A broad interest exists in developing structure-based indicators to use as proxies for other attributes that are difficult to assess, such as biological diversity. Summary variables that account for stand-scale forest structural complexity could facilitate the comparison among stands and provide a means of ranking stands in terms of their potential contribution to biodiversity. We developed an index of structural heterogeneity (SHI) for beech forests in southern Italy: (i) we established a preliminary list of 23 structural variables obtained from data routinely collected in forest inventories; (ii) we quantified these variables in a set of 64 beech-dominated stands encompassing a wide range of variability in the Cilento, Vallo di Diano and Alburni National Park; (iii) we identified a core set of attributes that take into account the main sources of structural heterogeneity identified in reference old-growth forests; and (iv) we combined these core attributes into a simple additive index (SHI). We identified eight core attributes that were rescaled to the range 0 to 10 using regression equations based on raw attribute data. The SHI was calculated as the sum of these attribute scores and then expressed as a percentage. The index performance was evaluated against ten reference old-growth beech stands in the Apennines. The index ranged between 38 and 79.1 (median=59.4) and was distributed normally for the calibration dataset. The SHI successfully discriminated between old-growth (range=71.9-99.9, median=85.1) and early-mature to mature forests. Furthermore, the SHI linearly increased with stand age and was higher in multi-layer high forests than in single- and double-layer forests. However, a large variation was detected within both management types and age classes. SHI could be helpful for foresters as a tool for quantifying and comparing structural heterogeneity before and after a silvicultural intervention aimed at restoring the structural complexity in second-growth stands.
The theoretical and practical relevance of the structural attributes of forest stands is being increasingly acknowledged (
A broad interest exists in developing structure-based indicators to use as proxies for other attributes that are difficult to assess. Several studies have tried to quantify the diversity of forest structures in a stand through the definition of synthetic indexes. Some of these indexes were designed to rank forest stands on the basis of management intensity (
Stand structural complexity is essentially a measure of the variety and relative abundance of different structural attributes in a given stand. Particular attention is usually paid to those attributes that quantify variation (
A stand-scale index of structural complexity may facilitate the comparison of stands based on their potential contribution to biodiversity (
Ranking stands according to their structural complexity may be challenging, since even ecologically similar forest stands within the same region may accumulate complexity in different ways (
Recently,
Here, we applied this methodology to develop a stand-scale index of structural heterogeneity for Apennine beech forests. To identify the suite of attributes to include in this index of structural heterogeneity, we first explicitly defined the main sources of structural complexity commonly reported for beech natural forests in Italy and southern Europe. We considered old-growth condition as the reference state, given that late successional forests, especially old-growth stands, are known to have a high horizontal and vertical structural diversification (
The aim of our study was thus: (i) to develop an index of structural heterogeneity for southern Apennine beech forests; (ii) to test whether different age classes and forest management types were characterized by different levels of structural heterogeneity; (iii) to compare levels of structural heterogeneity among some well-studied reference old-growth beech stands in the Apennines and forest stands that do not display old-growth attributes.
We hypothesized that, with the exception of recently established stands, which were not considered in this work, stand structural heterogeneity increases with forest age, and differs according to the management type, with double- or multi-layered uneven-aged stands having significantly higher structural heterogeneity values than even-aged single-layered stands. We also expected old-growth to be more heterogeneous than other early-mature to mature forest stands.
Data were collected in forest stands dominated by beech (
Sampling units were identified based on an aligned systematic design, overlaying a 500 m grid on the beech forest distribution map. Plots were selected with the help of the Park planning records and photo-interpretation of digital aerial photographs (Flight IT2000, nominal resolution 1 m), with a nominal scale of 1:25.000. We excluded areas whose structural heterogeneity could derive from recent harvesting, i.e., by the creation of stumps and the co-occurrence of remnant trees (or standards) and very young trees. We focused on early-mature to mature forests where structural heterogeneity is likely due to natural forest dynamics (tree senescence and death, establishment of natural regeneration and gap dynamics, etc).
Overall, 64 sampling units across 12 forest areas were selected, encompassing an area of about 137 km2 (
A circular plot of 20-m radius (1256 m2) was established in each unit. Living trees in each plot were calipered in concentric circular areas with a radius of 4, 13 and 20 m, with thresholds of minimum diameter at breast height (DBH) of 2.5, 10 and 50 cm, respectively. Height was measured using a Haglof Vertex on one out of ten sampled trees, chosen randomly. For the remaining trees, the height was estimated with a traditional H=f(DBH) model calculated on the basis of the trees whose height was measured. In the intermediate circular area (13 m radius), the length and diameter of all the lying deadwood components (with a minimum diameter ≥ 10 cm) were measured, as well as their decay level according to
As an additional dataset, we selected 10 beech-dominated stands in central Italy with old-growth features (such as high density of large living trees, high amount of deadwood, uneven-aged structure), whose high structural heterogeneity was previously reported in several studies (
Living trees and deadwood were sampled within a 1 ha plot in all the study sites except two (Cozzo Ferriero, 0.16 ha; Fosso Cecita, 0.45 ha), in which plot size was reduced because of the very steep slopes. The position, species, DBH (minimum threshold of 3 cm) and height of every tree in the plot were recorded, as well as the position, diameter and length (or height) of standing dead trees, downed dead trees, snags and stumps. Deadwood pieces were sampled if they met the following requirements: minimum diameter > 5 cm, more than half the base of their thicker end lying within the plot, length > 1 m. Further details are reported by
An index of structural heterogeneity was built according to the methodology proposed by
We compiled a preliminary list of 23 structural variables that may easily be derived from routinely collected data in forest monitoring programmes and NFIs (
To define the suite of attributes to be considered in the final core set of attributes, we first listed the main sources of structural complexity occurring in beech natural forests, as reported in recent literature on old-growth forests in southern Europe (
When designing an index of structural complexity, one should focus on attributes that: (i) have a low kurtosis, since a high kurtosis would indicate similar values of an attribute for several sites; (ii) may help to distinguish between categories of interest (
Both logarithm and square-root transformations were applied to improve the distribution of attributes showing a high kurtosis (<2): in the selection stage, we only retained the transformation that most improved their distribution.
Since we assumed that structural complexity increases with age, early- and late-successional stands were separated choosing an age threshold of 100 years (which represents the usual harvest return interval for beech forests in the Apennines) and then compared. Differences between the two mentioned groups were tested for each structural variable by Mann-Whitney test.
Pairwise correlation between variables was calculated using the Spearman’s
To help in variable selection, we also estimated whether they were more or less difficult to sample and/or calculate. Variables that only need tree diameter data (
Finally, we created a core set of structural attributes that included a variable for each of the eight sources of structural complexity listed in
A score ranging from 0 to 10 was assigned to each attribute in the core set based on linear regression through quartiles (
Finally, a Structural Heterogeneity Index (SHI) was obtained by summing the scores in the range 0-10 assigned to each variable in the core set, and then expressed as a percentage. Score weighting was not considered here because: (i) it could imply an arbitrary choice; (ii) index performances were found to be independent of the weighting of attributes (
SHI values calculated for the 64 forest plots in the Cilento National Park were tested for significant differences between broad classes based on age and management type using non-parametric Kruskal-Wallis test. We used multiple regression to test whether the SHI significantly increases with age, and covariates with management type as well as with other environmental attributes (
As an additional evaluation of index performances, we calculated the SHI on the set of 10 beech-dominated, old-growth forests from central Italy (Fig. S1.1, Tab. S1 -
All analyses were performed using the software package R 2.14.1 (
Beech stands in the Cilento National Park showed index values ranging between 38 and 79.1 (median 59.4) and showing no departure from normal distribution (Fig. S5.1). The SHI appeared to have a minimum in the 80 year-age class, and a slow increase with age (
The SHI of the selected old-growth stands ranged between 71.7 and 99.9 (median 85.1). The SHI was significantly higher for reference old-growth stands than for the beech stands included in the main dataset (H = 27.7, df = 2, p < 0.001 -
The high SHI values observed in the old-growth stands stem from the high scores of index subcomponents, whose relative importance varied greatly across stands (
We applied an acknowledged methodology to obtain an index of structural heterogeneity for southern Italy beech forests. Forest structural heterogeneity, as indicated by the SHI, linearly increased with stand age and was higher for multi-layer high forests than for single- and double-layer forests.
The positive relationship between the SHI and stand age closely matched our expectations, although there was a marked variability both within management types and age classes. Such variability was likely due to the wide range of site conditions, soil fertility, disturbance and forest management histories found in the study area. In particular, the SHI revealed a very high degree of variation in single- and double-layer high forests. These management types encompassed stands with highly variable amounts of growing stock (whose index sub-scores ranged between 2 and 10 for single-layered and between 1.1 and 10 for double-layered stands), height standard deviations (scores ranging between 2.6-10 and 3.3-10, respectively) and coarse woody debris volumes (1.1-10 and 1.1-9.4, respectively).
Although part of this variability can be accounted for by differences in age classes and altitude, the relationship observed between stand age and complexity should be considered only as a general trend. Since the management history of most stands is only partially known (harvest archives in most municipalities of the study area date back no longer than the 1990s), stands were only classified into broad age classes on the basis of expert opinion. More detailed stand age data, including accurate dendrochronological reconstructions of their past disturbances, would be required to quantify the actual rate at which forest complexity increases over time.
Besides stand age, most of the remaining variability is likely to be dependent on the different disturbance and harvesting histories of these stands. Forests in southern Italy are characterized by a peculiar history: in the 19th century, a forest law prescribed clearcuts with the release of 45 standards per hectare to be applied indistinctly to all the forests of the Kingdom of the two Sicilies (which included southern Italy and Sicily). However, this law was never extensively applied, and most of the forests continued to be subject to selective cuttings (
In an operational context, structural indicators may prove very useful to distinguish old-growth forests from younger developmental stages, as well as to rank forests along “old-growthness” gradients (
Recently,
In this study, old-growth stands showed very high SHI values, sometimes close to the maximum as in the case of the “Sasso Fratino” stand. This result confirmed that this index effectively captures aspects of structural heterogeneity recognized as important in reference beech old-growth forests in southern Europe (
Since SHI successfully distinguished between old-growth and younger stands, the question arises whether this index could also be used to assess the “naturalness” of a forest stand. The concept of “naturalness” is related to the degree to which forest ecosystems are characterized by natural processes and/or the absence of human influences (
There is a great need for simple tools that can help forest managers to improve stand biodiversity (
One of the main advantage of the SHI is that it simply consists of the sum of scores for each structural attribute, obtaining a “synthetic” index of stand complexity. On the other hand, similar SHI values may mask different underlying source of heterogeneity. This was the case of the “Abeti Soprani” and “Monte Cimino” old-growth stands (
The SHI relies upon input data routinely acquired by almost all the NFIs in the world (
Forest structural indicators may also be sensitive to the field methods used for their assessment, such as plot size and minimum DBH thresholds (
In conclusion, indicators based on key structural parameters are of considerable interest as practical surrogates for attributes that are normally too expensive or difficult to measure, such as biodiversity or ecosystem functioning. The common assumption that the structural, functional, and compositional attributes of a stand are inter-dependent (
We thank the
Distribution of forests in the
PCA of standardized structural variables, axes 1-2. (Green): Live trees structural variables; (Red): Deadwood-related variables; (Blue): Tree Height-related variables. (BA): Basal area; (RangeDBH): range of diameter distribution; (QMDBH): quadratic mean diameter; (LivVol): growing stock; (Ndbh): number of diameter classes; (StemDens): Stem density; (TreeRich): tree species richness. (CWD): Coarse woody debris volume; (Stumps): volume of stumps; (Snags): volume of standing dead trees broken above 1.3 m; (StDw): volume of standing dead trees; (SnagStDw): volume of standing dead trees (including snags); (NsnagsStDW): number of standing dead trees (including snags); (BASnagSt): basal area of standing dead trees (including snags); (DWtot): Deadwood (standing + CWD) total volume; (DWLivRatio): living wood/Deadwood volume ratio; (CWDI): coarse woody debris index (see
Boxplot of SHI across age classes (A) and structural types (B). Small numbers below the boxes represent the sample size. (HF): High forest.
SHI comparison between early-mature to mature, and old-growth stands. The boxplots refer to the SHI of managed beech forests in the Cilento National Park (left) with those of a set of beech forests with old-growth features located throughout the Apennines (right). Small numbers below the boxes represent the sample size.
List of the eight sources of structural heterogeneity considered in the present study and their ecological importance for forest biodiversity. This list represents the basis to select the structural attributes for constructing the SHI (Structural Heterogeneity Index).
Sources of structural heterogeneity | Description | References |
---|---|---|
Vertical heterogeneity (VH) | Stands containing a variety of tree heights are likely to contain a variety of tree ages and, consequently, a high vertical and horizontal heterogeneity. Horizontal and vertical patterns of trees significantly affect demographic processes, resource distribution ( |
|
Compositional diversity (CH) | The presence of a mix of shade-tolerant and shade-intolerant tree species may produce a multi-layered canopy. Compositionally diverse tree layers may favour herb-layer diversity, since different tree species may have different light transmittance and litter quality. | |
Uneven-agedness (UA) | In forested landscapes where small to intermediate scale disturbance events are dominant, an uneven-aged structure may indicate a natural development of the stand, or the application of close-to-nature silvicultural practices. The variability in tree size may also be an indicator of the diversity of niches occurring within a stand that could be used by a wealth of animal and plant organisms. | |
Density of large living trees (LLT) | Large living trees store a large amount of carbon and provide habitat functions for a number of threatened or ecologically important forest species. These functions relate to the great variety of niches that large trees offer, including rough bark, trunk hollows, exposed deadwood, sapflows, dead branches and dead tops. | |
Growing stock (GS) | Higher living above-ground biomass indicates the degree to which a stand effectively accomplishes its function of storing carbon. Owing to greater levels of biomass, old-growth stands were shown to attenuate surface temperature more effectively than managed stands, hosting a higher proportion of forest specialist herb-layer species | |
Total deadwood volume (DW-TOT) | Deadwood is a key ecosystem feature supporting high levels of biodiversity, for instance providing diverse niches for many specialized and saproxylic organisms. Such organisms include those with low dispersal capabilities that need long-term availability of deadwood substrate, whose absence in intensively managed stands may cause local or regional extinction of several species. | |
Deadwood decay classes (DW-DC) | The absence of deadwood in one or several decay phases strongly indicates a break in the continuity of deadwood supplies, typically due to a combination of recent harvesting and deadwood removal. This may affect the continuity of nutrient supply to the forest floor, and the diversity and abundance of saproxylic organisms. | |
Standing deadwood, dead trees and snags (DW-ST) | Standing dead trees and snags may bear niches such as tree hollows, cavity strings and cracks that are important for a variety of species such as breeding birds, mammals and invertebrates, as well as for lichens and bryophytes. |
Core set of structural variables and selection criteria. (W): Wilcoxon test (equivalent to Mann-Whitney); (Source): the source of heterogeneity indicates whether the variable could be a proxy of one of the eight features described in
Structural indicators | Kurtosis | Medians | Function as a surrogate (significant |
Source | Samplingefficiency | |||
---|---|---|---|---|---|---|---|---|
<100 yrs (n=26) | > 100 yrs (n=38) | W | Prob. | |||||
Living volume | 0.23 | 421.17 | 524.38 | 330 | 0.025 | Basal Area (0.53); no. DBH classes (0.53); no. trees DBH> 40 cm (0.68); Height (0.67) | GS | 1 |
no. trees DBH>40 cm | -0.11 | 1 | 6.5 | 186 | 0.001 | Living volume (0.66); DBH range (0.66); Height (0.78); Density of standing deadwood (-0.55); Basal area of standing deadwood (-0.50) | LLT | 3 |
Diameter diversity (Gini-Simpson index) | -0.81 | 0.6 | 0.72 | 575 | 0.273 | Living stem density (-0.64); Quadratic mean DBH (0.71) | UA | 2 |
Height standard deviation | -0.34 | 3.53 | 5.47 | 321 | 0.018 | - | VH | 2 |
CWD index | -0.99 | 2 | 2 | 516.5 | 0.754 | Lying CWD Volume (0.85); no. decay classes (0.75); Total log length (0.86) | DW-DC | 1 |
log (Tree species richness) | 0.17 | 1.1 | 0.69 | 573.5 | 0.229 | - | CH | 3 |
log (basal area of standing deadwood) | -0.31 | 0.85 | 0.18 | 323.5 | 0.018 | Height (-0.50); Snags volume (0.68); Standing dead trees volume (0.86); Total Standing deadwood (0.98); Total deadwood (0.73); Density of standing deadwood (0.89); Dead/Living wood ratio (0.76) | DW-ST | 3 |
sqrt (total deadwood) | 0.04 | 4.7 | 4.63 | 526.5 | 0.662 | Standing dead trees volume (0.65); Total Standing deadwood (0.77); Density of standing deadwood (0.68); Basal area of standing deadwood (0.61); Dead/Living wood ratio (0.96) | DW-TOT | 1 |
Regression equations used to assign a score to attributes on a scale of 0-10, obtained from 64 beech dominated forest stands in the
Attribute | Regression equation | R2 |
---|---|---|
Living volume | Score = -2.021 + X · 0.016 | 0.938 |
no. Large Living Trees DBH> 40 cm | Score = 3.274 + X · 0.595 | 0.952 |
DBH diversity (Gini-Simpson index) | Score = -2.233 + X · 14.034 | 0.969 |
Height standard deviation | Score = 1.815 + X · 0.811 | 0.918 |
CWD index | Score = 3.750 + X · 1.25 | 0.900 |
Log (Tree species richness) | Score = -2.511 + log(X) · 9.053 | 0.900 |
Log (basal area of standing deadwood) | Score = 3.536 + log(X) · 4.221 | 0.924 |
sqrt (Total deadwood volume) | Score = 1.167 + sqrt(X) · 1.083 | 0.999 |
SHI values and scores for its sub-components calculated for 10 beech stands with old-growth features located throughout the Apennines. Scores for each structural variable was obtined by the regression equations reported in
Stand | Livingvolume | No. trees DBH>40 cm | DBH diversity(Gini-Simpson) | Heightsd | CWDindex | Log (tree sp. richness) | Log (BA stand dw) | Sqrt(total dw) | SHI |
---|---|---|---|---|---|---|---|---|---|
Abeti Soprani | 6.9 | 10.0 | 10.0 | 6.8 | 10.0 | 10.0 | 9.4 | 10.0 | 91.4 |
Collemeluccio | 6.7 | 10.0 | 10.0 | 6.6 | 10.0 | 10.0 | 4.8 | 5.7 | 79.8 |
Cozzo Ferriero | 10.0 | 10.0 | 10.0 | 8.4 | 10.0 | 0.0 | 7.1 | 10.0 | 81.8 |
Fonte Novello | 10.0 | 10.0 | 10.0 | 7.3 | 10.0 | 0.0 | 10.0 | 10.0 | 84.1 |
Gargano-Pavari | 8.4 | 10.0 | 10.0 | 10.0 | 10.0 | 7.4 | 9.1 | 10.0 | 93.7 |
Monte Cimino | 10.0 | 10.0 | 9.3 | 10.0 | 10.0 | 10.0 | 5.0 | 7.3 | 89.5 |
Monte di Mezzo | 9.0 | 10.0 | 10.0 | 7.8 | 10.0 | 10.0 | 5.4 | 6.7 | 86.2 |
Monte Sacro | 5.3 | 10.0 | 8.3 | 7.9 | 10.0 | 0.0 | 5.9 | 10.0 | 71.7 |
Sasso Fratino | 10.0 | 10.0 | 10.0 | 10.0 | 10.0 | 10.0 | 10.0 | 9.9 | 99.9 |
Val Cervara | 3.7 | 10.0 | 8.6 | 8.6 | 10.0 | 0.0 | 10.0 | 10.0 | 76.1 |
Location and synthetic description of 10 old-growth stands used to evaluate the performance of the SHI.
Construction of the CWD index.
Structural variables, kurtosis and correlations.
Distribution of estimated ages across forest types in Cilento National Park.
Statistical distribution of the SHI.