Quercus cerris L. (Turkey oak, Fagaceae) acts as a dominant species in a large part of the oak forests of central and southern Italy (Blasi et al. 2004; Di Pietro et al. 2020b; Terzi et al. 2021). In the Molise region, the Q. cerris forests are widely distributed within the hilly, submontane and lower montane belt, ranging in altitude between 300 and 1200 m a.s.l. (Blasi and Paura 1995; Blasi et al. 2004, 2005). In particular, a wide Q. cerris belt separates thermophilous mixed oak woods with Q. cerris, Q. pubescens Willd. and Q. frainetto Ten. of the hilly belt from the beech woods of the montane belt. The Q. cerris belt is linked to Flysch substrates or the occurrence of large karst basins on limestone bordered by mild slopes covered by leached rich in clay soils. However, the abundance of Q. cerris woods in the Molise region is also due to the application of forestry management (e.g., different forest systems and silvicultural systems) that have advantaged this species over other tree species (e.g., Q. petraea (Matt.) Liebl. and Q. frainetto). The high-forest system of the Turkey oak woods, which is typically adopted in many parts of Italy and especially in the Molise region, finds its historical reason in the production of railway sleepers and, only marginally, of firewood (Di Martino 1996). Owing to the importance of Turkey oak for the regional economy, and paying attention to the ongoing climate change, we carried out an in-situ study aimed at identifying possible phenotypic response and plant functional traits (PFTs) variations from Q. cerris seedlings coming from wood stands characterised by different environmental conditions and forestry treatment. PFT variations have proved to be useful for establishing ecological and evolutionary trends for plant species and communities. They have often been used to interpret plant-environment relationships, to make previsions on possible consequences and to quantify a wide range of natural and/or human-driven processes in terms of species or community ecology (Pèrez-Harguindeguy et al. 2013; Adler et al. 2014; Pérez-Ramos et al. 2019). The measurements of the functionality patterns of seedlings of Q. cerris growing under different forest management, could provide useful information on the health state of the woods as well as address forest management policies in the next future. Surprisingly the available literature on such studies (i.e., observations in situ on Q. cerris seedlings) is relatively scarce in Italy. Most of the ecological studies performed on Q. cerris dealt with adult trees (e.g., Gratani and Foti 1998) or analysed, in greenhouses or under controlled conditions, the response of woody species seedlings to various types of induced stress (Otieno et al. 2005; Aref et al. 2013; Karavin et al. 2013; Chiatante et al. 2015; Abdala-Roberts et al. 2018; Torres-Ruis et al. 2019). Other studies have focused on the phenotypic plasticity of a species as a measure of its adaptability and fitness to environmental and climatic changes (Valladares et al. 2000, 2014; Pérez-Ramos et al. 2019). Field studies on the vitality of plants using PFTs, however, are more frequently applied to grassland communities (Bolzan et al. 2007; Catorci et al. 2013; Chelli et al. 2019; Baltieri et al. 2020) than to forest ones.

This study is part of an interdisciplinary and broader line of research that has been undertaken by the Laboratory of Systematic Botany and Floristics of the University of Molise in the last decade, which aims to shed light on the morphological, ecological, and bio-molecular features of Quercus species (Antonecchia et al. 2015; Fortini et al. 2015a, 2015b; Di Pietro et al. 2016; Conte et al. 2019; Di Pietro et al. 2020a, 2020c; 2021). Specifically, the present paper aims to evaluate the response of Quercus cerris seedlings to different ecological and forest management conditions, using PFTs and the analysis of morphological trait variations.

To obtain coenological information on the three Q. cerris stands a field-sampling campaign was carried out during the 2019 summer period using the phytosociological approach (Braun-Blanquet 1964). Species and syntaxa nomenclature in the phytosociological table followed Bartolucci et al. (2018) and Biondi et al. (2014), respectively. The following Ellenberg indicator values (L - light availability; T - temperature) and edaphic factors (U - soil moisture; N - soil nutrient) as reported in Pignatti (2005) were also considered to infer possible ecological differences between the three Q. cerris stands investigated. For the indicator values that were missing for some species in Pignatti (2005), reference was made to Sürmen et al. (2014). Subsequently, Pignatti-Ellenberg index was applied to the matrix of phytosociological relevés by calculating the weighted averaging using species abundances to evaluate the average response of each forest community to the abiotic factors associated with Ellenberg indicator values, following what is reported in Gristina and Marcenò (2008) from the unpublished data of Pignatti and Bona (2007).

The in-field collection of Quercus cerris seedlings followed mainly the protocol reported in Gottardini et al. (2016) according to the following steps: 1) identification of a sampling stand, with a random collection of thirty 3 to 4-year-old seedlings, starting from the centre of the site and maintaining a distance of 6–7 m between each seedling collected, choosing seedlings with fully expanded leaves free from damage from herbivores or pathogens (Garnier et al. 2001); 2) eradication of the seedlings by taking an entire clod of soil with a spade up to 30 cm deep and subsequent removing of the soil from the root through water; 3) measurements of the soil pH (classes of pH soil according to National Soil Survey Center 1998) precisely where the individual seedlings were eradicated; 4) measurement of the bulk density of the soil through a metal cylinder (classes of bulk density soil according to National Soil Survey Center 1998); 5) application of a damp cotton cover to each seedling and store it in dark in the cooler and 6) measurements of the chlorophyll leaf content by the at LEAF CHL PLUS instrument (FT Green LLC, Wilmington, US), whose function is comparable to the SPAD 502 (Minolta) (Zhu et al. 2012; Novichonok et al. 2016). The measurements were repeated three times in the central part of the lamina of three leaves directly in the field.

The first stage of field sampling was carried out during three consecutive days (24, 25 and 26 June 2019) under the same climatic conditions (no wind and no clouds). A total of 90 seedlings were collected and further analysed. The chlorophyll contents of the leaves of the seedlings were determined in August 2019 for 30 different seedlings in each stand due to difficulties in obtaining the instrumentation for measuring the chlorophyll content during the June 2019 field campaign. Laboratory measurements followed Pérez-Harguindeguy et al. (2013). Stem length, root length and leaf area of each seedling were measured on fresh materials. Three leaves for each seedling were fixed on a sheet and scanned using the Epson GT15000 scanner, and the leaf area (LA, cm²) was calculated using the ImageJ software (Rasband 1997–2018). Fresh leaf mass saturated (LFMsat.) was calculated (after seven hours from the collection) and dry leaf mass (LDM), was calculated after drying leaves for 72 h at 60 °C until a stable weight was achieved. The leaf measures were used to evaluate the following plant functional traits: specific leaf area (SLA), leaf dry matter content (LDMC), leaf thickness (Lth), leaf water content (LWC) and leaf mass per area (LMA) (Table 2).

It should be noted (Table 2) that some of the PFTs investigated express approximately the same concept providing redundant information (e.g., SLA/LMA or LDMC /LWC). Accordingly, we have decided to make a further selection considering three PFTs only, namely SLA, LDMC and Lth. A spreadsheet (Excel 2016) containing all the field and laboratory data was arranged and processed with XLSTAT 2020 3.1.3 software (Addinsoft 2005–2020). The data underpinning the analysis reported in this paper are deposited at “Mendeley Data” at https://doi.org/10.17632/6m2d4rxc7n.1. All the data were previously subjected to a normality check and then to a non-parametric comparison analysis, according to the Kruskal-Wallis method, with a paired comparison according to the Dunn method (with Bonferroni correction for significance, set at alpha = 0.05).

All three wood stands investigated were assigned to the same association, which is Roso arvensis-Quercetum cerridisUbaldi 2003. Nevertheless, owing to their different silvicultural treatment and ecological features, the three forest stands exhibited clear floristic-coenological differences. The BSL stand is dominated by Q. cerris in the upper tree layer and Carpinus orientalis in the lower tree layer, and the coverage of the shrub layer was found to be very high (40–90%). The PEi values were high for T and L. The BF stand is characterised by a dominance of Quercus cerris, with Carpinus betulus having a subordinate role, and by a high coverage by the herb layer (75–90%). The PEi values were found to be high for T and L. The SC stand is characterised by the co-dominance of Quercus cerris and Carpinus betulus in the upper three layer and by an almost totally missing shrub layer. The PEi values were high for U and N and low for T and L.

The stem and root lengths and the leaf areas of the seedlings were found to be rather similar in the three wood stands. Even if the Q. cerris seedlings of the three wood stands investigated are phenotypically very close each other, they showed a different response to environmental conditions if the PFTs values are considered. In the ecological literature, high values of SLA are normally related to environments characterised by being rich in nutrients and/or by shaded conditions. The higher values of SLA showed by the saplings of the SC stand compared to those of the BF and the BSL testify to the propensity of the SC seedlings for an investment in carbon oriented to a rapid growth of the saplings rather than to store material for longevity. High SLA values are related to a lower occurrence of dense tissues, which, as a negative counterpart, makes these seedlings more palatable to grazing herbivores. This negative effect is balanced by a more rapid production of new leaves, linked to a higher availability in nutrients, water, as suggested by the PEi values (see Fig. 3) (Wilson et al. 1999; Shipley et al. 2002; Wright et al. 2004; Pontes et al. 2007).

In both the BSL and the BF stands, the values of LDMC are higher than those measured in the SC, showing, therefore, for the seedlings of the BSL and the BF a greater propensity to longevity (Dijkstra and Lambers 1989; Niemann et al. 1992; Garnier and Laurent 1994; Kazakou et al. 2006). The low SLA and the high LDMC values in the BSL and the BF highlight different strategies compared to the SC stand. We could indirectly address these differences to a lower presence of resources for leaf development. The low Lth values of the SC stand are related to seedlings with thinner leaves. This is another PFT that identifies leaves with a greater propensity to grow. The Lth values are normally negatively correlated with photosynthesis and growth, while being positively correlated with leaf longevity and construction costs (Vile et al. 2005).

Considering that PFT values, such as SLA, LDMC and Lth, are in many cases used to estimate the growth rate of seedlings (Poorter and Garnier 1996; Wright and Westoby 2001), it can be assumed that the SC seedlings grow faster than those of the BF and the BSL. In broad terms, high values in SLA and low values in LDMC correspond to a rapid production of biomass, whereas the low values in SLA and high values in LDMC indicate the conservation of nutrients through long-lived leaves (Reich 1998; Garnier et al. 2001). Assuming that the SC stand seedlings benefit from a greater number of resources that lead these seedlings to exhibit higher SLA values, the next step could be making hypothesis on which kinds of resources are important. In this case, the response derives directly from the PEi, which indicates, for the SC stand, a higher availability of nutrients (N) and soil moisture (U). The flat geomorphological profile of the SC stand and the clayey substrate may play a crucial role in leading to a high degree of soil water retention. While refraining from drawing up a ranking of PEi by importance, we underline the effect that soil moisture can have for the deciduous forest communities occurring in Mediterranean or sub-Mediterranean environments, these latter being usually characterised by more or less prolonged dry periods during late spring and summer.

The pH of the soil, on the other hand, would seem to provide values in contrast, as it goes from 4.5 (SC) to 4.7 (BF) and 5.6 (BSL). These values are more consistent with a typical oligotrophic Quercetalia robori-petraea wood than with eutrophic Quercus cerris woods of Erythronio-Carpinion. Instead, the pH values of the BSL site were perfectly in line with this latter alliance. It would be interesting to understand if these low pH values of the SC and the BF could at least partially explain why the montane-belt Quercus cerris forests often give rise to almost monophytic communities, or in any case, communities lacking other species of oaks (see for example Di Pietro and Tondi 2005). It is possible that in addition to the altitude, the low soil pH values lead Quercus cerris to be less subjected to the competition from other oak species that are common in this area (e.g. Q. pubescens and Q. frainetto), whereas acidophilic oaks (e.g., Q. petraea and Q. robur) that potentially could take advantage from such low pH values are missing due to their rareness in southern Italy.

Another resource that could explain the high SLA values observed for the SC stand could be the light radiation. The PEi values showed a significant difference between the low value (5.19) of the SC stand and the values of the BF and the BSL stands (5.932 and 5.627, respectively).

Observing the headline of the phytosociological table, it emerges that the SC wood stand, while showing a high coverage index at the upper tree layer is characterised by an almost total lack of shrub layer. On the contrary, the BSL wood stand shows the highest cover indices for the shrub layer. It is possible, therefore, that in the SC population, the high coverage of the tree layer was much more effective in reducing the amount of light available for the herb layer than the dense shrub layer occurring in the BSL stand. Accordingly, the high Lth values observed in the BSL stand could be due to the low coverage value of the upper tree layer. In this case, a shrub layer, composed of prostrate species, such as Hedera helix and Rubus ulmifolius occurring in the BSL would not seem to represent an effective impediment to light in reaching the herb layer.

As far as values of CHL are concerned, these were found to be higher in the SC stand than in the other two stands. A greater quantity of chlorophyll favours a greater absorption capacity of solar radiation. Based on published studies (e.g., Curran et al. 1990; Filella et al. 1995; Gitelson et al. 2003) higher chlorophyll content suggests also a higher photosynthetic potential. This is in accordance with other PFT values observed in this study, such as the lower Lth values and with the higher soil moisture, which would turn out to be higher in the SC stand than in the other two stands (see Fig. 2). The soil moisture is one of the factors that is normally considered as influencing the photosynthetic potential (Zhang et al. 2011; Aref et al. 2013). All these results seem to agree that the seedlings of the SC stand, due to the lower light irradiance and higher soil moisture, are more prone to a rapid development than the seedlings of the BF and the BSL stands. The latter (BLS), growing on slightly drier substrates and greater light intensity, exhibits seedlings developing thicker leaves to cope with lower water availability and higher irradiance (Wright et al. 2004; Poorter et al. 2009).

In this study, phenotypic parameters, and some plant functional traits of Quercus cerris seedlings were tested for the first time in natural forest communities to highlight possible adaptation strategies to environmental conditions. The results do not show a significant link between phenotypic plasticity and environmental parameters. It emerged (at least in our case), that the phenotypic expression is not related with the growth strategy of the seedlings. Accordingly, the phenotypic plasticity of the Quercus cerris seedlings should not be considered a target element to detect the fitness to ensure the growth of the tree. On the contrary, PFT values allowed for highlighting significant differences between the three wood stands considered and for hypothesising possible strategies adopted by Quercus cerris in the first years after germination, depending on the environmental characteristics of the sites. Apparently, the PFT values evidenced that the seedlings of Selva di Castiglione (SC) were following a different development strategy than those of the Bosco della Ficora (BF) and Bosco di San Leo (BSL). Thus, the SC seedlings strategy, expressed by higher values of SLA and CHL and lower values of LDMC and Lth leads to the production of thinner leaves and testifies a greater investment in carbon aimed to a rapid growth of the seedling. The presence of more favorable conditions in the SC stand for the development of the seedlings is supported by the Pignatti-Ellenberg index (PEi) values, which confirm for this site a greater availability of nutrients and soil humidity.

Beyond contributing to a relatively still poor national PFTs database with the first data of in situ PFTs from southern Italy regarding a species of enormous territorial value such as Quercus cerris, this study emphasises the importance of investigating on the relationships between young plant development and environmental conditions. The in situ studies on plant functional traits of seedlings of forest species, providing crucial information on their ecology, may assume a great importance to address forest management practices. This information is likely to prove useful in the near future, precisely in the light of recent Italian and European policies linked to the containment of global warming. The guidelines for the “urban and extra-urban forest plan” of the National Recovery and Resilience Plan (Mission 2; Component 4; Investment 3.1) recently published by the Italian Ecological transition Ministry require specific references to natural forest communities and native woody flora to propose models of new urban forests that are consistent with the environmental characteristics of the sites and with the natural potential vegetation types. Yet, the new EU forestry strategy, which plans to plant 3 billion more trees in the European territory by 2030, can become an opportunity to apply the principles of ecology and vegetation science to sustainable land management.