Introduction

Plant reproductive phenology can be briefly described as the timing of certain events in the plant life cycle, such as flower and fruit production, which are generally triggered by environmental cues (Richardson et al. 2013; Lima et al. 2021). The changes in plant reproductive phenology frequently affect the very demographic process in a plant population, as well as the interactions among different plant species, and between plants and co-occurring animals in the community (Morellato et al. 2016; Kharouba et al. 2018; Brown et al. 2022; Vogel 2022) among other processes. Nowadays, the shift in plant reproductive phenology that is induced by climate change is a common phenomenon which is described worldwide, although it is modulated by factors such as the climatic region, the type of biome, and the plant species, among others (Cleland et al. 2006; Rosenzweig et al. 2008; Richardson et al. 2013; Mo et al. 2017; Pérez-Ramos 2020; Vogel 2022). However, there remain many uncertainties on the consequences of changes in reproductive phenology of most plants, and the vulnerability of different ontogenetic stages of plants to changes in climate continue to be a major challenge (Richardson et al. 2013; Caignard et al. 2017).

Seed fall timing is crucial in the reproductive cycle of a plant, because the dropped propagules endure a set of climatic conditions and biological interactions that are frequently critical for the success or shortage of the seedlings (Nathan and Muller-Landau 2000). This is especially relevant in plant species that show recalcitrant or desiccation-sensitive seeds, as is the case with most oak species (Walter et al. 2013). In opposition to dry seeds, recalcitrant seeds do not form seed banks, but they do germinate relatively fast. Hence, the ecological conditions at the microsite where the seed arrives are of utmost importance for the seed and the fate of the seedling. However, many plant species show a long seed dropping season that frequently lasts several months. Consequently, the variability in the environmental conditions that are suffered by these seeds can be important, especially in temperate regions and other regions with high seasonality. This variability includes not only climate and microclimate but also frequently biological interactions, such as pre- and post-dispersive seed predation (Leiva and Diaz-Maqueda 2016; Sone et al. 2016).

Holm oak (Quercus ilex L. subsp. ballota (Desf.) Samp) and cork oak (Q. suber L.) are two Mediterranean oak species that show long acorn falling seasons (Hasnaoui 2008; Sunyer et al. 2014). These species occupy large areas in the Mediterranean Basin, and are dominant in a type of seminatural, savannah-like ecosystem, that is named locally as dehesas in Spain and Montado in Portugal. The dehesas are very characteristic in the western and southwestern Iberian Peninsula, and they are composed by scattered oak trees (30 trees/ha on average) and a grassland matrix. Oaks are generally in monospecific or mixed stands with Q. ilex subsp. ballota and Q. suber being the major species, while species like Q. faginea Lam. or Q. pyrenaica Willd among others, are locally abundant (Costa et al. 2006, 2017). The dehesas are primarily devoted to livestock rearing and represent the largest agroforestry ecosystem in Europe (Eichhorn et al. 2006). The dehesas are also regarded as a traditional system of resource use with important economic value (Campos et al. 2020; Costa et al. 2006), which sustains a high level of biodiversity and provides important ecosystem services, such as high-quality meat production from cattle breeds, lambs, Iberian pig products, traditional knowledge conservation, recreation, eco-tourism, etc. (Garrido et al. 2017). However, the long-term persistence of these ecosystems is compromised because of limited oak regeneration (Gea-Izquierdo et al. 2006; Pulido et al. 2013; Leiva and Vera 2015), together with the adults that die off due to diseases caused by exotic pathogen (mainly the root disease by Phytophthora cinnamomic) (Duque-Lazo et al. 2019; Dorado et al. 2023).

One of the most important limitations to oak regeneration in the Mediterranean climate is the summertime drought, which represents a bottleneck for the survival of young seedlings (Leiva et al. 2013; Pulido et al. 2013; Morilla et al. 2023), a stressor that will likely get worse in the future, due to the expected increase in aridity in the Mediterranean region of Europe (IPCC 2021). However, the expected climate changes could also affect earlier stages in the oak regeneration cycles, such as the survival of dropped acorns and the establishment of the seedlings during the springtime, before the onset of the summertime drought. This process would represent an additional bottle neck for oak regeneration, thus reducing the pool of seedlings that would be ready to withstand the summertime drought. To the best of our knowledge, to date, this process has not been studied in detail. We can take advantage of the extended acorn-dropping phenology in Q. ilex subsp. ballota and Q. suber (Hasnaoui 2008; Sunyer et al. 2014) to investigate its effect on the success of seedling establishment, as well as to infer potential consequences of climate change in the regeneration cycle of these oak species. To this end, the use of phenological studies to shed light on plant response to climate change is becoming increasingly important in recent years (Thackeray et al. 2016).

The specific goals of this study are to: (I) analyse whether differences in acorn dropping time and potentially related differences in predispersal predation of acorns (i.e., acorn infection by insect larvae) have any significant effect on acorn fate and seedling establishment in Q. suber subsp. ballota and Q. ilex; (II) determine whether acorn infection by insects interacts with acorn predation by rodents; and (III) assess whether the kinetics of seedling establishment and seedling height increase are affected by acorn dropping and sowing time. For this purpose, we carried out two field experiments, in a representative dehesa, in two successive years. In the first experiment, the potential success of propagules was inferred from independent studies on the inherent ability of acorns to achieve success (in the absence of predation), and on the rate of acorn predation by rodents. In the second experiment, the two factors (i.e., inherent acorn success and acorn predation by rodents) were jointly studied, and the results were used to validate the first-year results. In the second experiment, the relationship among air temperature and kinetics of seedling emergence and height increase was also analysed. In addition, we conducted a field survey on temporal changes in acorn infection by carpophagous insects. We hypothesised that (I) late dropping and sowing of the acorns will result in a lower number of established seedlings in late springtime, and that these seedlings will have less developed (i.e., shorter) shoots than the seedlings from early dropped acorns. We also hypothesised that (II) acorn infection by carpophagous insects will be higher at the beginning of the acorn dropping season, and that the infected acorns will suffer less predation by rodents than the non-infected acorns.

Material and method

Focal species; acorn and seedlings characteristics

Q. ilex subsp. ballota and Q. suber are Mediterranean oak species in the subgenus Sclerophyllodris and Cerris, respectively (Castroviejo 1986–2012). The two species show recalcitrant seeds with sensitivity to desiccation, and they do not form dormant seed banks (Xiao et al. 2010; Walter et al. 2013). Acorn dropping seasons last from October to January or February and occasionally to March both in Q. ilex subsp. ballota and in Q. suber (Sunyer et al. 2014; Hasnaoui 2008), although peak acorn-dropping is earlier in Q. ilex subsp. ballota (Gómez-Casero et al. 2007). After acorns fall, they germinate relatively soon, and the hypocotyl develops and produces a large taproot that stores much of the acorn’s reserves over the wintertime (Kabeya and Sakai 2003). Epicotyl emergence, shoot enlargement, and leaf development take place a few months later, with emergence times in the field of 130 to 170 days (González-Rodríguez et al. 2012; Leiva et al. 2013).

Site description

The study was carried out in a Mediterranean dehesa, located in Doñana Natural Park, Southern Spain, 35 km inland from the Atlantic Coast (37º 14’ 33’’N; 6º 19’ 37’’W. Villamanrique de la Condesa Municipality). The topography is very flat, and the mean elevation is 39 masl. The climate is Mediterranean with wet, mild winters and long, dry summers. Mean annual temperature is 17.7 ºC, while mean annual rainfall is 465.6 mm. The coldest month is January with an average lowest temperature of − 0.1ºC and mean minimum temperature of 4.4 ºC. This month also concentrates the highest number of frost days (i.e., 3.7 days/month on average). The warmest month is August with average highest temperature of 41.8 ºC and mean maximum temperature of 34.6 ºC. Rainfall concentrates from September to May (99% of the annual rainfall), while precipitations are negligible from June to August (i.e., 0.1 to 2.1 mm/month on average) (the values rely on data from a nearby station; ICTS DOÑANA, Palacio-Meteorological Station; period 2010-11 to 2021-22). Soils are sandy, derived from quartzitic sand. The vegetation is a savannah-like formation composed by scattered oaks in the overstory, and an annual grassland layer in the understory. Q. suber is the dominant oak species while Q. ilex subsp. ballota is subdominant, and mostly occurs in patches. The combined oak density is ca. 30 individuals/ha on average. The area is devoted to livestock rearing, and is grazed by free range cattle (110 individuals grazing on 350 ha) and a goat herd (500 individuals which graze for ca. 5 h/day). In addition to livestock, rodents (Apodemus sylvaticus and Mus spretus) are major acorn predators, consuming as much as 90% experimental acorns when livestock is excluded (Leiva and Vera 2015). Other, less abundant wild acorn predators occurring in the area are rabbits (Oryctolagus cuniculus) and badgers (Meles meles). Acorn infection by insect larvae is due to the weevil Curculio elephas, which is the most frequent predispersal predator of acorns, and to the moth species Cydia fagiglandana. After oviposition of the adults, the larvae of these species develop inside the acorns, consume the endosperms, and then leave the acorns when they drop to the ground (Leiva and Fernández Alés 2005).

Experimental design

Two similar experiments were conducted in two successive years (2015-16 and 2016-17). In the first year, the importance of acorn predation by rodents and the inherent propagule failure due to other causes were tested separately, and the final success of propagules was predicted from these results. In this experiment, sound acorns as well as acorns infected by insect larvae were included. In the second-year experiment, both acorn predation by rodents and inherent failure in acorn survival and seedling development were jointly tested, and the results on the realised propagule success were used to validate the first year experiment. The kinetics of seedling emergence and seedling height increase were also analysed in the second year experiment. The climate data for the two study years (Fig. 1) were recorded at the same weather station mentioned above (ICTS DOÑANA). In addition, a field survey on the kinetics of acorn infection was carried out during the first year study.

Fig. 1
figure 1

Climate data during the first (a) and second year (b) experiments (2015-16 and 2016-17, respectively). Two solid arrows indicate early and late acorn sowing or acorn addition to the plot, respectively. Dotted arrow indicates end of the experiment. Prec.= precipitation, T max.= mean maximum temperature, T high = highest temperature, T min. = mean minimum temperature, T low = lowest temperature

Full size image

2.3.1. First year studies. (i) Predicted effect of acorn dropping time and acorn infection on propagule success and (ii) field survey on the kinetics of acorn infection.

The study was conducted in a representative stand of ca. 10 ha in size (experimental design represented graphically in the supplementary online resource S1). To develop the experiment on predicted effect of seed dropping time on acorn fate, we randomly established 9 locations (GSP determined points) at a minimal distance of 30 m between points. At each location we installed two mobile exclosures (a total of 18 mobile exclosures sized 2.40 × 1.20 m side and 1.20 m height). In each location, one exclosure was designed to exclude livestock and mid-sized acorn predators, but to allow rodents (hereinafter, rodents allowed). This exclosure was made of iron mesh with mid-sized cells (4 × 10 cm). The other exclosure was designed for the additional exclusion of rodents, besides livestock and mid-sized predators (hereinafter, rodents excluded). This exclosure was identical to the previous exclosure, but an additional wire mesh with small cells (0.5 × 0.5 cm) was attached to the former structure. Within each exclosure (i.e., block) six plots (40 × 40 cm side) were established, and six respective treatments were applied (a single replicate of each treatment per exclosure). The treatments consisted of supplying acorns of different categories, combining the two oak species (Q. suber and Q. ilex subsp. ballota), by way of three acorn classes: (i) Early dropped acorns infected by insect larvae (hereinafter, early-infected group); (ii) Early dropped acorns apparently not infected by larvae (hereinafter, early-sound group); (iii) Late dropped acorns apparently sound (hereinafter, late-sound group). Each plot was supplied with 8 acorns. Thus, we used a total of 432 acorns in the rodent allowed exclosures (i.e., 48 acorns per exclosure, half of which were from Q. ilex subsp. ballota, and the rest from Q. suber), and the same standard was applied to rodent excluded exclosures (i.e., a total of 864 acorns in the whole experiment).

Regarding acorn harvesting, the early-dropped acorns were collected as soon as there were enough acorns on the ground (October 10th, 2015), while the late dropped acorns were collected after peak acorn fall (January 20th, 2016). The infected acorns were only included in the experiment when they were abundant in the field (i.e., at the time that early-dropped acorns were collected), based on the temporal pattern that has been described in other studies (Miller and Schlarbaum 2005; Sunyer et al. 2014), and also based on our results from the field survey on acorn infection (see below). All of the acorns that were used in the experiment were collected beneath the canopies of five maternal trees per oak species (i.e., Q. ilex subsp. ballota and Q. suber). These maternal trees were at a minimal distance of 75 m from each other. The collected acorns (ca. 150 acorns per tree) were mixed in a single lot per oak species Acorns were collected at dawn in accordance with the method described by Leiva and Vera (2015). In the study site, the goat herd, conducted by the shepherd, used the stand between ca. 11am and 4pm. Thus, most of the acorns laying on the ground during the daytime are consumed by the goat herd and free-ranging cattle, while the acorns laying on the ground at dawn are most probably fresh acorn, that have fallen during the previous evening and night (Leiva and Vera 2015). Once the acorns were collected, they were moved to the greenhouse facilities of the University of Seville. In the case of the early-dropped acorns, they were tested for infection by insect larvae or apparent non-infection, before the experiment began. For this purpose, the acorns were incubated for five days in several open containers. During this period, the containers were checked daily and the acorns from which larvae emerged and exhibited exit-holes were collected and included in the group of the infected acorns (early-infected group). The acorns that did not exhibit exit-holes were subjected to the float method (Gribko and Jones 1995), selecting those acorns that sank and discarding those acorns that were floating (early-sound group). In the case of the late dropped acorns, they were directly subjected to the float method, selecting the acorns that sank (late-sound group). All of the acorns were kept in containers in the greenhouse for one week, and then they were returned to the field in order to begin the experiment. During the storage period, greenhouse temperature, humidity, and photoperiod were controlled to simulate ambient (exterior) conditions in Seville (October and January for early and late acorns, respectively).

The acorns in different categories were added to the field plots, placed on the ground surface, and kept until winter-end (April 15th, 2016). Then, the fences were opened, and the propagules (acorns at different ontogenetic development) were monitored. We distinguished among them the following three categories: (I) Healthy acorns, those acorns that remained turgid and apparently undamaged. (II) Failed acorns, those acorns that were dried or rotten and remained ungerminated in the plots, mainly within the rodent excluded fences. (III) Predated acorns, those acorns that were heavily gnawed by rodents, plus those acorns that were missing from the plots. Missing acorns were considered to have been predated by rodents because, in the study area, acorn consumption by rodents is overwhelmingly greater than acorn caching (Leiva and Vera 2015). This is a characteristic behaviour in Apodemus sylvaticus (Gómez et al. 2019), constituting the majority mice species in the study area. Missing acorns (almost restricted to the rodent allowed fences) were calculated as the difference between supplied acorns and acorns in the other categories.

The field survey on changes in the relative abundance of acorn infested by insect larvae during the seed dropping season was conducted at the same time as the above-described experiment. For this purpose, we randomly selected eight oak trees per species (Q. suber and Q. ilex subsp. ballota), which exhibited a non-negligible acorn production at the end of summertime, before the beginning of acorn dropping. Then, the fallen acorns laying on the ground were sampled fortnightly from mid-October to mid-January (i.e., seven sampling events) by placing fifteen mobile quadrat grids (50 × 50 cm in size) beneath each canopy. The acorns occurring inside the quadrat grids were stored in plastic bags, individualised by source tree, and were carried out to the greenhouse facilities of the University of Seville. Similarly to the previous case, the acorns from each sampling event were placed in open containers, individualised per source tree, and then incubated for ca. 20 days to allow the insect larvae to exit the acorns. The containers were monitored every 3 days by removing and identifying the emerged larvae, in accordance with the study by Leiva and Fernández-Alés (2005). When the larvae emergence stopped or was very infrequent (ca. three weeks duration), the number of acorns exhibiting exits holes was recorded, and the infection rate per tree was calculated. As in previous case, during the incubation period, greenhouse temperature, humidity, and photoperiod were controlled to simulate ambient (external) conditions in Seville (October – January).

2.3.2. Second year experiment: Realised effect of acorn dropping and sowing time on (i) seedling establishment and related variables, and (ii) on the kinetics of seedling emergence and seedling height increase.

The second-year experiment was developed in the same stand and using similar exclosures as in the previous year (experimental design graphically represented in supplementary online resource S2). In this case, we only used exclosures against livestock and mid-sized acorn predators (i.e., rodent allowed design) that mimic rotational grazing situations, where livestock are temporally excluded from a paddock in order to improve oak regeneration (Pulido et al. 2013). The exclosures were larger (11.5 m2 base area x 1.20 m height) than in the previous experiment because of increased space requirements for seedling development. A total of 5 exclosures were used, placed at 5 respective randomly determined locations that were ca. 30 m apart. Within each exclosure (block) four plots (100 × 220 cm side) were established, and four respective treatments were applied (i.e., one single treatment replicate per exclosure; 4 treatments with 5 replicates per treatment in total). The treatments were similar to treatments of the previous year, and consisted of sowing 40 acorns per plot and growing the seedlings until springtime. In this case, the treatments were a factorial combination of acorns of the two oak species (Q. ilex subsp ballota and Q. suber) by two acorn fall and sowing times, early sowing and late sowing, thus obtaining two cohorts of seedlings (hereinafter, cohort1 and cohort2, respectively). Thus, a total of 800 acorns were sown in the whole experiment, 400 acorns per oak species, with half for cohort1 and the rest for cohort2. In this case, we only used acorns that were apparently sound. Acorn collection and subsequent sorting process in the greenhouse were similar to the previous year, avoiding the incubation step for larval emergence and directly selecting the acorns that passed the floating test (i.e., apparently sound acorns). Then, the acorns were tagged and marked with coloured plastic strips, in accordance with the study by Leiva and Sobrino-Mengual (2022), in order to facilitate their localisation in the field. Field sowing was undertaken one week after acorn collections, on November 10th, 2016, and January 13th, 2017, for cohort1 and cohort2, respectively. The acorns were sown at 25 cm x 31 cm spacing and at 5 cm burial depth. At the time of sowing, the coloured strips were kept unburied, on soil surface, in order to be able to look for the acorns later. The experiment lasted until mid-Springtime (May 2017), the end of the wet season. Then, emerged seedlings were carefully checked and recorded, while for non-emerged seedlings the source acorns were carefully searched for in the soil, in order to determine whether they had been predated or not. The next categories of propagules were distinguished: (1) Unpredated seedlings (i.e., live propagules showing the stem and at least one fully expanded leaf, and without any predated part); (2) Predated seedlings (i.e., as with category 1, but showing any predated part; frequently, this would be the cotyledons); (3) Dead acorns (i.e., dried acorns, either with or without hypocotyl or epicotyl); and (4) Predated acorns (i.e., missing acorns and acorns highly gnawed by rodents). From these categories we also calculated: (5) Total predation (i.e., the sum of categories 2 and 4); and (6) Total seedlings (i.e., sum of categories 1 and 2).

In parallel to the development of the experiment we analysed the kinetics of seedling emergence and height increase. Seedling emergence (number of seedlings emerged per plot) was monitored fortnightly, starting when emergence began and ending at the end of the experiment (4 measurements in total: on March 15th, April 04th, April 26th, and May 17th). Thus, per each replicate and treatment, we also have 4 repeated measures on seedling emergence (100 measurements in total). In addition, the seedling height, a basic indicator for seedling growth, was also measured. For this purpose, we randomly selected a single emerged seedling per plot, tagged it, and measured the shoot height using a caliper. Measurements were repeated 4 times (at the same times described above). Thus, concerning seedling height, we also had 100 measurements in total.

Statistical analysis

The statistic was performed using IBM SPSS Statistic v26 software package. In the first-year experiment, data from the rodent allowed exclosures were analysed independently from the data from the rodent excluded exclosures, in order to disentangle the importance of acorn predation from the importance of acorn failure due to other causes. We used Generalised Linear Mixed Models (GLMMs) for binomial distribution data with the logit as link function for the two response variables (i.e. acorn predation and acorn failure in rodents allowed, and rodent excluded fences, respectively). In these models, exclosure (block) was defined as random effect, while oak species and acorn class were defined as fixed effect. From these data (i.e., predation and failure), the predicted propagule success was derived. It was calculated as the product of non-predated acorns (in the rodent allowed exclosures ) by non-failed acorns (in the rodent excluded fences). These final data on potential success of propagules (response variable) were also analysed using GLMMs, defining normal distribution data and the unit as link function in this case (data normality and homoscedasticity were tested by Shapiro-Wilks and Levene tests, respectively). As in the previous case, exclosure was defined as random effect, while oak species and acorn class were defined as fixed effects in the model.

Similarly to the previous case, data from the second year experiment were analysed using GLMMs. For these response variables, that were measured only once at the end of the experiment (i.e. total seedlings, predated seedlings, etc.; see Table 1), the binomial distribution data and the logit as link function were defined. In these models exclosure (block) was defined as random effect, while species and seedling cohort were defined as fixed effects.

For these data on the kinetic of seedling emergence and height increase, the GLMMs for repeated measures were used. Concerning cumulated seedling emergence (response variable), binomial distribution and logic as link functions were defined. In addition, exclosure (block) was defined as random effect, species and cohort were defined as fixed effects, and time (sampling dates) was included as repeated effect in this model. Concerning seedling height (response variable), normal distribution data and unit as link function were defined after testing data normality and homoscedasticity (Shapiro-Wilks and Levene tests, respectively). As in the previous case, the exclosure (block) was defined as random effect, species and cohort as fixed effects and time (sampling date) was included as repeated effect in this model.

Results

First year studies

The results of the field survey on the kinetics of acorn infection by insect larvae showed that the weevil Curculio elephas and the moth Cydia fagiglandana were the carpophagous larvae that emerged from the acorns. The infection rate ranged from 2.5 to 16% in Q. suber and from 0.8 to 5% in Q. ilex subsp. ballota (Fig. 2), being higher in the first half of the acorn dropping season, and decreasing onwards. This temporal pattern was similar in the two oak species, although Q. suber reached higher infection rate than Q. ilex subsp. ballota, especially in the first half of the sampling period.

Fig. 2
figure 2

Percentage of acorns infected by carpophagous insect larvae (mean values ± se bars) in acorns of Q. ilex subsp ballota and Q. suber, collected in the experimental location during the first year experiment. (M = mid; E = early; Oct., Nov., Dec., Jan. = October, November, December, January, respectively)

Full size image

Regarding the experiment on predicted effect of acorn dropping time on acorn fate, acorn predation, in the rodent allowed exclosures, exhibited the same pattern across treatments in the two oak species (Fig. 3a); the predation rate increased from early-infected acorns to late-sound acorns, being low to intermediate in early-sound acorns. Thus, concerning dropping time, predation was ca. double in the late-dropped and sound acorns (43 and 46% predation in Q. ilex subsp. ballota and Q. suber, respectively) than in the early-dropped and sound acorns (21 and 19% predation in Q. ilex subsp. ballota and Q. suber, respectively). On their side, larvae infection decreased acorn predation by rodents relative to the sound acorns (15 and 7% predation in Q. ilex subsp. ballota and Q. suber, respectively, for early-infected acorns, and 21 and 19% predation in Q. ilex subsp. ballota and Q. suber respectively for early-sound acorns). The GLMM for propagule predation (Table 1) indicated a non-significant effect of the oak species, but highly significant effect of the acorn class (while exclosure i.e., block, has a non-significant effect). The model coefficients were high and significant for the late-sound acorns, both in Q. ilex subsp. ballota and in Q. suber, which experienced much more predation than the early-sound acorns (i.e., the reference category), while the model coefficients were low and non-significant (in the case of Q. ilex subsp ballota) for the early-infected acorns; thus, indicating small differences in predation between this acorn class and the early-sound acorns (i.e., the reference category).

Table 1 First year experiment. Statistical results (GLMMs) for two response variables measured under different types of exclosure (A and B) and estimated variable (C). Significance of the effects (left-side part of the table) and model coefficients for different acorn classes (right-side part of the table). Ref. = reference category, Q. ilex = Q. ilex subsp. ballota. *** = P < 0.001; ** = P < 0.01; ns = non-significant
Full size table
Fig. 3
figure 3

First year experiment. Percentages (mean values ± se bars) of propagules (acorns in different stages) (relative to supplied acorns) that: were predated in locations allowed to rodents (a), failed in development (generally dried or rotten acorns) in locations excluded against rodents (b), potentially succeeded (product of non-predated propagules and non-failed propagules) (c)

Full size image

Acorn failure under rodent exclusion (Fig. 3b) exhibited the opposite pattern among the same treatments as acorn predation (Fig. 3a). Thus, failure rates were higher in the early-dropped acorns, either infected or sound, than in the late-dropped and sound acorns. This pattern was also consistent in the two oak species. However, Q. ilex subsp. ballota experienced ca. double the failure rates (25 to 69% failure) of Q. suber (10 to 27% failure). Generalized Linear Mixed Models indicated significant differences between species as well as between acorn classes, while exclosure (block) effect was non-significant (Table1). Model coefficients were high (absolute value) and significant for the late-sound acorns, indicating lower failure in this class than in the early-sound acorns (i.e., the reference category). In opposition, the model coefficients were non-significant for the early-infected acorns, which did not differ in failure rate from the early-sound acorns (the reference category).

The potential propagules success, calculated as the product of non-predated and non-failed acorns, ranged from low (9% success) to medium values (40% success) in different treatments (Fig. 3c), with the late-sound acorns exhibiting the highest potential success, both in Q. ilex subsp. ballota and in Q. suber. Generalized Linear Mixed Models indicated non-significant differences between species (Table 1), but significant differences among acorn classes (block effect, i.e. exlosure, was also non-significant in this case). The model coefficients were only significant for the late-sound acorns, which differed in potential success from the early-sound acorns (i.e., the reference category), while the model coefficients were non-significant for the early-infected acorns.

Second year experiment

Realised effect of acorn drop and sowing time on seedling establishment success.

Overall, 41 to 54% sown acorns were successfully established as seedlings at the end of the second-year experiment (Fig. 4a), with higher seedling establishment in cohort2 than in cohort1. A Generalized Linear Mixed Model (Table 2) indicated non-significant differences between the two oak species, but significant differences between the two cohorts. However, model coefficients for seedling cohort was only significant for Q. suber (block effect, i.e. exclosure, was non-significant for total seedlings, nor was it significant for any other response variable in Table2). Concerning predated seedlings (Fig. 4b), most of them exhibited total or partial removal of the cotyledons attached to the stems, while a few seedlings also exhibited clipped tips. The rest of the seedling shoots were not predated. The average number of predated seedlings varied between 6 and 30% in different treatments, whereas the cohort2 seedlings were much more predated than the cohort1 seedlings in the two oak species. Results from GLMM indicated significant differences between the cohorts, but not between the oak species (Table 2). Model coefficients were significant and very high (absolute value) for cohort2, indicating important differences relative to cohort1 (i.e., the reference group). The opposite pattern was found concerning the non-predated seedlings (Fig. 4c), with a higher percentage of these seedlings in cohort1 than in cohort 2. As in the previous case, results from GLMM indicated significant differences between the two cohorts, but non-significant differences between the two oak species (Table 2). The model coefficients for cohort was only significant in Q. ilex subsp. ballota because in this species the non-predated seedlings in cohort2 were ca. half of those in cohort1 (i.e., the reference category).

Fig. 4
figure 4

Second year experiment. Percentages of propagules (i.e., acorns, developped seedlings) (relative to sown acorns) in the early and late cohorts (Coh1 and Coh2, respectively) in the two oak species, at the end of the experiment (mean values ± se bars). Total seedlings (a) refers the sum of sound seedlings (c) plus predated seedlings (d). Total predation (b) is the sum of predated seedlings (d) plus predated acorns (f). Q. ilex = Quercus ilex subsp. ballota

Full size image
Table 2 Second year experiment. Statistical results (GLMMs) for different response variables. Significance of the effects (left-side part of the table) and model coefficients for different seedling cohorts (right-side part of the table). Coh = cohort; ref = reference category; Q. ilex = Q. ilex subsp. ballota; *** = P < 0.001; ** = P < 0.01; *=P < 0.05; ns = non-significant
Full size table

With reference to the propagules that did not emerge as seedlings, the results on the dead acorns, predated acorns, and total predation (Fig. 4d, e, f) were very similar to the results from the first year experiment (Fig. 3a, b). Thus, the percentage of dead acorns was lower, and the percentage of predated acorns and the total predation were higher in cohort2 than in cohort1. In addition, Q. ilex subsp. ballota experienced higher predation than Q. suber, in terms of total predation (Fig. 4e, f). Results from GLMMs (Table2) indicated significant differences between species (in predated acorns and total predation) and between cohorts in the three variables (i.e., dead acorns, predated acorns, total predation). The Model coefficients for cohort2 exhibited the highest significant absolute values for Q. ilex subsp. ballota, indicating important differences between the two cohorts in the number of predated acorns and in total predation in this species, compared to Q. suber.

Kinetics of seedling emergence and seedling height increase

The kinetics of seedling emergence was quite different between the two cohorts, as well as between the two oak species (Fig. 5a, b). In cohort1 about a quarter of sown acorns had emerged as seedlings at the beginning of the sampling period (24.8 and 28% emergence in Q. ilex subsp. ballota and Q. suber, respectively, in mid-March) while a few seedlings had emerged in cohort2 at the same time (6.5 and 11.5% emergence in Q. ilex subsp. ballota and Q. suber, respectively). Then, seedling emergence increased at a fast rate in cohort2 and a slow rate in cohort1 (from early April to mid-May), resulting in similar final emergence between cohorts in Q. ilex subsp. ballota and in higher final emergence in cohort2, compared to cohort1 in Q. suber (Fig. 5a, b). Results from GLMM on cumulated seedlings emergence (Table 3) indicated a non-significant block effect (i.e., exclosure), but significant differences between the two oak species. The cohort and the time effects were also highly significant. Model coefficients for different sampling dates indicated this differential temporal pattern of seedling emergence between the two cohorts (Table 3). Thus, in cohort2 most coefficients were significant, high (absolute values) and negative, indicating lower cumulated emergence at each sampling date than at the end of the study (i.e., mid-May; the reference category). In cohort1 the model coefficients were generally low (absolute values) and most of them were non-significant after mid-March, indicating early stabilisation of seedling emergence in this cohort. This pattern was noted in the two oak species although it was more pronounced in Q. suber.

Table 3 Second year experiment. Statistical results (GLMM) for different effects on cumulative seedlings emergence (left-side part of the table) and model coefficients for different dates (right-side part of the table), May 17 is the reference category (ref.). Coh = cohort; *** = P < 0.001; ** = P < 0.01; *=P < 0.05; ns = non-significant
Full size table
Fig. 5
figure 5

Second year experiment. Cumulated number of emerged seedlings (mean values ± se bars) in cohort1 and cohort2 in Q. ilex subsp. ballota (a) and Q. suber (b) measured four times during springtime 2018

Full size image

Concerning seedling height (Fig. 6a, b), the observed kinetics was similar to that of seedling emergence. Seedling height was notably higher (i.e., ca. 4 cm higher) in cohort1 than in cohort2 at the beginning of the sampling period (i.e., mid-March), both in Q. ilex subsp. ballota and Q. suber. However, these differences decreased over time and the two cohorts exhibited similar seedling height at the end of the experiment (i.e., mid-May). Results from GLMM (Table 4) indicated a non-significant effect of exclosure (i.e., block) nor of the oak species, but highly significant cohort and time effects. The Model coefficients for seedling height at different sampling dates were in general high (absolute value) in cohort2. In this cohort they were significant for longer (early and late April) than in cohort1 (early April). This indicate a recurrent increase in seedling height at different dates, relative to the end of the experiment (i.e., mid-May; the reference category) that was more noticeable in cohort2. In cohort1 the model coefficients were low (absolute values) or non-significant in many cases, in the two oak species, indicating earlier stabilisation of seedling height in this cohort.

Table 4 Second year experiment. Statistical results (GLMM) for different effects on seedling height (left-side part of the table) and model coefficients for different dates (right part of the table), May 17 is the reference category (ref.). Coh = cohort ; *** = P < 0.001; ** = P < 0.01; *=P < 0.05; ns = non-significant
Full size table
Fig. 6
figure 6

Second year experiment. Seedlings height (mean values ± se bars) in Q. ilex subsp. ballota (a) and Q. suber (b) in the two seedling cohorts measured four times during springtime 2018

Full size image

Discussion

The factors assessed in this study, i.e., acorn dropping and sowing time, or time of acorn addition to soil surface, as well as acorn infection by insect larvae, showed significant effects on different processes undergone by propagules of Q. ilex subsp. ballota and Q. suber, such as acorn and seedling predation by rodents and propagule failure in ontogenetic development. However, Q. suber and Q. ilex subsp. ballota differed in their response to acorn dropping and sowing time, or time of acorn addition to soil surface, in terms of seedling establishment, with potential implications for the two species to withstand global warming.

Kinetics of acorn infection by insect larvae and its effect on potential propagule success.

One of the aims of this study was to assess the temporal changes in acorn infection by insect larvae in our study site, and to compare them with the pattern found in other studies on temperate oak species. In those other studies, the infection rate of the fallen acorns is highest at the beginning of the acorn-dropping season and decreases onwards (Miller and Schlarbaum 2005; Sunyer et al. 2014), which is likely due to premature abscission of the infected acorns. The oviposition by adult insects induces sap outflow under the cupule, producing partial acorn dehydration and falling off (Yi and Yang 2010; Canelo et al. 2021). Our results from the field survey during the first-year study (Fig. 2) corroborate the same temporal pattern of acorn infection by insects, both in Q. ilex subsp. ballota and in Q. suber, although the infection rates were moderate in Q. suber and low in Q. ilex subsp. ballota.

Our results also indicated that the acorn infection by insect larvae conditions acorn predation by rodents, with significant decrease of rodent predation on infected acorns compared to sound acorns (Fig. 3; Table 2). This tendency was the same in the two oak species (Fig. 3a) although it was more conspicuous in Q. suber, the only species in which the model coefficient was significant (Table 2). A low preference of mice for the infected acorns agrees with studies on several oak species, including Q. ilex subsp. ballota and Q. pubescens (Sunyer et al. 2014), and is likely related to decreased food quality and quantity of the damaged endosperm (Hou et al. 2010; Sone et al. 2016). However, this pattern can be reversed (i.e., highest rodent predation on the infected acorns), whether or not insect larvae remain inside the acorns (Perea et al. 2012).

Importantly, infection by insect larvae of early-dropped acorns had no significant effect on propagule failure (i.e., non-significant model coefficients for early-infected acorns, Table1) nor on potential propagule success (Fig. 3c; Table1). The effects of the attack of carpophagous insects on acorn germination are very heterogeneous, depending on factors such as the oak species, the acorn size, or the particular acorn-end where the larvae develop (Hou et al. 2010; Sone et al. 2016). However, it is noteworthy that in our study, the non-infected and early-dropped acorns (i.e., early-sound class) already exhibited low to medium success (66 and 27% failure in Q. ilex subsp. ballota and Q. suber, respectively; Fig. 3b) which contributes to explaining the low effect of acorn infection on propagule failure. High acorn germination failure at early acorn falling has also been found in studies on Q. leucotrichophora in the Central Himalaya region in Asia (Tewari et al. 2017). In summary, under the conditions of this study, an inherently high failure rate of the early-fallen acorns, combined with the fact that these acorns suffer the highest infection by insect larvae (Fig. 2) resulted in the moderate importance of larval infection for the developmental success of propagules of Q. ilex subsp. ballota and Q. suber.

Predicted propagule success and realised seedling establishment in early and late fallen acorns

A major focus of this study was to understand whether seed dropping time has any significant effect on factors such as post-dispersal acorn predation by rodents, intrinsic failure rate in propagule development, and the resulting success or interruption of seedling establishment. Concerning these questions, we found very consistent results from both the first-year predictive experiment and the second-year experiment, so that the two experimental approaches proved to be useful and complementary. Overall, the late-dropped sound acorns were significantly more predated by rodents than the early-dropped sound acorns in the first-year experiment (Fig. 3a; Table1), as well as the fact that the cohort2 seedlings were significantly more predated by rodents (i.e., total predation, acorn and seedling predation; Fig. 4b, e, f; Table2) in the second year experiment. These results agree with a previous study on acorns of Q. ilex subsp. ballota, in which a higher predation by rodents was found at the end of, rather than before, the acorn-fall season (Leiva and Diaz-Maqueda 2016). This pattern is also consistent with studies reporting the maximal rodent population density during late-autumntime and wintertime in some Mediterranean areas (Moreno and Kufner 1988). In addition, Moran-López et al. (2015) found a temporal decoupling between the peak of holm-oak acorn fall and the peak of rodent predatory activity in Central Spain, especially in the southernmost oak populations.

Despite the above results, the negative impact of increased rodent predation on late-dropped acorns was overcompensated by the higher inherent capacity for ontogenetic development (i.e., low failure rate) of these late-dropped acorns, compared with the early-dropped acorns (Fig. 3b; Table1). As a result, the late-dropped acorns exhibited the highest potential success in the first-year experiment (Fig. 2c; Table 2), and the highest number of established seedlings at the end of the second-year experiment (i.e., total seedlings in cohort2; Fig. 4a). Thus, in line with the results on predispersal predation (i.e., acorn infection by insects, Fig. 3), in this experiment acorn properties appear as a major driver for seedling establishment over predation by rodents.

Kinetics of seedling emergence and height increase in different cohorts

As expected, seedling emergence and seedling height, in the second-year experiment, were initially (i.e., mid-March) higher in the seedlings from early-dropped and sown acorns (i.e., cohort1 seedling) than in the cohort2 seedlings, both in Q. ilex subsp. ballota and in Q. suber (Figs. 5 and 6). However, contrary to expectations, the early dropping and sowing of the acorns did not result in a higher number of emerged seedlings, nor in higher seedling height, at the end of the experiment (i.e., mid-May) (Figs. 5 and 6; Tables 3 and 4). Indeed, the opposite was found to be true for final seedlings emergence in Q. suber, whereby emerged seedlings in cohort2 overcompensated the emerged seedlings in cohort1 (Fig. 5b; Table 3). Thus, the cohort2 seedlings experienced a faster emergence and height increase than the cohort1 seedlings, especially in Q. suber.

These differential kinetics between the two seedlings cohorts could be related to physiological dormancy (PD), a mechanism that affects seeds germination in many plant species (Merouani et al. 2001), and implies the need of cold stratification for quick and uniform germination. In the Quercus genus, PD may affect any one of root development (i.e., hypocotyl PD), shoot development (i.e., epicotyl PD) or both, roots and shoots (Sun et al. 2021). The Quercus species frequently undergo epicotyl PD but not hypocotyl PD, i.e., the hypocotyl germinates and elongates quickly without cold stratification, while the epicotyl requires cold stratification for quick elongation. Species described as exhibiting this PD pattern include Q. suber as well as Q. petraea (Matt.) Liebl., Q. prinus L., Q. robur L., or Q. alba L. (Xia et al. 2015; Sun et al. 2021). When these species germinate under relatively warm temperatures, they show increased epicotyl dormancy and delayed shoot growth compared to the germination under colder temperatures (Farmer 1977). This was the process that presumably took place in our cohort1 and cohort2 seedlings, especially in Q. suber but also in Q. ilex subsp. ballota, as the temperature was warmer at early acorn sowing (20 ºC mean maximum temperature and 29 ºC highest temperature) and it was colder at late acorn sowing (15 ºC mean maximum temperature and 20 ºC highest temperature) (Fig. 1b). In addition, the lowest temperature was below freezing (-2 ºC) at late-acorn sowing, while it was above freezing (+ 3 ºC) at early-acorn sowing. These results were similar to the temperature changes observed in the first year study (Fig. 1a), and reflect the natural temperature transition from autumntime to wintertime. In summary, colder conditions at cohort2 acorn sowing are likely to be related to higher cohort2 seedling emergence in Q. suber and to faster seedling height gain in the two oak species. Q. ilex subsp. ballota has been described as not exhibiting hypocotyl PD while the information on epicotyl PD in this species is insufficient (Sun et al. 2021). The results of the present study suggest that epicotyl PD is also a likely mechanism in Q. ilex subsp. ballota.

According to the above results, the role of acorn-dropping time and associated temperature changes in determining the availability of established seedlings in late springtime, (i.e., before the onset of the Mediterranean summer-drought) is species-specific. Q. suber seedlings establishment benefitted from late-acorn dropping and sowing time, while Q. ilex subsp. ballota seedlings establishment was indifferent to this factor. Conversely, final seedling height was non-significantly affected by acorn-dropping and sowing time, in either Q. ilex subsp. ballota or Q. suber (Fig. 6; Table 4).

It is noteworthy that we have not included in this study the belowground component of the seedlings, which is crucial for seedlings in withstanding the summertime-drought in the Mediterranean region (Morilla et al. 2023). The lack of hypocotyl PD, both in Q. suber and Q. ilex subsp. ballota (Xia et al. 2015; Sun et al. 2021), and the fact that the temperature differentially affects roots and shoots growth (Farmer 1977; Reich et al. 1980; Teskey and Hinckley 1981) suggests that early-acorn dropping could provide seedlings with more developed roots and better access to deep soil water. Therefore, there is a need to focus on these effects in more detail, in future studies.

Implications of the results in the context of climate change

Temperature rise and rainfall decrease are among the major predicted climate changes in the European Mediterranean region, where rainfall irregularity and extreme climatic events are also expected to increase (IPCC 2021). The aridity increases undoubtedly represent a strong stressor for survival of the young oak seedlings, which already are very vulnerable to the summertime-drought characteristic of the Mediterranean climate (Ramírez-Valiente et al. 2018; Morilla et al. 2023). However, our results on the decreased emergence in Q. suber seedlings when the acorns were sown under the warmer autumntime conditions, suggest that temperature warming and the consequent lack of cold stratification will negatively affect the Quercus species, in its exhibiting physiological dormancy, and could contribute to differential regeneration success between Q. suber and Q. ilex subsp. ballota. In addition, extreme climatic events and rainfall irregularity will certainly affect these processes in a direction that is difficult to predict.

Conclusions

According to this study, the developmental success of propagules of Q. ilex subsp. ballota and Q. suber, and seedlings establishment in springtime, were dependent on acorn–falling phenology and were species-specific. This was presumably due to the temperature conditions at acorn-dropping and sowing time, which promoted, or otherwise, cold stratification. Additionally, acorn maturity also likely contributes to phenological differences in propagule development and seedling establishment. Notwithstanding, some of the major pre- and post-dispersal acorns and seedling predators (weevil and moth larvae and mice, respectively) had a relatively low effect on propagule success under the conditions of this study. However, this assessment could be modified, depending on potential climate-induced changes in the population dynamic of predators, and the direction of these changes in each specific region (Jiang et al. 2011; Čepelka et al. 2020).