Content uploaded by Ismet Yener
Author content
All content in this area was uploaded by Ismet Yener on Oct 08, 2023
Content may be subject to copyright.
Citation: Yener, I.; Guvendi, E.
Predicting and Mapping Dominant
Height of Oriental Beech Stands
Using Environmental Variables in
Sinop, Northern Turkey. Sustainability
2023,15, 14580. https://doi.org/
10.3390/su151914580
Academic Editor: Surendra
Singh Bargali
Received: 29 August 2023
Revised: 25 September 2023
Accepted: 1 October 2023
Published: 8 October 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
sustainability
Article
Predicting and Mapping Dominant Height of Oriental Beech
Stands Using Environmental Variables in Sinop,
Northern Turkey
Ismet Yener 1,* and Engin Guvendi 2
1Department of Forest Engineering, Faculty of Forestry, Artvin Coruh University, Artvin 08100, Turkey
2Department of Forestry, Kürtün Vocational School, Gümü¸shane University, Gümü¸shane 29810, Turkey;
eguvendi28@gumushane.edu.tr
*Correspondence: yener@artvin.edu.tr
Abstract:
The dominant height of forest stands (SDH) is an essential indicator of site productivity
in operational forest management. It refers to the capacity of a particular site to support stand
growth. Sites with taller dominant trees are typically more productive and may be more suitable for
certain management practices. The present study investigated the relationship between the dominant
height of oriental beech stands and numerous environmental variables, including physiographic,
climatic, and edaphic attributes. We developed models and generated maps of SDH using multilinear
regression (MLR) and regression tree (RT) techniques based on environmental variables. With this
aim, the total height, diameter at breast height, and age of sample trees were measured on 222 sample
plots. Additionally, topsoil samples (0–20 cm) were collected from each plot to analyze the physical
and chemical soil properties. The statistical results showed that latitude, elevation, mean annual
maximum temperature, and several soil attributes (i.e., bulk density, field capacity, organic carbon,
and pH) were significantly correlated with the SDH. The RT model outperformed the MLR model,
explaining 57% of the variation in the SDH with an RMSE of 2.37 m. The maps generated by both
models clearly indicated an increasing trend in the SDH from north to south, suggesting that elevation
above sea level is a driving factor shaping forest canopy height. The assessments, models, and maps
provided by this study can be used by forest planners and land managers, as there is no reliable data
on site productivity in the studied region.
Keywords: stand productivity; site factors; multiple linear regression; regression tree; modeling
1. Introduction
Site quality is a combination of climatic, physiographic, edaphic, and biotic factors
affecting the potential of trees to produce aboveground biomass in the forest. A common
indicator of site productivity in forestry is the site index (SI), referring to the stand dominant
height (SDH) at a standard age (e.g., 50 or 100 years) [
1
]. The calculation of the SI is based
on the SDH and age measurements of 100 trees per hectare (also known as h100) [
1
,
2
].
The SI is an essential parameter in ecology and forest management when deciding on
afforestation and reforestation locations [
3
]. Measuring the SDH and tree ages on the ground
is challenging because of time, labor, and cost restrictions [
4
,
5
]. Therefore, many researchers
seek to predict or model site productivity cost-effectively. The initial studies on modeling
forest site productivity used only parametric techniques, e.g., multilinear regression (MLR).
Recently, some studies have also used non-parametric techniques, such as fuzzy logic, an
artificial neural network (ANN), the general additive model (GAM), and the classification
and regression tree (CART), which are more accurate and precise than the former ones to
estimate the growth and height of dominant trees [
6
–
8
]. Among the previously mentioned
techniques, CART is a non-parametric technique used to model site productivity [
9
,
10
].
Recently, studies using the CART technique have increased worldwide [4,7,11–16].
Sustainability 2023,15, 14580. https://doi.org/10.3390/su151914580 https://www.mdpi.com/journal/sustainability
Sustainability 2023,15, 14580 2 of 20
Many ecological variables, including climatic, topographic, and edaphic variables,
have been used in site productivity modeling studies because of their vital effects on plant
growth. Of these factors, the climate plays a crucial role in characterizing global forests’
distribution, carbon storage, and development and is also directly related to biomass
production. For example, temperature governs photosynthesis and carbon loss, and pre-
cipitation is responsible for available water, affecting the nutrient uptake, leaf area index,
and overall stand productivity. Therefore, moving any plant species away from the opti-
mum climatic conditions may create negativities in development [
17
,
18
]. The topographic
variables, especially elevation along with slope and aspect, which control the spatial and
temporal distribution of climatic parameters such as temperature, precipitation, and solar
radiation, have a prominent influence on stand productivity and species composition [
19
].
The last ecological factor used in such studies is soil, whose physical and chemical proper-
ties (texture, bulk density, available water capacity, organic carbon, electrical conductivity,
and pH) have an essential effect on plant productivity via either root growth and nutrient
uptake or directly or indirectly altering soil aeration and respiration. Soils may also affect
the tree species that can be established in a particular area, resulting in functional traits
found there [20].
Recently, modeling studies have used ecological variables extracted from freely avail-
able spatial datasets, e.g., satellite images, which make modeling efforts more time, labor,
and cost efficient. The normalized difference vegetation index (NDVI) is one of the most
used variables obtained from images and correlated with site productivity and biomass be-
sides others, such as the normalized difference water index (NDWI), vegetation difference
index (VDI), and topographic wetness index (TWI) [5,21–23].
While topographic variables (e.g., latitude–longitude, elevation, slope, aspect, and
distance to the sea) can be extracted from maps generated using a digital elevation model
(DEM) [
24
], climatic (e.g., temperature and precipitation) and edaphic variables (e.g.,
organic carbon, clay content, and bulk density) are available from some digital platforms,
such as worldclim.org [
25
] and OpenLandMap.org [
26
–
28
]. Recently, some other databases
or platforms, such as Google Earth Engine (GEE) and Microsoft’s Planetary Computer
(MPC), which encourage users to analyze, manipulate, and download spatial data, have
been used by some researchers [29].
Oriental beech (Fagus orientalis Lipsky) is a commercial tree species in Turkey. Its
growth and productivity need to be determined practically in a timely manner. This species
natively spreads from the Balkans and Turkish Thrace to the Caucasus and Crimea, crossing
the Strandja Mountains, Istanbul, and Kocaeli Peninsula. It also has a narrow distribution
area in northern Aegean [
30
]. This species has a distribution area of 1,878,049 ha, covering
around 8% of Turkey’s forestlands, and ranks fourth among all native species in terms of
area coverage [31].
The present study aimed to model and map the SDH of oriental beech stands by regres-
sion tree (RT) and multilinear regression techniques using readily available environmental
variables. It is believed that the results of this study are vital in decision-making processes
in forest management and land planning.
2. Materials and Methods
2.1. Site Description
The research area is within the border of Ayancık, Sinop, and Turkeli Forest Enterprises,
affiliated with the Sinop Regional Directorate of Forestry. The study area is situated between
41
◦
40
0
50
00
–42
◦
05
0
53
00
N latitudes and 34
◦
13
0
28
00
–35
◦
12
0
40
00
E longitudes (Figure 1). There
is a sharp increase in elevation in the study area, except for the narrow coastal plains
in Ayancik. The minimum altitude starts from sea level and reaches 1500–1800 m in the
Eastern part of the Mountains ˙
Isfendiyar along Sinop Province [32].
The area’s geological structure mainly comprises Upper Cretaceous, Eocene, and
Neogene-aged sedimentary rocks and Quarternary-aged marine deposits [
33
]. There are
mostly four great soil groups in the study area according to the USDA soil taxonomy:
Sustainability 2023,15, 14580 3 of 20
hapludults, dystrudepts, hapludalfs, and udivitrands affiliated to the order of ultsiols,
inceptisols, alfisols, and mollisols, respectively [
34
]. The climate in Sinop is a typical Black
Sea regime: while the mean annual total precipitation ranges from 675 mm to 1012 mm,
the mean annual temperature changes between 13.2
◦
C and 14.1
◦
C. More than 75% of
the annual precipitation falls in the winter and fall. The minimum and maximum mean
temperature changes between 7.4–11.0
◦
C and 16.7–18.6
◦
C, respectively [
35
]. The area’s
land cover comprises mostly forest (62%) and agricultural lands (35%). The remaining area
includes pasture/grasslands, artificial surfaces, water bodies, wetlands, and bare lands
(Figure 1).
Sustainability 2023, 15, x FOR PEER REVIEW 3 of 20
The area’s geological structure mainly comprises Upper Cretaceous, Eocene, and Ne-
ogene-aged sedimentary rocks and Quarternary-aged marine deposits [33]. There are
mostly four great soil groups in the study area according to the USDA soil taxonomy:
hapludults, dystrudepts, hapludalfs, and udivitrands affiliated to the order of ultsiols, in-
ceptisols, alfisols, and mollisols, respectively [34]. The climate in Sinop is a typical Black
Sea regime: while the mean annual total precipitation ranges from 675 mm to 1012 mm,
the mean annual temperature changes between 13.2 °C and 14.1 °C. More than 75% of the
annual precipitation falls in the winter and fall. The minimum and maximum mean tem-
perature changes between 7.4–11.0 °C and 16.7–18.6 °C, respectively [35]. The area’s land
cover comprises mostly forest (62%) and agricultural lands (35%). The remaining area in-
cludes pasture/grasslands, artificial surfaces, water bodies, wetlands, and bare lands (Fig-
ure 1).
Figure 1. Location of the study area and its land cover classes according to the Cover Corine Land
[36].
The vegetation in Sinop is mainly composed of two types: humid forests and pseudo
maquis. While the former type includes oriental beech (Fagus orientalis), oak (Quercus sp.),
black pine (Pinus nigra), Calabrian pine (Pinus brutia), fir (Abies nordmanniana), and Scotch
pine (Pinus sylvestris), the latter includes laurel (Laurus nobilis), Irish strawberry (Arbutus
unedo), saunders (Arbutus andrachne), heather (Erica arborea), phllyrea (Phillyrea latifolia),
bushy juniper (Juniperus oxycedrus), rockrose (Cistus sp.), and terebinth (Pistacia sp.), along
with other fruit trees
[
37
]
.
Figure 1.
Location of the study area and its land cover classes according to the Cover Corine Land [
36
].
The vegetation in Sinop is mainly composed of two types: humid forests and pseudo
maquis. While the former type includes oriental beech (Fagus orientalis), oak (Quercus sp.),
black pine (Pinus nigra), Calabrian pine (Pinus brutia), fir (Abies nordmanniana), and Scotch
pine (Pinus sylvestris), the latter includes laurel (Laurus nobilis), Irish strawberry (Arbutus
unedo), saunders (Arbutus andrachne), heather (Erica arborea), phllyrea (Phillyrea latifolia),
Sustainability 2023,15, 14580 4 of 20
bushy juniper (Juniperus oxycedrus), rockrose (Cistus sp.), and terebinth (Pistacia sp.), along
with other fruit trees [37].
2.2. Data Collection and Analyses
2.2.1. Field Data Collection
The study was carried out during the summer seasons of 2008, 2009, and 2011 in the
oriental beech stands of the regional forestry directorate. To that end, 222 circular sample
plots (0.04 ha) were distributed to the entire area in GIS using a stratified random sampling
design [
38
]. We ensured the sample plots were fully stocked and showed no insect or
disease trace.
Some environmental variables regarding topography, climate, and soil were used to
predict/model the SDH, computed by taking the average height of four dominant trees.
The average age of the stands was calculated by counting four dominant trees’ rings on the
cores at breast height (1.3 m) and then adding ten years to each for seedling age [1,39].
The longitude (LONG), latitude (LAT), aspect (ASP), and elevation (ELEV) data were
recorded by a handheld GPS, and the ground slope (SLP) was measured using an incli-
nometer on each plot. The slope position (SPOS) was noted according to the approach
described in Schoeneberger’s study [40].
A soil pit was dug up to the bedrock and described according to Cepel’s method [
41
],
and disturbed soil samples were taken from the topsoil (at 0–20 cm soil depth) in each plot.
Some physical (sand, silt, and clay content and available water capacity) and chemical (pH,
organic carbon, electrical conductivity, and lime) properties were also determined on each
sample in addition to measuring the soil depth.
In the next step, the soil samples were transferred to the lab. The soil texture and
available water capacity (AWC) were determined following the methods of Bouyoucos [
42
]
and pressure desorption [
43
]. The soils’ pH and EC, organic carbon (OC), and lime content
were determined following the glass electrode, Scheibler [
43
], and Walkley–Black wet
digestion methods, respectively [44].
2.2.2. Spatial Data Extraction
We used independent variables extracted from the maps shown in Figures 2and 3.
These maps were generated from various data sources and web platforms, like the DEM
and GEE. The LONG, LAT, and ELEV datasets were extracted from the DEM using ArcMap
conversion tools (i.e., feature to raster) and spatial analyst tools (i.e., extract values to
points). The slope (SLP) and aspect (ASP) values were obtained from the surfaces created
using the DEM through the ArcMap spatial analyst tool. Then, the aspect values were
transformed to the radiation index (TRASP), ranging from 0.0 on the NNE slopes to 1.0 on
the SSW slopes, which was calculated using Equation (1) [45].
TRASP = [1 −cos ((π/180) (ASP −30))]/2 (1)
Additionally, the distance to the sea (DTS) was created using spatial analyst tools (i.e.,
distance) [
46
]. The coastline vector dataset was obtained from Kelso and Patterson [
47
]. The
other variable to be correlated to the SDH was the normalized difference vegetation index
(NDVI) obtained from NASA using the Google Earth Engine platform (Didan, 2015). The
NDVI, calculated by Equation (2), serves as a measure of the quantity, health, and spread
of green plants within a region, achieved by assessing the spectral reflectance disparity
between the red (Red) and near-infrared (NIR) bands of an image [48]:
NDVI = (NIR −Red)/(NIR + Red) (2)
Sustainability 2023,15, 14580 5 of 20
Sustainability 2023, 15, x FOR PEER REVIEW 5 of 20
Figure 2. The sources of topographic variables used to model dominant height ((a): LONG, (b): LAT,
(c): ELEV, (d): DTS, (e): TRASP, (f): SLP, (g): NDVI).
Figure 2.
The sources of topographic variables used to model dominant height ((
a
): LONG, (
b
): LAT,
(c): ELEV, (d): DTS, (e): TRASP, (f): SLP, (g): NDVI).
Sustainability 2023,15, 14580 6 of 20
Sustainability 2023, 15, x FOR PEER REVIEW 6 of 20
Figure 3. The sources of independent variables used to model dominant height ((a): MAT, (b): MA-
MINT, (c): MAMAXT, (d): MATP, (e): BD, (f): FC, (g): OC, (h): pH).
Figure 3.
The sources of independent variables used to model dominant height ((
a
): MAT,
(b): MAMINT, (c): MAMAXT, (d): MATP, (e): BD, (f): FC, (g): OC, (h): pH).
Sustainability 2023,15, 14580 7 of 20
The climate variables such as the mean annual temperature (MAT), mean annual
minimum temperature (MAMINT), mean annual maximum temperature (MAMAXT), and
total precipitation (MATP) were extracted from the spatial climate surfaces developed
by Yener [
49
] at a spatial resolution of 0.005
◦×
0.005
◦
(approximately 600 m). The soil
variables, such as the bulk density (BD), OC, field capacity (FC), and pH, were extracted
from the maps provided by Hengl and Wheeler [26] and Hengl [27,28].
2.2.3. Analysis and Mapping
First, Pearson’s correlation analysis and partial dependence plots (PDP) were used to
analyze the linear and nonlinear relations between the SDH and the other variables. The
smoothed PDPs were created for each explanatory variable using the pdp package in the R
programming language [
50
]. The PDPs demonstrated whether each predictor affected the
response variable while preserving the average of the remaining predictors [
8
,
14
]. After
determining the correlations between the variables, MLR and RT analyses were performed
to model the SDH using the site variables. While the linear and nonlinear relations were
determined using R [51], DTREG was used in the modeling [52].
After the RT and MLR models were developed, potential dominant height maps
regarding oriental beech stands were generated using Map Algebra in the spatial analyst
tools in ArcMap [
46
]. In this process, the model equations were entered into the software,
and the topographic, climatic, and edaphic surfaces in the raster format were introduced to
the program as the independent variables (to be used in equations) and then the tool was
run [
53
,
54
]. While the resolution of the predicted productivity maps was 0.005, equaling
600 m, the used surfaces’ resolution was 30 m for the topographic, 250 m for the edaphic,
and 600 m for the climatic attributes.
3. Results and Discussion
This section presents three subheadings: field-based data, relationships between the
stand dominant height (SDH) and spatial data, and predicting and mapping SDH.
3.1. Field-Based Data
The field data comprised the stand dominant height, stand age, and various soil prop-
erties, including the soil depth, texture, available water capacity, pH, electrical conductivity,
organic matter, and lime (Table 1). While the oriental beech SDH ranged between 14.6 m
and 33.4 m with an average of 22.8 m, the stand ages ranged between 29 and 148 with
an average of 73 years. The soil depths across the sample plots changed between 40 and
145 cm, averaging 105.2 cm. Oriental beech is well grown moderate and deep soils [
55
].
The average soil depth in oriental beech stands ranged from 70 cm to 95.2 cm in other
studies [56–59].
The soil texture in the sampled plots was mainly composed of loamy clay (42.0%),
heavy clay (14.3%), sandy clay loam (13.4%), sandy loam (11.8%), and others (18.5%)
(Figure 4a).
Changing topography, climate, parent material, and living organisms’ impact shape
the soil texture over time. The soil texture of oriental beech forests varies from heavy
clay [60] to loam and sandy loam [61].
The pH ranged between 3.8 and 7.4, with an average of 5.3 (Table 1, Figure 4b).
Strongly acid soils dominate the study area with an area share of 45% (Figure 4b). The soil
pH range in our study is consistent with those reported in other studies [
62
–
65
]. The EC was
between 0.1 mS/cm and 6.48 mS/cm, with an average of 0.9 mS/cm. The salinity classes
in the study area were formed as non-saline (91%), slightly saline (8%), and moderately
saline (1%) (Figure 4c). The soil salinity of beech stands is generally low and characterized
as non-saline in most studies [
57
,
60
,
66
]. The soils’ organic matter and lime content were
2–10.5% and 0–17.4%, with an average of 3.4% and 1.9%, respectively (Table 1, Figure 4d).
The soil organic matter content classes were characterized as high (39%), moderate (31%),
very high (20%), and others (10%) (Figure 4d). Topographic and climatic factors affected
Sustainability 2023,15, 14580 8 of 20
the soil OC, along with the soil texture, and the stand characteristics. Therefore, the average
OC content of beech stands varied from 1.9–3.0% [
61
,
64
] to 3.6–3.9% [
55
,
60
,
67
]. The AWC
ranged between 1.9% and 23.3%, with an average of 12.3% (Table 1).
Table 1. Some descriptive statistics regarding observed and digitally extracted variables.
Variables Abbr. Minimum Maximum Mean SE SD
Digitally extracted
variables
Longitude (◦)
LONG
34.3 35.0 34.7 <0.1 0.2
Latitude (◦) LAT 41.7 42.0 41.9 <0.1 0.1
Elevation (m) ELEV 12.0 1352.0 630.9 21.8 324.6
Distance to the sea (m) DTS 692.5 24,935.8 12,279.0 385.3 5741.2
Transformed aspect
TRASP
0.0 2.0 1.1 <0.1 0.7
Slope (%) SLP 5.0 84.4 31.8 1.1 16.8
Vegetation index NDVI 0.6 0.7 0.7 <0.1 <0.1
Mean temperature (◦) MAT 6.8 14.1 10.6 0.1 1.8
Minimum mean
temperature (◦)
MINMT
2.5 11.3 7.0 0.1 2.2
Maximum mean
temperature (◦)
MAXMT
11.5 17.5 14.8 0.1 1.5
Total precipitation (mm) TP 806.4 1072.9 925.4 3.6 54.3
Soil bulk density (kg/m3)BD 10.0 13.8 11.5 0.1 0.9
Soil field capacity (%) FC 30.0 40.0 33.5 0.1 2.0
Soil organic carbon (%) OC 0.8 2.8 1.6 <0.1 0.4
Soil pH pH 5.3 6.7 6.0 <0.1 0.3
Observed variables
Stand dominant height (m) SDH 14.6 33.4 22.8 0.2 3.7
Stand age (year) SA 29.0 148.0 73.3 1.4 21.2
Soil depth (cm)
SDEPTH
40.0 145.0 105.2 1.2 18.2
Sand content (%)
SAND
21.0 90.0 51.3 0.9 14.3
Silt content (%) SILT 3.0 39.0 18.5 0.4 6.0
Clay content (%) CLAY 5.0 60.0 30.2 0.8 12.0
Available water capacity (%)
AWC 1.9 23.3 12.3 0.2 3.3
Soil pH pH 3.8 7.4 5.3 0.1 0.8
Soil electrical conductivity
(mS/cm) ECe 0.1 6.48 0.9 0.1 0.9
Soil organic carbon (%) OC 0.2 10.5 3.4 0.1 2.0
Soil lime content (%) LIME 0.0 17.4 1.9 0.1 2.1
Sustainability 2023,15, 14580 9 of 20
Sustainability 2023, 15, x FOR PEER REVIEW 9 of 20
Figure 4. Distribution of sample plots to soil properties classes ((a): texture, (b): acidity, (c): salinity,
(d): organic matter).
3.2. Relationships between the Stand Dominant Height (SDH) and Spatial Data
This study used digitally extracted data as the explanatory variables to model and
map the oriental beech stands’ dominant height. While the Pearson correlation deter-
mined the relationships between the SDH and the explanatory variables (Figure 5), the
partial dependence plots visualized these relations (Figure 6). The variables included were
topographic (i.e., LONG, LAT, ELEV, DTS, TRASP, SLP, and NDVI), climatic (i.e., MAT,
MAMINT, MAMAXT, and MATP), and edaphic (i.e., BD, FC, OC, and pH).
Assessing the spatial data retrieved from various resources (Table 1), we saw that our
sample plots were located between 34.3°–35.0° N longitudes and 41.7°–42.0° E latitudes.
The ELEV ranged from 12 m near sea level to 1352 m with an average of 630.9 m. The
average DTS was 12,279 m in the area and reached up to 24,935 m. The slope of the sample
plots changed between 5 and 84.4%, with an average of 31.8% corresponding to steep
slopes. The NDVI also assessed as part of the topographic variables averaged 0.66, ranging
between 0.59 and 0.75, which is relatively high.
Of these variables, the ELEV was positively correlated (r = 0.36), and the LAT (r =
−0.25) and NDVI (r = −0.24) were negatively correlated with the SDH. No significant rela-
tionships existed with the other topographic variables (p > 0.05) (Figure 5). The DTS, LAT,
and ELEV are essential factors in determining the regional climate parameters, such as
temperature, precipitation, and radiation, affecting forest productivity [19,68]. While the
latitude impacts those parameters’ distribution, increasing the ELEV shortens the growing
period due to the decreasing temperature and increases the relative humidity [69]. The
negative effect of reducing the ELEV on the SDH could be attributed to the optimum alti-
tudinal zone for the oriental beech stands in Sinop, which is between 600 m and 1200 m
with an average of 1000 m [70].
Figure 4.
Distribution of sample plots to soil properties classes ((
a
): texture, (
b
): acidity, (
c
): salinity,
(d): organic matter).
3.2. Relationships between the Stand Dominant Height (SDH) and Spatial Data
This study used digitally extracted data as the explanatory variables to model and
map the oriental beech stands’ dominant height. While the Pearson correlation deter-
mined the relationships between the SDH and the explanatory variables (Figure 5), the
partial dependence plots visualized these relations (Figure 6). The variables included were
topographic (i.e., LONG, LAT, ELEV, DTS, TRASP, SLP, and NDVI), climatic (i.e., MAT,
MAMINT, MAMAXT, and MATP), and edaphic (i.e., BD, FC, OC, and pH).
Assessing the spatial data retrieved from various resources (Table 1), we saw that our
sample plots were located between 34.3
◦
–35.0
◦
N longitudes and 41.7
◦
–42.0
◦
E latitudes.
The ELEV ranged from 12 m near sea level to 1352 m with an average of 630.9 m. The
average DTS was 12,279 m in the area and reached up to 24,935 m. The slope of the sample
plots changed between 5 and 84.4%, with an average of 31.8% corresponding to steep
slopes. The NDVI also assessed as part of the topographic variables averaged 0.66, ranging
between 0.59 and 0.75, which is relatively high.
Of these variables, the ELEV was positively correlated (r = 0.36), and the LAT (
r = −0.25
)
and NDVI (r =
−
0.24) were negatively correlated with the SDH. No significant relationships
existed with the other topographic variables (p> 0.05) (Figure 5). The DTS, LAT, and ELEV
are essential factors in determining the regional climate parameters, such as temperature,
precipitation, and radiation, affecting forest productivity [
19
,
68
]. While the latitude impacts
those parameters’ distribution, increasing the ELEV shortens the growing period due to
the decreasing temperature and increases the relative humidity [
69
]. The negative effect
of reducing the ELEV on the SDH could be attributed to the optimum altitudinal zone for
the oriental beech stands in Sinop, which is between 600 m and 1200 m with an average of
1000 m [70].
Sustainability 2023,15, 14580 10 of 20
Sustainability 2023, 15, x FOR PEER REVIEW 10 of 20
Figure 5. Correlation matrix showing the linear relationships between explanatory variables and
SDH. The blank boxes indicate the nonsignificant relationships between variables, and the numbers
in the boxes show the correlation coefficients.
Figure 6. Partial dependence plots showing the nonlinear relations between dependent and inde-
pendent variables.
Figure 5.
Correlation matrix showing the linear relationships between explanatory variables and
SDH. The blank boxes indicate the nonsignificant relationships between variables, and the numbers
in the boxes show the correlation coefficients.
Sustainability 2023, 15, x FOR PEER REVIEW 10 of 20
Figure 5. Correlation matrix showing the linear relationships between explanatory variables and
SDH. The blank boxes indicate the nonsignificant relationships between variables, and the numbers
in the boxes show the correlation coefficients.
Figure 6. Partial dependence plots showing the nonlinear relations between dependent and inde-
pendent variables.
Figure 6.
Partial dependence plots showing the nonlinear relations between dependent and indepen-
dent variables.
Sustainability 2023,15, 14580 11 of 20
Therefore, increasing the elevation improved the SDH. The altitudinal zone could also
explain the negative correlation between the LAT and SDH. This is because increasing the
LAT means decreasing the elevation (r =
−
0.9) (Figure 5). The partial dependency plots,
showing nonlinear relations, indicated that the SDH increased with the ELEV increasing up
to 1000 m and then decreased (Figure 6), which could be proof of the optimum altitudinal
zone being below 1000 m for oriental beech stands. Similar results on the effect of the
ELEV on stand productivity were found in other studies [
7
,
12
,
14
,
58
,
71
]. For example, Alavi
et al. [
7
] also attributed increased productivity in oriental beech stands of Hyrcania, Iran,
to increasing the ELEV to the optimum altitudinal zone, above 1500 m. Güner et al. [
72
]
linked this positive effect of the ELEV on Anatolian black pine productivity to improved
precipitation at higher altitudes. However, other authors [
4
,
14
,
15
] reported conflicting
findings due to deteriorated ecological conditions with increasing ELEV, such as shortening
the growing period, decreasing fine-textured soil, and reducing decomposition due to
the condensing temperature. The reducing effect of LAT on stand productivity was also
reported in some studies [
14
,
73
]. Klinka et al. [
73
] reported a decrease of 2.9 m in the
Norway spruce SI and 2.5 m in the Douglas fir SI with a 1
◦
increase in latitude and 100 m
in elevation.
Another topographic variable correlated with the SDH was the NDVI, whose relatively
higher value above 0.6 in the study area indicates dense vegetation cover. Its weak negative
correlations with the SDH could also be linked to somewhat higher altitudes (r =
−
0.69 be-
tween the NDVI and ELEV, Figure 5), regarded as the oriental beech’s growing optimum. It
is also associated with decreasing precipitation in beech sites due to going southward, neg-
atively affecting the NDVI (r = 0.76 between the NDVI and LAT,
Figures 3d and 5) [74,75]
.
On the contrary, some other researchers found positive correlations between the NDVI and
site productivity [21,22,76–78].
The climatic variables used in this study were extracted from the climate surfaces devel-
oped by Yener [
49
]. According to this dataset, the study area’s climate is characterized by an
average of 10.6
◦
C mean, 7.0
◦
C mean minimum, and 14.8
◦
C mean maximum temperatures
with minimum–maximum values of 6.8–14.1
◦
C, 2.5–11.3
◦
C, and
11.5–17.5 ◦C
, respec-
tively (Table 1, Figure 3a–c). The precipitation in the study area ranged from
806.4 m
m to
1072.9 m
m and averaged 925.4 mm (Table 1, Figure 3d). Climate, one of the main ecological
factors affecting terrestrial ecosystems at global and local scales, is the driving force for
the biogeochemical cycle in nature [
17
,
69
]. Variable climatic conditions also affect biomass
and soil litter decomposition and carbon accumulation [
79
]. Therefore, the MAT, MAMINT,
MAMAXT, and MATP were used in this study; however, the MAMAXT was the only
climatic factor significantly correlated with the SDH (r = −0.39) (Figure 5).
Our finding is similar to those reported in other studies [
12
,
13
,
80
,
81
]. The negative
effect of the MAMAXT on the SDH may be attributed to enhanced summer drought [
80
].
Seltmann et al. [
13
] reported improved Norway spruce growth because of the low temper-
ature and high AWC. However, the temperature in most studies [
14
,
15
,
82
,
83
] positively
affected tree growth, attributed to prolonged growing periods, enhancing microbial activ-
ities and improving decomposition and soil attributes. Although no significant effect of
precipitation on the SDH was observed in this study, some other studies [
13
,
72
,
80
,
81
,
83
]
reported that it somehow affects stand productivity. Likewise, we detected no significant
climatic variables, except for the MAMAXT. The partial dependence plots (Figure 6) showed
that the SDH increased with an increasing MAT and MAMINT and MATP up to about
10 ◦C, 8 ◦C, and 950 mm, respectively, and then suddenly decreased.
The digitally extracted BD, FC, OC, and pH averaged 11.5 kg/m
3
, 33.5%, 1.6%, and
6.0, respectively, and ranged between 10.0 and 13.8 kg/m
3
, 30 and 40%, 0.8 and 2.8%, and
5.3 and 6.7, respectively (Table 1, Figure 3e–h). All the digitally extracted edaphic variables
significantly affected the SDH: while the BD (r =
−
0.34) and pH (
−
0.25) were negatively
associated with the SDH, the FC (r = 0.26) and OC (r = 0.43) positively affected it (Figure 5).
The positive effect of the FC or FC-based AWC on stand productivity was reported in
other studies [
8
,
13
,
58
,
59
]. Soil moisture or water is one of the most important ecological
Sustainability 2023,15, 14580 12 of 20
factors affecting plant growth, especially in influencing soil temperature, aeration, microbial
activity, and nutrient availability and diminishing the toxic material concentrations, besides
directly providing water to the plants [
84
]. The situation in which the OC was positively
correlated with the SDH could be attributed to the provided functions of organic matter on
soil quality through an increased cation exchange and available water capacity, improving
soil aggregates, aeration, and porosity and providing nutrients [
85
,
86
]. Similar findings
were reported by Aertsen et al. [
87
] and Subedi and Fox [
88
]. Results, in contrast to ours,
were also found by some other researchers [
4
,
58
], attributing their outcomes to decreased
pH with increasing organic matter. Unlike the FC and OC, the BD and pH negatively
affected productivity in the present study. An indicator of soil acidity, pH mainly affects
some processes, such as nutrient availability, nitrification, microbial activity, and heavy
metal toxicity [
89
]. Most of the studies [
71
,
87
,
88
] reported a positive effect of pH on
productivity, attributing it to improving soil conditions, such as nutrient availability and
microbial activity. However, some other researchers [
7
,
64
] reported contrasting results in
disagreement with our study. Our results showed that pH is one of the limiting factors for
the dominant height growth of oriental beech. An increasing pH negatively affects some
elements’ availability in the soil solution, such as phosphorus, iron, zinc, and manganese [
7
].
The BD is calculated by summing the soil pores and solids to the soil volume; its unit is
gr/cm
3
. A low BD indicates a loose and highly porous structure due to the high organic
matter in soils [
86
], consistent with our finding in Figure 5, showing r =
−
0.7 between the
BD and OC. Increases in BD generally result in adverse effects on tree development through
the negatively affected mentioned soil properties [
90
]. As a matter of fact, Irmak [
91
]
and Cepel [
41
] stated that the effect of soil properties on stand development becomes
more evident when the climate is not very variable, and the species moves away from the
optimum conditions.
3.3. Predicting and Mapping SDH
The presented study aimed to predict the SDH from spatial data, including topo-
graphic, climatic, and edaphic variables, using MLR and RT analyses and then generating
potential SDH maps using those models. We randomly selected 75% of the sample plots
as the training datasets in this context, withholding the remaining 25% as validation. The
model results showed that MLR accounted for 26% and 21% of the variation in the SDH,
with an RMSE of 3.16 m and 3.27 m for training and validation, respectively. RT out-
performed MLR, explaining 57% and 25% of the variation, with an RMSE of 2.37 m and
3.33 m
for training and validation, respectively (Figure 7). The RT model included at least
one variable from any group, such as topographic, climatic, and edaphic; however, MLR
included only climatic and topographic variables.
Sustainability 2023, 15, x FOR PEER REVIEW 13 of 20
In an RT analysis, a particular site productivity indicator, such as the dominant
height and site index, is classified by considering the predictors (independent or explan-
atory variables). The precise value of the predictor optimally splits the data at any branch
of the tree [100].
The RT model in this study included seven independent variables: edaphic (BD, pH,
OC, and FC), topographic (LAT and NDVI), and climatic (MAMAXT). The soil BD, with a
variable importance of 37.5%, followed by the LAT (18.5%) and pH (13.9%) (Figures 8 and
9), is the first and primary variable controlling the SDH. At this point, a BD below or equal
to 1.1 gr/cm
3
resulted in an SDH of 24.6 m. However, a BD above that value corresponded
to an SDH of 21.4 m., which was 13% lower than the previous one, suggesting that a de-
creased BD (below 1.1 gr/cm
3
) improves the SDH. The second important node was an LAT
above 41.8° interacting with a BD equal to or below 1.1 gr/cm
3
, an OC equal to or below
2.1%, and an FC above 34.5%, resulting in a 27.8 m SDH, the highest value in this model
followed by the BD (equal to or below 1.1 gr/cm
3
) and OC (above 2.1%) interactions corre-
sponding to the 27.2 m SDH.
Figure 7. Correlations and error metrics between observed and predicted SDH ((a): RT model, (b):
MLR model).
Figure 8. Variable importance in models.
Figure 7.
Correlations and error metrics between observed and predicted SDH ((
a
): RT model,
(b): MLR model).
The most critical variable in the models was MAMAXT for MLR and BD for RT, with
an importance of 62.3% and 37.5%. The most critical factor group affecting the SDH was the
Sustainability 2023,15, 14580 13 of 20
climate, with an importance rate of 62.3%, and the soil, with an importance rate of 61.2%
for the MLR and RT models (Figure 8).
Sustainability 2023, 15, x FOR PEER REVIEW 13 of 20
In an RT analysis, a particular site productivity indicator, such as the dominant
height and site index, is classified by considering the predictors (independent or explan-
atory variables). The precise value of the predictor optimally splits the data at any branch
of the tree [100].
The RT model in this study included seven independent variables: edaphic (BD, pH,
OC, and FC), topographic (LAT and NDVI), and climatic (MAMAXT). The soil BD, with a
variable importance of 37.5%, followed by the LAT (18.5%) and pH (13.9%) (Figures 8 and
9), is the first and primary variable controlling the SDH. At this point, a BD below or equal
to 1.1 gr/cm
3
resulted in an SDH of 24.6 m. However, a BD above that value corresponded
to an SDH of 21.4 m., which was 13% lower than the previous one, suggesting that a de-
creased BD (below 1.1 gr/cm
3
) improves the SDH. The second important node was an LAT
above 41.8° interacting with a BD equal to or below 1.1 gr/cm
3
, an OC equal to or below
2.1%, and an FC above 34.5%, resulting in a 27.8 m SDH, the highest value in this model
followed by the BD (equal to or below 1.1 gr/cm
3
) and OC (above 2.1%) interactions corre-
sponding to the 27.2 m SDH.
Figure 7. Correlations and error metrics between observed and predicted SDH ((a): RT model, (b):
MLR model).
Figure 8. Variable importance in models.
Figure 8. Variable importance in models.
Many studies [
4
,
12
,
15
,
16
,
71
,
72
,
82
,
92
] have used MLR and RT for estimating for-
est productivity. Our model results are consistent with those reported in other stud-
ies
[4,5,12,15,72,92–94]
. The most satisfied results in these outperformed models were
reached as an adjusted R
2
of 0.85 and RMSE of 1.17 m for the training dataset and an R
2
of 0.54 and RMSE of 1.91 m for validation by Alavi et al. [
7
] in the Hyrcanian oriental
beech forests of Iran using edaphic and physiographic variables. Mohammadi et al. [
5
]
also predicted oriental beech productivity, e.g., stand volume using MLR and RT tech-
niques, and RT outperformed with an R
2
of 0.67 (percentage RMSE = 30%). Some other
statistical approaches have also been implemented to model forest productivity, such
as the complementary methodological approach [
95
], random forest analysis [
14
,
96
,
97
],
Chapman–Richards model [83,98], and linear mixed effects models [99].
In an RT analysis, a particular site productivity indicator, such as the dominant height
and site index, is classified by considering the predictors (independent or explanatory
variables). The precise value of the predictor optimally splits the data at any branch of the
tree [100].
The RT model in this study included seven independent variables: edaphic (BD,
pH, OC, and FC), topographic (LAT and NDVI), and climatic (MAMAXT). The soil
BD, with a variable importance of 37.5%, followed by the LAT (18.5%) and pH (13.9%)
(
Figures 8and 9
), is the first and primary variable controlling the SDH. At this point, a
BD below or equal to 1.1 gr/cm
3
resulted in an SDH of 24.6 m. However, a BD above
that value corresponded to an SDH of 21.4 m., which was 13% lower than the previous
one, suggesting that a decreased BD (below 1.1 gr/cm
3
) improves the SDH. The second
important node was an LAT above 41.8
◦
interacting with a BD equal to or below 1.1 gr/cm
3
,
an OC equal to or below 2.1%, and an FC above 34.5%, resulting in a 27.8 m SDH, the
highest value in this model followed by the BD (equal to or below 1.1 gr/cm
3
) and OC
(above 2.1%) interactions corresponding to the 27.2 m SDH.
Sustainability 2023,15, 14580 14 of 20
Sustainability 2023, 15, x FOR PEER REVIEW 14 of 20
Figure 9. Regression tree model to predict SDH of oriental beech stands.
Figure 9. Regression tree model to predict SDH of oriental beech stands.
Sustainability 2023,15, 14580 15 of 20
This RT model suggests that it is vital to establish oriental beech stands where the BD
is below 1.1 gr/cm
3
and the OC is above 2.1%, or the BD is equal to or below 1.1 gr/cm
3
,
the OC is equal to or below 2.1%, the FC is above 34.5%, and the LAT is above 41.8
◦
. It
also suggests that the sites with a relatively higher BD (>1.1 gr/cm
3
) with a pH equal to or
below 6.15 and an LAT below 41.9
◦
are more productive than sites with more than 6.15 pH.
Afif-Khouri et al. [
71
] and Menendez-Miguelez et al. [
82
] also used RT to predict the
stand productivity in NW Spain chestnut coppice stands. While extractable Mg and annual
temperature were the only determinants in the former study, the latter research determined
the clay content of soils and the spring and summer precipitation as nodes in the RT. On
the other hand, the MLR model comprises the MAMAXT, LAT, and NDVI, which were also
included in the RT model.
We also mapped the potential productivity of the oriental beech forest in the study
area. The resolution of the maps shown in Figure 10 is 0.005
◦
(600 m/pix). According
to the maps, the stand dominant height was between 22.4 and 43.3 m for MLR and 17.3
and 27.8 m for the RT model. However, with some differences in the model maps, the
productivity increases toward the area’s inner part, especially to the southwest (Figure 10).
This improvement could be explained by enhanced ecological conditions in this part, such
as an elevated OC (Figure 3g) and FC (Figure 3f), a decreased BD (Figure 3e), and reaching
the optimum altitudinal zone in terms of air temperature (Figure 2c), suggesting that coastal
areas are unsuitable for oriental beech in this region.
Sustainability 2023, 15, x FOR PEER REVIEW 16 of 20
Figure 10. Spatial distribution of oriental beech SDH across the study area. Map generated by MLR
(a) and RT (b) models.
4. Conclusions
Because there is insufficient knowledge of the spatial variation of species-specific site
productivity, generating dominant height maps is crucial for forest management prac-
tices, such as reforestation and plantation. The maps generated in this study could help
forest planners and land managers visualize the most productive oriental beech sites rap-
idly. Our approach can also be used in locations with different vegetation types if site-
specific spatial data are available.
Figure 10. Spatial distribution of oriental beech SDH across the study area. Map generated by MLR
(a) and RT (b) models.
Sustainability 2023,15, 14580 16 of 20
The other point is decreased productivity in the region’s eastern part, which can
be attributed to the increased temperatures and decreased organic matter. Other re-
searchers
[6,23,101–103]
have also mapped forest productivity at variable spatial reso-
lutions, ranging from 5 m to 1000 m. They used soil, climate, and terrain attributes based on
different statistical techniques, like RF, RT, Chapman–Richards, and MLR. While Swenson
et al. [
103
] predicted and mapped the SI for Douglas fir stands with an R
2
of 0.55 and RMSE
of 6.1 m in the USA, Jiang et al. [
101
] found those values to be an R
2
of 0.64 and RMSE
of 4.6 m for softwood and an R
2
of 0.36 and RMSE of 4.2 m for hardwood species also in
the USA.
4. Conclusions
Because there is insufficient knowledge of the spatial variation of species-specific site
productivity, generating dominant height maps is crucial for forest management practices,
such as reforestation and plantation. The maps generated in this study could help forest
planners and land managers visualize the most productive oriental beech sites rapidly. Our
approach can also be used in locations with different vegetation types if site-specific spatial
data are available.
Based on the results, it is concluded that the RT model outperformed the MLR model
with more accurate estimates and maps. The RT model emphasizes that edaphic factors
should be given more importance than others, particularly in oriental beech afforestation
and reforestation works. Using freely available spatial datasets with more spectral data
and the RT technique, researchers can develop new site productivity models and generate
wall-to-wall maps of their area of investigation. Thus, site productivity can be assessed in a
spatially explicit manner.
Author Contributions:
E.G.—field and laboratory studies, writing and reviewing; I.Y.—original draft
preparation, methodology, statistical analyses and mapping with GIS, writing, reviewing and editing.
All authors have read and agreed to the published version of the manuscript.
Funding:
This work was supported by the TÜB˙
ITAK—TOVAG (The Scientific and Technological
Research Council of Turkey—Research Committee of Agriculture, Forestry and Veterinary), with
Project Number: 107O752.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments:
Some of the data used in this study were obtained from the doctoral thesis of
the second author. The authors thank Can Vatandaslar for his valuable English editing. We also
acknowledge the reviewers and the journal editors for their helpful comments and suggestions to
improve the manuscript.
Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design
of the study; in the collection, analyses, or interpretation of the data; in the writing of the manuscript;
or in the decision to publish the results.
References
1.
Skovsgaard, J.P.; Vanclay, J.K. Forest site productivity: A review of the evolution of dendrometric concepts for even-aged stands.
Forestry 2008,81, 13. [CrossRef]
2. García, O.; Batho, A. Top Height Estimation in Lodgepole Pine Sample Plots. West. J. Appl. For. 2005,20, 64–68. [CrossRef]
3.
Noordermeer, L.; Gobakken, T.; Næsset, E.; Bollandsås, O.M. Predicting and mapping site index in operational forest inventories
using bitemporal airborne laser scanner data. For. Ecol. Manag. 2020,457, 117768. [CrossRef]
4.
Yener, I.; Altun, L. Predicting Site Index for Oriental Spruce (Picea orientalis L. (Link)) Using Ecological Factors in the Eastern
Black Sea, Turkey. Fresenius Environ. Bull. 2018,27, 3107–3116.
5.
Mohammadi, J.; Shataee, S.; Babanezhad, M. Estimation of forest stand volume, tree density and biodiversity using Landsat ETM
plus Data, comparison of linear and regression tree analyses. Procedia Environ. Sci. 2011,7, 299–304. [CrossRef]
Sustainability 2023,15, 14580 17 of 20
6.
Laamrani, A.; Valeria, O.; Bergeron, Y.; Fenton, N.; Cheng, L.Z.; Anyomi, K. Effects of topography and thickness of organic layer
on productivity of black spruce boreal forests of the Canadian Clay Belt region. For. Ecol. Manag. 2014,330, 144–157. [CrossRef]
7.
Alavi, S.J.; Ahmadi, K.; Dormann, C.; Serra-Diaz, J.M.; Nouri, Z. Assessing the dominant height of oriental beech (Fagus orientalis
L.) in relation to edaphic and physiographic variables in the Hyrcanian Forests of Iran. Biotechnol. Agron. Soc. Environ.
2020
,24,
262–273. [CrossRef]
8.
Littke, K.M.; Harrison, R.B.; Zabowski, D. Determining the Effects of Biogeoclimatic Properties on Different Site Index Systems of
Douglas-fir in the Coastal Pacific Northwest. For. Sci. 2016,62, 503–512. [CrossRef]
9. Ozkan, K. Modelling ecological data using classification and regression tree technique (CART). Turk. J. For. 2012,13, 1–4.
10.
Prasad, A.M.; Iverson, L.R.; Liaw, A. Newer Classification and Regression Tree Techniques: Bagging and Random Forests for
Ecological Prediction. Ecosystems 2006,9, 181–199. [CrossRef]
11.
Yuan, Z.Q.; Ali, A.; Wang, S.P.; Gazol, A.; Freckleton, R.; Wang, X.G.; Lin, F.; Ye, J.; Zhou, L.; Hao, Z.Q.; et al. Abiotic and biotic
determinants of coarse woody productivity in temperate mixed forests. Sci. Total Environ. 2018,630, 422–431.
12.
Gulsoy, S.; Cinar, T. The Relationships between Environmental Factors and Site Index of Anatolian Black Pine (Pinus Nigra Arn.
Subsp. Pallasiana (Lamb.) Holmboe) Stands in Demirci (Manisa) District, Turkey. Appl. Ecol. Environ. Res.
2019
,17, 1235–1246.
[CrossRef]
13.
Seltmann, C.T.; Wernicke, J.; Petzold, R.; Baumann, M.; Munder, K.; Martens, S. The relative importance of environmental drivers
and their interactions on the growth of Norway spruce depends on soil unit classes: A case study from Saxony and Thuringia,
Germany. For. Ecol. Manag. 2021,480, 118671. [CrossRef]
14.
Zhang, H.Y.; Feng, Z.K.; Wang, S.; Ji, W.X. Disentangling the Factors that Contribute to the Growth of Betula spp. and Cunninghami
lanceolata in China Based on Machine Learning Algorithms. Sustainability 2022,14, 8346. [CrossRef]
15.
Velioglu, E.; Guener, S.T.; Karakurt, H.; Tastan, Y.; Yavuz, Z.; Tugrul, D. Relationships between site index and ecological variables
of trembling poplar forests (Populus tremula L.) in Turkiye. Environ. Monit. Assess. 2023,195, 308.
16.
Aertsen, W.; Kint, V.; van Orshoven, J.; Ozkan, K.; Muys, B. Comparison and ranking of different modelling techniques for
prediction of site index in Mediterranean mountain forests. Ecol. Model. 2010,221, 1119–1130. [CrossRef]
17.
Orlandini, S.; Bindi, M.; Howden, M. Plant biometeorology and adaptation. In Biometeorology for Adaptation to Climate Variability
and Change; Springer: Berlin/Heidelberg, Germany, 2009; pp. 107–129.
18.
Pan, Y.; Birdsey, R.A.; Phillips, O.L.; Jackson, R.B. The structure, distribution, and biomass of the world’s forests. J. Annu. Rev.
Ecol. Evol. Syst. 2013,44, 593–622.
19.
Stage, A.R.; Salas, C. Interactions of elevation, aspect, and slope in models of forest species composition and productivity. For. Sci.
2007,53, 486–492.
20.
Becknell, J.M.; Powers, J.S. Stand age and soils as drivers of plant functional traits and aboveground biomass in secondary tropical
dry forest. Can. J. For. Res. 2014,44, 604–613. [CrossRef]
21.
Brito, V.V.; Rubilar, R.A.; Cook, R.L.; Campoe, O.C.; Carter, D.R.; Mardones, O. Evaluating remote sensing indices as potential
productivity and stand quality indicators for Pinus radiata plantations. Sci. For. 2021,49, e3316. [CrossRef]
22.
Gazol, A.; Rozas, V.; Arribas, S.C.; Ponce, R.A.; Rodriguez-Puerta, F.; Gomez, C.; Olano, J.M. Stand characteristics modulate
secondary growth responses to drought and gross primary production in Pinus halepensis afforestation. Eur. J. For. Res.
2023
,142,
353–366.
23.
Zushi, K. Regional estimation of Japanese cedar (Cryptomeria japonica D. Don) productivity by use of digital terrain analysis. J. For.
Res. 2007,12, 289–297. [CrossRef]
24.
NASA/METI/AIST, Japan Spacesystems and U.S./Japan ASTER Science Team. ASTER Global Digital Elevation Model [Data
Set]. 2019. Available online: https://lpdaac.usgs.gov/products/astgtmv003/ (accessed on 9 September 2022).
25.
Hijmans, R.J.; Cameron, S.E.; Parra, J.L.; Jones, P.G.; Jarvis, A. Very high resolution interpolated climate surfaces for global land
areas. Int. J. Climatol. 2005,25, 1965–1978. [CrossRef]
26.
Hengl, T.; Wheeler, I. Soil Organic Carbon Content in x 5 g/kg at 6 Standard Depths (0, 10, 30, 60, 100 and 200 cm) at 250 m Resolution
(Version v02) [Data Set]; Zenodo: Geneva, Switzerland, 2018.
27.
Hengl, T. Clay Content in % (kg/kg) at 6 Standard Depths (0, 10, 30, 60, 100 and 200 cm) at 250 m Resolution (Version v02) [Data Set];
Zenodo: Geneva, Switzerland, 2018.
28.
Hengl, T. Soil Bulk Density (Fine Earth) 10 x kg/m-Cubic at 6 Standard Depths (0, 10, 30, 60, 100 and 200 cm) at 250 m Resolution (Version
v02) [Data Set]; Zenodo: Geneva, Switzerland, 2018.
29.
Xu, C.; Du, X.; Jian, H.; Dong, Y.; Qin, W.; Mu, H.; Yan, Z.; Zhu, J.; Fan, X. Analyzing large-scale Data Cubes with user-defined
algorithms: A cloud-native approach. Int. J. Appl. Earth Obs. Geoinf. 2022,109, 102784. [CrossRef]
30. OGM. Ormanlarımızda Yayılı¸s Gösteren Asli A˘gaç Türleri; Orman Genel Müdürlü˘gü: Ankara, Turkey, 2009; p. 48.
31. OGM, Orman Varlı ˘gımız. TC Çevre Ve Orman Bakanlı˘gı; Orman Genel Müdürlü˘gü Yayınları: Ankara, Turkey, 2020.
32.
Guvendi, E. Saf Do˘gu Kayını (Fagus orientalis Lipsky.) Ormanlarının Ekolojik Tabanlı ˙
Idare Sürelerinin Belirlenmesi (Sinop-
Ayancık-Türkeli Örne˘gi). Ph.D. Thesis, Karadeniz Technical University, Trabzon, Türkiye, 2013.
33. MTA. Geological Map of Turkey; General Directorate of Mineral Research and Exploration: Ankara, Turkey, 2019.
34. Hengl, T.; Nauman, T. Predicted USDA SOIL Great Groups at 250 m (Probabilities), v01; Zenodo: Geneva, Switzerland, 2018.
35. Özdemir, N. One of the effective natural disaster in Sinop: The landslide. Dicle Univ. J. Ziya Gökalp Fac. Educ. 2005,5, 67–106.
Sustainability 2023,15, 14580 18 of 20
36.
Cover Corine Land, European Union, Copernicus Land Monitoring Service. 2018. Available online: https://sdi.eea.europa.eu/
catalogue/copernicus/api/records/71c95a07-e296-44fc-b22b-415f42acfdf0?language=all (accessed on 9 September 2022).
37. Yıldız, S. Sinop ve Çevresinin Bitki Örtüsü. Master’s Thesis, Istanbul University, Istanbul, Türkiye, 2000. (In Turkish).
38. Pennock, D.; Yates, T.; Braidek, J. Soil sampling designs. Soil Sampl. Methods Anal. 2008,2, 25–37.
39. van Laar, A.; Akça, A. Forest Mensuration; Springer: Dordrecht, The Netherlands, 2007.
40.
Schoeneberger, P.J. Field Book for Describing and Sampling Soils; National Soil Survey Center, Natural Resources Conservation
Service, U.S. Department of Agriculture: Lincoln, NE, USA, 2002.
41. Cepel, N. Forest Ecology; Istanbul University, Faculty of Forestry: ˙
Istanbul, Turkey, 1978.
42.
Bouyoucos, G.J. Hydrometer method improved for making particle size analyses of soils. Agron. J.
1962
,54, 464–465. [CrossRef]
43.
Kalra, Y.P.; Maynard, D.G. Methods Manual for Forest Soil and Plant Analysis; Forestry Canada, Northwest Region, Northern
Forestry Centre: Edmonton, AB, Canada, 1991; p. 116.
44.
Walkley, A. A critical examination of a rapid method for determining organic carbon in soils-Effect of variations in digestion
conditions and of inorganic soil constituents. Soil Sci. 1947,63, 251–264. [CrossRef]
45.
Bolton, D.K.; White, J.C.; Wulder, M.A.; Coops, N.C.; Hermosilla, T.; Yuan, X. Updating stand-level forest inventories using
airborne laser scanning and Landsat time series data. Int. J. Appl. Earth Obs. Geoinf. 2018,66, 174–183. [CrossRef]
46. ESRI. ArcMap 10.2, 10.2; ESRI: Redlands, CA, USA, 2014.
47. Kelso, N.V.; Patterson, T. Natural earth vector. Cartogr. Perspect. 2009,64, 45–50. [CrossRef]
48.
Tucker, C.J. Red and photographic infrared linear combinations for monitoring vegetation. Remote Sens. Environ.
1979
,8, 127–150.
[CrossRef]
49.
Yener, ˙
I. Development of high-resolution annual climate surfaces for Turkey using ANUSPLIN and comparison with other
methods. Atmósfera 2022,37, 425–444. [CrossRef]
50. Greenwell, B.M. pdp: An R package for constructing partial dependence plots. R J. 2017,9, 421. [CrossRef]
51.
Team, R.C. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2013.
52.
Sherrod, P.H. DTREG Predictive Modeling Software. 2020. Available online: http://www.dtreg.com (accessed on 9
September 2022).
53.
Tomlin, C.D. Geographic Information Systems and Cartographic Modeling; Prentice Hall: Englewood Cliffs, NJ, USA, 1990; Vo
lume 2
49.
54. Tomlin, C.D. Map algebra: One perspective. Landsc. Urban Plan. 1994,30, 3–12. [CrossRef]
55.
Yıldız, O.; Bircan, ¸S. Batı Karadeniz kıyı bölgesi’nde yeti¸sen Kayın (Fagus orientalis Lipsky) odununun fiziksel ve mekanik
özelliklerinin yeti¸sme ortamı de˘gi¸skenleriyle ili¸skisi. Anadolu Orman Ara¸stırmaları Derg. 2022,8, 61–72. [CrossRef]
56.
Göl, C.; Çelik, N.; Çakır, M.; Gül, E. Türkmen Da˘gı (Evkondu Tepe) do ˘gu kayını (Fagus orientalis Lipsky.) ormanlarının bazı
yeti¸sme ortamı özellikleri. Süleyman Demirel Üniversitesi Orman Fakültesi Derg. Seri A 2008,1, 48–60.
57.
Özel, H.; Ertekin, M. Ecological conditions in the natural group regeneration areas of oriental beech (Fagus orientalis Lipsky.) in
Bartın and Devrek district. Bartın Orman Fakültesi Derg. 2010,12, 47–64.
58.
Yılmaz, M. Effects of soil and environmental factors on the site productivity of pure Oriental beech forests in Akku¸s Region of
Turkey. Eurasian J. For. Sci. 2019,7, 107–120.
59.
Yilmaz, M.; Usta, A.; Ozturk, I. Relationships between site productivity and environmental variables of Oriental beech forests in
northeastern Turkey. Austrian J. For. Sci. 2022,139, 239–262.
60.
Bolat, I.; Kara, O.; Tunay, M. Seasonal Changes of Microbial Biomass Carbon, Nitrogen, and Phosphorus in Soil Under an Oriental
Beech Stand. Forestist 2022,72, 259–265. [CrossRef]
61.
Azaryan, M.; Vajari, K.A.; Amanzadeh, B. Soil properties and fine root morphological traits in relation to soil particle-size fractions
in a broad-leaved beech (Fagus orientalis Lipsky) forest. Acta Oecologica 2022,117, 103852.
62.
Akbas, M.; Tufekcioglu, A. Contribution of the Root Component to Soil Respiration in Oriental Beech Stands in Artvin, Turkey.
For. Sci. 2022,68, 399–409.
63.
Amolikondori, A.; Vajari, K.A.; Feizian, M.; Di Iorio, A. Influences of forest gaps on soil physico-chemical and biological properties
in an oriental beech (Fagus orientalis L.) stand of Hyrcanian forest, north of Iran. Iforest-Biogeosci. For.
2020
,13, 124–129. [CrossRef]
64.
Guner, S.T. Relatıonshıps between Sıte Index and Ecologıcal Varıables of Orıental Beech Forest in the Marmara Regıon of Turkey.
Fresenius Environ. Bull. 2021,30, 6920–6927.
65.
Mohammadi, M.F.; Jalali, S.G.; Kooch, Y.; Said-Pullicino, D. The effect of landform on soil microbial activity and biomass in a
Hyrcanian oriental beech stand. Catena 2017,149, 309–317. [CrossRef]
66.
Kooch, Y.; Parsapour, M.K.; Egli, M.; Moghimian, N. Forest floor and soil properties in different development stages of Oriental
beech forests. Appl. Soil Ecol. 2021,161, 103823.
67.
Bakhshandeh-Navroud, B.; Vajari, K.A.; Pilehvar, B.; Kooch, Y. The interactions between tree-herb layer diversity and soil
properties in the oriental beech (Fagus orientalis Lipsky) stands in Hyrcanian forest. Environ. Monit. Assess.
2018
,190, 425.
[CrossRef]
68.
Bailey, R.G. Ecosystem Geography: From Ecoregions to Sites; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2009.
69. Spurr, S.H.; Barnes, B.V. Forest Ecology; Wiley: Hoboken, NJ, USA, 1991.
70. Atalay, ˙
I. Türkiye’nin Ekolojik Bölgeleri (Ecoregions of Turkey); TC Orman Bakanlı˘gı Yayını: Ankara, Turkey, 2002.
Sustainability 2023,15, 14580 19 of 20
71.
Afif-Khouri, E.; Alvarez-Alvarez, P.; Fernandez-Lopez, M.J.; Oliveira-Prendes, J.A.; Camara-Obregon, A. Influence of climate,
edaphic factors and tree nutrition on site index of chestnut coppice stands in north-west Spain. Forestry
2011
,84, 385–396.
[CrossRef]
72.
Güner, ¸S.T.; Comez, A.; Ozkan, K.; Karatas, R.; Celik, N. Modelling the productivity of Anatolian black pine plantations in Turkey.
Istanb. Univ. Fac. For. 2016,66, 159–172.
73.
Klinka, K.; Wang, Q.; Carter, R.E.; Chen, H.Y.H. Height growth-elevation relationships in subalpine forests of interior British
Columbia. For. Chron. 1996,72, 193–198. [CrossRef]
74.
Jiao, K.W.; Gao, J.B.; Liu, Z.H. Precipitation Drives the NDVI Distribution on the Tibetan Plateau While High Warming Rates May
Intensify Its Ecological Droughts. Remote Sens. 2021,13, 1305. [CrossRef]
75.
Shi, L.; Fan, H.J.; Yang, L.Y.; Jiang, Y.T.; Sun, Z.H.; Zhang, Y.L. NDVI-based spatial and temporal vegetation trends and their
response to precipitation and temperature changes in the Mu Us Desert from 2000 to 2019. Water Sci. Technol.
2023
,88, 430–442.
[CrossRef] [PubMed]
76.
Bollandsas, O.M.; Orka, H.O.; Dalponte, M.; Gobakken, T.; Naesset, E. Modelling Site Index in Forest Stands Using Airborne
Hyperspectral Imagery and Bi-Temporal Laser Scanner Data. Remote Sens. 2019,11, 1020. [CrossRef]
77.
Wang, J.; Rich, P.M.; Price, K.P.; Kettle, W.D. Relations between NDVI and tree productivity in the central Great Plains. Int. J.
Remote Sens. 2004,25, 3127–3138. [CrossRef]
78.
Erasmi, S.; Klinge, M.; Dulamsuren, C.; Schneider, F.; Hauck, M. Modelling the productivity of Siberian larch forests from Landsat
NDVI time series in fragmented forest stands of the Mongolian forest-steppe. Environ. Monit. Assess. 2021,193, 200. [PubMed]
79.
Main-Knorn, M.; Moisen, G.G.; Healey, S.P.; Keeton, W.S.; Freeman, E.A.; Hostert, P. Evaluating the Remote Sensing and
Inventory-Based Estimation of Biomass in the Western Carpathians. Remote Sens. 2011,3, 1427–1446. [CrossRef]
80.
Fang, K.Y.; Frank, D.; Zhao, Y.; Zhou, F.F.; Seppa, H. Moisture stress of a hydrological year on tree growth in the Tibetan Plateau
and surroundings. Environ. Res. Lett. 2015,10, 034010. [CrossRef]
81.
Hynes, A.; Hamann, A. Moisture deficits limit growth of white spruce in the west-central boreal forest of North America. For.
Ecol. Manag. 2020,461, 117944. [CrossRef]
82.
Menendez-Miguelez, M.; Alvarez-Alvarez, P.; Majada, J.; Canga, E. Effects of soil nutrients and environmental factors on site
productivity in Castanea sativa Mill. coppice stands in NW Spain. New For. 2015,46, 217–233.
83.
Zhou, Y.; Lei, Z.; Zhou, F.; Han, Y.; Yu, D.; Zhang, Y. Impact of climate factors on height growth of Pinus sylvestris var. mongolica.
PLoS ONE 2019,14, e0213509.
84.
Wilde, S.A. Forest soils, their properties and relation to silviculture. In Forest Soils, Their Properties and Relation to Silviculture; S. A.
Wilde. Ronald Press: New York, NY, USA, 1958.
85. Fitzpatrick, E.A. An Introduction to Soil Science; Longman Singapore Publishers Pte Limited: Singapore, 1995.
86. Weil, R.R.; Brady, N.C. The Nature and Properties of Soils; Pearson: London, UK, 2016.
87.
Aertsen, W.; Kint, V.; De Vos, B.; Deckers, J.; Van Orshoven, J.; Muys, B. Predicting forest site productivity in temperate lowland
from forest floor, soil and litterfall characteristics using boosted regression trees. Plant Soil 2012,354, 157–172.
88.
Subedi, S.; Fox, T.R. Predicting Loblolly Pine Site Index from Soil Properties Using Partial Least-Squares Regression. For. Sci.
2016,62, 449–456. [CrossRef]
89. Aydın, M.; Kılıç, ¸S. Toprak Bilimi; Nobel Yayınları: Çankaya, Turkey, 2020.
90. Binkley, D.; Fisher, R. Ecology and Management of Forest Soils; John Wiley & Sons: Hoboken, NJ, USA, 2012.
91. Irmak, A. Orman Ekolojisi; Ta¸s Matbaası: Istanbul, Turkey, 1970; p. 367.
92.
Berberoglu, S.; Evrendilek, F.; Ozkan, C.; Donmez, C. Modeling forest productivity using envisat MERIS data. Sensors
2007
,7,
2115–2127. [CrossRef] [PubMed]
93.
Gulsoy, S.; Suel, H.; Celik, H.; Ozdemir, S.; Ozkan, K. Modeling Site Productivity of Anatolian Black Pine Stands in Response to
Site Factors in Buldan District, Turkey. Pak. J. Bot. 2014,46, 213–220.
94.
Li, A.M.; Huang, C.Q.; Sun, G.Q.; Shi, H.; Toney, C.; Zhu, Z.L.; Rollins, M.G.; Goward, S.N.; Masek, J.G. Modeling the height of
young forests regenerating from recent disturbances in Mississippi using Landsat and ICESat data. Remote Sens. Environ.
2011
,
115, 1837–1849.
95.
Yue, C.F.; Kahle, H.P.; von Wilpert, K.; Kohnle, U. A dynamic environment-sensitive site index model for the prediction of site
productivity potential under climate change. Ecol. Model. 2016,337, 48–62. [CrossRef]
96.
Park, H.J.; Park, S.I.; Kim, H.S.; Lee, E.S.; Kim, H.J.; Lee, S.H. Selection of geographical factors using the random forest analysis
method for developing the site index equation of Pinus densiflora stands in Republic of Korea. For. Sci. Technol. 2019,15, 19–23.
97.
Dos Reis, A.A.; Franklin, S.E.; Acerbi, F.W.; Ferraz, A.C.; de Mello, J.M. Classification of Eucalyptus plantation Site Index (SI) and
Mean Annual Increment (MAI) prediction using DEM-based geomorphometric and climatic variables in Brazil. Geocarto Int.
2022
,
37, 1256–1273. [CrossRef]
98.
Scolforo, H.F.; McTague, J.P.; Burkhart, H.; Roise, J.; Alvares, C.A.; Stape, J.L. Site index estimation for clonal eucalypt plantations
in Brazil: A modeling approach refined by environmental variables. For. Ecol. Manag. 2020,466, 118079. [CrossRef]
99.
Li, X.Y.; Duan, A.G.; Zhang, J.G. Site index for Chinese fir plantations varies with climatic and soil factors in southern China. J.
For. Res. 2022,33, 1765–1780.
100.
McKenney, D.W.; Pedlar, J.H. Spatial models of site index based on climate and soil properties for two boreal tree species in
Ontario, Canada. For. Ecol. Manag. 2003,175, 497–507. [CrossRef]
Sustainability 2023,15, 14580 20 of 20
101.
Jiang, H.Q.; Radtke, P.J.; Weiskittel, A.R.; Coulston, J.W.; Guertin, P.J. Climate- and soil-based models of site productivity in
eastern US tree species. Can. J. For. Res. 2015,45, 325–342.
102.
Restrepo, H.I.; Orrego, S.A.; Salazar-Uribe, J.C.; Bullock, B.P.; Montes, C.R. Using biophysical variables and stand density to
estimate growth and yield of Pinus patula in Antioquia, Colombia. Open J. For. 2019,9, 195–213.
103.
Swenson, J.J.; Waring, R.H.; Fan, W.; Coops, N. Predicting site index with a physiologically based growth model across Oregon,
USA. Can. J. For. Res. 2005,35, 1697–1707. [CrossRef]
Disclaimer/Publisher’s Note:
The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
