Genetic variation in Arabidopsis thaliana for night-time leaf conductance


M. A. Christman. Fax: 801 581 4668; e-mail:


Night-time leaf conductance (gnight) and transpiration may have several adaptive benefits related to plant water, nutrient and carbon relations. Little is known, however, about genetic variation in gnight and whether this variation correlates with other gas exchange traits related to water use and/or native habitat climate. We investigated gnight in 12 natural accessions and three near isogenic lines (NILs) of Arabidopsis thaliana. Genetic variation in gnight was found for the natural accessions, and gnight was negatively correlated with native habitat atmospheric vapour pressure deficit (VPDair), suggesting lower gnight may be favoured by natural selection in drier habitats. However, there were also significant genetic correlations of gnight with daytime gas exchange traits expected to affect plant fitness [i.e. daytime leaf conductance, photosynthesis and intrinsic water-use efficiency (WUEi)], indicating that selection on daytime gas exchange traits may result in indirect selection on gnight. The comparison of three NILs to their parental genotypes identified one quantitative trait locus (QTL) contributing to variation in gnight. Further characterization of genetic variation in gnight within and among populations and species, and of associations with other traits and native habitats will be needed to understand gnight as a putatively adaptive trait.


A diverse range of C3 and C4 plant species exhibit significant night-time stomatal opening and transpirational water loss (Musselman & Minnick 2000; Caird, Richards & Donovan 2007a; Dawson et al. 2007; Marks & Lechowicz 2007). Although this night-time water loss is not accompanied by carbon gain, it can be a substantial portion of a plant's total daily water loss, with reports ranging from 5–30% depending on species and ambient conditions (Benyon 1999; Snyder, Richards & Donovan 2003; Bucci et al. 2004, 2005; Daley & Phillips 2006; Caird et al. 2007a; Caird, Richards & Hsiao 2007b; Fisher et al. 2007; Kavanagh, Pangle & Schotzko 2007; Scholz et al. 2007). A number of hypotheses have been suggested for the potential adaptive value, or lack thereof, of night-time stomatal conductance (gnight) and transpiration (Enight) (Donovan, Linton & Richards 2001; Snyder et al. 2003, 2008; Barbour et al. 2005; Daley & Phillips 2006; Caird et al. 2007a; Dawson et al. 2007; Fisher et al. 2007; Marks & Lechowicz 2007; Scholz et al. 2007). From an evolutionary perspective, these hypotheses can be grouped into three general categories that are not mutually exclusive: (1) that night-time water loss is not a substantial cost, and thus, under weak selection or selectively neutral; (2) that substantial gnight and Enight are an indirect effect of genetic and mechanistic links to daytime gas exchange rates, which are the more important traits and under stronger selection; and (3) that substantial gnight and/or Enight confer a direct benefit on plant growth and fitness in some habitats by enhancing various aspects of plant nutrient, water and carbon relations. All of these categories make the assumption that there is a genetic basis for variation in gnight, but little is known about the genetic variation for gnight within species and thus the potential for direct or indirect selection on this trait.

There is growing evidence that there is considerable environmentally induced variation in gnight and Enight, and that regulation of these traits can be attributed to some of the same factors well known to regulate daytime conductance (gday) and transpiration (Eday) (Caird et al. 2007a). For example, decreasing water availability (drought and salinity) and higher atmospheric vapour pressure deficit (VPDair) are associated with decreasing gnight (Rawson & Clarke 1988; Donovan et al. 1999; Daley & Phillips 2006; Caird et al. 2007a,b; Cavender-Bares, Sack & Savage 2007; Dawson et al. 2007; Kavanagh et al. 2007). Whereas these studies have focused on correlations of environmental variation and gnight, comparisons of species and populations in ‘common garden’ experiments that minimize environmental variation provide a means of identifying underlying genetic differentiation for a trait (Arntz & Delph 2001). For example, patterns of population genetic differentiation for daytime gas exchange traits [e.g. gday, photosynthesis (A), water-use efficiency (WUE)] and their associations with native habitat support the hypothesis that genetic differentiations in these traits are adaptive (reviewed in Arntz & Delph 2001). For daytime gas exchange traits, the further use of phenotypic selection analyses has gone on to test adaptive hypotheses for gas exchange traits (e.g. Dudley 1996; Sherrard & Maherali 2006; Donovan et al. 2007).

For night-time gas exchange traits, however, only a few common garden studies have identified species differences (Jordan, Brodribb & Loney 2004; Howard & Donovan 2007; Marks & Lechowicz 2007) and intraspecific variation among cultivars of agricultural and horticultural plants (see Supplementary Table S1 in Caird et al. 2007a). In a common garden study with woody saplings of 21 species, Marks & Lechowicz (2007) found that high nocturnal sap flux was characteristic of species that are generally associated with lower shade tolerance, but not with differences in soil water availability of native habitats. Associations among population level genetic differences in gnight and habitat have not been tested.

If habitat associations are found with variation among populations for a putatively adaptive trait, then this suggests that the population differentiation may be the evolutionary outcome of direct selection on that trait (Arntz & Delph 2001). However, population differentiation may also be caused by other processes including genetic drift, historical artefact or indirect selection because of underlying genetic correlations. Phenotypic correlations (including both environmentally and genetically based variation) have been found between gnight and gday for desert plant species and Tasmanian conifers (Snyder et al. 2003; Jordan et al. 2004), and between nocturnal sap flux and leaf N and stem growth for saplings of 21 deciduous tree species (Marks & Lechowicz 2007). These correlations can be used to suggest how plant traits are functionally coordinated, and thus, may be providing a benefit or cost for plant growth and fitness. An understanding of the underlying genetic correlations can further suggest whether traits are independently affected by selection or jointly affected. For example, daytime gas exchange traits related to water use and WUE are thought to be under selection in water-limited habitats (Dudley 1996; Arntz & Delph 2001; McKay, Richards & Mitchell-Olds 2003; Donovan et al. 2007, but see Sherrard & Maherali 2006). If there is a strong genetic correlation between gnight and daytime gas exchange traits, then patterns of gnight may be greatly influenced by selection on the daytime rates, regardless of whether variation in gnight has any effect on plant fitness.

We investigated within-species genetic variation in maximum gnight using natural accessions of the model C3 annual plant Arabidopsis thaliana (L.) Heynh. Substantial gnight has been observed for a few accessions of Arabidopsis, and there is evidence for specific mutants having altered the magnitude of gnight (Lasceve, Leymarie & Vavasseur 1997; Leymarie, Lasceve & Vavasseur 1998, 1999; Dodd et al. 2005). The values of gnight are well above the functionally defined cuticular conductances generally reported [minimum gnight when stomata are forced closed with desiccation or abscisic acid (ABA)] (Caird et al. 2007a,b; Cavender-Bares et al. 2007). Additionally, Arabidopsis shows substantial within-species genetic variation in integrated WUE estimated from leaf δ13C, and quantitative trait loci (QTL) have been identified that contribute to variation in leaf δ13C (McKay et al. 2003; Hausmann et al. 2005; Masle, Gilmore & Farquhar 2005).

In this growth chamber experiment, we explored within-species genetic variation by comparing 12 naturally occurring accessions of Arabidopsis representing populations from a variety of habitats around the world (Redei 1970) that differ in water relations traits (McKay et al. 2003; Juenger et al. 2005). We grew the plants and measured night-time and daytime gas exchange under well-watered conditions that would maximize the potential for high gnight. We asked three questions: (1) Is there genetic variation in maximum gnight of Arabidopsis? (2) Is gnight correlated to climate characteristics of the region where the accessions were collected (precipitation and VPDair)? (3) Are there genetic correlations between gnight and daytime gas exchange rates and δ13C? Additionally, we used near isogenic lines (NILs) known to have divergent δ13C and/or gday (McKay et al. 2003; Juenger et al. 2005) to investigate whether gnight and gday are under similar genetic control.


Plant material

We studied 12 accessions of Arabidopsis thaliana selected to represent populations from locations across the globe with contrasting temperature and precipitation (Table 1; McKay et al. 2003). Seeds were obtained from the Arabidopsis Biological Resource Center (ABRC,, and seed for each accession was increased using single seed descent. Of these accessions, nine are spring annuals, shorter living and capable of flowering without vernalization, and three are winter annuals, longer living and requiring vernalization to flower. Using latitude and longitude data provided by the Nottingham Arabidopsis Stock Center, the following climate data were obtained for each accession from a long-term (40 years) data set (New, Hulme & Jones 1998): mean annual VPDair, mean monthly VPDair, mean annual precipitation and mean monthly precipitation.

Table 1.  The origin locations (latitude, longitude and elevation), life history characteristics and number of days to flowering [least square (LS) mean ± SE] of the 12 natural Arabidopsis thaliana accessions in the study
AccessionAbbreviationLocationLatitudeLongitudeElevation(m)Winter/spring annualDays to flowering (d)
  1. DNF, did not flower.

6673Col-0Columbia, USAN38.5W92.550Spring24.4 ± 2.7
902Cvi-0Cape Verde IslandsN16W241200Spring23.0 ± 1.8
1102Db-1Tenne, GermanyN50.5E8.5400–500Spring25.4 ± 2.7
1122Edi-0Edinburgh, UKN56E3100–200WinterDNF
922HodjaTajikistanN43E1 Spring25.4 ± 2.7
903Kas-1Kashmir, IndiaN35E771580Winter24.4 ± 2.7
20Ler-0Landsberg erecta; GermanyN51E12 Winter25.9 ± 2.7
1643Oy-1Oystese, NorwayN60E61–100Spring24.9 ± 3.7
1504Sei-0Seis am Schlern, ItalyN46.5E11.51000–1500Spring25.8 ± 1.8
1640Tsu-1Tsushima, JapanN33E1361–100Spring25.0 ± 2.5
1584Van-0Vancouver, CanadaN49.5W1231–100Spring26.9 ± 3.7
1638Ws-3Wassilewskija, RussiaN52E30100–200Spring21.3 ± 2.1

We included three NILs known to have divergent δ13C and/or gday as compared to the natural accessions making up the background for these NILs (McKay et al. 2003; Juenger et al. 2005). The NILs delta 2.1 and delta 3.1, described by Juenger et al. (2005), contain a small portion of Cvi-1's chromosomes 2 and 3, respectively, in a Ler-2 background. The NIL ColFRI has the genomic region containing a functional FRI introgressed from the Sf-2 accession into a Columbia background, which carries a null, recessive FRIGIDA allele (Lee & Amasino 1995; Johanson et al. 2000). FRI affects flowering time by its role in the vernalization response (Johanson et al. 2000), and introgression of a functional FRI has been found to increase both δ13C and flowering time (McKay et al. 2003).

The experimental design was a complete randomized block with three blocks and two replicates of each natural accession or NILs per block (90 plants total). Each replicate was an individual plant in a pot constructed from a 50 mL centrifuge tube with lid containing a small hole for plant growth. The bottom of the centrifuge tube was cut off, and the soil-filled tube was placed on top of a soil-filled Conetainer (Stuewe & Sons, Corvallis, OR, USA) to allow added volume for root growth (see McKay et al. 2001; Juenger et al. 2005). Each pot was initially planted with four seeds and thinned to one plant after germination. After planting, the pots were placed in a dark, cold room at 4 °C for 7 d and then transferred to a growth chamber with a 12 h photoperiod at the UC Davis Controlled Environment Facility. For growth and gas exchange measurements of the natural accessions and NILs, the daytime growth chamber conditions averaged approximately 28.1 °C and 34.9% relative humidity (RH), and night-time conditions averaged approximately 17.5 °C and 43.1% RH.

Gas exchange measurements

Gas exchange measurements (g, E, A) were obtained using a LI-6400 portable photosynthesis system (Li-Cor, Inc., Lincoln, NE, USA) with a custom-built cuvette that enclosed the whole canopy (rosette) of Arabidopsis (see McKay et al. 2001; Juenger et al. 2005). Preliminary measurements of several plants for gas exchange through 30 h intervals (logged every 10 s) indicated that Arabidopsis gnight remained high and steady through the night. Thus, daytime measurements were made during the interval 3–5 h after chamber lights came on and night-time measurements were made during the interval 2–4 h before chamber lights came on the following morning. During night-time measurements, a headlamp with a green safe light with intensity not detectable by an LI-190 sensor [0 µmol m−2 s−1 photosynthetic photon flux density (PPFD); Li-Cor] was used to avoid promoting stomatal opening. There is no evidence from previous trials that the very low (non-detectable) intensity of green light used for these measurements caused any change in stomatal aperture. The cuvette enclosed the whole rosette of each plant, and measurements were logged after stability was attained in the chamber. For daytime and night-time measurements, cuvette PPFD was 350 and 0 µmol m−2 s−1, respectively, and CO2 was maintained at 400 µmol mol−1 CO2. Temperature and RH in the gas exchange cuvette were set to approximate ambient growth chamber conditions (approximately 28 °C and 35% RH during daytime measurements, and 18 °C and 43% RH during night-time measurements).

Each block was measured on a different day for comparisons of gas exchange rates among natural accessions and NILs. Plants of some accessions did not germinate or died prior to gas exchange measurements, resulting in some blocks having less than two plants of each accession available for sampling each day. After the gas exchange measurements were completed for an individual plant, leaf areas were obtained from a digital photograph of the canopy, and image analysis was performed using Scion Image Software program (Scion Corporation, Frederick, MD, USA). Intrinsic WUE (WUEi) was calculated as A/gday.

Plant biomass and flowering time

Plants in two of the three blocks were harvested following gas exchange measurements for above-ground biomass determination, and leaf N and carbon isotope ratio (δ13C). Plants were dried at 60 °C and weighed for dry biomass. Leaf tissue was ground and analysed for leaf N and δ13C at the UC Davis Stable Isotope Facility ( Within a species where leaf size and morphology are similar, leaf δ13C reflects WUE integrated over the lifetime of the leaf (Farquhar, Ehleringer & Hubick 1989; Ehleringer, Phillips & Comstock 1992). A higher (less negative) leaf δ13C reflects greater WUE. Plants in the third block were allowed to continue growing so that the number of days to flowering could be calculated as the number of days from germination to bolting.


Natural accessions were compared for gas exchange characters (gnight, Enight, gday, A) with two-way analysis of variance (anova) (PROC GLM, SAS Institute, Inc., Cary, NC, USA) with accession and block as fixed effects. Ler-0, delta 2.1 and delta 3.1 were also compared for gas exchange characters (gnight, Enight, gday, A) with two-way anova. Col-0 and ColFRI were compared using a two-way anova. Data were log transformed and weighted by the inverse of the variance as necessary to meet assumptions of normality and homoscedasticity. Reported results are back-transformed least square (LS) means and SEs. For the 12 accessions, we calculated the correlations between characteristics of their native habitat (mean annual VPDair, mean monthly VPDair, mean annual precipitation, mean monthly precipitation) and plant traits (using accession means). Additionally, we calculated genetic correlations (using accession means) and phenotypic correlations (using individual plants) among the follow traits: gnight, gday, A, WUEi, δ13C, leaf N, pre-flowering above-ground biomass and number of days to flowering, except for phenotypic correlations where the need for destructive harvests resulted in traits collected on separate blocks (see Table 2).

Table 2.  Correlation coefficients (r) for genotypic correlations (above the diagonal, n = 12) and phenotypic correlations (below the diagonal, n = 47 for gas exchange traits, 28–38 for above-ground biomass, leaf N and δ13C and 13–14 for flowering time) for natural accessions of Arabidopsis thaliana
 gnightgdayAWUEiδ13CLeaf NBiomass (above-ground)Flowering time
  1. Asterisks indicate significance, *P < 0.05, **P < 0.01, ***P < 0.001.

  2. N/A, traits only collected on non-overlapping blocks.

Leaf N−0.28−0.54**−0.66***−0.15−0.37*−0.75**0.05
Biomass (above-ground)0.330.46*0.72***0.43*0.59***−0.63***0.13
Flowering time0.75**0.66*0.28−0.64*N/AN/AN/A


Genetic variation in gnight

Each of the Arabidopsis natural accessions exhibited substantial gnight. The 12 accessions differed significantly for gnight (Fig. 1), with gnight varying 2.5-fold. Under our experimental conditions, with daytime and night-time VPDleaf averaging 1.6 and 0.6 kPa, respectively, Enight was 23–33% of daytime rates among accessions, representing a substantial fraction of total daily water loss occurring during dark, non-photosynthetic times. The accessions also differed significantly for gday (P < 0.001), A (P < 0.001), WUEi (P < 0.001), δ13C (P < 0.001), leaf N (P = 0.01), above-ground biomass (P < 0.001) and flowering time (P < 0.001).

Figure 1.

Night-time total leaf conductance (gnight) in 12 natural accessions of Arabidopsis thaliana. Different letters indicate significantly different gnight values among the accessions (P < 0.05). Conductance data are back-transformed least square (LS) means and SE (n = 2–6 per accession).

Correlations of gnight with characteristics of accession native habitats

Among accessions, gnight was negatively correlated with the native habitat mean annual VPDair (Fig. 2). When the relationship between gnight and monthly mean VPDair was examined, the maximum r values occurred during the spring (February–March) and fall (September–November) months. There was no correlation between gnight and native habitat mean annual or mean monthly precipitation (r = 0.11, P = 0.76 and r < 0.26, P > 0.44 for all months, respectively). For daytime traits, the relationships of gday and A to mean annual VPDair trended in the same negative direction, but were not significant (Fig. 2). Mean annual VPDair was positively correlated with δ13C, but not with WUEi (Fig. 2).

Figure 2.

Correlations between native habitat mean annual atmospheric vapour pressure difference (VPDair) and accession traits in the common garden experiments: night-time leaf conductance (gnight), daytime leaf conductance (gday), photosynthesis (A), carbon isotope discrimination (δ13C) and intrinsic water-use efficiency (WUEi) for the Arabidopsis thaliana accessions. The ‘Cvi’ accession is excluded here because of unavailability of VPDair data. Data are back-transformed least square (LS) means.

Genetic correlations of gnight with other traits

We found highly significant genetic correlations among the gas exchange traits (Table 2); gnight was positively correlated with gday and A, and negatively correlated with WUEi (Fig. 3); gday was positively correlated with A and negatively correlated with WUEi (Table 2). There was no genetic correlation of gnight or gday with leaf N, above-ground biomass or flowering time. Phenotypic correlations generally paralleled the genotypic correlations in direction and significance for the gas exchange traits, but not for leaf N, above-ground biomass and flowering time (Table 2).

Figure 3.

Scatterplots representing genetic correlations for night-time leaf conductance (gnight) with daytime leaf conductance (gday), photosynthesis (A), integrative water use efficiency (δ13C) and intrinsic water-use efficiency (WUEi; A/gday) among natural accessions of Arabidopsis thaliana. Data are back-transformed least square (LS) means.

NIL comparisons to natural accessions

NILs delta 2.1 and delta 3.1 did not differ in gnight when compared to the Ler-0 natural accession from which they were derived (P = 0.18; Fig. 4), but they did have a higher gday than Ler-0 (P = 0.01). In contrast, ColFRI plants had a lower gnight but a similar gday when compared to the Col-0 natural accession from which it was derived (P = 0.05 and P = 0.74, respectively).

Figure 4.

Total leaf conductance (g) during the day (open bars) and night (shaded bars) in Arabidopsis thaliana accession Ler-0 and two near isogenic lines (NILs) derived from a cross between Ler-0 and Cvi (left) and accession Col-0 and ColFRI NIL (right). Plants were grown in a common garden growth chamber under identical conditions. Note the difference in scales between panels. Capital letters denote significant differences in gday, and lower case letters denote significant differences in gnight. Data are back-transformed least square (LS) means and SE (n = 2–6).


The stomata of Arabidopsis did close partially during the dark, but gnight remained substantially greater than cuticular conductance throughout the night, similar to previous reports for the species (Leymarie et al. 1998, 1999; Dodd, Parkinson & Webb 2004; Dodd et al. 2005). We found within-species genetic variation for gnight among the 12 accessions, as well as for the other traits: gday, A, WUEi, δ13C, leaf N, above-ground biomass and flowering time. The variation in gnight is consistent with that previously reported for δ13C and flowering time for a larger set of natural accessions of Arabidopsis (McKay et al. 2003), and consistent with genetic variation for daytime gas exchange characters for other plant species (reviewed in Arntz & Delph 2001). Within-species genetic variation for gnight suggests there is potential for selection to operate on this trait if it is directly or indirectly related to plant growth and fitness. However, variation alone does not necessarily mean that selection for or against the trait has occurred.

We assessed relationships between climate characteristics of the native habitats and the accession plant traits in a common garden setting to determine if the accession differentiation was consistent with a hypothesized evolutionary outcome of selection on these traits (Arntz & Delph 2001). Because gnight can lead to water loss at night with no concomitant carbon gain, the potential for a high gnight might be disadvantageous in water-limited habitats, but under weak selection or selectively neutral in wetter habitats. VPDair can be a proxy for habitat water availability because it is the atmospheric end of the leaf-to-air vapour pressure gradient driving plant water loss: habitats with higher temperatures and lower precipitation have a higher VPDair. Comstock & Ehleringer (1992) found that more of the among-population genetic variation in WUE for an arid shrub (Hymenoclea salsola) was explained by models including the source habitat VPDair as compared to just including precipitation or temperature. The accession differences in gnight among Arabidopsis under well-watered conditions, and the negative correlation between mean annual VPDair and gnight suggest that an inherently lower gnight and associated night-time water loss were favoured by selection in drier habitats. However, there was also a significant correlation between mean annual VPDair and δ13C, and non-significant trends between mean annual VPDair and gday, A and WUEi, raising the issue of which of these traits is actually under direct selection. This common garden study cannot determine whether the observed among-accession variation in gnight is likely the result of direct selection on gnight or the result of indirect selection mediated through gday, A, WUE or other unmeasured traits, but assessing genetic correlations can provide some insights into the potential for indirect selection.

There were significant genetic correlations between gnight and daytime gas exchange traits, such that the accessions with the highest gnight also had the greatest gday and A, and the lowest WUEi. The genetic correlations could reflect pleiotropy, close linkage of loci individually affecting each trait and/or linkage disequilibrium among loci that are not physically linked, presumably by correlated selection on both traits. Given the expectation for selection in water-limited habitats to act on WUEi and its components of gday and A, it is then reasonable to suggest that the accession differentiation in gnight may be caused by direct selection on WUEi, gday or A, and associated indirect selection on gnight. This could lead to accession differentiation in gnight that was correlated with native habitat aridity even if gnight was selectively neutral or slightly disadvantageous. Studies of these plants with phenotypic selection analysis would be necessary to actually test whether there is direct or indirect selection on gnight.

For the natural accessions, higher gnight was associated with lower WUEi. Although there was also a negative trend for gnight with integrative WUE (δ13C), the genetic correlation was not significant. The relationship between gnight and WUE is logical from the point of view that there are two mechanisms – carbon gain and water loss – that affect WUE and increased water loss as by Enight would increase water loss without carbon gain, thereby lowering whole-plant WUE. However, it is worth considering that the often-used measures of WUEi and integrated WUE (δ13C) are leaf-level surrogates for whole-plant WUE that are related to daytime carbon gain and water loss, and thus, do not account for water loss by Enight. Positive correlations between gnight and both A and gday may be driving the relationship with WUEi because it is calculated as the ratio of A to gday. Similarly for δ13C, gnight being highly correlated to gday may be the indirect cause for the trend. Regardless, the implications for the lack of significant relationship between gnight and δ13C are significant in that night-time water loss is typically ignored or considered negligible in C3 and C4 plants. This may be particularly important in crop species, some of which are known to have substantial gnight (Rawson & Clarke 1988; Musselman & Minnick 2000; Caird et al. 2007b) and where δ13C plays a large role in selection criteria for new cultivars better suited to habitats with contrasting water availability.

The delta 2.1 and delta 3.1 NILs had substantially lower WUE (δ13C) and higher gday, but similar gnight, compared with plants of the genetic background genotype Ler-0. ColFRI plants had opposite phenotypes with lower gnight in ColFRI plants than in Col-0 plants, but no difference in gday. These differences are the first we know of that show a differential effect for gday and gnight because of specific genomic regions. This result also demonstrates a new physiological function of FRIGIDA, suggesting network models of flowering time may be missing a phenotypic dimension. Even with this start of knowledge of specific genetic effects on gnight, the next challenge is to elucidate the exact mechanism(s) underlying variation, because many genes function to affect gas exchange and WUE via multiple mechanisms. For example, the Arabidopsis ERECTA mutation in Ler has been shown to have a large effect on δ13C via mechanisms affecting both transpiration, including stomatal density and conductance, and photosynthesis, including mesophyll cell distribution and reduced Amax (Masle et al. 2005). Thus, for the natural variation observed in this study, it is likely that multiple genes acting through multiple pathways ultimately affect gnight.

Although there is a highly significant genetic correlation between gnight and gday among the natural accessions, it is not altogether surprising that the two processes are not under identical physiological or genetic control. There are several effectors of stomata which should affect gday and gnight separately (i.e. factors affecting photosynthetic processes would affect gday but not gnight), and thus it is expected that some genomic regions will affect one process but not the other. For the delta 2.1 and delta 3.1 NILs in this study, it is probable that the genomic region that differs from the parental genotype is responsible for factor(s) which affect(s) gday but not gnight; the opposite would be true for FRIGIDA. However, the tight correlation between gnight and gday may also be caused by factors which affect stomata in similar ways, such as morphological or anatomical differences among accessions that affect both gnight and gday in the same manner (e.g. stomatal size and density). Differences in stomatal size and density could at least partially explain the observed variation in gnight among these Arabidopsis accessions as well as the tight gnightgday correlation. Furthermore, there is evidence that the magnitude of gday affects the magnitude of gnight through effects on A and carbohydrate metabolism (Easlon, personal communication), which could be a physiological mechanism behind the observed correlation. Thus, gnight is likely affected by several genetic and physiological factors which may influence its magnitude separately or in tandem with gday.

The emphasis on gnight and water availability in this common garden Arabidopsis study can be contrasted with the results from the common garden study with saplings of 21 tree species of Marks & Lechowicz (2007). Although that study used a different method (sap flux) to assess the integrated effects of gnight and Enight that were measured as whole-canopy instantaneous gas exchange in this study, the methods at different scales have generally been shown to be in good agreement (Caird et al. 2007a,b; Marks & Lechowicz 2007). For the tree saplings, high nocturnal sap flux was not associated with species differences in soil water availability of native habitats, but was associated with fast growing shade-intolerant species. The strong cross-species correlation between nocturnal sap flux and leaf N suggests that high nocturnal sap flux may sustain carbohydrate export and other processes driven by dark respiration needed to support fast growth (Marks & Lechowicz 2007). In Arabidopsis, there were no genetic or phenotypic correlations of gnight with leaf N or pre-flowering above-ground biomass which would suggest a similar mechanistic or fitness advantage. The differences between the two studies are perhaps not surprising given the differences in life history and sampling scales for the annual Arabidopsis and a set of deciduous woody species.

In summary, our finding of an association between Arabidopsis variation in gnight and aridity of the native habitats suggests that selection may have shaped this within-species variation, but the strong genetic correlations with daytime gas exchange traits raise the possibility that selection on gnight could have been indirect. Further characterization of genetic variation in gnight within and among populations and species, and of associations with other plant traits, plant fitness and native habitats will be needed to understand gnight as a putatively adaptive trait.


This research was supported by an NSF graduate research fellowship (M.A.C.), and NSF grants IBN-0416581 (J.H.R.), DEB-0419969 (J.H.R. and J.K.M.), DEB-0419542 (E.L.S.), DEB-0420111 (T.E.J.) and IBN-041627 (L.A.D.).