Measuring metabolic rates of small terrestrial organisms by fluorescence-based closed-system respirometry

One of the basic processes of life is the oxidation of substrates, termed metabolism (Kleiber, 1961). Simple measures of O₂ consumption to infer metabolic rate (MR) are difficult to compare across diverse species because patterns of ecology and evolution influence both MR (Isaac and Carbone, 2010; White and Kearney, 2014) and its method of measurement (Lighton, 2008; Vleck, 1987; Withers, 2001). Calculating MR has a long heritage from Lavoisier in the 1770s (Hulbert and Else, 2004; Lighton, 2008), and there are many methods for quantifying MR, but indirect calorimetry measuring O₂ consumption and/or CO₂ production has gained ascendancy owing to the ease of measuring gas exchange by living organisms (Kleiber, 1961; Lighton, 2008; Withers et al., 2016).

Two respirometry methodologies (closed and open system) are based on a simple rationale – the measurement of O₂ uptake and/or CO₂ excretion over a known time course. In a closed system, the organism is kept in a sealed container of known air volume and the change in the O₂ (and/or CO₂) composition of air is measured over a time interval (Lighton, 2008; Vleck, 1987). Typically, closed systems are constrained to one-off measurements, from the start to end of a trial, allowing only an average MR to be calculated. Closed systems can be optimised to measure small organisms, either by reducing chamber volume or by extending measurement time to maximise the amount of gas exchange. Conversely, open, flow-through systems allow continual, real-time measurement of MRs in changing states of metabolic activity (Withers, 2001). Measurement of small organisms in an open system, however, requires low flow rates to allow O₂ depletion levels to remain within detectable limits of the analyser, constraining open systems to a lower mass limit.

In both open and closed respirometry, the measurement of O₂ partial pressure (P_O2; kPa) has traditionally been made using paramagnetic (Hill, 1972; Nunn et al., 1964), fuel-cell (Torda and Grant, 1972) or polarographic gas analysers (Clark et al., 1953), whereas CO₂ partial pressure (P_CO2) measurements are determined using infra-red gas analysers (Fowler, 1949). More recently, a novel technique has emerged that measures P_O2 by predictable and repeatable fluorometric quenching of metal organic dye located inside the respirometry chamber (e.g. Köster et al., 2008; Szela and Marsh, 2005; Bradford et al., 2013). Such fluorescent analysers have been used for several years to measure MRs of small, aquatic organisms (Alton et al., 2012; Bywater et al., 2014; Köster et al., 2008; Szela and Marsh, 2005) but less commonly for small, terrestrial or diving organisms (Bradford et al., 2013; Matthews and Seymour, 2010).

Typically, closed-system respirometry has been used to measure the P_O2 change from the start to the end of a trial, by analysing an air sample from the chamber, and average MR is calculated (Vleck, 1987). A key advantage of fluorometric O₂ measurement, however, is that the non-invasive nature of the in situ fluorescence dye allows for sequential, repeated measurement of the same organism in the sealed metabolic chamber (e.g. in aquatic systems; Alton et al., 2012; Szela and Marsh, 2005). The automation of such repeated measurements has recently been specifically developed for measuring O₂ consumption in an aerial closed respirometry system of seeds (Bradford et al., 2013) and leaf tissue of Arabidopsis (Scafaro et al., 2017). Previous studies have used standardised O₂ depletion as a proxy for MR of seeds from highly selectively bred cultivars; for example, in relation to the seed quality of tomato and lettuce varieties (Bello and Bradford, 2016; Bradford et al., 2013). Here, we refine the handling and processing of closed-system fluorometric data to calculate MR in traditional units (e.g. ml O₂ h⁻¹, or ml O₂ g⁻¹ h⁻¹), which allows the quantitative comparative analysis of MR variation expected for diverse, wild organisms.

Making repeated measurements of P_O2 and continuously calculating MR provides all the benefits normally ascribed to flow-through respirometry, including the ability to detect periods of altered metabolic states such as metabolic depression (Guppy and Withers, 1999). By aligning repeated-measurement fluorescence oxygen-sensing technology with broader respirometric theory and practice, we believe this represents a major advance in the measurement of MR for the great diversity of organisms too small for existing techniques. However, the location of the fluorescent sensor within a sealed aerial respirometry chamber is problematic, compared with conventional closed-system respirometry. Air sealed within the respirometric chamber is subject to P_O2 changes not only by organismal metabolism but also by changes in air pressure within the sealed chamber (e.g. owing to ambient temperature, water vapour pressure and P_CO2 changes). This substantially alters some of the assumptions and calculations of standard closed-system respirometry (Vleck, 1987).

Here, we have adapted several of the metabolic calculations of standard closed-system respirometry (Vleck, 1987) to align repeated-measurement fluorescence closed-system aerial respirometry with expectations of conventional closed system respirometry. Since repeated-measurement closed-system aerial respirometry has been exclusively used in the field of seed biology, we have used seeds as model species to apply these calculations for resting MR (RMR). Although the calculation of MR can provide insight into patterns and processes contributing to biological variation, miscalculation of MR can easily result from methodological and arithmetical errors (Lighton, 2008). We discuss these errors, and describe how they may be accounted for, eliminated or reduced in magnitude.

We used Q₂ technology (ASTEC Global, The Netherlands) for repeated-measurement fluorometric respirometry of seeds in air-filled 1800 µl closed plastic vials (Bradford et al., 2013; Scafaro et al., 2017). The fluorometric O₂ detector is suspended from a robotic arm above a series of vials laid out in a grid; the inner surface of the vial cap is impregnated with a fluorescent dye. Automated movement of the detector provides repeated measurement of internal P_O2 for vials at user-determined intervals (Scafaro et al., 2017). The fluorometer is calibrated before each set of measurements using standards containing ambient air (designated as fractional standard atmospheric O₂; i.e. Fs_O2=1.0) and O₂-free air (Fs_O2=0.0) above a saturated sodium dithionite solution (Scafaro et al., 2017). All organismal O₂ measurements are presented as standardised O₂ fractions.

The calculations we report here account for chamber size. The Q₂ system has standardised chamber types and sizes, ranging from 5500 µl chambers on a 24-well plate (384 chambers for 16 plates) to 500, 1800 and 2200 µl chambers on a 48-well plate (768 chambers for 16 plates), and 515 µl chambers on a 96-well plate (1536 chambers for 16 plates); we used 24- and 48-well plates. We make the general assumption that the respirometry chamber is sealed, impermeable to gases, and rigid (constant volume).

We outline here the calculation of absolute MR (V̇_O2; ml O₂ h⁻¹) and highlight some assumptions that currently remain untested for repeated-measures closed-system fluorometric respirometry, modifying conventional closed respirometry calculations (Vleck, 1987). Automated repeated and long-duration measurements possible with fluorometric respirometry result in large volumes of raw, ‘noisy’ data for calculation. We have developed an R script that automates our calculations, constructed around the specific data file structure of the Q₂ software, but these should be readily translatable to other data from other systems. The R script is available in the supplementary material, and is heavily annotated to provide guidance to its use (Script 1).

The initial chamber volume at dry standard temperature and pressure (STPD) conditions must be known, and corrected for the space occupied by the organism(s) in each chamber (Vleck, 1987).

where T_a is the initial chamber temperature (°C), P_b is the initial barometric pressure (kPa) and SWVP is the saturation water vapour pressure (kPa).

The measurement of P_O2 by fluorometry occurs inside the sealed respirometry chamber, unlike conventional closed-system respirometry, where an air sample is removed from the chamber for analysis. Simplistically, the P_O2 is dependent on barometric pressure inside the closed chamber at the time of measurement (P_b; kPa), but P_b inside the respirometry chamber is not known (or measurable) over repeated measurements. The calculation of P_CO2 depends on the percent CO₂ content of chamber air (which the Q₂ does not measure), and chamber pressure. The partial pressure of N₂ inside the respirometry chamber is not known, but we assume that the molar N₂ content of the air inside the respirometry chamber remains constant, i.e. it is metabolically inert and does not exchange with N₂ inside the organism(s). The water vapour pressure of the air inside the respirometry chamber is determined by the SWVP, which we account for in calculating v_g(STPD) (Eqn 3).

To calculate MR, P_O2 (from Fs_O2) must be converted to the fractional O₂ content (F_O2; %O₂/100). In a conventional closed respirometry system (e.g. Vleck, 1987), a dry CO₂-free gas sample is analysed at the start (i.e. initial Fi_O2) and at the end (i.e. final Fe_O2) of the respirometry period, and the difference is used to calculate V̇_O2 (i.e. the P_O2 is measured and is converted to F_O2 at STPD). The F_O2 can also be measured for a second dry sample that still contains CO₂ to calculate CO₂ production, and it can be measured for a further moist sample to calculate evaporative water loss (Vleck, 1987). In essence, conventional closed-system respirometry is considered an ambient pressure system, with initial and final P_O2 values measured at atmospheric pressure and converted to Fi_O2 and Fe_O2 (which, as well as v_g, are STPD).

However, repeated-measurement fluorometry differs in that the respirometry chamber remains sealed, and the P_O2 is measured in situ. This has the advantage of allowing repeated measurements of chamber P_O2, but has the disadvantage that any change in P_b within the chamber (e.g. resulting from fluctuating temperature, CO₂ accumulation and evaporation or condensation) will proportionally influence the P_O2. As far as we can tell, such changing partial pressures between successive measurements are not currently accounted for using commercially available automated fluorescence analysers; indeed, the descriptions regarding the temperature corrections described in the Q₂ user manual (Anonymous, 2014) refer to analysis corrections made in response to the fluorometer sensor temperature, not in situ correction.

Changes in water vapour pressure inside the chamber by evaporation or condensation also alter the P_O2. In the worst case, an initially dry chamber atmosphere will almost certainly reach H₂O vapour saturation throughout a metabolic trial because of evaporation from the organism(s), which would increase internal chamber P_b. This is an important consideration for calculating v_g of the chamber prior to sealing it. As such, we advocate allowing the respirometry chamber to reach saturation water vapour pressure prior to sealing, to avoid any effects of subsequent evaporation on barometric pressure in the chamber (hence, measured P_O2). The potential errors associated with these approaches are discussed below. Once the respirometry chamber is sealed, however, and assuming the chamber is rigid, changing water vapour pressure affects P_O2 and O₂ mole fraction in opposite directions. Effectively, changes in P_O2 and mole fraction of O₂ balance each other out, which is the previously unreported basis of how the Q₂ fluorescence analyser measures P_O2 independent of chamber humidity (Bello and Bradford, 2016; Bradford et al., 2013; Van Asbrouck and Taridno, 2009).

where Fs_O2,sample denotes the raw standardised value and Fs_O2,blank denotes the raw standardised value measured for the ‘blank’ chamber.

where Fs_O2 denotes the raw standardised value, t₀ denotes the initial measurement and t_n denotes the measurement after n periods.

Repeated sampling complicates the simple calculation of MR by Vleck (1987) using Fi_O2 and a single Fe_O2, and we suggest for repeated measurements that the previous O₂ measurement be assigned as Fi_O2 and the next as Fe_O2 to calculate MR over the elapsed interval between the two measurements. This introduces substantial MR variation as the signal to noise ratio between consecutive measurements of F_O2 made over a short period of time is quite low, so we advocate a moving window average (Kenney and Keeping, 1962) or linear regression (Crispin and White, 2013) to account for this variation. This effectively applies a statistical smoothing across the data to further minimise any changes due to noise between consecutive measurements, such as negative MRs, that result from temperature and pressure oscillations in the respirometry chambers, and signal noise. Data averaging may be less necessary for organisms with higher MR.

Following Vleck (1987), RER can be assumed to range between 0.7 and 1; the maximum error associated with assumption of 0.7 or 1.0 is 6.4% and the maximum potential error is 3.2% for an assumed RER of 0.85 (Vleck, 1987).

In summary, there are a number of points where repeated-measures closed-system fluorometric respirometry differs from standard closed-system respirometry. In the latter, CO₂ and H₂O can be removed prior to measurement of the P_O2 (Lighton, 2008; Vleck, 1987), but this is not possible for fluorometric measurements because P_O2 is measured inside the sealed respirometry chamber. Rather, any evaporation after sealing the respirometry chamber needs to be accounted for to calculate v_g(STPD), changes in P_O2 over time by changes in ambient temperature and sensor drift need to be accounted for, and the contribution of CO₂ production must be accounted for by assuming an RER. If done so appropriately, in the manner that we have developed here, then the equation for calculation of V̇_O2 for repeated-measures fluorometric aerial closed-system respirometry reduces to the same as for conventional closed-system respirometry (Eqn 7).

We have used seeds as model organisms to demonstrate the calculation of MR from P_O2 measurement by repeated-measures fluorometric closed-system respirometry. We measured seed RMR, which has rarely been done by fluorometric respirometry, but has been done using other methodologies (Dalziell and Tomlinson, 2017; Garwood and Lighton, 1990). However, there are several complications when measuring RMR of seeds that must be considered.

Firstly, metabolic processes in dry seeds are limited by solute mobility, and MR is not readily detectable in seeds with moisture contents in equilibrium with an RH below 90% (Bewley et al., 2013; Walters et al., 2005). Consequently, seeds need to be hydrated sufficiently for metabolic processes to be initiated prior to respiratory measurements being made. Secondly, seeds must not be fully imbibed during measurements of respiration, as germination (a metabolically active process that produces much higher MR compared with quiescent seeds) may occur during the trial. Therefore, in our metabolic trials, to ensure germination would not occur, seeds were pre-equilibrated to 95% RH prior to respirometry measurements. The signal to noise ratio is significantly reduced in quiescent seeds compared with germinating seeds, and to compensate for this we have measured more than one individual seed per chamber. This may offset some of the intrinsic utility suggested for repeated-measures fluorometric respirometry in making single seed measurements (as per Bello and Bradford, 2016), but this is countered somewhat by our ability to measure many replicate chambers at once. Finally, the interactions between seed dormancy status and MR is not well understood (Dalziell and Tomlinson, 2017), and the relationship between seed dormancy and germination/respiration is unclear (Bello and Bradford, 2016; Booth and Sowa, 2001; Bradford et al., 2007; Footitt and Cohn, 1995). As such, we measured RMR of dormant and non-dormant seeds (see methods below).

We measured seeds of five species of Fabaceae native to Western Australia. Using seeds from a single family (i.e. a phylogenetically constrained group) limited the potential influences of embryo type and size, as well as seed dormancy, on seed MR (Dalziell and Tomlinson, 2017). Seeds were collected from wild populations in the Pilbara region of Western Australia between 2010 and 2013 and stored in a controlled environment room at 15°C and 15% RH prior to experiments. Prior to respirometry measurements, all seeds were X-rayed (Faxitron Specimen Radiography System MX-20 Cabinet, Tucson, AZ, USA) to determine seed fill (Erickson et al., 2016). Seeds that did not contain an intact embryo or showed any internal damage were discarded. Seeds were immersed in hot water (95°C) for 2 min to alleviate physical dormancy (Erickson et al., 2016), then blotted dry with paper towel. To reduce the metabolic effect of microbial contamination, all seeds (both dormant and non-dormant) were sealed in porous nylon-mesh bags and soaked in a biocide [2% solution of Plant Preservative Mixture (Plant Cell Technology, Washington, DC, USA), with two drops of the surfactant Tween 20 (Sigma-Aldrich, Castle Hill, NSW, Australia)] for 30 min, after which seeds were rinsed in autoclave-sterilized, reverse-osmosis purified water. All seeds were then placed in respirometry chambers (14 replicate 1800 µl chambers containing seven seeds for each dormant and non-dormant treatment), and placed inside an air-tight electrical enclosure box (NHP Fibox, NHP Electrical Engineering Products, Sydney, Australia) containing a non-saturated solution of LiCl (∼4.8 g per 100 ml H₂O; Hay et al., 2008) to allow seeds to equilibrate to 95% RH. The box was placed in an incubator at 20°C for 10 days. To ensure that they had reached 95% RH at the end of the incubation period, non-dormant seeds were tested with a hygrometer (HygroPalm, Rotronic, Crawley, UK). Chambers were then removed from the RH box, capped and measured at 25°C (regulated using an in-built temperature controller) at 30 min intervals over a period of 2 days, using the standard data acquisition software. These data were further analysed according to Eqns 1–7 using the accompanying R script (Script 1). To calculate the initial v_g, we used the wet mass of seeds at 95% RH, multiplied by the number of seeds in each respirometry chamber, and P_b was measured using a portable weather station (TechBrands XC0394, Electus Distributions, Rydalmere, NSW, Australia). Subsequent to measurement, to ensure successful dormancy break and seed viability, all seeds were removed from the chambers and plated in Petri dishes containing water agar (0.7% w/v), placed in an incubator at 25°C, and assessed for germination over a period of 28 days.

The majority of seed RMR should result from the combined activity of the cotyledons, plumule and radicle, which we broadly considered as the ‘embryo’. To compare the RMR of seeds between species, and to understand the interspecific effects of seed mass and embryo mass, dry masses of whole seeds and excised embryos were determined. To determine embryo masses, a subsample of 30 non-dormant seeds of each species were imbibed for 24 h, and the embryos were excised using a pair of fine forceps and a scalpel under a binocular microscope. The embryos and the seed coats were separated, and split into three replicates (i.e. 10 seeds per replicate) and placed in individual metal pans and dried for 48 h at 105°C. Average dry masses of the 30 embryos and remaining seed contents were measured using a five-point balance (Mettler Toledo, Switzerland).

Oxygen depletion was linear for all species and resulted in constant RMR following our calculations. RMR ranged between 0.26±0.062 and 2.01±0.71 µl O₂ h⁻¹ for non-dormant seeds. Dormant seeds had RMR approximately 53% lower than this (Table 1). These values are within the range reported for aerobic metabolism of seeds, showing similar metabolic depression in dormant seeds (Dalziell and Tomlinson, 2017; Garwood and Lighton, 1990).

The biological sources of error described below contribute to both interspecific and intraspecific variation in MR. Although these form the basis of the biological patterns that most ecologists and evolutionary biologists are interested in, they also contribute potential errors to the measurement of MR, and need to be appropriately addressed (Felsenstein, 1985).

Mass is the most pervasive influence on MR of organisms, overpowering the combined influences of ecology and evolution (Darveau et al., 2002; Gillooly et al., 2001). Previous repeated-measures fluorometric respirometry studies (Bello and Bradford, 2016; Bradford et al., 2013) have not considered the influence of seed mass, which has substantial implications for the calculation of initial gas volume and the expression of MR corrected for metabolically active mass. Calculation of RMR in absolute terms (i.e. millilitres O₂ consumed per hour) does not account for the potentially complex allometric effects of mass that have a pervasive influence on RMR (Dalziell and Tomlinson, 2017). RMR of seeds (Table 1) shows that, even in closely related taxa, mass correction and embryo mass correction markedly influence the species comparisons. We have avoided phylogenetic influence to a large degree by measuring closely related species, but phylogenetically independent contrast analyses (Felsenstein, 1985) should be considered for broader interspecific comparisons, especially because the phylogenetic diversity of plant seeds is greater than that of any animal group studied to date. Finally, we specifically avoided using seeds with complex dormancy types, as there is a limited understanding about how these kinds of dormancy may affect RMR, and populations of seeds with physical dormancy are simple to manipulate into non-dormant fractions. Nevertheless, differences in RMR between dormant and non-dormant seeds (Table 1) are interspecifically variable, even in closely related species, and patterns are variable depending on different mass corrections. Investigating these patterns was not our aim here, but the data that we have presented indicate that a substantial amount of research is required to do so.

Methodological errors can contribute biological errors, e.g. some seeds became contaminated with fungal and/or bacterial infections. Alternatively, when dealing with multiple individuals in a single chamber, some replicates might include more-or-less metabolically inactive individuals. These errors result in much quicker or slower O₂ depletion rates, respectively. We have removed trials where O₂ depletion rates were greater than two standardised residuals different to average O₂ depletion rates (Dataset 1). We have not specifically included this process in the calculation algorithm (Script 1) because this is one of several statistical approaches to identify outliers.

As previously stated, although the biological variation described above contributes to patterns of interest to ecologists and evolutionary biologists, these patterns need to be appropriately accounted for so as to not influence the MR measured by repeated-measures fluorometric aerial closed-system respirometry.

To account for organism volume in the respirometer (Vleck, 1987), mass is corrected to volume by using an estimate of organism density. Seed density typically ranges between 1.0 and 1.5 g cm⁻³ (Gupta and Das, 1997; Nelson, 2002). Resulting errors in gas volume by not correcting organismal volume can be considerable; for four 100 mg seeds in a 2200 µl respirometry chamber, error ranges from an ∼16% (ρ=1.0) to 24% (ρ=1.5) overestimate of V̇_O2 (Fig. 1A). As such, we advocate an appropriate organism mass density be accounted for and suggest assuming (if necessary) the median of published values of 1.2 g cm⁻³ for seeds. This results in a nominal error of approximately 1.25% if ρ is between 1.0 and 1.5 g cm⁻³.

The next potential error in the calculation of V̇_O2 also relates to v_g, on the basis of initial water vapour density, to avoid the influences of initial and changing water vapour pressure on v_g. Because saturation water vapour pressure is temperature dependent (Parrish and Putnam, 1977), the water present in saturated air changes with temperature, and hence error is also temperature dependent. Exploring the worst-case scenario where chamber atmosphere was initially dry (0% RH) but we assume saturation, the v_g error at 10°C is approximately 1% underestimated (Fig. 1B). As temperature increases to 40°C, this increases to approximately 7%. In a more realistic scenario, where chamber humidity is actually 50%, the error is approximately 2.7% at 10°C, and increases to 13% at 40°C. Although these errors are small at biologically appropriate temperatures, they need to be minimised to standardise MR measurements, especially under experimental treatments where a broader range of temperatures may be applied.

We have made several assumptions relating to P_CO2, and how this influences the changing conditions in the respirometry chamber and the subsequent effects on quantifying MR. The first effect, that initial F_CO2 is zero, is minor if the initial F_CO2 is actually at atmospheric levels, resulting in <0.04% error in v_g (current atmospheric Fi_CO2 is 0.000406; https://www.esrl.noaa.gov/gmd/ccgg/trends/). The second effect, where CO₂ accumulates throughout the measurement period, is substantial, if accumulated metabolic CO₂ production is not accounted for. RER is usually between 0.7 and 1.0, and the maximum possible errors in calculated V̇_O2 will occur when estimated RER=0.7 when actually RER=1, or vice versa (Fig. 1C) (Vleck, 1987). Overestimates of V̇_O2 result if actual RER is less than assumed RER, and underestimates result if the reverse is true, with a maximum error of approximately 6.5% (Vleck, 1987). On this basis, minimal error is obtained by assuming RER=0.85, restricting the maximum possible error to ±3.2% (Vleck, 1987). Prolonged closure of the metabolic system can cause more substantial errors of approximately 10–13% because of the greater storage capacity of organismal water for CO₂ than for O₂ (Malte et al., 2016).

Hypoxia has a potential effect on MR per se. A rule of thumb for terrestrial animals suggests that O₂ levels below 12% (atmospheric) begin to affect MR (Bradford and Seymour, 1988; Seymour et al., 1995; Wheatly, 1981); 12% O₂ is equivalent to approximately 58% of initial O₂ using the standardised Q₂ scale. As O₂ becomes limiting in the chamber, it decreases MR over time (Fig. 2A), resulting in an asymptotic O₂ depletion curve rather than the expected straight line. This can be corrected by analysing data prior to the time at which hypoxia may have occurred. However, during dynamic state applications, such as germination, O₂ limitation generates unpredictable outcomes. Peak MR, such as that estimated at the midpoint of O₂ depletion curves during germination (R₅₀; Bello and Bradford, 2016; Bradford et al., 2013) are potentially underestimated because they have been measured in hypoxic atmospheres and this potentially reduces the MR of the seeds.

The approaches to utilising repeated-measures fluorometric aerial closed-system respirometry explored here potentially expand the comparative approach to seed biology, enabling quantitative evaluation of metabolic adaptations to a range of environmental and biological factors, as explored in animals (Bartholomew, 1964; Cloudsley-Thompson, 1991; Degen, 1997). Although evolutionary patterns in the MR of animals are well documented and continue to be debated (Uyeda et al., 2017), they have yet to be explored for plant seeds. Although our study reveals the complexity of calculating accurate RMR from repeated-measures fluorometric respirometry, even greater and less predictable complexity is entailed in making dynamic state calculations, and these will require further exploration. Nevertheless, repeated-measurement fluorescence sensors extend the capabilities of closed-system respirometry in ways that potentially revolutionise the study of MR for small taxa. Although measurement of V̇_CO2 using infra-red gas analysers has been accomplished for quite small taxa (Ayton et al., 2016; Lighton and Fielden, 1995; Terblanche et al., 2004), fluorometric closed-system respirometry has the capacity to extend this lower size limit substantially, including different life history stages such as eggs and larvae (Köster et al., 2008; Szela and Marsh, 2005). Measurements of the MR of such organisms, particularly as a response of thermal performance, are often crucial to understanding ecophysiological responses to changing environments (Kearney et al., 2010; Tomlinson et al., 2018). As energetics is the basis of many processes structuring ecosystems (McNab, 2002; Tomlinson et al., 2014), we expect great potential insights to be gained from measurement of MR using repeated-measurement fluorometric respirometers for small organisms.

Dr Shane Turner provided useful insight into seed physiology throughout the development of this analysis.