Short-term dynamics of Quercus ilex advance regeneration in a Pinus nigra plantation after the creation of small canopy gaps

Aim of the study: The aim of the research is to analyse the role of Quercus ilex advance regeneration in the stand regeneration of pine plantations after small-sized canopy openings, and to assess the influence of the forest stand and the canopy opening. The performance of the advance regeneration under the pine plantation is also examined. Area of study: A Pinus nigra plantation in dry Continental Mediterranean climate in eastern Spain. Material and methods: The tree regeneration of ten canopy openings of 0.17-0.43 ha was monitored during five years after treatment. It was also sampled in 0.12 ha-plots in the non-treated pine plantation surrounding the openings. Main results: An important increase in the height of Q. ilex regeneration was observed in the openings, unlike what was found in the intact pine plantation. In the pine plantation, stand density showed a moderate positive influence on the density of Q. ilex regeneration, whereas in the canopy gaps Q. ilex height was negatively influenced by stand density before the opening. Research highlights: The canopy opening triggered a response in Q. ilex advance regeneration, although height growth rates seemed to reduce over time. The results support the view that promoting Q. ilex in pine plantations may require different management strategies depending on the characteristics of the pine overstorey and on the density and size of the advance regeneration.


Introduction
The evergreen holm oak (Quercus ilex L) forests represent the potential vegetation in more than half of Mediterranean Spain (Maldonado et al., 2002). Their extension is nowadays much more limited, due to the historical spread of agriculture, abusive logging and overgrazing. In the second half of the 20 th century the overexploitation was reduced and many of the degraded lands were recolonised or reforested with pines (Ortuño, 1990). In recent decades the decrease in the economic profitability of timber and traditional forest products has led to abandonment of forest management (Domínguez-Torres and Plana, 2002). Several alternative uses have been proposed to promote forest management, like hunting and mushrooms. In lands suitable for the growing of the prized black truffle (Tuber melanosporum Vitt.) the experts propose to convert pine plantations and shrublands to understocked oak stands in order to improve the habitat for T. melanosporum and promote forest profitability (Reyna et al., 2004;Diette and Lauriac, 2005).
The plantations of Mediterranean pines are frequently invaded by seedlings of Q. ilex (Gómez, 2003). Many oak species regenerate through the accumulation of advance regeneration below overstorey (Johnson et al., 2002). The shoot growth of these seedlings is suppressed by adult trees and they often dieback and resprout. Some of these seedlings show a rapid shoot growth when the overstorey is disturbed (Johnson et al., 2002). Retana et al. (1999) studied the behaviour of Q. ilex seedling under forest canopies and concluded that shade favours seedling emergence and survival during the f irst year; however, light def icit li-Short-term dynamics of Quercus ilex advance regeneration in a Pinus nigra plantation after the creation of small canopy gaps S. Garcia-Barreda 1 * and S. Reyna 1,2 mits seedling growth and often interacts with water deficit.
The role of the advance regeneration of Q. ilex in the stand regeneration after canopy disturbance is controversial. In Q. ilex coppice forests, Retana et al. (1999) and Gracia et al. (2001) concluded that its role is minor in the short term, since the advance regeneration did not show release after thinning the overstorey. The stumps resprout vigorously, rapidly outcompeting seedlings . By contrast, its role in Mediterranean pines stands, which do not resprout, has not been thoroughly examined.
In this study, we analyse tree regeneration in a P. nigra plantation with Q. ilex advance regeneration during the five years after experimental small-sized canopy openings aimed to improve the habitat for T. melanosporum, and we focus on the role of the advance regeneration. The influence of the characteristics of the forest stand and the canopy opening on tree regeneration is also assessed, as well as the influence of forest stand characteristics on the advance regeneration under the intact pine plantation.

Study site
The study was conducted in El Toro, Valencian Community, eastern Spain (39°59' to 40°1' N; 0°44' to 0°46' W, 990-1,050 m a.s.l.). The climate is Continental Mediterranean, with a mean annual rainfall of 500-550 mm and a mean annual temperature of 11.9-12.4°C. The soils are calcixerepts developed on a quaternary calcareous glacis with less than 5% slope.
The area was reforested between 1958 and 1969 with a mixture of the non-native Pinus nigra Arnold subsp. nigra and the native P. nigra subsp. salzmannii Franco. Formerly, it was cultivated with cereal crops. Scattered adult Q. ilex and Quercus faginea Lam. form a lower tree layer. These oaks already existed before the plantation, in the field boundaries.
In March 2000, an experimental silvicultural treatment was executed with the aim of improving the habitat for T. melanosporum fruiting (Reyna et al., 2004). Circular canopy openings were created around the truffle-producing grounds by systematically cutting down all the pines and removing the shrubs. Woody coarse debris were chipped and spread throughout the opening. Adult oaks were retained and slightly pruned.
When the treatment was executed, the density of adult pines ranged from 1,000 to 2,450 trees ha -1 , with Assman top height from 9 to 12 m; the density of adult Quercus ranged from 29 to 259 trees ha -1 , with 62% Q. ilex and 38% Q. faginea; and the understorey shrubs (Juniperus sp. pl., Prunus spinosa L, Quercus coccifera L and Genista sp. pl.) and herbs cover was low (Table 1).

Data collection
Tree regeneration (height < 130 cm) was measured in ten canopy openings created in 2000 in the matrix of the pine plantation. The opening was settled as an expanded gap, with its borderline delimited by the trunks of the trees bordering the gap. The openings were sampled one year after the treatment execution, and again the second year, the fourth and the fifth.
The height and crown surface area were recorded for all the tree regeneration in the opening. The regeneration of Q. ilex and Q. faginea was classified as recent acorn seedlings (with the acorn or cotyledon scatters still attached) or seedling resprouts (individuals subject to recurrent cycles of shoot dieback and resprouting). The

180
S. Garcia-Barreda and S. Reyna / Forest Systems (2013) 22(2), 179-188 origin of the latter (root sprout or acorn) was not discerned, as the seedlings were not dug up. The oak regeneration coming from stump resprouts of adult trees was excluded from data analysis; they accounted for less than 0.4% of all the sampled individuals. In young P. nigra seedlings, subspecies were not distinguished. The density of regeneration of Q. ilex, Q. faginea and P. nigra was computed. For Q. ilex, the percent soil covered by the regeneration was also calculated, as well as the median height of the individuals. The median was preferred instead of the mean because the height of Q. ilex shoots followed a lognormal distribution and a logarithmic transformation was performed. The median height of Q. faginea and P. nigra was not calculated, because in most plots there were less than 30 individuals.
The following characteristics of the forest stand and the canopy openings were determined from fieldwork and aerial photographs (Table 1): site index (Assman top height of P. nigra subsp. salzmannii at the age of 60 was estimated using site index curves from Gómez-Loranca, 1996), pre-treatment canopy cover, residual BA in the opening (in which Q. ilex and Q. faginea were the only trees) as an estimate of shading and competition from residual adult trees, the size of the canopy opening (edge effect is higher in smaller-sized gaps) and the relative abundance of the allochthonous P. nigra subsp. nigra in the dominant tree layer around the openings (in which P. nigra and P. salzmannii accounted for 95% of the trees).
A control plot was sampled in the matrix of pine plantation surrounding each canopy opening. Each control consisted of three square subplots of 400 m 2 , randomly located around the opening, at a distance of 5-10 m from the borderline. The plots were sampled in 2001 and again in 2007.

Statistical analysis
An information-theoretic approach (Burnham and Anderson, 2002) was used to assess the time trend of tree regeneration after the opening and the influence of forest stand characteristics. For the analysis of Q. ilex and Q. faginea regeneration, a set of 32 models was developed that included all the combinations of the following predictor variables: time from treatment, site index, pre-treatment canopy cover, post-treatment BA and opening size. Since P. nigra does not regenerate through advance regeneration, in the analysis of P. nigra regeneration pre-treatment canopy cover was excluded; the relative abundance of P. nigra subsp. nigra was included instead to assess regeneration differences between the two subspecies.
These a priori models were constructed with linear mixed models (LMM) in which time was included as a repeated measures variable, stand characteristics were included as fixed predictors and an unstructured covariance matrix was specified (SPSS, 2006). The response variables were log transformed when the model assumptions were violated, as well as time from treatment.
In the various analyses performed, none of the competing models was strongly supported as the best (w i > 0.9), so model averaging was employed to account for model selection uncertainty (Burnham and Anderson, 2002). The following statistics are provided for inference: the model-averaged parameter estimate, which shows the direction of a predictor effect on the response; its 90% confidence interval (CI); and the sum of the Akaike weights (Σw i ), which indicates the relative importance of a predictor. Predictor variables were considered important in explaining variation in the response when the CI did not overlap zero and Σw i approached one. The best models from each set (with the lower Kullback-Liebler distance to the observed data) are also provided (Supplementary Tables S1, S2).
The characteristics of Q. ilex and Q. faginea seedling bank in the control pine stand were also assessed with an information-theoretic approach. The following predictors were tested: time (sampling of 2001 vs. 2007), site index, canopy cover, and BA of adult Quercus (which grow in the lower tree layer).
A post-hoc analysis was conducted to explore the time trend of Q. faginea regeneration density. The width of the rings of dominant P. nigra subsp. salzmannii (20 trees distributed throughout the study site) was used as a proxy for the effect of weather on trees (Martín-Benito et al., 2008).

Density of Q. ilex regeneration
In the canopy openings, model averaging indicated that time from treatment was the only predictor whose effect on the density of Q. ilex regeneration was clearly supported by the data (the 90% CI did not overlap zero) while showing a high relative importance (Σw i ). The relation was positive (Table 2) and resulted in a rather linear increase during the five years after treatment (Fig. 1a).
In the control pine stand, model averaging of Q. ilex regeneration density indicated that time was the only predictor with both a 90% CI not overlapping zero and a high Σw i (Tables 3, 4). Canopy cover showed a moderate positive influence on the density (Table 4).
Seedling resprouts were the only regeneration type present in the openings one year after treatment, whereas in the fifth year they were 14 times more abun-

182
S. Garcia-Barreda and S. Reyna / Forest Systems (2013) 22 (2), 179-188  dant than recent acorn seedlings. In the control pine stand, seedling resprouts of Q. ilex accounted for almost all the individuals in 2001, and they were 6 times more abundant than recent acorn seedlings in 2007.

Density of Q. faginea regeneration
In the openings, none of the predictors used in the model averaging showed both a 90% CI not overla-pping zero and a high Σw i . However, time from treatment and site index showed a moderate positive influence on the density of Q. faginea regeneration, and post-treatment BA showed a moderate negative influence (Table 2, Fig. 1a).
Since time did not show a strong and straightforward effect on Q. faginea regeneration density (Fig. 1a), a post-hoc analysis was conducted to compare three different time patterns: linear, quadratic and related to P. nigra ring width. Ring width reached a much lower Dynamics of Q. ilex advange regeneration in a P. nigra plantation after canopy opening 183  In the pine plantation, model averaging of Q. faginea regeneration density indicated that time was the only predictor with both a 90% CI not overlapping zero and a high Σw i (Tables 3, 4).
Seedling resprouts were the only regeneration type found in the openings one year after treatment, whereas in the fifth year they were 15 times more abundant than recent acorn seedlings. In the pine stand, seedling resprouts of Q. faginea accounted for almost all the individuals in 2001, and they were 31 times more abundant than recent acorn seedlings in 2007.

Density of P. nigra regeneration
Model averaging of P. nigra seedling density in the openings revealed that relative abundance of adult P. nigra subsp nigra around the opening, opening size and site index showed a 90% CI not overlapping zero while having a high Σw i . The effect of the two former on the density of P. nigra regeneration was negative, while the effect of the latter was positive (Table 5).
Five years after the opening only 9% of the seedlings were two or three years old and the rest were one year old. In the pine plantation, none of the seedlings found (Table 3) was more than three years old.

Height of Q. ilex regeneration
Model averaging of Q. ilex median height in the openings revealed that time from treatment, site index, pre-treatment canopy cover and post-treatment BA presented both a 90% CI not overlapping zero and a high Σw i (Table 2). Time showed a positive relation with the median height of Q. ilex regeneration, with an apparent curvilinear pattern reflecting an increase that levelled off with time (Fig. 1b). Site index showed a positive effect, whereas the effect of pre-treatment canopy cover and post-treatment BA was negative. The increase in height from the fourth to the fifth year after treatment did not show a negative correlation with the density of Q. ilex regeneration in the fourth year (Pearson's r = 0.59; 90% CI: 0.06, 0.86).
Model averaging of Q. ilex median height in the pine stand indicated that time was the only predictor with both a 90% CI not overlapping zero and a high Σw i , with median height being lower in 2007 than in 2001 (Tables 3, 4).
On average 8.3 individuals ha -1 recruited into the adult stratum (height > 130 cm) between the first and the fifth year after the canopy opening, which account for 1.5% of the seedling resprouts found in the first year. In the pine stand, only one individual of Q. ilex regeneration recruited into the adult stratum in the ten plots between 2001 and 2007, whereas five individuals reduced their height from above to below 130 cm.

Soil cover by Q. ilex regeneration
In the openings, model averaging indicated that time from treatment was the only predictor with both a 90% CI not overlapping zero and a high Σw i (Table 2). Its relation with the percent soil cover by Q. ilex regeneration was positive, and resulted in an apparently linear increase during the five years after treatment (Fig. 1b).
In the pine stand, model averaging of the soil cover by Q. ilex regeneration indicated that time was the only predictor with both a 90% CI not overlapping zero and a high Σw i (Tables 3, 4).

Discussion
Canopy disturbance usually triggers a period of rapid change in the vegetation. During this stand initiation stage it is usually difficult to accurately predict changes, although of high interest because future stand characteristics are highly influenced by the initial stage (Johnson et al., 2002).
In our study site the canopy openings were created in the matrix of a 31-42 year-old P. nigra plantation 184 S. Garcia-Barreda and S. Reyna / Forest Systems (2013) 22 (2), 179-188 with high density and a sparse lower tree layer of oaks. No established P. nigra regeneration was found under this canopy. Lucas-Borja et al. (2011) found that BA ranging 25-40 m 2 ha -1 (similar to those in our study site, Table 1) are optimum for the emergence of P. nigra subsp. salzmannii seedlings in Central Spain, although Gómez-Aparicio et al. (2006) showed the negative effect of shade on seedling biomass. Tíscar-Oliver (2007) found that seedling survival is highly dependent on the first summer drought. By contrast, Q. ilex regeneration was relatively abundant under the pine canopy, and its density was of the same order of magnitude than that found by Ruiz-Benito et al. (2012) in Spanish P. nigra plantations and by Gómez-Aparicio et al. (2009) in pine plantations in southern Spain, although much lower than the density found by Gracia et al. (2001) in Q. ilex coppice stands of northeastern Spain. In southern Spain, Urbieta et al. (2011) found that Q. ilex regeneration was much more abundant in oak than in pine forests, although the maximum was obtained in mixed stands.
European jays abundantly disperse Quercus acorns in pine stands (Gómez, 2003). However, seedling resprouts were much more frequent than young acorn seedlings. Q. ilex regeneration was small-sized and no evidences of a clear positive time trend in height growth were found. These are characteristic features of oak species that form seedling banks and perpetuate through repeated shoot dieback and resprout (Johnson et al., 2002).
After the canopy opening Q. ilex was also the most abundant tree regeneration. The scarcity of recent acorn seedlings indicates the dominant role that the advance regeneration played in forest dynamics during the first years after the opening. Recruitment of new acorn seedling was also observed, but the survival of young Q. ilex seedlings in full sunlight is usually low (Marañón et al., 2004;Prévosto et al., 2011b) and therefore it is difficult to predict if they will increase the density of Q. ilex meaningfully.
Q. ilex is considered as intermediate in shade tolerance (Valladares and Niinemets, 2008), with seedling recruitment and early survival being higher under full canopy than in full sunlight. In our pine stands, the canopy cover positively related to the density of Q. ilex regeneration. The harvesting operations and the sudden exposition to full sunlight may have hampered the survival of some young seedlings and seedling resprouts with low root reserves (Dillaway et al., 2007). However, after the opening the density of Q. ilex re-generation increased linearly with time, indicating the vigour of the advance regeneration to resprout and survive in full light. Retana et al. (1999) and Marañon et al. (2004) pointed out that, despite its shade tolerance, Q. ilex seedlings are unlikely to grow into saplings with low light levels. Gómez-Aparicio et al. (2009) found that in southern Spain the recruitment of Q. ilex seedlings was at maximum at pine densities similar to those in our study site, whereas the density of saplings was maximum at lower densities. In a Pinus halepensis Mill. shelterwood in southern France, Prévosto et al. (2011a) found that the growth of sown Q. ilex was higher in the stands with lower BA (10 m 2 ha -1 ). We observed an important increase in the height of Q. ilex regeneration in the years immediately after the canopy opening, whereas in the pine plantation no evidences of a clear height increase were found during the same period of time. This indicates that under the pine canopy there were Q. ilex seedling resprouts that retained the ability to grow rapidly after release, in contrast to the lack of response reported by Gracia et al. (2001) in thinned coppice forests. This tends to support the view that at least part of the seedling resprouts is able to recruit into the adult stratum when they are released.
However, the evolution of Q. ilex median height suggests a reduction in the growth rate of the regeneration over time. This reduction is not apparently related to intraspecific competition among seedling resprouts, since the density of regeneration was low and the correlation between density and height increase was not negative. Another factor which could influence height growth in small openings is the edge effect; however in our openings the size did not appear to be as important as the other predictors. This could be an effect of the narrow range of this variable (Table 1): York et al. (2003) studied group selection cuttings with a larger range of sizes (0.1-1 ha) in California and found that opening size was positively related to seedling height growth.
Another factor frequently affecting the behaviour of oak advance regeneration is the competition of the understorey shrubs (Johnson et al., 2002). However, in our study site the development of the understorey was moderate (Table 1). This is probably related to the former agricultural use and to the high stand density of the pine stand, although under the Continental-Mediterranean climate of central Spain the density of the woody understorey is usually low (Costa et al., 1996).
On the other hand, the canopy cover before treatment and the residual (post-treatment) BA of adult oaks negatively related to the height of Q. ilex regeneration. The residual BA of adult oaks showed a negative effect despite being always lower than 2.5 m 2 ha -1 . Gracia et al. (2001) found that in thinned Q. ilex coppice forests with BA 15-17.5 m 2 ha -1 the growth of Q. ilex seedlings was suppressed. This suppressive effect could be caused by shading, root competition or topsoil properties (Puerta-Piñero et al., 2006). The negative influence of the canopy cover before treatment agrees with the results of Dillaway et al. (2007), where the light levels under close canopy positively related to the root diameter and reserves of Quercus alba L advance regeneration, the latter being related to the vigour of advance regeneration. This suggests that a previous thinning of the pines could have improved the growth of seedling resprouts after opening.
In the case of Q. faginea the advance regeneration also played an important role in the openings. In contrast with Q. ilex, the increase in Q. faginea density was not as steady, and it was moderately influenced by site index and by the residual adult oaks. This suggests a less vigorous resprouting and survival of seedling resprouts in full sunlight. Q. faginea is a semi-deiciduous species that exhibits biological features similar to those of Q. ilex, but it is considered more sensitive to drought (Sanz-Pérez et al., 2007). The fact that the increase in the density of Q. faginea regeneration in the openings was better explained by annual weather conditions than by time from treatment supports this view.
The regeneration of P. nigra did not establish for the moment in the openings, and the emergence of seedlings was influenced by site index. This limitation in the effective recruitment is similar to the difficulties in regeneration after wildfire and clearcutting encountered for P. nigra subsp. salzmannii in Spain, where this taxon is considered quite shade-tolerant compared to other native Spanish pines (Ordóñez et al., 2004;Tíscar-Oliver, 2007).
On the other hand, P. nigra subsp. nigra has been traditionally considered as shade-intolerant, although Chauchard et al. (2006) found that it did not behaved as such in southeastern France. In Spain we know of no study about its shade tolerance. Due to the high mortality of young P. nigra seedlings, we could not determine which taxon they belonged to. However, the negative relation of seedling density in the openings with the abundance of adult P. nigra subsp. nigra around the openings suggests that the emerging seedlings were mostly P. nigra subsp. salzmannii.
The current lack of effective regeneration of P. nigra in the openings does not necessarily involve the loss of pines in the medium term. In burned Mediterranean pine forests, Gracia et al. (2002) observed that P. nigra subsp. salzmannii regenerated abundantly only after the resprouting Q. faginea covered the soil (from the 4 th to the 15 th year), and in the 37 th year pines dominated the tree layer, pointing to a nurse effect. In our study site, the soil cover by Q. ilex is low five years after the opening (Fig. 1b), so a long-term monitoring would be necessary to discern if P. nigra establishment is restricted by site conditions or by the scarce canopy cover.
Our results show that the canopy opening triggered a response in Q. ilex advance regeneration. However, since the monitoring suggests a reduction in median height growth over time, the effective recruitment of Q. ilex into the dominant tree layer will depend on the proportion of seedling resprouts capable of maintaining a competitive height growth in the new conditions. Our results also suggest that the stand density before treatment and the stand density of oaks after treatment are detrimental to the height growth of this regeneration.
The management objective in the study site was to achieve an open oak-dominated stand around the truffle-producing grounds in order to improve the habitat for T. melanosporum fructification. The size of the openings was kept to the minimum necessary in order to enhance the compatible management of truffle and other forest uses (Reyna et al., 2004). In canopy openings with sizes ranging 0.17-0.43 ha we did not find clear evidences that the opening size hindered the release of Q. ilex seedling resprouts; and Q. ilex regeneration enjoyed a competitive advantage over P. nigra. The statistical models predict that soil cover by Q. ilex regeneration will be on average 4% in the 10th year and 8% in the 20 th year. This will add to the canopy cover of residual adult oaks, which was on average 12% one year after the opening and 13% in the fourth year. Soil cover by Q. ilex will likely remain below 30% in the short term, thus suitable for T. melanosporum production (Reyna et al., 2004) without additional silvicultural operations.