Stomatal conductance in a tropical xerophilous shrubland at a lava substratum
- First Online:
- Cite this article as:
- Barradas, V.L., Ramos-Vázquez, A. & Orozco-Segovia, A. Int J Biometeorol (2004) 48: 119. doi:10.1007/s00484-003-0195-x
- 75 Views
Diurnal variation in leaf stomatal conductance (gs) of three xerophilous species (Buddleia cordata, Senecio praecox and Dodonaea viscosa) was measured over a 10-month period during the dry and wet seasons in a shrubland that is developing in a lava substratum in Mexico. Averaged stomatal conductances were 147 and 60.2 (B. cordata), 145 and 24.8 (D. viscosa) and 142.8 and 14.1 mmol m–2 s–1 (S. praecox) during the wet and dry season respectively. Leaf water potential (Ψ) varied in a range of –0.6 to –1.2 (S. praecox), –0.6 to –1.8 (B. cordata) and –0.9 to –3.4 MPa (D. viscosa) during the same measurement periods. Stomata were more sensitive to changes in irradiance, air temperature and leaf–air vapour pressure difference in the rainy season than the dry season. Although stomatal responses to Ψ were difficult to distinguish in any season (dry or rainy), data for the entire period of measurement showed a positive correlation, stomata tending to open as Ψ increased, but there is strong evidence of isohydric behaviour in S. praecox and B. cordata. A multiplicative model relating gs to environmental variables and to Ψ accounted for 79%–83% of the variation of gs in three sites (pooled data); however, the performance of the model was poorer (60%–76%) for individual species from other sites not included in the pooled data.
KeywordsBuddleia cordataDodonaea viscosaIsohydric behaviourLeaf water potentialSenecio praecox
One of the main features of some regions in the tropics is the restriction of precipitation to a part of the year where almost 90% occurs during the rainy season (de Ita-Martínez and Barradas 1986; Murphy and Lugo 1986; Rojo 1994). In Pedregal de San Angel (south of Mexico City), a tropical shrubland community is established on a lava substratum, where precipitation is highly seasonal and the soil is poorly developed and limited only to rock crevices. Although the climate in the region is subhumid, the low capacity for water retention by the lava substratum results in arid conditions for the plants growing in Pedregal. Because of these features the shrubland of Pedregal has been previously classified as xerophilous (Rzedowski 1954). Plants in Pedregal are mainly grouped into four life forms: perennial, deciduous, biannual and annual. Thus, the annual distribution of precipitation and the low water retention of the substratum can be relevant to the survival of perennial species, which may show physiological adaptations to tolerate water stress.
Stomatal function plays a key role in the water economy of plants (e.g. Raschke 1976; Mansfield and Wilson 1981; Jones 1992; Meinzer et al. 1993). Therefore, the study of diurnal and seasonal responses of stomata to environmental conditions is essential to an understanding of differences in the performance of plants under water stress and the elucidation of adaptative processes. However, the knowledge of stomatal conductances in tropical communities is very poor (Körner 1995). Furthermore, data in communities such as Pedregal de San Angel are scarce (Rojo 1994) and the knowledge of their ecophysiology is even poorer.
Stomatal response to environmental variables has largely been studied through empirical models (e.g. Jarvis 1976; Whitehead et al. 1981; Fanjul and Barradas 1985; Jones 1992) and it is not clear whether or not the parameters of those models (e.g. Dolman et al. 1991; Pitman 1996) are specific to each plant type (or a single canopy) at the time of the measurements or are more generally applicable.
The objective of the work described here was to characterize the stomatal behaviour of three dominant plant species (Buddleia cordata, Dodonaea viscosa and Senecio praecox) during the dry and wet seasons in a tropical shrubland established on a lava substratum in Mexico. Data were used to estimate the parameters of a multiplicative model to analyse and predict stomatal conductance from environmental data and leaf water potential and also to test the performance of the model.
Materials and methods
Measurements were made at the reserve Pedregal de San Angel of the Universidad Nacional Autonoma de Mexico (19°19′ N, 99°11′ W), a suburban natural area of 124 ha in the south of Mexico City. The mean annual rainfall of the site is 803 mm (mean of 21 years) and almost 93% occurs during the rainy season (May to October). Extreme temperatures are registered in April (29.5 °C) and January (–1.1 °C) (Sánchez-Huerta 1990). Winds are light and predominantly from the northeast (Barradas et al. 1999).
The vegetation in the Pedregal is dominated by xerophilous shrubs growing on a thick layer of lava (0.5–10 m) that was deposited by the eruption of the Xitle volcano approximately 2,500 years ago (Rzedowski 1954). This vegetation is diverse in its species composition, with an estimated 301 species grouped in 254 genera and 61 families (Valiente-Banuet and de Luna García 1990). Soils are shallow and poorly developed, slightly acid and confined to the rock crevices. Four flat areas were selected, each with plants of the three selected species that were less than 10 m apart, therefore it was possible to take measurements from the three species on the same day.
Measurements were made on mature leaves of three dominant species: Senecio praecox L. (Compositae), Buddleia cordata HBK (Loganiaceae) and Dodonaea viscosa L. Jacq. (Sapindaceae) 1.5–2.5 m high. S. praecox and B. cordata are hypostomatous and D. viscosa is amphistomatous.
S. praecox is a succulent deciduous shrub with glabrous and oblong leaves (2–18 cm long and 2–10 cm wide). This species produces some leaves after blooming (end of the dry season) that persist for a very short period. B. cordata is an evergreen small tree with oblong and pubescent leaves (5–24 cm long and 1.5–10.5 wide) which reduces its canopy leaf area during the dry season. D. viscosa is an evergreen shrub with oblong and glabrous leaves (5–12 cm long and 2–4 cm wide). Leaf area indices (n = 6) were 1.1 (similar in dry and wet seasons) for S. praecox, 2.54 (wet season) and 1.79 (dry season) for B. cordata, and 2.25 (similar in dry and wet seasons) for D. viscosa as estimated with a canopy analyser (LAI-2000, LI-COR Ltd., Lincoln, Nebraska, USA).
Measurement of stomatal conductance and leaf water potential
Stomatal conductance (gs) was measured in two individuals of each species at four sites on at least five fully expanded leaves per plant, with a steady-state diffusion porometer (LI-1600, LI-COR, Lincoln, Nebraska, USA). Initially measurements were made on both upper and lower leaf surfaces in D. viscosa. However, the conductance of the upper surface was less than 5% of that of the lower surface. Consequently only measurements of the lower surface were made afterwards. Measurements of leaf water potential (Ψ) were made with a pressure chamber (PMS, Corvallis, Oregon, USA) as described by Scholander et al. (1964, 1965) and Turner (1981) on at least two leaves per tree.
Photosynthetically active radiation (Q), air temperature (Ta) and relative humidity (RH) were determined next to each measured leaf with a quantum sensor (LI-190SB, LI-COR Ltd., Lincoln, Nebraska, USA), a fine wire thermocouple and a humicap sensor (Vaisala, Helsinki, Finland). Leaf temperature (TL) was also measured. Thermocouples were mounted in the porometer. The leaf–air vapour pressure difference (VPD) was calculated from Ta, TL and RH measurements.
In addition, air temperature and precipitation were measured at 2 m above the soil surface in a clear area of the Pedregal with a thermistor (SK2011, Skye Instruments, Llandrindod Wells, UK) and a tipping-bucket rain gauge (Met One) respectively. These instruments were connected to a datalogger (Skye Instruments, Llandrindod Wells, UK), and data were recorded every hour.
All measurements, physiological and climatological, were made in a 10-month period including the dry and rainy seasons of 1996 on 4 days every 2 weeks from 0700 hours to 1800 hours local time at intervals of 1 h. During the rainy season leaves were sometimes wet in the morning, thus measurements started when leaves were dry (around 1000 hours local time).
Stomatal conductance model
Data collected during dry and wet seasons in sites 1, 2 and 3 of the Pedregal were used to fit the boundary-line functions to analyse and model the response of gs to the driving variables considered in this study, using the multiplicative model (Eq. 1). After that, the data sets generated in site 4 (shrubs 1 and 2) were used to test the performance of the model.
Stomatal conductance was higher in the rainy season than in the dry season in the three species studied. However, among species the average gs values were similar during the wet season (x = 147.6, 145.4 and 142.8 mmol m–2 s–1 for B. cordata, D. viscosa and S. praecox respectively) (F(2, 450) = 2.6, P > 0.05) but different in the dry season (x = 60.2, 24.8 and 14.1 mmol m–2 s–1 respectively) (F = 152.6(2, 499), P < 0.0000).
Leaf water potential was higher during the wet season than in the dry season in the three species. The highest and lowest average values of Ψ were recorded in S. praecox (–0.6 to –1.23 MPa) and D. viscosa (–0.9 to –3.38 MPa) respectively, whereas B. cordata registered intermediate values (–0.6 to –1.8 MPa) in the dry season. The average leaf water potential was significantly different among the three species studied, both in the dry season (F = 510.7(2, 464), P = 0.0000) and in the rainy season (F = 642.2(2, 574), P = 0.0000).
Ψ, measured on a typical day of the dry season (6 May) varied over the range –0.70 to –0.98 MPa (S. praecox), –0.78 to –1.48 MPa (B. cordata) and –1.55 to –2.65 MPa (D. viscosa). During the wet season Ψ ranged from –0.63 to –0.93 MPa for S. praecox, from –0.73 to –1.25 MPa for B. cordata and from –1.35 to –2.20 MPa for D. viscosa. However, it is important to note that Ψ for B. cordata and S. praecox always remained constant after 1000 hours local time during the day in both the dry and wet seasons, but for D. viscosa it fluctuated (Fig. 3).
The boundary-line analysis
Values of parameter A, the asymptotic value of gs and parameter B, the sensitivity of gs to changes in irradiance (Eq. 2), obtained from measurements on Buddleia cordata, Dodonaea viscosa and Senecio praecox. r2 is the coefficient of determination and the standard error of the mean is shown in parentheses
A (mmol m–2 s–1)
B (mmol m–2 s–1)
A (mmol m–2 s–1)
B (mmol m–2 s–1)
Calculated values of the maximal gs, or A for B. cordata, were 51% and 56% higher than the values for S. praecox and D. viscosa respectively during the dry season. Nevertheless, there were no significant differences in gsMAX among species in the rainy season. D. viscosa showed a drastic reduction in gsMAX (76%) from the rainy season to the dry season, whereas this reduction was only 47% in B. cordata (Table 1).
Values and standard errors (in parentheses) of cardinal (Tmin and TMAX) and optimal (TO) temperatures for stomatal function derived from Eq. 3 for B. cordata, D. viscosa and S. praecox in the dry and rainy seasons
In both seasons stomatal conductance in the three species tended to decrease linearly as VPD increased. Stomatal sensitivity to VPD was similar in D. viscosa and S. praecox (–127.7 and –123.6 mmol m–2 s–1 kPa–1, r2 = 0.93, 0.97, in the wet season; –15.8 and –16.5 mmol m–2 s–1 kPa–1, r2 = 0.92, 0.96, in the dry season respectively). B. cordata was approximately 73% more sensitive to VPD (–26.8 mmol m–2 s–1 kPa–1, r2 = 0.98) than the other two species in the dry season, but about 24% less sensitive to VPD (–96.2 mmol m–2 s–1 kPa–1, r2 = 0.97) in the wet season.
Data collected during dry and wet seasons in sites 1, 2 and 3 of the Pedregal were used to fit the boundary-line functions to model the response of gs to the driving variables considered in this study, using the multiplicative model (Eq. 1). After that, the data sets generated in site 4 (shrubs 1 and 2) were used to test the performance of the model.
Multiplicative model comparison of pooled data of three sites and for the data set of two individuals of each species from site 4 not included in the model for B. cordata (1), D. viscosa (2) and S. praecox (3). RMS Root mean square error, r2 the coefficient of determination, n the sample size
Three sites pooled data
Previous studies of the effect of microenvironmental and physiological variables on stomatal conductance have shown that there is a considerable diversity of responses of gs to different microclimatic and physiological factors (Q, Ta, VPD and Ψ) (Fanjul and Barradas 1985; Roberts et al. 1990; Pitman 1996; Meinzer et al. 1997a; Comstock and Mencuccini 1998). Although there are no stomatal conductance data reported for plant communities growing on a lava substratum, the data collected in the Pedregal for B. cordata, D. viscosa and S. praecox showed that measured values of gsMAX were similar to those observed in Buddleia asiatica and Dubautia scabra in a cinder substratum at Hawaii Volcanoes National Park (Matson 1990) and slightly higher in both seasons than those reported by Fanjul and Barradas (1985) in Coccoloba liebmannii and Jacquinia pungens, two tropical deciduous species. Values of gsMAX during the wet season in the three species were slightly lower than those observed in humid tropical forests (Körner 1995; Pitman 1996), but much higher than those presented for Simmondsia chinenses and Encelia asperifolia in the desert of central Baja California (Nilsen et al. 1990). However, values of gsMAX for the three species studied in the dry season were very similar to those of species native to the desert of Central Baja California. Therefore data obtained for B. cordata, D. viscosa and S. praecox are in reasonable agreement with other results obtained in different plant communities.
The drastic change of stomatal conductance observed from the dry to the wet season indicates that the control of water loss becomes more efficient as water stress increases in the species studied. Furthermore, the relative stomatal closure showed by B. cordata and D. viscosa in the afternoon is a typical mechanism to avoid dehydration when evaporative demand is high (Fanjul and Barradas 1985). The decrease in stomatal conductance has been attributed to a decrease in Ψ (Beadle et al. 1978; Schulze and Hall 1982; Kramer 1988; Comstock and Mencuccini 1998), to increasing VPD and Ta (Schulze and Hall 1982; Schulze 1986; Maroco et al. 1997; Meinzer et al. 1997a) or to a combination of these factors. However, stomatal conductances of B. cordata and D. viscosa were not affected by Ψ during the dry or wet season. Nevertheless, the effect of Ψ on gs was significant when these variables were analysed for the whole period of measurements. This response was probably due to the low variability of Ψ in either the dry or wet season, but, when data for the two seasons were combined, the variation of Ψ was sufficient to explain the variation of gs.
During the wet and dry seasons stomatal conductance in the three species was highly correlated to VPD and Ta but not to Ψ. This suggests that the control of stomatal opening is mainly a function of the atmospheric evaporative demand rather than a response to the plant water status (Meinzer 2002). Therefore it appears, for these species, that VPD is a more important factor than Ψ in controlling stomatal conductance. In B. cordata stomatal conductance was less affected by VPD than in D. viscosa and S. praecox in the wet season, but the opposite occurred during the dry season. This apparently greater sensitivity to VPD may be a useful adaptation to low ambient humidities encountered in the dry season. Nonetheless, stomatal responses to VPD have been analysed by subjecting plants to a series of steps of 1–2 h duration with increasing VPD (Bunce 1985; Aphalo and Jarvis 1991; Franks et al. 1997), with an observed reduction of gs at each step, but Monteith (1995) showed that the effects of changing VPD accounted for changes in transpiration rate.
Daily courses of gs and Ψ observed during the wet and dry seasons on B. cordata and S. praecox (Fig. 3) suggest that stomatal conductance responded to transpiration rather than leaf water potential (Meinzer et al. 1997b). Therefore, stomatal control allowed Ψ to remain almost constant after mid morning by progressively closing stomata to balance the increase of evaporative demand. This mechanism also allowed Ψ during the dry season to be similar to that of the wet season, by a drastic stomatal closure which balanced the decrease of water availability (Tardieu and Simonneau 1998; Meinzer 2002). This isohydric behaviour has also been observed in Pisum sativum, the pea (Bates and Hall 1981), Lupinus albus, the lupin (Henson et al. 1989), Saccarum officinarum, sugarcane (Saliendra and Meinzer 1989), Zea mays maize (Tardieu et al. 1993) and Populus euramericana, the poplar (Tardieu and Simonneau 1998). Nevertheless, this stomatal behaviour is more difficult to distinguish in D. viscosa since leaf water potential remained fluctuating in both seasons (dry and wet) although stomatal conductance decreased significantly after late morning (1000 hours local time). Furthermore, gs/Ψ scatter diagrams (Fig. 5) show a wider range of leaf water potential for D. viscosa (2.5 MPa) than for S. praecox (0.7 MPa) and B. cordata (1.5 MPa) fitting a better regression of gs versus Ψ. This may imply that Ψ fluctuates as soil water availability decreases and acts as a signal to regulate stomatal conductance to avoid dehydration (Comstock and Mencuccini 1998; Hubbard et al. 2001) showing probably anisohydric behaviour like that of Prunus persica, the peach (Steinberg et al. 1989), Prunus dulcis the almond (Wartinger et al. 1990), Glycine max, the soybean (Allen et al. 1994), Helianthus annus, the sunflower (Tardieu et al. 1996) and barley Hordeum vulgarae (Borel et al. 1997).
However, the effect of Ψ on gs for tropical species has not been widely studied, especially not in lava plant communities. Evidently the root system is also important (Borchert 1994) for explaining the variation of water availability and Ψ, although little is known about it in tropical plant communities.
The values of the coefficient of determination (r2) for the boundary-lines are indicative of a good approach to stomatal function ranges. The fit of gs versus Q, Ta and VPD showed a higher sensitivity of gs to the driving variables during the wet season than during the dry season in the three species studied. The changes in stomatal sensitivity to Ta and VPD can be attributable to a wider range of stomatal function registered in the dry season as a possible response to the differences measured between extreme air temperatures in the dry and wet seasons. This differential response of gs to the microenvironment (Ta and VPD) between seasons could be due to an acclimation of stomata to the thermal variation of the environment.
Predicted gs values in the three species agreed closely with the results from the multiplicative model. In general, values of coefficients of determination of the model are indicative of a better agreement between observed values of gs and values calculated from the model (Table 3). The model coefficients also indicate that the model does not explain efficiently the stomatal conductance variation throughout the whole period of measurements. Although this model is not efficient for explaining stomatal variation for the species studied, the method of boundary-line functions is a strong tool for analysing separately the stomatal responses to different driving variables (Q, Ta, VPD, Ψ).
It has been demonstrated from studies in tropical forests and pastures that empirical approximations, such as the multiplicative model of stomatal conductance, are specific to the site (Wright et al. 1996). Results reported here corroborate the site specificity but also that this model is individual-specific (Table 3). The relatively poor fit of the multiplicative model to data from individuals, whatever the initial formulation of the model, shows that the model is site- and/or individual-specific. Therefore, care is needed when attempting to extrapolate the model parameters to other sites and individuals.
We thank Daniel Degollado-Zaldívar and Marithza G. Ramírez-Gerardo for technical assistance. We also thank two anonymous reviewers for helpful comments on the manuscript. The research was supported by a grant from Dirección General de Asuntos del Personal Académico, UNAM (IN-210995).