development of a tree -ring network for the italian peninsula

TREE -RING BULLETIN, Vol. 52, 1992
DEVELOPMENT OF A TREE -RING NETWORK
FOR THE ITALIAN PENINSULA
FRANCO BIONDI
Laboratory of Tree -Ring Research
The University of Arizona
Tucson, AZ 85721
USA
ABSTRACT
This article describes the analysis of tree -ring collections from standing trees of sixteen species
at twenty sites distributed throughout the Italian Peninsula. Visual and numerical crossdating
among ring widths allowed the computation of standard and residual tree -ring chronologies.
Relationships among chronologies were identified by Spearman's coefficient of rank correlation,
using Bonferroni's inequality to adjust significance level. The oldest living tree sampled to date is a
963 -year old palebark pine (Pinus leucodermis Ant.) at Parco del Pollino. Individuals more than
two centuries old were identified at eleven sites for eight species. The tree -ring network so far consists of twenty -two chronologies for nine species at nineteen sites. Seven conifer species account
for ten chronologies and two angiosperm species account for the remaining twelve chronologies.
The most represented species is Fagus sylvatica L., with eleven chronologies distributed over the
entire peninsula and highly correlated with one another. The order of autoregressive models fitted
to the data never exceeded two. In particular, the order of autoregressive models fitted to Fagus sylvatica chronologies decreased with decreasing age of sampled trees. Based on the significant coefficients of rank correlation, residual chronologies of Fagus sylvatica could be separated into northern, central, and southern groups. This points to the existence of broad regions distributed along a
latitudinal gradient, corresponding to large -scale climatic regimes over the Italian Peninsula.
In dieser Arbeit wird die dendrochronologischen Auswertung von 16 Baumarten an 20 über
ganz Italien verteilten Standorten vorgestellt. Visuelles sowie statistisches "cross -dating"
ermöglichte die Berechnung von Standard- und Residualchronologien. Die Beziehungen zwischen
den Chronologien wurden mit Hilfe des Rangkorrelationskoeffizienten nach Spearman beschrieben
und mit der Ungleichung nach Bonferroni auf Signifikanz geprüft. Der in dieser Untersuchung
älteste, datierte lebende Baum ist eine 963 jährige Schlangenhautkiefer (Pinus leucodermis) am
Parco del Pollino. Insgesamt wurden an elf Standorten und von acht Arten einzelne Bäume gefunden, die älter als 200 Jahre sind. Bislang besteht das Chronologien -Netz aus 22 Chronologien von
neun Arten an 19 Standorten. Sieben Nadelbaumarten ergaben zehn und zwei Laubbaumarten die
12 übrigen Chronologien. Dabei ist die Baumart Fagus sylvatica mit elf über die ganze Halbinsel
verteilten und eng untereinander korrelierten Chronologien am häufigsten vertreten. Die Ordnung
der jeweils berechneten autoregressiven Modelle lag niemals höher als zwei. Insbesondere bei
Fagus sylvatica wurde deutlich, daß mit abnehmendem Alter der Bäume auch die Ordnung der
angepaßten Modelle geringer wurde. Anhand der Signifikanz der Rangkorrelationen konnten die
Residualchronologien von Fagus sylvatica in nördliche, mittlere, und südliche Gruppe geteilt werden. Diese Aufteilung deutet auf großräumige Regionen entlang eines Breitengradgradienten hin,
die mit großflächig einheitlichen Klimamustern auf der italienischen Halbinsel übereinstimmen.
Cet article décrit l'analyse d'échantillons provenant de 16 espèces arbres vivants, issus de 20
sites distribués à travers la Péninsule Italienne. Des synchronisations visuelles et numériques ont
permis le calcul de chronologies standards et résiduelles. Les relations entre chronologies ont été
vérifiées par le calcul du coefficient de corrélation de rang de Spearman en utilisant la statistique de
Bonferroni pour ajuster le niveau de signification.Le plus vieil arbre vivant échantillonné est un
Pinus leucodermis Ant. de 963 ans. Des individus de plus de deux siècles ont été trouvés chez huit
espèces provenant de onze sites. Le réseau dendrochronologique est composé de vingt -deux
chronologies provenant de neuf espèces issues de dix -neuf sites. Sept espèces de conifères ont produit dix chronologies, tandis que les douze restantes correspondent à chronologies, tandis que les
douze restantes correspondent à deux espèces d'Angiospermes. Fagus sylvatica L. est `espèce la
16
BIONDI
mieux représentée avec onze chronologies distribuées sur la péninsule entière et fortement correllées entre elles. L'ordre des modèles autorégressifs ajustés aux données n'exède jamais deux. En
particulier, l'ordre des modèles utilisés pour Fagus sylvatica decroit lorsque l'àge des arbres échantillonnes augmente. Sur base des coefficients de corrélation de rang significatifs, les chronologies
résiduelles de Fagus sylvatica pourraient ètre séparées en groupes septentrional, central et méridional. Ceci indique l'existence de larges régions distribuées suivant un gradient latitudinal, correspondant aux régimes climatiques á large échelle couvrant la Péninsule Italienne.
INTRODUCTION
Tree -ring networks have been established not only in semiarid zones but also in humid,
temperate, and cold regions all over the world (Hughes et al. 1982; Jacoby and Hornbeck
1987; Schweingruber et al. 1991). In Italy, a few tree -ring chronologies exist for alpine and
northern regions (Bebber 1990; Brugnoli and Gandolfo 1991; Nola 1988; Schweingruber
1985). To date, published tree -ring chronologies for the Italian peninsula are limited to
conifers growing on a handful of mountain sites (Bräker and Schweingruber 1989; Santini and
Martinelli 1991; Serre- Bachet 1985). The scarcity of Italian tree -ring chronologies does not
reflect a scarcity of forests. Total forest coverage amounts to more than 8.6 million hectares,
about 28% of the entire national territory (Bortolotti 1989). Since most flatland and low -hill
sites are occupied by urban settlements or devoted to agricultural and industrial production,
about 60% of Italy's forests are located in mountainous areas, where human presence is
reduced and the need to prevent soil erosion and landslides is increased.
Proxy records of past environmental changes are not abundant in southern Europe and the
Mediterranean Basin, where dendrochronological studies may provide much needed scientific
information (Serre- Bachet 1992). The development of crossdated tree -ring chronologies is
necessary to the dendroclimatic reconstruction of climatic variation at timescales of decades to
centuries (Fritts 1976). Furthermore, tree -ring analysis provides accurate estimates of plant
longevity, which relates to the multisided questions concerning senescence and death among
plants (Loehle 1988; Molisch 1938). Although Italian forests have been managed for centuries,
old individual trees and forest stands still exist, especially in remote areas (Biondi 1988;
Società Botanica Italiana 1971). This article outlines the development of a network of tree -ring
chronologies for the Italian peninsula using core samples extracted from living trees.
MATERIALS AND METHODS
Sampling sites were selected throughout the Italian Peninsula, between 39.5° and 46.5°
north latitude and between 7° and 16.5° east longitude (Figure 1). Criteria for establishing the
network of tree -ring sites were: (a) homogeneous horizontal distribution over the entire peninsula; (b) broad elevational range, from sea level to timberline; (c) presence of presumably old
trees belonging to species with datable rings (Nardi Berti 1979; Società Botanica Italiana
1971). Field collections took place during the summer months of 1988, 1989, and 1990.
Tree selection focused on single or grouped trees that showed the best combination of old
age and trunk health. Old age was defined by large stem diameter, strongly tapered trunk, large
branches, flat crown top, and dominant position among surrounding trees. Trunk health was
defined by lack of large scars, wood mushrooms, or bark stains. In some cases, vigorous,
locally dominant trees were selected regardless of their age to compare their ring patterns with
those of the old trees. Dead standing trees were also sampled to extend the tree -ring record as
far back in time as possible. Wood cores 5 mm thick were extracted using Swedish increment
Development of a Tree -Ring Network for the Italian Peninsula
17
Figure 1. Locations of collection sites. Site information is given in Table 1.
borers ranging in length from 40 to 75 cm, depending on trunk diameter. Quantitative and
qualitative information on sites (i.e., location, elevation, slope, exposure, stand composition,
and density) and trees (i.e., trunk diameter, height, crown health, and core position) was
recorded in the field and later entered into a computerized database. Whenever possible, information on site history, forest management, climatic regime, pedology, and other factors was
gathered from published and unpublished literature.
All increment cores were glued to wooden mounts, sanded with a belt sander, and polished by hand until the smallest rings were clearly visible. Ring counting provided a preliminary estimate of tree age. Crossdating was ascertained by skeleton plots (Stokes and Smiley
1968) and by visually comparing samples under a binocular microscope. Ring widths were
measured to the nearest 0.01 mm. Crossdating was independently verified by another
researcher and by numerical techniques (Holmes 1983). Ring width was modeled as an exponential function of both a deterministic component, the intrinsic growth trend, Y (Fritts 1976),
and a stochastic component, the autoregressive process, I (Box and Jenkins 1976):
BIONDI
18
wt =
L
e (Y
+
where:
wit = ring width at year t in specimen i.
The autoregressive process was defined as:
lit = El - Op(B)11it + co it
where:
Op(B) = autoregressive operator, B being the backward shift operator and Op the polynomial in
B of order p; p is also the order of the autoregressive process (Box and Jenkins 1976);
coif = white -noise series.
Standard tree -ring chronologies (Figure 2) were developed by removing only the intrinsic
growth trend, as follows:
nt
where:
E (ln wár -
t
i=1
nt
It = average ring index at year t;
In = logarithmic transformation, used to obtain homoscedastic time series (Davis 1986);
yit = intrinsic growth trend at year t in specimen i; it was estimated by a cubic spline with 50%
variance reduction at a frequency of one cycle per 50 years (Cook and Peters 1981);
nt = number of specimens that included year t; to minimize errors, no chronology value was
computed using just a single specimen, hence nt >_ 2 always.
Residual tree -ring chronologies (Figure 3) were computed by removing both components, as
follows:
nt
-wt -
E (B)(ln wit - yát)
i=1
nt
where:
tot = average residual ring -index at year t.
Selection of the best autoregressive model for the data followed guidelines given by Biondi
and Swetnam (1987).
Dendrochronological parameters, such as mean sensitivity and signal -to -noise ratio (Fritts
1976), were computed for the standard tree -ring chronologies. Such parameters are based on
ring -width measurements, and may be used to quantify crossdating quality (Fritts and Shatz
1975). Other indicators of crossdating quality are the percentage of variance explained by the
first principal component (Morrison 1983) of the ring -width measurements, the number of
locally absent rings, and the square ratio between number of ring- widths included in the
chronology and number of rings originally collected. The last parameter, called dating efficiency, or D, ranges from 0 (no crossdating) to 1 (total crossdating), although both extremes
do not apply to published chronologies. The square term emphasizes the nonlinear relationship
Development of a Tree -Ring Network for the Italian Peninsula
19
--......-.,.,.r.,
MTFSS
MEORS
TARAS
MEPSS
CLPSS
MSFSS
SRPPS
CRFSS
PCPPS
MBFSS
CAPES
FUFSS
MAFSS
MNRRS
POFSS
VAFSS
BSFSS
BRFSS
PAPNS
GELDS
PRFSS
1700
1750
1800
1850
1900
1950
2000
Figure 2. Time -series graphs of standard tree -ring chronologies. Chronology codes are
explained in Table 2.
between original sample size and final chronology replication. In other words, the square term
stresses that a chronology built from 400 rings out of 1600 collected rings is worse, in terms
of dating efficiency, than a chronology built from 400 rings out of 800. The advantage of this
parameter over, for instance, the number of increment cores collected and measured is that a
single core may be measured only in part or, on the other hand, it can be broken into more
than one ring -width series.
20
BIONDI
MTFSR
Nr+wW4www
MEORR
TARAR
MEPSR
CLPSR
+,N..V MSFSR
SRPPR
CAFSR
PCPPR
MBFSR
CAPER
FUFSR
MAFSR
MNAAR
POFSR
VAFSR
BSFSR
BAFSR
PAPNR
GELDR
PAFSR
,
.
.
1
.
1700
.
.
.
1
.
1750
.
.
1800
.
1
1850
1900
1950
2000
Figure 3. Time -series graphs of residual tree -ring chronologies. Chronology codes are
explained in Table 2.
The exploratory nature of the analysis called for robust indicators of relationships among
chronologies and for stringent significance levels. Spearman's coefficient of rank correlation
(Sokal and Rohlf 1981) was calculated between all possible pairs of residual tree -ring
chronologies. Since each chronology was tested for correlation with 21 other chronologies, the
level of significance was adjusted using Bonferroni's inequality (Morrison 1983). Hence, only
p- values less than 0.05/21 = 0.0024 were considered significant.
Development of a Tree -Ring Network for the Italian Peninsula
21
RESULTS AND DISCUSSION
A total of 568 increment cores was collected from standing trees of 16 species at 20 sites
(Table 1). No sampled stand could be considered free of human influence, and site conditions
varied greatly in terms of elevation, topography, substrate, soil, and vegetation types.
However, most sampled trees were located in forested areas that are not easily accessible or
are set aside for conservation purposes as parks and natural reserves. Conifers were more
represented than broadleaf species, with eleven compared to five species. Pinus was the most
sampled genus, with six species at nine sites. The oldest living tree was a 963 -year -old pale bark pine (Pinus leucodermis Ant.) sampled in July 1989 at Parco del Pollino. Individuals
more than two centuries old were identified at eleven sites for eight species (Table 1). They
included a 535- year -old Abruzzo pine (Pinus nigra var. italica Hoch.), a 501- year -old
sycamore maple (Acer pseudoplatanus L.), a 475 -year -old yew (Taxus baccata L.), a 417 year -old European beech (Fagus sylvatica L.), a 309 -year -old silver fir (Abies alba Mill.), and
a 287 -year -old European larch (Larix decidua Mill.). The maximum number of annual xylem
rings counted on a single specimen sets a lower bound for species longevity, because increment cores were taken at least 1.0 m above germination level and most cores did not include
the stem pith.
Crossdating occurred at every site for at least one sampled species. Dating problems were
frequent and differed from species to species and from site to site. Brera (Table 1) was the
only site where the number of crossdated samples was too small to produce a reliable tree -ring
chronology. Brera, a botanical garden in the city of Milan, was also the only site where samples were collected from an exotic species. Four tree -ring chronologies, Quercus robur L. at
Lagoni di Mercurago and Fagus sylvatica L. at Maiella, Foresta Umbra, and Parco del
Pollino, were developed from fewer than ten samples (Table 2). They were included in the
tree -ring network because of good crossdating among specimens, but additional collections
are planned at those sites to increase sample size.
Crossdating among annual ring patterns of sampled trees indicates by itself the existence
of an environmental factor, climate, that limits tree growth at large spatial scales (Hughes et
al. 1982). However, trees growing in managed forests under temperate climates usually show
poorly synchronous growth variation (Schweingruber 1988). Most published tree -ring
chronologies (NOAA 1992) are not computed from a constant number of specimens throughout their length, and their earliest time span is commonly based on a single specimen. To minimize the likelihood of dating errors, the earliest time span of the chronologies presented here
was computed using at least two, three, five, six or eight specimens, depending on crossdating
quality and on the availability of nearby tree -ring chronologies for the same species (Table 2).
The tree -ring network for the Italian Peninsula developed so far consists of 22 chronologies for 9 species at 19 sites (Table 2). Seven conifer species accounted for ten chronologies,
and two angiosperm species accounted for the remaining twelve chronologies. The most
represented species was Fagus sylvatica L., with eleven chronologies distributed over the
entire peninsula (Table 2, Figure 1). Pine species were the second largest group, with six
chronologies, also scattered from northern to southern Italy. The development of chronologies
at Parco d'Abruzzo, Parco del Circeo, San Rossore, Campolino, and Parco del Pollino is
described elsewhere (Biondi 1988, 1992, 1993; Biondi and Visani 1993). The tree -ring
chronologies in Biondi 1992 are available from the International Tree -Ring Data Bank
(NOAA 1992).
Mean sensitivity and signal -to -noise ratios (Table 2) of standard tree -ring chronologies
BIONDI
22
Table 1. Summary of 1988 -90 tree -ring collections.'
Table 1. Summary of 1988 -90 tree -ring collections.)
Site and Species
1-Caneiglio
a. Fagua sylvatica
b. Abiea alba
Tree
Core
DEB
Height
(cm)
(a)
Elevation
Slope
N
R
(%)
(m)
5
2
14
3
58- 63
70- 86
26-28
32-34
1160-1265
1220-1240
40- 48
45- 60
149
104
1468
251
8
16
82 -139
15 -24
1430 -1480
20- 27
209
2319
12
24
50- 66
21 -27
925- 975
7- 51
67
1209
12
27
67 -126
27 -35
750- 765
22- 71
226
3528
44- 81
57- 80
15 -22
18 -28
290- 305
300- 320
5- 60
0- 45
89
84
868
447
158
617
3206
667
2- Valzurio
Pague aylvatica
3 -Monte Mottarone
Pague aylvatica
4 -Monte Barro
Pague aylvatica
5- Lagoni di Mercurago
a. Pinas sylvestris
b. Quercua robar
7(1) 13(1)
8
4
6 -Brera
Gingko biloba
2
6
91 -127
18 -20
16
23
24 -106
4 -11
2
6
35- 38
16
32
25- 38
8
17
14
28
120
0
7 -Monte Nero
a. Abies alba
b. Pínus mugo
1585 -1710
1625 -1640
15- 80
10- 20
309
148
8 -10
490- 530
14- 40
86
40- 63
17 -21
1185 -1215
44- 68
99
1408
45- 93
10 -28
1570 -1720
0- 64
155
3110
95 -122
16 -27
1603 -1663
50- 65
287
3598
41- 74
20 -27
1515 -1555
23- 42
79
1675
991 -1316
836 -1126
20- 80
15- 60
361
125
4523
418
6
8 -Ca' del Lupo
Pinus sylvestris
-
2095
9 -Monte della Scoperta
ragua aylvatica
10- Campolino
Picea excelsa
11 -Monte Gerbonte
Larix decidua
10(1) 19(1)
12 -Testa d'Alpe
Abies alba
14
28
13 -Badia Prataglia
a. Pagos aylvatica
b. Abiea alba
16(1) 30
2
4
43 -136
102 -103
8 -30
38 -39
7
14
12
86-151
105-147
21-27
26-29
0-
1
6
0- 20
0- 20
148
158
1483
1304
4
2
8
2
62- 64
60- 80
11-13
13-15
1110-1190
1400-1410
65- 70
150-200
142
515
629
977
a. Pague aylvatica
10
b. Acer pseudoplatanus 2
19
4
59 -123
121 -150
19 -30
27 -29
1320 -1375
1350 -1360
0- 20
15- 16
211
308
2122
1046
6
8
10
108 -151
56 -101
27 -31
16 -24
13 -20
720- 770
25- 90
725- 750
0- 15
0- 55
0- 40
186
66- 90
475
1008
472
2453
69-261
41- 76
80-207
79-123
18-34
5-22
12-48
24-25
1290-1895
1310-1760
1680-1850
1238-1260
0- 53
55-110
24- 75
417
535
501
225
10158
7837
4185
978
70- 89
73- 97
25 -27
25 -26
22 -24
116
207
188
1845
1009
847
963
350
188
204
18707
1850
1110
14-San Rossore
a. Pinua pinea
b. Quercus robur
15-Maiella
a. ragua sylvatica
b. Pinus nigra
0
16 -Bosco di S. Antonio
17- Foresta Umbra
a. ragua aylvatica
b. Pinus halepensis
c. Taxas baccata
3
4
5
18-Parco d'Abruzzo
22(2) 43(4)
a. Pagus sylvatica
b. Pinus nigra
16(2) 32(4)
14
c. Acer pseudoplatanus 8
d. Quercus cerna
3
6
9
18
6
6
13
94
19 -Parco del Circeo
a. Pinus pines
b. Quercua cerria
c. Quercua irainetto
3
3
84 -108
20 -Parco del Pollino
90 -188
a. Pinas leucodermis 21(3) 42(6)
8
74 -121
4
b. Pague sylvatica
110 -154
4
10
c. Quercua cerria
d. Abies alba
3
5
104 -151
4
18
25
23
-24
-25
-36
-35
-50-
20-
0
35
45
1790 -2050
1535 -1930
920- 950
1500 -1540
0
0
0
28 -100
20- 40
15- 25
10- 25
611
Site numbers are the same as in Figure 1.
Numbers of trees and cores in parentheses indicate specimens taken from dead trees.
Nmm: Maximum number of rings counted on a single specimen.
Rte: Total number of rings counted on collected specimens.
Development of a Tree -Ring Network for the Italian Peninsula
23
Table 2. Statistics for standard tree -ring chronologies developed to date.'
Site, Species, (Chronology ID)
First Last
Lc
T
C
D
LAR
PC1
MS
SD
At
S/N p
( %)
1- Cansiglio
a. Fagus sylvatica (CAFSS /R)
1886
1988
103
3- 5
5 -11 .45
0
41
.15
.18
.40
1.8
1
1796
1988
193
2- 8
3 -16 .79
4
56
.26
.24
.22
4.6
2
1943
1989
47
5 -12
8 -23 .65
0
25
.09
.09
.15
2.8
0
1867
1988
122
4 -12
5 -25 .46
2
37
.15
.16
.32
5.1
1
1913
1925
1988
1988
76
64
3- 7
2- 4
5 -13 .78
46
35
.11
.16
.17
.20
.68
.47
5.5
0.8
1
3- 8 .82
1
0
1839
1989
151
5 -12
5 -17 .33
0
35
.13
.15
.38
4.3
2
1912
1989
78
3 -16
5 -32
.89
0
24
.11
.13
.47
3.2
2
1903
1989
87
4- 8
5 -17 .79
1
52
.24
.23
.17
6.8
0
1858
1988
131
3 -12
5 -24 .61
0
24
.11
.11
.10
2.6
0
1737
1988
252
3 -10
3 -19 .84
25
66
.26
.26
.28 15.2
1
0
42
.11
.13
.37
7.9
1
2- Valzurio
Fagus sylvatica (VAFSS /R)
3- Mottarone
Fagus sylvatica (MTFSS /R)
4 -Monte Barro
Fagus sylvatica (MBFSS /R)
5- Mercurago
a. Pinos sylvestris (MEPSS /R)
b. Quercus robur (MEQRS /R)
7 -Monte Nero
a. Ables alba (MNAAS /R)
8 -Ca' del Lupo
Pinus sylvestris (CLPSS /R)
9 -Monte della Scoperta
Fagus sylvatica (MTFSS /R)
1
10- Campolino
Picea excelsa (CAPES /R)
11 -Monte Gerbonte
Larix decidua (GELDS /R)
12 -Testa d'Alpe
Abres alba (TAAAS /R)
13 -Badia Prataglia
a. Fagus sylvatica (BAFSS /R)
14 -San Rossore
a. Pinua pines (SRPPS /R)
1915
1988
74
2 -14
3 -28 .92
1773
1988
216
2 -13
3 -25 .38
1
31
.15
.17
.35
5.1
1
1897
1988
92
3- 7
5 -13
.54
5
31
.13
.16
.49
1.4
1
1849
1988
140
1- 4
2- 7 .68
0
52
.19
.19
.27
0.1
2
1785
1988
204
1- 9
2 -17 .74
2
52
.29
.33
.45
8.6
2
.67
0
63
.27
.24
.18
3.1
2
15- Maiella
a. Fagus sylvatica (MAFSS /R)
16 -Bosco di S. Antonio
a. Fagus sylvatica (BSFSS /R)
17- Foresta Umbra
a. Fagus sylvatica (FUFSS /R)
18 -Parco d'Abruzzo
a. Fagua sylvatica (PAFSS /R)
b. Pinus pigra (PAPNS /R)
19 -Parco del Circeo
a. Pinos pinea (PCPPS /R)
20 -Parco del Pollino
a. Pinus leucodermis (POPLS /R)
b. Fagus sylvatica (POFSS /R)
1852
1988
137
1670
1760
1988
1987
319
228
3 -22
5 -41 .71
5 -16 .14
17
0
39
38
.19
.11
.18
.14
.25
.55
9.5
2.6
2
3- 8
1885
1988
104
4- 9
5 -18 .86
6
42
.13
.17
.54
5.4
2
1 -21
2 -39 .82
7
5- 7 .24
1
38
59
.15
.20
.32
3- 4
.14
.20
8.0
1.9
2
2
1036
1824
1988
1988
953
165
3
6
.31
1
The last letter of the Chronology ID identifies standard (S) and residual (R) chronologies.
Chronology length, from the first to the last year.
Lc:
T, C:
Number of measured trees and cores.
D:
Dating efficiency, i.e. the square ratio between number of measured rings and number of collected rings.
LAR: Number of measured locally absent rings.
Amount of variance explained by the first principal component of the ring -width measures.
PC1:
MS:
Mean sensitivity.
Standard deviation.
SD:
Autocorrelation of first order.
A1:
Signal -to-noise ratio.
S/N:
p:
Order of the autoregressive model used to obtain the residual chronology.
were small compared to commonly reported values (Fritts 1976). Mean sensitivity was ? 0.2
in some Fagus sylvatica chronologies and in the Larix decidua chronology at Monte
Gerbonte, whose signal -to -noise ratio was the only one to exceed ten. Low mean sensitivity
and signal -to -noise ratios were most likely caused by the limited number of available specimens, the conspicuous human disturbance, and the combination of climatic factors limiting
tree growth at each site.
Locally absent rings were included in twelve tree -ring chronologies (Table 2). The greatest number of locally absent rings was included in the Larix decidua chronology at Monte
Gerbonte and in the Fagus sylvatica chronology at Parco d'Abruzzo. Locally absent rings
were included in seven of the eleven beech chronologies, in one of the two Pinus sylvestris
chronologies, in both Pinus pinea chronologies, and in the Pinus leucodermis chronology.
Locally absent rings may also have occurred elsewhere because not all collected rings were
BIONDI
24
crossdated. For instance, no locally absent ring was included in the Pinus nigra chronology at
Parco d'Abruzzo, although they were likely to occur before A.D. 1750, when collected specimens could not be crossdated because of many contiguous microrings (Biondi and Visani
1993).
Autoregressive models of first and second order were able to remove the autocorrelation
(Table 2). In particular, the order p of selected autoregressive models for beech tree -ring series
decreased from two to one to zero with decreasing age of beech trees. When most specimens
were taken from trees older than 150 years - as at Valzurio, Parco d'Abruzzo, Bosco di
Sant'Antonio, Maiella, Foresta Umbra, and Parco del Pollino - the best model was a second order autoregressive process (Table 2). This model always included a positive relationship
with the previous year but a negative relationship between values two years apart (Biondi
1993). At Monte Barro, Cansiglio, and Badia Prataglia, where very few trees exceeded 100150 years of age, the best model was a first -order autoregressive process (Table 2) with a positive parameter estimate. At Mottarone and Monte della Scoperta, where most tree ages were
less than 100 years, no significant autocorrelation remained after removing the growth trend,
hence p was equal to zero (Table 2). Further research is needed to rigorously test if cambial
age affects autocorrelation of wood growth in Fagus sylvatica and possibly in other tree species.
Coefficients of rank correlation were computed between every possible pair of residual
chronologies (Table 3). The Fagus sylvatica chronologies showed the best agreement with one
another. The only exception was the Mottarone site (Table 2), which showed no strong synchrony with other beech chronologies. Trees at that site were young, fast growing, and thus
scarcely sensitive to year -to -year climatic variations. By comparing site locations (Figure 1)
with significant correlations (Table 3), it was possible to identify a northern, a central, and a
southern group of chronologies. The northern group centered around the beech chronology at
Valzurio (VAFSR), which also correlated significantly with northern chronologies of coniferous species in mountain areas, such as Abies alba at Monte Nero (MNAAR) and Larix decidua at Monte Gerbonte (GELDR). The central group clustered around the beech chronology at
Badia Prataglia (BAFSR), which correlated significantly with every other beech chronology
except Mottarone, Foresta Umbra, and Parco del Pollino. The southern group was best represented by the beech chronology at Parco d'Abruzzo (PAFSR), which also correlated significantly with other southern chronologies of coniferous species in mountain areas, such as Pinus
nigra at Parco d'Abruzzo (PAPNR) and Pinus leucodermis at Parco del Pollino (POPLR;
Table 3).
Annual stem growth of old European beeches at Parco d'Abruzzo, whose precipitation
and temperature regimes could be described as Mediterranean mountain climate, is greatly
affected by the accumulation and melting of winter snowfall (Biondi 1993). Radial growth is
mainly limited by summer drought and relies heavily on the accumulation and melting of winter snowpack to meet moisture demand in the growing period. The widespread occurrence of
old beech individuals and forest stands and the climatic sensitivity displayed by their annual
rings is likely to allow further refinement of existing dendroclimatic reconstructions for southern Europe and the Mediterranean Basin.
The Pinus leucodermis chronology at Parco del Pollino (Table 2; Biondi and Visani
1993) covered a larger area and a longer time span than Serre- Bachet's (1985) chronology for
the same species. Spurious correlations between discrete time series may arise when values
are autocorrelated (Monserud 1986). Hence, autocorrelation in Serre- Bachet's chronology,
abbreviated SBPLS, was removed before computing correlations between chronologies. The
1
13:.43(121)
13:.40( 86)
1:.46(102)
18a:.28(215)
9:.39( 86)
17:.27(135)
15:.33(138)
18a:.37(135)
20b:.58(163)
4-MBFSR
9-MSFSR
13-BAFSR
15-MAFSR
16-RSFSR
17-FUFSR
18a-PAFSR
9:.40( 63)
4:.31(121)
2:27(191)
18a:.20(227)
18a:.17(317)
7-MNAAR
11-GELDR
18b-PAPNR
20a-POPLR
18a:.58(163)
5-MEQRR
18a:.26(251)
2:.25(148)
17:32(135)
15:37(138)
20b:.32(135)
20b:.29(163)
18a:.37(138)
4:.43(121)
16:.26(202)
15:.39( 86)
2:37(121)
1:.35(102)
2:35 (102)
15:.30(138)
17:.37(135)
15:27(135)
13:26(202)
13:.35(138)
9:.40( 86)
1:.33(102)
13:.33(191)
4:.33(102)
16:29(163)
13:.28(215)
18a:.25(202)
16:33(138)
15:.35(138)
18a:.22(191)
Chronology ID are identical to those used in Tables 1 and 2 and in Figures 1, 3, and 4.
Each row of Significant Correlations is ordered by magnitude of correlation.
The number of observations used to compute each correlation is given in parentheses.
Only correlations having p- values < 0.0024 were considered significant and included in this table.
20b-POFSR
4:37(121)
2:22(191)
13:.46(102)
2-VAFSR
Significant Correlations
With Fagua sylvatica Chronologies
1-CAFSR
Chronology
Other
Fagus
16:25(202)
20b:.30(138)
2:.33(191)
SBPLR:.66(824)
20a:.33(227)
11:26(251)
5:.40 (63)
7:.31(121)
7:25(148)
18b:.33(227)
SBPLR:.30(214)
18b:.20(227)
11:27(191)
Significant Correlations
With Other Chronologies
Table 3. Spearman's coefficients of rank correlation between pairs of residual tree -ring chronologies.'
20a:.17(317)
26
BIONDI
*So*
I)*
ky% Aiii40001
WOofoiv
1000
1200
SBPLR
490,1404
SBPLS
4#01$
POPLR
$&40,
feliWii*SA(1444iikikie4 POPLS
1400
1600
1800
2000
Figure 4. Time -series graphs of standard and residual Pinus leucodermis chronologies.
SBPLS is Serre -Bachet's (1985) chronology; SBPLR is the residual chronology derived by
fitting an AR(3) model to SBPLS. The other two chronologies are the standard and residual
chronologies developed in this study (Table 2).
residuals from a third -order autoregressive model, or AR(3), formed the prewhitened chronology, abbreviated SBPLR (Table 3, Figure 4). A second -order autoregressive model, or AR(2),
was selected to transform POPLS into POPLR (Table 2). Most likely, the type of standardization used to remove the growth trend in the present study left a smaller amount of autocorrelation in the average ring indices than the one used to develop SBPLS. The correlation between
the two Pinus leucodermis chronologies was the highest between any pair of chronologies
(Table 3).
The generally low correlation between chronologies for coniferous species agreed with
previous dendroclimatic studies of European species. Densitometric methods based on X -ray
techniques have been suggested to maximize climatic signals present in annual rings of
conifers growing on cold -moist sites near alpine and northern timberline (Schweingruber
1988). Since densitometric methods are not very effective when applied to angiosperms, ring
width is the main parameter available for such species. It is therefore fortunate, from a den drochronological standpoint, that tree -ring chronologies derived from ring -width series of
European beech growing in the Italian Peninsula are highly synchronous, to the extent of
reflecting broad geographic regions and climatic patterns.
CONCLUSIONS
The tree -ring network for the Italian Peninsula consists of 22 tree -ring chronologies at 19
sites. No sampled stand could be considered free of human influence, and site conditions varied greatly in terms of elevation, topography, substrate, soil, and vegetation types. However,
most sampled trees were located in forested areas that are not easily accessible or are set aside
for conservation purposes as parks and natural reserves. Eleven sites yielded individuals more
than two centuries old for eight tree species.
Development of a Tree -Ring Network for the Italian Peninsula
27
Fagus sylvatica is the most promising species in terms of plant longevity, widespread distribution, crossdating quality, and climatic sensitivity. Significant coefficients of rank correlation between beech tree -ring chronologies highlight the existence of three broad regions distributed along a latitudinal gradient, corresponding to large -scale climatic patterns across the
Italian Peninsula. Future research should disentangle the multivariate information embedded
in tree rings of Italian species and exploit existing chronologies for dendroclimatic reconstructions.
In southern Europe and the Mediterranean Basin, dendrochronological studies provide
much needed proxy records of past environmental changes (Dutilleul and Till 1992; Meko
1985). The new Italian network presented here is another mosaic stone needed to complete
existing tree -ring databases, such as those developed by Schweingruber (1985) and SerreBachet (1992). Once combined with ongoing research efforts in other Mediterranean countries
(Biger and Liphschitz 1992; Creus et al. 1992; Gadbin 1992; Guibal 1992; Gutiérrez 1989;
Kuniholm and Striker 1987), dendrochronology of Italian species will contribute significantly
to our understanding of climate -tree growth relations in temperate environments.
ACKNOWLEDGMENTS
Tree -ring sampling was supported, in part, by the Dr. M. Aylwin Cotton Foundation, the
Parco Nazionale d'Abruzzo, and the Centro Ricerche Termiche e Nucleari/ENEL. Data processing was funded, in part, by the Consiglio Nazionale delle Ricerche, the Centro Ricerche
Termiche e Nucleari/ENEL, and The University of Arizona. I owe thanks to Giorgio
Schenone, Franco Tassi, and Malcolm K. Hughes for their support. A few foresters, park
rangers and volunteers helped me during field collections. Richard Holmes, Bob Lofgren, and
Henri Grissino -Mayer wrote software used for data processing. Simona Visani was most helpful in dating and measuring a number of samples. Jeff Dean, Paul Sheppard, and two anonymous reviewers provided comments on the original manuscript.
REFERENCES
Bebber, A.E.
Una cronologia del lance (Larix decidua Mill.) delle Alpi orientali italiane. Dendrochronologia 3:201206.
Biger, G., and N. Liphschitz
Tree ring and dendrohistorical investigations in Israel - the use of imported timber in 19th century
1992
1990
Israel. In "Tree Rings and Environment: Proceedings of the International Dendrochronological
Symposium, Ystad, South Sweden, 3 -9 September 1990," edited by T. S. Bartholin, B. E. Berglund, D.
Eckstein, F. H. Schweingruber, and O. Eggertsson, pp. 30 -33. Lundqua Report, No. 34. Lund.
Biondi, F.
1988
Studi dendrocronologici in località appenniniche: primo rapporto. Informatore Botanico Italiano
20(1):515 -521.
1992
1993
Four tree -ring chronologies for the Italian peninsula. In "Tree Rings and Environment: Proceedings of
the International Dendrochronological Symposium, Ystad, South Sweden, 3 -9 September 1990," edited
by T. S. Bartholin, B. E. Berglund, D. Eckstein, F. H. Schweingruber, and O. Eggertsson, pp. 41-44.
Lundqua Report, No. 34. Lund.
Climatic signals in tree rings of Fagus sylvatica from the central Apennines, Italy. Acta Oecologica
14:57 -71.
Biondi, F., and T.W. Swetnam
Box -Jenkins models of forest interior tree -ring chronologies. Tree -Ring Bulletin 47:71 -95
1987
Biondi, F., and S. Visani
Cronologie dendroanulari per la Penisola Italiana. Annali della Facoltà di Agraria, Università Cattolica
1996
di Milano, 33(1):3 -23
BIONDI
28
Bortolotti, L.
Foreste. In: Lega per l'Ambiente, Ambiente Italia, pp. 99 -107. ISEDI Petrini Editore, Torino.
1989
Box, G.E.P., and G.M. Jenkins
Time Series Analysis: Forecasting and Control, revised edition. Holden -Day, Oakland.
1976
Bräker, O.U., and F.H. Schweingruber
1989
Standorts- Chronologien Teil 2: Apenninen -Halbinsel. Eidgenössische Anstalt für das forstliche
Versuchswesen, Birmensdorf.
Brugnoli, A., and C. Gandolfo
Analisi dendroclimatica sull'abete rosso (Picea abies (L.) Karst.) del Trentino orientale: primi risultati.
1991
Monti e Boschi 42(6):51 -56.
Cook, E.R., and K. Peters
The smoothing spline: a new approach to standardizing forest interior tree -ring width series for dendro1981
climatic studies. Tree -Ring Bulletin 41:45 -53.
Creus, J., M.d.M. Génova, A.F. Cancio, and A.P. Antelo
New dendrochronologies for Spanish Mediterranean zone. In "Tree Rings and Environment:
1992
Proceedings of the International Dendrochronological Symposium, Ystad, South Sweden, 3 -9
September 1990," edited by T. S. Bartholin, B. E. Berglund, D. Eckstein, F. H. Schweingruber, and O.
Eggertsson, pp. 76 -78. Lundqua Report, No. 34. Lund.
Davis, J.C.
Statistics and Data Analysis in Geology, second edition. Wiley, New York.
1986
Dutilleul, P., and C. Till
Evidence of periodicities related to climate and planetary behaviors in ring -width chronologies of Atlas
1992
cedar (Cedrus atlantica) in Morocco. Canadian Journal of Forest Research 22:1469 -1482.
Fritts, H.C.
Tree Rings and Climate. Academic Press, New York.
1976
Fritts, H.C., and D.J. Shatz
Selecting and characterizing tree -ring chronologies for dendroclimatic analysis. Tree -Ring Bulletin
1975
35:31 -40.
Gadbin, C.
1992
Guibal, F.
1992
Influence of data and calibration processing on response function: the example of Pinus pinea L. In
"Tree Rings and Environment: Proceedings of the International Dendrochronological Symposium,
Ystad, South Sweden, 3 -9 September 1990," edited by T. S. Bartholin, B. E. Berglund, D. Eckstein, F.
H. Schweingruber, and O. Eggertsson, pp. 109 -111. Lundqua Report, No. 34. Lund.
Comparative analysis of three Mediterranean pine series: Pinus halepensis Mill., Pinus pinea L., Pinus
mesogeensis Fieschi et Gaussen. In "Tree Rings and Environment: Proceedings of the International
Dendrochronological Symposium, Ystad, South Sweden, 3 -9 September 1990," edited by T. S.
Bartholin, B. E. Berglund, D. Eckstein, F. H. Schweingruber, and O. Eggertsson, pp. 132 -136.
Lundqua Report, No. 34. Lund.
Gutiérrez, E.
1989
Dendroclimatological study of Pinus sylvestris L. in southern Catalonia (Spain). Tree -Ring Bulletin
49:1 -9.
Holmes, R.L.
Computer -assisted quality control in tree -ring dating and measurement. Tree -Ring Bulletin 43:69 -78.
1983
Hughes, M.K., P.M. Kelly, J.R. Pilcher, and V.C. LaMarche, Jr., Editors.
Climate from Tree Rings. Cambridge University Press, Cambridge.
1982
Jacoby, G.C., Jr., and J.W. Hornbeck, Compilers
Proceedings of the International Symposium on Ecological Aspects of Tree -Ring Analysis. U.S.
1987
Department of Energy, Washington, DC.
Kuniholm, P.I., and C.L. Striker
Dendrochronological investigations in the Aegean and neighboring regions, 1983 -1986. Journal of
1987
Field Archaeology 14:385 -398.
Loeble, C.
Tree life history strategies: the role of defenses. Canadian Journal of Forest Research 18:209 -222.
1988
Meko, D.M.
Temporal and spatial variation of drought in Morocco. In: Proceedings of the Conference on Drought,
1985
Water Management and Food Production, pp. 55 -82. Kingdom of Morocco, Agadir.
Molisch, H.
The Longevity of Plants. E.H. Fulling, New York.
1938
Monserud, R.A.
Time -series analyses of tree -ring chronologies. Forest Science 32:349 -372.
1986
Development of a Tree -Ring Network for the Italian Peninsula
29
Morrison, D.F.
Applied Linear Statistical Methods. Prentice -Hall, Englewood Cliffs.
1983
Nardi Berti, R.
La struttura anatomica del legno ed il riconoscimento dei legnami italiani di più corrente impiego.
1979
Istituto del Legno, Consiglio Nazionale delle Ricerche, Firenze, Italy.
NOAA
International Tree -Ring Data Bank. National Geophysical Data Center, Boulder, Colorado.
1992
Nola, P.
Dendrocronologia della quercia nella Pianura Padana. Dendrochronologia 6:179 -192.
1988
Santini, A., and N. Martinelli
1991
Indagine dendroecologica sull'abete bianco (Abies alba Mill.) di Serra San Bruno (CZ). Giornale
Botanico Italiano 125(6):895 -906.
Schweingruber, F.H.
Dendro -ecological zones in the coniferous forests of Europe. Dendrochronologia 3:67 -75.
1985
Tree Rings: Basics and Applications of Dendrochronology. Kluwer, Boston.
1988
Schweingruber, F.H., K.R. Briffa, and P.D. Jones
Yearly maps of summer temperatures in Western Europe from A.D. 1750 to 1975 and Western North
1991
America from 1600 to 1982: Results of a radiodensitometrical study on tree rings. Vegetatio 92:5 -71.
Serre -Bachet, F.
Une chronologie pluriséculaire du sud de l'Italie. Dendrochronologia 3:45 -66.
1985
Annual temperature reconstructions for the Mediterranean area (1700- 1979). In "Tree Rings and
1992
Environment: Proceedings of the International Dendrochronological Symposium, Ystad, South
Sweden, 3 -9 September 1990," edited by T. S. Bartholin, B. E. Berglund, D. Eckstein, F. H.
Schweingruber, and O. Eggertsson, pp. 293 -297. Lundqua Report, No. 34. Lund.
Società Botanica Italiana
1971
Censimento dei biotopi di rilevante interesse vegetazionale meritevoli di conservazione in Italia.
Gruppo di lavoro per la conservazione della natura, Camerino.
Sokal, R.R., and F.J. Rohlf
Biometry, second edition. W.H. Freeman and Co., New York.
1981
Stokes, M.A., and T.L. Smiley
An Introduction to Tree -Ring Dating. The University of Chicago Press, Chicago.
1968

Download Report