cones
opened. That is, for these species, serotiny is clearly a quanti-
tative trait, and the differences between fi re regimes are in the
frequency distribution of serotinous cones. This contrast with
other pine species in which the extremes phenotypes (seroti-
nous and nonserotinous trees) are dominant and the serotiny
level varies more among sites than among individuals (e.g.,
Muir and Lotan, 1985 ). The observed large individual vari-
ability, even within site and fi re regime, may be the consequence
of some variability in the selective regime. This phenotypic
variability might allow P. halepensis and P. pinaster to recruit
in a variety of conditions, as crown-fi res might sporadically
occur in the low fi re frequency environment, and fi re-independent
gaps may also occur in the high fi re frequency environment.
The different serotiny levels between the two Mediterranean
pines may refl ect differences in strategies to cope with environ-
mental unpredictability. Pinus halepensis has a high proportion
of serotinous cones mostly allocated to a few recent cohorts; in
contrast, P. pinaster has a lower proportion of serotinous cones,
but they are distributed over a longer time ( Fig. 2 , Appendix S3).
In our populations, the maximum age of the serotinous cones
was 17 yr for P. halepensis and 23 yr for P. pinaster . This dif-
ferent temporal strategy between the species matches with the
the cones (mean slopes for HiFi = −0.076 and LoFi = 0.024,
t = 2.16, df = 4.07, p = 0.09). The spatial scale at which auto-
correlation was signifi cant for HiFi populations of P. halepen-
sis was always less than 200 m for both the proportion of
serotinous cones and the maximum age of closed cones ( Fig. 3 ).
For P. pinaster , none of the populations showed a signifi cant
spatial pattern of serotiny, and there were no difference in spa-
tial structure between fi re regimes, neither for the proportion of
serotinous cones nor for the maximum age of the cones ( Table 1 ).
Fig. 2. Frequency distribution of trees in relation to their maximum
closed cone age for the two species studied ( Pinus halepensis and P. pinaster ).
Dotted lines are means for species; means for each population are given in
Table 1 . The gray pattern in the stacked bars corresponds to the proportion
of trees in HiFi populations (i.e., with dominant fi re-dependent recruit-
ment), the white to the proportion of trees in LoFi populations (i.e., with
dominant fi re-independent recruitment). The frequency distribution of
cone age is provided in Appendix S3.

6AMERICAN JOURNAL OF BOTANY [Vol. 100
Despite this cost, retaining cones for longer, as in P. pinaster ,
ensures postfi re recruitment even after consecutive years of low
crop production. In addition, it might also ensure a more geneti-
cally diverse recruitment, although this effect is not observed in
serotinous cones of Banksia ( Barrett et al., 2005 ; Ayre et al., 2010 ).
observed lower seed longevity in P. halepensis than in P. pinaster
(20 and 30 yr respectively; Catalán, 1991 ; Tapias et al., 2004 ).
There is evidence that seed germinability decreases with the age
of the serotinous cone ( Cowling and Lamont, 1985 ; Daskalakou
and Thanos, 1996 ; Barrett et al., 2005 ; Crawford et al., 2011 ).
Fig. 3. Autocorrelation diagrams of serotiny in HiFi populations of Pinus halepensis (only sites with a signifi cant trend are shown; Table 2 ). Autocor-
relogram refers to the proportion of serotinous cones (sites HH1, HH2, HH4, HH5) and to the maximum cone age (site HH2). Large colored dots indicate
a signifi cant Moran index, which was computed for up to one-half the maximum lag distances in each population.

7
December 2013] HERNÁNDEZ-SERRANO ET AL.—FIRE STRUCTURES PINE SEROTINY
LITERATURE CITED
A BDEL MALAK , D . , AND J . G . P AUSAS . 2006 . Fire regime and post-fi re
Normalized Difference Vegetation Index changes in the eastern
Iberian peninsula (Mediterranean basin). International Journal of
Wildland Fire 15 : 407 – 413 .
A YRE , D. , E. O’BRIEN , K . O TTEWELL , AND R . W HELAN . 2010 . The ac-
cumulation of genetic diversity within a canopy-stored seed bank.
Molecular Ecology 19 : 2640 – 2650 .
B ARRETT , L . G . , T . H E , B. B. LAMONT , AND S. L. KRAUSS . 2005 . Temporal
patterns of genetic variation across a 9-year-old aerial seed bank of
the shrub Banksia hookeriana (Proteaceae). Molecular Ecology 14 :
4169 – 4179 .
B ATES , D . , M . MAECHLER , AND B . BOLKER . 2011 . lme4: Linear mixed-
effects models using S4 classes. R package version 0.999375-42.
Website http://CRAN.R-project.org/package=lme4 .
B ORCHERT , M. 1985 . Serotiny and cone-habit variation in populations of
Pinus coult eri (Pinaceae) in the southern Coast Ranges of California.
Madroño 32 : 29 – 48 .
B UCCI , G. , S. C. GONZÁLEZ-MARTÍNEZ , G . L E-PROVOST , C . P LOMION , M.
M. RIBEIRO , F . S EBASTIANI , R . A LÍA , AND G . G . V ENDRAMIN . 2007 .
Range-wide phylogeography and gene zones in Pinus pinaster Ait.
revealed by chloroplast microsatellite markers. Molecular Ecology
16 : 2137 – 2153 .
B UDDE , K . , M . H EUERTZ , A . H ERNÁNDEZ-SERRANO , J. G. PAUSAS , G. G.
V ENDRAMIN , M . V ERDÚ , AND S . C . G ONZÁLEZ-MARTÍNEZ . 2013 . In situ
genetic association for serotiny, a fi re-related trait, in Mediterranean
maritime pine ( Pinus pinaster Aiton). New Phytologist .
C ATALÁN , G. 1991 . Semillas de árboles y arbustos forestales. Ministerio
de Agricultura Pesca y Alimentación, Instituto de la Conservación de
la Naturaleza (ICONA), Madrid.
C LIMENT , J . , R . T APIAS , J. A. PARDOS , AND L . G IL . 2004 . Fire adaptations
in the Canary Islands pine ( Pinus canariensis ). Plant Ecology 171 :
185 – 196 .
C OWLING , R. M. , AND B. B. LAMONT . 1985 . Variation in serotiny of
three Western Australian Banksia species along a climatic gradient.
Australian Journal of Ecology 10 : 345 – 350 .
C RAWFORD , A. , J. PLUMMER , R . P ROBERT , AND K. STEADMAN . 2011 . The in-
fl uence of cone age on the relative longevity of Banksia seeds. Annals
of Botany 107 : 303 – 309 .
D ASKALAKOU , E. N. , AND C. A. THANOS . 1996 . Aleppo pine ( Pinus
halepensis ) postfi re regeneration: The role of canopy and soil seed
banks. International Journal of Wildland Fire 6 : 59 – 66 .
D E LUCAS , A. I. , S. C. GONZÁLEZ-MARTÍNEZ , G . G . V ENDRAMIN , E . HIDALGO ,
AND M . H EUERTZ . 2009 . Spatial genetic structure in continuous and
fragmented populations of Pinus pinaster Aiton. Molecular Ecology
18 : 4564 – 4576 .
In contrast, allocating serotinous cones to the most recent
crops maintains high seed viability levels but might jeopar-
dise reproduction under consecutive stressful years; that is,
consecutive years with very low seed production may strongly
reduce the canopy seed bank and imperil postfi re regeneration.
The contrasting fi re regimes have implications, not only in
favoring serotiny, but also on the regeneration dynamics of
the populations, leading to differences in the spatial struc-
ture of the recruitment. Provided that serotiny is heritable
( Perry and Lotan, 1979 ; Budde et al., 2013 ), the phenotypic
spatial structure generated by fi re may be a consequence of
the spatial genetic structure of the population. Indeed, there
is evidence of spatial genetic structure for serotinous species
in fi re-prone ecosystems ( Ayre et al., 2010 ). In HiFi environ-
ments, recruitment always occurs at times of high availabil-
ity of space and resources (i.e., after fi re) that allows the
offspring of a tree to recruit close to their mother tree. This
aggregated spatial pattern of closely related individuals is
expected to produce a phenotypic spatial aggregation of her-
itable traits. In LoFi environments, regeneration is driven by
gap dynamics, and thus spatial aggregation is not expected.
In addition, tree mortality should erase the initial aggregated
patterns of recruitment with time since disturbance ( Steinitz
et al., 2011 ). We found that serotiny was spatially autocor-
related in most HiFi populations of P. halepensis but never
in the less serotinous P. pinaster . Pinus halepensis was more
serotinous but it is also the species with a greater seed crop
( Tapias et al., 2001 ), probably leading to higher postfi re seed
release, that in turn may produce stronger signatures of the
initial seed shadow in the adult population. However, other
differences in the biology of these two species might explain
this differential pattern.
Our results suggest that fi re can generate signifi cant fi ne-
scale spatial genetic structure even in outcrossing forest trees
with long-distance dispersal where this structure is typically
weak (for an example in Mediterranean pines, see reviews:
Vekemans and Hardy, 2004 ; De Lucas et al., 2009 ). The hy-
pothesis of differential spatial structure driven by different fi re
regime deserves further testing using fi ne-scale population ge-
netic analyses, which should enable a better understanding of
the demographic and selective roles of fi re in shaping plant
populations.
T ABLE 3. Relative strength of the relationship between serotiny of Pinus species and fi re regime for different studies (see Appendix S2 for details) expressed
as Pearson correlation ( r ) and standardized effect size [Fisher transformation, Z ( r )]. N refers to the number of plots for which the relationship was
studied. Studies may be repeated in different rows if they use different serotiny variables (Serotiny measure; see Appendix S2 for further details). Data
sorted by the relative strength between serotiny and fi re (decreasing order).
Rank Species r N Z ( r ) Serotiny measure Reference
1 P. banksiana 0.976 17 2.205 % serotinous trees Radeloff et al. (2004)
2 P. halepensis 0.941 8 1.746 % closed cones Present study
3 P. contorta 0.905 20 1.499 % of serotinous cones (mean) Muir and Lotan (1985)
4 P. coulteri 0.879 12 1.373 % of serotinous trees Borchert et al. (1985)
5 P. pinaster 0.850 23 1.256 % closed cones per tree Tapias et al. (2004)
6 P. halepensis 0.800 7 1.099 % closed cones (mean) Goubitz et al. (2004)
7 P. pinaster 0.710 23 0.887 % closed cones (mean) Tapias et al. (2004)
8 P. canariensis 0.690 8 0.848 % serotinous trees Climent et al. (2004)
9 P. pinaster 0.681 7 0.831 % closed cones Present study
10 P. contorta subsp. latifolia 0.585 25 0.670 % serotinous trees (young) Schoennagel et al. (2003)
11 P. rigida 0.584 166 0.669 % serotinous trees Givnish (1981)
12 P. banksiana 0.284 30 0.284 % serotinous cones Gauthier et al. (1996)
13 P. banksiana 0.189 30 0.191 % serotinous and quasi-serotinous trees Gauthier et al. (1996)
14 P. contorta subsp. latifolia 0.092 25 0.092 % serotinous trees (old) Schoennagel et al. (2003)

8AMERICAN JOURNAL OF BOTANY [Vol. 100
G AUTHIER , S . , Y . B ERGERON , AND J . P . S IMON . 1996 . Effects of fi re regime
on the serotiny level of jack pine. Journal of Ecology 84 : 539 – 548 .
G IVNISH , T. J. 1981 . Serotiny, geography and fi re in the pine barrens of
New Jersey. Evolution; International Journal of Organic Evolution
35 : 101 – 123 .
G ÓMEZ-GONZÁLEZ , S . , C . T ORRES-DÍAZ , C . B USTOS-SCHINDLER , AND E.
G IANOLI . 2011 . Anthropogenic fi re drives the evolution of seed
traits. Proceedings of the National Academy of Sciences, USA 108 :
18743 – 18747 .
G OUBITZ , S . , R . N ATHAN , R . R OITENBERG , A . S HMIDA , AND G . N E’EMAN . 2004 .
Canopy seed bank structure in relation to: fi re, tree size and density.
Plant Ecology 173 : 191 – 201 .
G RIVET , D . , F . S EBASTIANI , R . ALÍA , T . B ATAILLON , S. TORRE , M . Z ABAL-
AGUIRRE , G. G. VENDRAMIN , AND S. C. GONZÁLEZ-MARTÍNEZ . 2011 .
Molecular footprints of local adaptation in two Mediterranean coni-
fers. Molecular Biology and Evolution 28 : 101 .
G RIVET , D . , F . S EBASTIANI , S. C. GONZÁLEZ-MARTÍNEZ , AND G. G. VENDRAMIN .
2009 . Patterns of polymorphism resulting from long-range coloniza-
tion in the Mediterranean conifer Aleppo pine. New Phytologist 184 :
1016 – 1028 .
H ADFIELD , J. D. 2010 . MCMC methods for multi-response general-
ized linear mixed models: The MCMCglmm R package. Journal of
Statistical Software 33 : 1 – 22 .
H ARDY , O. J. 2009 . AutocorQ. A program to characterise and test the
spatial autocorrelation of quantitative traits. http://ebe.ulb.ac.be/ebe/
Software.html .
H EDGES , L. V. , AND I . O LKIN . 1985 . Statistical methods for meta-analysis.
Academic Press, Orlando, Florida, USA.
H E , T. , J. G. PAUSAS , C. M. BELCHER , D. W. SCHWILK , AND B. B. LAMONT .
2012 . Fire-adapted traits of Pinus arose in the fi ery Cretaceous . New
Phytologist 194 : 751 – 759 .
K EELEY , J. E. , J. G. PAUSAS , P . W . R UNDEL , W. J. BOND , AND R. A. BRADSTOCK .
2011 . Fire as an evolutionary pressure shaping plant traits. Trends in
Plant Science 16 : 406 – 411 .
K EELEY , J. E. , AND P. H. ZEDLER . 1998 . Evolution of life histories in Pinus.
In D. M. Richardson [ed.], Ecology and biogeography of Pinus , 219–
250. Cambridge University Press, Cambridge, UK.
L AMONT , B. B. , AND N. J. ENRIGHT . 2000 . Adaptive advantages of aerial
seed banks. Plant Species Biology 15 : 157 – 166 .
L AMONT , B. B. , D. C. LE MAITRE , R. M. COWLING , AND N . J . E NRIGHT .
1991 . Canopy seed storage in woody plants. Botanical Review 57 :
277 – 317 .
M IDGLEY , J. J. 2000 . What are the relative costs, limits and correlates of
increased degree of serotiny? Austral Ecology 25 : 65 – 68 .
M OREIRA , B . , Ç . T AVSANOGLU , AND J. G. PAUSAS . 2012 . Local versus re-
gional intraspecifi c variability in regeneration traits. Oecologia 168 :
671 – 677 .
M UIR , P. S. , AND J. E. LOTAN . 1985 . Disturbance history and serotiny in
Pinus contorta in Western Montana. Ecology 66 : 1658 – 1668 .
O JEDA , F . , J . G . P AUSAS , AND M . V ERDÚ . 2010 . Soil shapes community
structure through fi re. Oecologia 163 : 729 – 735 .
P AUSAS , J. G. 2004 . Changes in fi re and climate in the eastern Iberian
Peninsula (Mediterranean basin). Climatic Change 63 : 337 – 350 .
P AUSAS , J. G. , G. A. ALESSIO , B . MOREIRA , AND G . C ORCOBADO . 2012 . Fires
enhance fl ammability in Ulex parvifl orus. New Phytologist 193 :
18 – 23 .
P AUSAS , J. G. , R. A. BRADSTOCK , D. A. KEITH , J . E . K EELEY , AND GCTE_
FIRE_NETWORK . 2004 . Plant functional traits in relation to fi re in
crown-fi re ecosystems. Ecology 85 : 1085 – 1100 .
P AUSAS , J. G. , AND S . F ERNÁNDEZ-MUÑOZ . 2012 . Fire regime changes in the
Western Mediterranean Basin: From fuel-limited to drought-driven
fi re regime. Climatic Change 110 : 215 – 226 .
P AUSAS , J. G. , AND J. E. KEELEY . 2009 . A burning story: The role of fi re in
the history of life. Bioscience 59 : 593 – 601 .
P AUSAS , J. G. , AND D. W. SCHWILK . 2012 . Fire and plant evolution. New
Phytologist 193 : 301 – 303 .
P ERRY , D. A. , AND J . E . L OTAN . 1979 . A model of fi re selection for serotiny
in lodgepole pine. Evolution 33 : 958 – 968 .
R ADELOFF , V. C. , D. J. MLADENOFF , R . P . G URIES , AND M . S . B OYCE . 2004 .
Spatial patterns of cone serotiny in Pinus banksiana in relation to differ-
ences in fi re regimes. Forest Ecology and Management 189 : 133 – 141 .
R OSENTHAL , R. 1991 . Meta-analytic procedures for social research. Sage,
Beverly Hills, California, USA.
S CHOENNAGEL , T. , M. G. TURNER , AND W. H. ROMME . 2003 . The infl u-
ence of fi re interval and serotiny on postfi re lodgepole pine density in
Yellowstone National Park. Ecology 84 : 2967 – 2978 .
S TEINITZ , O . , D . T ROUPIN , G. G. VENDRAMIN , AND R . N ATHAN . 2011 . Genetic
evidence for a Janzen–Connell recruitment pattern in reproductive off-
spring of Pinus halepensis trees. Molecular Ecology 20 : 4152 – 4164 .
T APIAS , R . , J . C LIMENT , J . A . P ARDOS , AND L . G IL . 2004 . Life histories of
Mediterranean pines. Plant Ecology 171 : 53 – 68 .
T APIAS , R . , L . G IL , P . F UENTES-UTRILLA , AND J. A. PARDOS . 2001 . Canopy
seed banks in Mediterranean pines of southeastern Spain: A compari-
son between Pinus halepensis Mill., P. pinaster Ait., P. nigra Arn.
and P. pinea L. Journal of Ecology 89 : 629 – 638 .
T INKER , D. B. , W. H. ROMME , W. W. HARGROVE , R . H . G ARDNER , AND M.
G. TURNER . 1994 . Landscape-scale heterogeneity in lodgepole pine
serotiny . Canadian Journal of Forest Research 24 : 897 – 903 .
V EKEMANS , X . , AND O . J . H ARDY . 2004 . New insights from fi ne-scale spa-
tial genetic structure analyses in plant populations. Molecular Ecology
13 : 921 – 935 .
V ERDÚ , M . , AND J . G . P AUSAS . 2007 . Fire drives phylogenetic clustering in
Mediterranean Basin woody plant communities. Journal of Ecology
95 : 1316 – 1323 .