Our study was performed in Central Spain, in the Autonomous Community of Castilla y León, the third European region in size and one of the most strongly damaged by wildfires. According to the European Forest Fire Information System (EFFIS), 1996 fires occurred in the region during 2008 affecting 152.64 km² (Quintano et al., 2011). The fire season occurs during the period of June-September, corresponding to the warm to hot and dry summer, typical of Mediterranean climate.

The study was carried out in Honrubia de la Cuesta (northern part of Segovia province), which is a Mediterranean ecosystem dominated by Pinus pinaster plantations established by the Spanish Forest Services in previously deforested areas (440901-443169 longitude-UTM, 4592 704-4590583 latitude-UTM, 750-880 m.a.s.l.). Here, a large wildfire burned 1200 ha of forest and canopies in August 2008 (Figure 1), time at which pine trees were about 40 years old. This site has a supra-Mediterranean climate with 3 months of dry season in the summer, a mean annual rainfall of 480-500 mm and mean temperatures ranging from 8 to 13 °C. The warmest month is July and the coldest January. These data were provided by the closest meteorological station (Linares del Arroyo 41º 31´ 40´´ N, 3º 32´ 72´´ W) located 15 km from the study area.

This area is composed of Paleozoic metamorphic rocks, dominated by Ordovician and Silurian shales (Barrenechea & Rodas, 1992). The soil is classified as Inceptisol suborder Xerept (Alvarez et al.,1993 ).

In the forest understory, sparse individuals of Cytisus scoparius (L.) Link, Quercus faginea Lam and Quercus ilex (L.) ssp. ballota (Desf.) Samp. were found. A number of species have been identified during mushroom forays in the area (pers. obs.). Ectomycorrhizal Collybia sp., Cortinarius cinnamomeus (L.) Fr., Cystoderma amianthinum (Scop.) Fayod, Hebeloma mesophaeum (Pers.) Quél., Inocybe sp., Laccaria laccata (Scop.) Cooke, Rhizopogon luteolus Fr., Tricholoma scalpturatum (Fr.) Quél., and saprophytic Astraeus hygrometricus, Baeospora myosura (Fr.) Singer, Collybia butyracea (Bull.) P. Kumm., Hemimycena sp., Hygrophorus gliocyclus Fr., Mycena pura (Pers.) P. Kumm., Mycena pura ssp. lutea (Gillet) Arnolds, were observed at sites not affected by fire. While Galerina sp., Gerronema sp., Omphalina sp., Pholiota carbonaria (Fr.) Sing., were found in places moderately affected by fire and only pyrophitic Pholiota carbonaria was found at those sites fully impacted by the fire. No ECM species were found to form sporocarps in areas affected by fire.

Three different sites were chosen within our study area according to the degree of damage caused by fire on vegetation and soil (Figure 1). Fire severity and soil damage were classified following Rincón & Pueyo (2010) and Vega et al. 2013 criteria: control, unburned site (hereafter UB) was established in an adjacent P. pinaster forest unaffected by fire at least in the preceding 40 years (dominant trees of approximately 40-cm diameter). Moderate fire severity site (hereafter MFS) had all pine crowns and upper barks burned. Here, the soil organic matter was not consumed and the ground surface remained intact after the fire. The soil was darkened and water repellent. The high fire severity site (hereafter HFS) had pines, canopy and understory litter totally burned and the entire humic soil organic layer consumed For HFS, the loss of soil structure was very evident and the rootlets were consumed. We devoted our greatest effort to select sites differing in fire severity but similar in terms of vegetation and local topography, aiming to avoid obvious major variation between sampling sites (Dias et al., 2010). That is, we aimed to select sites such that observed differences between them could be linked to differences in fire intensity.

In mid-June 2009, uniform areas were sampled within the three above mentioned sites. A total of 15 intact soil blocks (22 x 22 x 20 cm), five per site, were extracted randomly with a minimum spacing of 100 m with a metal cube with a sharpened edge at HSF, MSF and adjacent UB sites. This minimum distance between sampling points was chosen in order to ensure low autocorrelation and provide estimates as independent from each other as possible (Lilleskov et al., 2004, Dias et al., 2010, Buscardo et al., 2015). Soil blocks were then placed into square plastic containers of the same dimensions and taken to the greenhouse facilities of the Forest Engineering Faculty, Palencia Campus, University of Valladolid. Nine evenly spaced soil samples, three from each of the previously sampled sites were also extracted using a cylindrical (2 cm radius, 20 cm deep, 250 cm³) soil borer (Taylor, 2002) for chemical analysis. The edaphic parameters analyzed included pH, organic matter, total nitrogen, phosphorous and potassium (N–P–K), sodium, magnesium, calcium, cation exchange capacity (CEC) and conductivity (Cond) (see details in Table 1).

Greenhouses were naturally lit, with controlled temperature (15-20°C) and humidity, maintaining the natural soil at field capacity. P. pinaster seeds (375 in total) were provided by a local nursery (Viveros Fuente Amarga, Cabezón de Pisuerga, Valladolid, Spain), these were surface sterilized with 30% hydrogen peroxide for 30 min and washed tree times with sterile water. Twenty five seeds were sown in each container following a square grid, ensuring a homogenous distribution. Pots were tap-watered daily and no nutrients were added. Four months after sowing, five randomly selected seedlings from each pot (75 in total) were extracted from the center of each container thus avoiding border effects.

After extraction from the containers, each plant was rinsed with tap water in a plastic tray and gently shaken to soften and remove adhering soil. Tap root length and shoot length were measured with a ruler to the nearest millimeter (Palfner et al., 2008). The root system of each plant was then divided into two vertical sections, from 0 to10 cm and from 10 to 20 cm from the plant root collar (Anderson et al., 2007).

All root tips were classified morphologically as mycorrhizal vs. non-mycorrhizal or “non vital” (Agerer, 1991, 1987-2002, de Román & de Miguel, 2005, Scattolin et al., 2008), using a Leica M3 dissecting microscope and a 15x magnification stereomicroscope. The percentage of mycorrhization per site, overall and at each of the two vertical sections, was calculated by dividing the number of mycorrhizal root tips by the total number of root tips (Brundrett et al., 1996). Shoot and root dry biomass were determined after drying the plant material at 70ºC for 48 h. in a drying oven (Sousa et al., 2011). Root/shoot ratio was derived from these measurements (Palfner et al., 2008).

One and two-way repeated measures ANOVA were applied to explore the influence that different fire damage levels (UB, MFS and HFS) and root section parts (upper and lower) had on the response variables (morphometric traits, mycorrhizal colonization and soil chemical characteristics). Non-independence of errors between seedlings grown in the same pot was accounted within the model. Post-hoc tests (P>0.05) (LSD for plant morphometric traits and percentage of mycorrhization and Tukey test for soil chemical variables) were applied to explore the differences between fire intensity levels. Normality of data was checked with a Kolmogorov-Smirnov test and we applied log transformations when necessary, namely for some soil nutrients. We used STATISTICA ’08 Edition software to perform these statistical analyses (StatSoft Inc. 1984-2008).

Finally, we carried out a DCA (Detrended Correspondent Analysis) with our complete dataset. The longest gradient length, indicative of how heterogeneous our data are, was below 3.0 standard deviations. Then, a Principal Component Analysis (PCA) was performed based on Pearson product-moment correlation coefficients (R Development Core Team, 2011). Fire severity parameters were coded as 1 (UB), 2 (MSF) and 3 (HSF). Only those variables showing a relatively strong relationship to the first two PCA axes (vector length greater than 0.5 units) are shown.

Mortality rates were low and did not affect more than three plants per container. Seedlings grown in high severity fire soils (HSF) had heavier and longer shoots (P < 0.01 and P=0.030, respectively; Figure 2) and longer roots (P=0.004) than those grown in moderate severity fire (MSF) and unburned (UB) soils (Figure 2). In turn, root length from MSF sites was also longer than that from UB sites (P=0.042). Root/shoot ratio was highest for HSF and MSF sites and lowest for UB samples (HSF vs MSF P=0.798; HSF vs UB P=0.0000; MSF vs UB P=0.0001) (Figure 2).

The proportion of total mycorrhization from both post-fire sites was significantly lower than that from undisturbed soils (HSF, P=0.006; MSF P=0.005). Regarding the two vertical sections of the root system, higher levels of mycorrhization (P<0.01) were observed in the upper (0-10cm) section compared to the lower one (>10 cm) for all three sites (Figure 3). At the upper root section, we observed a higher colonization rate in UB sites when compared to MSF (P=0.012) and HSF (P=0.044) sites (Figure 3). The same trend was followed by colonization in the lower root section, but in this case no differences were observed between seedlings from MSF soils compared to the other two sites. Regarding soil chemical characteristics, the main differences between sites were a lower N, K and organic matter content and a higher pH and P content in burnt soils (Table 1).

The PCA grouped samples from HFS and MFS sites in the positive area of Axis 1 whereas UB ones were located in the negative area (Figure 4). The two axes explained 68.85% of the variation present in the samples (48.47% axis 1 and 20.38% axis 2). Unburned soil samples appeared associated with higher values of organic matter, N, Ca and cation exchange capacity (CEC), but lower values of pH, Mg, Na and conductivity. The opposite was true for both burned (HFS and MFS) sites (Table 1, Table 2). Regarding seedling vegetative traits, the PCA revealed a positive correlation between shoot biomass with P and fire severity. Axis 1 also reflected a relation between higher levels of mycorrhization (total, 0-10 cm and 10-20 cm) with the unburned sites.

Seedlings in the high severity fire soil showed higher shoot height, root length and root/shoot ratio and also more than 10% of the shoot biomass of those growing in unburned soil. Similar biometric results have been found by Pausas et al. (2003) in Pinus halepensis Mill. seedlings in eastern Iberian Peninsula under three fire severity classes, where seedlings from sites most affected by fire grew significantly more. Those differences may lie in the changes in soil chemistry caused by different fire severities and leading to higher fertility (see below) (Pausas et al., 2003).

Maximum ECM colonization rate was found in soils from unburned sites, although it did not vary significantly among high and moderate severity fire sites. Our results are in agreement with a large body of research on Mediterranean dry forests (de Román & Miguel, 2005, Martín-Pinto et al., 2006b, Buscardo et al., 2010, Hernández-Rodríguez et al., 2013) and specifically with some studies conducted also with P. pinaster (Buscardo et al., 2011, Gassibe et al., 2011, Sousa et al., 2011). Nonetheless, the overall evidence in this field cannot be considered as conclusive yet. For example, Rincón & Pueyo (2010) found that fungal richness and colonization of P. pinaster seedlings did not depend on fire severity, but on time elapsed after fire. Indeed, time span after fire seems to mediate the evolution of fungal communities, from fire-adapted taxa to later-stage ones (Rincón et al. 2014).

Regarding the relationship between vertical root distribution (upper and lower sections) and mycorrhization rate, our results revealed how the upper part of the roots had the highest ECM rates across all fire damage levels (Visser, 1995, Torres & Honrubia, 1997). In fact, differences between upper and lower root parts across fire damage levels were kept constant (Figure 3). This was true also in the site most severely affected by fire, where higher colonization rates would have been expected at the lower (deeper) root parts. This could have come as a result of higher exposure of the top layer to very high temperatures and buffering of the lower layer (Bastias et al., 2006, Kipfer et al., 2010). Our results might indicate a particularly high resistance of some kind of propagules to fire. Here, molecular taxonomy techniques can reveal the existence of interesting pyrophitic taxa (Rincón et al. 2014).

Our experimental design did not allow separating root physiological or anatomical effects from those derived from propagule abundance or soil properties at the two depth levels within sites. For example, seedling root distribution and length varies according to soil depth (Wallander et al., 2004), in this sense post-fire seedlings tend to extend their root system vertically as seen in our results by increasing mainly tap-root length therefore also changing the distribution and structure of the lateral roots (Palfner et al., 2008). Also the mycelium of ECM fungi which is usually most abundant in the superficial organic soil layers in undisturbed ecosystems (Visser, 1995, Neville et al., 2002, Wallander et al., 2004) may influence root morphology and architecture through the formation of short lateral roots and root tips (Ostonen et al., 2009, Kubisch et al., 2015), therefore affecting seedling growth (Jones et al., 2003). There is also evidence of the stratification of fungal communities (Dickie et al., 2002, Rosling etal., 2003; Anderson etal., 2007) between the 0–10 and 10–20 cm sections of soil profile, but is not always the case as reported by Anderson etal. (2007) where non stratified homogeneous ECM communities were present within the 20 cm of soil depth after two years after fire.

Nonetheless, data from our unburned site provide a suitable baseline against which we can compare the two other scenarios. The higher mycorrhization rate in the top layer of UB plant roots might be related to either root physiology or propagule abundance. Lower mycorrhization rates in MFS and HFS in that same top layer compared to UB plants, is likely due to different soil properties. Under natural conditions, mycorrhization rates in upper layers could be even higher due to mechanisms of fungal colonization such as spore dispersion by wind or rodents, which were suppressed in our experiment. Also, the same rationale applies to the lower layer. Studies similar to ours have also found lower fungal richness (Smith etal., 2004) or different fungal community structures (Bastias et al., 2006) in the upper 10 cm layer of soils variously affected by fire but interestingly, no trend at a depth of 10–20 cm (Bastias et al., 2006).

Considering mycorrhization rates alone, our P. pinaster seedlings attained smaller sizes in those soils where mycorrhization was highest i.e. in unburned soils. Nonetheless, given that potential mycorrhizal inoculum was confounded with soil chemistry, direct conclusions cannot be drawn. Inclusion in our study of seedlings grown on sterilized soils from the three studied sites, would have allowed measuring the effect of soil chemistry alone on plant growth. Notwithstanding, a negative effect of mycorrhization on plant growth rates under nursery conditions has been previously reported (Stenström et al., 1985, 1990, Le Tacon et al., 1992), even though global evidence does support the beneficial effect of ECM species on P. pinaster plant growth (Sousa et al., 2012).