Changes in fine-root production, phenology and spatial distribution in response to N application in irrigated sweet cherry trees

Abstract

Factors regulating fine-root growth are poorly understood, particularly in fruit tree species. In this context, the effects of N addition on the temporal and spatial distribution of fine-root growth and on the fine-root turnover were assessed in irrigated sweet cherry trees. The influence of other exogenous and endogenous factors was also examined. The rhizotron technique was used to measure the length-based fine-root growth in trees fertilized at two N rates (0 and 60 kg ha⁻¹), and the above-ground growth, leaf net assimilation, and air and soil variables were simultaneously monitored. N fertilization exerted a basal effect throughout the season, changing the magnitude, temporal patterns and spatial distribution of fine-root production and mortality. Specifically, N addition enhanced the total fine-root production by increasing rates and extending the production period. On average, N-fertilized trees had a length-based production that was 110–180% higher than in control trees, depending on growing season. Mortality was proportional to production, but turnover rates were inconsistently affected. Root production and mortality was homogeneously distributed in the soil profile of N-fertilized trees while control trees had 70–80% of the total fine-root production and mortality concentrated below 50 cm depth. Root mortality rates were associated with soil temperature and water content. In contrast, root production rates were primarily under endogenous control, specifically through source–sink relationships, which in turn were affected by N supply through changes in leaf photosynthetic level. Therefore, exogenous and endogenous factors interacted to control the fine-root dynamics of irrigated sweet cherry trees.

Introduction

Fine roots represent the most active and dynamic part of the root system and are responsible for most water and nutrient uptake in plants. They are traditionally defined as non-woody, small-diameter (<2 mm), often short-lived (<1 year) roots, accounting for most of the root system length in tree species (Pregitzer et al. 1997, Wells and Eissenstat 2001, 2003, Rytter 2013). Given that fine roots are continually emerging, aging, and dying throughout the growing season, the composition of the fine fraction of root systems is very heterogeneous, with roots of different ages, lengths, diameters, colors, branching orders and degrees of mycorrhizal colonization (Eissenstat and Yanai 1997, Wells and Eissenstat 2003). Newly born fine roots are white and exhibit a high capacity for water and nutrient uptake. As they age, they turn brown and show a sharp decrease in respiratory activity and nutrient uptake rates (Comas et al. 2000, Bouma et al. 2001, Volder et al. 2005, Baldi et al. 2010). However, older roots may significantly contribute to total nutrient uptake as they account for larger surface areas compared with younger roots, at least in tree seedlings (Hawkins et al. 2014).

The balance between production and mortality, together with turnover rates, determines the total amount of fine roots available for resource acquisition at each time point, i.e., the standing crop. Fine-root production follows a time course through the growing season, i.e., root phenology, which varies among species. Root phenology is also strongly affected by exogenous factors, especially by the soil temperature, water content, and nutrient availability (Yunhuan et al. 2006, Fukuzawa et al. 2013, Noguchi et al. 2013), resulting in production patterns changing from year to year and with varying climatic conditions (Atkinson and Wilson 1980, Eissenstat et al. 2006, Abramoff and Finzi 2015). Recent studies have suggested that root phenology is more strongly related to soil temperature than to soil water or N availability (Tierney et al. 2003, Majdi and Öhrvik 2004, Steinaker et al. 2010, Fukuzawa et al. 2013, Rytter 2013). Despite these antecedents, it is still widely accepted that temperate fruit trees typically have a bimodal pattern of root production (Eissenstat et al. 2006). However, different fruit species have shown a wide range of root patterns, varying from unimodal and bimodal to evenly distributed, e.g., in apple (Psarras et al. 2000), grapevine (Comas et al. 2005), peach (Baldi et al. 2010) and sweet cherry (Atkinson and Wilson 1980, Gratacós et al. 2008). Moreover, the root phenology of fruit trees can also be affected by rootstock genotype (Atkinson and Wilson 1980, van Hooijdonk et al. 2011), tree age (Wu et al. 2012) and cultural practices, such as above-ground pruning (Eissenstat and Duncan 1992, Comas et al. 2000, 2005), crop load adjustment (Rosecrance et al. 1996, Morinaga et al. 2003), vegetation control on the orchard floor (Parker and Meyer 1996) and plant density (Atkinson and Wilson 1980), making it difficult to establish generalizations in this regard.

Endogenous factors, such as source–sink relationships, and even hormonal control, are also likely to be important drivers of root phenology (Berman and DeJong 2003, Tierney et al. 2003, Comas et al. 2005, Abramoff and Finzi 2015). This should be particularly true in fruit-bearing trees due to the presence of fruits, which are considered to be stronger sinks for carbohydrates in comparison with shoots and roots (Grossman and DeJong 1995, Flore and Layne 1999, Génard et al. 2008). Indeed, several studies have shown that fruiting reduces root growth, especially during the final stage of fruit growth when fruit has maximum sink strength (Williamson and Coston 1989, Glenn and Welker 1993, Grossman and DeJong 1995, Inglese et al. 2002, Morinaga et al. 2003, Mimoun and DeJong 2006, Basile et al. 2007, Abrisqueta et al. 2008). Further evidence of the effect of source–sink relationships on root phenology might be the decoupling frequently registered between root and shoot growth in woody species (Steinaker et al. 2010, Abramoff and Finzi 2015). In temperate fruit trees, the major fine-root peaks are generally registered several days or weeks after the main phase of shoot extension, e.g., in young apple trees (Psarras et al. 2000), in mature apple trees (Wells and Eissenstat 2001), in young sweet cherry trees (Bonomelli et al. 2012) and in young peach trees (Abrisqueta et al. 2008). This pattern may be interpreted as a strategy to avoid shoots and fine roots, the two major resource-acquiring tissues, becoming competitive sinks for stored and newly fixed C (King et al. 2002, Wells and Eissenstat 2003).

On the other hand, the replacement rate of the fine roots, i.e., the turnover rate, is relatively fast in tree species (0.4–2.8 year⁻¹; Gill and Jackson 2000, King et al. 2002, Fukuzawa et al. 2013, McCormack et al. 2014), as fine roots live for a short time. Calculations are mainly based on dividing the annual root production or mortality by the average, maximum, or minimum root standing crop (Gill et al. 2002, Fukuzawa et al. 2013, McCormack et al. 2014). The turnover rate has also been calculated as the inverse of the fine-root lifespan as determined using minirhizotrons (Burton et al. 2000, Majdi et al. 2005) or using biomass-based methods (Nadelhoffer 2000). The different methods have their own limitations and can produce dissimilar results, making it difficult to estimate this variable (Gaul et al. 2009). In temperate fruit trees, the information about the turnover rates of fine roots is scarce. However, the median lifespan, i.e., number of days to 50% mortality of a root population, typically varies between 30 and 100 days (Bouma et al. 2001, Wells and Eissenstat 2001, Wells et al. 2002), and given that survivorship curves are frequently asymmetrical, this means that most, if not all, fine roots that emerge each year turn brown or senesce before the end of their first season.

The fine-root lifespan, and consequently the turnover rate, varies among species, and is influenced by several additional factors (see McCormack and Guo 2014). In general, the root lifespan decreases as soil temperatures increase (Burton et al. 2000, Gill and Jackson 2000, Pregitzer et al. 2000a, Tierney et al. 2003, Majdi and Öhrvik 2004). Studies focused on nutrient and water availability have shown positive and negative effects depending upon the methodology, ecosystem, species, or study scale (see Eissenstat et al. 2013).

Specifically concerning the impact of the N fertilization and soil N availability on fine-root processes in tree species, several studies have examined this topic. Some works have reported increases in the standing crop length, fine-root length production and mortality, and fine-root biomass with improved soil N availability (Pregitzer et al. 1993, 2000b, Kubiske et al. 1998, King et al. 1999, 2002, Majdi 2001, Adams et al. 2013, Noguchi et al. 2013). Other studies have reported null effects (Phillips et al. 2006, Rytter 2013) or even the opposite result (Jia et al. 2010, Rytter 2013). Similarly, studies evaluating the effect of N on root lifespan have had inconsistent results (see Eissenstat et al. 2013, McCormack and Guo 2014). Some have shown a decrease in the fine-root lifespan (or an increase in turnover rate) in N-rich soils, both in woody (Kubiske et al. 1998, Johnson et al. 2000, Pregitzer et al. 2000b, Majdi 2001) and herbaceous species (Van Der Krift and Berendse 2002, Bai et al. 2008). However, evidence for an increase in the fine-root longevity (Pregitzer et al. 1993, Burton et al. 2000, Baldi et al. 2010, Adams et al. 2013, Noguchi et al. 2013) or even null effects (Guo et al. 2008, Baldi et al. 2010, Adams et al. 2013, Rytter 2013) also exists. Concordantly, there are different hypotheses about this topic (see Nadelhoffer 2000, Rytter 2013), with most resulting in absolute and proportionally lower fine-root biomass in N-fertile sites through an absolute decrease in the fine-root production, with or without an influence on the fine-root longevity (Nadelhoffer et al. 1985, Vogt et al. 1986).

In summary, there is currently limited knowledge about below-ground processes and fine-root dynamics, particularly in fruit tree species. Control mechanisms of exogenous and endogenous factors are also poorly understood. In this context, the rhizotron technique was applied in an irrigated sweet cherry (P. avium cv ‘Bing’ on Gisela^®6 rootstock) orchard during two growing seasons to assess the effects of long-term N addition on the temporal and spatial distribution of fine-root length-based growth and on fine-root turnover rates. Additionally, this study seeks to improve the understanding of the endogenous and exogenous control of fine-root dynamics in temperate fruit trees. Specifically, the objectives were as follows: (i) to examine the N fertilization effects on standing length, production, mortality, and turnover rates of fine roots distributed in different soil layers and growing seasons; (ii) to assess the seasonal and inter-seasonal variation in fine-root production and mortality in central Chile, relating these variations to environmental factors, such as soil temperature and N availability; and (iii) to examine the extent to which root and above-ground phenology are related in sweet cherry trees. We hypothesized that: (i) a non-limiting N availability in the soil will increase length-based production and mortality during the growing season in irrigated sweet cherry trees, and the relative increments of both variables will lead to an increase in seasonal fine-root standing length; (ii) a non-limiting N availability in the soil will shorten the lifespan (or increase the turnover rate) of fine roots, which in turn will produce an increase in the seasonal fine-root production; and (iii) the seasonal patterns of length-based production and mortality rates of fine roots in irrigated sweet cherry trees growing in a temperate climate will be primarily influenced by exogenous factors, such as soil temperature and N availability, while endogenous factors will have an effect mainly limited to the fruiting period.

Materials and methods

Study site and plant material

The study was conducted in a 5-year-old orchard of sweet cherry cv. ‘Bing’ on Gisela^®6 rootstock (Prunus cerasus × Prunus canescens) located in Chile’s central region (34°08′S and 70°43′W). The orchard was planted in August 2006 as part of a wider N optimization experiment (see Bonomelli and Artacho 2013), with trees spaced 2.5 m apart within rows and 4.5 m apart between rows (889 trees ha⁻¹). The ‘Black Tartarian’ cultivar was used as a pollinizer in a tree proportion of 11%. The climate of the area is a Mediterranean type with rainy, cool, wet winters and hot, dry summers (Meteorological Office of Chile 2014). The annual precipitation (mean 500 mm) is concentrated in the winter months (between May and September), and the mean annual potential evaporation is 1.242 mm. The soil at the site is classified as Fluventic Haploxeroll according to Soil Taxonomy-USDA; it presents a clay-loam texture, is more than 1.5 m in depth and exhibits no physical limitations. The initial soil chemical characteristics (0–30 cm depth) were as follows: organic matter content (Walkley–Black method) 2.0%; pH (soil : water, 1 : 2.5) 6.6; P (Olsen method) 23 mg kg⁻¹; exchangeable K (ammonium acetate method) 242 mg kg⁻¹; and medium to high levels of bases, S and micronutrients. The soil analysis was repeated in 2012, showing similar results. Thus, all nutrients in the soil, except N, were at sufficient levels and did not limit the growth or fruit production of the trees.

Experimental design

The length-based fine-root growth was studied during two growing seasons (2011–12 and 2012–13) using the rhizotron method. The basal experiment was established in 2006 as a four-replicate completely randomized design with three treatments (0, 60 and 120 kg N ha⁻¹), with experimental units corresponding to a group of nine trees (three trees per row in three adjacent rows; see Bonomelli and Artacho 2013). In the present study, two treatments were used (0 and 60 kg N ha⁻¹), in a three-replicate completely randomized design, which were ongoing since the orchard planting in 2006. The experimental unit was a rhizotron installed next to a tree. However, the eight nearby trees had received the same N treatment since orchard planting and were used for destructive measurements and for soil sampling from the surrounding area.

Each season, the N fertilization was applied as urea (46% N) in four N splits, commencing when the soil temperature reached 15 °C (mid-October) and ending in late February. Specifically, trees were fertilized on 12 October 2011, 2 November 2011, 7 December 2011 and 29 February 2012 in the 2011–12 season and on 10 October 2012, 31 October 2012, 21 November 2012 and 20 February 2013 in the 2012–13 season. The N fertilizer was applied manually below the two-drippers nearest to each tree and was immediately incorporated with irrigation.

Both control and N-fertilized trees received supplemental irrigation through a double-drip line system with 4 l h⁻¹ pressure-compensating drip emitters located every 0.9 m along the rows (approximately four emitters per tree). Irrigation was applied two to three times per week from mid-October to end-March, with an average irrigation rate of 250 m³ ha⁻¹ week⁻¹. The total amount of water applied through the irrigation system was 5531 and 5883 m³ ha⁻¹ in the 2011–12 season and 2012–13 season, respectively. The irrigation was designed to maintain the soil water content near to field capacity. The volumetric water content at field capacity (0.32) was derived from soil texture analysis following Saxton and Rawls (2006).

Soil and air measurements

The soil volumetric water content and temperature were monitored, respectively, with 10HS and EC-T sensors (Decagon Devices Inc., Pullman, WA, USA) installed at 20 and 40 cm depth below the drippers that irrigated the trees with rhizotrons. The air temperature and relative humidity were registered with a combined sensor (Decagon Devices Inc.) located at 2 m height. All sensors were connected to EM50 or EM5b data loggers (Decagon Devices Inc.), registering data every 15 min. The maximum and minimum daily air temperature values and relative humidities were used to estimate the daily values of vapor-pressure deficit (VPD).

The N availability in the soil was assessed monthly at 0–20, 20–40 and 40–60 cm depth from soil samples collected around the trunks of nearby trees, i.e., trees without rhizotrons but receiving the same treatment. Soil samples at each depth were composited by five subsamples obtained in a radial array within a 75 × 75 cm quadrant of a tree trunk. The extraction of NO₃-N and NH₄-N was realized with 2.0 M KCl, and the determination was performed with steam distillation (Maynard and Kalra 1993, Sadzawka et al. 2006).

Above-ground measurements

The growth of shoots, trunks and fruits in each growing season was evaluated on trees with rhizotrons, i.e., three trees per treatment. Four lateral shoots per tree were selected and marked on representative branches and their length measured weekly from October to January. Fruit growth was assessed from early October to harvest on a weekly basis by measuring the equatorial diameters with a digital caliper (Digimess, Buenos Aires, Argentina) for 10 designated fruits per tree. The fruit yield per tree was measured at harvest, which occurred on 30 November 2011 (70 days after full bloom; DAFB) in the 2011–12 season, and on 4 December 2012 (62 DAFB) in the 2012–13 season. At these times, the equatorial diameter of 40 additional fruits per tree was evaluated. The trunk diameter at 0.1 m above the graft union was measured weekly from early October (bud break) to the end of March (dormancy) during both growing seasons. The tree phenology was evaluated weekly in experimental trees, and the dates of key phenological events were registered.

The nutritional status of the trees was evaluated in each growing season by foliar analysis from leaves collected in mid-summer (mid-January) from the middle third portion of newly formed shoots. According to the reference standards reported by Reuter and Robinson (1997) for sweet cherry trees, the foliar N concentration was normal in trees with a 0 kg N ha⁻¹ treatment and was slightly high (3.10%) in trees receiving 60 kg N ha⁻¹. The leaves from both N treatments presented normal values of P, K, Zn, Cu and Mn, slightly low concentrations of Ca (1.10%) and Mg (0.27%) and slightly high concentrations of B (75 ppm).

The net leaf photosynthetic rate was measured around midday from mid-October 2011 to the end of February 2012 with a CI-340 portable gas analyser (H₂O/CO₂) (CID, Inc., Camas, WA, USA) and leaf chamber clip (6.25 cm² surface) with a quantum sensor, air temperature and humidity sensors. The gas exchange unit was operated in the open mode at a flow rate of 0.3 l min⁻¹ and an ambient CO₂ partial pressure. Biweekly evaluations were performed until the end of December 2011, and subsequently every 3 weeks, completing nine measurements in the 2011–12 season. During the 2012–13 season, only four measurements were taken due to instrument loss caused by a fire in our laboratory. Three measurements per leaf of three leaves per experimental tree (nine leaves per N treatment) were taken around midday (1100–1400 h) on clear days on fully expanded, sunny, healthy leaves located at breast height.

Rhizotron installation and length-based measurements of fine roots

The rhizotrons were wooden boxes (1.2 m wide by 1.2 m long and 1.2 m deep) with glass (1.2 m wide by 1.2 m long and 8 mm thick) forming the root observation window, and were installed on 11 August 2011 on the north face of each experimental tree and at 0.5 m from the trunk within the row. Thus, three rhizotrons per treatment were examined. An excavation was carried out manually to introduce this structure into the soil, minimizing root damage as much as possible. Then, the soil was backfilled according to the original order of the soil strata, and ensuring a tight soil–window contact. To exclude light and stabilize soil temperature, the access door located on the top of the rhizotrons was constructed with an insulated panel (100 mm thick), and the glass window was covered with expanded polystyrene (80 mm thick; 10 kg m⁻³ density). A 1 × 1 inch (2.5 ×  2.5 cm) grid with identificatory letters and numbers was drawn on the edges of the rhizotron windows to facilitate further image processing and root measurements. The measurement area of each glass window was 36 inches (90 cm) wide and 40 inches (100 cm) long.

From September 2011 to April 2012 and from September 2012 to May 2013, digital images of the roots visible on the rhizotron glass were collected every 7 ± 1 days. During winter, two measurements were performed (25 May and 12 July 2012). On each sampling date, eight images per rhizotron (each 3648 × 2736 pixels) were captured with a digital camera (SONY Cybershot W190, 12.1 MP). The camera was mounted on a self-constructed photographic stand provided with eight fixed photo-shot positions and located 60 cm from the glass window. Using Adobe Photoshop CS3 software (Adobe Systems Inc., San Jose, CA, USA), a single image of the soil profile was created from individual images previously edited with a digital grid. Subsequently, each rhizotron image was scaled to its actual size in AutoCAD^®2007 software (Autodesk Inc., San Rafael, CA, USA). Only white and fine roots (<2 mm diameter) were measured in each image. The roots were traced manually with digital lines, and the total length was obtained using the AutoCAD^®2007 tool ‘sum of length of lines’. During the study period, 426 images of 0.9 m² soil profiles were analysed.

The standing root length (SRL) was estimated by summing the length of all white and fine roots present on each rhizotron on each sampling date for the entire soil profile (0–100 cm) and for different soil layers (0–25, 25–50, 50–75 and 75–100 cm), and it was expressed as the root length per glass rhizotron area observed (m m⁻²). The root production and mortality rates for each sampling period (7 ± 1 days) were calculated as the difference in the SRL between two consecutive sampling dates and were then expressed on a daily basis (m m⁻² day⁻¹). Positive values were assumed to be root production as a result of the growth of pre-existing roots and/or the growth of new roots that appeared on the glass between sampling dates. Negative values were assumed to be root mortality, comprising the disappearance of pre-existing roots on the glass or the brown pigmentation of the roots. Therefore, in this study, mortality was assessed as the death of the fine roots as functional organs, which was assumed to occur when roots turn brown (Comas et al. 2000). Monthly estimates of root production and mortality rates were calculated by averaging the weekly rates registered during the 4 weeks of each month and were also expressed as m m⁻² day⁻¹. Total fine-root length-based production (FRLP) and mortality (FRLM) (m m⁻² season⁻¹) were obtained by accumulating the weekly values over the growing season.

The turnover rate (season⁻¹) was calculated as the average of the turnover rates for production and mortality for each experimental tree, according to Eqs (1) and (2), respectively (Burton et al. 2000, Gill et al. 2002, Fukuzawa et al. 2013, McCormack et al. 2014).

(1)

(2)

where FRLP is the total fine-root production in length (m m⁻² season⁻¹), FRLM is the total fine-root mortality in length (m m⁻² season⁻¹), and SRL_max is the maximum SRL (m m⁻²) observed during each growing season. The average fine-root lifespan was estimated by dividing the number of days of the observation period in each season (295 and 238 days in 2011–12 and 2012–13, respectively) by the turnover rates (King et al. 2002).

Statistical analysis

Based on triplicate measurements per treatment, the means and standard error (SE) were calculated for all parameters. A repeated-measures analysis of variance (ANOVA) was used to examine the effects of the time of observation, N rate, soil depth and their interaction on the standing length, root production and mortality rates for each growing season, as well as on the exogenous soil conditions (temperature, water content and N availability). The between-subject effect was evaluated as the N rate and soil depth, and the within-subject effect was the time of observation (week or month). The assumption of sphericity was assessed with Mauchly’s test, and the degrees of freedom were adjusted by the Greenhouse–Geisser method if the assumption was violated.

The inter-seasonal differences in root and soil parameters were analysed using one-way ANOVA, the previous assessment of homogeneity of variance using Barlett’s test, and the normality of the data using Shapiro–Wilks’s test. The means were separated by Tukey’s honestly significant difference test. Differences at P < 0.05 were defined as clearly significant, and differences between P = 0.10 and P = 0.05 were defined as marginally significant, considering the high variability and expense of replication in root studies (Comas et al. 2005). The root parameters that were measured once in each season, including the total root production and mortality, turnover rates and lifespan, were analysed by a factorial ANOVA to examine the effects of the growing season, N rate, soil depth and their interaction.

The relationships between the monthly rates of fine-root length production and mortality or total fine-root length production and mortality for the entire soil profile (0–100 cm depth) and at each soil layer, as well as soil (temperature, water content and N availability) and air (VPD) environmental variables, were tested with Pearson’s product-moment correlations. Additionally, relationships between the total fine-root length production and mortality for the entire soil profile and between the monthly cumulative fine-root production or mortality and cumulative soil temperature were analysed using regression analysis.

To examine the timing of root growth at each soil depth and for the entire soil profile, the data of SRL, root production rates and mortality rates were analysed separately by the N rate with a repeated-measures ANOVA, using time of observation (week or month) as the within-subject effect. A Dunnett’s test was used to separate the means by comparing each measurement against a preselected measurement (control group).

All analyses were performed with the STATISTICA 6.0 software (Statasoft Inc., Tulsa, OK, USA), except for the regression analyses, which used Graphpad Prism version 5.0 (Graphpad Software Inc., San Diego, CA, USA).

Results

Tree phenology and environmental conditions

Full bloom occurred 12 days later in 2012 as compared with 2011 (3 October 2012 vs 21 September 2011), but the harvest maturity of the fruit was reached at a similar date and growing degree-days (Table 1). A linear regression analysis applied to the growing degree-days data accumulated from bud-break to full bloom showed that heat units were accumulated at a slower rate in 2012–13 than in 2011–12 (Y = 11.44 + 0.75X; R² = 0.98; P < 0.01), as the slope of the line was less than 1 (P < 0.01). In turn, a more rapid accumulation of heat units between full bloom and harvest occurred in the 2012–13 season, given that the slope of the line adjusted (Y = −18.80 + 1.13X; R² = 0.99; P < 0.01) was greater than 1 (P < 0.01). This result explains why the fruit maturity was reached at a similar date and growing degree-days in both seasons.

Table 1. .

Date of occurrence and growing degree-days (°C day⁻¹) for the main phenological stages of sweet cherry trees growing in central Chile.

Stage	Date		Growing degree-days1
	2011–12	2012–13	2011–12	2012–13
Full bloom	21 September	3 October	—	—
Petal fall	5 October	14 October	221	227
Pit hardening	2 November	7 November	421	430
Harvest	30 November	4 December	710	710
Shoot growth cessation	14 December	19 December	890	880
Early-leaf fall	28 March	3 April	2284	2196

¹From 1 September; base temperature: 7 °C.

The air temperature and VPD showed a typical seasonal pattern for a temperate environment (Figure 1a). The monthly mean maximum air temperature reached 31 °C in the summer months in both seasons. This peak lasted 4 months in 2011–12 (December–March) and 1 month in 2012–13 (February). Then, the maximum temperatures dropped to 15 °C in mid-winter (July). At this time, the monthly mean minimum temperature was −0.76 °C (data not shown). The monthly course of VPD was similar to the air temperature, with maximum mean values observed in mid-summer (almost 1.6 kPa), but with an early occurrence of the minimum mean values (May) (Figure 1a).

Figure 1. .

Environmental conditions during the period of study: (a) monthly mean maximum air temperature (AT), and monthly mean VPD at the experimental site; (b) monthly mean maximum soil temperature (ST, average values for 20 and 40 cm depth) for the N treatments; and (c) monthly mean volumetric soil water content (SWC) at 20 cm depth for the N treatments. Shaded areas indicate the irrigation period. FC, field capacity; PWP, permanent wilting point. Bars are the standard errors (n = 3 for soil water content and n = 6 for soil temperature).

The mean air temperatures during the growing season (September–May) were 16.8 and 16.1 °C in 2011–12 and 2012–13, respectively, and the respective values for the soil temperature were 19.5 and 18.9 °C at 20 cm depth. No differences in the soil temperature due to N fertilization were detected in any season (P > 0.05), even though the soil was ∼1 °C warmer in spring–summer and 1 °C colder in autumn–winter in the no-N treatment than in the with-N treatment (Figure 1b). The soil depth was not a significant factor (20 vs 40 cm; P > 0.05), probably due to the uppermost sensor being located too deep (20 cm), and the temperature fluctuations were dampened by the soil (Pregitzer et al. 2000a).

As an overall effect, the inter-seasonal variation in soil temperature was not significant (P > 0.05). However, a univariate analysis showed significant differences at the beginning (September–December) and at end of the season (March–May), when the soil was warmer by 1–2 °C in 2011–12 in comparison with 2012–13 (P < 0.05). The temperature followed a typical seasonal pattern for temperate soils (P < 0.01), i.e., the soil temperature increased from spring to mid-summer and then began to decline. The monthly mean maximum temperatures peaked from January to February in each season (24–25 °C), and then diminished to 9–10 °C from July to August (Figure 1b).

The soil volumetric water content was maintained near to field capacity during the irrigation period (October–March), independent of the N treatment (Figure 1c). Moreover, it was not affected by the N rate (P > 0.05) or time of observation (P > 0.05), confirming an adequate irrigation strategy in both seasons. The soil depth was a significant factor only during 2011–12 (P < 0.05), when the soil at a 40 cm depth had a higher mean water content than did that at a 20 cm depth (0.32 vs 0.29 cm³ cm⁻³). The inter-seasonal variation was significant (P < 0.01), with lower mean values in 2011–12 in comparison with the 2012–13 season (0.31 vs 0.34 cm³ cm⁻³).

The soil N availability was increased by N fertilization in each season (P < 0.01), even at the deepest soil layer (Figure 2). The soil depth was a significant factor only during 2011–12 (P < 0.01), as well as the N rate × soil depth interaction (P < 0.01). At this time, the soil N availability in the no-N treatment was higher at a 20 cm depth in comparison with 20–40 and 40–60 cm, particularly at the beginning and at end of the season, while the soil N availability at the N-fertilized treatment was higher in the superficial layers after each N application (Figure 2). There was a significant temporal variation in each season (P < 0.01), and the differences depended on the N rate (significant N rate × time interaction; P < 0.01). In the control treatment, the N availability at each soil depth remained low during most of each growing season, increasing towards the end of each season (February–March) with mild soil temperatures, ample soil water availability and trees entering into dormancy. In the N-fertilized treatment, the N availability at each soil depth increased after each N split, and at the end of the season. Additionally, the 2011–12 season had a higher mean value than did the 2012–13 season (14.7 vs 11.9 mg kg⁻¹; P < 0.01) (Figure 2).

Figure 2. .

Monthly mean mineral N in soil at 0–20, 20–40 and 40–60 cm depth for the N treatments. Bars are the standard errors (n = 3). Arrows indicate the N applications in the N-fertilized treatment. Asterisks denote differences between N treatments (Tukey’s test; *P < 0.10; **P < 0.05).

Tree growth and fruit yield

N treatments produced significant differences in the magnitude and dynamic of the shoot growth (see Figure S1 available as Supplementary Data at Tree Physiology Online). At the end of the 2011–12 season, the mean shoot length was higher in the N-fertilized (35.2 ± 3.7 cm) than in the control treatment (24.5 ± 3.9 cm), and the differences during the entire season led to an overall significant effect of N (P < 0.05). Additionally, the shoot growth dynamic by N regime was characterized by significantly different sigmoidal curves in both seasons (P < 0.01), but the N effect on the magnitude of shoot growth was not significant at the end of the 2012–13 season (29.1 ± 1.9 vs 31.8 ± 3.3 cm; P > 0.05) (see Figure S1a available as Supplementary Data at Tree Physiology Online). Using a functional approach (Hunt 2003), log(Gaussian) curves were fitted to the data of shoot growth rate. According to this analysis, the maximum growth rate of shoots was higher in trees receiving N (P < 0.01), but it occurred at the same time in both treatments (end October to early November) (see Figure S1b available as Supplementary Data at Tree Physiology Online). In 2011–12, the maximum rates were 0.87 ± 0.03 and 0.58 ± 0.05 cm day⁻¹, respectively, for N-fertilized and control trees. In 2012–13, the respective values were 0.99 ± 0.07 and 0.77 ± 0.04 cm day⁻¹. At the whole-tree level, trees fertilized with N had a similar number of shoots per tree to control trees, but the mean shoot length was higher. This resulted in trees with a higher total shoot length, but not in a higher total leaf area (Table 2).

Table 2. .

Vegetative growth variables and fruit yield of trees of sweet cherry ‘Bing’ on Gisela^®6 fertilized with 0 and 60 kg N ha⁻¹. Values are means with SE in parentheses (n = 3).

N rate (kg N ha−1)	Trunk cross-sectional area1		Shoot length2			Total leaf area2  (cm2 tree−1)	Cumulative fruit yield3 (kg tree−1)
	Absolute increment (cm2 tree−1)	Relative increment (cm2 cm−2)	Total length (m tree−1)	Mean length (cm shoot−1)	Total number (no. tree−1)
0	41.9 (2.9)	0.4 (0.1)	28.1 (1.5)	19.7 (3.6)	150 (20)	45.6 (1.3)	6.6 (3.0)
60	47.8 (1.9)	0.5 (0.1)	40.9 (4.7)	28.8 (2.3)	142 (5)	71.0 (14.2)	14.6 (3.5)
P-value	0.156	0.148	0.060	0.105	0.704	0.150	0.186

¹Calculated on the basis of the difference between measurements at the beginning of the 2011–12 season and at the end of the 2012–13 season.

²Measurements from destructive harvest of trees at the end of the 2012–13 season.

³Sum of fruit yields of the 2011–12 and 2012–13 seasons.

No differences in tree growth were detected between the N rates (P > 0.05), as expressed by trunk cross-sectional area (TCSA). At the beginning of the 2011–12 season, the mean TCSA was 62.3 ± 10.6 and 64.9 ± 12.9 cm² per tree for the control and N-fertilized trees, respectively; at the end of the 2012–13 season, the respective values were 104.2 ± 7.8 and 104.4 ± 26.7 cm² (data not shown). In terms of seasonal TCSA increment, the absolute and relative values were higher in trees from the N-fertilized treatment, but the differences were not statistically significant (Table 2). The fruit yield was not affected in either the seasonal yield (data not shown; P > 0.10) or the cumulative yield (Table 2). However, the N addition changed the fruit size distribution at harvest (P < 0.05), increasing the proportion of fruit in the range of 24–26 mm in diameter in both seasons, and in the range of 26–28 mm in diameter in the 2012–13 season (data not shown).

Standing crop length

The fine-root dynamics, observed as weekly changes in the standing crop length, showed a significant seasonal variation in each N treatment (P < 0.01). During the first season of measurements (2011–12), a single root peak was registered for the entire soil profile (0–100 cm) in control trees (P < 0.01) (Figure 3). This peak lasted from mid-November to the end of December (7 weeks), coinciding with fruits in the last growth stage and with shoots with declining rates of growth. The peak for the entire soil profile was a result of consecutive short-duration single peaks (2–4 weeks) that occurred below 25 cm depth (P < 0.05) (Figure 3). During 2012–13, control trees also showed a single peak, which started at a similar time as that observed in the previous season, but with a shorter duration (2 weeks). Significant peaks at different soil depths were registered only below 50 cm in depth (P < 0.05) (Figure 3). In contrast, N-fertilized trees have three significant and different peaks for the complete soil profile in both seasons (P < 0.01), occurring from early November to late February (Figure 3). During 2011–12, three significant peaks occurred at intermediate segments in the soil profile (P < 0.05), while only the last two root peaks were registered at 75–100 cm depth, and no peaks were observed in the uppermost layer. A similar behavior was shown during 2012–13 (Figure 3).

Figure 3. .

Seasonal variation in the standing length of white fine roots for the entire soil profile (0–100 cm) and at different soil depths (0–25, 25–50, 50–75 and 75–100 cm) during two growing seasons (2011–12 and 2012–13). Bars are the standard errors (n = 3). Arrows indicate the N applications in the N-fertilized treatment. Dashed lines indicate the occurrence of the different phenological stages: FB, full bloom; FH, fruit harvest; CSG, cessation of shoot growth.

N fertilization also had a significant effect on the magnitude of standing crop length in both growing seasons (P < 0.01), with trees subjected to non-limiting N availability showing a higher mean value at each soil depth (no significant N rate × soil depth interaction; P > 0.05) (Figure 3). The mean values across observation time points and soil layers were respectively 3.13 and 5.94 m m⁻² in 2011–12 for control and N-fertilized trees; and 1.09 and 3.90 m m⁻² in 2012–13. Thus, a noticeable variation across seasons was noted, with the mean standing crop length values 80% higher in 2011–12 than in 2012–13 (5.00 vs 2.80 m m⁻²). Significant differences due to soil depth were observed only in 2011–12, when the standing length was higher in deeper soil layers (50–75 and 75–100 cm) in both N treatments (P < 0.01) (Figure 3).

Fine-root production and mortality rates

On a monthly basis, the production and mortality rates showed a strong seasonal variation (P < 0.01), and temporal differences depended upon the N rate (significant time × N rate interaction; P < 0.05). In trees without N, a single peak in root production was observed in November of each season for the complete soil profile (P < 0.01), and subsequently, the rates sharply decreased to minimal values (<0.05 m m⁻² day⁻¹; Figure 4a). The same temporal behavior was registered at soil layers below 25 cm depth (P < 0.10) (Figure 4c). In trees receiving N, the maximum production rates were maintained from November to January each season for the entire soil profile (P < 0.01) (Figure 4b) and at soil layers located below 25 cm depth (P < 0.10) (Figure 4d). From January onwards, the rates gradually diminished to minimal values in March each season (<0.05 m m⁻² day⁻¹). In turn, the peaks in root mortality rates were more prolonged and followed the peaks in root production rates for both N regimes (Figure 4a and b). In trees without N, the maximum root mortality rates for the complete soil profile were close to fruit harvest (December–January in 2011–12 and November–December in 2012–13; P < 0.10), and then the rates decreased smoothly (Figure 4a). In trees with N application, the maximum mortality rates for the entire soil profile extended until February or March each season (P < 0.10) (Figure 4b). Similarly to the production rates, steady root mortality rates were registered at the surface layer of the soil in both N treatments. At deeper soil layers, consecutive single peaks in root mortality were observed in both treatments (Figure 4c and d). For example, in the control treatment, the maximum mortality rates occurred in November 2011 at 25–50 cm depth, in December 2011 at 50–75 cm depth, and in January at 75–100 cm depth, and slow, steady rates were registered during the rest of the season. In the N-fertilized treatment, the maximum mortality rates were evident from November 2011 to February 2012 at 25–50 cm depth, from December 2011 to January 2012 at 50–75 cm depth, and in February 2012 at 75–100 cm depth.

Figure 4. .

Mean fine-root production and mortality rates for the entire soil profile (0–100 cm depth) (a and b) and at different soil depths (0–25, 25–50, 50–75 and 75–100 cm) (b and c) during two growing seasons (2011–12 and 2012–13). Bars are the standard errors (n = 3). Dashed lines indicate the occurrence of the different phenological stages: FB, full bloom; FH, fruit harvest; CSG, cessation of shoot growth; ELF, early leaf fall.

The application of N fertilization significantly enhanced the rates of root production and mortality in both seasons (P < 0.01), independent of the soil depth (no significant N rate × soil depth interaction; P > 0.05) (Figure 4). The mean production rates during 2011–12 were 0.06 and 0.13 m m⁻² day⁻¹ for the control and N-fertilized treatments, respectively, across soil layers. The corresponding values during 2012–13 were 0.04 and 0.10 m m⁻² day⁻¹. The mean mortality rates were very close to the production rates. As expected, the inter-seasonal variation was significant (P < 0.01), with higher production and mortality rates in 2011–12 than in 2012–13. An effect of the soil depth was detected in 2011–12 (P < 0.01), but not in 2012–13 (P > 0.10) (Figure 4c and d). At this time, both the production and mortality rates were lower at the surface layer (0–25 cm depth).

Total fine-root production and mortality

N addition enhanced the total root production and mortality in each season (P < 0.001), independent of the soil depth (no significant N rate × soil depth interaction; P > 0.05) (see Figure S2 available as Supplementary Data at Tree Physiology Online). During 2011–12, 16.9 and 35.2 m m⁻² of fine roots were produced for the entire soil profile in control and N-fertilized treatments, respectively; of that total, 93% (15.7 m m⁻²) and 95% (33.4 m m⁻²) turned brown or disappeared at the end of the season. During 2012–13, trees without and with N respectively produced 9.8 and 27.5 m m⁻² of fine roots, and the corresponding mortality values were 7.9 m m⁻² (80%) and 25.4 m m⁻² (92%) (see Figure S2 available as Supplementary Data at Tree Physiology Online). Both variables were positively correlated, as demonstrated by a linear regression analysis (see Figure S2 available as Supplementary Data at Tree Physiology Online) showing a single line to be statistically suitable for representing the data from different seasons, soil depths and N treatments (Y = 0.939X; R² = 0.99, P < 0.001). Also, the total production was higher during the first season (P < 0.05), with 49% more root length produced in 2011–12 than in 2012–13 (27.9 vs 18.7 m m⁻²) across soil depths and N rates (see Figure S2 available as Supplementary Data at Tree Physiology Online).

The soil depth was a significant factor only during 2011–12 (P < 0.01), when the 0–25 cm layer had a significantly lower total root production and mortality, independent of the N treatment (data not shown). However, in terms of the relative distribution, the effect of soil depth on the total root production and mortality depended upon the N fertilization in both seasons (significant N rate × soil depth interaction; P < 0.05) (Figure 5). In 2011–12, control trees had a total root production and mortality concentrated below 50 cm depth, without differences between the 50–75 and 75–100 cm layers. The 0–25 cm layer accounted for <1% of the total production and mortality in these trees. Trees with N had similar proportions of total root production and mortality at layers below 25 cm depth, and only the surface layer had a lower proportion (<10%). Similar results were found during the 2012–13 season (Figure 5).

Figure 5. .

Total root production and mortality by depth as a fraction of the production and mortality for the entire soil profile during two growing seasons (2011–12 and 2012–13). Bars are the standard errors (n = 3). Different lower-case letters denote differences between the soil depths in the 0 kg N ha⁻¹ treatment during both seasons, and different upper-case letters denote differences between the soil depths in the 60 kg N ha⁻¹ treatment during both seasons (Tukey’s test; P < 0.05).

Fine-root turnover and lifespan

N fertilization diminished the lifespan (P = 0.055) and hastened the turnover rate (P = 0.059) of fine roots in 2011–12, and this effect was independent of the soil depth (no significant N rate × soil depth interaction; P > 0.05) (Table 3). No differences due to N regimes were detected in 2012–13 (P > 0.10). In turn, the effect of the soil depth was significant in both studied seasons (P < 0.01), with fine roots located in the surface soil layer having higher turnover rates and shorter lifespans than fine roots at layers below 25 cm depth (Table 3). Additionally, the inter-seasonal variation in both parameters was very high (P < 0.01), with values of fine-root turnover rates 50% higher and values of lifespan 50% smaller in 2012–13 than in the 2011–12 season (Table 3).

Table 3. .

Fine-root turnover and lifespan during two growing seasons (2011–12 and 2012–13) in trees of sweet cherry ‘Bing’ on Gisela^®6 fertilized with 0 and 60 kg N ha⁻¹. Values are means with SE in parentheses (n = 3). Different superscript letters denote differences between the N treatments or between the soil depths as applicable (Tukey’s test; normal letters, P < 0.05; underlined letters, P < 0.10).

Soil depth (cm)	Root turnover rate (season−1)			Lifespan (days)
	0 kg N ha−1	60 kg N ha−1	Mean	0 kg N ha−1	60 kg N ha−1	Mean
2011–12
0–25	2.15 (0.08)	2.04 (0.06)	2.09 (0.05)b	137 (5)	145 (4)	142 (3)a
25–50	1.19 (0.12)	1.66 (0.17)	1.47 (0.15)a	251 (26)	182 (20)	209 (22)ab
50–75	1.26 (0.07)	1.74 (0.17)	1.55 (0.15)a	236 (12)	173 (17)	198 (19)ab
75–100	1.23 (0.20)	1.45 (0.28)	1.36 (0.17)a	247 (40)	219 (40)	230 (26)b
Mean	1.46 (0.10)a	1.72 (0.08)b		218 (14)b	179 (11)a
2012–13
0–25	3.57 (0.52)	2.98 (0.42)	3.27 (0.32)b	68 (10)	82 (11)	75 (7)a
25–50	2.14 (0.03)	2.39 (0.03)	2.26 (0.06)a	112 (2)	100 (1)	106 (3)b
50–75	2.53 (0.31)	2.27 (0.10)	2.40 (0.16)a	97 (13)	105 (5)	101 (7)b
75–100	2.39 (0.31)	2.02 (0.17)	2.20 (0.18)a	103 (14)	120 (11)	112 (9)b
Mean	2.57 (0.20)a	2.36 (0.13)a		97 (7)a	103 (5)a

Influence of exogenous and endogenous factors on fine-root production and mortality

The total length-based fine-root production and mortality were significantly correlated to the mean N availability during the growing season. Consistently, the total root production (r = 0.90; P < 0.01) and mortality (r = 0.91; P < 0.01) for the entire soil profile (0–100 cm) increased with the mean seasonal soil N availability at 0–20 cm depth (see Table S1 available as Supplementary Data at Tree Physiology Online). The N availability at soil depths other than 0–20 cm was not as good a predictor. The total production and mortality at different soil layers was also significantly and positively correlated with the mean seasonal N mineral at 0–20 cm depth, but with lower correlation coefficients (see Table S1 available as Supplementary Data at Tree Physiology Online). Despite these results, the seasonal dynamics of fine-root production and mortality were not linearly related to the seasonal dynamic of soil mineral N, as expressed by non-significant coefficients of correlation between the monthly mean soil mineral N and the monthly production and mortality rates (P > 0.10; data not shown).

The monthly rates of fine-root production at different soil layers were, in general, not correlated with the monthly mean soil temperature at 20 and 40 cm depths. Meanwhile, the correlations with the monthly rates of fine-root mortality were positive and significant along the entire soil profile (r = 0.43, P < 0.01 and r = 0.46, P < 0.01 for soil temperature at 20 and 40 cm depth, respectively), as well as at practically each soil layer (see Table S2 available as Supplementary Data at Tree Physiology Online). Similarly, the monthly mean soil water content at 20 and 40 cm depth was not correlated with production rates at 0–100 cm depth or at the different soil layers of soil profile. In turn, the correlations with mortality rates were negative and significant, although exclusively with soil water content at 20 cm depth (see Table S2 available as Supplementary Data at Tree Physiology Online).

In general terms, quadratic equations were statistically suitable for describing the relationships between the monthly cumulative production and mortality and the cumulative soil temperature at 20 cm depth, accounting for 80–98% of the production/mortality variability (Figure 6). A clear separation of the N treatments was observed, demonstrating the strong effect of an improved soil N availability on fine-root production and mortality throughout the growing season (Figure 6).

Figure 6. .

Relationship between the cumulative soil temperature (°C day⁻¹; base temperature, 7 °C) at 20 cm depth with (a) the monthly cumulative root production and (b) the monthly cumulative root mortality for the entire soil profile (0–100 cm) during two growing seasons (2011–12 and 2012–13). Lines are the curves fitted to data. Sy.x is the standard deviation of the residuals.

Endogenous factors, expressed as the shoot and fruit phenology, showed consistent relationships with fine-root production from season to season, but depending upon the N treatment (Figure 7). In trees with N, a peak of root production (P < 0.05) was observed when fruit growth was minimal, probably during pit hardening (end-October to mid-November). The shoot growth showed a similar behavior (P < 0.05), although more precocious. Later, the root production and shoot growth rates diminished when fruit growth was resumed (Phase II), but the reduction in growth rate was faster in the roots. Finally, in December, a new root peak (P < 0.05) was registered after fruit removal and with shoots growing at minimal rates. In trees with no N application, a single root production peak was detected after the pit hardening stage (P < 0.05) and was preceded by a shoot growth peak (P < 0.05). No significant root production peaks were observed after fruit harvest or cessation of shoot growth (Figure 7).

Figure 7. .

Growth rates of shoots and fruits, and fine-root production rates for the entire soil profile (0–100 cm depth) during the period of active growth of the aerial organs. Data from the 2011–12 season. Bars are the standard errors (n = 3). Dashed lines indicate the occurrence of the different phenological stages: FB, full bloom; FH, fruit harvest.

Discussion

The effects of N fertilization, and consequently of an improved soil N supply, were evident on the below- and above-ground growth of trees. The N influence on shoot growth was consistent from year to year. The shoot growth dynamic changed, resulting in a higher mean final shoot length in trees receiving N. In contrast, the TCSA and seasonal TCSA increment did not vary despite N treatments having been applied for six consecutive seasons. N application did not affect fruit yield, but the proportion of fruit >26 mm in diameter increased. An increase in fruit size due to N fertilization is not uncommon in fruit species (Klein et al. 1989, Wargo et al. 2003). This effect has been primarily attributed to an increase in cell numbers per fruit in apple (Cheng et al. 2007, Xia et al. 2009), highlighting the importance of a high N supply during early fruit development.

N application to soil drastically and consistently altered the seasonal growth dynamic of fine roots in irrigated and fruit-bearing sweet cherry trees, both expressed as production and mortality rates and as changes in standing crop length. In terms of the standing crop length, the number and magnitude of peaks for the entire soil profile differed between N-fertilized and control trees, but the peaks were coordinated in time from season to season within each N treatment (Figure 3). In no-N trees, a single peak was registered close to fruit harvest and to cessation of shoot growth in both seasons. In N-fertilized trees, three significant peaks were observed for the complete soil profile, which lasted for 2–4 weeks. In general terms, these peaks took place shortly before fruit harvest (November), around the cessation of shoot growth (end-December), and late in the growing season (January or February). During winter, the standing crop length was minimal in both treatments (∼0.5 m m⁻²), indicating a very low survival of white roots and/or a very low production of new roots (Figure 3). Although the seasonal patterns were quite consistent from season to season, it is very difficult to generalize or even to compare with other studies. The wide range of root phenological patterns registered in temperate fruit trees and the multiplicity of factors affecting them in time and space makes this a difficult task.

In terms of monthly rates, trees from both N regimes showed a concentrated pattern of root production, although with differences in the persistence and magnitude of maximum rates or peaks (Figure 4). Trees with an ample soil N supply maintained rates of production for the entire soil profile at maximum values during three consecutive months in each season (November–January; Figure 4b). Subsequently, root production rates gradually diminished. In turn, control trees were able to sustain this peak during a single month (November), and later, production rates dropped sharply to minimal values (Figure 4a).

Considering that the seasonal dynamics of soil N availability (P > 0.10; data not shown), soil temperature and soil water content (see Table S2 available as Supplementary Data at Tree Physiology Online) were not correlated with the seasonal dynamics of fine-root production rates, the significant differences in root phenology between N regimes suggest a primary dependence of root production upon non-environmental factors, e.g., source–sink relationships. In fact, soil N availability influenced the source–sink relationships as expressed by a significant effect on the leaf photosynthetic capacity. The leaves of trees receiving N had higher mean net assimilation rates during the growing season than no-N trees (6.52 vs 7.57 µmol CO₂ m⁻² s⁻¹; P < 0.10), and this effect was particularly observed at the end of the season (data not shown). Apparently, N-fertilized trees had a delayed onset of leaf senescence, maintaining green leaves with higher levels of photosynthesis for a longer time, which in turn allowed them to support higher rates of root production, as has been suggested by King et al. (1999) for poplar trees. In this context, it is well known that autumnal leaf senescence in deciduous fruit trees is accompanied by a decrease in N content in leaves due to an N withdrawal to perennial organs (Millard and Grelet 2010), which reduces their photosynthetic capacity (Xu and Baldocchi 2003), and that the onset of leaf senescence is regulated, in part, by exogenous factors like N limitations (Lim et al. 2007, Angelis et al. 2012).

Further evidence that root production is mainly under endogenous control is the temporal synchronization of the peaks of root production rates at soil layers below 25 cm depth in both treatments (Figure 4). If soil temperature was the most determining factor, then the peaks of root production along the soil profile should follow the spatial pattern of soil temperature peaks, characterized by a time lag imposed by soil thermal conductivity (Pregitzer and King 2005), which was not observed. These results do not agree with studies on forest and grass species reporting positive associations between fine-root production rates and soil (air) temperature (Tierney et al. 2003, Majdi and Öhrvik 2004, Steinaker et al. 2010, Fukuzawa et al. 2013, Rytter 2013). However, none of these studies included fruit tree species, and most were performed in climates with cooler temperatures than that of the temperate climate of central Chile (e.g., boreal, cool temperate and continental climates). It is likely that the soil temperature exerts strong effects on root production at extremes, e.g., below 10 °C and above 35 °C for temperate plants (Comas et al. 2005), which is not the case for the agro-ecosystem of this study.

If the primary dependence of the root production rates on source–sink relationships proposed here is correct, then it should be clearly expressed during the fruit growth period. In sweet cherry trees, this period is very limited (63–70 DAFB in this study) and is simultaneous with shoot and root growth; thus, a strong and early competition for carbohydrates between tree organs is expected. Indeed, decreases in daily fruit growth rate, e.g., during pit hardening and after fruit harvest, coincided with increases in the daily root production rate and daily shoot growth rate in trees with an ample N supply (Figure 7). Therefore, the effect of source–sink relationships on root production in sweet cherry trees is particularly strong during fruit development, as has been extensively reported for peach trees (Williamson and Coston 1989, Glenn and Welker 1993, Grossman and DeJong 1995, Inglese et al. 2002, Mimoun and DeJong 2006, Basile et al. 2007, Abrisqueta et al. 2008). In trees without N, the alternating growth pattern of fruit and roots was not clearly observed (Figure 7). Moreover, a small root peak was observed during fruit growth, and no peaks were registered after fruit removal or after the cessation of shoot growth. It is likely that control trees experienced a permanent source limitation, first driven by the size of the N storage pool (defined by N supply in previous seasons; Millard and Grelet 2010) and then by a limited N uptake from the soil, thus preventing the resumption of root growth when competition for C was low or null.

Fine-root production and mortality were simultaneously occurring processes (Figure 4). However, peaks in root mortality rates were more prolonged and followed peaks in root production by a lag of 1 month. In control trees, the maximum root mortality rates were observed during 2 months of each season. In trees with N addition, the maximum values extended until February or March, coinciding with the period of highest air and soil temperatures (Figure 1a and b). After peaks, the mortality rates progressively diminished in both N regimes and, unlike the production rates, were seldom equal to zero even during the winter months (Figure 4). Additionally, the root mortality rates peaked progressively later with increasing depth in both treatments, indicating a strong influence of the soil temperature (Figure 4c and d).

As an overall effect, N addition enhanced the length-based root production rates throughout each growing season, coinciding with various studies conducted with forest species (Pregitzer et al. 1993, 2000b, Kubiske et al. 1998, Majdi 2001, King et al. 2002, Adams et al. 2013, Noguchi et al. 2013). However, evidence of decreased or unchanged root production also exists. Mortality rates were also increased by N, and were very close to production rates. The production and mortality rates, as well as the SRL values were significantly higher in the first season (2011–12), which may have resulted from an excessive root proliferation during the season immediately after the installation of the rhizotron. Similar observations have been performed in minirhizotron studies, and root proliferation has been attributed to root pruning and to nutrient release in microsites near the newly installed minirhizotron (Joslin and Wolfe 1999, Wells and Eissenstat 2003). In minirhizotron studies, a waiting period of 6–12 months between minirhizotron installation and image collection is recommended to avoid artifacts in root data and analysis (Johnson et al. 2001). In this study, we started to collect the root images 1 month after rhizotron installation. However, two growing seasons of root observations were carried out to deal with this limitation. Moreover, the results from the second season were similar in terms of temporal pattern (but not in magnitude), confirming the validity and reliability of the findings of this study.

The N effect on the dynamics and magnitude of root production rates resulted in a higher total length-based fine-root production and mortality at the end of each season (see Figure S2 available as Supplementary Data at Tree Physiology Online). Consequently, the total fine-root production and mortality were correlated with the mean seasonal N availability, particularly at 20 cm soil depth, and this parameter alone accounted for 80% of the variation in both variables (see Table S1 available as Supplementary Data at Tree Physiology Online). The total root mortality was proportional to production according to a significant linear regression (see Figure S2 available as Supplementary Data at Tree Physiology Online). The slope of the adjusted line to the data of different seasons, soil depths and N rates was 0.94, indicating that practically all white fine roots born during each season turned brown or disappeared within the season, as has been previously reported in apple trees (Psarras et al. 2000).

Across N rates, the mean total fine-root production and mortality were concentrated in the deeper soil layers in 2011–12 (Figure 4). In relative terms, the soil depth effect was consistent in both seasons. In control trees, 70–80% of the total fine-root production and mortality was concentrated below 50 cm depth, and the surface layer was practically unexplored. In contrast, trees receiving N had similar proportions of total root production and mortality in different soil layers, and only in the first season, the surface layer has a significantly lower participation (Figure 5). These distributions of the root production/mortality are unusual, particularly that of the control trees, in comparison with studies with forest trees reporting a higher length-based fine-root production and mortality in the surface soil than at deeper soil layers (Burton et al. 2000, Comas et al. 2005, Yunhuan et al. 2006) . However, these studies were carried out with the minirhizotron at soil depths lower than 50 cm or under colder climates. In contrast, the studies carried out under similar conditions to our study, i.e., irrigated fruit species, warm climates, deep soil profile, high tree density, report fine-root distributions similar to that of N-fertilized trees (Atkinson and Wilson 1980, Basile et al. 2007, Abrisqueta et al. 2008). Probably, the high temperatures experienced in the summer months in our experimental area limit the fine-root growth in surface soil layers as well as the soil N mineralization, forcing the roots of control trees to find N in deeper soil layers with more water and mild temperatures.

The length-based turnover rates of the fine roots of irrigated sweet cherry trees (1.5–2.6 season⁻¹; Table 3) were in line with the values for a temperate forest calculated from global datasets (Finér et al. 2011) and for loblolly pine (2.02–2.49 year⁻¹; King et al. 2002). Fertilization with N diminished the lifespan and consequently hastened the turnover rate of fine roots, although only in 2011–12. At this time, N-fertilized trees showed mean lifespan values that were 20% lower and mean turnover rates that were 18% higher than for control trees (Table 3). These results are consistent with the hypothesis proposed by Burton et al. (2000) and with simulations performed by Eissenstat et al. (2000) using a cost–benefit model (Eissenstat and Yanai 1997). In both cases, a shorter lifespan in response to N addition would be explained by higher respiratory rates derived from a higher N concentration in roots, leading to a rapid depletion of cytosolic sugar and faster root turnover. In fact, the N concentration in fine roots was significantly higher in N-fertilized trees harvested in previous (Bonomelli and Artacho 2013) and subsequent seasons (data not published) at the same experimental site.

Fine roots located in the surface layer had shorter lifespans and consequently faster turnover rates than in deeper layers during both studied seasons (Table 3). Given that the effect of soil depth did not interact with the N rates, these results were caused by factors other than the N supply. Baddeley and Watson (2005) mentioned as possible causes the exposure of surface roots to higher fluctuations in soil temperature and water content, or even to higher levels of herbivory. Another explanation is a higher root respiratory activity associated with the higher temperatures (Eissenstat et al. 2000, McCormack and Guo 2014) commonly registered at surface soil during spring–summer (Pregitzer and King 2005).

In summary, an improved soil N availability increased the rates of fine-root production and mortality, changed the fine-root phenology, and enhanced the seasonal root production and mortality of irrigated sweet cherry trees during the two seasons of this study. Less consistent was the N effect on turnover rates, which suggests that a greater standing length in trees with an ample N supply was reached mainly as a result of higher production rates rather than by higher turnover rates. The seasonal dynamics of root production and mortality and soil N availability were not correlated, indicating a basal influence of N fertilization on root processes rather than immediate and transitory effects after N applications. Root production and mortality peaked during warmer months in both N regimes, although mortality peaks followed production peaks. Only root mortality rates were correlated with soil temperature and soil water content. Root production rates were primarily under endogenous control, specifically through source–sink relationships, which in turn were affected by N supply through changes in leaf photosynthetic level. This was traduced in an alternating growth pattern of fruits and roots during the fruiting period. Therefore, exogenous and endogenous factors interacted to control the fine-root dynamics of irrigated sweet cherry trees.

Supplementary data

Supplementary data for this article are available at Tree Physiology Online.

Conflict of interest

None declared.

Funding

This study was funded by Postgraduate Thesis in Industry Program of CONICYT, Project 78111202, with financial and operational support of Vivero Rancagua S.A.