Abstract

Non-structural carbohydrates (NSC) in plant tissue are frequently quantified to make inferences about plant responses to environmental conditions. Laboratories publishing estimates of NSC of woody plants use many different methods to evaluate NSC. We asked whether NSC estimates in the recent literature could be quantitatively compared among studies. We also asked whether any differences among laboratories were related to the extraction and quantification methods used to determine starch and sugar concentrations. These questions were addressed by sending sub-samples collected from five woody plant tissues, which varied in NSC content and chemical composition, to 29 laboratories. Each laboratory analyzed the samples with their laboratory-specific protocols, based on recent publications, to determine concentrations of soluble sugars, starch and their sum, total NSC. Laboratory estimates differed substantially for all samples. For example, estimates for Eucalyptus globulus leaves (EGL) varied from 23 to 116 (mean = 56) mg g−1 for soluble sugars, 6–533 (mean = 94) mg g−1 for starch and 53–649 (mean = 153) mg g−1 for total NSC. Mixed model analysis of variance showed that much of the variability among laboratories was unrelated to the categories we used for extraction and quantification methods (method category R2 = 0.05–0.12 for soluble sugars, 0.10–0.33 for starch and 0.01–0.09 for total NSC). For EGL, the difference between the highest and lowest least squares means for categories in the mixed model analysis was 33 mg g−1 for total NSC, compared with the range of laboratory estimates of 596 mg g−1. Laboratories were reasonably consistent in their ranks of estimates among tissues for starch (r = 0.41–0.91), but less so for total NSC (r = 0.45–0.84) and soluble sugars (r = 0.11–0.83). Our results show that NSC estimates for woody plant tissues cannot be compared among laboratories. The relative changes in NSC between treatments measured within a laboratory may be comparable within and between laboratories, especially for starch. To obtain comparable NSC estimates, we suggest that users can either adopt the reference method given in this publication, or report estimates for a portion of samples using the reference method, and report estimates for a standard reference material. Researchers interested in NSC estimates should work to identify and adopt standard methods.

Introduction

Non-structural carbohydrates (NSC) are products of photosynthesis, provide substrates for growth and metabolism and can be stored by the plant. Consequently, NSC play a central role in plant response to the environment (Chapin et al. 1990, Kozlowski 1992). Major theories of plant defense and growth such as the ‘growth–differentiation balance hypothesis’ (Loomis 1932), the ‘carbon/nutrient hypothesis’ (Bryant et al. 1983), revisions to the ‘hydraulic limitation hypothesis’ (Ryan et al. 2006) and the ‘carbon limitation hypothesis’ (Körner 2003) all outline a role for NSC, but that role has yet to be firmly established or rejected. In more recent years, NSC of woody plants have received wider attention in the context of understanding drought-induced mortality (Grünzweig et al. 2008, McDowell et al. 2008, Galiano et al. 2011, Muller et al. 2011, Piper 2011, Adams et al. 2013, Duan et al. 2013, Hartmann et al. 2013, Mitchell et al. 2013, 2014, O’Brien et al. 2014, Sevanto et al. 2014, Dickman et al. 2015), altitudinal boundaries for forests (Hoch et al. 2002, Hoch and Körner 2003, Handa et al. 2005, Li et al. 2008, Fajardo et al. 2011, 2012, 2013, Fajardo and Piper 2014), growth limitation (Sala et al. 2010, 2012, Piper and Fajardo 2011, Palacio et al. 2014) and plant survival under poor resource conditions (Kobe 1997, Strauss and Agrawal 1999, Haukioja and Koricheva 2000, Lusk and Piper 2007, Quentin et al. 2011, Piper and Fajardo 2014).

Several major questions about the role and regulation of stored carbohydrates in woody plants remain unanswered, such as their role in indicating plant carbon balance, helping plants cope with stress, and whether control of storage and use is active, passive or more complex (Chapin et al. 1990, Sala et al. 2011, 2012, Wiley and Helliker 2012). The many uncertainties about how NSC are involved in the regulation of whole-tree carbon metabolism make predictions of growth and productivity under environmental change difficult (Ryan 2011).

Many carbohydrates can comprise NSC such as monosaccharides (glucose and fructose), disaccharides (sucrose), polysaccharides (starch and fructans), oligosaccharides (raffinose) and sugar alcohols (inositol, sorbitol and mannitol) (Rastall 1990, Stick and Williams 2010). Sucrose, fructose and glucose are generally, but not always, the predominant soluble sugars, and starch is the pivotal non-soluble longer-term storage compound (Mooney 1972, Chapin et al. 1990); many studies focus on these four carbohydrates while measuring plant NSC. The diversity of carbohydrates and matrices (tissue structural and biochemical characteristics), and the search for reliable and inexpensive methods that can be used for the large number of samples in environmental plant physiology studies, has led to the development of many analytical methods to determine the identity and amount of carbohydrates in plant tissue (Table 1, Table S1 available as Supplementary Data at Tree Physiology Online; Gomez et al. 2003). Within any given plant species, a wide range of NSC values have been reported in different studies (Table 2). Potential explanations for these differences include plant age and growing conditions, but the extraction and quantification methods may also have a major impact on the results (Rose et al. 1991, Chow and Landhäusser 2004). For 8–12-month-old Eucalyptus globulus saplings, leaf total NSC concentration varied between 28 and 224 mg g−1 when measured using three different soluble sugar and starch extraction methods, and three different quantification methods (Table 2). Studies have also used the same extraction and assay methods to analyze different tissues (leaves, stems, roots) that consist of different matrices (Table 2), despite evidence that different matrices can have a profound impact on the analytical results (Smeraglia et al. 2002, Matuszewski et al. 2003, Thompson and Ellison 2005, Santiago da Silva et al. 2012). For example, the phenolics and tannins in many conifer needles can interfere with enzymatic/colorimetric techniques (Ashwell 1957), but not all plant tissues contain these chemicals. Given such variability in NSC estimates, we believe that there is an urgent need to compare estimates of NSC of standard samples for different laboratories around the world, with the laboratories using the same methods as in their recent publications.

Table 1.

Summary of the primary solvents and assays used for extraction and quantification methods to estimate soluble sugars and starch in five plant materials. The method categories also vary in the number of extractions, duration, temperature and standards. For further details on each specific method, please refer to Tables S1 and S2 available as Supplementary Data at Tree Physiology Online. AA, α-amylase; Amylo., amyloglucosidase; DMSO, dimethyl sulfoxide; EtOH, ethanol; FRUC, fructose; G6PDH, glucose-6-phosphate dehydrogenase; GHK, glucose hexokinase; GLUC, glucose; GOPOD, glucose oxidase/peroxidase-o-dianisidine; H2SO4, sulfuric acid; HCl, hydrochloric acid; HClO4, perchloric acid; 1H-NMR, proton nuclear magnetic resonance; HPAEC-PAD, high-performance anion exchange chromatography-pulsed amperometric detection; HPLC, high-performance liquid chromatography; KOH, potassium hydroxide; MCW, methanol : chloroform : water; NA: not attributed; NaOH, sodium hydroxide; PGI, phosphoglucose isomerase; SUC, sucrose. Soluble sugar methods include 31 laboratories and starch methods 28 laboratories. Two laboratories have used two methods to estimates the soluble sugars, whereas one laboratory did not estimate starch.

 Strength No. of extraction Combination Duration (min) Temperature (°C) No. of laboratories 
Soluble sugars 
 Extraction methods 
  EtOH or MeOH 70–80%1 1–5 EtOH or W 2–60 60–100 19 
  W – 1–3 – 10–60 65–100 
  MCW – 1–3 – 5 to overnight 4–60 
 Absorbance Reagents Standards No. of laboratories 
 Quantification methods 
  HPLC – – Trehalose or mannitol 
  HPAEC-PAD – – GLUC, FRUC, SUC 
  1H-NMR – – GLUC, FRUC 
  Enzymatic 340 G6PDH + GHK + PGI + Invertase GLUC, FRUC, SUC 10 
  Colorimetric 620 Anthrone GLUC 
 490 Phenol GLUC 
 Duration (min) Temperature (°C) No. of laboratories Starch 
 Gelatinization methods 
  None – – 
  NaOH 30–180 50–100 
  DMSO 100 
  KOH 30 95 
  EtOH 30 100 
  AA 30 85–90 
  Others2 NA–90 120 
 Reagent/enzyme No. of extraction Temperature (°C) Duration (min h−1No. of laboratories 
 Digestion/extraction methods 
  Acid HClO4 Room temperature 16–20 h 
H2SO4  Autoclave 3.5 min 
HCl  100 6 min 
  Enzymatic Amylo. 1 or 2 45–100 30 min to 24 h 16 
AA + amylo. 55–100 3–30 min 
37–100 1 min to 16 h 
 Absorbance Reagent Standard No. of laboratories 
 Quantification methods 
  HPLC – – GLUC 
  HPAEC – – GLUC 
  Enzymatic 340 G6PDH + HK GLUC 10 
  Colorimetric 620–630 Anthrone GLUC 
 490 Phenol GLUC 
 510–5253 GOPOD GLUC 
 Strength No. of extraction Combination Duration (min) Temperature (°C) No. of laboratories 
Soluble sugars 
 Extraction methods 
  EtOH or MeOH 70–80%1 1–5 EtOH or W 2–60 60–100 19 
  W – 1–3 – 10–60 65–100 
  MCW – 1–3 – 5 to overnight 4–60 
 Absorbance Reagents Standards No. of laboratories 
 Quantification methods 
  HPLC – – Trehalose or mannitol 
  HPAEC-PAD – – GLUC, FRUC, SUC 
  1H-NMR – – GLUC, FRUC 
  Enzymatic 340 G6PDH + GHK + PGI + Invertase GLUC, FRUC, SUC 10 
  Colorimetric 620 Anthrone GLUC 
 490 Phenol GLUC 
 Duration (min) Temperature (°C) No. of laboratories Starch 
 Gelatinization methods 
  None – – 
  NaOH 30–180 50–100 
  DMSO 100 
  KOH 30 95 
  EtOH 30 100 
  AA 30 85–90 
  Others2 NA–90 120 
 Reagent/enzyme No. of extraction Temperature (°C) Duration (min h−1No. of laboratories 
 Digestion/extraction methods 
  Acid HClO4 Room temperature 16–20 h 
H2SO4  Autoclave 3.5 min 
HCl  100 6 min 
  Enzymatic Amylo. 1 or 2 45–100 30 min to 24 h 16 
AA + amylo. 55–100 3–30 min 
37–100 1 min to 16 h 
 Absorbance Reagent Standard No. of laboratories 
 Quantification methods 
  HPLC – – GLUC 
  HPAEC – – GLUC 
  Enzymatic 340 G6PDH + HK GLUC 10 
  Colorimetric 620–630 Anthrone GLUC 
 490 Phenol GLUC 
 510–5253 GOPOD GLUC 
1

Strength used for the first extraction. With further extractions, strength varied between 30 and 80% for ethanol, and 0% when water is used.

2

Includes shaking, autoclaving, boiling and ultrasound.

3

Method using the Megazyme® kit.

Table 2.

Procedures for soluble sugar, starch measurements and NSC concentrations and mean values for E. globulus and Prunus persica and for Pinus edulis for various environmental response studies. AA, α-amylase; Amylo., amyloglucosidase; BA, β-amylase; DMSO, dimethyl sulfoxide; Dig., digestion; Enz., enzymatic; EtOH, ethanol; Extr., extraction; FRUC, fructose; GLUC, glucose; GOPOD, glucose oxidase/peroxidase-o-dianisidine; HCl, hydrochloride acid; HPLC, high-performance liquid chromatography; L, leaf; MCW, methanol : chloroform : water; mo: month-old; Quant., quantification; R, roots; spec, spectrophotometry; S, stem; St, starch; SUC, sucrose; TSS, total soluble sugars; UPLC, ultra performance liquid chromatography; W, water; yo, year-old.

References Age Tissue Sample weight (mg) Soluble sugars
 
Starch
 
Concentration (mg g−1) in the literature
 
Extr. Quant. (assay) Dig. Quant. (assay) GLUC FRUC SUC TSS St Total NSC 
E. globulus 
Shvaleva et al. (2006) ∼12 mo 20 EtOH Spec. (anthrone) HCl Spec. 620    72–83 49–56 115–117 
  50        32–45 29–32 78–88 
Eyles et al. (2009a) 11 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    105 94 199 
          40 79 118 
          33 100 132 
Eyles et al. (2009b) ∼16 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    46 93 140 
Merchant et al. (2010) ∼12 mo 40 MCW GC   12   
O’Grady et al. (2010) >6 yo L (at 7 m high) 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    56 64 120 
  L (at 15 m high)         19 37 56 
Quentin et al. (2010) ∼8 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    93–106 37–39 130–145 
Pinkard et al. (2011) ∼3–4 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    142 93 187 
Quentin et al. (2011) >6 yo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)      145 
            60 
            63 
Barry et al. (2012) 18 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    60 16 76 
          24 32 
          28 40 67 
Drake et al. (2013) 100 EtOHx2 Spec 630 (anthrone)      6–14   
  R (tap)         7–16   
Duan et al. (2013) 8 mo 20 EtOHx2 + W Spec 620 (anthrone) AA + amylo. Spec. 515    83–90 33–140 117–224 
          32–60 2–8 35–62 
          10–24 1–2 12–25 
Eyles et al. (2013) 7 mo 50 EtOHx1 UPLC Amylo. Spec. 490 (phenol) 18 22 54 92 146 
Mitchell et al. (2013) 6 mo 20 EtOHx2 + W Spec 620 (anthrone) AA + amylo. Spec. 515 (GOPOD)    85 120 206 
          20 13 33 
          46 30 76 
Gauthier et al. (2014)1 <6 mo EtOHx3 Enz. Amylo. Spec 515 (GOPOD)    10 17 
P. persica 
Moing et al. (1992) 2 mo EtOHx2 HPLC Amylo. HPLC 11.1 5.69 36.7 95 89 184 
Nii (1997) 37–38 yo EtOH Spec (anthrone)      78 77 155 
Tworkoski et al. (1997) 5–6 yo 200 EtOH HPLC Amylo. Spec.    38–158 33–48 86–191 
          44–77 39–45 83–122 
Escobar-Gutiérrez et al. (1997) 2.5 mo EtOHx2 HPLC Amylo.  39 10 53 215 135 350 
Inglese et al. (2002) 3 yo 150 EtOH Enz. Amylo. Enz.     6–9  
Graham (2002) 2 mo 50 MCW HPLC Amylo. Spec. 450 16 57 52 109 
       11 54 33 88 
       20 15 106 26 132 
Leite et al. (2004) 11 yo S (Oct) 10 EtOHx2 HPLC Amylo. Enz. 27 69 65.5 134 
  S (Feb)      14 35 74 16 90 
Bonhomme et al. (2005) 4 yo 10 EtOHx2 HPLC Amylo. Enz. 15 28 72 12 84 
Gordon et al. (2006) 2 yo HPLC Amylo.      150 
Dichio et al. (2007) >3 yo EtOH Spec. 625 (anthrone) Amylo. Spec. 425 (GOPOD)    120 125 
          100 10 110 
          240 60 300 
Li et al. (2007)2 5 yo 15 000 EtOHx3 HPLC Amylo. Spec. (GOPOD) 51 25 76 
Cheng et al. (2009)2 8 yo 15 000 EtOHx3 HPLC Amylo. Spec. (GOPOD) 11 43 23 65 
Weibel et al. (2008) 4–5 yo  EtOH Spec. (anthrone)        260 
           160 
            190 
P. edulis 
Adams et al. (2013)3 15–25 yo 12 Enz. Amylo. Enz.    10–56 0–185 19–216 
Anderegg and Anderegg (2013) 10–15 yo EtOH + MCW Spec 595 BA + amylo. Spec. 595 5–10   0–30  30–60 
       10–28   8–21  45–95 
Dickman et al. (2015)3 L (2007) 12 Enz. Amylo. Enz. 10 16 
  L (2008)      
  L (2009)      12 16 
Sevanto et al. (2014)3 15–25 yo 12 Enz. Amylo. Enz. 13–27 1–19 27–36 6–31 36–59 
References Age Tissue Sample weight (mg) Soluble sugars
 
Starch
 
Concentration (mg g−1) in the literature
 
Extr. Quant. (assay) Dig. Quant. (assay) GLUC FRUC SUC TSS St Total NSC 
E. globulus 
Shvaleva et al. (2006) ∼12 mo 20 EtOH Spec. (anthrone) HCl Spec. 620    72–83 49–56 115–117 
  50        32–45 29–32 78–88 
Eyles et al. (2009a) 11 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    105 94 199 
          40 79 118 
          33 100 132 
Eyles et al. (2009b) ∼16 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    46 93 140 
Merchant et al. (2010) ∼12 mo 40 MCW GC   12   
O’Grady et al. (2010) >6 yo L (at 7 m high) 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    56 64 120 
  L (at 15 m high)         19 37 56 
Quentin et al. (2010) ∼8 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    93–106 37–39 130–145 
Pinkard et al. (2011) ∼3–4 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    142 93 187 
Quentin et al. (2011) >6 yo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)      145 
            60 
            63 
Barry et al. (2012) 18 mo 50 EtOHx1 Spec. 490 (phenol) Amylo. Spec. 490 (phenol)    60 16 76 
          24 32 
          28 40 67 
Drake et al. (2013) 100 EtOHx2 Spec 630 (anthrone)      6–14   
  R (tap)         7–16   
Duan et al. (2013) 8 mo 20 EtOHx2 + W Spec 620 (anthrone) AA + amylo. Spec. 515    83–90 33–140 117–224 
          32–60 2–8 35–62 
          10–24 1–2 12–25 
Eyles et al. (2013) 7 mo 50 EtOHx1 UPLC Amylo. Spec. 490 (phenol) 18 22 54 92 146 
Mitchell et al. (2013) 6 mo 20 EtOHx2 + W Spec 620 (anthrone) AA + amylo. Spec. 515 (GOPOD)    85 120 206 
          20 13 33 
          46 30 76 
Gauthier et al. (2014)1 <6 mo EtOHx3 Enz. Amylo. Spec 515 (GOPOD)    10 17 
P. persica 
Moing et al. (1992) 2 mo EtOHx2 HPLC Amylo. HPLC 11.1 5.69 36.7 95 89 184 
Nii (1997) 37–38 yo EtOH Spec (anthrone)      78 77 155 
Tworkoski et al. (1997) 5–6 yo 200 EtOH HPLC Amylo. Spec.    38–158 33–48 86–191 
          44–77 39–45 83–122 
Escobar-Gutiérrez et al. (1997) 2.5 mo EtOHx2 HPLC Amylo.  39 10 53 215 135 350 
Inglese et al. (2002) 3 yo 150 EtOH Enz. Amylo. Enz.     6–9  
Graham (2002) 2 mo 50 MCW HPLC Amylo. Spec. 450 16 57 52 109 
       11 54 33 88 
       20 15 106 26 132 
Leite et al. (2004) 11 yo S (Oct) 10 EtOHx2 HPLC Amylo. Enz. 27 69 65.5 134 
  S (Feb)      14 35 74 16 90 
Bonhomme et al. (2005) 4 yo 10 EtOHx2 HPLC Amylo. Enz. 15 28 72 12 84 
Gordon et al. (2006) 2 yo HPLC Amylo.      150 
Dichio et al. (2007) >3 yo EtOH Spec. 625 (anthrone) Amylo. Spec. 425 (GOPOD)    120 125 
          100 10 110 
          240 60 300 
Li et al. (2007)2 5 yo 15 000 EtOHx3 HPLC Amylo. Spec. (GOPOD) 51 25 76 
Cheng et al. (2009)2 8 yo 15 000 EtOHx3 HPLC Amylo. Spec. (GOPOD) 11 43 23 65 
Weibel et al. (2008) 4–5 yo  EtOH Spec. (anthrone)        260 
           160 
            190 
P. edulis 
Adams et al. (2013)3 15–25 yo 12 Enz. Amylo. Enz.    10–56 0–185 19–216 
Anderegg and Anderegg (2013) 10–15 yo EtOH + MCW Spec 595 BA + amylo. Spec. 595 5–10   0–30  30–60 
       10–28   8–21  45–95 
Dickman et al. (2015)3 L (2007) 12 Enz. Amylo. Enz. 10 16 
  L (2008)      
  L (2009)      12 16 
Sevanto et al. (2014)3 15–25 yo 12 Enz. Amylo. Enz. 13–27 1–19 27–36 6–31 36–59 
1

Values reported in g m−2.

2

Estimations were made on fresh weight.

3

No fertilizer used.

Several other factors suggest that a comparison of the NSC of standard samples would be worthwhile. First, such a comparison would allow plant ecophysiologists studying the role of NSC and regulation to assess and compare their own results. Second, the composition of NSC can vary widely among species, tissues and seasons (Hoch et al. 2003, Landhäusser and Lieffers 2003, El Zein et al. 2011, Richardson et al. 2013, Dickman et al. 2015), and this diversity further contributes to potential misinterpretation when comparing results from studies that use different methods. Finally, knowledge of the comparability of quantitative estimates of NSC would benefit articles that review NSC among studies to formulate hypotheses about the regulation of plant carbon and growth mechanisms (Körner 2003, Ainsworth and Rogers 2007, McDowell et al. 2008). To our knowledge, no study has addressed the comparability of NSC among different laboratories.

Our primary objective was to assess whether soluble sugar, starch and total NSC concentrations could be compared across laboratories that use NSC estimates to understand plant response to a variety of biotic and abiotic factors. Many of these studies focused on NSC estimates in woody species, so our common samples were from trees. We answered the question of inter-laboratory comparability in NSC quantification by sending sub-samples of five different tissue samples (leaf, root and stem) that we have hypothesized varied widely in NSC, matrix structure and chemistry, to 29 laboratories. The laboratories evaluated the samples using their own ‘in-house’ protocols of NSC extraction and quantification (see Tables S1 and S2 available as Supplementary Data at Tree Physiology Online).

Our second objective was to determine whether estimates from an individual laboratory were consistent across the five standard samples. If a laboratory’s estimates were high, low or similar relative to all laboratories for a given sample, would the same rank apply for the other four standard samples? Consistency among samples would indicate the reliability of comparing relative change within and among laboratories.

The third objective was to determine whether any differences among laboratory estimates were related to the methods of extraction and/or quantification of soluble sugars and starch, and if variability among laboratories differed by sample. Because our first objective was the primary purpose for the study, our ability to test the third objective suffered by having to group extraction and quantification methods into broad categories. This grouping and our sample of laboratories precluded testing factors that may be important sources of variability because of lack of replication. These factors include the number, temperature and duration of extractions and the gelatinization of starch. We partially addressed this issue by investigating the effect of different extraction methods on sugar estimates in a single laboratory using a common quantification method.

Materials and methods

Non-structural carbohydrate analyses of standard samples in different laboratories

We selected five samples for our standards: leaves (EGL), roots (EGR) and stem (EGS) of E. globulus, Pinus edulis needles (PEN) and Prunus persica leaves (PPL). We selected these samples because a priori knowledge suggested that they differed in the concentration of soluble sugars and starch, and had very different structural or chemical matrices that would challenge NSC extraction. Each substrate was homogenized, irradiated at 27.8 kGy for microbiological control to meet international quarantine requirements and then homogenized. Method S1 available as Supplementary Data at Tree Physiology Online describes the collection and handling of samples used.

Sub-samples of the same five dried and ground samples were sent to 29 laboratories around the world (Austria, Australia, Canada, Chile, Estonia, France, Germany, Japan, Israel, The Netherlands, Spain, Switzerland and the USA), where each laboratory used their own protocol to analyze the samples in triplicate (see Method S2, Tables S1 and S2 available as Supplementary Data at Tree Physiology Online). One laboratory (Q) only provided sugar estimates, and two other laboratories (L1, L2; Z1, Z2) provided sugar estimates from two different methods. The number of estimates for starch was 28, the number of estimates for total soluble sugars was 31, and the independent estimates for sugars from laboratories L and Z were combined with the single starch estimate from those laboratories to yield 30 NSC estimates. One laboratory had samples analyzed twice by two separate investigators, which we counted as six laboratory replicates, but only one starch, soluble sugar and NSC mean estimate. Table 1 summarizes the procedures used in this study to measure soluble sugars and starch in plant tissues and Tables S1 and S2 available as Supplementary Data at Tree Physiology Online provide more detailed methods. All data were reported as mg g−1 of dry mass.

Different methods for soluble sugar extraction within a single laboratory

We selected four methods of soluble sugar extraction: 80% ethanol (80% EtOH), 70% methanol (70% MeOH), methanol : chloroform : water (MCW) at 80 °C (MCW 80) and MCW at ambient laboratory temperature (MCW amb). Individual soluble sugars (glucose, fructose, sucrose) were extracted from 20 mg of dried plant tissue for each of the five samples for each of the four methods. Alcohol methods (EtOH) were derived from Gomez et al. (2002), and ternary solvent methods (MCW) from Dickson and Larson (1975). All four methods were conducted within the same laboratory (see Method S3 available as Supplementary Data at Tree Physiology Online).

Other methods

We also performed an analysis of the effect of microwaving duration to halt enzymatic activity (see Method S4 available as Supplementary Data at Tree Physiology Online), and the effect of particle size (see Method S5 available as Supplementary Data at Tree Physiology Online) in single laboratories.

Statistical analyses

For Objective 1, we used a general linear mixed model analysis to determine differences in estimates among laboratories with laboratory and sample types as fixed effects and the extraction and quantification categories (below) as random effects. For Objective 2, we used Spearman rank correlations for laboratory ranks among all sample pairs to evaluate the consistency of laboratory estimates for samples with different chemical constituents. Correlations were estimated for total soluble sugars, starch and total NSC.

For Objective 3, we used a different general linear mixed model analysis, with extraction and quantification groups and sample as fixed effects, and laboratory as a random effect. We could not perform one overall test with laboratories and methods, because methods were confounded with laboratory. We grouped methods according to the type of solvent used for the extraction methods (EtOH, EtOH + W, MCW, W for the soluble sugars; and Acid, AA + amylo., Amylo. for starch) and by the type of quantitative assay for the quantification methods (HPLC, Enz., Spec. 490, Spec. 620 and Spec. 510). High-performance anion exchange chromatography-pulsed amperometric detection (HPAEC-PAD) and proton nuclear magnetic resonance (1H-NMR) were grouped with HPLC. Both sugar and starch concentrations were log-normally distributed and all components were transformed for analysis. Least squares means (LSM) were back-transformed to original units after estimation of the model parameters. Other differences in laboratory protocols (differences among the number, temperature and duration of extractions or methods used for the gelatinization of starch) were not considered as factors within the method because of the lack of replication. General linear mixed model analyses were done using SAS PROC GLIMMIX (SAS Institute 2012). The proportion of the variance explained by the method categories compared with sample and laboratory was evaluated using the method of computing R2 for generalized linear mixed models described in Nakagawa and Schielzeth (2013). We assessed how differences among method categories compared with differences among samples and laboratories by comparing the R2 for models with only the method category as a fixed factor with (i) R2 for models with only sample category as a fixed factor and (ii) with the R2 for the full model with sample and method as fixed factors and laboratory as a random factor. R2 measures were computed using the ‘R’ statistical package version 3.1.2 (R Development Core Team 2014) and the MuMIn library.

We examined the differences between soluble sugar extraction methods on total NSC in the same laboratory with an analysis of variance for each sample type (α = 0.05). For all tests and all experiments, we set α at 0.05. Participants were assured of anonymity in the experiment, and the results were coded by letters.

Results

Objective 1: estimates for soluble sugars, starch and total NSC for the same samples varied substantially among laboratories

Estimates for individual sugars, total soluble sugars, starch and total NSC differed among laboratories (P < 0.001, Figure 1), with a large range for all components. For example, in E. globulus leaves (EGL), laboratory estimates ranged from 23 to 116 mg g−1 (coefficient of variation (CV) 35%) for total soluble sugars, 6–533 mg g−1 (CV 102%) for starch and 53–649 mg g−1 (CV 69%) for total NSC (Figure 1). Laboratory estimates for Prunus leaves (PPL, average CV = 87% for sugars, starch and total NSC) were more variable than those for other samples (average CV = 54–69% for all NSC components). Starch estimates were more variable among laboratories (CV 87–120%) than were soluble sugars and total NSC (CV 24–71% for sugars and 44–71% for total NSC, Figure 1). For all samples and NSC components, 10–57% of the laboratories were within the 95% confidence intervals estimated for the means. Laboratories were most consistent for starch estimated for the Eucalyptus leaf, stem and root samples (EGL, EGS, EGR, 16 of 28 laboratories were within the 95% confidence intervals), and least consistent for sugar estimates for Eucalyptus leaves (4 of 30 laboratories) and total NSC estimated for Pinus leaves (8 of 30 laboratories) and Prunus leaves (3 of 30 laboratories). The subset of the laboratories that identified sucrose and glucose + fructose (n = 20) were relatively consistent, having an average of 51% or 10 of 20 laboratory estimates within the 95% confidence intervals (range = 7–14 laboratories, Figure 1a). The interaction between laboratory and sample type was highly significant for sugars, starch and total NSC (P < 0.001), indicating that differences among laboratories differed with sample type.

Figure 1.

Laboratory estimates of (a) sucrose, glucose + fructose, total soluble sugar, and (b) starch and NSC for five samples: EGL, PEN, PPL, EGR and EGS, with means (text and solid line), range, coefficient of variation (CV) and 95% confidence interval (dashed lines). Estimates are ranked by sugar extraction category: W, water; EtOH + W, ethanol water mixture; MCW, methanol : chloroform : water; EtOH, ethanol. Estimates differed substantially among laboratories and within method categories.

Figure 1.

Laboratory estimates of (a) sucrose, glucose + fructose, total soluble sugar, and (b) starch and NSC for five samples: EGL, PEN, PPL, EGR and EGS, with means (text and solid line), range, coefficient of variation (CV) and 95% confidence interval (dashed lines). Estimates are ranked by sugar extraction category: W, water; EtOH + W, ethanol water mixture; MCW, methanol : chloroform : water; EtOH, ethanol. Estimates differed substantially among laboratories and within method categories.

The range of estimates varied substantially with method and sample types (Figure 1, Figure S1 available as Supplementary Data at Tree Physiology Online). For example, NSC in the PPL sample showed high variability among laboratories (Figure 1, Figure S1a available as Supplementary Data at Tree Physiology Online), and estimates for soluble sugars varied largely within each method of extraction and quantification, except for the water extraction (W) (see Figure S1a available as Supplementary Data at Tree Physiology Online). In comparison, NSC in the EGS sample had the lowest variability among laboratories (Figure 1b) and estimates varied less within each method (see Figure S1b available as Supplementary Data at Tree Physiology Online).

Objective 2: laboratories had similar rankings for all five common samples

Laboratory rankings were consistent for most sample pairs (Table 3; Figure 2), with higher rank correlations for starch (0.41–0.91, mean = 0.71) and total NSC (0.45–0.84, mean = 0.60) than for soluble sugars (0.11–0.83, mean = 0.44). This consistency shows that laboratories with estimates below, above or near the mean for one sample tend to have a similar ranking for that carbohydrate relative to other laboratories for other samples.

Table 3.

The Spearman rank correlation indicates correlations for laboratories between sample pairs of 0.11–0.83 (mean = 0.44) for soluble sugars, 0.41–0.91 (mean = 0.71) for starch and 0.45–0.84 (mean = 0.60) for total NSC. These results suggest starch has the most consistency among laboratory ranks for the different samples.

*P < 0.05,

**P < 0.01.

 EGL EGR EGS PEN PPL 
Soluble sugars 
 EGL      
 EGR 0.33     
 EGS 0.11 0.73**    
 PEN 0.29 0.52** 0.41*   
 PPL 0.83** 0.39* 0.37* 0.41*  
Starch 
 EGL      
 EGR 0.69**     
 EGS 0.59** 0.87**    
 PEN 0.47* 0.83** 0.91**   
 PPL 0.41* 0.68** 0.84** 0.81**  
Total NSC 
 EGL      
 EGR 0.59**     
 EGS 0.49** 0.69**    
 PEN 0.45* 0.84** 0.64**   
 PPL 0.49** 0.54** 0.55** 0.72**  
 EGL EGR EGS PEN PPL 
Soluble sugars 
 EGL      
 EGR 0.33     
 EGS 0.11 0.73**    
 PEN 0.29 0.52** 0.41*   
 PPL 0.83** 0.39* 0.37* 0.41*  
Starch 
 EGL      
 EGR 0.69**     
 EGS 0.59** 0.87**    
 PEN 0.47* 0.83** 0.91**   
 PPL 0.41* 0.68** 0.84** 0.81**  
Total NSC 
 EGL      
 EGR 0.59**     
 EGS 0.49** 0.69**    
 PEN 0.45* 0.84** 0.64**   
 PPL 0.49** 0.54** 0.55** 0.72**  
Figure 2.

Correlations of laboratory ranks among all sample pairs that show the worst and best correlations for soluble sugars, starch and total NSC. Plots show that laboratory rankings can be consistent for the different samples. Spearman rank correlations for all sample pairs are given in Table 3. Solid lines are the 1 : 1 line.

Figure 2.

Correlations of laboratory ranks among all sample pairs that show the worst and best correlations for soluble sugars, starch and total NSC. Plots show that laboratory rankings can be consistent for the different samples. Spearman rank correlations for all sample pairs are given in Table 3. Solid lines are the 1 : 1 line.

Objective 3: extraction and quantification methods affect NSC estimates, but the effect is lower than variability among laboratories

We investigated whether the methods used to extract or quantify NSC could explain the variability in NSC results among laboratories (Table 4; Figure 3). When analyses were pooled across laboratories and samples, NSC estimates did not differ by sugar or starch extraction or quantification methods (Table 4, P = 0.07–0.84, Figure 3c, e, g, and i: LSM). Across laboratories and samples, starch estimates were lower for ethanol + water sugar extraction than for the other three sugar extraction categories (Figure 3b: LSM, P < 0.05), but did not differ by starch extraction or quantification categories (Figure 3d and h: LSM). Across laboratories and samples, sugar estimates did not vary by extraction method category (Figure 3a: LSM), but did by sugar quantification method category (Figure 3f: LSM, P < 0.05), with the Spec. 620 colorimetric method producing higher estimates than the HPLC, enzymatic or Spec. 490 method. A principal component analysis showed that within a method, the estimates for soluble sugars were more variable than were estimates for starch (see Figures S2 and S3 available as Supplementary Data at Tree Physiology Online).

Table 4.

The general linear mixed model analysis with laboratory as a random factor showed some differences for extraction and quantification methods for sugars and starch concentrations and interactions between extraction and quantification methods and sample for sugars, starch and total NSC. The interactions suggest that a method performs differently for different samples. df, degree of freedom; Num., numerator; Den., denominator.

 Soluble sugars (SS)
 
Starch
 
Total NSC
 
Num. df Den. df F P-value Num. df Den. df F P-value Num. df Den. df F P-value 
Sample 426 63.4 <0.0001 387 152 <0.0001 386 122 <0.0001 
SS extraction 28 2.1 0.123 25.01 9.2 0.0003 25.01 2.6 0.074 
SS quantification 27.95 5.6 0.004 – – – – 25.01 25.0 0.443 
Starch extraction – – – – 26.01 3.1 0.064 26.02 0.12 0.837 
Starch quantification – – – – 24 1.3 0.306 24.01 1.9 0.141 
Sample × SS extraction 12 426 11.6 <0.0001 12 387 5.1 <0.0001 12 386 11.7 <0.0001 
Sample × SS quantification 12 426 7.54 <0.0001 – – – – 12 386 386 <0.0001 
Sample × Starch extraction – – – – 391 4.7 <0.0001 390 3.5 0.0007 
Sample × Starch quantification – – – – 16 383 15.0 <0.0001 16 382 10.7 <0.0001 
 Soluble sugars (SS)
 
Starch
 
Total NSC
 
Num. df Den. df F P-value Num. df Den. df F P-value Num. df Den. df F P-value 
Sample 426 63.4 <0.0001 387 152 <0.0001 386 122 <0.0001 
SS extraction 28 2.1 0.123 25.01 9.2 0.0003 25.01 2.6 0.074 
SS quantification 27.95 5.6 0.004 – – – – 25.01 25.0 0.443 
Starch extraction – – – – 26.01 3.1 0.064 26.02 0.12 0.837 
Starch quantification – – – – 24 1.3 0.306 24.01 1.9 0.141 
Sample × SS extraction 12 426 11.6 <0.0001 12 387 5.1 <0.0001 12 386 11.7 <0.0001 
Sample × SS quantification 12 426 7.54 <0.0001 – – – – 12 386 386 <0.0001 
Sample × Starch extraction – – – – 391 4.7 <0.0001 390 3.5 0.0007 
Sample × Starch quantification – – – – 16 383 15.0 <0.0001 16 382 10.7 <0.0001 
Figure 3.

Differences in LSM for all samples and for individual samples (EGL, PEN, PPL, EGR, EGS) for the extraction and quantification methods for soluble sugars, starch and total NSC show that method category generally had little effect on NSC difference, perhaps because of high within-method variance. Error bars are standard errors for the LSM. Total soluble sugars results are grouped by sugar extraction (a) and quantification (f) method. Starch results are grouped by sugar (b) and starch (d) extraction method and starch quantification method (h). Total NSC results are grouped by sugar (c) and starch (e) extraction methods, and for sugar (g) and starch (i) quantification methods. Significant differences (*) among methods within each tissue were assessed with Tukey–Kramer test (α = 0.05).

Figure 3.

Differences in LSM for all samples and for individual samples (EGL, PEN, PPL, EGR, EGS) for the extraction and quantification methods for soluble sugars, starch and total NSC show that method category generally had little effect on NSC difference, perhaps because of high within-method variance. Error bars are standard errors for the LSM. Total soluble sugars results are grouped by sugar extraction (a) and quantification (f) method. Starch results are grouped by sugar (b) and starch (d) extraction method and starch quantification method (h). Total NSC results are grouped by sugar (c) and starch (e) extraction methods, and for sugar (g) and starch (i) quantification methods. Significant differences (*) among methods within each tissue were assessed with Tukey–Kramer test (α = 0.05).

An analysis of R2 for model components showed that the differences in method category in our analysis accounted only for a small portion of differences in NSC among laboratories. R2 for total soluble sugars with sugar extraction method category was 0.05 and 0.12 for sugar detection method category, compared with 0.30 for sample and 0.66–0.69 for the full model. R2 for starch with starch extraction method category was 0.10 and 0.11 for starch detection method category, compared with 0.23 for sample and 0.88 or 0.92 for the full model; sugar extraction method category had an R2 of 0.33. R2 for total NSC with sugar extraction method category was 0.09, 0.04 for sugar detection method category, 0.01 for starch extraction method category and 0.09 for starch detection category compared with 0.37 for sample and 0.79–0.84 for the full model. Additionally, differences between the highest and lowest LSM for the overall effect of methods categories were small compared with the differences among laboratories (compare Figure 3 with Figure 1).

Objective 3: method effects differ by sample

Sample and method had significant interactions (Table 4, P < 0.0001), with the foliar samples (EGL, PEN and PPL) showing more variation among method categories than the wood samples (EGR, EGS). For example, the sugar extractions with water (W and EtOH + W) yielded lower soluble sugar and total NSC estimates for the foliar samples (EGL, PEN and PPL), while having less effect on woody samples (EGR and EGS, Figure 3a and c). Starch concentration differences among extraction and quantification methods in woody samples were similar to those for foliar samples (Figure 3b, d and h). Colorimetric quantification (Spec. 490 and Spec. 620) of starch and soluble sugars almost always produced higher estimates for soluble sugars, starch and total NSC than did the HPLC or enzymatic methods (Figure 3f–i).

Objective 3: single laboratory tests of soluble sugar extraction methods, microwaving and particle size

Soluble sugar extraction methods influenced sugar estimates when samples were quantified in the same laboratory using the same method. Estimates of total soluble sugars were affected by extraction methods for all samples (P < 0.05) except EGL (P > 0.10). Differences among sugar extraction methods tested in the same laboratory (Figure 4) were relatively minor compared with differences among laboratories (Figure 1a), with the largest differences occurring for the MCW extractions at different temperatures (Figure 4).

Figure 4.

Means and standard errors for soluble sugars by extraction method for samples processed in one laboratory and using the same quantification method. Results show that extraction method can affect estimates especially for PEN and PPL samples. In all samples MCW-based methods produced consistently lower estimates than alcohol-based methods. Different letters indicate significant difference at α = 0.05 according to F-protected LSD test.

Figure 4.

Means and standard errors for soluble sugars by extraction method for samples processed in one laboratory and using the same quantification method. Results show that extraction method can affect estimates especially for PEN and PPL samples. In all samples MCW-based methods produced consistently lower estimates than alcohol-based methods. Different letters indicate significant difference at α = 0.05 according to F-protected LSD test.

Microwaving small samples (<5 g) of P. edulis at 800 W required 180 s to deactivate enzymes. No microwaving or 90 s of microwaving was not effective at halting the conversion of sucrose and starch to glucose + fructose. At 300 s, starch and NSC increased, suggesting conversion of non-NSC compounds to NSC (see Method S4 available as Supplementary Data at Tree Physiology Online, Figure 5). Grinding Pinus banksiana tissues to a smaller particle size (<105 μm) yielded higher starch and total NSC estimates for root tissues (but not needles or stem) compared with extractions of larger particle size (<400 μm, Method S5, Figure S4 available as Supplementary Data at Tree Physiology Online).

Figure 5.

Effect of microwaving samples <5 g at 800 W on amount of glucose + fructose (Gluc + Fruc), sucrose (Suc), starch and total NSC for foliar (a) and twig (b) samples of P. edulis. See Method S4 available as Supplementary Data at Tree Physiology Online for details on the method. At 0 and 90 s microwaving time, sucrose hydrolyzing and starch debranching enzymes are still active, leading to lower sucrose levels, higher glucose + fructose levels and higher starch levels because debranching enzymes make starch more accessible to the enzymatic assay. At 180 s and above, enzymes are deactivated, yielding consistent sucrose and glucose + fructose. At 300 s, starch starts to gelatinize, again making it more accessible to the assay. Orthogonal contrasts for trend with microwaving time: glucose + fructose, quadratic for leaf and twig, P < 0.05; sucrose, linear for leaf and twig, P < 0.001; starch, quadratic for leaf and twig, P < 0.01; total NSC, quadratic for leaf and twig, P < 0.01.

Figure 5.

Effect of microwaving samples <5 g at 800 W on amount of glucose + fructose (Gluc + Fruc), sucrose (Suc), starch and total NSC for foliar (a) and twig (b) samples of P. edulis. See Method S4 available as Supplementary Data at Tree Physiology Online for details on the method. At 0 and 90 s microwaving time, sucrose hydrolyzing and starch debranching enzymes are still active, leading to lower sucrose levels, higher glucose + fructose levels and higher starch levels because debranching enzymes make starch more accessible to the enzymatic assay. At 180 s and above, enzymes are deactivated, yielding consistent sucrose and glucose + fructose. At 300 s, starch starts to gelatinize, again making it more accessible to the assay. Orthogonal contrasts for trend with microwaving time: glucose + fructose, quadratic for leaf and twig, P < 0.05; sucrose, linear for leaf and twig, P < 0.001; starch, quadratic for leaf and twig, P < 0.01; total NSC, quadratic for leaf and twig, P < 0.01.

Discussion

Absolute estimates of NSC are not comparable among laboratories (Objective 1)

Results demonstrate that estimates of soluble sugar, starch and total NSC provided by different laboratories in this study cannot be compared, even if they are obtained with the same general methods. Laboratories differed substantially in estimates for sugars, starch and total NSC, and the variability across laboratories and even within a method category was unexpectedly large. Therefore, comparing values for any NSC component across studies in the literature (e.g., Ainsworth et al. 2002, Morgan et al. 2003, Wittig et al. 2009) should not be done, either for individual studies or for meta-analyses, unless the study accounts for laboratory effects.

Relative differences within a single laboratory can be consistent and meaningful (Objective 2)

The Spearman rank correlation analysis of sample pairs showed that laboratory ranks were fairly consistent among the five samples for starch, but less so for soluble sugars and total NSC. These results suggest that relative differences among treatments and species within a laboratory can be meaningful. While we did not explicitly test how laboratories would perform using the same substrate with two different NSC concentrations, preserving laboratory rank across such a diverse sample cohort was a significant finding in this experiment. Therefore, an assessment of relative responses of different treatments to a control may be robust, especially for starch, and meaningful within and between studies.

Method differences explained only some of the variability among laboratories, but meeting Objective 1 compromised our ability to identify these differences (Objective 3)

Differences among methods, as captured by our extraction and quantification group approaches, were generally small relative to the differences among laboratories. However, fulfilling our primary objective (to identify whether NSC estimates could be compared among laboratories) compromised the ability to identify differences between methods. We can interpret these results to mean that: (i) real differences among methods would exist, and variation among laboratories would be minimized if the laboratories using the same method followed the same protocols exactly for extraction and quantification; or (ii) NSC quantification is such a highly variable and sensitive procedure that even minor differences among laboratories’ procedures not captured in an explicit protocol would cause variation among laboratories using the same method. We suspect that both explanations play a role in the low ability of ‘methods’ to explain laboratory differences.

Variation in protocols within a method category may have contributed to the lack of significant differences among methods. For example, the number, temperature and duration of extractions and the method of starch gelatinization (Table 2, Tables S1 and S2 available as Supplementary Data at Tree Physiology Online) are known to affect soluble sugar and starch estimates (Yemm and Willis 1954, MacRae et al. 1974, Rose et al. 1991, Johansen et al. 1996, Shi et al. 2002, Gomez et al. 2003, Kim et al. 2003). We were surprised at the variability among laboratories in these factors, and even laboratories using the same ‘method’ differed in these important factors. Variability of method application within a method category yielded little or no replication for these factors, and limited the evaluation to broad method categories. As an example of how these factors might contribute to differences among laboratories, yet not appear in our methods analysis, we found that higher temperature increased sugar concentration for MCW extracts in two of the four samples (Figure 4).

The lack of differences among soluble sugar extraction method categories (P = 0.12, Table 4), coupled with the small differences between different methods within a single laboratory (Figure 4), suggests that variation in the application of extraction methods across laboratories was larger than the effect of the extraction solvent. However, despite laboratory differences in protocol, we could still detect an effect of soluble sugar quantification methods on sugar estimates (Figure 3, P = 0.004). These differences may result from the fact that different methods quantify different sugars. This result suggests that systematic differences in quantification, especially between colorimetric and HPLC-based methods, might be interpreted and possibly corrected.

We also did not assess the effect of other factors such as air temperature, level of expertise of the person conducting the analyses or quality of the laboratory equipment. Such factors might contribute to the variability among laboratories, even for those using the same general method, but they have not been assessed.

Method effects differ by sample (Objective 3)

Non-structural carbohydrate components exist within a complex and varied chemical matrix and need to be extracted from this matrix for analysis. Procedures to extract NSC from the matrix can free the target compound, but also convert other compounds into the target. Maximizing the extraction while minimizing the conversion is the goal of procedures, but may not always occur (Hansen and Møller 1975, Thompson and Ellison 2005, Santiago da Silva et al. 2012, Huang and Fu 2013). In our study, soluble sugar estimates for Eucalyptus and Prunus leaves differ with the sugar quantification method (colorimetric methods generate higher estimates than do HPLC or enzymatic methods, Figure 3; see Note S1 available as Supplementary Data at Tree Physiology Online). Clearing interfering compounds from the solvent might minimize these effects (Thompson and Ellison 2005), as would avoiding acid use during sugar extraction (Chow and Landhäusser 2004). The significant interactions between sample type and methods also suggest that different extraction and quantification protocols will give different results for NSC in samples with different matrices.

How can we make quantitative, comparable estimates of the true value of NSC components?

Determination of the role and regulation of NSC is governed by what we can measure (Dietze et al. 2014). Our study demonstrates that laboratories and methods produce widely different and non-comparable estimates and progress in plant science will be limited until this problem is resolved, although relative differences in NSC have been and will continue to be important for many questions. Being able to compare between and within studies and knowing the true value are essential for a mechanistic understanding of NSC pools and fluxes (Ryan 2011), especially for questions about the role of NSC in ecosystem productivity, stress responses and plant adaptations. Relative differences within and across studies are valuable for testing many hypotheses, and this study shows that these have values, particularly for starch.

Comparability might be solved using two approaches: either adopt a standard method and report values for certified reference material (CRM) or embrace a central laboratory for all processing. A standard method would require a detailed and easily applied protocol, from sample collection to quantification, so that any laboratory can reproduce values for the CRM. Another solution to the comparability problem would be to establish and adopt a central laboratory for all NSC analyses, similar to the calibration laboratories of the Global Atmosphere Watch program (http://www.wmo.int/pages/prog/arep/gaw/qassurance.html) or the US National Atmospheric Deposition Program (http://nadp.sws.uiuc.edu). A central laboratory could use different methods for samples of different characteristics and still maintain comparability among samples. Both approaches can be criticized for the lack of flexibility and freedom they impose on the scientific community, and raise the practical issue of what to do with the existing costly analytical equipment. Adopting a standard method for NSC determination in plants would likely be more practical than establishing a central facility, but would impose an investment for laboratories to comply with the selected standard. Adoption of either approach would depend on the cooperation of the science community.

Our results provide some insights into which methods might give the most homogenous results (i.e., those less affected by random error). High performance liquid chromatography was the quantification method with the least variable results, whereas colorimetric assays exhibited more variability (Figure 1, Figure S1 available as Supplementary Data at Tree Physiology Online). High performance liquid chromatography methods (including HPAEC-PAD and 1H-NMR) are increasingly chosen by laboratories because of (i) their high resolution, even with a small amount of sample and (ii) reproducibility due to a close control of parameters affecting the efficiency of separation and quantification (Giannoccaro et al. 2008, Raessler et al. 2010). However, the HPLC process is time-consuming, laborious and expensive—especially for carbon balance studies where only the total amount of glucose equivalents may be of interest. In addition, HPLC still relies on sugar and starch extractions that vary substantially with solvent and other method details.

Colorimetric methods are less expensive than other techniques, rapid and can detect all types of sugars, and therefore are still widely used; nevertheless, they have major drawbacks, including: (i) the necessity to prepare a calibration curve using a series of standards because different carbohydrates give different absorbance responses (see Dubois et al. 1956, Hall 2013); (ii) the use of toxic and dangerous chemicals; and (iii) possible interference of metabolites with the concentrated sulfuric acid (Ashwell 1957).

The enzymatic method also produced relatively consistent results and allowed for the measurement of individual sugars. This method requires expertise for timing of enzyme additions, checking for cross contamination (converting non-targeted oligosaccharides) and maintenance of a precise pH for NADPH. In this study, three laboratories using the enzymatic method reported negative results for sucrose (Figure 1; Table S1 available as Supplementary Data at Tree Physiology Online). Negative results are not normally reported and are usually assumed to be zero, but indicate that something went wrong in the assay. This might be caused by inappropriate extraction (hydrolyzing sucrose into glucose and fructose) or too low pH (leading to NADPH degradation following the addition of invertase, the enzyme enabling the quantification of sucrose). To solve these issues, cross-validation with HPLC or NMR should be performed each time a new sample type is analyzed.

Best practice in other plant chemical analyses generally use CRMs to ensure comparability of results (e.g., Quevauviller et al. 1994, Clement et al. 1996, Saunders et al. 2004). Unfortunately, CRM for carbohydrates do not currently exist. Many laboratories use pure sugar and/or starch standards (n = 15 in our study) to define recovery of known concentrations of specific sugars. However, these standards do not account for the effect of plant matrix which may generate incomplete carbohydrate extraction or yield compounds that interfere with quantification (Emons et al. 2004). A CRM is accompanied by a certificate, which specifies property values of the material: before the certificate is delivered, a procedure establishes material traceability to an accurate realization of the unit, and for which each certified value is accompanied by an uncertainty at a stated level of confidence (Emons et al. 2004). Certified reference materials are a key element of analytical data quality assurance and are used for four main purposes: (i) instrument calibration; (ii) method validation, in particular for assessment of the reliability of a method; (iii) ensuring the traceability of measurement results; and (iv) statistical quality control (Emons et al. 2004). Certified reference material for NSC will likely require several samples with different matrices, sugar and starch concentrations. Integration of CRMs into NSC analysis should be standard practice to improve comparability among laboratories.

In addition to the difficulty of quantitatively assessing soluble sugars and starch, studies assessing NSC may miss important components that could represent a substantial fraction of NSC. Most studies assessing NSC have focused on analyzing the three ‘major’ sugars (sucrose, glucose, fructose) and starch, and assume that this pool represents the NSC available to the plant—a reasonable assumption for most trees (Hoch et al. 2003, Hoch and Körner 2005). A few studies suggest that we should sometimes look deeper. For example, sorbitol is found in high concentrations in PPL (Zhang et al. 2013) and quercitol in droughted Eucalyptus astringens leaves (Arndt et al. 2008), and raffinose concentration was greater than that of starch in birch buds (Ruuhola et al. 2011).

Conclusions and recommendations for the future

We conclude that absolute values of NSC, total soluble sugars, starch and individual sugars cannot be directly compared among laboratories, even among laboratories that use a method in the same method category. Differences relative to a control may have value with a single laboratory and for comparisons among laboratories for starch, but less so for total NSC and for soluble sugars. Differences in absolute values among laboratories were poorly related to our broad method categories, but many factors that may contribute to different estimates could not be assessed in our analyses.

Our study shows that developing methods to produce reliable, absolute and comparable estimates of NSC and its components in plant tissue will be a serious challenge because of high variability in methods currently in use, lack of absolute standards and little information about the causes of the high variation in estimates among laboratories. Our team discussed the benefits and pitfalls of proposing a standard method for sample collection, storage, processing, extraction and quantification as a first step towards achieving comparability among laboratory estimates. Team members mostly supported the publication of a standard method (although there was less agreement about the particular method), but there were also strong arguments against such an approach. The small differences among method categories and the high variability of laboratory processes within the method categories in this study suggest that adopting a standard method would have a higher likelihood of producing comparable estimates across studies. A standard method would at least insure that differences among studies are not because of methodological differences. However, neither this study nor any other of which we are aware has identified a ‘best’ method. Arguments against proposing a standard method are (i) that we do not have the data to support selecting any particular method, (ii) laboratories that change methods will lose a connection to past studies, (iii) laboratories that do not adopt the proposed standard method risk having difficulty in publishing their results, and (iv) there was disagreement over what the proposed method should be—with the largest disagreements over the sample size (50 mg samples processed in ∼10 ml vials versus 10 mg samples processed in standard 96-well plates) and sample storage prior to processing (to freeze or not).

Recognizing the different viewpoints of our team members, to help the research community move towards NSC analysis that is comparable both among and within laboratories, we propose:

  • A reference method for sample collection and storage, sample processing, sugar extraction, starch extraction and quantification. We use the term ‘reference method’ to identify the method as one that can indicate comparability among laboratory estimates, as distinct from a ‘standard method’ that might imply a ‘best’, fully vetted method. Our data showed that water extractions gave the least variability among laboratories for soluble sugar extraction (see Figure S2A available as Supplementary Data at Tree Physiology Online), and that the α-amylase + amyloglucosidase extractions gave the least variability for starch (see Figure S3A available as Supplementary Data at Tree Physiology Online). Although water is the optimal extraction solvent for low molecular weight sugars and exhibited the least variability, it can also dissolve interfering hydrophilic polysaccharides and proteins. Extraction in aqueous alcohol can minimize this problem, and provide a high recovery of low molecular weight sugars. Standardization of alcohol strength and the number, temperature and duration of extractions is important to minimize variability in the results (see Figure S2a available as Supplementary Data at Tree Physiology Online). Using these results, the discussion about methods in Note S2 available as Supplementary Data at Tree Physiology Online, and the results for microwave duration and intensity (Figure 5) and particle size (see Figure S4 available as Supplementary Data at Tree Physiology Online), we recommend the method detailed in Figure 6 be adopted as a reference method. HPLC and variants showed the least variability among quantification methods because of its precision, but perhaps also because HPLC procedures incorporate filtration to remove interfering compounds. However, the reference method does not include a filtration or quantification step. We ended the reference method with extraction, because our study does not provide the data to support a recommendation for the adoption of the expensive HPLC quantification and filtration steps.

  • That laboratories adopt the reference method for sample collection and storage, sample processing, sugar and starch extraction and filtration, or laboratories retain their current methods but analyze a portion of a study’s samples with the reference method for sample collection and storage, sample processing, sugar and starch extraction and filtration. Samples selected for analysis with the reference method should span the range of NSC values identified using the laboratory’s current methods and results should be reported in publications. Laboratories retaining methods different from the reference method should provide a rationale for their use and a full description of the method. Following either of these recommendations would aid both in-house procedures and comparability among studies.

  • Researchers should implement standard procedures of internal quality control and include a detailed description of this procedure to the method. Analytical results should evaluate and present ‘measurement uncertainty’, given by the sample replicates, starch and sugar standards and NSC values for the peach leaf standard (SRM 1547). While SRM 1547 does not have certified estimates for NSC and its components, it is a widely available and standardized sample.

  • Certified reference materials and laboratory inter-calibration should be developed and applied in all NSC analyses. The development of an appropriate range of CRMs will require coordination within the research community to ensure that the CRMs represent the range of tissues and matrices of interest. Once CRMs have been developed, an indication of quality control should be published with all NSC results, to aid in more effective among-laboratory comparisons.

  • The research community, including ecologists and biochemists, should work to develop a small set of standard methods that are appropriate for particular samples and questions and test the reference method.

Figure 6.

Instructions for sample collection, handling, preparation and sugar and starch extraction for reference method.

Figure 6.

Instructions for sample collection, handling, preparation and sugar and starch extraction for reference method.

The problem we have highlighted here, that NSC estimates are not comparable among different laboratories, will likely limit understanding of plant response to environmental stress. While our study focused on NSC determination in woody vegetation, a similar range of methods is used in non-woody species (e.g., Campo et al. 2013, Jaikumar et al. 2014, Kagan et al. 2014, King et al. 2014), and our results are likely to be relevant to the broader plant science community. A more unified approach to NSC analysis and standardization of methods will contribute to better understanding of plant responses to environment and management.

Supplementary data

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

Conflict of interest

None declared.

Funding

M.G.R. was funded by McMaster fellowship (1158.C). S.P. was funded by Juan de la Cierva contract (MCI project) and project ARBALMONT/786-2012 (OPAN, MAAMA, Spain). F.P. was funded by Fondecyt 11121175. Ü.N. and M.T. were funded by the Estonian Ministry of Education and Science, grant IUT-8-3. N.G.M. and L.T.D. were funded by DOE-BER. H.D.A. was funded by LANL-LDRD. J.M.-V. was funded by the Spanish Government (CGL 2010-16376). S.H. was funded by the Montana Institute on Ecosystems’ Graduate Enhancement Award from NSF EPSCoR Track-1 NSF-IIA-1443108. Valuable comments from Dr Mauricio Mencuccini (University of Edinburgh), Dan Binkley (Colorado State University) and two anonymous reviewers were also greatly appreciated.

Acknowledgments

A.G.Q. is grateful to members of the participating laboratories for their work and co-operation. We would like to thank Marc Vandame, Nicole Sonderegger, Laurianne Rouan, Mathieu Moreau, Alieta Eyles, Elena Lahoz, Leo Goudzwaard, Arjen van Peppel, Efrat Neuhaus, Julie Lavergne, Catherine Deborde, Britta Jahn-Humphrey, Dugald Close and Widad Al-Shawi for their help and contribution with the laboratory analyses.

References

Adams
HD
Germino
MJ
Breshears
DD
Barron-Gafford
GA
Guardiola-Claramonte
M
Zou
CB
Huxman
TE
(
2013
)
Nonstructural leaf carbohydrate dynamics of Pinus edulis during drought-induced tree mortality reveal role for carbon metabolism in mortality mechanism
.
New Phytol
 
197
:
1142
1151
.
Ainsworth
EA
Rogers
A
(
2007
)
The response of photosynthesis and stomatal conductance to rising [CO2]: mechanisms and environmental interactions
.
Plant Cell Environ
 
30
:
258
270
.
Ainsworth
EA
Davey
PA
Bernacchi
CJ
et al
(
2002
)
A meta-analysis of elevated [CO2] effects on soybean (Glycine max) physiology, growth and yield
.
Glob Change Biol
 
8
:
695
709
.
Anderegg
WRL
Anderegg
LDL
(
2013
)
Hydraulic and carbohydrate changes in experimental drought-induced mortality of saplings in two conifer species
.
Tree Physiol
 
33
:
252
260
.
Arndt
SK
Livesley
SJ
Merchant
A
Bleby
TM
Grierson
PF
(
2008
)
Quercitol and osmotic adaptation of field-grown Eucalyptus under seasonal drought stress
.
Plant Cell Environ
 
31
:
915
924
.
Ashwell
G
(
1957
)
Colorimetric analysis of sugars
.
Methods Enzymol
 
3
:
73
105
.
Barry
KM
Quentin
A
Eyles
A
Pinkard
EA
(
2012
)
Consequences of resource limitation for recovery from repeated defoliation in Eucalyptus globulus Labilladière
.
Tree Physiol
 
32
:
24
35
.
Bonhomme
M
Rageau
R
Lacointe
A
Gendraud
M
(
2005
)
Influences of cold deprivation during dormancy on carbohydrate contents of vegetative and floral primordia and nearby structures of peach buds (Prunus persica L. Batch)
.
Sci Hortic (Amsterdam)
 
105
:
223
240
.
Bryant
JP
Chapin
FS
III
Klein
DR
(
1983
)
Carbon/nutrient balance of boreal plants in relation to vertebrate herbivory
.
Oikos
 
40
:
357
368
.
Campo
L
Monteagudo
AB
Salleres
B
Castro
P
Moreno-Gonzalez
J
(
2013
)
NIRS determination of non-structural carbohydrates, water soluble carbohydrates and other nutritive quality traits in whole plant maize with wide range variability
.
Span J Agric Res
 
11
:
463
471
.
Chapin
FS
Schulze
E
Mooney
HA
(
1990
)
The ecology and economics of storage in plants
.
Annu Rev Ecol Syst
 
21
:
423
447
.
Cheng
J
Fan
P
Liang
Z
Wang
Y
Niu
N
Li
W
Li
S
(
2009
)
Accumulation of end products in source leaves affects photosynthetic rate in peach via alteration of stomatal conductance and photosynthetic efficiency
.
J Am Soc Hortic Sci
 
134
:
667
676
.
Chow
PS
Landhäusser
SM
(
2004
)
A method for routine measurements of total sugar and starch content in woody plant tissues
.
Tree Physiol
 
24
:
1129
1136
.
Clement
R
Keith
L
Siu
K
(
1996
)
Reference materials for environmental analysis
 .
Lewis Publishers
,
Boca Raton, FL
.
Dichio
B
Xiloyannis
C
Sofo
A
Montanaro
G
(
2007
)
Effects of post-harvest regulated deficit irrigation on carbohydrate and nitrogen partitioning, yield quality and vegetative growth of peach trees
.
Plant Soil
 
290
:
127
137
.
Dickman
LT
McDowell
NG
Sevanto
S
Pangle
RE
Pockman
WT
(
2015
)
Carbohydrate dynamics and mortality in a piñon-juniper woodland under three future precipitation scenarios
.
Plant Cell Environ
 
38
:
729
739
.
Dickson
RE
Larson
PR
(
1975
)
Incorporation of 14C-photosynthate into major chemical fractions of source and sink leaves of cottonwood
.
Plant Physiol
 
56
:
185
193
.
Dietze
MC
Sala
A
Carbone
MS
Czimczik
CI
Mantooth
JA
Richardson
AD
Vargas
R
(
2014
)
Nonstructural carbon in woody plants
.
Annu Rev Plant Biol
 
65
:
667
687
.
Drake
PL
Mendham
DS
Ogden
GN
(
2013
)
Plant carbon pools and fluxes in coppice regrowth of Eucalyptus globulus
.
For Ecol Manag
 
306
:
161
170
.
Duan
H
Amthor
JS
Duursma
RA
O’Grady
AP
Choat
B
Tissue
DT
(
2013
)
Carbon dynamics of eucalypt seedlings exposed to progressive drought in elevated [CO2] and elevated temperature
.
Tree Physiol
 
33
:
779
792
.
DuBois
M
Gilles
KA
Hamilton
JK
Rebers
PA
Smith
F
(
1956
)
Colorimetric method for determination of sugars and related substances
.
Anal Chem
 
28
:
350
356
.
El Zein
R
Maillard
P
Bréda
N
Marchand
J
Montpied
P
Gérant
D
(
2011
)
Seasonal changes of C and N non-structural compounds in the stem sapwood of adult sessile oak and beech trees
.
Tree Physiol
 
31
:
843
854
.
Emons
H
Linsinger
TPJ
Gawlik
BM
(
2004
)
Reference materials: terminology and use. Can’t one see the forest for the trees?
Trends Analyt Chem
 
23
:
442
449
.
Escobar-Gutiérrez
A
Moing
A
Gaudillère
J-P
(
1997
)
Time course of carbohydrates concentration in mature leaves of peach seedlings [Prunus persica (L.) Batsch] during the night
.
ISHS Acta Horticulturae 465: IV International Peach Symposium
, pp
337
344
.
Eyles
A
Pinkard
EA
Mohammed
C
(
2009a
)
Shifts in biomass and resource allocation patterns following defoliation in Eucalyptus globulus growing with varying water and nutrient supplies
.
Tree Physiol
 
29
:
753
764
.
Eyles
A
Pinkard
EA
O’Grady
AP
Worledge
D
Warren
CR
(
2009b
)
Role of corticular photosynthesis following defoliation in Eucalyptus globulus
.
Plant Cell Environ
 
32
:
1004
1014
.
Eyles
A
Pinkard
EA
Davies
NW
Corkrey
R
Churchill
K
O’Grady
AP
Sands
P
Mohammed
C
(
2013
)
Whole-plant versus leaf-level regulation of photosynthetic responses after partial defoliation in Eucalyptus globulus saplings
.
J Exp Bot
 
64
:
1625
1636
.
Fajardo
A
Piper
FI
(
2014
)
An experimental approach to explain the southern Andes elevational treeline
.
Am J Bot
 
101
:
788
795
.
Fajardo
A
Piper
FI
Cavieres
LA
(
2011
)
Distinguishing local from global climate influences in the variation of carbon status with altitude in a tree line species
.
Glob Ecol Biogeogr
 
20
:
307
318
.
Fajardo
A
Piper
FI
Pfund
L
Körner
C
Hoch
G
(
2012
)
Variation of mobile carbon reserves in trees at the alpine treeline ecotone is under environmental control
.
New Phytol
 
195
:
794
802
.
Fajardo
A
Piper
FI
Hoch
G
(
2013
)
Similar variation in carbon storage between deciduous and evergreen treeline species across elevational gradients
.
Ann Bot
 
112
:
623
631
.
Galiano
L
Martínez-Vilalta
J
Lloret
F
(
2011
)
Carbon reserves and canopy defoliation determine the recovery of Scots pine 4 yr after a drought episode
.
New Phytol
 
190
:
750
759
.
Gauthier
PPG
Crous
KY
Ayub
G
et al
(
2014
)
Drought increases heat tolerance of leaf respiration in Eucalyptus globulus saplings grown under both ambient and elevated atmospheric [CO2] and temperature
.
J Exp Bot
 
65
:
6471
6485
.
Giannoccaro
E
Wang
Y-J
Chen
P
(
2008
)
Comparison of two HPLC systems and an enzymatic method for quantification of soybean sugars
.
Food Chem
 
106
:
324
330
.
Gomez
L
Rubio
E
Augé
M
(
2002
)
A new procedure for extraction and measurement of soluble sugars in ligneous plants
.
J Sci Food Agric
 
82
:
360
369
.
Gomez
L
Jordan
MO
Adamowicz
S
Leiser
H
Pagès
L
(
2003
)
Du prélèvement au dosage: réflexions sur les problèmes posés par la mesure des glucides non structuraux chez les végétaux ligneux
.
Cah Agric
 
12
:
369
386
.
Gordon
D
Rosati
A
Damiano
C
Dejong
T
(
2006
)
Seasonal effects of light exposure, temperature, trunk growth and plant carbohydrate status on the initiation and growth of epicormic shoots in Prunus persica
.
J Hortic Sci Biotechnol
 
81
:
421
428
.
Graham
CJ
(
2002
)
Nonstructural carbohydrate and prunasin composition of peach seedlings fertilized with different nitrogen sources and aluminum
.
Sci Hortic (Amsterdam)
 
94
:
21
32
.
Grünzweig
JM
Carmel
Y
Riov
J
Sever
N
McCreary
DD
Flather
CH
(
2008
)
Growth, resource storage, and adaptation to drought in California and eastern Mediterranean oak seedlings
.
Can J For Res
 
38
:
331
342
.
Hall
MB
(
2013
)
Efficacy of reducing sugar and phenol–sulfuric acid assays for analysis of soluble carbohydrates in feedstuffs
.
Anim Feed Sci Technol
 
185
:
94
100
.
Handa
IT
Körner
C
Hättenschwiler
S
(
2005
)
A test of the treeline carbon limitation hypothesis by in situ CO2 enrichment and defoliation
.
Ecology
 
86
:
1288
1300
.
Hansen
J
Møller
I
(
1975
)
Percolation of starch and soluble carbohydrates from plant tissue for quantitative determination with anthrone
.
Anal Biochem
 
68
:
87
94
.
Hartmann
H
Ziegler
W
Trumbore
S
(
2013
)
Lethal drought leads to reduction in nonstructural carbohydrates in Norway spruce tree roots but not in the canopy
.
Funct Ecol
 
27
:
413
427
.
Haukioja
E
Koricheva
J
(
2000
)
Tolerance to herbivory in woody vs. herbaceous plants
.
Evol Ecol
 
14
:
551
562
.
Hoch
G
Körner
C
(
2003
)
The carbon charging of pines at the climatic treeline: a global comparison
.
Oecologia
 
135
:
10
21
.
Hoch
G
Körner
C
(
2005
)
Growth, demography and carbon relations of Polylepis trees at the world’s highest treeline
.
Funct Ecol
 
19
:
941
951
.
Hoch
G
Popp
M
Körner
C
(
2002
)
Altitudinal increase of mobile carbon pools in Pinus cembra suggests sink limitation of growth at the Swiss treeline
.
Oikos
 
98
:
361
374
.
Hoch
G
Richter
A
Körner
C
(
2003
)
Non-structural carbon compounds in temperate forest trees
.
Plant Cell Environ
 
26
:
1067
1081
.
Huang
Y-B
Fu
Y
(
2013
)
Hydrolysis of cellulose to glucose by solid acid catalysts
.
Green Chem
 
15
:
1095
1111
.
Inglese
P
Caruso
T
Gugliuzza
G
Pace
L
(
2002
)
Crop load and rootstock influence on dry matter partitioning in trees of early and late ripening peach cultivars
.
J Am Soc Hortic Sci
 
127
:
825
830
.
Jaikumar
NS
Snapp
SS
Flore
JA
Loescher
W
(
2014
)
Photosynthetic responses in annual rye, perennial wheat, and perennial rye subjected to modest source : sink ratio changes
.
Crop Sci
 
54
:
274
283
.
Johansen
HN
Glitsø
V
Bach Knudsen
KE
(
1996
)
Influence of extraction solvent and temperature on the quantitative determination of oligosaccharides from plant materials by high-performance liquid chromatography
.
J Agric Food Chem
 
44
:
1470
1474
.
Kagan
IA
Kirch
BH
Thatcher
CD
Teutsch
CD
Pleasant
RS
(
2014
)
Chromatographic profiles of nonstructural carbohydrates contributing to the colorimetrically determined fructan, ethanol-soluble, and water-soluble carbohydrate contents of five grasses
.
Anim Feed Sci Technol
 
188
:
53
63
.
Kim
S
Kim
W
Hwang
IK
(
2003
)
Optimization of the extraction and purification of oligosaccharides from defatted soybean meal
.
Int J Food Sci Technol
 
38
:
337
342
.
King
JR
Conway
WC
Rosen
DJ
Oswald
BP
Williams
HM
(
2014
)
Total nonstructural carbohydrate trends in deeproot sedge (Cyperus entrerianus)
.
Weed Sci
 
62
:
186
192
.
Kobe
RK
(
1997
)
Carbohydrate allocation to storage as a basis of interspecific variation in sapling survivorship and growth
.
Oikos
 
80
:
226
233
.
Körner
C
(
2003
)
Carbon limitation in trees
.
J Ecol
 
91
:
4
17
.
Kozlowski
TT
(
1992
)
Carbohydrate sources and sinks in woody plants
.
Bot Rev
 
58
:
107
222
.
Landhäusser
SM
Lieffers
VJ
(
2003
)
Seasonal changes in carbohydrate reserves in mature northern Populus tremuloides clones
.
Trees
 
17
:
471
476
.
Leite
G
Bonhomme
M
Lacointe
A
Rageau
R
Sakr
S
Guilliot
A
Maurel
K
Pétel
G
Couto-Rodriguez
A
(
2004
)
Influence of lack of chilling on bud-break patterns and evolution of sugar contents in buds and stem tissues along the one-year-old shoot of the peach trees. VII International Symposium on Temperate Zone Fruits in the Tropics and Subtropics
.
Acta Hortic
 
662
:
61
71
.
Li
M-H
Xiao
W-F
Wang
S-G
Cheng
G-W
Cherubini
P
Cai
X-H
Liu
X-L
Wang
X-D
Zhu
W-Z
(
2008
)
Mobile carbohydrates in Himalayan treeline trees I. Evidence for carbon gain limitation but not for growth limitation
.
Tree Physiol
 
28
:
1287
1296
.
Li
WD
Duan
W
Fan
PG
Yan
ST
Li
SH
(
2007
)
Photosynthesis in response to sink—source activity and in relation to end products and activities of metabolic enzymes in peach trees
.
Tree Physiol
 
27
:
1307
1318
.
Loomis
W
(
1932
)
Growth-differentiation balance vs. carbohydrate-nitrogen ratio
.
Proc Am Soc Hortic Sci
 
29
:
240
245
.
Lusk
CH
Piper
FI
(
2007
)
Seedling size influences relationships of shade tolerance with carbohydrate-storage patterns in a temperate rainforest
.
Funct Ecol
 
21
:
78
86
.
MacRae
JC
Smith
D
McCready
RM
(
1974
)
Starch estimation in leaf tissue—a comparison of results using six methods
.
J Sci Food Agric
 
25
:
1465
1469
.
Matuszewski
BK
Constanzer
ML
Chavez-Eng
CM
(
2003
)
Strategies for the assessment of matrix effect in quantitative bioanalytical methods based on HPLC-MS/MS
.
Anal Chem
 
75
:
3019
3030
.
McDowell
N
Pockman
WT
Allen
CD
et al
(
2008
)
Mechanisms of plant survival and mortality during drought: why do some plants survive while others succumb to drought?
New Phytol
 
178
:
719
739
.
Merchant
A
Peuke
AD
Keitel
C
Macfarlane
C
Warren
CR
Adams
MA
(
2010
)
Phloem sap and leaf δ13C, carbohydrates, and amino acid concentrations in Eucalyptus globulus change systematically according to flooding and water deficit treatment
.
J Exp Bot
 
61
:
1785
1793
.
Mitchell
PJ
O’Grady
AP
Tissue
DT
White
DA
Ottenschlaeger
ML
Pinkard
EA
(
2013
)
Drought response strategies define the relative contributions of hydraulic dysfunction and carbohydrate depletion during tree mortality
.
New Phytol
 
197
:
862
872
.
Mitchell
PJ
O’Grady
AP
Tissue
DT
Worledge
D
Pinkard
EA
(
2014
)
Co-ordination of growth, gas exchange and hydraulics define the carbon safety margin in tree species with contrasting drought strategies
.
Tree Physiol
 
34
:
443
458
.
Moing
A
Carbonne
F
Rashad
MH
Gaudillère
J-P
(
1992
)
Carbon fluxes in mature peach leaves
.
Plant Physiol
 
100
:
1878
1884
.
Mooney
HA
(
1972
)
The carbon balance of plants
.
Annu Rev Ecol Syst
 
3
:
315
346
.
Morgan
PB
Ainsworth
EA
Long
SP
(
2003
)
How does elevated ozone impact soybean? A meta-analysis of photosynthesis, growth and yield
.
Plant Cell Environ
 
26
:
1317
1328
.
Muller
B
Pantin
F
Génard
M
Turc
O
Freixes
S
Piques
M
Gibon
Y
(
2011
)
Water deficits uncouple growth from photosynthesis, increase C content, and modify the relationships between C and growth in sink organs
.
J Exp Bot
 
62
:
1715
1729
.
Nakagawa
S
Schielzeth
H
(
2013
)
A general and simple method for obtaining R2 from generalized linear mixed-effects models
.
Methods Ecol Evol
 
4
:
133
142
.
Nii
N
(
1997
)
Changes of starch and sorbitol in leaves before and after removal of fruits from peach trees
.
Ann Bot
 
79
:
139
144
.
O’Brien
MJ
Leuzinger
S
Philipson
CD
Tay
J
Hector
A
(
2014
)
Drought survival of tropical tree seedlings enhanced by non-structural carbohydrate levels
.
Nat Clim Change
 
4
:
710
714
.
O’Grady
AP
Eyles
A
Worledge
D
Battaglia
M
(
2010
)
Seasonal patterns of foliage respiration in dominant and suppressed Eucalyptus globulus canopies
.
Tree Physiol
 
30
:
957
968
.
Palacio
S
Hoch
G
Sala
A
Körner
C
Millard
P
(
2014
)
Does carbon storage limit tree growth?
New Phytol
 
201
:
1096
1100
.
Pinkard
EA
Eyles
A
O’Grady
AP
(
2011
)
Are gas exchange responses to resource limitation and defoliation linked to source : sink relationships?
Plant Cell Environ
 
34
:
1652
1665
.
Piper
FI
(
2011
)
Drought induces opposite changes in the concentration of non-structural carbohydrates of two evergreen Nothofagus species of differential drought resistance
.
Ann For Sci
 
68
:
415
424
.
Piper
FI
Fajardo
A
(
2011
)
No evidence of carbon limitation with tree age and height in Nothofagus pumilio under Mediterranean and temperate climate conditions
.
Ann Bot
 
108
:
907
917
.
Piper
FI
Fajardo
A
(
2014
)
Foliar habit, tolerance to defoliation and their link to carbon and nitrogen storage
.
J Ecol
 
102
:
1101
1111
.
Quentin
AG
Pinkard
EA
Beadle
CL
Wardlaw
TJ
O’Grady
AP
Paterson
S
Mohammed
CL
(
2010
)
Do artificial and natural defoliation have similar effects on physiology of Eucalyptus globulus Labill. seedlings?
Ann For Sci
 
67
:
203
.
Quentin
AG
Beadle
CL
O’Grady
AP
Pinkard
EA
(
2011
)
Effects of partial defoliation on closed canopy Eucalyptus globulus Labilladière: growth, biomass allocation and carbohydrates
.
For Ecol Manag
 
261
:
695
702
.
Quevauviller
P
Astruc
M
Ebdon
L
Desauziers
V
Sarradin
PM
Astruc
A
Kramer
GN
Griepink
B
(
1994
)
Certified reference material (CRM 462) for the quality control of dibutyl- and tributyl-tin determinations in coastal sediment
.
Appl Organomet Chem
 
8
:
629
637
.
Raessler
M
Wissuwa
B
Breul
A
Unger
W
Grimm
T
(
2010
)
Chromatographic analysis of major non-structural carbohydrates in several wood species—an analytical approach for higher accuracy of data
.
Anal Methods
 
2
:
532
538
.
Rastall
RA
(
1990
)
Methods in plant biochemistry. Volume 2. Carbohydrates
 .
Academic Press
,
London, UK
.
R Development Core Team
(
2014
)
R: a language and environment
 
for statistical computing
 .
R Foundation for Statistical Computing
,
Vienna, Austria
.
http://www.R-project.org (26 May 2015, date last accessed)
.
Richardson
AD
Carbone
MS
Keenan
TF
Czimczik
CI
Hollinger
DY
Murakami
P
Schaberg
PG
Xu
X
(
2013
)
Seasonal dynamics and age of stemwood nonstructural carbohydrates in temperate forest trees
.
New Phytol
 
197
:
850
861
.
Rose
R
Rose
CL
Omi
SK
Forry
KR
Durall
DM
Bigg
WL
(
1991
)
Starch determination by perchloric acid vs enzymes: evaluating the accuracy and precision of six colorimetric methods
.
J Agric Food Chem
 
39
:
2
11
.
Ruuhola
T
Keinanen
M
Keski-Saari
S
Lehto
T
(
2011
)
Boron nutrition affects the carbon metabolism of silver birch seedlings
.
Tree Physiol
 
31
:
1251
1261
.
Ryan
MG
(
2011
)
Tree responses to drought
.
Tree Physiol
 
31
:
237
239
.
Ryan
MG
Phillips
N
Bond
BJ
(
2006
)
The hydraulic limitation hypothesis revisited
.
Plant Cell Environ
 
29
:
367
381
.
Sala
A
Piper
F
Hoch
G
(
2010
)
Physiological mechanisms of drought-induced tree mortality are far from being resolved
.
New Phytol
 
186
:
274
281
.
Sala
A
Fouts
W
Hoch
G
(
2011
)
Carbon storage in trees: does relative carbon supply decrease with tree size?
In:
Meinzer
FC
Lachenbruch
B
Dawson
TE
(eds)
Size and age-related changes in tree structure and function
 .
Springer
,
Heidelberg
, pp
287
306
.
Sala
A
Woodruff
DR
Meinzer
FC
(
2012
)
Carbon dynamics in trees: feast or famine?
Tree Physiol
 
32
:
764
775
.
Santiago da Silva
CM
Habermann
G
Marchi
MRR
Zocolo
GJ
(
2012
)
The role of matrix effects on the quantification of abscisic acid and its metabolites in the leaves of Bauhinia variegata L. using liquid chromatography combined with tandem mass spectrometry
.
Braz J Plant Physiol
 
24
:
223
232
.
SAS Institute
(
2012
)
Sas/Stat 12.1 User’s Guide: The Glimmix Procedure
.
SAS Publishing
, pp
2916
3263
.
Saunders
S
Karduck
P
Sloof
WG
(
2004
)
Certified reference materials for micro-analysis of carbon and nitrogen
.
Microchimica Acta
 
145
:
209
213
.
Sevanto
S
McDowell
NG
Dickman
LT
Pangle
R
Pockman
WT
(
2014
)
How do trees die? A test of the hydraulic failure and carbon starvation hypotheses
.
Plant Cell Environ
 
37
:
153
161
.
Shi
J
Mazza
G
le Maguer
M
(
2002
)
Functional foods: biochemical and processing aspects
 .
Vol. II
.
CRC Press
,
London, UK
.
Shvaleva
AL
Silva
FCE
Breia
E
et al
(
2006
)
Metabolic responses to water deficit in two Eucalyptus globulus clones with contrasting drought sensitivity
.
Tree Physiol
 
26
:
239
248
.
Smeraglia
J
Baldrey
SF
Watson
D
(
2002
)
Matrix effects and selectivity issues in LC-MS-MS
.
Chromatographia
 
55
:
S95
S99
.
Stick
RV
Williams
S
(
2010
)
Carbohydrates: the essential molecules of life
 .
Elsevier
,
London, UK
.
Strauss
SY
Agrawal
AA
(
1999
)
The ecology and evolution of plant tolerance to herbivory
.
Trends Ecol Evol
 
14
:
179
185
.
Thompson
M
Ellison
SLR
(
2005
)
A review of interference effects and their correction in chemical analysis with special reference to uncertainty
.
Accred Qual Assur
 
10
:
82
97
.
Tworkoski
T
Glenn
D
Welker
W
(
1997
)
Carbohydrate and nitrogen partitioning within one-year shoots of young peach trees grown with grass competition
.
HortScience
 
32
:
1174
1177
.
Weibel
A
Reighard
G
Rajapakse
N
DeJong
T
(
2008
)
Dormant carbohydrate reserves of two peach cultivars grafted on different vigor rootstocks
.
ISHS Acta Horticulturae 903: IX International Symposium on Integrating Canopy, Rootstock and Environmental Physiology in Orchard Systems
, pp
815
820
.
Wiley
E
Helliker
B
(
2012
)
A re-evaluation of carbon storage in trees lends greater support for carbon limitation to growth
.
New Phytol
 
195
:
285
289
.
Wittig
VE
Ainsworth
EA
Naidu
SL
Karnosky
DF
Long
SP
(
2009
)
Quantifying the impact of current and future tropospheric ozone on tree biomass, growth, physiology and biochemistry: a quantitative meta-analysis
.
Glob Change Biol
 
15
:
396
424
.
Yemm
EW
Willis
AJ
(
1954
)
The estimation of carbohydrates in plant extracts by anthrone
.
Biochem J
 
57
:
508
514
.
Zhang
C
Shen
Z
Zhang
Y
Han
J
Ma
R
Korir
NK
Yu
M
(
2013
)
Cloning and expression of genes related to the sucrose-metabolizing enzymes and carbohydrate changes in peach
.
Acta Physiol Plant
 
35
:
589
602
.