Seasonal variation in chemistry, but not morphology, in roots of Quercus robur growing in different soil types

Patterns of root traits among different root orders and their variation across seasons are of considerable importance for soil resource acquisition and partitioning in forest ecosystems. We evaluated whether morphological, anatomical and biochemical traits varied among root orders of Quercus robur (L.) sampled across spring, summer and fall seasons and growing in two different soil types with contrasting site fertility. We found no consistent differences in root diameter and specific root length in relation to soil type or growing season. There was, however, a strong seasonal variation in patterns of nitrogen (N) concentration among root orders. During spring and summer, N concentration was highest in the most distal, absorptive portion of the root system. At the end of the growing season, we observed a sharp decline in the N concentration of these lower-order, absorptive roots and an increase in N concentration of the higher-order, transport roots. The specific mechanisms driving the seasonally changing N concentration remain unclear but are likely related to different functions of lower-order roots for absorption and higher-order roots for structure and storage. Future work should identify how common the observed seasonal changes in N concentration are across species and determine what specific environmental cues plants or roots use to trigger shifts in resource allocation within the root branching hierarchy.


Introduction
Fine-root systems are highly diverse, adaptive structures that supply the plant with growth-limiting soil resources (e.g., nitrogen (N), phosphorus and water) and must cope with frequent changes in their surrounding environment. Throughout a single growing season, fine roots respond to wide variations in external temperature and moisture as well as internal changes in the availability and allocation of plant resources. In early spring, nutrient acquisition is often restricted due to low soil temperatures or lack of new carbohydrates to support the N assimilation process, while at the same time bud break marks a sharp turning point with increasing plant uptake demand and, eventually, influx of new photosynthate ( Taylor 1967cited after Chapin et al. 1990, Fagerström and Lohm 1977. Later, when intensive tree growth (leaf and stem extension, bole wood) is reduced during summer, roots must continue to function in increasingly variable conditions, which may include periodic drought or water-logging, while plant nutrient and carbohydrate supply is split between maintenance respiration and storage (Chapin et al. 1990). Finally, as trees prepare to overwinter and leaves senesce, carbohydrate accumulation, as a means of frost protection, or carbohydrate depletion with retranslocation to more proximal storage organs may be observed in fine roots during (Aerts 1996, Freschet et al. 2010. However, it is generally not known whether different orders of fine roots (i.e., most distal, tip ended roots are first (lowest) order and second-order roots starting at the junction of first-order roots etc. as characterized by Fitter (1982)) act as a source or sink of N compounds during transitions over the fall season.
Given that different roots within the fine-root branching hierarchy often play functionally different roles (Guo et al. 2008, Pregitzer 2008, accurate assessment of root conditions and changes during spring, summer and fall requires information of multiple traits made on separate root branching orders. Previous studies have sought to determine the effect of variation in moisture and temperature on root characteristics within absorptive but not transport fine-root orders (Wells et al. 2002, Hishi andTakeda 2005). Absorptive fine roots represent roots with greater capacity for nutrient and water uptake, whereas transport fine roots occur higher in the branching hierarchy and play primarily structural and transport functions (McCormack et al. 2015). Fewer studies describe seasonal changes ( Pregitzer et al. 2002), yet there is evidence that different root orders are likely to respond differently across seasons (Xia et al. 2010, Yin et al. 2014). Furthermore, adaptation by plants to occupy stands of different fertility likely involves adaptation to different seasonal patterns of nutrient utilization (discussed by Chapin and Kedrowski 1983), for example, to minimize loss of nutrients in resource poor sites (Aerts 1999). Broad generalizations regarding adaptation of functional root traits across seasons therefore require analysis in multiple sites with contrasting fertility.
Quercus robur is adapted to growth both within poor, dry and acidic soils as well as in nutrient rich, humid and more neutral soil types (Szymański 1986, Ceitel 2006. As the fine roots adapt to local conditions we could expect changes in root morphology and chemistry to maximize root function in each environment (Eissenstat 1992, Marschner 1995, Reich et al. 1997. For example, it is expected that in environments rich in water and nutrients plants may produce a network of relatively thin roots with low average diameter (AD) and high specific root length (SRL) (Schenk 2006, Brassard et al. 2009) to enhance high resource acquisition (Chen and Brassard 2013). On the other hand, in resource poor stands, where the cost of building roots is high, plants may favor roots of higher AD and lower SRL, which may enable longer functional lifespan due to better protection from environmental hazards (Eissenstat et al. 2000, Yanai and thereby reducing resource losses through tissue turnover. However, these assumptions have not been well tested and it is still necessary to determine how fineroot functional traits respond in distinct soil types as well as across multiple seasons, which may also impact root functional traits (Wang et al. 2006, Jia et al. 2011. Furthermore, in the case of species in which physiological composition and tissue construction is strongly determined by genotype, e.g., Quercus petraea (Matt.) Liebl., root morphology may not be expected to respond to changes in the local environment. Additionally, high interplant competition may also change apparent patterns of nutrient availability and demand, which could also affect patterns of root construction Leuschner 2009a, 2009b).
In the current study, we examined key root traits that are likely to be related to nutrient acquisition across six root orders of 40-year-old individuals of Q. robur growing in different soil types during spring, summer and fall seasons. Targeted root traits included root morphology and structure (AD and SRL), anatomy (primary vs secondary tissue development), as well as indicators of physiological activity and resource storage (concentrations of nitrogen (N), starch, soluble carbohydrates, total non-structural carbohydrates (TNC)). We address the following questions: (i) do structural, anatomical and biochemical properties of fine-root orders vary across season and between soil types? (ii) Do anatomical breaks between roots of primary vs secondary development reflect changes in nitrogen storage across seasons?

Study site
This study was conducted at a common garden located in the Siemianice experimental forest established in 1970 in central Poland (52°14′87″N and 18°06′35″E). Before the common garden experiment was established the site was occupied by a Scots pine (Pinus sylvestris L.) plantation. The Scots pine plantation encompassed two soil types: one with relative high fertility and one with relatively low fertility (see below). After clear cutting of the Scots pine, 14 tree species were planted in three replicate (400 m 2 each, 20 × 20 m) monospecific plots at each soil type. Details of site preparation and forest cultivation are presented by Reich et al. (2005). This experimental setting allows us to explore how root trait characteristics (morphological, anatomical and biochemical) change in response to different soil properties. In this study we utilized plots planted with a common oak species, Q. robur. We collected root samples from two plots at each soil type.

Soil analysis
We collected four soil samples of similar volume from the 0-10 cm soil depth from each oak plot in both soil types, after removal of decomposed litter. At the laboratory, samples from each plot were sifted by hand to remove rocks and plant material and then air dried at room temperature, crushed and passed through a 1-mm sieve as described by Reich et al. (2005). The organic carbon (C) % and N % were determined by the Tyurin and Kjeldahl volumetric methods, respectively. The pH water of the soil sample was measured in a soil : water suspension. Hydrolytic acidity (Hh) and cation exchangeable bases (CEB) were determined by the Kappen method (Raczuk 2001). Cation exchange capacity (CEC) was estimated using the following equation: CEC = CEB + Hh. The extent of base saturation of the soil sorption complex was calculated as follows: %VCEB = CEB/CEC × 100, and the degree of hydrogen saturation of the soil sorption complex was defined with %VHh = Hh/CEC × 100 (Ostrowska et al. 1991). Moreover, at each plot soil types were classified via excavation pit to the a depth >1 m, according to WRB or Soil Taxonomy nomenclatures, respectively. The first soil type was a relatively more fertile, Brunic Luvisol/Arenic Hapludalfs and second soil type was relatively less fertile and classified as Brunic Arenosol/Typic Udipsamments (Table 1).

Root sampling
Root samples for trait characterization were collected in early April, late June and in early October of 2012 from each of the soil types. These dates were chosen to coincide with spring leaf emergence, summer and fall leaf senescence. Within each plot we randomly selected two subplots where root samples were collected to 10 cm depth using a shovel. Accordingly, four soil blocks (20 × 20 × 10 cm) were excavated for each harvest time at each soil type. Samples were taken up to 2 m from the trunk. The soil blocks were placed into plastic bags and transported to the Institute of Dendrology in Kórnik, Poland. Soil blocks were stored in the refrigerator (∼5 °C) until further analysis for up to 1 week.
Once removed from the soil, the roots were divided into individual orders according to the method described by Pregitzer et al. (2002). Dissections of roots from first through sixth order were made using a steel scalpel. During the dissections we paid careful attention to keep the roots moist. Furthermore, only roots that appeared to be alive, determined based on texture and visual appearance, were collected for analyses. In the case of first-order root tips, we studied only fibrous roots sensu Zadworny and Eissenstat (2011), and avoided pioneer roots that are morphologically and functionally different from fibrous roots (i.e., pioneer roots primarily function for soil exploration and form the framework of the root system (Bagniewska-Zadworna et al. 2012)). During dissections it was noted, and later confirmed by anatomical studies, that nearly all root tips were colonized by mycorrhizal fungi to some extent and no segregation between mycorrhizal and non-mycorrhizal root tips was made.
In fall 2014, standing root biomass was estimated in both soil types by soil core sampling as described by Jagodziński and Kałucka (2011). Briefly, eight randomly selected cores per plot were collected (4.7 cm diameter to 20 cm depth). After harvesting, roots were cleaned over 2-mm sieves and then all roots ≤2 mm diameter manually sorted using tweezers. All roots classified as dead based on visual and textual cues were discarded. We then determined the dry biomass of live roots by drying them at 65 °C for 3 days and then weighing. Data were expressed in g m -2 . The Brunic Luvisol soil was characterized by lowest fine-root biomass (260.24) g m -2 , whereas root biomass at the Brunic Arenosol soil type with lower fertility was significantly higher (487.2) (P < 0.001; Student's t-test). This observation reflects the general pattern of higher biomass allocation to fine roots at stands of lower fertility.

Frequency and biomass allocation of root order
In the fall of 2014, in order to estimate the frequency of the different root orders in both soil types, intact root branches were dissected into individual orders, up to sixth order, as described above. We used three branches per plot, and two plots per stand. The number of each root order within the branch was calculated. After dissection, the roots were dried at 65 °C for 3 days and weighed. The per cent contribution of each root order to the total branch weight was calculated. The percentage was calculated for each order on each branch and then these per cent values were averaged for each soil type. Based on anatomical results, we divided the mass of individual orders into assigned groups of absorptive roots and transport roots (see Anatomical analysis section). The mass from a given root order that was added to the absorptive pool was proportional to the percentage of those roots that expressed only primary development. For example, if there were 100 mg of third-order roots, but only 50% showed primary development, 50 mg was added for the calculation of total mass of absorptive roots. By summing the mass contribution of roots with absorptive function we determined whether the total roots with absorptive function varied between soil types.

Morphological and chemical analysis
Roots of each order were scanned with an Epson Perfection V700 PHOTO desktop scanner (transmitting light system) in gray scale at 300 dpi. Images of individual root orders were analyzed for length and diameter for each subplot separately using WinRHIZO software (Regent Instruments, Inc., Québec, QC, Canada). The roots were then dried at 65 °C for 3 days and Table 1. Average value of basic characteristics of Brunic Luvisol and Brunic Arenosol soil types including pH, hydrolytic acidity (Hh), soil extractable Al (Al.), exchangeable acidity (Hw), cation exchangeable bases (CEB), cation exchange capacity (CEC), the degree of base saturation of soil sorption complex (VCEB), the degree of hydrogen saturation of soil sorption complex (VHh), hygroscopic water, loss on ignition, organic carbon, organic matter, total nitrogen, carbon-to-nitrogen ratio (C : N) and bulk density within 10 cm depth. To explore patterns of carbohydrate allocation within different soil types and across seasons, concentrations of nonstructural carbohydrates (soluble carbohydrates and starch) were determined as described by Oleksyn et al. (2000). Sugars were extracted from oven-dried and ground tissue in methanolchloroform-water. The tissues remaining after extraction were used to determine starch concentration. Starch in the insoluble material was converted to glucose with amyloglucosidase and oxidation using the peroxidase-glucose oxidase complex. Using a UV-1700 Pharma Spec (Shimadzu, Kyoto, Japan) spectrophotometer, concentrations of soluble sugars were measured at a wavelength of λ = 625 nm, following a color reaction with anthrone, while starch concentrations were measured at λ = 450 nm following the reaction with dianisidine. Concentrations of soluble sugars and starch (with glucose as a standard) were expressed as a percentage of dry mass.

Anatomical analysis
For anatomical analyses, 30 roots were chosen randomly from each root order from each plot, resulting in a total of 60 roots analyzed per order, soil type and season. Following order dissections, selected roots were immediately fixed in 2% formaldehyde (Polysciences, Inc., Warrington, PA, USA) and 2% glutaraldehyde (Polysciences) in 0.1 M cacodylate buffer (Polysciences) according to Bagniewska-Zadworna et al. (2012). For lowerroot orders, the entire root was fixed. In higher-order roots where lengths frequently exceeded 15 mm, the root samples were limited to 10 mm. After 24 h in fixing buffer, the roots were washed once in 0.05 M cacodylate buffer, washed twice in deionized water and then dehydrated in gradually increasing series of ethyl alcohol (Polish Chemical Reagents, Gliwice, Poland) for 1 h in each concentration (10, 30, 50, 70, 90, 96 and 100%). Roots were infiltrated and embedded using Technovit 7100 (Heraeus Kulzer, Wehrheim, Germany) directly after the final dehydration in 100% ethyl alcohol. Cross sections of 5 µm thickness (embedded in Technovit 7100) were obtained using a rotary microtome (Leica RM2265) and stained with 0.5% toluidine blue (Sigma, St Louis, MO, USA) dissolved in 1% sodium tetraborate (Sigma). For consistency, we dissected the basal part of the tip ended roots (closer to the junction to another order), as considerable differences in the tissue differentiation can be found in the range of millimeters. Observations were performed under an Axioskope 20 microscope (Carl Zeiss, Jena, Germany). Root orders were classified as having primary or secondary growth based on the presence or absence of vascular cambium and phellem (the presence of both is indicative of secondary development and reduced root absorptive function as expanded phellem inhibits ion and water uptake) (Peterson et al. 1999, Enstone et al. 2001, Meyer and Peterson 2011 marking their primary function for transport and structural support.

Statistical analysis
The effect of soil type, root order and season were analyzed by ANOVA. For traits expressed as percentages, the Bliss angular transformation was used for statistical analyses to meet the assumption of normality, but the figures present non-transformed data. Furthermore, Student's t-test was used to compare soil and root characteristics between Brunic Luvisol and Brunic Arenosol soil. All statistical relationships were considered significant at P < 0.05. All analyses were conducted using Statistica version 8.0 (StatSoft, Inc., Tulsa, OK, USA).

Root morphology and anatomy
Significant differences were found between soil types in regard to the number of roots of a given order per higher root branch order across first-, second-and third-order roots ( Table 2). We also observed that despite higher numbers of roots within a particular order at Brunic Luvisol, the root order frequency and Trait variation in oaks root orders across seasons 647 Table 2. The number of roots per order per branch (Root no.), the order frequency proportion and order mass proportion of each root order (I-VI) sampled from Q. robur growing in Brunic Luvisol and Brunic Arenosol soils. The order frequency proportion was based on the percentage contribution (based on root number) of a given order to total root number for all root orders (I-VI). The order mass proportion was based on the mass proportion of each root order to the total branch weight. The Student's t-test was used to assess significance of differences between Brunic Luvisol and Brunic Arenosol root order characteristics and significance is indicated as *P < 0.05, **P < 0.01 and ***P < 0.001 and ns (i.e., not significant); n = 6 for each soil type per each order. root order mass contribution throughout the branching hierarchy were not different between the two soil types (Table 2). There were no significant differences in root morphology and structure between roots from the different soil types (Table 3). Root diameter increased and SRL decreased with branch order at each harvest time in each soil type (Figure 1), with a mean diameter increase from 0.2 to 0.7 mm at the Brunic Luvisol soil and to 0.8 mm at the Brunic Arenosol soil. At the same time, SRL decreased sevenfold from ∼36 to 5 m g −1 in both soil types. Secondary development increased with branch order in a similar pattern in both soil types. All roots of the first three orders were characterized by primary development with little to no secondary development in the first two orders and <30% in third-order roots representing a transition zone between primary and secondary growth. In contrast, nearly all higher-order roots (fourth through sixth order) exhibited secondary growth with expanded secondary xylem and a peridem layer.

Relationship between soil type, nitrogen concentration and carbohydrates
The N concentration varied across seasons (P = 0.015) and between soil types (P = 0.047; Table 3) being on average 6.5% higher in Brunic Luvisol than Brunic Arenosol soil. Nitrogen concentration varied significantly with root orders ( Table 3), but surprisingly the expected reduction in N concentration with increasing root order was not consistent across all seasons. During spring and summer we observed a consistent decrease in N concentration from first to sixth order by 56% (Figure 2). However, during    fall, at the time of leaf senescence, this pattern was modified by a marked decrease in the N concentration in first-, second-and third-order roots coupled with an increase in N concentration between the third-and fourth-order roots. This shift was observed in both the Brunic Luvisol and Brunic Arenosol soils with increases of 140 and 65%, respectively, in fourth-order roots (Figure 2). The alteration in N concentration also resulted in a concomitant change in the observed C : N ratios (Table 3; Figure 2). No significant variability within soluble carbohydrates, starch and TNC concentrations existed with respect to soil types (P > 0.05) and soil types × season (P > 0.05) ( Table 3). We observed a significant variation in soluble carbohydrates and TNC across root orders (P < 0.001) where a trend of increasing their concentration with branching hierarchy existed (Figure 3), while starch concentration showed a lack of significance (P > 0.942). Furthermore, there were few consistent patterns in carbohydrates among seasons, but we did observed an increase in starch concentration in the third-and fourth-order roots during April, but only in the Brunic Luvisol soil type (Figure 3).

Discussion
Fine roots enable plants to acquire limiting resources from the soil and must do so under a range of environmental conditions. Accordingly, fine roots also play an important role enabling plants to adapt to variation in local and seasonal environments. In this study we observed patterns of morphological, anatomical and physiological traits relating to nutrient acquisition across three seasons and between two contrasting soil types. Surprisingly, we found that patterns of root morphology were generally similar across all seasons, i.e., spring, summer and fall, and neither AD nor SRL differed between Brunic Arenosol and Brunic Luvisol soil types (Table 1; Figure 1). These morphological parameters are considered to be indicators of uptake capacity (Leuschner et al. 2004, Ostonen et al. 2007) and construction cost (Eissenstat and Yanai 1997) and should be related to root adaptation to varied soil conditions (Ostonen et al. 2006, Jagodziński and Kałucka 2010, Jagodziński and Kałucka 2011. Specific root length is a trait that previously was observed to respond to different levels of soil fertility among sites as was discussed by Ostonen et al. (2007). One explanation for a lack of observed differences may be strong links of AD and SRL to genotype in Q. robur and that there is little phenotypic plasticity in these traits as they respond to different environmental pressures. This is in accordance with some previous studies showing limited plasticity in root morphological traits across studies and growing seasons (Pregitzer et al. 2002, Comas and Eissenstat 2004, Lee et al. 2014). The striking difference in seasonal patterns of N concentration across root orders (Figure 2) is most likely related to root anatomy and the potential storage function of higher-order roots with secondary development. In our study, higher N concentration was observed during fall leaf senescence in the fourth to sixth orders, which differ in structure and function from absorptive, first-to third-order roots. Based on the proportion of biomass estimated to be absorptive vs transport fine roots from Trait variation in oaks root orders across seasons 649

robur grown in Brunic Luvisol and Brunic
Arenosol soil types harvested in April (circles), June (squares) and October (rhombuses). Significant differences among seasons but within the same root order are indicated as *P < 0.05, **P < 0.01 and ***P < 0.001 (ANOVA). Mean values are given ±SE; n = 4 for each root order, season and soil type.
at Institute of Geographic Sciences and Natural Resources Research,Chinese Aca on December 3, 2015 http://treephys.oxfordjournals.org/ Downloaded from root samples collected in fall 2014, and the N concentration of each root order in the fall season, we found that the proportion of total fine-root N contained in transport fine-root roots was approximately three times that contained in absorptive fine roots (∼25% in absorptive vs 75% of N in transport fine root). This was partially the result of the low biomass fraction of fine roots classified as absorptive, which constituted on average 35% (at Brunic Luvisol) and 29% (at Brunic Arenosol) of total fine-root biomass (first through sixth orders). However, it was also largely the product of the striking increases in N concentration in the fourth-to sixth-order roots in the fall season. It is important to note that if all roots <1 mm were treated as a single homogeneous fine-root pool these patterns of N mobilization and storage would be undetectable.
Based on the structure and function of the higher-order roots (e.g., fourth to sixth), we assert that the observed patterns of N concentrations within these roots are likely not related to the high metabolic activity (Reich et al. 2008) or nutrient uptake (Comas and Eissenstat 2004). These roots are characterized by secondary development and primarily serve as a structural framework for lower-order roots and to transport nutrients and water acquired in lower orders. Furthermore, these roots are not colonized by mycorrhizal fungi, so it is not possible to link increases in N concentrations with increased contributions of fungal mycelium that contain N (i.e., chitin). As a result, we propose that the large pool of N within transport fine roots that accumulates during fall is most likely reserved to support new growth during spring when environmental conditions inhibit high uptake from cold soil.
The location of stored N in higher-order fine roots may be critical to ensure maximum nutrient retention over winter and allow rapid deployment of new root growth in early spring. Given that the finest root orders with only primary development (i.e., first to third) are the most susceptible to frost (Colombo 1994, Figure 3. Temporal pattern of soluble carbohydrates, starch and TNC concentrations across root orders (first to sixth) of Q. robur grown in Brunic Luvisol and Brunic Arenosol soil types harvested in April (circles), June (squares) and October (rhombuses). Significant differences among seasons but within the same root order are indicated as *P < 0.05 (ANOVA). Mean values are given ±SE; n = 4 for each root order, season and soil type.  Colombo et al. 1995) and might undergo planned obsolescence (Bagniewska-Zadworna et al. 2014), it is reasonable that N accumulation takes place primarily in fourth-and fifth-order roots as a way of limiting nutrient loss from first-to third-order roots, which may be damaged or die over winter. The higher root orders are better protected with cork layers and also contain stored carbohydrates (Figure 3), which can further protect these roots from low temperatures (Sakai 1983, Boldingh et al. 2000, Norisada et al. 2005 in addition to supporting root growth before leaves emerge in spring (Wang et al. 2006). Thus, storing N in better protected, higher-order roots with secondary development may improve resource retention and decrease loss over winter, which may be important attributes for nutrientconservative species (Norby et al. 2000) such as Q. robur, and which together would indicate discrete functions among higherand lower-order roots. Furthermore, maintaining local reserves of N in fourth-and fifth-order roots also enables rapid growth of new absorptive fine roots to capture pulses of nutrient availability as soils begin to warm. This is consistent with previous studies which have found that spring Scots pine growth depends mainly on nutrients stored from previous growing seasons ( Fagerström and Lohm 1977).
Finally, it has been well established that N may be resorbed or withdrawn from leaf tissues prior to senescence (Chapin and Kedrowski 1983, Nambiar and Fife 1991, Aerts 1996. Compared with leaves, consistent patterns of root senescence are less clear. Therefore, we cannot directly infer a relationship between reduced N concentration in lower-order roots and root senescence (Meier et al. 1985, Freschet et al. 2010. However, it is also possible that some of the observed N decrease in lower-order roots is related to broader patterns of root senescence and nutrient retranslocation. Regardless of the specific mechanism, seasonal patterns of N cycling with increasing root N concentration in higher-order roots in fall likely serves as a strategy to retain N in plant tissues (i.e., minimize loss) and ensure an adequate resource supply for activity or new production of absorptive roots in spring.
Plant strategies for root growth and allocation of resources to different root orders has important consequences for the ability of a root system to satisfy the immediate needs of a plant for water and nutrients. Furthermore, strategies for internal cycling and storage of limiting resources also have implications for the ability of a root system to grow and support future plant demands for soil resources. In this study, dissection of roots into separate root orders revealed important changes in the patterns of N concentration within finest root orders (<1 mm in diameter). Of particular importance was the observation that while N concentration generally decreased with increasing root order in spring and summer, there was a striking increase in the N concentration of fourth-and fifth-order roots in fall. While increased N concentration is generally considered to be an indicator of metabolic activity, in this case it is more likely that the increased N in higher-order roots of secondary development is a reflection of their potential role as storage organs where resources can be retained over winter and support root activity and growth in early spring. Further work is needed to quantify the size of this storage pool in relation to more traditionally identified pools of N and carbohydrates stored in woody stems and large branches.