Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
HAL Id: hal-03237154
https://hal.inrae.fr/hal-03237154
Submitted on 26 May 2021
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Distributed under a Creative Commons Attribution - NonCommercial - NoDerivatives| 4.0
International License
Co-occurrence patterns of tree-related microhabitats: A
method to simplify routine monitoring
Laurent Larrieu, Alain Cabanettes, Benoit Courbaud, Michel Goulard,
Wilfried Heintz, Daniel Kozák, Daniel Kraus, Thibault Lachat, Sylvie Ladet,
Jörg Müller, et al.
To cite this version:
Laurent Larrieu, Alain Cabanettes, Benoit Courbaud, Michel Goulard, Wilfried Heintz, et al.. Co-
occurrence patterns of tree-related microhabitats: A method to simplify routine monitoring. Ecological
Indicators, Elsevier, 2021, 127, �10.1016/j.ecolind.2021.107757�. �hal-03237154�
Ecological Indicators 127 (2021) 107757
1470-160X/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Co-occurrence patterns of tree-related microhabitats: A method to simplify
routine monitoring
Laurent Larrieu
a
,
b
,
*
, Alain Cabanettes
a
, Benoit Courbaud
c
, Michel Goulard
a
, Wilfried Heintz
a
,
Daniel Koz´
ak
d
, Daniel Kraus
e
, Thibault Lachat
f
, Sylvie Ladet
a
, J¨
org Müller
g
, Yoan Paillet
h
,
Andreas Schuck
i
, Jonas Stillhard
j
, Miroslav Svoboda
d
a
Universit´
e de Toulouse, INRAE, UMR DYNAFOR, Castanet-Tolosan, France
b
CNPF-CRPF Occitanie, Tarbes, France
c
University Grenoble Alpes, INRAE, UR LESSEM, France
d
Czech University of Life Sciences, Czechia
e
Bayerische Staatsforsten A¨
oR (BaySF), FB Rothenburg, Germany
f
Bern University of Applied Sciences, School of Agricultural, Forest and Food Sciences HAFL, & Swiss Federal Institute for Forest, Snow and Landscape Research WSL,
Birmensdorf, Switzerland
g
Bavarian Forest National Park & Department of Animal Ecology and Tropical Biology, University of Würzburg, Germany
h
INRAE, UR EFNO, Nogent sur Vernisson, France & Univ. Grenoble Alpes, INRAE, UR Lessem, France
i
European Forest Institute, Bonn Ofce, Germany
j
Swiss Federal Research Institute WSL, Birmensdorf, Switzerland
ARTICLE INFO
Keywords:
TreM monitoring
Biodiversity-friendly forest management
ABSTRACT
A Tree-related Microhabitat (TreM) is a distinct, well-delineated morphological singularity occurring on living or
standing dead trees, which constitutes a crucial substrate or life site for various species. TreMs are widely
recognized as key features for biodiversity. Current TreM typology identies 47 TreM types according to their
morphology and their associated taxa. In order to provide a range of resolutions and make the typology more
user-friendly, these 47 TreM types have been pooled into 15 groups and seven forms. Depending on the accuracy
required and the time available, a user can now choose to describe TreMs at resolution levels corresponding to
type, group or form. Another way to more easily record TreMs during routine management work would be to use
co-occurrence patterns to reduce the number of observed TreMs required. Based on a large international TreM
database (2052 plots; 70,958 individual trees; 78 tree species), we evaluated both the signicance and the
magnitude of TreM co-occurrence on living trees for 11 TreM groups. We highlighted 33 signicant co-
occurrences for broadleaves and nine for conifers. Bark loss, rot hole, crack and polypore had the highest num-
ber of positive co-occurrences (N =8) with other TreMs on broadleaves; bark loss (N =4) had the highest number
for conifers. We found mutually exclusive occurrences only for conifers: Exposed Heartwood excluded both
dendrotelm and sap run. Among the four variables we tested for their positive contribution to signicant co-
occurrences, tree diameter at breast height was the most consistent. Based on our results and practical consid-
erations, we selected three TreM groups for broadleaves, and nine for conifers, and formed useful short lists to
reduce the number of TreM groups to assess during routine forest management work in the eld. In addition,
detecting potential similarities or associations between TreMs has potential theoretical value, e.g. it may help
researchers identify common factors favouring TreM formation or help managers select trees with multiple
TreMs as candidates for retention.
* Corresponding author.
E-mail addresses: laurent.larrieu@inrae.fr (L. Larrieu), alain@o-art.fralai, n@o-art.fr (A. Cabanettes), benoit.courbaud@inrae.fr (B. Courbaud), michel.
goulard@inrae.fr (M. Goulard), wilfried.heinz@inrae.fr (W. Heintz), daniel.kraus@baysf.de (D. Kraus), thibault.lachat@bfh.ch (T. Lachat), sylvie.ladet@inrae.fr
(S. Ladet), joerg.mueller@npv-bw.bayern.de (J. Müller), yoan.paillet@inrae.fr (Y. Paillet), andreas.schuck@e.int (A. Schuck), jonas.stillhard@wsl.ch
(J. Stillhard), svobodam@d.czu.cz (M. Svoboda).
Contents lists available at ScienceDirect
Ecological Indicators
journal homepage: www.elsevier.com/locate/ecolind
https://doi.org/10.1016/j.ecolind.2021.107757
Received 29 May 2020; Received in revised form 23 April 2021; Accepted 25 April 2021
Ecological Indicators 127 (2021) 107757
2
1. Introduction
A Tree-related Microhabitat (TreM) is a distinct, well-delineated
morphological singularity occurring on living or standing dead trees,
which constitutes a crucial substrate for species (Larrieu et al., 2018).
Cavities, conks of lignivorous fungi and dead branches are examples of
TreMs. TreMs are widely recognized key features of biodiversity (Bütler
et al., 2013) and are useful indirect indicators for biodiversity (e.g.
Winter and M¨
oller, 2008; Paillet et al., 2018; Basile et al., 2020).
Therefore, Asbeck et al. (2021) have suggested using them as a moni-
toring tool to address biodiversity conservation issues in forest
ecosystems.
Larrieu et al. (2018) identied TreMs according to their morphology
and their associated taxa and allocated them into 47 types, the most
precise category, 15 groups, and seven forms, the more generic category,
by following a hierarchical way. Depending on the accuracy required
and the time available, a user can choose the suitable level to record
TreMs in the eld. For example, forest managers can record TreM forms
(e.g. cavities l.s.) during tree marking to estimate TreM diversity at the
stand scale whereas TreM groups (such as woodpecker breeding cav-
ities) could be applied in routine surveys and inventories like the na-
tional forest inventories, while elaborating management plans or for
Natura 2000 site evaluations. Researchers could use TreM types (e.g.
small, medium or large woodpecker breeding cavities) for more
exhaustive scientic surveys (Larrieu et al., 2018).
Another possible way to simplify and speed-up TreM recording
during routine management work would be to use non-random TreM co-
occurrence patterns (i.e. when TreM distribution on the tree is co-
dependent), to reduce the number of types to observe. In other words,
managers could use a shorter list of TreMs as a surrogate for the full list
that indicate the presence of further TreMs with a high probability.
However, TreM co-occurrence patterns are poorly known. Preliminary
studies revealed co-occurrence patterns at the tree scale for European
beech (Fagus sylvatica L.), pubescent oak (Quercus pubescens), holm oak
(Quercus ilex), silver r (Abies alba Mill.) and Douglas r (Pseudotsuga
menziesii Franco) (Larrieu and Cabanettes, 2012; Regnery et al., 2013a;
Winter et al., 2015; Puverel et al., 2019). However, these studies used
databases with a narrow geographical range and a limited number of
observed trees. Larrieu and Cabanettes (2012) highlighted, for example,
that bark loss and rot-holes co-occur in both beech and r while other
co-occurrences are tree-species specic: rot-holes and saproxylic fungi
co-occur only on beech, and dendrotelms and bark loss only on r.
Winter et al. (2015) showed TreM co-occurrence patterns for Douglas
r, e.g. for bark pockets and rot-holes. These results suggest that co-
occurrence patterns may be different between broadleaves and conifers.
Besides tree-species, other tree and stand features also inuence
these co-occurrence patterns. First, a greater diameter at breast height
(dbh) increases the probability of TreMs co-occurring on the same tree
(e.g. Winter and M¨
oller, 2008; Vuidot et al., 2011; Regnery et al., 2013a;
Larrieu et al., 2014; Courbaud et al., 2017; Asbeck et al., 2019).
Therefore, dbh is likely to be a crucial driver of co-occurrence patterns.
Second, pioneer species such as Salix spp., Populus spp. and Betula spp.
are relatively short-lived (Rameau et al., 1993) and individuals often
seem to simultaneously bear several TreM types early in their life cycles,
especially TreMs linked to reduced competitive ability (e.g. crown
deadwood) or early senescence (e.g. conks of polypores). In contrast,
small individuals of long-lived, shade-tolerant species such as Fagus
sylvatica and Quercus petraea for broadleaves, or Abies alba and Picea
abies for conifers (Rameau et al., 1993) rarely bear several TreM types
simultaneously (e.g. Larrieu et al., 2014). We therefore hypothesized
that tree species with distinct life cycles and succession dynamics would
show different co-occurrence patterns. Third, the CODIT system
(COmpartmentalization of Decay In Trees; Shigo, 1984) describes the
reaction of a tree following a trunk injury in order to limit the volume of
wood affected by pathogens. Tree species compartmentalize the decay in
unique ways and with a range of effectiveness, and exhibit a variety of
CODIT proles. While some species like the oaks can inhibit the spread
of pathogens within their organism by creating both chemical and
anatomical boundaries (Shigo, 1984), other tree species like the poplars
(Populus spp.) are less able to protect themselves and wood decay can
quickly affect a larger part of the trunk, thus creating, for example, wide
rot-holes. We hypothesized that the type of CODIT prole would affect
the development of saproxylic TreMs (i.e. those that involve decaying
wood) and would therefore inuence TreM co-occurrence. Fourth,
Winter et al. (2015) showed that management intensity has an impact on
TreM co-occurrence patterns for Douglas r. Although there are prop-
ositions for indices to assess management intensity (e.g., Kahl & Bauhus,
2014), the data required to calculate these indices are only seldomly
assessed during eld measurements. However, the time since the last
harvest is often available, at least broadly speaking, and can be used as a
proxy for management intensity to quantify its effect on TreM co-
occurrence.
Our study focused on living trees and co-occurrence patterns among
a set of TreMs at the tree scale. We expected that (i) co-occurrence
patterns of TreMs will differ between broadleaves and conifers, and
that (ii) tree dbh, time since last harvest (as a proxy for management
effect), succession dynamics of tree species and compartmentalization
capacity would drive co-occurrence patterns.
The practical outcome of this study was to develop short and
manageable lists to efciently record TreMs during routine eld visits.
We thus aim to provide forest managers with a practical tool to better
take into account the biodiversity associated with TreMs.
2. Materials and methods
2.1. Data
We collected data from a large range of temperate and boreal forests
from Northern Iran to Western Europe (Fig. 1; Table 1SM in Supple-
mentary Material). These forests cover a wide range of degrees of
naturalness, from regularly harvested stands to primeval forests (see e.g.
Commarmot et al., 2013; REMOTE project https://www.remoteforests.
org). The datasets from the managed stands cover various forest types
and silvicultural regimes and do not focus on TreM-rich stands only.
Each dataset provided was standardized according to the TreM typology
by Larrieu et al. (2018). However, since the typologies used by the eld
agents recording the TreMs differed slightly, we were not able to follow
exactly the same typology as Larrieu et al. (2018). In order to optimize
the available data, we designated eleven TreM subgroups (Table 1), very
close to the 15 TreM groups described by Larrieu et al.’s (2018), and
discarded several TreM types that were rarely recorded or recorded with
protocols that differed too much to be merged (see Table 1 for the TreM
types analyzed). In addition, TreMs belonging to the form “Epiphytic
and epixylic structures” (Larrieu et al., 2018) - namely bryophytes, li-
chens, lianas, ferns and mistletoes - were not included since they have
been rarely recorded. Finally, the eleven TreM subgroups used, hereafter
referred to simply as TreMs, encompassed 24 TreM types.
Overall 70,958 living trees (including 54,740 broadleaves, 16,218
conifers and 78 tree species) from 2,052 plots were used for the calcu-
lations. According to Larrieu et al., 2018, TreMs occur on both living and
standing dead trees. However, we analyzed co-occurrence in living trees
only since snags are not routinely included in tree-marking for
harvesting.
2.2. Analyses
All calculations were performed with R v3.0.0 (R Development Core
Team, 2018).
2.2.1. Presence/absence of non-random TreM co-occurrences
The data used was an absence/presence matrix for the eleven TreM
subgroups, with one row for each tree observed. To quantify the nature
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
3
of a co-occurrence, we counted for the corresponding pair of TreMs (e.g.
crack and polypore) the number of (1,1) in the data matrix for the col-
umns associated to this pair (in this example, the rst column for crack
and the second column for polypore) meaning that both TreMs are
present on the same tree; there is no co-occurrence if both TreMs are
absent or if only one is present. If this count is low when a TreM is
present (e.g. crack) but the other (e.g. polypore) is often absent, the co-
occurrence can be qualied as negative; if the count is high since the two
TreMs are often both present on the same tree, the co-occurrence can be
qualied as positive. To decide if the co-occurrence is signicantly
positive/negative or random, we compared this count with a similar one
calculated on a sample where pairs of 0 and 1 are obtained by resam-
pling on the 0/1 vectors observed for each TreM of the considered pairs
independently; so we did a resampling test for each TreM pair. This
resampling procedure (the R-script is provided in the supplementary
material) gave min, mean and max counts and when an observed count
fell inside the min–max interval, the corresponding pair was considered
to be random. Otherwise, the co-occurrence was considered non-random
(negative or positive). As we observed considerable heterogeneity of
presences and co-occurrences at the plot level, we did the resampling at
each plot level to tackle specic plot characteristics (number of pres-
ences for each TreM and number of trees involved). We ran 10,000 it-
erations of this resampling. We then calculated, for each pair of TreMs,
the difference between the observed co-occurrence frequency (i.e. the
count of trees in the database that bore the TreM pair) and the mean
frequency obtained by the 10,000 iterations of the resampling; we called
this difference “the magnitude of the co-occurrence“. The results for
broadleaves and conifers were treated separately. Graphical represen-
tations (Figs. 2 and 3) were inspired by those provided in the “co-occur”
package (Grifth et al, 2016).
2.2.2. Modeling non-random co-occurrences to highlight key factors
To analyze the effect of four explanatory variables (detailled below)
on the probability of co-occurrence for each pair of TreMs at the tree
scale, we used for the 42 non-random co-occurrences found (33 for
broadleaves and 9 for conifers) a logistic model with a binomial error
distribution and a logit link-function (GLMM approach, glmer function,
R-package lme4; Bates et al., 2015). The dependent variable was a binary
variable (presence/absence of co-occurrence) since at least one of the
TreMs in the respective combination was present. For each combination
of TreMs, we considered only the trees bearing at least one of the TreMs
since we were looking for co-occurrence. Excluding trees without TreMs
did not affect the binomial distribution of the variable. As explanatory
variables, we used: (i) tree dbh, (ii) time since the last harvest on the plot
(ve classes: 1- <15 years, 2- from 15 to 30y, 3 - from 30 to 50y, 4 - from
50 to 100y and 5 - unharvested for at least 100y), (iii) tree-dynamic status
(two categories: long-lived and shade-tolerant species, and pioneer/post-
pioneer together in order to balance tree numbers between categories
since post-pioneers were underrepresented in the dataset) according to
Rameau et al. (1993), and (iv) compartmentalization capacity according
to Shigo’s, 1984 concept (two classes: weak and high; Gilman, 2011;
Oven and Torelli, 1999; Schneuwly-Bollschweiler and Schneuwly, 2012;
Dujesiefken and Liese, 2015; Table 5SM). We used the plot identity as a
random-effect variable (i.e. (1|SitePlot) since several plots were nested
in the same site). Using tree-species succession status instead of simply
tree species allowed us to include rarely observed tree species and to
follow a functional approach to stand dynamics. It should be noted that
compartmentalization capacity was not pertinent for conifers in our study
since all the conifers we assessed have a high compartmentalization
capacity according to the literature. The number of trees distributed
among the ve value classes of time since the last harvest was sometimes
very irregular. When the number of trees in a class was too small or
equalled zero, the model could not be calculated correctly or did not
converge; the variable time since the last harvest was therefore removed
from the model. We systematically used VIF >3 (Zuur et al., 2010) as
the cut-off point to remove collinear variables (vif.mer function).
Thirty-six models were tested for each signicant co-occurrence and
that separately for broadleaves and conifers (31 for broadleaves and 5
for conifers). We then used the MuMIn package (Barton, 2019) to
calculate the Second-order Akaike Information Criterion and R
2
values
(r.squaredGLMM and r.squaredLR) for each of the 36 models. The sig-
nicance of each explanatory variable was tested with the Anova func-
tion (R-package car; Fox and Weisberg, 2011). The signicance of the
different levels of the factorized variables was calculated with the model.
Fig. 1. Map of the TreM datasets; symbols identify the datasets; numbers indicate the dataset IDs shown in Table 1SM.
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
4
Table 1
TreM forms, groups, subgroups (level created for this study to optimize available data) and types (from Larrieu et al., 2018, and Kraus et al., 2016 for the illustrations);
TreMs belonging to the form “Epiphytic and epixylic structures” (Larrieu et al., 2018) were not included.
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
5
avg function (R-package MuMIn) based on the calculation of the condi-
tional average model.
2.2.3. Selecting tree-related microhabitat combinations for monitoring
We used the results obtained on TreM co-occurrence to identify the
combinations of TreMs that gave the most complete representation of
TreM diversity while reducing the monitoring effort as much as possible.
For this purpose, we assigned a score to each combination of TreMs
(from one to ten TreMs) based on their co-occurrences with other TreMs
and the reliability of their observation. Our reasoning was that a TreM
strongly co-occurring with others could be a proxy indicator for a larger
group of TreMs and should therefore have a higher score. In addition, we
considered that TreMs with high observational reliability (i.e. no man-
agement, season or observer bias highlighted in literature) and a high
occurrence rate (i.e. above median) should have higher scores. We
calculated two scores for each combination: one for broadleaved trees
and one for coniferous trees. We assigned a score of 0 to any combina-
tion of TreMs that did not have any signicant co-occurrences with any
other TreMs. For the other combinations, the total score was the
weighted sum of ve criteria (see below). Each criterion had a value
Fig. 2. TreM co-occurrences for broadleaves (top panels) and conifers (bottom panels). The left panels show positive co-occurrences while the right panels show
exclusive ones. Although only the results with p <0.0001 were considered signicant, here we show the whole range of signicance levels for a broader overview of
TreM relationships. X-axis labels are abbreviations of the full names of the TreM-subgroups indicated along the Y-axis, i.e. RH: rot hole, De: dendrotelm, RC: root
concavity, BL: bark loss, EH: exposed heartwood, Cr: crack, CD: crown deadwood, BC: burr canker, Po: polypore. Since plots with mixed stands were counted twice, i.e. for
both broadleaves and conifers, total plot number exceeds the total indicated in Table 1SM.
Fig. 3. Magnitude of TreM co-occurrences for broadleaves (left panel) and conifers (right panel). Expected co-occurrence (X-axis) corresponds to the average number
of co-occurrences between the 2 TreMs, resulting from a random reallocation of the TreMs observed on each plot over all the trees belonging to that plot. Each dot
corresponds to a co-occurrence between 2 TreMs (55 possible pairs). Values along the axes correspond to the number of trees bearing a TreM pair in the whole dataset
(for broadleaves and conifers, 1,859 and 902 plots respectively). The dashed black lines delimit the range of values (min and max) calculated for the random
assumption (p =0.0001; see Material and method section). Only the strongest 10% of the magnitudes are identied (see Tables 4SM and 5SM for magnitude values):
CD: crown deadwood, BL: bark loss, Cr: crack, RH: rot hole, BW: breeding woodpecker hole, Po: polypore, EH: exposed heartwood, SR: sap run, De: dendrotelm, RC:
root concavity.
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
6
between 0 and 1, which reected the mean of the values for each TreM
in the combination (see Table 4SM). We weighted the values to obtain a
clear hierarchy among the criteria. Our rst criterion was non-depen-
dence on management, with a weight of 5. We considered this criterion
the most important of all since harvesting can drastically modify both
TreM occurrence (Larrieu et al., 2012; Lassauce et al., 2013; Paillet et al.,
2017) and their co-occurrence (Winter et al., 2015). The second crite-
rion was co-occurrence with TreMs not included in the combination, with a
weight of 4. We considered this criterion highly important since our
main aim was to reduce the number of TreMs to observe. The third
criterion was the number of occurrences in the database, with a weight of
3. This criterion focused on the most frequent TreMs to increase the
probability of observing at least one TreM on the short list whatever the
stand; this criterion is also important in terms of data collection and
training practitioners in TreM observation. The fourth criterion was
TreM life-span (i.e. permanent versus temporary) with a weight of 2. This
criterion was deemed somewhat less important even though TreM
longevity makes year-round observation possible. The fth criterion was
observer effect (according to Paillet et al., 2015) with a weight of 1. We
included this criterion because the presence of an observer effect in
certain eld records could lead to stand mischaracterization. We
selected the best TreM combinations to create short-lists encompassing
from one to ten TreMs. We then analyzed how the total weighted scores
of these short lists varied as a function of the number of TreMs making
up the list, for conifers and broadleaves separately.
3. Results
3.1. Non-random TreM co-occurrences
We highlighted 33 non-random positive co-occurrences for broad-
leaves while we found seven positive and two mutually-exclusive co-
occurrences for conifers (p <0.0001; Fig. 2). All the TreMs on broad-
leaves showed at least one signicant co-occurrence with another TreM.
Burr canker never co-occurred with any other TreMs on conifers. Bark
loss, rot-hole, crack and polypore showed the highest number of positive
co-occurrences with other TreMs for broadleaves (N =8) as bark loss (N
=4) did for conifers. We found signicant mutually-exclusive co-oc-
currences only for conifers: exposed heartwood with dendrotelm and sap
run. Six co-occurrences were shared by broadleaves and conifers: Crown
deadwood with polypore, bark loss with sap run, bark loss with crack, root
concavity with crown deadwood, rot hole with bark loss, and nally
breeding woodpecker hole with bark loss. Dendrotelm with crack was the
only co-occurrence specic to conifers.
We found a wide range of magnitude values, mainly for broadleaves
(Fig. 3). The strongest magnitudes were observed for the co-occurrences
of bark loss with crack for broadleaves and breeding woodpecker hole with
bark loss for conifers.
3.2. Key factors for high-magnitude non-random co-occurrences
Among a set of four variables tested for their positive contribution to
signicant co-occurrences (i.e. dbh, time since the last harvest, tree-species
category in dynamic succession and compartmentalization capacity of the
tree species), dbh was the variable with the highest consistency. It
showed a signicant (p <0.05) effect on the likelihood of two TreMs co-
occurring for 88% and 71% of the high magnitude (i.e. the 10% stron-
gest magnitudes) non-random co-occurrences for broadleaves and co-
nifers respectively (Tables 2SM and 3SM). Longer time spans without
harvesting (time classes 4 and 5, both above 50 years) favored co-
occurrences between breeding woodpecker hole and both bark loss and
crown deadwood, rot hole and crown deadwood, bark loss and both exposed
hardwood and crack for broadleaves, and co-occurrences between bark
loss and crack for conifers. For broadleaves, a shorter time without
harvesting (time class 2, 15–30 years) showed a positive effect on the co-
occurrence of rot hole with root concavity and bark loss with polypore,
while it had a signicant negative effect on co-occurrences between
crown deadwood and polypore, bark loss and crack, root concavity and
crown deadwood. Tree-species category and compartmentalization capacity
were sometimes collinear. Therefore, we were unable to evaluate their
contribution for all the co-occurrence combinations. However, for
broadleaves, tree-species category in dynamic succession did have a sig-
nicant, though sometimes opposite, effect for half of the co-occurring
pairs. For example, breeding woodpecker hole had a mainly positive ef-
fect – often of high magnitude, as when it was combined with exposed
heartwood, but a negative effect among long-lived and shade-tolerant
species when it was combined with polypores. Compartmentalization ca-
pacity had a signicant effect for pairs including rot hole.
3.3. Selecting tree-related microhabitat assemblages for monitoring
The relationship between the scores of the best TreM combinations
and the number of monitored TreMs showed a bell-shaped curve both
for conifers and broadleaves (Fig. 4). The maximum score was reached
quickly for broadleaves, at three TreMs, whereas it was reached much
more slowly for conifers, requiring nine TreMs (Table 2).
3.3.1. Broadleaves
For broadleaves, several combinations of only three TreMs showed
signicant co-occurrences with all the unmonitored TreMs. The assem-
blage crack +burr-canker +crown deadwood had the highest score
(Table 2); it displayed strong co-occurrence with unmonitored TreMs
and it involved TreMs with relatively frequent occurrences, low sensi-
tivity to management, long life span and low observer effects. This
combination score was very similar to the scores obtained by combi-
nations of four TreMs.
3.3.2. Conifers
For conifers, the score increased slowly with the number of moni-
tored TreMs in the combination because co-occurrences were infre-
quent. Adding a new TreM to the combination did not result in a strong
increase in co-occurrence with the remaining TreMs. The maximum
score was reached for a combination of nine TreMs: breeding woodpecker
hole +exposed heartwood +polypore +root concavity +rot hole +sap run
in addition to the three TreMs selected above for broadleaves (Table 2).
Adding dendrotelm to this combination decreased the overall score
because of the sensitivity of dendrotelm to management.
4. Discussion
Based on a large-scale database combining 11 TreM groups, we
showed signicant high-magnitude co-occurrences between TreMs at
the tree scale. We also showed that these co-occurrences are more
frequent on broadleaves than on conifers, and that dbh had a consistent
effect on the co-occurrence, while life traits of trees (i.e. category in
dynamic succession and compartmentalization capacity) and forest
management had a lesser effect.
4.1. Co-occurrence between TreMs vary with tree species groups
Most of the co-occurrences between TreMs on broadleaves are likely
due to the propensity of some species to form certain types of micro-
habitats (e.g. crown deadwood in oaks, Paillet et al., 2019) that may, in
turn, lead to the occurrence of other TreMs linked to the same process (in
this case: crack and bark loss; Larrieu, 2014). More generally, the vital
status of a given tree is known to be a strong driver of microhabitat
dynamics (e.g. Vuidot et al., 2011; Larrieu and Cabanettes, 2012). We
can assume that when the vitality of a tree decreases, TreMs linked with
the decaying process appear (i.e. saproxylic TreMs). The patterns of co-
occurrence we observed in this study, where we worked only with living
trees, conrm this assumption. We found mutually-exclusive co-occur-
rences for conifers only. This is in accordance with the results of Winter
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
7
et al. (2015) who only found a slightly exclusive co-occurrence between
bark pockets and broken tree parts on Douglas r (Pseudotsuga menziesii
Mirb. Franco) while studying TreM co-occurrence patterns in European
Beech (Fagus sylvatica L.) and Douglas r forests. Although our TreM
group exposed heartwood is quite similar to the group broken tree parts
used by Winter et al. (2015), we could not consolidate the two results
since we were not able to analyze bark pockets through our database. At
the same level of signicance (p <0.0001), our results were in accor-
dance with co-occurrences highlighted by Larrieu and Cabanettes
(2012) for (i) European beech, between rot hole and root concavity, and
(ii) Silver r, between bark loss and crack, and sap run and rot hole.
4.2. Tree diameter mainly drives TreM co-occurrence patterns
The effect of tree dbh on TreM co-occurrence probability had not
been evaluated before the present study. For our dataset, dbh was the
most relevant variable explaining co-occurrence patterns, both for
broadleaves and conifers. Generally, the larger the tree, the greater
variety of TreMs it bears (e.g. Larrieu and Cabanettes, 2012; Paillet
et al., 2019). Thus, a larger dbh favors TreM co-occurrence both “by
sampling” (larger trees have more chances to have several types of mi-
crohabitats) and by ontogeny (the same processes apply for different
TreMs). Dbh is used as a proxy of tree-age since it is easier to record in
the eld than age. However, several TreMs might be linked with age
rather than with dbh since they are more likely to occur over a long
period, e.g. polypores (Boddy, 2008). Certain TreMs such as lightning
scars might benet from both age and dbh since lightning strikes on
trees are quite rare in temperate forests and a large dbh often accom-
panies tree dominance and canopy exposure. Finally, TreMs such as
woodpecker breeding holes require trees large enough to provide
adequate trunk volume (Rolstad et al., 2000). Moreover, the ontogenic
stage of the tree (i.e. juvenile, adult, mature and senescent, based on the
number of replications of the species-specic architectural unit, which is
only slightly correlated to age) can lead to TreM occurrence since e.g.
the senescent stage is characterized by the presence of sun-lit dead
branches. Therefore, the link between dbh and TreM co-occurrence
might actually hide the real links with age or ontogenic stage
(Rutishauer et al., 2011). For the few TreM co-occurrences that could be
assessed, we found mostly positive effects for a longer time without
harvesting, though there were three signicant negative effects for
TreMs that are rare in managed stands, such as polypore and crack.
Management might reduce co-occurrence for these TreMs in several
ways: (i) applying a low-rotation dbh is likely to reduce the number of
large trees in the stand (e.g. Asbeck et al., 2019); (ii) TreM-bearing trees
are often marked to be cut, thus reducing their proportion (Winter and
M¨
oller, 2008, Larrieu et al., 2012), particularly in broadleaf-dominated
Fig. 4. Scores of monitored TreM combinations. The score of the best combination of monitored TreMs is shown for different numbers of monitored TreMS, for
broadleaves (solid line) and conifers (dotted line).
Table 2
Best TreM assemblages revealing potential candidates for a short list of TreMs
for monitoring as a proxy for the set of the 11 TreMs studied; for the calculation
of the combined score, see Materials and Methods.
Broadleaves Number of
TreMs
observed
Best assemblages (i.e. highest total
scores)
Total
score
2 crack +polypore 4.911
3 crack +burr canker +crown deadwood 7.004
4 crack +burr canker +crown deadwood
+exposed heartwood
7.003
5 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
6.898
6 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
+sap run
6.672
7 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
+sap run +polypore
6.662
8 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
+sap run +polypore +breeding
woodpecker hole
6.733
9 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
+sap run +polypore +breeding
woodpecker hole +rot hole
6.694
10 crack +burr canker +crown deadwood
+exposed heartwood +root concavity
+sap run +polypore +breeding
woodpecker hole +rot hole +
dendrotelm
5.212
Conifers 5 burr canker +crack +crown deadwood
+exposed heartwood +bark loss
4.425
6 burr canker +crack +exposed
heartwood +polypore +root concavity
+bark loss
4.532
7 burr canker +crack +breeding
woodpecker hole +crown deadwood +
exposed heartwood +rot hole +sap run
5.759
8 burr canker +crack +breeding
woodpecker hole +exposed heartwood
+polypore +root concavity +rot hole
+sap run
6.304
9 burr canker +crack +breeding
woodpecker hole +exposed heartwood
+polypore +root concavity +rot hole
+sap run +crown deadwood
6.353
10 burr canker +crack +breeding
woodpecker hole +exposed heartwood
+polypore +root concavity +rot hole
+sap run +crown deadwood +
dendrotelm
5.124
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
8
stands (Larrieu et al., 2014); and (iii) managers tend to eliminate trees
with trunk-borne TreMs, which strongly reduce the tree commercial
value, as is the case for polypores, since a conk indicates that the wood is
already decaying (Stokland et al., 2012) and is therefore unsuitable for
timber. All of these choices lead to a reduction in TreM diversity
(Larrieu et al., 2012) and thus, mechanistically, the the reduction of co-
occurrences. Winter et al. (2015) showed that management affects TreM
occurrence patterns in European Beech (Fagus sylvatica L.) and Douglas
r (Pseudotsuga menziesii Mirb. Franco) forests by strongly reducing the
number of signicant co-occurrences. Furthermore, they found that
management promotes co-occurrences not observed in more natural
unmanaged forests; these combinations include cavities and broken tree
parts or bark pockets and bark injuries for beech, and broken tree parts
and bark injuries for Douglas r (see Winter et al., 2015 for TreM
denitions). In our case, it seems that management – through time since
the last harvest – has relatively moderate effects.
To date, no studies have investigated the relationship between tree-
species life-traits and TreM co-occurrence. It is well known that all
woodpeckers excavate their breeding cavities in the part of the trunk
degraded by fungi (Schepps et al., 1999; Jackson and Jackson, 2004;
Matsuoka, 2008; Zahner et al., 2012). The Black woodpecker (Dryocopus
martius) may even trigger the colonization by the fungi, thus facilitating
cavity drilling (see e.g. Puverel et al., 2019). Therefore, woodpecker
breeding cavities and fungi are linked through functional processes.
However, conks of fungi may appear several years after the tree has
actually been colonized by the mycelium (Conner et al., 1976; Jackson
and Jackson, 2004) and thus shift the visible co-occurrence in time.
Pioneer broadleaves are often used by woodpeckers as breeding trees,
particularly birches (Betula spp.; e.g. Pakkala et al., 2019) and poplars
(Populus spp.; e.g. Hebda et al., 2017). This may be due to the fact that
they are susceptible to saproxylic fungi rather early in their life cycle.
They also have a weak compartmentalization capacity (see Table 5SM),
which allows the fungus, once introduced, to spread quickly inside the
wood (Kahl et al., 2017). These traits favor the creation of a large vol-
ume of favorable substrate for breeding holes.
For a tree, investing in defence against pathogens is a trade-off with
growth speed and life span (Loehle, 1988). Fast-growing broadleaved
pioneers, for example, are good at colonizing open areas and competing
with low ground/shrub vegetation, but they generally have a weak
compartmentalization capacity since their investment in defence bar-
riers is very low and they are short-lived (Morris et al., 2016). The strong
relationship between tree-species succession dynamics and compart-
mentalization capacity explains why we often found colinearity between
these variables in our models. We revealed a signicant positive effect of
a weak compartmentalization capacity for the TreM pair rot hole/bark
loss. This suggests that most bark loss leads to the development of a rot-
hole for broadleaved pioneers, since these trees are not able to isolate
the wound area effectively. However, another process might be involved
since we found no signicant difference between pioneers and long-
lived and shade-tolerant tree species for the co-occurrence of this
TreM pairs as a function of dbh (Fig. 1 SM).
5. Conclusion: Limitations and elds of application
5.1. TreM co-occurrences as clues to better understanding TreM
formation processes
Studies viewing TreMs as key features for biodiversity at the stand
level are quite recent (Winter and M¨
oller, 2008). Current knowledge of
TreM formation and dynamics is limited and is based only on expertise
or cross-sectional (synchronic) data (see e.g. Courbaud et al., 2017).
However, there is no doubt that certain TreMs are linked through dy-
namic processes; for example, we found a positive co-occurrence be-
tween bark loss and rot hole both for broadleaves and conifers. Indeed,
bark loss will irrevocably evolve towards a rot hole if the wound favors
infection by wood-decomposing fungi and if the bearing tree is not able
to overlay the wound. Although TreM life spans may be very different
(ranging from a few weeks for myxomycetes to several decades for large
rot-holes), TreMs evolve slowly on average. Therefore, obtaining
diachronic data would require both permanent plots dedicated to this
topic and long-term funding for periodic monitoring hard to imagine
given the area and time required to obtain enough trees in a dbh range
equivalent to the one in our synchronic data. In this context, TreM co-
occurrence patterns might help to identify certain TreM dynamic re-
lationships (e.g. shift of dominance between two TreMs when tree-dbh
increases), or at least to identify local conditions that lead to the for-
mation of different TreMs on a given tree. These patterns could guide
specic studies, as improved knowledge of TreM dynamics may lead to
better management of a continuous TreM supply, both at the stand and
forest levels.
Despite the large number of trees in our database, it was not possible
to perform analyses at the TreM-type level since some TreMs were rarely
recorded. Furthermore, due to the heterogeneity of the TreM denitions
in the available datasets, we were also unable to analyze all the TreM
groups sensu Larrieu et al. (2018). Further research should analyze co-
occurrence patterns on standing dead trees since they signicantly
bear TreMs (Larrieu and Cabanettes, 2012; Vuidot et al., 2011; Regnery
et al., 2013a; Paillet et al., 2017). However, thanks to the size of our
database and the conservative approach we used (signicance with a p-
value <0.0001), our results can benet forest managers during routine
practices (tree-marking, inspection visits or plot assessments) or can
provide input for management planning based on sound and robust
scientic data.
5.2. A short list of TreMs for monitoring based on co-occurrence patterns
Monitoring based on a limited number of TreMs inevitably di-
minishes the practitioner’s ability to precisely assess the full TreM di-
versity in a forest. However, the best-performing TreM lists we selected
(three TreMs for broadleaves and nine TreMs for conifers) are charac-
terized by a strong co-occurrence with unmonitored TreMs. The pres-
ence of these TreMs in a forest therefore indicates that TreM richness is
probably high in this forest.
Firstly, knowing co-occurrence frequencies can help managers
develop efcient strategies for the retention of TreM-bearing trees
(habitat-trees; Bütler et al., 2013). Indeed, if co-occurrence frequency is
high, managers may be able to conserve a wide range of TreM types
simply by protecting the habitat-trees which bear multiple TreMs. In
contrast, if co-occurrence frequency is low, managers must retain
different habitat-trees for each TreM type, or to target the habitat-trees
bearing the rarest TreMs.
Secondly, since practitioners often have limited time for tree
marking, reducing the number of TreMs to be monitored could help
forest managers incorporate TreM observation and recording, a time-
consuming process (Cosyns et al., 2019). Since every TreM has a mini-
mum required size for recording (Larrieu et al., 2018), shorter TreM
lists/TreM guides with only a few threshold size values to remember
may make TreM assessment more efcient and may also lead to higher
acceptance to do such assessments. However, if practitioners use a TreM
short-list rather than a more comprehensive one, they must be careful
not to reduce the time they dedicate to observing the trees. Since the
listed TreMs are not only important per se but are also surrogates for
other TreMs, missing them inadvertently could lead to signicant in-
formation loss and thus a higher likelihood that such a tree may be
marked for removal. Paillet et al. (2015) highlighted a signicant fa-
miliarity (i.e. the observer has already observed the TreM) observer ef-
fect for cracks, for instance, thus highlighting the need for careful
training. Using a short list of TreMs for monitoring does not justify
reducing the overall number of TreM-bearing trees to retain while
marking, since the density of habitat-trees is an important driver for
species richness and for species composition for taxa such as saproxylic
beetles, polypores, hoveries, bats and birds (Paillet et al., 2018;
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
9
Regnery et al., 2013b; Bouget et al., 2013; Winter and M¨
oller, 2008;
Larrieu et al., 2019). Furthermore, actively selecting trees bearing
different TreMs is the most efcient way to ensure TreM diversity at the
stand scale (Asbeck et al., 2020).
The higher number of TreMs selected for conifers as compared to
broadleaves was mainly due to the lower number of co-occurrence pairs
observed on conifers. Breeding woodpecker hole was selected in our best
TreM combination for conifers. This TreM is often targeted for biodi-
versity conservation or integrative forest management approaches.
Thus, many forest managers are used to assessing this TreM in their daily
work. Furthermore, breeding woodpecker hole is often deemed a keystone
feature for biodiversity since a wide range of taxa uses or depends on this
TreM (Bobiec et al., 2005; Roberge and Angelstam, 2004). Crown
deadwood was selected for both broadleaves and conifers. This form of
deadwood is crucial for numerous saproxylic taxa (e.g. Bouget et al.,
2011) and is very rarely assessed, even during deadwood monitoring
(see e.g. Larrieu et al., 2019). Although bark loss had a high number of
positive co-occurrences with other TreMs for both conifers and broad-
leaves, it was not selected in our procedure, partly because we assigned a
strong weight to the variable management effect, and this negatively
inuenced the score for bark loss. Indeed, bark loss can be a common
feature resulting from timber harversting (Larrieu et al., 2012) since
trees are often wounded along skidding trails . This may then lead to a
local overestimation of occurrence of other TreMs. Moreover, Paillet
et al. (2015) highlighted a double observer effect for bark loss (both
recording duration and familiarity effects).
For studies that aim at analyzing the relationship between TreMs and
biodiversity at the stand level, we recommend using the TreM-type level
to ensure a precise description of the stand; indeed, local conditions can
inuence co-occurrence patterns and there are many highly specialized
species whose habitat cannot be characterized by a group of TreMs.
CRediT authorship contribution statement
Laurent Larrieu: Conceptualization, Methodology, Resources,
Writing - original draft. Alain Cabanettes: Methodology, Formal anal-
ysis, Writing - original draft. Benoit Courbaud: Methodology, Software,
Formal analysis, Resources, Writing - original draft. Michel Goulard:
Methodology, Formal analysis, Writing - original draft. Wilfried Heintz:
Data curation. Daniel Koz´
ak: Resources, Writing - review & editing.
Daniel Kraus: Resources, Writing - review & editing. Thibault Lachat:
Resources, Writing - review & editing. Sylvie Ladet: Data curation,
Writing - review & editing. J¨
org Müller: Resources, Writing - review &
editing. Yoan Paillet: Resources, Writing - review & editing. Andreas
Schuck: Resources, Writing - review & editing. Jonas Stillhard: Re-
sources, Writing - review & editing, Resources, Writing - review &
editing. Miroslav Svoboda: Resources, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgments
Part of this research was funded by the French Ministry in Charge of
Ecology (Convention Cemagref-DEB (MEEDDAT), Action GNB; the
program “Biodiversité, Gestion Foresti`
ere et Politiques Publiques”
(BGF), convention GNB 10-MBGD-BGF-1-CVS-092, no CHORUS 2100
214651. The Uholka dataset was collected in a joint project of the Swiss
Federal Institute for Forest, Snow and Landscape Research WSL, the
Carpathian Biosphere Reserve and the Ukrainian National Forestry
University, with nancial support from the State Secretariat for Educa-
tion, Research and Innovation SERI, Switzerland. The REMOTE dataset
was collected within the MSMT projects (CZ.02.1.01/0.0/0.0/16_019/
0000803 and LTT20016). We thank Sergey Zudin (EFI) for data man-
agement and validation of the I +TreM Dataset as well as the German
Federal Ministry of Nutrition and Agriculture (BMEL) for funding the
projects Integrate +and Informar, which allowed EFI to build this
extensive I +TreM Dataset. We also express our sincere thanks to all the
data providers and numerous collaborators who recorded data in the
eld. We thank Vicki Moore for reviewing the English in the previous
manuscript and Fr´
ed´
eric Gosselin for his valuable comments on the rst
draft.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.ecolind.2021.107757.
References
Asbeck, T., J. Grossmann, Y. Paillet, N. Winiger, and J. Bauhus. 2021. The Use of Tree-
Related Microhabitats as Forest Biodiversity Indicators and to Guide Integrated
Forest Management. Current Forestry Reports 7: 59–68. DOI:10.1007/s40725-020-
00132-5.
Asbeck, T., Pyttel, P., Frey, J., Bauhus, J., 2019. Predicting abundance and diversity of
tree-related microhabitats in Central European montane forests from common forest
attributes. FEM 432, 400–408.
Asbeck, T., Messier, C., Bauhus, J., 2020. Retention of tree-related microhabitats is more
dependent on selection of habitat trees than their spatial distribution. Eur. J. Forest
Res. https://doi.org/10.1007/s10342-020-01303-6.
Barton, K., 2019. Package ‘MuMIn’. Model selection and model averaging base on
information criteria. Version 1.43.15. Vienna, Austria: R Foundation for Statistical
Computing.
Basile, M., Asbeck, T., Jonker, M., Knuff, A.K., Bauhus, J., Braunisch, V., Mikusi´
nski, G.,
Storch, I., 2020. What do tree-related microhabitats tell us about the abundance of
forest-dwelling bats, birds, and insects? J. Environ. Manage. 264, 110401 https://
doi.org/10.1016/j.jenvman.2020.110401.
Bates, D., Maechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-effects models
using lme4. J. Stat. Softw. 67 (1), 1–48. https://doi.org/10.18637/jss.v067.i01.
Bobiec, A., Gutowski, J.M., Audenslayer, W.F., 2005. The Afterlife of the Tree. WWF
Poland, Warszawa.
Boddy, L. (Ed.), 2008. Ecology of Saprotrophic Basidiomycetes. Academic Press, Elsevier,
London.
Bouget, C., Brin, A., Brustel, H., 2011. Exploring the “last biotic frontier”: are temperate
forest canopies special for saproxylic beetles? For. Ecol. Manage. 261, 211–220.
Bouget, C., Larrieu, L., Parmain, G., Nusillard, B., 2013. In search of the best local habitat
drivers for saproxylic beetle diversity in temperate deciduous forests. Biodivers.
Conserv. 22, 2111–2130.
Bütler, R., Lachat, T., Larrieu, L., Paillet, Y., 2013. Habitat trees: key elements for forest
biodiversity. (Chap 2.1, pp 84–92). In: Kraus, D., Krumm, F. (Eds.), 2013. Integrative
approaches as an opportunity for the conservation of forest biodiversity. European
Forest Institute, p. 284.
Commarmot, B., Br¨
andli, UB., Hamor, F., Lavnyy, V., eds (2013). Inventory of the largest
primeval beech-forest in Europe. A Swiss-Ukrainian scientic adventure.
Birmensdorf, WSL; L’viv, Ukrainian national forestry university; Rakhiv, Carpathian
biosphere reserve.
Conner, R.N., Orson Jr., K.M., Adkisson, C.S., 1976. Woodpecker dependence on trees
infected by fungal heart rots. Wilson Bull. 88, 575–581.
Cosyns, H., Kraus, D., Krumm, F., Schulz, T., Pyttel, P., 2019. Reconciling the tradeoff
between economic and ecological objectives in habitat-tree selection: a comparison
between students, foresters, and forestry trainers. Forest Sci. 65 (2), 223–234.
https://doi.org/10.1093/forsci/fxy042.
Courbaud, B., Pupin, C., Letort, A., Cabanettes, A., Larrieu, L., 2017. Modelling the
probability of microhabitat formation on trees using cross-sectional data. Methods
Ecol. Evol. 8 (10), 1347–1359.
Dujesiefken, D., Liese, W., 2015. The CODIT Principle: Implications for Best Practices.
International Society of Arboriculture, USA.
Fox, J., Weisberg, S., 2011. An {R} Companion to Applied Regression, Second Edition.
Thousand Oaks CA: Sage. URL: http://socserv.socsci.mcmaster.ca/jfox/Books/Com
panion.
Gilman, E.F., 2011. An Illustrative Guide to Pruning, Third edition. Delmar Publishers.
Grifth, D.M., Veech, J.A., Marsh, C.J., 2016. Cooccur: probabilistic species co-ccurrence
analysis in R. J. Stat. Softw. 69, 17p. https://doi.org/10.18637/jss.v069.c02.
Hebda, G., Wesolowski, T., Rowinski, P., 2017. Nest sites of a strong excavator, the Great
Spotted Woodpecker Dendrocopos major, in a primeval forest. Ardea 105 (1), 61–71.
Jackson, J.A., Jackson, B.J.S., 2004. Ecological relationships between fungi and
woodpecker cavity sites. Condor 106 (1), 37–49.
Kahl, T., Arnstadt, T., Baber, K., B¨
assler, C., Bauhus, J., Borken, W., Buscot, F., Floren, A.,
Heibl, C., Hessenm¨
oller, D., Hofrichter, M., Hoppe, B., Kellner, H., Krüger, D.,
Linsenmair, K.E., Matzner, E., Otto, P., Purahong, W., Seilwinderm, C., Ernst-Detlef
Schulze, E.D., Wende, B., Weisserm, W.W., Gossner, M.M., 2017. Wood decay rates
of 13 temperate tree species in relation to wood properties, enzyme activities and
organismic diversities. For. Ecol. Manage. 391, 86–95.
L. Larrieu et al.
Ecological Indicators 127 (2021) 107757
10
Kahl, T., Bauhus, J., 2014. An index of forest management intensity based on assessment
of harvested tree volume, tree species composition and dead wood origin. Nature
Conserv. 7, 15–27.
Kraus, D., Bütler, R. ,Krumm, F., Lachat, T., Larrieu, L., Mergner, U. ,Paillet, Y., Rydkvist,
T., Schuck, A., Winter, S., 2016. Catalogue of tree microhabitats – Reference eld
list. Integrate+Technical Paper. 16p.
Larrieu, L., Cabanettes, A., 2012. Species, live status, and diameter are important tree
features for diversity and abundance of tree microhabitats in subnatural montane
beech-r forests. Can. J. For. Res. 1433–1445.
Larrieu, L., Cabanettes, A., Delarue, A., 2012. Impact of silviculture on dead wood and on
the distribution and frequency of tree microhabitats in montane beech-r forests of
the Pyrenees. Eur. J. Forest Res. 131 (3), 773–786.
Larrieu, L., Cabanettes, A., Brin, A., Bouget, C., Deconchat, M., 2014. Tree microhabitats
at the stand scale in montane beech–r forests: practical information for taxa
conservation in forestry. Eur. J. Forest Res. 133, 355–367. https://doi.org/10.1007/
s10342-013-0767-1.
Larrieu, L., Paillet, Y., Winter, S., Bütler, R., Kraus, D., Krumm, F., Lachat, T., Michel, A.
K., Regnery, B., Vandekerkhove, K., 2018. Tree related microhabitats in temperate
and Mediterranean European forests: a hierarchical typology for inventory
standardization. Ecol. Indic. 84, 194–207.
Larrieu, L., Cabanettes, A., Gouix, N., Burnel, L., Bouget, C., Deconchat, M., 2019. Post-
harvesting dynamics of the deadwood prole: the case of lowland beech-oak
coppice-with-standards set-aside stands in France. Eur. J. Forest Res. 138 (2),
239–251.
Larrieu, L., 2014. Les dendro-microhabitats : facteurs cl´
es de leur occurrence dans les
peuplements forestiers, impact de la gestion et relations avec la biodiversit´
e
taxonomique. PhD Thesis. Institut National Polytechnique de Toulouse (INP
Toulouse).
Lassauce, A., Larrieu, L., Paillet, Y., Lieutier, F., Bouget, C., 2013. The effects of forest age
on saproxylic beetle biodiversity: implications of shortened and extended rotation
lengths in a French oak high forest. Insect Conserv. Divers. 6 (3), 396–410.
Loehle, C., 1988. Tree life history strategies: the role of defenses. Can. J. For. Res. 18,
209–222.
Matsuoka, S., 2008. Wood hardness in nest trees of the Great Spotted Woodpecker
Dendrocopos major. Ornithol. Sci. 7, 59–66.
Morris, H., Brodersen, C., Schwarze, F.W.M.R., Jansen, S., 2016. The parenchyma of
secondary xylem and its critical role in tree defense against fungal decay in relation
to the CODIT model. Front. Plant Sci. 7, 1665. https://doi.org/10.3389/
fpls.2016.01665.
Oven, P., Torelli, N., 1999. Response of the cambial zone in conifers to wounding. Phyton
(Horn, Austria) 39 (3), 133–137.
Paillet, Y., Coutadeur, P., Vuidot, A., Archaux, F., Gosselin, F., 2015. Strong observer
effect on tree microhabitats inventories: a case study in a French lowland forest.
Ecol. Ind. 49, 14–23.
Paillet, Y., Archaux, F., Boulanger, V., Debaive, N., Fuhr, M., Gilg, O., Gosselin, F.,
Guilbert, E., 2017. Snags and large trees drive higher tree microhabitat densities in
strict forest reserves. For. Ecol. Manage. 389, 176–186.
Paillet, Y., Archaux, F., du Puy, S., Bouget, C., Boulanger, V., Debaive, N., Gilg, O.,
Gosselin, F., Guilbert, E., 2018. The indicator side of tree microhabitats: a multi-
taxon approach based on bats, birds and saproxylic beetles. J. Appl. Ecol. 55 (5),
2147–2159.
Paillet, Y., Debaive, N., Archaux, F., Cateau, E., Gilg, O., Guilbert, E., 2019. Nothing else
matters? Tree diameter and living status have more effects than biogeoclimatic
context on microhabitat number and occurrence: An analysis in French forest
reserves. PLoS One 14 (5), e0216500.
Pakkala, T., Tiainen, J., Pakkala, H., et al., 2019. Nest tree characteristics of the Lesser
Spotted Woodpecker (Dendrocopos minor) in boreal forest landscapes. Ornis fennica
96 (4), 169–181.
Puverel, C., Abourachid, A., B¨
ohmer, C., Leban, J.M., Svoboda, M., Paillet, Y., 2019. This
is my spot: what are the characteristics of the trees excavated by the Black
Woodpecker? A case study in two managed French forests. For. Ecol. Manage. 453
(2019), 117621.
R Development Core Team, 2018. R: a language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna.
Rameau, J.C., Mansion, D., Dum´
e, G., 1993. Flore Foresti`
ere Française. IDF, Paris.
Regnery, B., Paillet, Y., Couvet, D., Kerbiriou, C., 2013a. Which factors inuence the
occurrence and density of tree microhabitats in Mediterranean oak forests? For. Ecol.
Manage. 295, 118–125.
Regnery, B., Couvet, D., Kubarek, L., Julien, J.F., Kerbiriou, C., 2013b. Tree
microhabitats as indicators of bird and bat communities in Mediterranean forests.
Ecol. Ind. 34, 221–230.
Roberge, J.M., Angelstam, P., 2004. Usefulness of the umbrella species concept as a
conservation tool. Conserv. Biol. 18 (1), 76–85.
Rolstad, J., et al., 2000. Black woodpecker nest sites: characteristics, selection, and
reproductive success. J. Wildl. Manage. 64 (4), 1053–1066.
Rutishauer, E., Barth´
el´
emy, D., Blanc, L., Nicolini, E.-A., 2011. Crown fragmentation
assessment in tropical trees: method, insights and perspectives. For. Ecol. Manage.
261, 400–407.
Schepps, J., Lohr, S., Martin, T.E., Schepps, J., Lohr, S., Martin, T.E., 1999. Does tree
hardness inuence nest-tree selection by primary cavity nesters? Auk 116, 658–665.
Schneuwly-Bollschweiler, M., Schneuwly, D.M., 2012. How fast do European conifers
overgrow wounds inicted by rockfall? Tree Physiol. 00, 1–8.
Shigo, A.L., 1984. Compartmentalization: a conceptual-framework for understanding
how trees grow and defend themselves. Annu. Rev. Phytopathol. 22, 189–214.
Stokland, J.N., Siitonen, J., Jonsson, B.G., 2012. Biodiversity in Dead Wood. Cambridge
University Press, Cambridge; UK.
Vuidot, A., Paillet, Y., Archaux, F., Gosselin, F., 2011. Inuence of tree characteristics
and forest management on tree microhabitats. Biol. Conserv. 144, 441–450.
Winter, S., M¨
oller, G.C., 2008. Microhabitats in lowland beech forests as monitoring tool
for nature conservation. For. Ecol. Manage. 255 (3–4), 1251–1261.
Winter, S., Hoeer, J., Michel, A.K., Boeck, A., Ankerst, D.P., 2015. Association of tree
and plot characteristics with microhabitat formation in European beech and
Douglas-r forests. Eur. J. Forest Res. 134 (2), 335–347.
Zahner, V., Sikora, L., Pasinelli, G., 2012. Heart rot as a key factor for cavity tree
selection in the black woodpecker. For. Ecol. Manage. 271, 98–103.
Zuur, A.F., Ieno, E.N., Elphick, C.S., 2010. A protocol for data exploration to avoid
common statistical problems. Methods Ecol. Evol. 1, 3–14.
L. Larrieu et al.
