Species, live status, and diameter are important tree features for diversity and abundance of tree microhabitats in subnatural montane beech–fir forests1

Abstract

Because quantitative data on the distribution of whole microhabitat sets are still lacking to indirectly assess taxonomic biodiversity in forests, we studied the distribution of seven key microhabitat types in 10 montane European beech (Fagus sylvatica L.) – silver fir (Abies alba Mill.) forests (Pyrénées, France) that had not been harvested for several decades. We examined 2105 live trees and 526 snags. Frequencies of cavities and dendrothelms were significantly higher on live beech than on fir. Sap runs were strictly found on live fir. Frequencies of cracks and saproxylic fungi were significantly higher on snags than on live trees. Seventy percent of live beeches but only 18% of firs carried one or more microhabitats. For both beech and fir and for each microhabitat type, we found, using the recursive partitioning method, one to three diameter thresholds that each corresponded to a significant change in the probability of microhabitat presence. When considering the whole microhabitat set, the most significant diameter thresholds were 42, 60, 73, and 89 cm for beech and 99 cm for fir. We suggest that forest managers conserve (i) mixed stands and (ii) beech with a diameter at breast height >90 cm and fir >100 cm. These rules should be adapted for each forest ecosystem.

Full text

Species, live status, and diameter are important
tree features for diversity and abundance of tree
microhabitats in subnatural montane beech–fir
forests1
Laurent Larrieu and Alain Cabanettes
Abstract: Because quantitative data on the distribution of whole microhabitat sets are still lacking to indirectly assess taxo-
nomic biodiversity in forests, we studied the distribution of seven key microhabitat types in 10 montane European beech
(Fagus sylvatica L.) –silver fir (Abies alba Mill.) forests (Pyrénées, France) that had not been harvested for several decades.
We examined 2105 live trees and 526 snags. Frequencies of cavities and dendrothelms were significantly higher on live
beech than on fir. Sap runs were strictly found on live fir. Frequencies of cracks and saproxylic fungi were significantly
higher on snags than on live trees. Seventy percent of live beeches but only 18% of firs carried one or more microhabitats.
For both beech and fir and for each microhabitat type, we found, using the recursive partitioning method, one to three diam-
eter thresholds that each corresponded to a significant change in the probability of microhabitat presence. When considering
the whole microhabitat set, the most significant diameter thresholds were 42, 60, 73, and 89 cm for beech and 99 cm for
fir. We suggest that forest managers conserve (i) mixed stands and (ii) beech with a diameter at breast height >90 cm and
fir >100 cm. These rules should be adapted for each forest ecosystem.
Résumé : Étant donné qu’il n’y a pas encore de données quantitatives sur la distribution d’ensembles complets de microha-
bitats qui permettraient d’obtenir une évaluation indirecte de la biodiversité dans les forêts, nous avons étudié la distribution
de sept types clés de microhabitats dans 10 forêts montagnardes (Pyrénées, France) de hêtre (Fagus sylvatica L.) et de sapin
(Abies alba Mill.) qui n’ont pas été récoltées depuis plusieurs décennies. Nous avons examiné 2105 arbres vivants et 526
chicots. La fréquence des cavités et des dendrotelmes était significativement plus élevée sur les tiges vivantes de hêtre que
sur le sapin. Des écoulements de sève ont été observés uniquement sur les tiges vivantes de sapin. La fréquence des fentes
et des champignons saproxyliques était significativement plus élevée sur les chicots que sur les arbres vivants. Il y avait au
moins un microhabitat sur 70% des hêtres mais sur seulement 18% des sapins. Tant sur le hêtre que sur le sapin et pour
chaque type de microhabitat, nous avons trouvé, à l’aide d’une méthode de partitionnement récursif, entre un et trois diamè-
tres seuils. Chaque diamètre seuil correspondait à un changement significatif dans la probabilité de la présence d’un micro-
habitat. Lorsqu’on tient compte de l’ensemble complet des microhabitats, les diamètres seuils les plus importants sont 42,
60, 73 et 89 cm chez le hêtre et 99 cm chez le sapin. Nous suggérons que les gestionnaires forestiers conservent (i) des peu-
plements mixtes ainsi que (ii) les hêtres et les sapins dont le diamètre à hauteur de poitrine est respectivement plus grand
que 90 et 100 cm. Ces règles devraient être adaptées pour chaque écosystème forestier.
[Traduit par la Rédaction]
Introduction
Using only a few species to assess taxonomic biodiversity
in forest ecosystems is not satisfactory and it is rather pref-
erable to use whole taxonomic groups (Lindenmayer and
Franklin 2002). Further, direct bioindicator records are very
expensive and require taxonomic specialists (Puumalainen et
al. 2003). Most importantly, the relationships between indica-
tor taxa and total biodiversity are not yet well established
(Lindenmayer et al. 2000; McElhinny et al. 2005). Therefore,
forest managers need alternative approaches to assess biodi-
versity in forests. Lindenmayer et al. (2000) suggested using
structure-based variables as indirect biodiversity indicators.
Indicators based on key structural factors have been shown
to be a practical and efficient way to ensure that taxonomic
biodiversity is taken into account in current forest manage-
ment (Larsson 2001). Similarly, Tews et al. (2004) proposed
using “crucial keystone structures”, such as dead wood, for
biodiversity management. Lindenmayer et al. (2006) pub-
Received 8 September 2011. Accepted 4 May 2012. Published at www.nrcresearchpress.com/cjfr on xx July 2012.
L. Larrieu. INRA, INPT/ENSAT/EIPURPAN, UMR 1201 Dynafor, F-31326 Castanet-Tolosan, France; Centre Régional de la Propriété
Forestière de Midi-Pyrénées, 7 chemin de la Lacade, F-31320 Auzeville Tolosane, France.
A. Cabanettes. INRA, INPT/ENSAT/EIPURPAN, UMR 1201 Dynafor, F-31326 Castanet-Tolosan, France.
Corresponding author: Alain Cabanettes (e-mail: alain.cabanettes@toulouse.inra.fr).
1This article is one of a selection of papers from the International Symposium on Dynamics and Ecological Services of Deadwood in
Forest Ecosystems.
1433
Can. J. For. Res. 42: 1433–1445 (2012) doi:10.1139/X2012-077 Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

lished a checklist of strategies to guide forest biodiversity
conservation that encompasses the maintenance of key ele-
ments of stand structural complexity.
Tree microhabitats (such as cavities or cracks) are key
components of forest stands (Michel and Winter 2009) be-
cause they host a wide taxonomic biodiversity (Speight
1989). So, they are relevant and practical proxies to assess
taxonomic biodiversity at the stand scale (Winter and Möller
2008). Several authors have demonstrated the important role
of very large trees (Ranius and Jansson 2000; Grove 2002;
Branquart et al. 2005; Bauhus et al. 2009), tree species (Gos-
selin and Larroussinie 2004), and snags (Jonsell and Weslien
2003) on taxonomic biodiversity. However, in most types of
forest, few data are available regarding the distribution of tree
microhabitats in natural stands and the links between tree
species, diameter, status (live tree or snag), and microhabitat
occurrence.
To improve our knowledge of the distribution of tree mi-
crohabitats in natural forests, and particularly to better under-
stand the role of tree species and tree diameter on
microhabitat occurrence, we observed tree microhabitats in
montane European beech (Fagus sylvatica L.) –silver fir
(Abies alba Mill.) forests that had not been logged for at least
60 years. Beech–fir forest is a common forest community,
present in most of the European mountain ranges, and has a
great economic and ecological importance. It is also a mixed
forest where biocenoses partially differ between broadleaved
trees and conifers (Nascimbene et al. 2009).
We chose a set of seven microhabitat types that seem to be
very important for taxonomic diversity in forests because the
associated taxonomic groups are numerous and varied or
very specific: empty cavities, cavities with mould, sporo-
phores of saproxylic fungi, dendrothelms (water-filled holes
in the wood), sap runs, missing bark, and cracks. Empty cav-
ities are used for protection against bad weather conditions or
predators by more than 25% of vertebrate species in north-
eastern North American forests (DeGraaf and Shigo 1985).
Moreover, in France, 41% of forest birds are cavity-dwelling
species (Blondel 2005). Cavities with mould are inhabited by
arthropods (Ranius 2002) and create favorable conditions for
epiphyte species of conservation concern (Fritz and Heil-
mann-Clausen 2010). Sporophores of saproxylic fungi sup-
port a varied insect fauna (Dajoz 2007), especially when
they are tough (polypores s.l.) or pulpy (e.g., oyster fungi).
Some parasitic fungi also use saproxylic fungi as a resource
(Lisiewska 1992). Dendrothelm-dwelling species are not nu-
merous (Dajoz 2007) but very specialized (Kitching 1971).
Sap runs host syrphid larvae (Speight et al. 2010) and are
used by the adults of several Coleoptera species (Alexander
2002). Cracks are important microhabitats for spiders (Stan-
ska et al. 2010), birds (Cramp 1980), flat bugs (Heiss and
Pericart 2007), and bats (Pénicaud 2000). Cavities, cracks,
and missing bark are indicators of natural forests (Michel
and Winter 2009; Remm and Lõhmus 2011).
This paper aims at (i) evaluating the role in terms of sup-
ply of microhabitats of the tree species that compose beech–
fir forests independently of their relative abundance and (ii)
identifying critical diameter thresholds for both microhabitat
abundance and diversity. Then, in the context of sustainable
management practices, we suggest some practical recommen-
dations and a management strategy to conserve microhabi-
tats.
Materials and methods
Studied forests and sampling design
The 10 studied forests (Table 1) are situated in the central
Pyrénées mountain range (Fig. 1) and have not been logged
for more than 60–100 years. They are natural habitats of
beech–fir forest (Bardat et al. 2004). However, stands host a
very variable proportion of fir, which is directly linked to his-
toric human intervention that favored beech at the expense of
fir (Métailié 2001). For the analysis, we pooled all of the
studied forests because local conditions of fertility were not
markedly different and we did not sample forests growing in
extreme conditions of infertility (e.g., site with PODZOSOL).
Observations were carried out in 2008 and 2009 on a sample
of 62 plots, 2105 live trees, and 526 snags (Table 1).
Although the leaf canopy may hinder observations, all of the
plots were set up in summer because these sites are covered
by snow for a large part of the year. Because of the presence
of leaves, we expected an underestimation of the number of
microhabitats on beech and on the other broadleaved species.
For the evergreen species such as fir or common yew (Taxus
baccata L.), data taken in the vegetation period or in winter
are more easily comparable.
Measurements and observation of microhabitats
Plots were circular and of variable sizes set up using the
No. 1 strip (return angle of 1/50) of a Bitterlich relascope
(Pardé and Bouchon 1988). This device uses a constant an-
gle. The measurement errors due to terrain slope are auto-
matically corrected by the relascope, which is very practical
in the mountains. The use of a relascope allows a high sam-
pling rate of large trees that are richer in microhabitats (Win-
ter and Möller 2008; Larrieu et al. 2011). Plot locations were
positioned approximately on a map before the field phase in
relation to accessibility. Then, the precise locations of plots
were established in relation to topographical constraints
(such as cliffs) and plot centres were always a minimum of
100 m apart. Trees were observed individually. We noted the
status (live tree or snag), identified the tree to species level,
and measured the diameter on the outside of the bark at
breast height (dbh) to the nearest centimetre when the diame-
ter exceeded 5 cm. We carefully examined the trunk from the
ground to the top of the tree to note microhabitats hosted on
the visible part of the trunk both beneath and within the tree
crown. We recorded a set of seven microhabitats types as fol-
lows. (i)“Empty”cavities with an entrance above 3 cm in
width. We did not use a device such as a camera mounted
on a telescopic pole to ensure the cavity volume. Therefore,
we pooled in this category all woodpecker breeding holes,
holes made by woodpeckers when feeding and deep enough
to host a vertebrate, deep cavities formed between roots, and
also natural cavities low enough on the tree to enable verifi-
cation that they were empty and that they were not at a stage
where mould could develop. (ii) Cavities with mould. We
pooled in this category the other natural cavities and cavities
with mould with an entrance above 3 cm in width and also
missing bark patches with an area above 100 cm2with wood
in a decay stage of more than 3 (in reference to a scale with
1434 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

Table 1. Main characteristics of the studied European beech (Fagus sylvatica)–silver fir (Abies alba) forests and sampling design.
Diameter at breast height (cm)
Forest Vegetation level Dominant tree
species History Studied area
per forest (ha) No. of
plots
No. of
observed
trees
No. of
observed
snags Live
beeches Beech
snags Live
firs Fir
snags
Plagnech de ton Lower montane Beech and fir Not logged since 1900 23 10 332 93 15.9 11.1 8.0 22.3
62.9 53.4 44.4 70.4
127.3 95.5 124.1 109.8
Reouère Lower montane Beech Old pasture woodland not
logged for more than
110 years
43 8 295 40 11.1 8.0
67.0 53.1
152.8 162.3
Auribareille Lower montane Fir Never logged 12 2 61 14 15.9 63.7 22.3 19.1
50.9 63.7 70.6 64.0
78.0 63.7 130.5 89.1
Genie loungue Lower montane Beech Not logged since 1900 16 5 122 14 21.6 11.8 47.4
54.4 25.6 47.4
114.6 33.1 47.4
Bugatet Upper montane/
subalpine Fir Not logged since 1908 18 5 185 36 7.0 22.9
49.2 68.0
111.4 140.1
Barrada Upper montane Fir Only logged once (1953)
by selective logging 13 5 221 63 11.1 11.1 19.1
20.3 58.6 61.2
25.5 171.9 125.7
Bosc nou Upper montane/
subalpine Beech and fir Never logged 13 5 178 55 29.0 9.5 21.6
36.1 57.5 59.7
42.7 114.6 101.9
Es tucoulets Upper montane Beech and fir Not logged since 1900 34 9 265 82 24.8 19.1 9.5 15.9
31.8 19.1 53.7 60.9
38.8 19.1 130.5 154.4
Es piches Upper montane Beech and fir Not logged since 1900 17 5 153 20 15.9 31.8 11.1 12.7
56.2 35.0 56.8 62.5
95.5 38.2 135.3 124.1
Ouderou Upper montane Fir Old pasture woodland not
logged for more than
100 years
25 8 293 109 14.3 12.7 10.2 13.0
36.0 28.2 65.9 65.4
78.0 60.5 135.9 133.7
Note: For diameters at breast height, the three values are successively the minimum, mean, and maximum.
Larrieu and Cabanettes 1435
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

five steps: see Table S12(describes in detail wood decay
stages)). (iii) Sporophores of saproxylic fungi without taxo-
nomic identification. We noted only tough fungi (polypores
s.l.) or pulpy fungi (e.g., Pleurotus ostreatus (Jacq.) P.
Kumm. 1871). (iv) Dendrothelms, when the entrance was
more than 3 cm in width. In the dry period, observation of
traces of water flow on the trunk allowed the diagnostic of
inaccessible dendrothelms. (v) Sap runs with a minimal
length of 10 cm. (vi) Missing bark (i.e., wood patches with
bark loss) of at least 100 cm2, with wood in a decay stage of
less than 2. (vii) Cracks in the tree trunk with a width of 1–
5 cm and situated over 1 m above the ground or bark in the
process of peeling and that formed a shelter. Their impor-
tance for several species of bats (Meschede and Heller 2003)
was the justification for pooling these two microhabitats and
using these thresholds.
On each tree, we counted every microhabitat type as often
as it appeared, except in the case of fungi, which were only
noted as presence–absence.
Beech (658 live trees observed) and fir (1310 live trees ob-
served) were the main tree species but we also observed 137
live trees of more than 10 secondary species (European
mountain-ash (Sorbus aucuparia L.), European white birch
(Betula pendula Roth), downy birch (Betula pubescens
Ehrh.), European ash (Fraxinus excelsior L.), sweet cherry
(Prunus avium (L.) L.), field maple (Acer campestre L.),
Norway ample (Acer platanoides L.), common yew, largeleaf
linden (Tilia platyphyllos Scop.), and Salix spp.). Secondary
species were pooled in the analysis because of their low
number. Stand maturity allowed us to explore a large diame-
ter gradient: 6–172 cm for fir, 11–153 cm for beech, and 5–
95 cm for other species.
Calculations and statistical procedures
The theoretical tree frequency per hectare was calculated
by allocating the coefficient N, related to its diameter (D), to
every tree observed in the relascope sampling (Pardé and
Bouchon 1988):
ND ¼p108½arctanð1=50Þ=pD2
where “arctan”is the trigonometrical “arctangent”function.
All statistical calculations were done using R software (R
development Core Team 2007).
Basic data were measured and analyzed at the level of the
individual tree: tree species, tree diameter, type, and number
of microhabitats. However, the effect of tree species was
tested using the average data per plot and per species. The
role of the tree species in the supply of microhabitats was
evaluated independently of their relative abundance.
Comparisons of frequencies, per species, of the trees that
bear microhabitats and of frequencies of microhabitat co-oc-
currences per species were carried out using the c2test (Sne-
decor and Cochran 1971).
The hypothesis of independence between the three species
categories (i.e., beech, fir, and the third that pooled all secon-
Fig. 1. The 10 studied forests are situated on the northern slopes of the Pyrénées mountain range (the southwest of France).
2Supplementary data are available with this article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/x2012-077.
1436 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

dary species) and the response variable (frequency of one or
all microhabitats or frequency of microhabitat type) was as-
sessed using multiple testing of resampled data (Hothorn et
al. 2006). The response variable was analyzed as a rank vari-
able. The pvalue obtained by this procedure was adjusted for
multiple comparisons using a step-down max-T approach. In
addition, for each response variable, a post hoc test (Tukey
all-pair comparisons) was applied to assess the differences
between each pair of categories. The corresponding pvalues
were adjusted for all comparisons performed here. This anal-
ysis procedure is based on implementation of the above pro-
cedures in the add-on package “coin”(Hothorn and Hornik
2005).
The relation between microhabitat frequency and tree di-
ameter was analyzed using tree-based regression and classifi-
cation models. We tested the hypothesis that there is a
threshold-based relation between the number of microhabitats
and dbh because we found this kind of relation in a previous
analysis of comparable data (Larrieu et al. 2011). These
threshold values were calculated by recursive partitioning
(Hothorn and Zeileis 2008). This approach allows simultane-
ous identification of a threshold and assessment of its signifi-
cance by means of a statistical test procedure. The thresholds
are derived from estimates of break points by means of max-
imally selected two-sample statistics. Their validity is judged
by multiple test procedures. Once the data set is divided into
Table 2. Tree species (European beech (Fagus sylvatica), silver fir (Abies alba), and
others) effect on microhabitat frequency: results for the comparison tests on live trees and
snags (significant results in bold).
Microhabitat Tree status All-categories
comparison (p)Pairwise
comparison p
Cavities with mould Live trees <0.001 Beech/others <0.001
fir/others 0.841
Fir/beech <0.001
Empty cavities Live trees 0.005 Beech/others 0.005
Fir/others 0.503
Fir/beech 0.161
Saproxylic fungi Live trees 0.424 Beech/others 0.424
Fir/others 0.779
Fir/beech 0.871
Dendrothelms Live trees <0.001 Beech/others <0.001
Fir/others 0.978
Fir/beech <0.001
Missing bark Live trees 0.988 Beech/others 0.976
Fir/others 0.999
Fir/beech 0.999
Cracks Live trees 0.212 Beech/others 0.211
Fir/others 0.688
Fir/beech 0.733
Sap runs Live trees <0.001 Beech/others 0.711
Fir/others 0.002
Fir/beech <0.001
Total Live trees <0.001 Beech/others 0.366
Fir/others 0.011
Fir/beech <0.001
Empty cavities Snags 0.005 Beech/others 0.512
Fir/others 0.001
Fir/beech 0.005
Saproxylic fungi Snags 0.48 Beech/others 0.569
Fir/others 0.483
Fir/beech 0.902
Dendrothelms Snags 0.21 Beech/others 0.153
Fir/others 0.633
Fir/beech 1.000
Cracks Snags 0.007 Beech/others 0.42
Fir/others 0.007
Fir/beech 0.04
Total Snags 0.022 Beech/others 0.593
Fir/others 0.064
Fir/beech 0.022
Note: We distinguished three categories of tree species: beech alone, fir alone, and other tree spe-
cies pooled. Values are pvalues from a c2test.
Larrieu and Cabanettes 1437
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

two subsets by the threshold with the highest explanatory
power, each subset is evaluated for additional thresholds.
This method provides a decision tree with pvalues for one
or more critical thresholds. Based on 10 000 bootstrap sam-
ples, a confidence interval (at 80%) was calculated for all
thresholds. The calculations were performed on “presence–
absence”data using the add-on package “party”(Hothorn
and Hornik 2006).
Furthermore, to make an explicit link with forest manage-
ment practices, we also discuss our results by using “man-
agement diameter thresholds”that separate diameter
categories used by all managers in French forestry to de-
scribe and manage stands. These management diameter
thresholds are 20 cm ≤small tree ≤25 cm, medium tree
≤50 cm, large tree ≤70 cm, and very large tree >70 cm
(Bastien and Gauberville 2011).
Results
Tree species effect (live trees and snags)
Tree species affects the occurrence of different microhabitat
types and the total number of microhabitats per tree,
independently of the relative densities of the different tree
species
The species of the live trees significantly affected the fre-
quency of empty cavities, cavities with mould, dendrothelms,
and sap runs (Table 2; Fig. 2). The first three microhabitats
were mostly associated with beech, whereas sap runs were
exclusively associated with fir. Secondary species played an
intermediary role. The frequencies of saproxylic fungi, miss-
ing bark, and cracks were not related to tree species. On
snags, in contrast, fir carried a higher quantity of empty cav-
Fig. 2. Microhabitat frequency distributions split by species for live trees (European beech (Fagus sylvatica), silver fir (Abies alba), and
others) (2.1) and snags (2.2) represented as boxplots. Each bar corresponds to the two interquartiles Q1 and Q3 of the distribution. The hor-
izontal central line is the median. The length of the whiskers (broken lines) is 1.5 × (Q3 –Q1). Outlying points are not represented. Different
letters indicate significant differences between species. The deviations in the boxplots are based on average plot data.
1438 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

ities and cracks than beech. The frequencies of dendrothelms
and saproxylic fungi were not influenced by snag tree spe-
cies. Fir carried all microhabitat types (seven) whereas beech
carried only a maximum of six.
In total, beech carried significantly more microhabitats
than fir on live trees (Table 2; Fig. 2). Secondary species did
not differ from beech and also carried significantly more mi-
crohabitats than fir. For snags, only fir differed from beech.
Beech and secondary species carried more than five times
more microhabitats on live trees than on snags whereas fir
carried twice as many microhabitats on snags than on live
trees (Fig. 3).
Tree species affects the density of microhabitat-bearing live
trees
Seventy percent of the beech trees carried one or more mi-
crohabitats whereas we observed microhabitats on only 18%
of firs. Others species carried as many microhabitats as beech
(75%).
Live tree species affects the conditional probability of
occurrence of microhabitat types
The following microhabitat associations on a given tree
were significantly more probable than random for both beech
and fir (Table 3): (i) cavities with mould and missing bark
and (ii) empty cavities, dendrothelms, and cracks. Concern-
ing beech only, the following associations were found: (i)
cavities with mould, empty cavities. and saproxylic fungi
and (ii) saproxylic fungi and dendrothelms. On fir only, the
following associations were significant: (i) empty cavities
and saproxylic fungi, (ii) dendrothelms and missing bark,
(iii) missing bark and cracks, and (iv) sap runs on the one
hand and dendrothelms, missing bark, and cracks on the
other hand.
Tree diameter effects on microhabitat richness
Tree diameter affects the presence of each microhabitat type
We found one to three significant diameter thresholds for
each microhabitat type where the probability of its presence
varied significantly (Table 4). Thresholds were less numerous
in the fir data than in the beech data. Multiple thresholds cor-
respond to microhabitats likely to occur more than once per
tree (this was only the case for cavities, dendrothelms. and
missing bark).
Tree diameter affects the total number of microhabitats and
the number of microhabitat types
For beech (Fig. 4 and Fig. 5), the four first diameter
thresholds, which correspond to a gradual increase in the me-
dian of total numbers of microhabitats from zero to four,
were significant and spaced at regular intervals (of roughly
15 cm). There was no significant threshold beyond four mi-
crohabitats per tree (between four and 16). For fir (Fig. 4
and Fig. 5), only the higher diameter threshold (99 cm) cor-
responded to a significant increase of the median number of
microhabitats from zero to one. Beyond one microhabitat
(between one and seven), we did not detect any significant
threshold. For beech and fir, the first microhabitat occurred,
respectively, at 41 and 60 cm dbh (median values).
Concerning the total number of microhabitat types per tree
(Table 4): (i) for beech, only the two first thresholds (42 and
60 cm, similar to the thresholds for the total number of mi-
crohabitats per tree) significantly increased the median of the
number of microhabitat types per tree from zero to one and
then from one to two and (ii) for fir, the same threshold as
for the total number of microhabitats per tree (99 cm) signifi-
cantly increased the median from zero to one.
The proportion of microhabitat-bearing trees increased
markedly with increasing diameter as specified for the man-
agement categories (Table 5). However, the shape of this re-
lationship differed between tree species.
As dbh increased, the first microhabitat types that occurred
on beech were cavities with mould and missing bark on small
trees, then empty cavities, dendrothelms, and saproxylic fungi
on medium trees, and finally cracks on large trees and very
large trees. For fir, cavities and missing bark occurred on
small trees, sap runs and cracks appeared on medium trees,
and finally dendrothelms and saproxylic fungi on large trees
and very large trees.
On average, when all the diameter categories were repre-
sented, a beech–fir stand carried 71 microhabitats/ha and a
total of the seven microhabitat types. Large and very large
trees carried 48% of the microhabitats.
Discussion
Only a few papers describe the role of tree species and tree
diameter for distribution patterns of a set of microhabitats. In-
deed, in most cases, authors focused on only one microhabitat
type. The most documented microhabitats are cavities (e.g.,
McClelland and Frissel 1975; Cline et al. 1980; Mannan et al.
1980; Fan et al. 2003b, 2005; Drapeau et al. 2005; Remm and
Lõhmus 2011) and dendrothelms (Kitching 1971; Vaillant
1978; Schmidl et al. 2008). Due to this lack of available results,
our work provides new insights on microhabitat key factors.
Fig. 3. Total microhabitat distributions split by tree status and spe-
cies (European beech (Fagus sylvatica), silver fir (Abies alba), and
others). Each bar corresponds to the two interquartiles Q1 and Q3 of
the distribution. The horizontal central line is the median. The
length of the whiskers (broken lines) is 1.5 × (Q3 –Q1). Outlying
points are not represented. Letters indicate if differences between
species are statistically significant or not.
Larrieu and Cabanettes 1439
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

Role of the tree species on microhabitat distribution
Live beech and secondary species trees carried
microhabitats more often than live fir and certain
microhabitat types were associated with beech or fir,
sometimes exclusively
Tree species influences the number of different microhabi-
tat types per tree as well as the occurrence of each microha-
bitat, and beech carries microhabitats more often than fir
(Vuidot et al. 2011). In keeping with this, we showed that, at
a given diameter, the proportion of microhabitat-bearing trees
was mostly higher in beech and the other broadleaves species
than in fir. However, we found that fir carried more microha-
bitat types than beech.
Beyond these general trends, each microhabitat shows par-
ticular variations.
The probability that a tree carries a cavity varies with tree
species (Fan et al. 2003b). Remm and Lõhmus (2011)
showed that cavity density is higher in deciduous forests
than in mixed forests. McClelland and Frissel (1975), Cline
et al. (1980), Mannan et al. (1980), as well as Drapeau et al.
(2005) pointed out that cavities are rare in live conifers.
However, Bull et al. (1997) revealed the important role of
grand fir (Abies grandis (Douglas ex D. Don) Lindl.), west-
ern redcedar (Thuja plicata Donn ex D. Don), and western
larch (Larix occidentalis Nutt.) in the supply of cavities for
wildlife in forests dominated by conifers of the Columbia
River basin (United States). On live trees, we found that cav-
ities were mostly linked to beech. Trunk cavities result from
woodpecker excavations or fungi colonization. Empty cavities
are sometimes shaped by tree roots, more frequently when
the slope is steep. Woodpeckers prefer broadleaved species
Table 3. Independence tests between microhabitat frequencies (microhabitat type co-occurrences).
Cavities
with mould Empty
cavities Saproxylic
fungi Dendrothelms Missing bark Cracks Sap runs
Cavities with
mould
—<0.000 0.010 0.09 0.007 0.56
Empty cavities 0.44 —0.75 0.01 0.2 0.011
Saproxylic fungi 0.156 0.0015 —0.017 0.58 0.96
Dendrothelms 1 0.007 0.05 —0.98 0.35
Missing bark <0.000 0.71 0.61 0.034 —0.75
Cracks 0.147 0.0005 11 <0.000 —
Sap runs 0.23 0.71 1 0.011 0.0001 0.002 —
Note: Values are pvalues from a c2test for European beech (Fagus sylvatica) and silver fir (Abies alba). Values above the diagonal concern beech
and those below the diagonal concern fir. Results in bold indicate significant “positive”dependence between the two microhabitat types.
Table 4. Diameter thresholds and confidence intervals per tree species (European beech
(Fagus sylvatica) and silver fir (Abies alba)), on live trees, for the presence of each
microhabitat and their total or type frequencies.
Beech Fir
Microhabitat Threshold Confidence
interval Threshold Confidence
interval
Cavities with mould 41 33–51 65 47–81
63 53–78 87 81–113
79 75–88
Empty cavities 41 40–49 57 54–76
65 48–88 94 64–99
126 86–126
Saproxylic fungi 100 100–121 61 61–94
Dendrothelms 43 42–73 99 81–103
86 60–93
Missing bark 110 72–110 47 31–75
98 60–99
Cracks 72 72–81 100 86–102
Sap runs 76 74–102
Total microhabitat fre-
quency 42 41–60 47 47–57
60 44–60 81 67–81
73 60–79 99 81–99
89 80–107
Microhabitat type fre-
quency 42 40–43 50 47–57
60 59–67 81 80–99
86 64–116 99 94–103
Note: The division level of the dichotomous branching is indicated as follows: bold, first level
division in the set; regular, second level division of the two subsets; italic, third level division. All
thresholds were statistically significant at p< 0.05.
1440 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

to construct their nest cavities because of resin runs in coni-
fers (Cramp 1980). In deciduous riverine forests of Estonia,
cavity occurrence is determined by tree species, with a strong
preference for European aspen (Populus tremula L.) for
woodpecker-excavated cavities and black alder (Alnus gluti-
nosa (L.) Gaertn.) for cavities created by fungi (Remm et al.
2006).
We found that dendrothelms were mainly linked to beech
and that fir rarely carries dendrothelms, in agreement with
Vaillant (1978). Silver fir very exceptionally provides favor-
able conditions to create dendrothelms, probably because of
a centripetal deterioration (the external layers rot quite
quickly and fall off, while the heart resists much longer).
Vaillant (1978) indicated in addition that beech and linden
(Tilia spp.) carry dendrothelms more often than European
white birch. Kitching (1971) indicated that dendrothelms oc-
cur most abundantly in European beech, but also occur in
European ash, sycamore, birch, linden, and silver fir in the
British Isles. Dendrothelms have been observed in other tree
species: yellow poplar (Liriodendron tulipifera L.), horse
chestnut (Aesculus hippocastanum L.), European chestnut
(Castanea sativa Mill.), European hornbeam (Carpinus betu-
lus L.), durmast oak (Quercus petraea (Matt.) Liebl.), elms
(Ulmus spp.), London plane (Platanus ×hybrida Brot.),
northern red oak (Quercus rubra L.), and black alder (Vail-
lant 1978; Schmidl et al. 2008).
Although they are indicated on many genera present in
montane beech–fir forests such as Acer,Betula,Fagus,Frax-
inus,Populus,Quercus,Tilia,Ulmus, and Salix (Speight et
al. 2010), sap runs were strictly linked to fir trees in our
studied forests. We personally observed sap runs on beech in
the Vosges Mountains (the northeast of France), but very
rarely.
In spite of its thin bark, beech did not bear more missing
bark than the other tree species, contrary to what we ob-
served in managed beech–fir stands (Larrieu et al. 2011). In
subnatural stands, missing bark forms mainly as fall scars of
dying trees or stones on steep slopes. Missing bark that we
observed on common yew mainly resulted from elk (Cervus
elaphus Linnaeus, 1758) bark peeling occurring in winters
Fig. 4. Recursive partitioning tree for European beech (Fagus sylvatica) and silver fir (Abies alba) diameter at breast height (dbh) (cm) from
total microhabitat frequency data. The pvalues indicate the level of statistical significance of each node and nindicates the numbers of trees
per group. Only live trees. The difference between the two first box plots is linked to the number of outlying points that are merged.
Larrieu and Cabanettes 1441
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

with exceptionally great snowfalls (personal observation), in
spite of the strong toxicity of the yew bark for animals (Cor-
nevin 1887). We did not observe any significant association
between missing bark and saproxylic fungi, although these
bark injuries facilitate parasitic fungi colonization (Giro-
mpaire and Ballon 1992). This is all the more surprising for
beech because its wood has a low resistance to fungi attacks
(Keller 1986).
The association between empty cavities and saproxylic
fungi was not significant in our beech data, probably because
the main woodpecker species living in the study forests,
black woodpecker (Dryocopus martius Linnaeus, 1758), dig
holes in trees colonized by fungi but that are apparently
healthy (Zahner et al. 2012), contrary to the other wood-
pecker species that dig their cavity nest in wood showing
clear signs of decay (Cramp 1980).
The secondary species were rich in microhabitats but they
did not play an important role because of their low abun-
dance in the stands, except for certain taxa that are strictly
associated with a given tree species.
Beech and fir are complementary in their supply of micro-
habitats: beech provides quantity and fir may provide more
diversity. Furthermore, the communities associated with a
given microhabitat differ depending on whether the tree is a
conifer or a broadleaved tree (Cramp 1980; Meschede and
Heller 2003; Dajoz 2007; Speight et al. 2010).
Snags and live trees were complementary in the supply of
microhabitats (Fig. 2)
Vuidot et al. (2011) showed the important role that snags
play in the supply of tree microhabitats by finding that snags
carry almost twice as many microhabitats as live trees. How-
ever, Fan et al. (2003b) showed, on the contrary, that cavity
abundance in the old-growth hardwood forests of the east-
central United States is about twice the percentage for live
trees than for snags. In fact, even though snags concentrate
Table 5. Proportion of microhabitat-bearing trees (European beech (Fagus syl-
vatica) and silver fir (Abies alba)) per management diameter category.
% of microhabitat-bearing trees
Management diameter category Beech Fir
Small trees (20 ≤dbh < 25 cm) 35 6
Medium trees (25 ≤dbh < 50 cm) 43 9
Large trees (50 ≤dbh < 70 cm) 78 21
Very large trees
dbh ≥70 cm 92 32
dbh ≥89 cm 99
dbh ≥99 cm 70
Note: dbh, diameter at breast height.
Fig. 5. Distribution for European beech (Fagus sylvatica) and silver fir (Abies alba) of the diameter at breast height (dbh) per number of
microhabitats. The vertical lines correspond to the dbh thresholds that are statistically significant and correspond to an increase of a minimum
of one unit in the median value for microhabitat number.
1442 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

certain microhabitats, particularly for fir in our data, their rel-
atively low density could explain their low contribution in
most stands.
In our data, cracks were linked to snags rather than to live
trees. The cracks that we observed on fir snags were mostly
in the form of bark in the process of peeling. One or 2 years
after the death of a fir, bark fragments and peels off slightly.
Before falling on the ground, this space under the bark lasts
several years, offering shelter for medium-sized animals such
as bats. Beech bark is very adhesive and peels off several
years after the death of the tree, only in small fragments of-
fering very little shelters to crack-dwelling mammals. How-
ever, this shelter can be used by other taxa, such as arboreal
spiders (Chai and Liu 1998), flat bugs (Heiss and Pericart
2007), or beetles (Alexander 2002).
We also observed that snags carried sporophores of sap-
roxylic fungi much more often than live trees. That said, a
high hygrometry in dead wood is necessary for saprophyte
fungi to develop carpophores, while lignicolous saprophyte
fungi parasitizing weak trees are less dependent on atmos-
pheric conditions.
Role of tree diameter in microhabitat distribution
Winter and Möller (2008) found a strong link between the
number of microhabitats and the diameter of the host tree.
Vuidot et al. (2011) revealed that diameter is the main factor
influencing the number and probability of occurrence of cav-
ities, cracks, and missing bark. In Douglas-fir (Pseudotsuga
menziesii var. menziesii (Mirb.) Franco) stands, the abun-
dance of many bark microhabitats increased with tree diame-
ter and several bark microhabitats were not observed in the
smaller diameter classes (Michel et al. 2011). In a review,
Fan et al. (2003a) concluded that cavity occurrence is
strongly related to tree diameter. So, large trees are more fa-
vorable than smaller trees to cavity creation (DeGraaf and
Shigo 1985). Dufour (2003) also showed a positive correla-
tion between tree diameter and cavity occurrence. From
Kitching (1971), dendrothelm density increases in beech trees
with diameter above 50 cm.
Our results showed that the number of microhabitats per
tree, and also the number of microhabitat types per tree,
markedly increased with tree diameter. We found statistically
significant diameter thresholds that could be used by forest
managers. Most of these thresholds are situated above dbh =
50 cm (73% for beech and 90% for fir), which is used by for-
est managers, in most of the managed stands in the Pyrénées
Mountains, as the diameter at which it is economically opti-
mal to cut trees (ONF 2006). The harvest of the trees with
dbh ≥50 cm significantly reduces the number of microhabi-
tats per hectare by 48% and leads to the total elimination of
one microhabitat out of six for beech and two microhabitats
out of seven for fir.
Practical recommendations to improve sustainable forestry
Beech and fir play complementary roles in the supply of
microhabitat diversity. Secondary species (Tilia spp., Acer
spp., Betula spp., etc.) are often scarce but often bear micro-
habitats. Furthermore, the broadleaved–conifer mixture (i)is
the natural stand composition in most montane forests, (ii)fa-
cilitates management of a complex vertical structure that is
favorable to several taxa (e.g., birds, Orthoptera), and (iii)is
a means to stabilize income for the small forest estates be-
cause the markets fluctuate and are sometimes favorable to
broadleaved trees and sometimes favorable to conifers. In
theory, we could also manage monospecific stands within a
single landscape to achieve a high level of taxonomic biodi-
versity at the landscape scale. However, this approach is
likely to provoke problems for species that are strictly associ-
ated with either broadleaved trees or conifers and that have a
low dispersal capacity, as they may be unable to disperse
across such a fragmented landscape. Thus, we suggest con-
serving mixed stands.
Very large trees play a significant role because they host
all microhabitat types and the proportion of microhabitat-
bearing trees is very high. We consider that current manage-
ment diameter thresholds at 50 cm (lower limit of the large
tree category) and 70 cm (lower limit of the very large tree
category) are relevant with respect to microhabitats. Indeed,
by taking into account the confidence intervals (CI) at 80%,
we consider them equivalent to the significant thresholds
that we found: respectively, 42 cm (CI 41–60 cm), 60 cm
(CI 44–60 cm), and 73 cm (CI 60–79 cm) for beech and
47 cm (CI 47–57 cm) and 81 cm (CI 67–81 cm) for fir. We
observed other diameter thresholds at 89 cm for beech (CI
80–107 cm) and 99 cm for fir (CI 81–99 cm) that are signifi-
cant with respect to the number of microhabitats per tree and
also to the number of microhabitat types per tree. Therefore,
for forest management, we suggest creating a supplementary
diameter category (“largest trees”) with a lower limit at
dbh = 90 cm for beech and at dbh = 100 cm for fir to better
take into account the ecological role of these trees.
To promote the idea of “largest trees”category and thus to
conserve microhabitat-bearing trees, we recommend develop-
ing silvicultural practices that allow, at the stand scale, a pro-
portion of the trees to finish their complete natural cycle. The
modeling of the microhabitat distribution at the stand level
and the analysis of the abundant bibliography focusing on
green retention trees will help us to work towards a consen-
sual management strategy for mixed montane forests by fix-
ing a proportion of trees to be conserved according to the
stand characteristics.
The largest living trees seem to play a key role in all forest
ecosystems. For example, they are key features from tropical
(Grove 2002) to boreal (Martikainen et al. 2000) domains for
the invertebrate assemblages. According to our results, con-
serving large trees and the diversity of tree species should
help to manage taxonomic biodiversity in all forest ecosys-
tems at the condition of defining the diameter thresholds and
the role of each tree species in each ecosystem type.
Acknowledgements
This study was financed with the help of European funds
(FEDER) and French grants (State and Conseil régional de
Midi-Pyrénées). We wish to thank the members of the
Groupe d’Étude des Vieilles Forêts des Pyrénées team and
particularly Mathilde Harel for her help in the field. We
thank also Marc Deconchat, Christophe Bouget, and the three
anonymous reviewers of the manuscript for their constructive
criticisms and their suggestions for the improvement of the
manuscript. Thanks also to Mark Hewison for checking the
English.
Larrieu and Cabanettes 1443
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

References
Alexander, K. 2002. The invertebrates of living and decaying timber
in Britain and Ireland —a provisional annotated checklist. English
Nature Research Reports No. 467. Department for Environment,
Food and Rural Affairs and English Nature, Cranfield University
at Silsoe, Bedfordshire, U.K.
Bardat, J., Bioret, F., Botineau, M., Boullet, V., Delpech, R., Géhu, J.
M., Haury, J., Lacoste, A., Rameau, J.C., Royer, J.M., Roux, G.,
and Touffet, J. 2004. Prodrome des végétations de France. MNHN,
Paris, France.
Bastien, Y., and Gauberville, C. (Coordinators). 2011. Vocabulaire
forestier. Ecologie, gestion et conservation des espaces boisés.
IDF, Paris, France.
Bauhus, J., Puettmann, K., and Messier, C. 2009. Silviculture for old-
growth attributes. For. Ecol. Manage. 258(4): 525–537. doi:10.
1016/j.foreco.2009.01.053.
Blondel, J. 2005. Bois mort et à cavités: leur rôle pour l’avifaune
cavicole. In Bois morts et à cavités: une clé pour des forêts
vivantes. Edited by Vallauri et al. Tec et Doc Lavoisier, Paris,
France. pp. 137–142.
Branquart, E., Vandekerkhove, K., Bourland, N., and Lecomte, H.
2005. Les arbres sur-âgés et le bois mort dans les forêts de Flandre,
de Wallonie et du Grand-Duché de Luxembourg. In Bois mort et à
cavités: une clé pour des forêts vivantes. Edited by Vallauri et al.
Tec et Doc Lavoisier, Paris, France. pp. 19–29.
Bull, E.L., Parks, C.G., and Torgesen, T.R. 1997. Trees and logs
important to wildlife in the interior Columbia River basin. U.S.
For. Serv. Gen. Tech. Rep. PNW-391.
Chai, L.Y., and Liu, C.Y. 1998. Preliminary investigation report on
the overwintering arthropod species in the cracks on stems of apple
trees in the north of Henan. Acta Agric. Univ. Henanensis, 32(1):
52–54.
Cline, S.P., Berg, A.B., and Wight, H.M. 1980. Snag characteristics
and dynamics in Douglas-fir forests, western Oregon. J. Wildl.
Manage. 44(4): 773–786. doi:10.2307/3808305.
Cornevin, C. 1887. Des plantes vénéneuses et des empoisonnements
qu’elles déterminent. Institut National Agronomique, Paris,
France.
Cramp, S. (Coordinator). 1980. Handbook of the birds of Europe, the
Middle East and North Africa. The birds of the Western Palearctic.
Oxford University Press, New York.
Dajoz, R. 2007. Les insectes des forêts. Rôle et diversité des insectes
dans le milieu forestier. 2nd ed. Tec et Doc Lavoisier, Paris,
France.
DeGraaf, R.M., and Shigo, A.L. 1985. Managing cavity trees for
wildlife in the Northeast. U.S. Gen. Tech. Rep. NE-101.
Drapeau, P., Nappi, A., St Germain, M., and Angers, V.A. 2005. Les
régimes naturels de perturbations, l’aménagement forestier et le
bois mort dans la forêt boréale québécoise. In Bois mort et à
cavités: une clé pour des forêts vivantes. Edited by Vallauri et al.
Tec et Doc Lavoisier, Paris, France. pp. 46–54.
Dufour, D. 2003. Étude de l’influence du bois mort sur l’avifaune
cavernicole en forêt feuillue. Mémoire de fin d’études. FUSAGx,
Gembloux.
Fan, Z.F., Larsen, D.R., Shifley, S.R., and Thompson, F.R. 2003a.
Estimating cavity tree abundance by stand age and basal area,
Missouri, USA. For. Ecol. Manage. 179(1–3): 231–242. doi:10.
1016/S0378-1127(02)00522-4.
Fan, Z.F., Shifley, S.R., Spetich, M.A., Thompson, F.R., III, and
Larsen, D.R. 2003b. Distribution of cavity trees in midwestern
old-growth and second-growth forests. Can. J. For. Res. 33(8):
1481–1494. doi:10.1139/x03-068.
Fan, Z.F., Shifley, S.R., Spetich, M.A., Thompson, F.R., and Larsen,
D.R. 2005. Abundance and size distribution of cavity trees in
second-growth and old-growth central hardwood forests. North. J.
Appl. For. 22(3): 162–169.
Fritz, O., and Heilmann-Clausen, J. 2010. Rot holes create key
microhabitats for epiphytic lichens and bryophytes on beech
(Fagus sylvatica). Biol. Conserv. 143(4): 1008–1016. doi:10.1016/
j.biocon.2010.01.016.
Girompaire, L., and Ballon, P. 1992. Conséquences de l'écorçage du
Cerf élaphe dans le massif des Vosges alsaciennes [Results of
barking by red deer in the Alsacian Vosges]. Rev. For. Fr. 44(6):
501–511. doi:10.4267/2042/26354.
Gosselin, M., and Larroussinie, O. 2004. Biodiversité et gestion
forestière: connaître pour préserver. Synthèse bibliographique.
Cemagref, Nogent sur Vernisson, France.
Grove, S.J. 2002. Tree basal area and dead wood as surrogate
indicators of saproxylic insect faunal integrity: a case study from
the Australian lowland tropics. Ecol. Indic. 1(3): 171–188. doi:10.
1016/S1470-160X(01)00016-4.
Heiss, E., and Pericart, J. 2007. Hemiptères Aradidae Piesmatidae et
Dipsocoromorphes euro-méditerranéens. Fédération Française des
Sociétés de Sciences Naturelles, Paris, France.
Hothorn, T., and Hornik, K. 2005. The coin package, conditional
inference procedure in a permutation tes-framework, 0.4-1.
Available from http://CRAN.R-project.org.
Hothorn, T., and Hornik, K.Z.A. 2006. Party: a laboratory for
recursive part(y)itioning. Available from http://CRAN.R-project.
org.
Hothorn, T., and Zeileis, A. 2008. Generalizes maximally selected
statistics. Biometrics, 64(4): 1263–1269. doi:10.1111/j.1541-0420.
2008.00995.x. PMID:18325074.
Hothorn, T., Hornik, K., Van de Wiel, M.A., and Zeileis, A. 2006. A
Legosystem for conditional inference. Am. Stat. 60(3): 257–263.
doi:10.1198/000313006X118430.
Jonsell, M., and Weslien, J. 2003. Felled or standing retained wood —
it makes a difference for saproxylic beetles. For. Ecol. Manage.
175(1–3): 425–435. doi:10.1016/S0378-1127(02)00143-3.
Keller, R. 1986. Identification et classement des bois français.
ENGREF, Nancy, France.
Kitching, R.L. 1971. An ecological study of water-filled tree-holes
and their position in the woodland ecosystem. J. Anim. Ecol. 40
(2): 281–302. doi:10.2307/3247.
Larrieu, L., Cabanettes, A., and Delarue, A. 2011. Impact of
silviculture on dead wood and on the distribution and frequency of
tree microhabitats in montane beech–fir forests of the Pyrenees.
Eur. J. For. Res. 131(3): 773–786. doi:10.1007/s10342-011-0551-
z.
Larsson, T.B. (Compiler). 2001. Biodiversity evaluation tools for
European forests. Ecol. Bull. 50:75–81.
Lindenmayer, B., and Franklin, J.F. 2002. Conserving forest
biodiversity; a comprehensive multiscaled approach. Island Press,
Washington, D.C.
Lindenmayer, D.B., Margules, C.R., and Botkin, D.B. 2000.
Indicators of biodiversity for ecologically sustainable forest
management. Conserv. Biol. 14(4): 941–950. doi:10.1046/j.
1523-1739.2000.98533.x.
Lindenmayer, D.B., Franklin, J.F., and Fischer, J. 2006. General
management principles and a checklist of strategies to guide forest
biodiversity conservation. Biol. Conserv. 131(3): 433–445. doi:10.
1016/j.biocon.2006.02.019.
Lisiewska, M. 1992. Macrofungi on special substrates. In Fungi in
vegetation science. Edited by W. Winterhoff. Kluwer, the Nether-
lands. pp. 151–182.
Mannan, R.W., Meslow, E.C., and Wight, H.M. 1980. Use of snags
by birds in Douglas-fir forests, western Oregon. J. Wildl. Manage.
44(4): 787–797. doi:10.2307/3808306.
1444 Can. J. For. Res. Vol. 42, 2012
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.

Martikainen, P., Siitonen, J., Punttila, P., Kaila, L., and Rauh, J. 2000.
Species richness of Coleoptera in mature managed and old-growth
boreal forests in southern Finland. Biol. Conserv. 94(2): 199–209.
doi:10.1016/S0006-3207(99)00175-5.
McClelland, B.R., and Frissell, S.S. 1975. Identifying forest snags
useful for hole-nesting birds. J. For. 73(7): 414–417.
McElhinny, C., Gibbons, P., Brack, C., and Bauhus, J. 2005. Forest
and woodland stand structural complexity: its definition and
measurement. For. Ecol. Manage. 218(1–3): 1–24. doi:10.1016/j.
foreco.2005.08.034.
Meschede, A., and Heller, K.G. 2003. Écologie et protection des
chauves-souris en milieu forestier. Le Rhinolophe, 16.
Métailié, J.-P. 2001. Un patrimoine historico-environnemental: les
forêts pastorales dans les Pyrénées. In Proceedings of “126 ème
congrès des sociétés historiques et scientifiques”,Toulouse,
France. pp. 33–47.
Michel, A.K., and Winter, S. 2009. Tree microhabitat structures as
indicators of biodiversity in Douglas-fir forests of different stand
ages and management histories in the Pacific Northwest, USA.
For. Ecol. Manage. 257(6): 1453–1464. doi:10.1016/j.foreco.
2008.11.027.
Michel, A.K., Winter, S., and Linde, A. 2011. The effect of tree
dimension on the diversity of bark microhabitat structures and
bark use in Douglas-fir (Pseudotsuga menziesii var. menziesii).
Can. J. For. Res. 41(2): 300–308. doi:10.1139/X10-207.
Nascimbene, J., Marini, L., and Nimis, P.L. 2009. Influence of tree
species on epiphytic macrolichens in temperate mixed forests of
northern Italy. Can. J. For. Res. 39(4): 785–791. doi:10.1139/X09-
013.
ONF. 2006. Schéma régional d’aménagement des forêts pyrénéennes.
ONF, Direction territoriale Sud-Ouest. Toulouse, France.
Pardé, J., and Bouchon, J. 1988. Dendrométrie. 2nd ed. ENGREF,
Nancy, France.
Pénicaud, P. 2000. Chauves-souris arboricoles en Bretagne: typologie
de 60 arbres-gîtes et éléments de l’écologie des espèces observées.
Le Rhinolophe, 14:37–68.
Puumalainen, J., Kennedy, P., and Folving, S. 2003. Monitoring
forest biodiversity: a European perspective with reference to
temperate and boreal forest zone. J. Environ. Manage. 67(1): 5–14.
doi:10.1016/S0301-4797(02)00183-4. PMID:12659799.
R Development Core Team. 2007. R: a language and environment for
statistical computing. Available from http://www.R-project.org.
Ranius, T. 2002. Osmoderma eremita as an indicator of species
richness of beetles in tree hollows. Biodivers. Conserv. 11(5):
931–941. doi:10.1023/A:1015364020043.
Ranius, T., and Jansson, N. 2000. The influence of forest regrowth,
original canopy cover and tree size on saproxylic beetles
associated with old oaks. Biol. Conserv. 95(1): 85–94. doi:10.
1016/S0006-3207(00)00007-0.
Remm, J., and Lõhmus, A. 2011. Tree cavities in forests —the broad
distribution pattern of a keystone structure for biodiversity. For.
Ecol. Manage. 262(4): 579–585. doi:10.1016/j.foreco.2011.04.
028.
Remm, J., Lõhmus, A., and Remm, K. 2006. Tree cavities in riverine
forests: what determines their occurrence and use by hole-nesting
passerines? For. Ecol. Manage. 221(1–3): 267–277. doi:10.1016/j.
foreco.2005.10.015.
Schmidl, J., Sulzer, P., and Kitching, R.L. 2008. The insect
assemblage in water filled tree-holes in a European temperate
deciduous forest: community composition reflects structural,
trophic and physicochemical factors. Hydrobiologia, 598(1):
285–303. doi:10.1007/s10750-007-9163-5.
Snedecor, G.W., and Cochran, W. 1971. Méthodes statistiques.
[Translated from English by H. Boelle and E. Camhaji.] ACTA,
Paris, France.
Speight, M.C.D. 1989. Saproxylic invertrebrates and their conserva-
tion. Rep. 42. Council of Europe, Strasbourg, France.
Speight, M.C.D., Castella, E., Sarthou, J.-P., and Monteil, C. 2010.
Species accounts of European Syrphidae (Diptera). Syrph the Net,
the database of European Syrphidae. Syrph the Net Publications,
Dublin, Ireland.
Stanska, M., Rozwaka, R., and Kowalczyk, K. 2010. Dipoena torva
(Thorell, 1875), spider living on tree stems. Lesne Prace
Badawcze, 71(2): 201–205.
Tews, J., Brose, U., Grimm, V., Tielborger, K., Wichmann, M.C.,
Schwager, M., and Jeltsch, F. 2004. Animal species diversity
driven by habitat heterogeneity/diversity: the importance of
keystone structures. J. Biogeogr. 31(1): 79–92. doi:10.1046/j.
0305-0270.2003.00994.x.
Vaillant, F. 1978. Les Systenus et leur habitat dendrotelme (Dipt.
Dolichopodidae). Bull. Soc. Entomol. Fr. 83:73–85.
Vuidot, A., Paillet, Y., Archaux, F., and Gosselin, F. 2011. Influence
of tree characteristics and forest management on tree micro-
habitats. Biol. Conserv. 144(1): 441–450. doi:10.1016/j.biocon.
2010.09.030.
Winter, S., and Möller, G.C. 2008. Microhabitats in lowland beech
forests as monitoring tool for nature conservation. For. Ecol.
Manage. 255(3–4): 1251–1261. doi:10.1016/j.foreco.2007.10.029.
Zahner, V., Sikora, L., and Pasinelli, G. 2012. Heart rot as a key factor
for cavity tree selection in the black woodpecker. For. Ecol.
Manage. 271:98–103. doi:10.1016/j.foreco.2012.01.041.
Larrieu and Cabanettes 1445
Published by NRC Research Press
Can. J. For. Res. Downloaded from www.nrcresearchpress.com by INRA on 07/31/12
For personal use only.