Improving floral nectar storage on filter paper for sugar recovery

Keywords: nectar elution, nectar storage methods, nectar storage on filter paper, nectar sugar analysis, zeolite.

Nectar is the main reward for flower visitors (Simpson and Neff 1981), and its most commonly measured sugars are sucrose, glucose, and fructose (Baker and Baker 1983b). Other sugars are sometimes present in small traces (e.g. mannose, arabinose, xylose, maltose, melibiose, raffinose, melezitose, stachyose; Baker and Baker 1983b), as well as amino acids, enzymes, lipids, phenolics, salts, and alkaloids (Baker and Baker 1973; Gottsberger et al. 1984; Galletto and Bernardello 2005; Nicolson and Thornburg 2005). Floral nectar sugar characteristics have been the focus of many investigations, including how viscosity, sugar composition and concentration, and energetic reward are shaped by microclimate (Corbet 1979), nectar-dwelling yeasts and microorganisms (Herrera et al. 2009), and flower age and structure (Percival 1961; Nicolson and Van Wyk 1998). The evolutionary and behavioural consequences of nectar characteristics also represent key aspects of the pollination literature (e.g. pollinator feeding rates, preferences, distributions, and plant fitness; Kingsolver and Daniel 1995; Brown et al. 2010; Amorim et al. 2013; Staab et al. 2017).

The accuracy of ecological studies involving nectar sugar measurements depends on the method of collection, laboratory analysis, and storage. Different methods are likely to produce inconsistent results. Nectar can be sampled using microcapillary pipettes, syringes, and graduated pipettes (Kearns and Inouye 1993; Corbet 2003), by using paper wicks (McKenna and Thomson 1988; Dungan et al. 2004; Dósa 2008), by washing cut flowers (Morrant et al. 2009), and by rinsing uncut flowers (Mallick 2000). Morrant et al. (2009) and Petit et al. (2011) demonstrated that sampling techniques affected results, with washing and rinsing yielding the greatest amount of sugar among the methods they compared.

When determining nectar sugar concentrations in the field, a hand-held refractometer is the most common instrument used. However, a refractometer is unsuitable if nectar volumes are too small (e.g. <0.5 μL; Kearns and Inouye 1993; Corbet 2003), or if a detailed sugar analysis is required (Aronne and Malara 2019). In this case, chromatography in the laboratory is required. Historically, paper chromatography (PC; Baker and Baker 1983a; Kearns and Inouye 1993), thin-layer chromatography (TLC; Lewis and Smith 1969; Baker 1977), and gas chromatography (GC; Baskin and Bliss 1969) were chromatographic procedures used to identify nectar sugars. However, high-performance liquid chromatography (HPLC) is now the preferred chromatographic method for nectar sugar analyses. It permits the identification of different sugars and quantifies their masses, whereas GC requires complex methods to generate volatile derivatives of nectar before analysis, and TLC and PC are slow to run, and their results are qualitative (Gilbert 1987). HPLC results are based on precise area counts of peaks that are measured with software, rather than visual comparisons or spot intensity matching techniques, as with TLC and PC (Coskun 2016). HPLC also offers different detector, software, stationary phase, and column options (Dhandhukia and Thakker 2011; Coskun 2016) to improve separation and provide the required sensitivity.

Between collection and laboratory analysis, nectar is often refrigerated, frozen, or preserved with an antimicrobial agent to prevent changes in sugar composition. However, these treatments may result in poor sugar concentration stability over time (Morrant et al. 2009). Nectar should not be stored for extended periods because sucrose is hydrolysed to glucose and fructose (Nicolson and Thornburg 2005). Enzymes such as invertase that are found in flowers during pre-secretion can alter sugar composition (Nichol and Hall 1988; Pate et al. 1998; Nicolson and Thornburg 2005; Ruhlmann et al. 2010). Invertase activity has been reported for several species (Baker and Baker 1983a; Pate et al. 1985; Heil et al. 2005; Shenoy et al. 2012), and could be a significant contributor to altered sugar characteristics during storage. Microbial growth can also affect nectar. For example, yeasts and bacteria affected sucrose, glucose, and fructose concentrations in studies by Herrera et al. (2009) and de Vega and Herrera (2012).

Considering the lack of nectar stability, storage conditions have the potential to affect the results of a nectar study greatly. We aimed to develop an effective method to store nectar on filter paper in the field. Filter paper is a common storage medium for nectar (Galletto and Bernardello 2005). Once the nectar has air-dried on the filter paper, it can be eluted by washing with a solvent before analysis. Elution methods are often complex, non-standardised, and mostly unreported in published work; storage methods are, likewise, rarely reported (Amato and Petit 2017).

Preliminary work on nectar that was stored on filter paper and conducted over 30 days showed that sugar recovery was not consistent for sucrose in Eremophila maculata nectar (Amato 2015). We hypothesised that elution volume, size and type of filter paper, and desiccating regime could have affected the results. Desiccation inhibits enzymes and microbial metabolism in the nectar and on the filter papers and prevents microbial growth.

Consequently, we designed this study to address the following objectives: (1) to test the effect of water volume and agitation on sugar recovery on the day of collection; (2) to determine whether filter paper size and type influence sugar recovery of stored nectar; and (3) to compare drying and storage methods for nectars stored on Whatman number 1 filter paper. The results will be useful to researchers investigating sugar contents of nectar when immediate analysis is not possible and the storage of samples is required.

Eremophila maculata ssp. brevifolia (Benth.) Chinnock (Scrophulariaceae) was used to address Objective 1 (to test the effect of water volume and agitation on sugar recovery on the day of collection). Eremophila maculata ssp. maculata (Ker Gawl.) F.Muell. was used to address Objectives 2 (to determine the influence of filter paper size and type on sugar recovery) and 3 (to compare drying and storage methods). We used the bird of paradise Strelitzia reginae Aiton (Strelitziaceae) as well for Objective 3. The experiments required large volumes of nectar, which all three species had. The experiments conducted are presented in a chronological order. Eremophila m. ssp. maculata nectar was used because E. m. ssp. brevifolia had stopped flowering after the first objective. We used Strelitzia reginae nectar because we wanted to determine whether different nectar characteristics influenced storage. Eremophila m. maculata nectar is hexose dominant, whereas S. reginae nectar is balanced for the three sugars. With a 100-μL pipette, we drew up the nectar of six individual E. m. maculata, eight E. m. brevifolia, and 12 S. reginae plants throughout Adelaide, South Australia (AGD94 –34.919822, 138.612568). We sampled all the flowers from each plant, combining nectar from over 500 Eremophila flowers and 50 S. reginae flowers for their respective experiments. All nectar collection occurred between 08:00 and 09:30. Nectar was pooled into 50-μL Eppendorf tubes and stored on ice during transportation to the laboratory. We sampled unbagged E. m. maculata flowers of all ages from May to September 2015 and S. reginae flowers in July 2016.

We used HPLC to measure sugar concentrations in nectars and in water eluted from filter paper on which nectar had been stored. All solutions were filtered using a 0.45-nm nylon membrane (Celltreat Scientific Products, Shirley, MA, USA) into a 7 × 40-mm, 800-μL crimp-top vial, and sealed with 8-mm ‘silver’ seals with PTFE/silicone liners (ThermoFisher Scientific, Waltham, MA, USA). We injected 100 μL of each sample by using a LS-3200 auto-sampler (SGE, Melbourne, Vic., Australia) and chromatographed with a carbohydrate, Pb²⁺ 300 × 7.8-mm column (Benson Polymeric, Inc., Reno, NV, USA) heated to 85°C with a column-heating module. Water as solvent was pumped using a Waters 600E solvent delivery system at a flow rate of 0.8 mL min⁻¹. Detection was achieved with an evaporative light-scattering detector (Alltech Model 3300 ELSD, Flemington, NJ, USA) heated to 45°C with a gain set at 8 or a differential refractive index (RI) detector (Waters Model 431), heated to 35°C. Data acquisition and handling were performed using Baseline 815 software from Waters (Dynamic Solutions of Millipore, Ventura, CA, USA). Calibration curves between 2 and 80 mg L⁻¹ were made from stock solutions of sucrose, glucose and fructose (Sigma−Aldrich, Saint Louis, MO, USA). Fresh calibration standards were made during every objective, and we recalibrated the HPLC frequently. The detector response was non-linear over this range, and so calibration curves were constructed using a second-order polynomial fit for both detectors.

The freshly collected nectar was pooled in a 10-mL vial and magnetically stirred for 5 min. A volume of 10 mg of nectar was diluted with 15 mL of Milli-Q water (EMD Millipore, Darmstadt, Germany), and assayed using HPLC to determine the sugar masses and concentrations. The synthetic nectar (control) was prepared using sugar concentrations similar to those of the floral nectars collected from E. m. maculata, E. m. brevifolia, and S. reginae, by weighing high-purity sucrose, glucose, and fructose in 10-mL volumetric flasks and dissolving with milli-Q water. The homogenised nectar and synthetic nectar were used for each experiment.

Recovery is the ratio of the observed mass to known (true) mass. Recovery calculations assess the accuracy of the experiment and allow the determination of whether low recoveries are the result of the analytical procedure or other external factors, such as losses during the sample storage. To evaluate the instrument’s accuracy, we assayed quality-control (QC) samples at five-sample intervals. QC samples of low (∼25 mg L⁻¹), high (∼70 mg L⁻¹), and similar nectar concentrations were prepared from determined volumes of stock solutions. The instrument accuracy was calculated as the percentage deviation from the known concentration of the QC or percentage recovery value:

QC recoveries of 100 ± 2% meant that any low sugar recoveries were unlikely to have resulted from the analytical procedure. Sugar recovery calculations were also used to determine the difference between the sugar masses recovered from the filter paper and calculated known sugar masses (determined from the liquid nectar at Day 0); they were expressed as sugar percentage recovery. We chose to express the spread of the data with two standard errors to represent 95% confidence that the sample mean was within two standard errors of the population mean.

To evaluate the effects of water volume and agitation time on extraction of sugars from the filter paper, nectar was loaded onto the filter paper, air-dried and extracted immediately to minimise any chance of sugar decomposition. We cut Whatman number 1 filter paper (GE Healthcare, Chicago, IL, USA) into squares of 1.5 × 1.5 cm, wearing gloves to prevent contamination, and spotted 10 μL of the homogenised nectar on each filter-paper square. The spotted filter paper was immediately weighed on a microbalance to determine nectar mass, and air-dried for 1 min. Spotted papers were stored in labelled 15- × 10-cm plastic Clickzip plastic sandwich bags (Hercules, THA) at room temperature until analysis.

We placed each spotted paper in a 70-mL stoppered container and subjected it to one of nine treatments to redissolve the nectar (n = 3 for each treatment). Treatments consisted of three volumes of milli-Q water (5, 10, and 15 mL) added to the stoppered vial × agitation with an auto-shaker for three durations (1, 5, and 15 min). The solution was assayed using HPLC to determine the mass of sucrose, glucose, and fructose removed from the paper.

Whatman number 1 filter paper was cut into 3.0- × 3.0-cm and 1.5- × 1.5-cm squares, and Whatman 7404-004 nylon membrane filter (GE Healthcare, Chicago, IL, USA) into 1.5- × 1.5-cm squares and spotted with homogenised natural and synthetic nectars to compare the effects of different sizes and types of paper on nectar sugar recovery. Spotted squares were immediately weighed on a microbalance to determine nectar mass, then air dried. They were stored in labelled Clickzip sandwich bags until analysis. Each spotted paper was placed in a 70-mL stoppered container with 15 mL of milli-Q water and agitated for 1 min before analysis.

To evaluate the effects of different drying treatments on nectar sugar recovery, Whatman filter paper number 1 was cut into 1.5- × 1.5-cm pieces. Homogenised nectar from E. m. maculata was stored, using the following treatments: (a) spotted on filter papers that had been desiccated overnight in a forced-draught oven at 50°C, re-dried in the oven overnight, placed in 15- × 10-cm plastic Clickzip plastic bags, and frozen (n = 2); and (b) spotted on filter paper, placed in Clickzip sandwich bags with eight pieces of ∼3- × 7-mm natural zeolite (Zeolite Australia Pty Ltd, Werris Creek, NSW, Australia) and stored at room temperature (n = 2). During subsequent nectar storage experiments, E. m. maculata was no longer flowering, so we used a different species for further testing. Homogenised S. reginae nectar was stored on (c) filter papers and placed in Clickzip sandwich bags, and stored at room temperature (n = 5), and (d) filter papers placed in Clickzip sandwich bags with silica gel (0.5-g Tyvek packets, DuPont, MI, USA) and stored at room temperature (n = 5). The homogenised liquid nectars of E. m maculata and S. reginae that were left over from the filter paper treatments were transferred to 7-mL bijou vials and stored in the refrigerator. As before, papers from all four treatments were placed in a 70-mL stoppered container with 15 mL of milli-Q water and agitated for 1 min. The extracts were assayed using HPLC to determine the masses of sugars recovered from the treatments at Days 0, 7, 14, 21 and 30. Concentrations of sugars in the homogenised liquid nectar samples stored refrigerated were also determined at each of the above time intervals by dissolving 10 mg of the refrigerated nectar in 15 mL of water and assaying the solution by using HPLC.

We used the statistical package SPSS (IBM Corp., Armonk, NY, USA) to conduct analyses (α = 0.05). For Objective 1, we used a multivariate analysis of variance (MANOVA) to compare the effect of different volumes and agitation times on sugar recovery. Univariate analysis of variance (ANOVA) identified effects on individual sugars (Field 2009), and we also analysed sucrose separately. For Objective 2, sugar recoveries from the collection day were compared among the three treatments with a MANOVA. For Objective 3, sugar recoveries were compared separately among the different treatments using a repeated-measure one-way ANOVA. Repeated contrasts were used to compare each storage day against the previous one. If sphericity was not met, a Greenhouse–Geisser correction was applied to all tests. Data were tested for normality using a Kolmogorov–Smirnov goodness-of-fit test. We conducted statistical analyses only for treatments with three or more replicates.

Mean (±2 s.e., n = 3) sucrose, glucose and fructose concentrations measured on the day of collection for E. m. brevifolia nectar were 0.8 ± 0.2% (w/w), 6.2 ± 0.2% (w/w) and 6.9 ± 0.2% (w/w) respectively. The MANOVA testing the effects of elution volume and agitation time indicated a significant effect of volume (F_3,10 = 47.72, P < 0.001), highlighting sucrose (univariate ANOVA F_1,12 = 115.28, P < 0.001) and glucose (F_1,12 = 6.13, P = 0.029). We had no data for glucose and fructose at 5 mL; because of the low dilution factor, their concentrations were not within the calibration range of the HPLC (2–80 mg L⁻¹). Therefore, we performed a separate univariate ANOVA with sucrose and found that volume (F_2,18 = 122.98, P < 0.001) and volume + agitation (F_4,18 = 4.12, P = 0.015) were significant, with the main effect for volume, indicating that large volume and small agitation time should be preferrable (Table 1).

Table 1.  Mean (±2 s.e.) percentages of sugar masses recovered from nectar of Eremophila maculata ssp. brevifolia stored on filter paper at Day 0 (n = 3), for three volumes of water and three agitation durations.

We found a significant effect of different paper sizes and types on nectar sugar recovery (F_6,10 = 3.749, P = 0.032). Separate univariate ANOVA on the sugar recoveries indicated that sucrose and glucose were significantly affected by the different paper sizes and types (F_2,6 = 6.00, P = 0.037; F_2,6 = 31.33, P < 0.001 respectively; Table 2). Sucrose recovery was better on the 1.5- × 1.5-cm nylon paper than the 3.0- × 3.0-cm filter paper (P = 0.044), but the opposite was true for glucose recovery (P = 0.049). Glucose recovery was better on the 1.5- × 1.5-cm filter paper than on the 3.0- × 3.0-cm filter paper (P = 0.007). The MANOVA did not highlight a significant effect of different paper size and type on nectar sugar recovery for synthetic nectar.

Table 2.  Mean (±2 s.e.) percentages of sugar masses recovered from the nectar of Eremophila maculata maculata (univariate ANOVA significant) and synthetic nectar (non-significant) at Day 0 when stored on 1.5- × 1.5-cm Whatman number 1 filter paper (FP – small), 3.0- × 3.0-cm Whatman number 1 filter paper (FP – large), and 1.5- × 1.5-cm nylon filter paper (Nylon – small); n = 3 for each treatment.

Sucrose in E. m. maculata nectar accounted for 2–6% of the total sugar mass. When sucrose, glucose, and fructose masses were combined and the total sugar recovery was calculated, the low recovery of sucrose had little effect on the total sugar mass and recovery. The differences in the total sugar mass recovery among the three treatments were small (107.1 ± 9.3% for 1.5- × 1.5-cm filter paper and 93.08 ± 3.8% for 3.0- × 3.0-cm filter paper, and 96.7 ± 1.9% for the nylon paper).

Mean (±2 s.e., n = 3) sucrose, glucose, and fructose concentrations (%, w/w) of E. m. maculata on the day of collection were 0.9 ± 0.1%, 10.5 ± 2.9% and 11.2 ± 1.6% respectively. Both storage treatments (desiccated and frozen, and zeolite at room temperature) of E. m. maculata nectar, which is hexose-rich, produced sucrose and glucose recoveries >100% on the day of collection (Fig. 1; n = 2).

Fig. 1.  Mean sugar percentages recovered from (a) Eremophila maculata ssp. maculata and (b) a synthetic control nectar stored on a desiccated filter paper (FP) that was then frozen and filter paper stored with zeolite over 30 days. Error bars show ±2 s.e.; n = 2 for each sample.

By Day 30, mean (±2 s.e., n = 2) sucrose recovery decreased to 68.1 ± 8.3% from the desiccated and frozen treatment and 68.6 ± 2.0% from paper stored with zeolite. Fructose on the desiccated and frozen papers had a poorer initial recovery on Day 0 (90.0 ± 8.3%) than when stored with zeolite (96.0 ± 3.5%). Synthetic nectar sugar recoveries were ∼100% and were relatively consistent from Day 0 to Day 30 (Fig. 1).

Strelitzia reginae nectar was richer in sucrose than was E. m. maculata nectar. Mean (±2 s.e., n = 5) sucrose, glucose, and fructose concentrations (%, w/w) of S. reginae on the day of collection were 4.79 ± 0.28%, 4.08 ± 0.19%, and 4.20 ±0.12% respectively. Time had a significant effect on mean nectar sucrose (F_4,32 = 15.88, P < 0.001), glucose (F_4,32 = 11.99, P < 0.001), and fructose (F_2.31,18.49 = 19.73, P < 0.001) recoveries for both filter paper and filter paper + silica gel treatments (Fig. 2; n = 5). Contrasts between storage days showed significant differences in sugar recovery from the previous test day at Day 21 for sucrose (F_1,4 = 13.13, P = 0.022) and for fructose (F_1,4 = 28.55, P = 0.022), and at Day 7 for glucose (F_1,4 = 8.32, P = 0.045) when nectar was stored on filter paper with silica gel. Similarly, sucrose (F_1,4 = 38.53, P = 0.003) and fructose (F_1,4 = 21.97, P = 0.001) were significantly different from the previous test day at Day 21, and glucose (F_1,4 = 27.52, P = 0.006) at Day 7 when stored on filter paper without silica gel. Glucose recovery had surpassed Day 0 recovery by 3.5 ± 2.8% when stored with silica gel and by 6.8 ± 3.0% when stored without silica gel by Day 7.

Fig. 2.  Mean sugar percentages recovered from (a) Strelitzia reginae and (b) a synthetic control nectar stored on filter paper (FP) and filter paper with silica gel over 30 days. Error bars show ±2 s.e.; n = 5 for each sample. Asterisk (*) indicates a statistically significant difference from the previous test day (see text).

We also found a significant difference in mean sucrose recovery of S. reginae nectar when stored on filter paper with vs without silica gel (F_1,8 = 24.96, P = 0.034). On Day 30, mean (±2 s.e., n = 5) sucrose recovery from the filter papers without silica gel was 92.6 ± 1.6%. The addition of silica gel showed a small improvement in sucrose recovery (97.0 ± 2.4%).

Sugar recoveries of the synthetic control nectar stored on filter paper with and without silica gel were ∼100% at the day of collection. Time had a significant effect on the mean sucrose (F_2.98,23.84 = 13.92, P < 0.001), glucose (F_2.55,20.42 = 3.69, P = 0.014), and fructose (F_2.310,18.49 = 14.44, P < 0.001) recoveries for both treatments for synthetic nectar stored on filter paper with and without silica gel (Fig. 2; n = 5). Contrasts between storage days showed a significant difference from the previous test day at Day 7 for sucrose (F_1,4 = 10.33, P = 0.032) when the paper was stored with silica gel. When the synthetic nectar was stored on filter paper without silica gel, Day 7 was significantly different from the previous test day for sucrose (F_1,4 = 174.99, P < 0.001) and fructose (F_1,4 = 65.47, P < 0.001; Fig. 2). By Day 7, the mean (±2 s.e., n = 5) sucrose and fructose recoveries for the filter papers without silica gel were 82.1 ± 3.1% and 85.7 ± 3.2% respectively. We found a significant difference in mean sucrose (F_1,8 = 63.66, P < 0.001) and fructose (F_1,8 = 25.57, P < 0.001) recovery for synthetic nectar between filter paper with silica gel, and filter paper without.

Concentrations of glucose and fructose measured in nectar from E. m. maculata stored on filter papers with zeolite over a 30-day period were relatively stable, whereas sucrose concentrations decreased by Day 30 for both treatments (Fig. 3). For nectar stored in the refrigerator, at Day 30, sucrose was not detected, glucose was similar, and fructose concentration had a 67.3% increase compared with the day of collection.

Fig. 3.  Mean sugar concentrations (%, w/w) in Eremophila maculata ssp. maculata nectar stored on filter paper (FP) with zeolite, and on desiccated filter papers and stored in the freezer, and stored in liquid form in the refrigerator over 30 days. Error bars show ±2 s.e.; n = 2 for each sample.

Time had a significant effect on the mean sucrose (F_1.60,6.38 = 30.68, P < 0.001), glucose (F_1.60,6.41 = 12.64, P = 0.017), and fructose (F_2.19,8.75 = 28.67, P < 0.001) concentrations of S. reginae liquid nectar stored in the refrigerator. Sucrose concentration had changed significantly from the previous test day by Day 14 (F_1,4 = 49.39, P = 0.002), glucose (F_1,4 = 6.82, P = 0.050) and fructose (F_1,4 = 7.89, P = 0.050) by Day 7 when liquid nectar was stored in the refrigerator (Fig. 4). When stored on filter paper, mean sucrose (F_1,4 = 19.245, P = 0.012) and fructose (F_1,4 = 29.51, P = 0.006) concentrations had changed by Day 21, and glucose (F_1,4 = 8.30, P = 0.045) by Day 7. For nectar stored in the refrigerator, at Day 30, sucrose and glucose concentrations had decreased by 48.2% and 7.1% respectively, whereas fructose had increased by 22.7% compared with Day 0.

Fig. 4.  Mean sugar concentrations (%, w/w) in Strelitzia reginae nectar stored on filter paper (FP), filter paper with silica gel, and in liquid form in the refrigerator over 30 days. Error bars show ±2 s.e.; n = 5 for each sample. Asterisk (*) indicates a statistically significant difference from the previous test day (see text).

In the field, biologists rarely know what the sugar constituents, concentrations, or proportions are for the nectars they collect; they rely on different methods of storage and analysis to measure sugars. We used several test plant species to demonstrate variations in the effectiveness of storing nectar and eluting it. The results showed that (a) only 1 min of agitation was required for nectar elution and a large volume of water was more effective, (b) small filter paper generally performed better for sugar recovery than did large filter paper, (c) for hexose-dominant E. m. maculata nectar, sugar concentrations were relatively stable and changed by Day 14 when stored on filter paper, compared with Day 7 when stored in its liquid form in the refrigerator. The addition of zeolite made a small improvement to sugar measurements at Day 14. (d) For balanced S. reginae nectar, significant differences in sugar recovery were found at Day 7 for glucose, and at Day 21 for sucrose and fructose with and without silica gel. The addition of silica gel made an overall improvement to sucrose recovery. Refrigeration of S. reginae liquid nectar led to a change in sucrose by Day 14 and glucose and fructose by Day 7. We caution that analyses were conducted on three to five replicates and may suffer from Type II errors; it is likely that a greater number of significant results would be generated from larger sample sizes.

Water volume had a significant effect on sugar recovery; the greatest water volume had the best result and an agitation time of 1 min was as effective as was 15 min. We did not expect that agitation time would affect sugar recovery because the movement of sugar from filter paper to water is likely to be very fast. We recommend eluting with a large volume of water and a short agitation time. The amount of water will vary depending on the detection limits of the instruments being used.

Our preliminary work (Amato 2015) showed that nectar sugar recoveries were not consistent over time for E. m. maculata, and sucrose recoveries from the filter paper were often less than 65% on the day of collection. Amato (2015) had not measured the size of the filter papers on which the nectar was stored. We hypothesised that the size of the filter paper affected the recovery of sucrose, with the filter paper structure detaining sucrose molecules, preventing complete dissolution into the water. Consequently, the larger the filter paper, the more opportunity existed for sugar to remain on the filter paper. Comparatively large volumes of water would dissolve the sugar to a greater extent and reduce the driving force (concentration gradient) for re-adsorption. Our results support the hypothesis that we generated after our preliminary work because small (1.5- × 1.5 cm) pieces of paper led to better sugar recovery than did larger (3.0- × 3.0 cm) ones for natural nectar, significantly so for glucose. Sucrose and glucose appeared to behave slightly differently on filter paper and nylon paper (Table 2), and paper type should be considered in the context of the experiment. For example, nylon may be preferable for high-sucrose nectars, but the unlikely mean sucrose recovery of 111% for synthetic nectar (Table 2) indicates that more work is needed to explore the functionality of nylon paper. We recommend using pieces of paper that are as small as possible, with no ‘blank space’ around the nectar spot, and a generous amount of water that still allows the accurate detection of sugars.

Sucrose, glucose, and fructose recoveries of E. m. maculata nectar stored on filter paper with and without zeolite experienced changes within 7 days. Zeolite had a marginally positive effect compared with desiccated and frozen nectar, but we had only two replicates. Eremophila m. maculata liquid nectar that was stored in the refrigerator experienced only small changes to its sugar concentrations for up to 7 days. By Day 30, no sucrose was detected in E. m. maculata, and mould was growing in E. m. maculata and S. reginae liquid nectars that were stored in the refrigerator. Statistical analysis of sucrose, glucose, and fructose recoveries of S. reginae nectar showed that glucose was not stable, and its recovery was greater than on the day of collection by Day 7. Sucrose and fructose were stable for 14 days. The addition of silica gel made a minor improvement to the overall mean sucrose recovery. Glucose and fructose concentrations from S. reginae liquid nectar stored in the refrigerator were stable for less than a week. The benefit of storing nectar in the refrigerator is that there is no need for complex washing techniques to prepare the samples. However, we would recommend storing nectar in the refrigerator for fewer than 7 days.

The decrease in sucrose in nectar from E. m. maculata over time did not result in a great change in the nectar sugar content because this plant’s nectar is very low in sucrose. By Day 30, at least one-third of the sucrose was lost in some treatments, but this amount represented less than 0.3% of the total nectar sugar mass in E. m. maculata. By contrast, by Day 30, we reported only a 3–7% loss of sucrose from the equally proportioned sugars in the nectar of S. reginae in all filter paper treatments. The increase in glucose recovery by Day 7 for S. reginae also made a relatively small difference to the total sugar mass.

Storage in an air-conditioned laboratory contrasts with storage in many field situations, such as the wet tropics. Zeolite and silica gel in those conditions may have a greater impact, which is why we recommend their use. As opposed to desiccation and freezing, storage with zeolite or silica gel can take place immediately after collection in the field and does not require specialised equipment.

Considering the microorganisms and other components present in nectar, we predicted that the synthetic nectars would have similar or better recoveries than their natural nectar counterparts. However, the synthetic nectar made to represent S. reginae sugar concentrations had the lowest recovery of all the treatments, even if silica gel improved recovery. Because this experiment was not conducted under sterile conditions, it is possible that microorganisms in the synthetic nectar used sugar carbon for growth, thus reducing sugar recovery. No further decreases in sugar recoveries after Day 7 suggest that microbial growth was inhibited after this time, perhaps as a result of depletion of other essential nutrients (e.g. phosphorus and nitrogen). Some nectars have stabilising compounds such as proteins and enzymes that protect the nectar from microbial infestation (Carter and Thornburg 2000; Nicolson and Thornburg 2005). Nectarin enzymes found in the floral nectar of Nicotiana produce hydrogen peroxide to create a microbe-free environment in the nectar (Carter and Thornburg 2004). Jaroenkit et al. (2008) reported that in cut flowers of S. reginae, mould growth was present on the stamens and stigma, but not in the nectar. These types of compounds may reduce microbial growth in S. reginae nectar. It is wise to use sterile methods and equipment, including filter paper, in methodological experiments involving nectar.

Sugar recoveries were often greater than 100%. Sucrose hydrolyses into glucose and fructose, thus generating a greater mass of the hexose sugars. Because of the low mass of sucrose present in our nectar, it is unlikely that the high recoveries of hexose sugars were the result of sucrose hydrolysis. We checked that sugar contamination of the syringe (used to inject samples into the HPLC vials) from previous samples was not the cause of high sugar recovery (not presented in Results). Washing the syringe between each sample with distilled water eliminated sugar from previous samples and did not influence sugar recovery. We washed and assayed blank filter papers, and found no positive interference, so recoveries above 100% could not be accounted for by contamination from the filter papers. We suggest that recoveries of over 100% are the result of interference from impurities within the nectars, water evaporation from the nectar when it was weighed and spotted on the filter paper, or both factors together. Biological samples with low concentrations often show positive interference from background impurities (chemical noise), which can overshadow the peak of the analyte (sucrose, glucose, and fructose) on the chromatogram and lead to inaccurate measurements (Guo et al. 2006). Nectar was weighed with a precision of 0.1 mg when spotting it onto the filter paper. Water may have evaporated from the nectar while the microbalance was stabilising, thus recording a smaller total mass than was actually added, a higher apparent sugar concentration, and resulting in initial sugar recoveries greater than 100%. Evaporation rates of nectar droplets would vary among samples because evaporation rates are governed by the size, shape, surface area and constituents of the droplet, as well as environmental factors such as humidity, vapour concentration, and temperature (Shahidzadeh-Bonn et al. 2006; Girard et al. 2008). In the field, nectar is often collected and volumetrically measured with microcapillary pipettes (Corbet 2003), whereas in the laboratory we measured it gravimetrically on a microbalance. A closed vessel such as microcapillary pipette may reduce evaporation and the lack of precision from gravimetric measurements before analysis, although this collection method is associated with other problems (Petit et al. 2011).

During the experiment, we had to change detectors from an ELSD to RI. RI is less sensitive than is ELSD (Clement et al. 1992). Although we found no notable difference between QC recoveries using the ELSD and RI detectors for all three sugars, RI can be more susceptible to interference from impurities, which may have influenced the results for our nectar samples. The accuracy and reliability of the data can be a direct result of the type of HPLC detector (or chromatographic procedure) chosen. Errors can be large (Karkacier et al 2003). We recommend that researchers quantify the sources of error when assaying nectar sugars, including those from the instruments.

Sampling from sucrose-poor nectars requires a low volume of water to ensure that the concentration can be measured within the range of detection of the instrument. Increasing the volume of water to rinse the samples can dilute sucrose concentrations to the point where they fall under the limit of detection of HPLC. Pulsed-amperometric detection (PAD) might overcome problems of diluting nectar with low sucrose concentrations. PAD is highly sensitive and accurate, offering a low limit of detection for simple carbohydrates (Bailey et al. 2014).

We have shown several sources of uncertainty and error during the process, from collection, storage, and elution, to HPLC analysis. Variances for replicates were sometimes large. Biologists need a level of certainty from their results, especially when most ecological investigations assay only one replicate per nectar sample. For fieldwork, we recommend the use of small pieces of filter paper with a desiccant, and later elution with as large a volume of water as permitted to achieve the desired sensitivity for analysis. Analysis should be performed within a few days if high accuracy is needed; storage as liquid in the refrigerator is not recommended. However, these recommendations do not apply to the recovery of amino acids (Power et al. 2017), EFNs, and one method does not fit all purposes. Detailed storage and laboratory methods, including elution techniques, QC procedures, and detailed analytical procedures, need to be reported in published works. We suggest that other factors that may influence sugar recoveries should be investigated, such as the mass of nectar spotted on the filter paper, different nectar concentrations and proportions, different amino acids compositions, and microorganisms.

The data that support this study are available in the article and accompanying supplementary material online.

The authors declare that they have no conflicts of interest.

This work was supported by a Lirabenda Endowment Fund Research Grant (Field Naturalists Society of South Australia); Nature Foundation SA Honours research grant; two Playford Memorial Trust Science and Engineering Scholarships; UniSA Rural and Isolated Scholarship; Australian Postgraduate Award to B. Amato.