Growth temperature influences the underlying components of relative growth rate: an investigation using inherently fast- and slow-growing plant species


O. K. Atkin. E-mail:


We examined the effect of growth temperature on the underlying components of growth in a range of inherently fast- and slow-growing plant species. Plants were grown hydroponically at constant 18, 23 and 28 °C. Growth analysis was conducted on 16 contrasting plant species, with whole plant gas exchange being performed on six of the 16 species. Inter-specific variations in specific leaf area (SLA) were important in determining variations in relative growth rate (RGR) amongst the species at 23 and 28 °C but were not related to variations in RGR at 18 °C. When grown at 18 °C, net assimilation rate (NAR) became more important than SLA for explaining variations in RGR. Variations in whole shoot photosynthesis and carbon concentration could not explain the importance of NAR in determining RGR at the lower temperatures. Rather, variations in the degree to which whole plant respiration per unit leaf area acclimated to the different growth temperatures were responsible. Plants grown at 28 °C used a greater proportion of their daily fixed carbon in respiration than did the 18 and 23 °C-grown plants. It is concluded that the relative importance of the underlying components of growth are influenced by growth temperature, and the degree of acclimation of respiration is of central importance to the greater role played by NAR in determining variations in RGR at declining growth temperatures.


Global climate change is resulting in increases in the concentration of atmospheric CO2 as well as changes in the daily, seasonal and annual mean temperatures experienced by plants. Numerous studies have examined the impacts of rising CO2 on plant growth (e.g. Poorter, Roumet & Campbell 1996; Lloyd 1999) making it possible to predict the response of plants to increasing atmospheric CO2. In most species, elevated CO2 stimulates photosynthesis and consequently growth (Gunderson & Wullschleger 1994). Inherently fast-growing species exhibit a greater absolute relative growth rate (RGR, increase in plant mass per starting unit mass and time) response to elevated CO2 than their slow-growing counterparts (Poorter 1993, 1998; Atkin et al. 1999). The response to temperature is, however, more complex. As most plant processes are temperature sensitive (Kramer & Kozlowski 1979), changes in temperature can have markedly different effects on the underlying physiological and biomass allocation parameters that contribute to plant growth. To model the response of plants to global climate change, it is necessary for the effect of temperature on the underlying components of plant growth to be more clearly understood.

Temperature can influence the RGR of a given plant, as can other abiotic factors such as water and nutrient availability and irradiance. In addition to these short-term environmentally induced differences in RGR, plants adapted to resource-limiting habitats (e.g. nutrient-poor, arctic, alpine or low-rainfall environments) often exhibit lower maximum RGR values than their counterparts from resource-rich environments (Grime & Hunt 1975; Chapin 1980). In most studies, variations in RGR amongst species are most closely associated with variations in the leaf area ratio (LAR, ratio of leaf area to plant dry mass), with plants having an inherently low RGR exhibiting a low LAR (e.g. Higgs & James 1969; Smeets & Garretsen 1986; Dijkstra & Lambers 1986; Poorter & Remkes 1990; Atkin, Botman & Lambers 1996a; Atkin et al. 1998). LAR is made up of two components; leaf mass ratio (LMR, proportion of plant mass allocated to leaves) and specific leaf area (SLA, ratio of leaf area to leaf dry mass). SLA is usually considered more important in determining LAR (and thus RGR) than LMR. However, net assimilation rate (NAR, increase in plant mass per unit leaf area and time) is more closely correlated with changes in RGR in some studies (Eagles 1967; Pons 1977; Oberbauer & Donnelly 1986; Popma & Bongers 1988; Veenendaal et al. 1996). Clearly, the importance of NAR and LAR (and thus SLA and LMR) in determining the RGR varies.

Most studies investigating the relationship between RGR and its underlying components in contrasting plant species have grown plants at a single growth temperature (typically 20–25 °C). The effects of different growth temperatures on RGR and its underlying components are not, however, well understood. Patterson (1993) examined the effect of different growth temperatures on the growth of Galega officinalis and Medicago sativa. Although the LAR of G. officinalis increased with increasing night-time temperature, it was not significantly influenced by day-time temperature. In M. sativa LAR was not affected by either night- or day-time growth temperature. The NAR of G. officinalis decreased with increasing night temperature whereas changes in NAR in M. sativa were not consistent. In an experiment by Bruhn, Leverenz & Saxe (2000) in which Fagus sylvatica seedlings were exposed to four temperature regimes ranging from 2·9 °C below ambient to 4·8 °C above ambient, it was found that NAR was not significantly influenced by growth temperature. Stirling et al. (1998) also found that a 3 °C elevation of growth temperature from ambient did not affect the NAR of five species of fast-growing annuals. None of the above studies compared the effect of temperature on the underlying components of RGR in a range of inherently fast- and slow-growing plant species.

If NAR were influenced by temperature, the effects of this on plant growth are likely to be partly mediated via its effect on photosynthesis and respiration, as NAR largely reflects the rate of photosynthetic carbon gain minus carbon released by respiration (Lambers & Poorter 1992). Respiration and photosynthesis are both sensitive to short-term changes in temperature, with the short-term temperature-response curve for these processes varying between species (Wardlaw 1979; Atkin, Edwards & Loveys 2000). However, the impact of long-term temperature changes on respiration and photosynthesis depends on the degree of thermal acclimation exhibited by these processes. Acclimation can be assessed in several ways. A common method is to determine whether respiration rates are homeostatic across a range of growth temperatures. However, in the event that acclimation does not result in perfect homeostasis, partial acclimation to a high growth temperature is still thought to have occurred if increases in specific respiration rates across growth temperatures are less than would be predicted based on short-term temperature responses (i.e. Q10, the proportional increase in rate per 10 °C rise) (Larigauderie & Körner 1995). In any case, acclimation of respiration results in reduced efflux of CO2 at higher temperatures (in warm-acclimated plants) or increased efflux at lower temperatures (in cold-acclimated plants). It is well known that in some plant species physiological processes acclimate to changes in temperature over periods as short as several days (Rook 1969; Pearcy 1977; Tranquillini, Havranek & Ecker 1986; Tjoelker, Reich & Oleksyn 1999a; Atkin, Holly & Ball 2000). However, other species exhibit little or no acclimation (Tjoelker, Oleksyn & Reich 1999b). The degree of acclimation is thus highly variable amongst species.

It has been shown that slow-growing plant species respire a greater proportion of their acquired carbon than fast-growing ones [i.e. the ratio of respiration to photosynthesis (R/P) is greater in slow-growing species; Poorter, Remkes & Lambers 1990; Atkin et al. 1996a] when plants are grown at 20 °C. It is not known, however, whether the interspecific differences in R/P observed at 20 °C are maintained when plants are grown at lower or higher temperatures. Changes in temperature may have markedly different effects on the efficiency of respiratory energy production and/or use in fast- and slow-growing species. Moreover, it is not known whether fast- and slow-growing species differ systematically in the extent to which photosynthesis and/or respiration acclimates. Inter-specific differences in the respiratory and photosynthetic acclimation may result in substantial changes in the R/P-values exhibited by individual species.

In this study, the impact of three growth temperatures on the physiology, biomass allocation and growth of 16 species differing in relative growth rate were examined. To obtain a wide range of RGRs, we selected trees, shrubs, grasses and forbs that originated from diverse geographical regions. The selected species are likely to be adapted to a wide range of temperature environments, with the temperature optima for growth potentially differing amongst the species. Although this may influence their response to the temperature treatments imposed in our experiments, comparisons of such a diverse range of species are necessary if we are to generalize about the response of RGR and its underlying components in contrasting plant taxa to temperature. Whole-plant gas exchange measurements were used to assess the effect of temperature on the underlying components of NAR and also whether changes in growth temperature alter the percentage of photosynthates that is respired (R/P) each day. Although pairs of closely related species from contrasting habitats were selected [enabling comparisons of the temperature response within phylogenetic independent contrasts (PICs)], phylogenetic analysis of the data was beyond the scope of the current paper. A future paper will assess whether the relationships observed in the current study among PICs also exist within each PIC.

Materials and methods

Plant material

Several fast- and slow-growing plant species with a range of growth forms were chosen. Table 1 shows all the species included in the study, whether they were fast- or slow-growing and the native distributions of each species. Seeds of Acacia melanoxylon and A. aneura were prepared as described by Atkin et al. (1998); once the radicle had emerged the germinated seeds were transferred to trays of 1 : 1, sand : vermiculite. Seeds from all other species were sown on trays of John Innes F2 compost. The seed of the Eucalyptus species required vernalization at 4 °C for 6 weeks prior to germination. Once the roots of all species had reached at least 3 cm in length the seedlings were carefully removed from the compost or sand/vermiculite and the roots thoroughly washed with deionized water. They were then transferred to 16 L hydroponic tanks filled with a fully aerated modified Hoaglands nutrient solution (Poorter & Remkes 1990). An additional nitrogen supply (2 mm NH4NO3) was added to tanks in which Luzula acutifolia was grown. The pH of all solutions was adjusted daily to 5·8, and the solution was changed weekly. Initial establishment of the plants took place at 22 ± 2 °C (day) and 14 ± 2 °C (night) with a 16 h day. Subsequent growth of plants on hydroponic culture took place in Conviron E15 growth cabinets (Conviron, Winnipeg, Canada) with a 14 h day, 300 µmol m−2 s−1 PPFD provided by a combination of 400 W metal halide and 400 W high-pressure sodium bulbs. The temperature treatments in the Conviron cabinets were constant 18, 23 and 28 °C.

Table 1.  Distribution and general habitats of all 16 species
Acacia aneuraS F.Muell. Ex Benth.Central regions AustraliaLow rainfall, high temps
Acacia melanoxylonF· R.BrCoastal regions of SE AustraliaHigh rainfall, moderate temps
Achillea millefoliumF· L.British Isles, Europe and W AsiaDry grasslands
Achillea ptarmicaS L.British Isles, Europe (excluding Med.),
Asia Minor, Caucasus, Siberia
Damp acid soils
Eucalyptus dumosaS· Cunn. Ex Schauer.SE AustraliaDry warm region
Eucalyptus delegatensisF· R.Baker.SE AustraliaCool damp, upper slopes
Geum rivaleS· Linn.British Isles, Europe (excluding Med.),
Caucasus, Asia Minor, Siberia, N America
Damp, shade
Geum urbanumF· L.British Isles (except Shetland), S,C,E Europe,
Azores, Crete, W. Asia, North Africa
Fertile soils, shade
Luzula acutifoliaS· Nordenskiold.Alpine/subalpine Australian AlpsAlpine, wet sites
Plantago euryphyllaS· BG Briggs.Alpine/subalpine Australian AlpsAlpine, dry sites
Plantago lanceolataF L.British Isles, W, C, S, E Europe, N AfricaLowland grassland
Plantago majorF· L.British Isles, Europe N Africa, N and C. AsiaLowland grassland
Poa costinianaS· J. Vickery.Alpine/subalpine Australian AlpsAlpine, wet sites
Poa trivialisF· L.British Isles, Europe, temperate Asia, N Africa,
Macronesia,N America
Lowland, damp grassland and woods
Silene dioicaF· L.British Isles, N Europe (excluding Med.),
C. Asia. N Africa, Greenland
Fertile soils, shade
Silene unifloraS· Roth.British Isles, W. EuropeLow nutrient soils

Growth analysis

Four harvests of six plants were performed over a period of 20–30 d; the first harvest occurred only after plants had been in hydroponics for at least 7 d. Harvests were performed at regular intervals, which varied depending on the growth rate of each species. Plants were divided into leaves, phyllodes (if present), stems and roots. Fresh and dry mass (following freeze drying) (Edwards Modulyo Freeze Drier, York, UK) was recorded. Leaf and/or phyllode area were measured using a Li-Cor LI-3000 A leaf area meter (Licor Inc., Lincoln, NE, USA). The leaf area of Poa costiniana was underestimated using the LI-3000 A as its leaves are folded; we corrected the leaf area by forcing the leaves flat and scanning them on a flat-bed scanner and measuring leaf area. The four replicate plant parts were pooled at each harvest and ground to a fine powder using either, a mortar and pestle or a hammer mill (31–700 Hammer Mill; Glen Creston, Stanmore, UK) and analysed by mass spectrometry (CE Instruments NA2100 Brewanalyser; ThermoQuest Italia S.p.A. Milan, Italy) for total C and N concentration.

The RGR was estimated as the slope of a linear regression through the natural logarithm of total dry mass plotted against time. In the first instance all four harvests were used in the calculation of RGR. However, in some cases there was evidence that interplant shading had reduced the growth rate at the final harvest; in such cases the final harvest was not included in the calculation of RGR values. Similarly, the first harvest of Poa trivialis was removed due to a lag in the recovery of the plants from root disruption upon transfer to hydroponics (as demonstrated by the lack of plant growth between the first and second harvests). The average NAR value for each species was calculated by dividing each species RGR value by the average LAR value for all harvests.

Gas exchange

Whole-plant gas exchange was measured on six of the 16 species used in the growth analysis; Silene uniflora, S. dioica, Acacia aneura, A. melanoxylon, Poa costiniana and P. trivialis. Seeds of these species were germinated and seedlings transferred to hydroponics tanks as described above. Once the 12 seedlings of each species had been in hydroponics for approximately 7 d (14 d in the case of A. aneura) they were placed in small plastic containers containing nutrient solution and cotton wool and transported by air to Utrecht University, The Netherlands. The plants were placed in identical conditions in Utrecht as described above for York and were only out of aerated hydroponics tanks for approximately 6 h. The plants were allowed to re-establish for 7 d prior to experimentation beginning. Whole shoot photosynthesis and dark respiration and root respiration were measured in four randomly selected plants of each species at each growth temperature (18, 23 and 28 °C) when the plants had reached a total fresh mass of approximately 3–6 g (Poorter & Welschen 1993). Intact plants were placed into cuvettes with the roots and shoots in separate compartments. The root compartment was filled with 1 L of aerated nutrient solution as describe above with the addition of 10 mm 2-(N-morpholino)ethanesulfuric acid. The irradiance and temperature were the same as that in the growth cabinets. Plants were allowed to equilibrate for 1·0–1·5 h after placing them in the cuvettes before a rate of photosynthesis was measured. Root respiration was measured immediately after the photosynthesis measurement. Shoot respiration was recorded after the plants had been in darkness for approximately 0·5 h. The CO2 fluxes from the shoot and root compartments were measured using a Licor 6262 infra red gas analyser (Licor Inc.) in an open system.

The percentage of daily net photosynthesis used in respiration (R/P) was calculated by the following formula:


where RSm is specific shoot dark respiration, Sm is shoot mass (g), 10/24 and 14/24 are the relative lengths of the dark and light periods, respectively, RRm is the specific root respiration, Rm is the root mass (g), Pm is specific net photosynthesis (i.e. CO2 uptake by carboxylation minus CO2 release by photorespiration and dark respiration in the light) and Lm is leaf mass (g). Rates of RSm, RRm and Pm calculated on a per second basis were converted to a per day basis via multiplication by 86 400. The resultant daily rates were then weighted according to the number of hours per day that each gas exchange parameter took place (i.e. RSm, RSm, and Pm daily rates were multiplied by 24/24, 10/24 and 14/24, respectively). RRm was assumed to be constant over light and dark periods as shown by Scheurwater et al. (1998). Although shoot respiration also continues in the light, establishing the rate of RSm in the light is problematic. In addition to the technical difficulties in estimating the rate of respiration in the light for whole shoots, the degree to which light inhibits leaf respiration is highly variable, especially amongst contrasting temperatures (Atkin et al. 2000). We therefore decided to only include RSm in darkness when calculating daily whole plant respiration. Furthermore, we calculated R/P using values of net photosynthesis rather than rates of gross photosynthesis (i.e. carboxylation). R/P-values are commonly based on rates of RSm in darkness and net photosynthesis (e.g. Poorter et al. 1990; Atkin et al. 1996a); our approach thus enables direct comparisons to be made with previous studies. In any event, calculation of gross photosynthesis would require accurate estimates of RSm in the light and photorespiration in whole shoots (neither of which were measured in our study).

Fresh and dry mass of leaves, stems and roots were determined. Tissues were freeze-dried in a Virtus, Unitop 600 SL freeze dryer (Gardiner, New York, USA). Leaf area was determined using a LiCor, 3100 leaf area meter (Licor Inc.).


All data were tested for normality and homogeneity using a Kolmogorov–Smirnov test in SPSS version 10 (SPSS Science, Birmingham, UK). Linear regressions were performed in Sigma Plot v5 (SPSS Science). Slope analysis of linear regressions was performed using GraphPad Prism v2 (GraphPad Software, San Diego, CA, USA). One-way analyses of variance with Tukey post-hoc tests and Students t-tests were also performed using GraphPad Prism v2.


The RGR values of the 16 species used in this study ranged from 22·6 mg g−1 d−1 for A. aneura grown at 18 °C to 258 mg g−1 d−1 for P. major also grown at 18 °C (Table 2). Six of the 16 species achieved their maximum RGR when grown at 18 °C whereas the remaining 10 reached their highest RGR when grown at 23 °C. None of the species grew fastest when the growth temperature was 28 °C.

Table 2.  Mean relative growth rate (RGR, mg g−1 d−1), specific leaf area (SLA, m2 kg−1), leaf mass ratio (LMR, g leaf g−1 plant) and net assimilation rate (NAR, g m−2 d−1) values of all 16 species when grown at 18, 23 and 28 °C. SLA and LMR values are the average of 18–24 plants sampled from four harvests (± SE)
18 °C23 °C28 °C18 °C23 °C28 °C18 °C23 °C28 °C18 °C23 °C28 °C
Acacia aneura 2363 50 6·4 ± 0·511·9 ± 0·710·9 ± 0·50·46 ± 0·030·63 ± 0·010·61 ± 0·01 8·6 7·77·9
Acacia melanoxylon10799 9911·3 ± 0·418·1 ± 0·721·1 ± 0·50·58 ± 0·030·52 ± 0·010·55 ± 0·0116·410·58·5
Achillea millefolium25119722317·9 ± 0·524·4 ± 0·728·1 ± 1·10·65 ± 0·010·58 ± 0·020·70 ± 0·0119·314·29·2
Achillea ptarmica18718215019·7 ± 0·630·1 ± 1·426·2 ± 1·80·58 ± 0·010·53 ± 0·020·63 ± 0·0115·513·29·1
Eucalyptus delegatensis 67150 5411·7 ± 0·617·1 ± 0·917·8 ± 0·70·74 ± 0·020·72 ± 0·010·67 ± 0·01 7·714·12·7
Eucalyptus dumosa 7811110510·9 ± 0·415·3 ± 0·619·1 ± 0·90·75 ± 0·020·69 ± 0·010·65 ± 0·01 9·710·68·4
Geum rivale103119 9625·8 ± 0·529·7 ± 0·628·4 ± 1·20·61 ± 0·010·64 ± 0·010·70 ± 0·01 6·6 5·25·1
Geum urbanum11316611124·7 ± 0·930·8 ± 0·929·5 ± 0·70·71 ± 0·010·67 ± 0·010·69 ± 0·01 5·5 5·65·4
Luzula acutifolia122138 5714·7 ± 0·416·7 ± 0·714·6 ± 0·50·58 ± 0·010·55 ± 0·010·56 ± 0·0114·214·96·4
Plantago euryphylla 60104 3911·2 ± 0·516·6 ± 0·515·2 ± 0·60·79 ± 0·010·74 ± 0·010·67 ± 0·01 6·0 8·53·9
Plantago lanceolata17826719315·9 ± 0·626·8 ± 0·929·7 ± 1·10·68 ± 0·020·64 ± 0·030·73 ± 0·0116·012·99·0
Plantago major25815222817·0 ± 0·725·2 ± 0·830·6 ± 0·50·75 ± 0·010·68 ± 0·010·69 ± 0·0112·6 8·09·4
Poa costiniana 87134 7921·9 ± 1·531·2 ± 1·321·2 ± 1·20·54 ± 0·020·56 ± 0·020·66 ± 0·02 8·6 7·77·8
Poa trivialis21019116329·0 ± 1·133·1 ± 1·137·8 ± 1·10·52 ± 0·020·51 ± 0·020·58 ± 0·0211·6 6·26·6
Silene dioica21522817717·9 ± 0·432·9 ± 0·933·7 ± 1·20·73 ± 0·010·63 ± 0·010·66 ± 0·0214·811·07·7
Silene uniflora16816016014·9 ± 0·522·6 ± 0·625·8 ± 0·50·65 ± 0·050·62 ± 0·010·69 ± 0·0117·213·28·1

As growth temperatures influenced the growth rate, we examined the underlying components of RGR to determine the mechanism(s) for such changes. There was considerable interspecific variation in the relationship between SLA and RGR (Fig. 1a). However, overall there was a positive relationship between the two factors when data from all growth temperatures were combined. In most species, SLA increased as growth temperature increased (Table 2; Fig. 1a), with the absolute increase in SLA being greater in the fast-growing species than in their slow-growing counterparts (Fig. 2a). The importance of SLA in explaining variations in RGR differed amongst the three growth temperatures (Fig. 1a). At 18 °C the relationship between SLA and RGR was not significant (P = 0·07), with large variations in RGR occurring in plants that exhibited limited variations in SLA. At higher growth temperatures SLA became important in explaining differences in RGR; the slope of the SLA versus RGR plot was steeper at 23 and 28 °C than at 18 °C (see legend of Fig. 1 for details). There was no significant difference between the slopes of the regressions at 23 and 28 °C. The r2 values of the plots were also greater at 23 and 28 °C than at 18 °C (see legend of Fig. 1). The SLA values of the 23 °C-grown plants were between those of the 18 and 28 °C treatments.

Figure 1.

Components of relative growth rate (RGR, mg g−1 d−1) at all three growth temperatures. (a) specific leaf area (SLA, m2 kg−1); (b) leaf mass ratio (LMR, g g−1); (c) net assimilation rate (NAR, g m−2 d−1) for plants grown at 18 (○), 23 (▪) and 28 (▴) °C. Each value is the mean of 18–24 plants sampled over four harvests. Values in (c) were calculated as described in the Materials and Methods. Linear regressions shown by a solid line are significant and by a dashed line are not significant. The equations for the linear regressions of SLA and NAR were: SLA18 = 11·5 + 0·038 × RGR, P = 0·07, r2 = 0·2; SLA23 = 10·3 + 0·088 × RGR, P = 0·045, r2 = 0·45; SLA28 = 12·5 + 0·095 × RGR, P = 0·0003, r2 = 0·62; NAR18 = 4·2 + 0·062 × RGR, P < 0·0001, r2 = 0·71; NAR23 = 5·8 + 0·030 × RGR, P= 0·002, r2 = 0·48; NAR28 = 4·0 + 0·028 × RGR, P < 0·0001, r2 = 0·67.

Figure 2.

Difference in (a) specific leaf area (SLA, m2 kg−1) and (b) net assimilation rate (NAR, m2 kg−1) values exhibited by plants grown at high temperatures minus those grown at low temperatures, for 16 species differing in RGR (mg g−1 d−1). All values are plotted against the maximum RGR exhibited by each species. A value of zero indicates there was no difference between the high and low temperature grown plants. Symbols: (○), 23 °C-values minus values at 18 °C; (•) 28 °C values minus values at 18 °C. The regressions shown are all significant, indicating that the effect of temperature differed between fast- and slow-growing species. The regression was not significant when the difference in SLA and NAR values exhibited by the 28 and 23 °C-grown fast- and slow-growing plants were compared (data not shown). Each point represents a single species.

Variations in LMR did not explain differences in RGR in most published studies, and in our study growth temperature had no systematic effect on LMR (Fig. 1b). There were some changes in LMR within species but these changes were not consistent across the range of RGRs.

The final component of growth to examine is NAR (Fig. 1c). NAR was positively correlated with RGR at all growth temperatures. For a given RGR, NAR was higher at 18 than at 28 °C. The extent to which variations in NAR were associated with variations in RGR was greater at 18 °C than at 23 and 28 °C (i.e. the slope of the NAR versus RGR plots was steepest for the 18 °C-grown plants; see legend of Fig. 1). In the slow-growing species, growth temperature had little effect on the NAR (Fig. 2b) whereas growth temperature had a greater impact on NAR in the fast-growing species (Fig. 2b). In the fast-growing species, NAR values for plants grown at 18 °C were substantially higher than the NAR values exhibited by the plants grown at 23 and 28 °C.

Figure 3 shows the RGR at each growth temperature as a proportion of the maximum RGR achieved by each species, plotted against maximum RGR. Although temperature affected the growth of most species, a greater proportional difference in RGR occurred in several of the slow-growing species (0·1–0·7 of the maximum) than in the fast-growing plants (typically 0·1–0·4 of the maximum).

Figure 3.

Ratio of the relative growth rate (RGR, mg g−1 d−1) exhibited at each temperature divided by maximum RGR exhibited by each species. A ratio of one indicates the temperature at which maximum RGR was exhibited in each species. Symbols: (○), RGR at 18 °C divided by the maximum RGR; (▪), RGR at 23 °C divided by the maximum RGR; (▴), RGR at 28 °C divided by the maximum RGR. Each point represents a single species.

To explain why the relationship between NAR and RGR changed with growth temperature, we examined the rates of daily whole shoot photosynthesis and daily whole plant respiration (both expressed on a leaf area basis) in six of the 16 species. NAR (g m−2 d−1) reflects photosynthetic carbon gain minus carbon released by respiration, according to:


where Pa is daily whole shoot net photosynthesis per unit leaf area (mmol CO2 m−2 d−1; Fig. 4a), Ra is whole plant respiration per unit leaf area (mmol CO2 m−2 d−1; Fig. 4b) and CC is whole plant carbon concentration (mmol C g−1; Fig. 5). Ra is the sum of root respiration over a 24 h period plus shoot respiration during the 10 h night (both expressed on a leaf area basis; see text associated with Eqn 1 for an explanation of how the daily rates of net photosynthesis and respiration were calculated). Changes in the relationship between NAR and RGR must therefore be associated with changes in the relationship between Pa, Ra and/or CC with RGR.

Figure 4.

Variations in daily CO2 exchange in six of the 16 selected species. (a) Daily whole shoot photosynthesis per unit leaf area (Pa, mmol CO2 m−2 d−1); (b) Daily whole-plant dark respiration per unit leaf area (Ra, mmol CO2 m−2 d−1); and (c) The ratio of daily whole plant respiration to daily photosynthesis (i.e. R/P) of S. dioica (•), S. uniflora (○), A. melanoxylon (▪), A. aneura (□), P. trivialis (▾) and P. costiniana (▿) grown and measured at 18, 23 and 28 °C as indicated. In (c), significant linear regression shown by solid lines and not significant by dashed lines. Each value is the mean of four replicate plants (± SE).

Figure 5.

Total whole plant carbon concentration (mmol C g dry mass−1) of all 16 species grown at 18 °C (○), 23 °C (▪) and 28 °C (▴) and their respective relative growth rates. Each value is the mean of 18–24 plants sampled from four harvests. Equations for the linear regressions shown are; CC18 = 31·6 + 0·0175 × RGR, P = 0·01, r2 = 0·37; CC23 = 35·1 − 0·023 × RGR, P = 0·01, r2 = 0·31; CC28 = 34·9 − 0·024 × RGR, P = 0·0002, r2 = 0·62.

Data were not available for Acacia aneura grown at 18 °C as their slow growth rate resulted in the plants not being large enough to conduct whole plant gas exchange measurements. The RGR values used in the following figures are those calculated from growth analysis performed on plants in York and presented in Fig. 1 and Table 2. To check that these values were accurate for the plants that were moved to Utrecht we estimated the RGR of the Utrecht plants from initial seedling mass and final fresh weight. There was a significant relationship between the two estimates of RGR (P = 0·001, r2 = 0·52; data not shown). There was no relationship between RGR and daily photosynthesis on an area basis (Pa) and no consistent change in Pa when plants were grown at different temperatures (Fig. 4a). Daily whole plant respiration (root plus shoot) expressed on a leaf area basis (Ra) was negatively correlated with RGR at all three growth temperatures, but for individual growth temperatures the relationship was only significant at 28 °C (P = 0·016, r2 = 0·8) (Fig. 4b). Carbon concentration was greater in fast-growing species in plants grown at 18 °C in comparison with those grown at 23 and 28 °C (Fig. 5). The slope between RGR and CC was positive and significant at 18 °C (P = 0·01, r2 = 0·32; Fig. 5). In contrast, at 23 and 28 °C the relationship between RGR and CC was negative, particularly at 28 °C (P = 0·0002, r2 = 0·62).

The percentage of daily photosynthesis respired (R/P) (Eqn 1) gives an indication of the efficiency of plant carbon use. Figure 4c shows that there was an overall negative correlation between R/P and RGR when all temperatures were considered together. However, for individual temperatures, the relationship was significant only in the case of the 18 °C treatment (P = 0·027, r2 = 0·84). For plants grown at 28 °C the correlation was strong but not significant and the slope close to that for the 18 °C plants. For plants grown at 23 °C there was no relationship between RGR and R/P. These relationships were maintained even when R/P was recalculated using both RSm at night and RSm in the light (assuming that light inhibited RSm by 50%) and RSm in the light was added to the measured rates of net photosynthesis (data not shown).

To further examine the changes in whole plant respiration rate we calculated the ratio of respiration rates exhibited by 18 °C-grown and measured plants divided by rates of Ra exhibited by warm-grown plants (i.e. 23 and 28 °C) measured at their respective growth temperatures (Fig. 6). Overall, the low/high temperature ratio was greater in the slow-growing species than in their fast-growing counterparts (P = 0·04, r2 = 0·42). This suggests that the slow-growing species exhibited a higher degree of respiratory acclimation to the low growth temperature than the fast-growing species (see Introduction for criteria for assessing acclimation).

Figure 6.

Ratio of whole-plant respiration per unit leaf area exhibited by low and high temperature grown plants (measured at their respective growth temperatures) plotted against maximum relative growth rate (RGR, mg g−1 d−1). Ratios are for 18 °C-grown and measured plants divided by rates exhibited by plants grown and measured at 23 °C (•), or divided by the rate of plants grown and measured at 28 °C (○). Each point is a single species. Solid line is linear regression for all data (P = 0·04, r2 = 0·42).

Figure 7 shows there was a significant (P = 0·04) negative relationship between daily whole plant Ra of Utrecht-grown plants plotted against NAR of the York-grown plants over all three growth temperatures. Variations in NAR were therefore driven in part by variations in Ra. There were clear differences among the growth temperatures with 18 °C-grown plants exhibiting the lowest and highest Ra and NAR values, respectively (Fig. 7).

Figure 7.

Net assimilation rate (NAR, g m−2 d−1) of York-grown plants plotted against whole plant respiration (Ra, µmol m−2 s−1) grown at 18 °C (○), 23 °C (▪) and 28 °C (▴). Each point is the mean of four plants (± SE). Linear regression fitted through all temperature treatments (Ra = 172·6 − 5·362 × NAR, P = 0·04, r2 = 0·2).

Rates of whole shoot photosynthesis on a mass basis (Pm), whole shoot dark respiration on a mass basis (RSm) and root respiration on a mass basis (RRm) for the six selected species are shown in Fig. 8. The Pm of the three slow-growing species (S. uniflora, A. aneura and P. costiniana) was not significantly affected by growth temperature (Fig. 8a). Silene dioica and A. melanoxylon exhibited a significant temperature response between 18 and 23 °C (P < 0·05). Only two species showed homeostasis of RSm (slow-growing A. aneura and P. costiniana), whereas the remaining species had significantly (P < 0·05) lower rates of RSm at 18 °C compared with the warmer temperatures (Fig. 8b). RRm was similar at all temperatures in S. uniflora; in contrast the five other species exhibited lower rates of RRm at 18 °C compared with those grown at 23 and 28 °C (Fig. 8c).

Figure 8.

Variations in CO2 exchange on a dry mass basis: (a) whole-shoot photosynthesis (Pm, nmol g−1 s−1); (b) whole-shoot dark respiration (RSm, nmol g−1 s−1); (c) intact root respiration (RRm, nmol g−1 s−1) for six selected species measured at three growth temperatures. The species are shown in order of decreasing RGR, with S. dioica and A. aneura being the fastest and slowest growing species, respectively. Each bar is the mean of four replicates (SE). Different letters indicate significant difference (P = 0·05).


We used a combination of growth analysis and gas exchange of a wide range of species grown at three different temperatures to examine the impacts of growth temperature on the underlying components of growth. Our results demonstrate that the partitioning of biomass to leaves was not altered by growth temperature (Fig. 1b). However, growth temperature had marked effects on SLA and NAR and their relative importance in determining the values of RGR.

The changing role of SLA and NAR

Over a wide range of RGR values SLA increased with increasing growth temperature in 14 of the 16 selected species (Fig. 1a; Table 2). The two exceptions were Luzula acutifolia where SLA was temperature insensitive, and Achillea ptarmica where SLA increased between 18 and 23 °C, and then decreased at 28 °C. A positive relationship between SLA and RGR was apparent when plants were grown at 23 and 28 °C (Fig. 1a). However, the correlation between SLA and RGR was not significant at 18 °C with very little difference in SLA values occurring amongst the contrasting species when grown at 18 °C (Fig. 1a). This finding suggests that the commonly held view that SLA is the most important variable in determining variations in RGR is not necessarily true of plants experiencing cool growth temperatures. Moreover, although it may be possible to predict the response of SLA at moderate growth temperatures along a range of RGR values, prediction of SLA at cooler temperatures may not be possible. In contrast, interspecific differences in variations in NAR were more important in determining interspecific differences in RGR in plants grown at 18 °C (Fig. 1c). When plants were grown at 23 and 28 °C, SLA and NAR combined to play an important role in determining variation in RGR.

Overall, NAR was higher in plants grown at 18 °C than at 23 or 28 °C for most RGR values (Fig. 1c). Given that NAR can be calculated using Eqn 2 there are three factors that could explain the higher overall NAR values at 18 °C relative to warm-grown plants: (1) higher photosynthetic rates per unit leaf area; (2) lower rates of whole plant respiration per unit leaf area; and/or (3) lower carbon concentrations. Figures 4a and 5 demonstrate that the overall higher NAR values exhibited by the 18 °C-grown plants were not due to higher rates of Pa or lower whole plant carbon concentration, respectively. The response of Ra to growth temperature therefore appears to be important in determining the overall response of NAR to growth temperature, with lower rates of Ra generally being associated with higher NAR values (Fig. 7). The lower rates of Ra in the 18 °C-grown plants could have resulted from either lower rates of respiration per unit plant mass and/or higher SLA × LMR values as:


As SLA was in fact lower in the 18 °C-grown plants (Fig. 1a; Table 2), and LMR was unaffected by growth temperature (Fig. 1b), the low Ra values must have been due to lower rates of respiration on a mass basis (Rm) (Fig. 8b & c) compared with plants grown at the warmer temperatures.

In addition to NAR being generally higher in the cold-grown plants for most RGR values, differences in NAR amongst the three growth temperatures were greater in the fast-growing species than in the slow-growing species (Fig. 2b). As a result, variations in NAR explained a greater proportion of the interspecific variations in RGR at 18 °C than at 23 or 28 °C (where variations in both NAR and SLA are both important in determining RGR values) (Fig. 1). Our data show that there was no correlation between whole shoot Pa and RGR (Fig. 4a), in agreement with previous studies that compared plants grown at a single temperature (e.g. Poorter et al. 1990; Atkin, Botman & Lambers 1996b). Moreover, our data demonstrate that the lack of correlation between Pa and RGR occurred at all three growth temperatures, with growth temperature not having a significant effect on Pa (Fig. 4a). The changing importance of NAR in determining variations in RGR at the three growth temperatures (Fig. 1c) was therefore probably not due to interspecific differences in the response of photosynthesis per unit area to growth temperature.

Another reason why NAR was singularly important in explaining interspecific variations in RGR at 18 °C could be that the species differ in the extent to which the carbon concentration of roots plus shoots was affected by growth temperature. For example, growth at the low temperature might result in the carbon per unit mass of the fast-growing species being relatively low, and/or that in the slow-growing species relatively high, compared with plants grown at the higher temperatures. However, the total plant carbon concentration (including structural and non-structural carbon) was actually higher in the fast-growing species grown at 18 °C relative to the 23 and 28 °C-grown (Fig. 5). The slow-growing species exhibited little change in carbon concentration with changing growth temperature. The changing role of NAR in determining variations in RGR was not due therefore to interspecific differences in plant carbon concentration.

Given that differences in Pa (Fig. 4a) or CC (Fig. 5) do not explain the greater importance of NAR in determining RGR at the lower temperatures (Fig. 1c), the selected species must have differed in the extent to which Ra was affected by growth temperature. Growth at the lower temperature must have suppressed carbon use in respiration to a greater extent in the fast-growing species than in the slow-growing ones. Conversely, Ra must have been more homeostatic amongst the growth temperatures (i.e. a higher degree of thermal acclimation) in the slow-growing species than in their fast-growing counterparts. Unfortunately, interplant variability and measurement error of measured rates of Ra preclude us stating whether or not the fast- and slow-growing species differed statistically as described above (Fig. 4b). Nevertheless, interspecific differences in the response of Pa (Fig. 4a) or carbon concentration (Fig. 5) were clearly not being responsible. Moreover, the ratio of Ra in cold-grown plants divided by that in warm-grown plants was greatest in the slow-growing species (Fig. 6) (see definition of acclimation in Introduction). We thus feel confident that interspecific differences in the degree of thermal acclimation of Ra were primarily responsible for the changing importance of NAR at the different growth temperatures.

The greater degree of acclimation of Ra to the lower growth temperature exhibited by the slow-growing species might result from a higher degree of acclimation of Rm and/or a greater decrease in SLA at the lower growth temperature (growth temperature did not systematically affect LMR values; see Eqn 3 above). Although all species exhibited a lower SLA at 18 °C compared with 23 and 28 °C (Fig. 1a), the absolute change in SLA was generally lower in the slow-growing species than in the fast-growing species (Fig. 2a). This suggests that interspecific differences in the effect of temperature on SLA alone cannot account for the greater degree of acclimation of Ra to lower temperatures in the slow growing species. Therefore, interspecific differences in acclimation of Rm (Fig. 8b & c) must play a key role in the importance of NAR in determining variations in RGR at lower growth temperatures.

Was the degree of acclimation of whole plant Rm greater in the slow-growing species? Acclimation can be quantified in various ways as discussed in the Introduction. On a dry mass basis two of the slow-growing species (P. costiniana and A. aneura) exhibited homeostasis of RSm between 18 and 28 °C (Fig. 8b). All the other species had lower rates of RSm at 18 °C than at either 23 or 28 °C. The only species that exhibited a temperature response that may indicate no acclimation (assuming a Q10 of 2·0), was the fast-growing acacia, A. melanoxylon (Fig. 8b). However, there did not appear to be any systematic difference in the degree of acclimation of root respiration amongst the species. In five out of the six species, RRm increased significantly with increasing growth temperature (Fig. 8c). The Q10 of root respiration has been shown to commonly range from 1·2 (Higgins & Spomer 1976) to 2·9 (Tjoelker et al. 1999b). Assuming this range of Q10 values the response seen in our data would suggest that P. costiniana and A. aneura show no acclimation of root respiration whereas the others may have partially acclimated. Silene uniflora exhibited homeostasis of root respiration. Taken together, it appears that the importance of NAR in determining variations in RGR at the lower growth temperature may have been due to a greater degree of thermal acclimation of Rm of the shoots, but not the roots, in the slow-growing species.

Balance between respiration and photosynthesis

It is often assumed that, because respiration is more temperature sensitive in the short term than photosynthesis, the R/P ratio will increase under growth conditions of elevated temperature (Woodwell 1983, 1990). However, such assumptions are rarely based on actual experimental data. Gifford (1995) found that when a diverse range of species was grown at constant temperatures ranging from 15 to 30 °C, R/P (using estimates of gross photosynthesis and assuming that leaf respiration occurs at the same rate in the light and dark) was constant. Similarly, a single leaf model published by Dewar, Medlyn & McMurtrie (1999) based on models by Thornley (1977) and McCree (1982) found that when daily temperature data were used in a 1 year simulation of the model, R/P (using modelled estimates of gross photosynthesis and leaf respiration) was constant at approximately 40%. In our study, there was no systematic difference in the R/P-values exhibited by plants grown at 18 and 23 °C (Fig. 4c). Moreover, no difference in R/P was found amongst S. uniflora plants grown at 18, 23 and 28 °C. However, four of the six species did exhibit higher R/P-values at 28 °C than at 18 and 23 °C (Fig. 4c). The percentage increase in R/P from 18 to 28 °C ranged from 36% for S. dioica to 66% for P. trivialis. The response of R/P to growth temperature therefore varies amongst species. Moreover, although R/P is relatively homeostatic at moderate growth temperatures, increases often occur when plants are grown at unfavourably high temperatures.

Plasticity of growth and its underlying parameters

In a study assessing the impact of elevated ambient CO2 concentration on the growth of a range of fast- and slow- growing Acacia species, Atkin et al. (1999) reported that the proportional difference in RGR between CO2 treatments was similar regardless of inherent RGR value. However, absolute differences in RGR between CO2 treatments increased with increasing RGR value (Atkin et al. 1999). This suggests that fast-growing species are more ‘plastic’ with respect to atmospheric CO2 concentration. In our study, there appeared to be larger differences (amongst the growth temperatures) in the proportion of maximum RGR achieved by the slow-growing species than in their fast-growing counterparts (Fig. 2). Consequently, absolute differences in RGR (amongst the growth temperatures) were generally similar in the slow- and fast-growing species (see Table 2). The slow- and fast-growing species did not therefore appear to differ in their plasticity of RGR response to growth temperature. In contrast, there were systematic interspecific differences in the SLA and NAR response to growth temperature; absolute differences in SLA and NAR (amongst the growth temperatures) were greater in the fast-growing species than their slow-growing counterparts (Fig. 2). Why then, were the growth temperature-mediated (absolute) differences in RGR similar in the fast- and slow-growing species? The answer lies in the fact that, for any RGR value, growth temperature affected SLA and NAR in opposite directions (i.e. high temperatures typically resulted in higher SLA and lower NAR values). Thus, the changes in NAR and SLA tended to cancel each other out, with the result that the absolute differences in RGR were not as great as predicted by the differences in NAR and/or SLA alone.

Thermal origin and RGR

The 16 species used in this study evolved in diverse habitats (Table 1). We had expected that the effect of growth temperature on RGR and its components would reflect the thermal origin of the individual species. However, the response of NAR and RGR to temperature was more closely associated with the inherent maximum RGR than with differences in thermal origin. This in turn suggests that slow-growing species from thermally contrasting environments (e.g. alpine, subalpine and arid environments) must exhibit similar proportional changes in RGR when exposed to different growth temperatures. For example, two of the slower growing species used in this study (A. aneura and P. costiniana) came from habitats exhibiting very different thermal regimes. The temperature range likely to be experienced by A. aneura is between a mean maximum of 36–40 °C and a mean minimum of 5–8 °C (Boland et al. 1985) whereas the temperature range for P. costiniana is a mean maximum of 9–12 °C and a mean minimum of 0–3 °C (Bureau of Meterology 2001). Yet the RGR of these two species responds similarly to different constant growth temperatures (Fig. 2; Table 2). One explanation for the common response of the slow-growing species is that the alpine species and species from arid habitats experience larger temperature extremes than faster-growing species from more moderate climates. Species native to habitats with large temperature variations during their growing season generally have a strong ability to adjust their photosynthesis, whereas species from thermally stable habitats show a poorer ability to acclimate photosynthetically (Berry & Björkman 1980; Björkman 1981; Öquist 1983). The temperature at which each species achieved their maximum RGR did not correlate with their thermal origin. Van Rijn et al. (2000) found that variation in growth characteristics among populations of Hordeum sponteneum was poorly related to environmental variables such as rainfall, humidity or temperature at the site of origin. Similarly, experiments on Agrostis curtisii populations from south Wales in the British Isles to Portugal showed that there were no significant correlations of plant responses to either CO2 or growth temperature with the climate of origin (Norton et al. 1999). It appears that thermal origin of a species is not a useful predictor of plant dry matter production in response to changing growth temperatures that may result from climate change.


We have shown that growth temperature can change the relative importance of the underlying components of RGR. Importantly, we have challenged the commonly accepted view that SLA is the most important factor driving Rgr; our study shows that this does not necessarily apply at lower growth temperatures as there was no significant relationship between SLA and RGR at 18 °C. NAR was more important in explaining variations in RGR at 18 °C than at 23 and 28 °C. The greater importance of NAR at the cooler growth temperatures was due to differing degrees of acclimation of respiration across the range of RGR values. Plants grown at 28 °C used a greater proportion of their daily fixed carbon in respiration than did the 23 and 18 °C-grown plants. The lack of correlation between changes in RGR and thermal environment of origin has implications for the prediction of the impact of rising global mean temperatures on plant growth. It is more likely that grouping of species as fast- and slow-growing in terms of their response to changing growth temperatures may produce more accurate predictions than generalizations based on thermal origin.


The authors thank David Sherlock for expert technical assistance and Dr John Evans and Dr Everard Edwards for useful comments on the manuscript. This work was funded by a NERC research grant to O.K.A./A.H.F. (GR3/11898).