4.1. Air-Soil Temperature and LAI, LT, CO, GF Correlations (Overall Course and Leaf Phenological Stages)
According to the overall results of this study, even though alternating dependent upon the altitudinal gradients, the overall intra-annual course of the LAI data has almost followed the similar pattern with the intra-annual air and soil temperature data (Fig. 4). As a matter of fact, there were significant positive correlation (r = 0.894; P = 0.000) between the intra-annual air temperatures and the average LAI data for all the altitudinal gradients (Table 1). Besides, the positive significant correlations were also valid between the intra-annual temperatures of the soils; particularly at the − 5 cm (r = 0.930; P = 0.000), -10 cm (r = 0.922; P = 0.000), -20 cm (r = 0.913; P = 0.000) depths with the intra-annual average LAI data (Table 1). Therefore, these significant correlations have confirmed and supported the direct positive influence of the air-soil temperature on the weekly, monthly, seasonal and inter-seasonal trend of the LAI data. Hence, relatively recent scientific studies have confirmed the influence of both the air and soil temperatures on the seasonal and inter-seasonal dimensional leaf development of the same or similar forest trees (Urban et al. 2014), and on their associated LAI, suggesting similarly high and significant correlations amongst these parameters (Öztürk et al. 2015). On the other hand, again alternating dependent upon the altitudinal gradients, the average LT, CO, GF data have all shown negative, besides significant correlations (P = 0.000) with both the air (r=-0.935, -0.935, -0.887) respectively, and upper soil temperatures (r=-0.96 for LT and CO, r = 0.92 for GF; down to first − 20 cm depths) respectively (Table 1). Thereby, these significant correlations have also confirmed and supported the direct negative influence of the air-soil temperature on the weekly, monthly, seasonal, and inter-seasonal trends of the LT, CO, GF data. The scientific literature has also predominantly suggested this situation, confirming the influence of the air-soil temperature on the seasonal and inter-seasonal alterations of these vegetation parameters. Furthermore, the negative correlation between the vegetation parameter CO and air-soil temperature is at so high level that CO estimations can be used for the prediction and explanation of in-situ measured air and soil temperatures as well as predicting and explaining some other ecological parameters within the different oak-hornbeam temperate forest types of central Europe (Hederová et al. 2023). Therefore, any shifting in air-soil temperature trend which could be a consequence of forest canopy closure change, in return will most probably affect that forest canopy gap and openness, and hence will presumably have direct impact on the associated light penetration through that forest canopy as also indicated in a relatively recent study by Kašpar et al. (2021). Although an increasing air-soil temperature trend directly cannot be pronounced for the spring, summer, and autumn seasons of the last 10 years (2012–2021; TSMS, 2022), an apparent increasing trend for the winter air-soil temperature stands out (Fig. 5), suggesting possible silent warming in winter and associated unexpected earlier budburst dates for the tree leaves. As a matter of fact, earlier budburst dates may either be beneficial in terms of water and soil conservation or be alarming from the point of soil moisture and nutrient depletion within the overall reservoir landscape.
Indeed, depending either on the hemispherical photograph images or on the seasonal course of the LAI, LT, CO and GF data acquired from their analyses, the leaf budburst and emergence of the first foliage has occurred during the second half of the March; principally the late March, when the mean air and soil (-10 cm) temperatures were 6.5°C and 6.9°C respectively (Fig. 4). During this leaf budburst and emergence stage the average LAI increased slightly with 0.38 m2 m− 2 (from 0.51 m2 m− 2 to 0.89 m2 m− 2) whereas the average percentages of LT and CO decreased slightly with about 3% (dropping only from 64–65% to 61–62% respectively) and the average percentage of GF descended with approximately 5% (from 19–14%) (Fig. 4). On the other hand, in their study for the forest habitats in a national park at the north-western Greece, which includes deciduous oak and beech species besides the coniferous trees, Stagakis et al. (2022) pointed out the period from mid-April to early May for their leaf budburst and first emergence. Considering the altitudinal differences between our study site (~ 200 m asl.) and their study site (ranging between 400 m asl. and 2637 m asl.), average summer and annual air temperature differences (approximately 4.5°C for both) between these two sites could have been the main cause for such one-month delay of leaf budburst and emergence dates. Furthermore, in their study within a floodplain forest of south-eastern Czechia at 150 m asl., which is close to the altitude of our study site (200 m asl.), Nezval et al. (2020) indicated almost the same periods and nearly dates for the leaf budburst and first emergence of the European hornbeams and then English oaks. They also verified almost the same air temperature values (5°C to 8°C) for those dates of leaf budburst and first emergence, which coincide with the results of our study together with the average altitude. However, average air and soil (-10 cm) temperatures were 7.4°C and 9.2°C for the second halves of the March along the last 10 years (2012–2021; TSMS 2022, Fig. 5), confirming that the late March could have been the period of leaf budburst and first emergence within the sample mixed deciduous stand of the reservoir landscape, at least for those last 10 years.
The leaf development stage that involves the unfolding, expansion and numerical increase of the deciduous tree leaves, and that has occurred during the two months period between the early April and early June, led to the average LAI exactly quadrupling (from 0.89 m2 m− 2 to 3.56 m2 m− 2), whereas leading the average percentages of LT and CO to drop more than one-fifth (from 61–62–12%), average percentage of GF to drop by a factor of fourteen (from 14–1%) (Fig. 4). During this stage, these average LAI values from the hemispherical photographs complied with the average LAI values from the LAI-2200C device (Fig. 4). In relation to this, during the same two months period of leaf development stage, the mean air and soil temperatures (-10 cm) were 13.7°C and 13.6°C respectively, which were very close to the average annual air and soil temperatures (both 13.3°C), and were around two-fold of the lead budburst and emergence stage (6.5°C for air and 6.9°C for − 10 cm soil) (Fig. 4). Correspondently, in their study site, being mature, mixed temperate forest of Switzerland, which involve diverse tree species including also the oaks, hornbeams and beeches such as our study site, Zahnd et al. (2023), broadly determined early May as the beginning of the leaf development stage for the same year (2021) of their study. According to their study, during the spring of the year 2021, the mean air temperature in their study site (550 m asl.) was 7.7°C, which was about 3.6°C lower than the mean spring temperature (~ 11.3°C for the year 2021) of our study site; the difference that corresponds to the lapse rate (about 1.0°C/100 m) compared to the altitudinal difference with our study site (~ 200 m asl.). Nonetheless, confirming this correspondence, the mean annual air temperature within their study site was 9.6°C, which reveals almost the same difference with 3.7°C compared to the mean annual air temperature (~ 13.3°C) of our study site, revealing also the one-month delay for the beginning and duration period of the leaf development stage. On the other side, the long-term average air and soil (-10 cm) temperatures for the two months period of leaf development stage (April and May) were 13.3°C and 15.6°C respectively along the last 10 years (2012 and 2021; TSMS 2022, Fig. 5), suggesting both the air-soil temperature and the duration period of leaf development had not changed much along that last decade. Macek et al. (2019) also remarked the jointly elevation and topography-driven maximum temperature for the best prediction of the plant community composition; particularly the understorey plant species.
The stationary leaf stage, principally when both the size and number of the tree leaves gain their maximum level and do almost not change, and that has been achieved during the early June and most probably lasted two and half months until the mid-August, led the average LAI fluctuate only in the relatively narrow range between 3.43 m2 m− 2 and 3.60 m2 m− 2 (Fig. 4). These values again complied with the average LAI values from the LAI-2200C device (Fig. 4). However, during this stationary stage, the average percentages of LT and CO were both within the relatively narrow ranges between 9% and 12% approximately whereas the average percentage of GF was within again the relatively narrow range between 1% and 2% approximately. Hence, during the same two and half months period of stationary leaf stage, the mean air and (-10 cm) soil temperatures were 22.5°C and 22.6°C respectively, which were very close to the average whole summer air and soil temperatures (22.6°C and 22.9°C respectively; Fig. 4). In their study for a temperate deciduous forest composed of mainly sessile oaks and European hornbeams such as our study, Soudani et al. (2021) monitored their leaf phenology at the close altitude (103 m asl.) in the western France where both the mean summer and annual air temperatures have been quite consistent with those in our study site. They determined that the mean LAI had achieved its’ maximum values during the early June and, had nearly lasted until the end of summer (late-August) almost the same duration defined in our study for the stationary leaf stage (Fig. 4).
On the other hand, the deciduous leaf senescence stage, that involves the period from the end of the stationary leaf stage to the leaf discoloration, to the leaf fading, to the leaf fall, and until the leafless stage, and that has approximately lasted during the three and half months period from the mid-August and until the end of the November, led the average LAI drop from 3.43 m2 m− 2 down to 0.85 m2 m− 2 (Fig. 4). These values also to some extent complied with the average LAI values from the LAI-2200C device (Fig. 4). In fact, depending on the meta-analysis of autumn phenology studies, Gill et al. (2015), referred to almost the same period such as our study for the autumn senescence of the deciduous trees within the closer latitudes of the northern hemisphere. Thus, during this leaf senescence stage, conversely the average percentage of LT climbed from 9% up to 65% and similarly average percentage of CO ascended from 10% up to 66%, whereas the average percentage of GF increased from 1% up to 12% approximately (Fig. 4). Nonetheless, during this three and half months of senescence stage, the mean air and soil (-10 cm) temperatures were 14.8°C and 15.9°C respectively, which were close to the long-term average air and soil (-10 cm) temperatures (14.8°C and 17.6°C) for the three and half months period of leaf senescence stage (from mid-August to the end of November) along the last 10 years (2012 and 2021; TSMS, 2022, Fig. 5), suggesting both the air-soil temperature and the duration period of leaf senescence had not changed much along that last decade. However, in their study for the European beeches and pedunculate oaks in Europe, which are two of the dominant temperate tree species in Europe, Chen et al. (2019) suggested the late-September and early-October for the senescence of the European beeches and the pedunculate oaks, and indicated that their senescence had more or less been delayed due to the global warming during the long term period between 1951 and 2013, particularly for the last half of that period.
The leafless stage, when the deciduous trees are almost bare without leaves or with their remaining last few dry leaves palely hanging on their branches, has been reached during the early December and lasted three months along the whole winter until the end of February. Presumably, this leafless stage again has been followed by first formation of the next year’s buds and most probably has proceeded during the first half of the next March. Indeed, during the leafless stage, the average LAI declined from 0.85 m2 m− 2 down to 0.50 m2 m− 2, when conversely, the average percentages of LT and CO were approximately both within their high ranges between 65% and 75% whereas the average percentage of GF was within the relatively narrow high range between 12% and 14% approximately. (Fig. 4). As a matter of fact, based on the review of the scientific studies depending upon the ground-based measurements of LAI, Bréda (2003) indicated that the LAI of a European forest, which involve also similar deciduous trees such as our study, had fallen below 1 m2 m− 2 particularly after the late-October and early-November, confirming the validity of approximately one-month delay for the initiation of leafless stage such as previous stages in Europe compared with our study. On the other hand, during that winter (December-2021, January and February-2022) in our study, the mean air and (-10 cm) soil temperatures were 5.0°C and 4.1°C respectively (Fig. 4), which were close to the long-term (10 years between 2012 and 2021; TSMS, 2022, Fig. 5) average winter air and soil (-10 cm) temperatures (4.6°C and 5.2°C) respectively (Fig. 4); despite a silent warming can be seen for the winter air temperature within the reservoir landscape.
4.2. Precipitation and LAI, LT, CO, GF Correlation (Overall Course and Leaf Phenological Stages)
However, Meier and Leuschner (2008) suggested the positive influence of the preceding year’s summer precipitation rather than the air temperature on the total number of the leaves of European beech trees per ground area within those beech forests of central Germany. Indeed, in our study, the correlation between the periodic total precipitations and the LAI data was negative and relatively insignificant (r=-0.230; P = 0.316, Table 1) indicating relatively weak influence of the precipitation on the development of LAI, compared to the air-soil temperature during the intervals between those field visits. The same insignificant situation was valid but reverse for the other canopy parameters; positive correlation coefficients, being r = 0.148, 0.148, 0.088 for the LT, CO, and GF respectively (Table 1). On the other hand, in their study, for the forests in the Flanders region of Belgium, Bequet et al. (2012) insisted on the previous year’s higher summer precipitation as one of the causes for the decrement in LAI within a beech stand whereas indicating positive correlation between the previous year’s spring precipitation and the LAI within an oak stand. For our study, although there was a positive yet insignificant correlation (r = 0.600; P = 0.285, Table 1) between the previous year’s early spring (early March to early April of 2020) precipitation and the LAI data of that particular term (2021), any apparent correlation was not valid for the whole spring season and even for that whole year (r=-0.162; P = 0.484, Table 1). However, being insignificant, correlation coefficients were negative for the other canopy parameters; LT, CO, GF (-0.600 for all, Table 1) for the previous year’s same term, early spring precipitation whereas the situation was not valid for that whole year (r = 0.079, 0.079, 0.047 respectively, Table 1). Nevertheless, increasing trend of the spring precipitation particularly along the last 10 years (Fig. 5) may possibly trigger and lead to the increment of the annual maximum LAI. Similarly, the study by Zhang et al. (2015) also remarked the correlation of the annual maximum LAI and precipitation gradients along the years for some of the dominant tree and shrub species within the central Loess Plateau of China.