Perspectives: Key factors determining the presence of Tree-related Microhabitats: A synthesis of potential factors at site, stand and tree scales, with perspectives for further research
Introduction
A Tree-related Microhabitat (TreM) is defined as “a distinct, well-delineated structure occurring on living or standing dead trees, that constitutes a particular and essential substrate or life site for species or species communities during at least a part of their life cycle to develop, feed, shelter or breed” (Larrieu et al., 2018). TreMs support a wide array of biodiversity (see Table 2 in Larrieu et al., 2018) that is not usually supported by other forest structures, such as deadwood items (Stokland et al., 2012). Several studies have highlighted the significant impact of an increase in TreM-bearing tree (hereafter called habitat-tree) density on species richness for several taxa (see e.g. Bouget et al., 2013, Bouget et al., 2014a, Bouget et al., 2014b, Larrieu et al., 2019, and Winter and Möller, 2008 for saproxylic beetles; Regnery et al., 2013a, Paillet et al., 2018 for bats and birds; Larrieu et al., 2019 for polypores and hoverflies; Basile et al., 2020 for insects and bats). Hence, some authors have suggested using TreMs as indirect biodiversity indicators in forest ecosystems and as tools to promote integrative forest management (Kraus and Krumm, 2013, Winter and Möller, 2008, Regnery et al., 2013b, Paillet et al., 2018, Bütler et al., 2013, Larrieu et al., 2018, Asbeck et al., 2021). However, at plot and stand scales, the relationship between TreM density and/or diversity with variations in biodiversity are not so straightforward. Indeed, this relationship is only partially consistent, for both species’ richness and composition, when considering a range of forest contexts (Bouget et al., 2013, Bouget et al., 2014a, Bouget et al., 2014b, Paillet et al., 2018). This is likely due to complex interactions between TreMs and other resources (e.g. deadwood items, flowering plants in clearings, water bodies; Larrieu, 2014), flaws in procedures for assessing taxa and TreMs (Larrieu and Bouget, 2017), time lags in the response of certain TreM-dwelling species to TreM presence (Herrault et al., 2016), as well as the spatial distribution of source populations (Komonen and Müller, 2018).
The spatial distribution of TreMs is not solely dependent on that of the TreM-bearing trees. Indeed, they are typically limited in availability, persisting from only a few days for lignicolous agarics, to several decades for large rot-holes. Thus, TreMs can be considered as Ephemeral Resource Patches (ERP, Finn, 2001). Furthermore, some of these structures are generated by stochastic events that occur very rarely (e.g. lightning scars), or have a very long development time (e.g. fully evolved rot holes). Numerous forest-dwelling species are continuity-dependent and therefore are restricted by both development time of a novel habitat and the time required to colonize that novel resource patch (Nordén et al., 2014). As a result, it is challenging to manage forests to ensure a continuous resource supply for TreM-dwelling taxa. To provide forest managers with practical recommendations for the conservation of TreM-dwelling taxa, i.e. which trees should be exempt from harvesting, numerous studies have attempted to identify key features at the tree level that are linked to TreM formation. They have highlighted the key roles of tree species, tree diameter at breast height (dbh) and status (i.e. living vs standing dead) for driving the occurrence and abundance of TreMs (Winter and Möller, 2008, Michel and Winter, 2009, Vuidot et al., 2011, Regnery et al., 2013b, Larrieu and Cabanettes, 2012, Larrieu et al., 2014b, Paillet et al., 2018, Paillet et al., 2019, Kozák et al., 2018, Asbeck et al., 2019). Notwithstanding the abundance of studies on the topic, to date, predictive models have mainly focused on only two basic tree features, namely dbh and species for living trees (Courbaud et al., 2017, Jahed et al., 2020); in some cases, a qualitative variable was used to separate managed and unmanaged forests (Courbaud et al., 2022). Dbh and tree-species are easy to record, and are also routinely assessed by forest managers for silvicultural and monitoring purposes. However, the power to predict TreM occurrence with these two tree features alone is often rather low, e.g. about 26 % in beech (Fagus sylvatica)-silver fir (Abies alba) forests (Larrieu et al., 2014a). Moreover, Courbaud et al. (2022) have shown that site effects are huge. However, these previous works have not been able to highlight what site features influence the presence and dynamics of TreMs.
For these reasons, the distribution of TreMs is currently difficult to predict, hampering the elaboration of appropriate management guidelines that take into consideration these crucial biodiversity features. This is particularly important as the need to take TreMs into account in silvicultural planning is increasingly acknowledged among forest managers. For example, TreMs have been incorporated into an index that is routinely used in the field by forest managers in France (Index of Biodiversity Potential, IBP; Larrieu and Gonin, 2008, Gosselin and Larrieu, 2020). At a larger spatial scale, a rapidly growing network of about 160 training plots (called “marteloscopes”) has been established across 22 countries, mainly in Europe, with the aim of improving managers’ knowledge about TreMs and inventory calibration, employing tree-marking exercises in the field (Kraus et al., 2021). Therefore, there is a critical need to better explain and predict TreM occurrence and the processes that lead to their formation, with the ultimate aim of encouraging forest managers to take TreM-associated biodiversity into account in their daily work routines.
In this paper, using a large international TreM database, we first quantify the explanatory power of the factors that currently feature in most of the available datasets, namely tree species, dbh, tree status (i.e. living or standing dead), time since last harvest and plot context for predicting TreM occurrence at the tree level.
Plot context is currently a “black box” which combines local environmental conditions, past and current management legacies, and local biotic features which might impact TreM formation in several ways. Environmental conditions determine tree species assemblages in relation to both biogeographic and bioclimatic contexts, as well as soil fertility. Soil fertility may determine the presence of epiphytic plants that are considered as TreMs when they climb on trunks. For example, ivy (Hedera helix) does not thrive on very acidic and nutrient-poor soils (Dumé et al., 2018). Thin soils which are prone to be often dry can promote dead wood in the crown of the trees (Breda et al., 2004). Furthermore, the dynamics of TreM formation has been shown to differ between tree species (Courbaud et al., 2017, Jahed et al., 2020) and not all tree species are likely to support the same type of TreMs (Vuidot et al., 2011, Larrieu and Cabanettes, 2012, Paillet et al., 2019). The presence of particular geological features, such as cliffs or mobile scree, may increase the density of trees that have bark loss or broken stems due to falling rock (Dorren and Berger, 2006, Stokes et al., 2005). For example, in the Black Forest (Germany), Asbeck et al. (2019) highlighted that increasing altitude favours the number of buttress-root concavities and epiphytic lichens, while mosses and mistletoe are more abundant at lower altitudes. However, the detailed effect of local conditions has, to date, not yet been well quantified. Furthermore, such observations may actually mask confounding effects, e.g. when altitude and slope are strongly and positively correlated, as is often the case in mountain areas. Forest management is known to influence both the density and the diversity of TreMs (e.g. Winter and Möller, 2008, Larrieu and Cabanettes, 2012), while the impact of harvesting persists over the long term (Bouget et al., 2014a, Bouget et al., 2014b, Paillet et al., 2015, Larrieu et al., 2016). In addition, certain biotic features may have an effect on the presence and abundance of TreMs, e.g. density of red deer (Cervus elaphus) in relation to food resource availability for bark loss (Verheyden et al., 2006), or the presence of black woodpecker (Dryocopus martius) for both breeding holes and feeding concavities (Bobiec et al., 2005).
Hence, secondly, in order to unpick the composition of this black box and to identify the most influential features, we consider a set of factors related to site, stand and tree features for which there is a strong assumption that they play a key role in TreM formation. The main goal here was to identify the most biologically relevant drivers, rather than relying on only the most widely available variables. An approach based on a selection of factors that have been identified in the literature as likely having a positive influence on TreM occurrence should help us to avoid focusing on spurious indirect relationships with no causal role in TreM formation.
Thirdly, based on a consideration of the trade-off between sampling effort and relevance for explaining the occurrence of TreMs, evaluated from both literature and based on our own expertise, we suggest a sub-set of features that i/ should be tested by further studies focusing on TreMs when widely available (e.g. via large scale databases), or ii/ should be recorded in the future by researchers in the field.
Section snippets
Predictive power of the features currently available in most TreM datasets
To quantify the predictive power of the features shared by most of the datasets, we used an international database which integrates 23 harmonized datasets, comprising 100,855 living trees and 10,354 standing dead trees belonging to 89 tree species (appendix; Table 1SM). For each of the eleven TreM subgroups that were designated by Courbaud et al. (2022), we built a Generalized Linear Mixed Model (GLMM) that described the presence/absence of this TreM group in relation to dbh, tree species, tree
Predictive power of the features currently available in the databases
The best full models explained, on average, around one third of the variance in TreM occurrence, from 15% for dendrotelms to 59.9% for buttress-root concavities (Table 1). Plot context was always the feature that explained the highest proportion of variance in TreM occurrence.
Additional features that may play a key role in TreM formation
From the literature, we identified 21 features which may play an important role in TreM formation: nine environmental site specific, two stand dependent and ten tree-related features (Table 2). The feature implicated in
Features that may play a key role in TreM formation based on biological processes
Ontogenic stage appears to provide more power for predicting TreM formation than tree age. During its ontogeny, a tree goes through four stages of development (Drénou, 2017; Fig. 1): young, adult, mature and senescent. During the young stage, the branching consists of a limited number of axis categories (architectural unit) which is characteristic of each tree species. The branches are both thin and ephemeral, so that they form a temporary crown. The adult stage corresponds to the duplication
Conclusion
Many environmental and tree-specific features that have rarely been considered until now by researchers studying TreMs appear to be promising candidates for improving the prediction of TreM occurrence and their dynamics. Several of these features can be easily measured in the field or extracted from large scale environmental databases. We suggest that future studies record a subset of nine features, in addition to variables already routinely recorded, to provide additional information to enable
Author contribution statement
LL, BC and CD designed the general structure of the manuscript. MG run statistical analysis. RB, DKo, DKr, FK, TL, JM, YP, AS, JS, MS, KV contributed to discussion of the results and the writing of the final document.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
We thank Mark Hewison both for his valuable comments and for reviewing the English. We also thank Sylvie Ladet for her expertise on the possibilities to evaluate certain features from GIS and her contribution to the metadata of the international TreM database. M. Svoboda and D. Kozák were supported by the project of the Czech ScienceFoundation (21-27454S).
References (121)
- et al.
Predicting abundance and diversity of tree-related microhabitats in Central European montane forests from common forest attributes
For. Ecol. Manage.
(2019) - et al.
Key features for saproxylic beetle diversity derived from rapid habitat assessment in temperate forests
Ecol. Ind.
(2014) - et al.
Applying ecological model evaluation: lessons learned with the forest dynamics model Samsara2
Ecol. Model.
(2015) The function and formation of buttresses
Trends Ecol. Evol.
(1993)- et al.
Comparison of two models for predicting the critical wind speeds required to damage coniferous trees
Ecol. Model.
(2000) - et al.
Developing and using statistical tools to estimate observer effect for ordered class data: The case of the IBP (Index of Biodiversity Potential)
Ecol. Ind.
(2020) Tree roots in canopy soils of old European beech trees. An ecological reassessment of a forgotten phenomenon
Pedobiologia
(2011)- et al.
Profile of tree-related microhabitats in European primary beech-dominated forests
For. Ecol. Manage.
(2018) - et al.
Co-occurrence patterns of tree-related microhabitats: A method to simplify routine monitoring
Ecol. Ind.
(2021) - et al.
Deadwood and tree microhabitat dynamics in unharvested temperate mountain mixed forests: a life-cycle approach to biodiversity monitoring
For. Ecol. Manag.
(2014)