Site Index Curves for Abies borisii-regis Mattf. and Fagus sylvatica L. Mixed Stands in Central Greece

Abstract

Despite their productivity, fir and beech forests in Greece lack site index curves. In this work, site index curves for Fagus sylvatica and Abies borisii-regis in central Greece were developed. Thirty plots were randomly established in the mixed stands of F. sylvatica–A. borisii-regis in Aspropotamos, central Greece, and two dominant trees, one from each species, were randomly selected and cut. Height–age measurements were collected through stem analysis. These data were used to develop site index curves for each species. The site index curves illustrate a growth rate difference between the two species, specifically in the worst sites, with fir growing faster than beech. Additionally, as trees age, the growth difference between the two species in the best sites decreases. Based on these results, F. sylvatica is found to be more site-sensitive than A. borisii-regis. In the new adverse conditions of global warming, an increased knowledge of the site sensitivity of the two species will help to develop appropriate treatments for the conservation of the studied mixed stands, or at least to minimize negative impacts.

1. Introduction

Good forest management and yield forecasting require an efficient assessment of site productivity. Site index modeling is the most popular method for assessing site productivity. In this context, we assessed the productivity of two of the most productive species in Greece, Abies borisii-regis Mattf. and Fagus sylvatica L. Fir forests comprise 8.34% of the woodland in Greece [1], much of which is A. borisii-regis (Bulgarian fir), which is endemic to the Balkans [2]. This species is considered to be a hybrid of A. alba Mill. and A. cephalonica Loud [3]. A. borisii-regis is very shade-tolerant and can survive for many years in dark conditions, growing extremely slowly [4]. In Greece, Abies spp. produce 31.23% of the merchantable volume of wood [1], mainly A. borisii-regis.

Regarding F. sylvatica, 5.17% of the woodland in Greece comprises European beech forests, which supply 20.05% of the merchantable wood volume [1]. It is also a shade-tolerant species [5]. In mountainous regions, stands of F. sylvatica are found in the most productive sites [2]. F. sylvatica and A. borisii-regis often form mixed forests in Greece, and based on their expansion and the wood they produce, they are considered to be important productive species in the country.

In order to develop good forest management practices in mixed stands of F. sylvatica–A. borisii-regis, as well as yield forecasting for the aforementioned species, an efficient assessment of the site productivity is needed. Clutter et al. (1983) [6] classified the techniques for quantifying site quality into direct and indirect. Direct methods use historical data, while indirect methods develop statistical relationships among forest biometrical variables (site index modeling). Direct methods, most inevitably, give much better site quality assessments compared with the indirect methods; however, historical data are unavailable for the majority of forest species. Site index modeling is the most popular process for assessing site productivity [7,8]. The site index is easily defined as the mean height of the dominant trees in a stand, at a base (reference) age [9]. Tree height is used because, in theory, it is sensitive to site productivity (i.e., volume), but is not sensitive to site density [10].

Data for the development of site index curves can be collected from temporary or permanent plots. Temporary plots are inexpensive and are the fastest source of data; however, sampling and data collection are based on the assumption that site indices are represented in all age classes [6,10]. This is rarely achieved in practice. Permanent plots provide the finest data for site indices, but are costly compared with temporary plots.

Over the last 10–15 years, different studies have been carried out to assess forest stand characteristics such as tree heights and wood volume, utilizing different aerial laser profiling and scanning techniques [11,12,13,14,15,16]. The majority of researches have concentrated on mature and old-growth forest, although tree attributes have been studied even in very young stands [17]. Researchers have concentrated on in-depth examinations of the correlations between laser-derived metrics and tree and canopy features, providing sufficient information for the effective application of laser data in large-scale forest surveys [13,15,17]. A strategy for a large-scale forest inventory based on data from scanning laser devices has been described and tested in many studies, showing that laser data may be used to calculate mean tree height [12,13,14,17,18,19], dominant height [17], stand basal area [13], and stand volume [13,15], with adequate accuracy compared to traditional approaches. Additionally, laser data may be used to determine mean stem diameter and stem number [11].

Previous research has demonstrated that trees of different ages produce laser canopy height distributions that have a wide variety of distinct shapes [11]. There is a potential for species-specific variation in the relationships between stand attributes and laser data metrics [14,20], while crown morphology has an effect on laser metrics [21]. It is possible that well-known parameters for stratification, like age class and site quality, could be useful for achieving effective and efficient inventory stratification [22].

When it comes to assessing stand productivity, aerial laser scanning data, despite the fact that they give significant information on stand characteristics, should absolutely be complemented with field measurements. It is important for an area-based method [23,24,25,26] to have a sufficient number of field plots, as well as for those field plots to be of a high quality. An area-based approach involves the development of prediction models between characteristics derived from remotely sensed data and field-measured forest inventory attributes obtained from sample plots [22,27]. When predicting forest inventory attributes such as basal area, stem volume, mean diameter, and mean height over inventory areas for the purposes of wood-procurement planning [28,29,30], airborne laser scanning and aerial imagery are used in conjunction with field-measured sample plots. Next, prediction models are applied at the grid level across the region of interest that was specified. Typically, the size of the field sample plot corresponds to the size of the prediction grid. Attributes of the stand-level forest inventory may then be determined by aggregating the grid-level forecasts made within each stand [27]. According to [31], in order for an area-based approach to combine sample plot data with remote sensing data, forest managers need objective and exact mean and summary information on growing stock at the plot level.

Regarding the significance of the research area’s geographical position in this paper, we emphasize that Aspropotamos is near the southern limits of F. sylvatica expansion in Greece, which is in Grameni Oxia [2]. The analysis of height growth of the two species, apart of a tool for the management of the F. sylvatica–A. borisii-regis stands, will enable forest practice to confront the challenges which will arise in the context of climate change, when even the existence of F. sylvatica or A. borisii-regis in a great part of the area will be doubtful.

Beech and fir species in mixed stands exhibit varied growth patterns, based on the observations of forestry experts in Greece during the development of management plans, with beech being more site-sensitive than fir. To assess the productivity of these species and provide statistical evidence for the aforementioned hypothesis, site index curves were developed for F. sylvatica and A. borisii-regis in Aspropotamos, central Greece, using data from randomly sampled and cut dominant trees in temporary plots established in mixed stands of the two species.

2. Materials and Methods

2.1. Study Area

This research was conducted in mixed stands of F. sylvatica and A. borisii-regis in Aspropotamos, which is in central Greece (coordinates of the mid-point 39°38′28.0793″ N, 21°17′17.7420″ E, Figure 1).

The study area has an elevation from 795 to 2279 m and an area of 20,168 ha. The greatest part of the study area is covered by A. borisii regis stands. Mixed stands of F. sylvatica and A. borisii-regis are scattered in the study area, while F. sylvatica and Pinus nigra stands are also present. The mixed stands of F. sylvatica–A. borisii-regis are unevenly aged with an irregular structure. In the evenly aged groups of these mixed stands, selective thinning (i.e., thinning from above—positive selection) is applied, while the shelterwood method is used in the regeneration process [32].

The mean temperature in the study area is 8.75 °C and the annual precipitation is 874.86 mm [32]. The parent material in the mixed stands is Pindos flysch [32].

2.2. Data Collection

In 2013, within the study area, 30 randomly selected plots, each measuring 0.1 hectares, were established in the mixed stands of F. sylvatica (European beech) and A. borisii-regis (Bulgarian fir). In total, 30 dominant trees of each species were cut. From each cut tree, cross sectional discs were taken at 0.3 m height (stump height, i.e., logging height), at breast height, at 3.3 m height, and every 3 m up to the top. The last disc was cut at a height of 5 cm bole diameter. In all discs, the annual growth rings were counted using RinnTech’s LINTAB system [33]. For the stem analysis, in order to calculate tree height corresponding to each age, the modified Carmean algorithm was used [34].

The data from stem analysis provided evidence of inherent variations in growth potential related to tree age, for seven fir trees and one beech tree. Although growth rate varies inevitably over time, a decreasing growth rate as the initial tree height increases renders these trees unsuitable for analyzing tree growth performance and efficiency. In other words, we should avoid using these trees in site productivity assessment because doing so would lead to an underestimation of the dominant height of young trees, and thus an overestimation of the expected performance and yield of young stands [35]. We excluded these seven fir trees and one beech tree from the development of the site index curves because their height growth deviated from that of the trees that were dominant throughout their life. This exclusion was applied in order to avoid an underestimation of heights through the site index models.

In order to reveal natural groupings within the height–age dataset, we applied a two-step cluster analysis [36], with automatic selections of clusters (groups) corresponding to site qualities. Such groupings could reveal different growth patterns, thus different site qualities, considering that F. sylvatica–A. borisii-regis stands are scattered in the study area and do not belong to uniform areas in terms of site quality. Clustering was performed using the SPSS statistical package [37] and by applying the Schwarz’s Bayesian Criterion (BIC) as the clustering criterion. For the height–age datasets of both species, two groups (i.e., two site qualities, I and II, in order of decreasing productivity) were distinguished, with a silhouette measure of cohesion and separation > 0.5, meaning that the overall clustering quality was good. Then, height–age models were fitted to each dataset of site quality in order to develop polymorphic site index curves, per species.

Summary statistics for the sampled trees are given in

Table 1. Statistical description of the sample.

		A. borisii-regis					F. sylvatica
		Mean	Standard Deviation	Min	Max	N of Trees	Mean	Standard Deviation	Min	Max	N of Trees
Whole sample	Age (years)	99.48	31.62	53.00	160.00	23	116.07	40.33	60.00	177.00	29
	Height (m)	26.26	3.53	20.32	33.27	23	24.81	3.62	17.81	32.42	29
Site I	Age (years)	85.25	22.76	53.00	151.00	16	104.00	39.31	60.00	177.00	19
	Height (m)	25.79	3.61	20.32	33.27	16	24.77	4.06	17.81	32.42	19
Site II	Age (years)	132.00	24.42	104.00	160.00	7	139.00	32.82	90.00	173.00	10
	Height (m)	27.33	3.35	22.59	31.65	7	24.89	2.81	18.73	28.77	10

.

2.3. Fitted Site Index Models

Forty two height–age models were fitted to the data [38]. Of the 42 models, those which had significant regression coefficients, i.e., had 95% confidence intervals not including zero, were further compared. The coefficient of determination (R²), the standard error of the estimate (SEE), and the root of the mean squared error (RMSE) were calculated for the models with significant regression coefficients. These statistics were used to select the best fitted site index model, for each site quality and species.

3. Results

3.1. A. borisii-regis Site Index Curves

Regarding the A. borisii-regis height–age models, those which did not have statistically significant regression coefficients had their comparison criteria marked “Not Applicable” (N/A). For the remaining models, the values for the comparison criteria (R², SEE, RMSE) are given in

Table 2. Statistical comparison of the A. borisii-regis height–age models.

Model No	Model	Site Quality I			Site Quality II
		R2 (Optimum Value: 1)	SEE (Optimum Value: Min)	RMSE (Optimum Value: Min)	R2 (Optimum Value: 1)	SEE (Optimum Value: Min)	RMSE (Optimum Value: Min)
1	H = b0 + b1t + b2t2 + b3t3	0.92	2.50	2.68	0.95	2.02	2.34
2	H = b0 + b1t + b2t2	0.92	2.51	1.50	0.95	2.07	1.01
3	H = b0 + b1t + b2t3	0.85	3.39	1.98	0.92	2.65	1.27
4	H = b1t + b2t2	0.87	3.15	20.52	0.92	2.65	30.43
5	H = b0 + b1$\sqrt{t}$ + b2t	0.89	2.91	7.74	0.93	2.49	7.39
6	H = b0 + b1$\sqrt{t}$ + b2t + b3t2	0.89	2.91	8.99	0.92	2.51	8.32
7	H = b0 + b1$\sqrt[3]{t}$ + b2t	0.88	2.97	1.73	0.93	2.44	1.17
8	H = b0 + b1$\sqrt[3]{t}$ + b2$\sqrt{t}$ + b3t	0.87	3.11	3.37	0.90	2.83	4.57
9	H = b0 + b1t + b2$\frac{1}{t}$	0.85	3.34	1.95	0.92	2.64	1.27
10	H = b0 + b1$\frac{1}{t}$ + b2$\frac{1}{t^2}$	0.48	6.23	3.63	0.42	6.93	3.33
11	H = b0 + b1$\frac{1}{t}$	0.22	7.64	4.46	0.18	8.24	3.96
12	H = b0 + b1$\frac{1}{t^2}$	0.07	8.37	4.89	0.04	8.91	4.28
13	H = b0 + b1t + b2lnt	0.87	3.11	1.81	0.92	2.53	1.21
14	H = b0 + b1lnt	0.77	4.17	2.43	0.78	4.25	2.04
15	H = b0 + b1lnt + b2(lnt)2	0.77	4.17	10.55	0.78	4.25	12.47
16	lnH = b0 + b1lnt	0.83	3.56	2.27	0.89	3.03	1.57
17	lnH = b0 + b1lnt + b2(lnt)2	0.84	3.42	2.32	0.93	2.45	4.93
18	lnH = b0 + b1lnt + b2(lnt)3	0.83	3.57	2.32	0.91	2.67	3.36
19	lnH = b0 + b1$\frac{1}{t}$	0.58	5.63	4.37	0.55	6.11	4.11
20	lnH = b0 + b1$\frac{1}{t}$ + b2$\frac{1}{t^2}$	0.81	3.74	2.97	0.83	3.81	2.88
21	lnH = b0 + b1$\sqrt{t}$ + b2t	0.89	2.92	7.97	0.93	2.42	7.34
22	lnH = b0 + b1$\sqrt[3]{t}$ + b2t	0.92	2.42	1.46	0.94	2.18	1.09
23	lnH = b0 + b1$\frac{1}{\sqrt{t}}$	0.87	3.08	2.76	0.91	2.77	2.58
24	lnH = b0 + b1t + b2$\frac{1}{\sqrt{t}}$	0.64	5.17	3.98	0.74	4.65	2.95
25	lnH = b1lnt + b2(lnt)2	0.88	3.05	7.20	0.83	3.78	7.59
26	lnH = b0 + b1$\sqrt{t}$ + b2$\frac{1}{t}$ + b3$\frac{1}{t^2}$	0.78	4.09	6.05	0.78	4.32	6.32
27	$H=b_1\left[1-e^{\left(-b_{2}t\right)}\right]$	0.90	2.69	1.63	0.94	2.28	1.16
28	$H=b_1{\left[1-e^{\left(-b_{2}t\right)}\right]}^{b_3}$	N/A	N/A	N/A	N/A	N/A	N/A
29	$H=b_1\left[1-e^{\left(-b_2{t}^{b_3}\right)}\right]$	0.91	2.55	2.17	N/A	N/A	N/A
30	$H=\frac{t^2}{b_0+b_{1}t+b_2{t}^2}$	0.92	2.38	1.40	0.96	1.80	0.87
31	$H=\frac{t}{b_0+b_{1}t+b_2{t}^2}$	0.00	8.66	214.47	0.00	9.10	285.04
32	$H=\frac{t^2}{{\left(b_0+b_{1}t\right)}^2}$	0.92	2.52	1.48	0.95	2.00	0.97
33	$H=e^{\left(b_0+b_1\frac{1}{t}\right)}$	0.92	2.45	1.44	0.96	1.81	0.87
34	$H=\frac{b_1}{1+b_2{e}^{-b_{3}t}}$	N/A	N/A	N/A	N/A	N/A	N/A
35	$H=b_1{e}^{\left(\frac{-b_2}{t}\right)}$	0.92	2.45	1.44	0.96	1.81	0.87
36	$H=b_1\left[1-b_2{e}^{\left(-b_{3}t\right)}\right]$	0.91	2.60	1.52	0.95	2.13	1.05
37	$H=b_0+b_1{e}^{\left(b_{2}t\right)}$	N/A	N/A	N/A	N/A	N/A	N/A
38	$H=b_1{e}^{\left(\frac{-b_2}{t^{b_3}}\right)}$	0.03	8.54	4.98	0.02	9.03	4.33
39	$H=b_1{e}^{\left(-b_{2}t\right)}$	0.67	5.01	3.00	0.79	4.19	2.09
40	$H=b_1\left[1-e^{{\left(\frac{-b_2}{t}\right)}^{b_3}}\right]$	0.30	7.24	9.44	0.25	7.88	8.21
41	$H=\frac{b_1}{1+b_2{e}^{-b_3{t}^{b_4}}}$	N/A	N/A	N/A	N/A	N/A	N/A
42	$H=b_1{e}^{\left(-b_2{e}^{-b_{3}t}\right)}$	N/A	N/A	N/A	N/A	N/A	N/A

H: height (m); t: age (years); b_i: regression coefficients.

. The optimum value for each criterion (R², SEE, and RMSE) is highlighted.

Based on Table 2, the selected regression model for the development of the polymorphic site index curves for A. Borisii-regis is the model no 30:

$$H=\frac{t^2}{72.771-0.181t+0.033t^2} (\mathrm{site} \mathrm{quality} \mathrm{I})$$

$$H=\frac{t^2}{183.539-1.035t+0.036t^2} (\mathrm{site} \mathrm{quality} \mathrm{II})$$

3.2. F. sylvatica Site Index Curves

Regarding the F. sylvatica height–age models, those which did not have statistically significant regression coefficients had their comparison criteria marked “Not Applicable” (N/A). For the remaining models, the values for the comparison criteria (R², SEE, RMSE) are given in

Table 3. Statistical comparison of the F. sylvatica height–age models.

Model No	Model	Site Quality I			Site Quality II
		R2 (Optimum Value: 1)	SEE (Optimum Value: Min)	RMSE (Optimum Value: Min)	R2 (Optimum Value: 1)	SEE (Optimum Value: Min)	RMSE (Optimum Value: Min)
1	H = b0 + b1t + b2t2 + b3t3	0.95	1.97	2.57	0.92	2.24	2.89
2	H = b0 + b1t + b2t2	0.95	2.02	2.03	0.89	2.70	2.52
3	H = b0 + b1t + b2t3	0.88	3.05	2.14	0.90	2.56	1.51
4	H = b1t + b2t2	0.91	2.71	33.97	0.90	2.56	42.07
5	H = b0 + b1$\sqrt{t}$ + b2t	0.93	2.27	9.36	0.91	2.41	7.89
6	H = b0 + b1$\sqrt{t}$ + b2t + b3t2	0.93	2.26	11.08	0.91	2.43	8.83
7	H = b0 + b1$\sqrt[3]{t}$ + b2t	0.93	2.35	1.65	0.91	2.37	1.40
8	H = b0 + b1$\sqrt[3]{t}$ + b2$\sqrt{t}$ + b3t	0.92	2.46	3.07	0.89	2.68	4.74
9	H = b0 + b1t + b2$\frac{1}{t}$	0.89	2.95	2.07	0.90	2.55	1.50
10	H = b0 + b1$\frac{1}{t}$ + b2$\frac{1}{t^2}$	0.48	6.35	4.46	0.41	6.19	3.64
11	H = b0 + b1$\frac{1}{t}$	0.22	7.79	5.47	0.18	7.31	4.31
12	H = b0 + b1$\frac{1}{t^2}$	0.06	8.54	6.00	0.04	7.89	4.65
13	H = b0 + b1t + b2lnt	0.92	2.55	1.79	0.91	2.44	1.43
14	H = b0 + b1lnt	0.82	3.71	2.61	0.78	3.83	2.25
15	H = b0 + b1lnt + b2(lnt)2	0.82	3.71	13.16	0.78	3.83	13.18
16	lnH = b0 + b1lnt	0.88	3.09	2.64	0.89	2.67	1.67
17	lnH = b0 + b1lnt + b2(lnt)2	0.86	3.30	5.17	0.91	2.47	3.41
18	lnH = b0 + b1lnt + b2(lnt)3	0.85	3.40	6.85	0.90	2.58	2.00
19	lnH = b0 + b1$\frac{1}{t}$	0.59	5.65	5.32	0.53	5.55	4.35
20	lnH = b0 + b1$\frac{1}{t}$ + b2$\frac{1}{t^2}$	0.84	3.57	3.54	0.78	3.76	3.12
21	lnH = b0 + b1$\sqrt{t}$ + b2t	0.93	2.30	9.77	0.91	2.37	7.87
22	lnH = b0 + b1$\sqrt[3]{t}$ + b2t	0.95	1.88	1.33	0.93	2.19	1.32
23	lnH = b0 + b1$\frac{1}{\sqrt{t}}$	0.92	2.53	3.04	0.88	2.82	2.73
24	lnH = b0 + b1t + b2$\frac{1}{\sqrt{t}}$	0.80	3.90	3.06	0.78	3.76	2.94
25	lnH = b1lnt + b2(lnt)2	0.93	2.34	8.49	0.85	3.15	8.00
26	lnH = b0 + b1$\sqrt{t}$ + b2$\frac{1}{t}$ + b3$\frac{1}{t^2}$	0.80	3.92	7.22	0.74	4.13	6.56
27	$H=b_1\left[1-e^{\left(-b_{2}t\right)}\right]$	0.95	2.03	1.49	0.92	2.28	1.58
28	$H=b_1{\left[1-e^{\left(-b_{2}t\right)}\right]}^{b_3}$	N/A	N/A	N/A	N/A	N/A	N/A
29	$H=b_1\left[1-e^{\left(-b_2{t}^{b_3}\right)}\right]$	0.96	1.76	1.27	0.93	2.18	1.54
30	$H=\frac{t^2}{b_0+b_{1}t+b_2{t}^2}$	0.96	1.75	1.24	0.93	2.11	1.24
31	$H=\frac{t}{b_0+b_{1}t+b_2{t}^2}$	0.58	5.70	21.83	0.64	4.81	15.58
32	$H=\frac{t^2}{{\left(b_0+b_{1}t\right)}^2}$	0.96	1.85	1.30	0.93	2.14	1.26
33	$H=e^{\left(b_0+b_1\frac{1}{t}\right)}$	0.96	1.81	1.30	0.93	2.13	1.26
34	$H=\frac{b_1}{1+b_2{e}^{-b_{3}t}}$	N/A	N/A	N/A	N/A	N/A	N/A
35	$H=b_1{e}^{\left(\frac{-b_2}{t}\right)}$	0.96	1.81	1.30	0.93	2.13	1.26
36	$H=b_1\left[1-b_2{e}^{\left(-b_{3}t\right)}\right]$	0.95	1.93	1.38	0.93	2.20	1.37
37	$H=b_0+b_1{e}^{\left(b_{2}t\right)}$	N/A	N/A	N/A	N/A	N/A	N/A
38	$H=b_1{e}^{\left(\frac{-b_2}{t^{b_3}}\right)}$	0.03	8.69	6.11	0.02	8.00	4.71
39	$H=b_1{e}^{\left(-b_{2}t\right)}$	0.74	4.49	3.31	0.79	3.72	2.24
40	$H=b_1\left[1-e^{{\left(\frac{-b_2}{t}\right)}^{b_3}}\right]$	0.30	7.39	11.95	0.25	7.00	9.11
41	$H=\frac{b_1}{1+b_2{e}^{-b_3{t}^{b_4}}}$	N/A	N/A	N/A	N/A	N/A	N/A
42	$H=b_1{e}^{\left(-b_2{e}^{-b_{3}t}\right)}$	N/A	N/A	N/A	N/A	N/A	N/A

H: height (m); t: age (years); b_i: regression coefficients.

. The optimum value for each criterion (R², SEE, and RMSE) is highlighted.

Based on Table 3, the selected regression model for the development of the polymorphic site index curves for F. sylvatica is the model no 30:

$$H=\frac{t^2}{70.977-0.508t+0.029t^2} (\mathrm{site} \mathrm{quality} \mathrm{I})$$

$$H=\frac{t^2}{126.982-1.100t+0.029t^2} (\mathrm{site} \mathrm{quality} \mathrm{II})$$

4. Discussion

F. sylvatica exhibits a greater growth rate difference between the best site quality I and the worst site quality II (different for each species) compared to that of A. borisii-regis (Figure 2). However, this growth rate difference is by far greater in the worst sites (for each species) compared to that in the best sites. The growth difference between the two species in the best site (for each species) decreases as the age of the trees increases. These results lead to the conclusion that F. sylvatica is a more site-sensitive species compared with A. borisii-regis.

Compared to the site index curves created for beech forests in two areas of Serbia, Žagubica and Rudnik, where dominant trees were used [39], the curves of beech in the mixed stands of the present study correspond approximately to the III and IV site classes in Serbia (for site I and site II of the present study, respectively; see Figure 3).

The same pattern is observed with the site index curves created for F. sylvatica stands in two areas of northern Greece (Figure 4), Arnea [40] and Vrondou [41]. In these cases, dominant and codominant trees were used, while the height–age curves were developed for breast height age.

There are no site index curves for A. borisii-regis to be compared with those developed for the mixed stands of Aspropotamos. However, ref. [42] developed a height–age model using the tallest tree in areas of 100 m² for the A. spp. forests in Greece. These forests comprise A. borisii-regis forests. The two curves of fir developed in the present study are deployed close to the curve of site quality II of the A. spp. study, at ages greater than 100 years. The site I curve is deployed above the curve of site quality II (in the A. spp. study), and the site II curve appears slightly below the aforementioned curve (Figure 5).

If the site curves for site I and site II of A. borisii-regis developed from the mixed stands are compared with the site index curves created for the A. cephalonica (one of the parental species of A. borisii-regis) forest in the Mount Taygetos in Greece [43], they appear below the curves for site classes I and II, respectively, of the A. cephalonica study (Figure 6). In the study of [43], dominant and codominant trees were used. If we consider the height–age at breast height curves of [43] for A. cephalonica, the curves of A. borisii-regis in the mixed stands move slightly towards the curves representing the best productivity sites in the A. cephalonica studies (Figure 6).

The range of sites where the studied mixed stands appear is rather narrow. Probably, in better sites than those studied in this research, F. sylvatica (as a site—sensitive species) would grow faster than A. borisii-regis, while in worse (than in the present study) sites the growth superiority of A. borisii-regis would be greater than that observed in the worst sites of this study. However, more research is needed in order to obtain a clearer picture of the two species’ height growth in mixed stands of different areas, as well as of A. borisii-regis in pure stands.

Based on the findings of this study, in the less productive sites of the mixed stand of Aspropotamos, A. borisii-regis has a height growth advantage compared to F. sylvatica. Forest practice in the area, and in areas with analogous ecological conditions, has to adjust the treatments applied to the mixed stands to accommodate this height growth pattern. Based on the specific structural conditions that exist in each stand [44], forest practice must apply the required silvicultural treatments in each case in order to enhance the participation of F. sylvatica in the worst sites of F. sylvatica–A. borisii-regis mixed stands. Generally, the findings of the present study in combination with the volume equations developed for F. sylvatica and A. borisii-regis mixed stands by [45] increase the ability to treat the mixed stands of the two species more efficiently in order to achieve the set management goals. In the new adverse conditions of global warming, an increased knowledge of the site sensitivity of the two species will be valuable for the development of the appropriate treatments for the conservation of the studied mixed stands, or at least for the minimization of negative impacts.

5. Conclusions

The site index curves that were formulated for beech and fir species in mixed stands located in central Greece demonstrated distinct growth patterns. The site index curves illustrated a growth rate difference between the two species, specifically in the worst sites, with fir growing faster than beech. Additionally, as trees aged, the growth difference between the two species in the best sites decreased. Based on these results, F. sylvatica was found to be more site-sensitive than A. borisii-regis. The aforementioned results hold significant potential for guiding forest management practices within the Aspropotamos region. The labor force of Aspropotamos is primarily associated with forestry, agriculture, and tourism due to the region’s dense forested terrain. The primary occupations of the inhabitants are agriculture and farming. The remaining inhabitants of the flatland engage in agricultural pursuits, including dynamic cultivation practices, agricultural enterprises, and greenhouse operations [46]. On a larger scale, given the current adverse conditions resulting from global warming, an improved comprehension of the site sensitivity of the two species is imperative in order to develop suitable conservation measures for the studied mixed stands, or, at the very least, to mitigate any negative impacts.