Chlorophyll fl uorescence varies more across seasons than leaf water potential in drought-prone plants

ABSTRAC T Among the effects of environmental change, the intensifi cation of drought events is noteworthy, and tropical vegetation is predicted to be highly vulnerable to it. However, it is not clear how tropical plants in drought-prone habitats will respond to this change. In a coastal sandy plain environment, we evaluated the response of six plant species to water defi cits across seasons, the relationship between their morphophysiological traits, and which traits would be the best descriptors of plants’ response to drought. Regardless of leaf succulence and phenology, responses between seasons were most strongly related to chlorophyll fl uorescence. In this study we have demonstrated that a better comprehension of how tropical species from drought-prone habitats cope with changes in water availability can be based on seasonal variation in leaf water potential and chlorophyll fl uorescence. Temporal variation in leaf water potential and chlorophyll fl uorescence was found useful for differentiating between groups of sandy soil species that are responsive or unresponsive to water availability. However, chlorophyll fl uorescence appeared to be a more sensitive descriptor of their seasonal and short-term responses.


INTRODUCTION
In the current scenario of large-scale environmental change, the search for functional traits which will allow us to predict the response of communities and species to changes in conditions and resources has been identifi ed as one of the 100 fundamental questions in ecology (Sutherland et al. 2013). Among the effects of environmental change, the intensification of drought events is noteworthy (McDowell et al. 2008(McDowell et al. , 2011, and several studies have identified plant physiological traits that may be descriptors for the responses of plants to water availability (Bartlett et al. 2012, Choat et al. 2012, Bhaskar and Ackerly 2006. In tropical environments seasonality is largely reflected in changes in water availability (Franco 2002), and temporal fluctuations in water availability are particularly important for plants growing in sandy environments, due to the low water retention capacity of sandy soils (Noy-Meir 1972). Although several studies of plant responses to drought 550 BRUNO H.P. ROSADO and EDUARDO A. DE MATTOS in tropical dry forests and savannas have been published (Bucci et al. 2005, Franco 2002, Choat et al. 2006, Méndez-Alonzo et al. 2012, it is still unknown how tropical plants in other tropical drought-prone habitats (e.g. coastal sandy plains) will respond to drought, and which are the best traits for predicting this response. Plants growing on sandy soils show higher water potential than those growing on clay soils, since sandy soils hold water at high water potential . One important process that facilitates highly conservative water use is homeostasis of leaf water potential (Ψ L ) at midday across seasons (Bucci et al. 2005, Maseda andFernández 2006). The ability to maintain constant Ψ L might be achieved by mechanisms such as greater stomatal control and hydraulic adjustments, which reduce transpiration (Maseda and Fernández 2006). Species with less stomatal control may exhibit greater decreases in Ψ L (Franks et al. 2007, Maseda andFernández 2006). An important aspect of species that are able to maintain high Ψ L is related to the avoidance of embolism (Sperry et al. 2002, Maseda andFernández 2006). However, species with lower Ψ L can maintain CO 2 gas exchange for longer periods during drought, under low soil water availability and/or high air vapor pressure defi cit (VPD). On the other hand, such species might be more susceptible to water loss (Maseda and Fernández 2006). Importantly, the greater the degree of stomatal closure (associated with high Ψ L ), the dissipation of excitation energy by photosynthetic carbon assimilation will be more reduced (Maseda and Fernández 2006). In addition, greater stomatal closure may result in an increase in carbon starvation due to a decrease in photosynthesis (McDowell et al. 2008, van der Molen et al. 2011. A further drawback of stomatal closure is that it may lead to light stress, i.e., photoinhibition caused by excess light that is not used in photosynthesis. Changes in the ratio of the yield of variable (F v ) and maximum (F m ) chlorophyll fluorescence in dark-acclimated leaves can be used to characterize the degree of photoinhibition in plants subjected to environmental stress (Baker 2008). A decrease in the F v /F m ratio indicates increased photoprotection, in which excess excitation energy is dissipated as heat, and to direct photodamage to photosystem II (PS II) centers (Krause 1988, Walker 1992. In addition, an increase in thermal dissipation in PS II is associated with a decrease in F m and minimum chlorophyll fluorescence (F 0 ), whereas damage to the reaction centers increases only F 0 (Krause 1988, Porcar-Castell et al. 2008). Rosado and de Mattos (2010) observed variations between wet and dry seasons in the pressurevolume relationship in species on the Brazilian coastal sandy plains, a species assemblage known as restingas. As expected for plants growing on dry sandy soils, they observed high values of Ψ L at the turgor loss point (Ψ TLP ), less negative osmotic pressure, and a low bulk modulus of elasticity in restinga species (Rosado and de Mattos 2010). However, most species have an average predawn F v /F m of less than 0.83 during the dry season, which indicates chronic photoinhibition. To date, how ever, no information is available about the physiologi cal responses of restinga plants to seasonal variation in water availability, or the possible relationship between such physiological responses and leaf phenology.
In this study we chose the two dominant species, which are able to maintain a high minimum Ψ L during the dry season (Rosado and de Mattos 2010), and four less dominant restinga species. These six species occur in a restinga in southeastern Brazil and exhibit a range of leaf succulence, from 147 to 1105 g m -2 , and leaf phenology characteristics, from successive to concentrated leaf production in wet seasons de Mattos 2007, 2010). Leaf succulence refl ects the amount of water stored in leaves and may act as a buffer for water loss (Lamont and Lamont 2000),  and variations in leaf phenology are commonly associated with the ability to deal with water defi cits (Reich andBorchert 1982, Borchert 1994). We thus addressed the following hypotheses: 1) the six species have different strategies to maintain high Ψ L across seasons based on distinct arrays of leaf phenology and leaf morpho-physiological traits; 2) seasonal responses of chlorophyll fl uorescence are dependent on seasonal variation in Ψ L and both Ψ L and chlorophyll fl uorescence are redundant in the description of species' seasonal responses to water availability. Although soil moisture availability is a function of input minus output of water and rooting depth, the number of rainless days (NRD) might be a useful surrogate for soil moisture availability, especially in sandy soils whose water retention capacity is low . Given the low water retention capacity of sandy soils and the occurrence of dry spells in both wet and dry seasons (Cavalin and de Mattos 2007), coastal sandy plants may be differentially predisposed to water shortage during rainless days with varying VPD during wet and dry seasons. Our third hypothesis is therefore as follows: Ψ L responds to higher VPD only when NRD is low, but responds to an increase in NRD, regardless of VPD.

STUDY AREA AND PLANT SPECIES
Our study was conducted in the Restinga of Jurubatiba National Park, (14 140 ha), located in the northern part of the state of Rio de Janeiro, Brazil (extending north-south between 22º00'S and 22º23'S and east-west between 41º15'W and 41º35'W). The maximum temperature (around 30 °C) occurs in January and the minimum (around 20 °C) in July (Henriques et al. 1986). The mean annual temperature is 22.6 ºC. The sandy soil is characterized by low water retention capacity, low nutrient content and low organic matter content. Average annual rainfall is 1100-1300 mm, with a monthly minimum of 41 mm in dry seasons and a monthly maximum of 189 mm in wet seasons (Araujo et al. 1998  We selected six shrub species based on variation in leaf succulence and phenology (Rosado and de Mattos 2007) in an area 300 m from the ocean: Clusia hilariana (average leaf succulence: 1105 g m -2 ) and Byrsonima sericea DC. (Malpighiaceae) BRUNO H.P. ROSADO and EDUARDO A. DE MATTOS (239 g m -2 ) are classified as successive leaf producers because they produce leaves throughout the year, while Protium icicariba (147 g m -2 ) , Ocotea notata (Nees) Mez (Lauraceae) (169 g m -2 ) , Eugenia umbelliflora Berg (Myrtaceae) (358 g m -2 ) , and Erythroxylum ovalifolium Peyr. (Erythroxylaceae) (194 g m -2 ) concentrate their leaf production during the wet season. Erythroxylum is the only species that shows brevi-deciduous behavior. For simplicity, we will refer to each species by its genus name. Clusia is an obligatory CAM plant (Franco et al. 1996) , whereas all other species are C 3 plants.

LEAF WATER POTENTIAL AND CHLOROPHYLL FLUORESCENCE
Leaf water potential was measured using a pressure chamber (Model 1000, PMS Instruments, USA) at predawn (Ψ pd ) and midday (Ψ md ) . The measurements were made using two completely exposed mature leaves from the north-facing side of the canopy of the same three individuals of each species throughout the study. For Erythroxylum measurements, samples were taken from terminal shoots because the leaves and petioles were too small to be used in the pressure chamber. When the difference in Ψ L between two leaves of each individual was greater than 0.5 MPa, we measured a third leaf to estimate the average. From April 2004 to August 2005, measurements of Ψ L were made during eight months chosen to represent the variation between the wet and the dry seasons. Twelve measurements were made at predawn and 18 at midday for each individual. To represent these data graphically we chose the days with the minimum values for each individual per month. The minimum values are good descriptors of species' responses to water deficits  because they indicate the maximum xylem tension that a given species is able to tolerate while being able to maintain physiological activity (Bhaskar and Ackerly 2006). At the same time , measurements of chlorophyll a fl uorescence were performed using a pulse-amplitude modulated fl uorometer with a fi ber optic (FMS-2, Hansatech, UK), on the same two leaves from each individual. The fi ber optic was kept at a constant distance and angle to the leaf surface using a leaf clip. The potential quantum yield of PSII, which is dependent on minimal (F 0 ) and maximal fl uorescence yield (F m ), was measured by applying a pulse of light strong enough to saturate all the PSII reaction centers after the leaf was dark-adapted for 30 min (de Mattos and Luttge 2001). The F v /F m ratio was determined just before dawn (F v /F m pd ) and at midday (F v /F m md ) by darkadapting leaves for 30 min. Despite the usual high variability between leaves and individuals, we found it useful to use the absolute values of F 0 and F m to indicate temporal variation in photoprotection or photodamage associated with stress (Porcar-Castell et al. 2008). The averages were calculated in the same way as for Ψ L . On each sampling occasion we measured relative humidity and air temperature at 30 min intervals, starting at predawn and ending at midday.

STATISTICAL ANALYSES
To test hypothesis 1, we evaluated the differences in Ψ pd , Ψ md and F v /F mpd , F v /F mmd , and F 0 and F m between seasons and among species using a twoway repeated-measures ANOVA (P < 0.05). Additionally, to evaluate changes in the response of species (ecophysiological performance) between seasons, we constructed a correlation matrix using the mean values of each variable for each species in the wet and dry season, respectively. The wettest period comprised the months between November 2004 and February 2005. We analyzed the correlation matrix with a Principal Component Analysis (PCA) using program MULTIV (Pillar 1997). We also used PCA to describe the association between Ψ L and the chlorophyll variables (hypothesis 2).
To test hypothesis 3, we ran a multiple line ar regression to determine the contribution of VPD and NRD to variation in Ψ md and Ψ pd . We chose the highest VPD on each day when Ψ L and F v /F m were measured at both predawn and midday, and NRD during the one, two, three or four weeks before the measurements were taken. We performed a separate multiple regression for each week because we used cumulative NRD. Two-way ANOVAs and multiple linear regressions were carried out in R 2.11.1 (http://www.R-project.org). of Ψ md throughout the year, which were greater than those of other species (Tables II and III) . Clusia showed a slightly lower Ψ md than Protium, but the difference was not signifi cant. Clusia and Erythroxylum showed signifi cantly lower values of Ψ pd in the dry season in comparison to the wet season, but Erythroxylum was the only species with a signifi cantly lower Ψ md in the dry season (Tables  II and III Fig. 2c). With the exception of Protium, all species showed a decline in F v /F mpd in the dry season (Tables II and  III). A decline in F v /F mmd in all species was also observed in September 2004. Clusia and Protium showed the highest F v /F mmd in the wet season (0.80 and 0.79, respectively), while Eugenia showed the lowest (0.66) ( Table II, Fig. 2d). The highest F v / F mmd values were observed for all species during July 2005. There were no signifi cant differences between seasons for any species (Tables II and III).

LEAF WATER POTENTIAL
The F 0pd values varied from 180 to 450, with the lowest values observed in August 2005 (Fig. 3). Byrsonima, Clusia and Eugenia showed signifi cant differences between seasons, with the highest values observed in September 2004 (Tables II and  III ). Eugenia showed the greatest variation in F 0pd throughout the year (Fig. 3a) . F mpd varied from 800 to 2000 (Fig. 3b). With F 0pd , Eugenia showed the greatest variation between seasons. Most species displayed a simultaneous marked decline in F mpd in September 2004 and August 2005. Only Eugenia showed signifi cant differences between seasons, with the lowest values of F mpd measured in the dry season (Table II).  TABLE II Midday leaf water potential (Ψ md ), predawn leaf water potential (Ψ pd ), potential quantum yield of photosystem II at midday (F v /F mmd ), potential quantum yield of photosystem II at predawn (F v /F mpd ), minimal fl uorescence yield (F 0pd ) and maximal fl uorescence yield (F mpd ) during dry and wet seasons. In each column, different letters represent signifi cant differences (two-way repeated-measures ANOVA, Tukey's test, P < 0.05).

Species
Season Multiple linear regressions indicated that only NRD showed a negative relationship with Ψ pd and F v /F mpd in Clusia, Erythroxylum and Eugenia. Both parameters were affected in Erythroxylum, while only to Ψ pd or F v /F mpd responded to NRD in Clusia and Eugenia, respectively (Table IV). For midday measurements, only Ocotea showed a signifi cant negative relationship between F v /F mmd and VPD ( Table V).
The first two axes of the PCA explained 41.89% and 26.90% of the variation observed in the six traits, respectively (Table VI). The fi rst axis was most strongly associated with F v /F mpd , F mpd and F 0pd . The second axis was most strongly associated with   For Erythroxylum, however, the response was more closely associated with the second axis, with the response from wet to dry season being negatively related to Ψ pd and Ψ md (Fig. 4).

TABLE IV
Coeffi cient of multiple regression between predawn leaf water potential (Ψ pd , MPa), F v /F mpd , daily maximum air vapor pressure defi cit (VPD, Pa KPa -1 ) and the number of rainless days (NRD). We selected the greatest VPD for each day on which measurements were taken and the number of rainless days during the one, two, three or four weeks before that. Asterisks and bold values indicate P < 0.05.

TABLE V Coeffi cient of multiple regression between temporal variation in midday leaf water potential (Ψ md, MPa), F v /F mmd , daily maximum air vapor pressure defi cit (VPD) and the number of rainless days (NRD).
We selected the greatest VPD on each day on which measurements were taken and the number of rainless days during the one, two, three or four weeks before that. Asterisks and bold values indicate P < 0.05.

DISCUSSION
Most of the six species did not differ across seasons in their ability to maintain Ψ md . Although the dominant species, Clusia and Protium had higher values of Ψ md throughout, the other species, except for Erythroxylum, maintained similar values throughout the year, which were consistently lower than the dominant species. These values were maintained despite differences in leaf morphology and phenology. This agrees with reports indicating that the maintenance of Ψ md occurs in a series of ways, in contrast with the maintenance of constant hydraulic conductance, which occurs through changes in architectural traits such as leaf area per sapwood area (Meinzer et al. 1999, Bucci et al. 2005, Maseda and Fernández 2006, Franks et al. 2007). In Protium, for instance, the ability to maintain a higher Ψ md throughout the year might be mediated by a substantial reduction in leaf number during the dry season (Rosado and de Mattos 2010). Regardless of the mechanisms involved, it is noteworthy that only the two dominant species exhibited higher and constant Ψ md , which suggests a higher safety margin for embolism . Our fi ndings support that in drought-prone habitats most species are able to maintain similar Ψ md between dry and wet seasons. However, only a small number of species are able to maintain a relatively high Ψ md , which may be an important trait related to the ability of these species to achieve dominance in sandy, drought-prone environments. F v /F mpd was lower in the restinga plants than the optimal value for most species, which indicates photoinhibition (Demmig-Adams and Adams 2006). However, this was less pronounced than Figure 4 -PCA ordination of six species in two seasons on the basis of six traits: predawn (Ψ pd ) and midday leaf water potential (Ψ md ), predawn (F v / F mpd ), midday chlorophyll fl uorescence (F v /F mmd ), predawn basal (F 0pd ) and maximum fl uorescence yield (F mpd ). Each species, in the wet and dry seasons (open and closed symbols, respectively) are represented as follows: Byrsonima (cross), Clusia (circles), Eugenia (hexagon), Erythroxylym (triangle), Protium (losangle) and Ocotea (square). Axis 1 and 2 represent 41.89% and 26.90% of total variance, respectively. in plants from other harsh environments, where F v /F m may reach values below 0.5 (Bellot et al. 2004, Zunzunegui et al. 2005, Ogaya et al. 2011. Nevertheless, there is a positive linear relationship between F v /F mpd and the leaf turnover rate for restinga species, which suggests that there are ecophysiological implications of even small decreases in F v /F mpd (Rosado and de Mattos 2010). The greatest decreases in F v /F mpd and F v / F mmd occurred after a rainless period of 30 days in September 2004. These decreases were probably caused by low soil water potential and high radiation load, as observed for woody species in semi-arid Mediterranean shrubland in Spain (Bellot et al. 2004).
The reversible decline in F v /F m reflects the protective down-regulation of PS II, which may prevent damage to the reaction centers when there is excess excitation energy (Walker 1992, Demmig-Adams and Adams 2006). However, a comparison of changes in F 0 and F m indicates whether species are exhibiting increased thermal dissipation of excitation energy or deactivation of the PSII reaction centers (Krause 1988, Porcar-Castell et al. 2008 Irrespective of leaf succulence and leaf phenology, with the exception of Erythroxylum, physiological changes of species across seasons were more strongly captured by chlorophyll fl uorescence variables than to Ψ L . The multivariate analysis showed that the responses of all species across seasons were positively associated with F 0pd , which suggests the occurrence of photodamage during the dry season. However, the differential displacement of each species between seasons in the multivariate space indicates different degrees of susceptibility to photoinhibition, with Eugenia showing the strongest changes between seasons. The only species for which changes in fl uorescence between seasons was also paralled by changes in Ψ md and Ψ pd , was Erythroxylum. For Erythroxylum, the negative relationship between Ψ md and Ψ pd was expected because of this species' higher wood density, which may decrease water transport to leaves (Rosado and de Mattos 2010). The less responsive behavior of Byrsonima, Clusia and Protium to NRD and VPD indicates that these species have the ability to decouple from the main factors that drive daily and seasonal variation in water availability. In fact, Byrsonima and Protium show a strong midday depression of stomatal conductance in both wet and dry seasons (Lignani 2008), and the CAM photosynthesis in Clusia (Franco et al. 1996) may explain the lack of a midday response to NRD and VPD. In Clusia, the negative relationship between Ψ pd and NRD might reflect nocturnal transpiration, typical in CAM species, which may lead to more negative Ψ pd as NRD increases. In addition, the unresponsiveness of Ψ md and Ψ pd to NRD and/or VPD in Byrsonima may be related to this species' higher leaf death rate BRUNO H.P. ROSADO and EDUARDO A. DE MATTOS and Ψ TLP (Rosado and de Mattos 2010). The less negative Ψ TLP associated with a higher rate of leaf shedding may be an escape strategy for Byrsonima, allowing it to avoid the possible adverse effects of NRD and/or VPD on Ψ md and/or Ψ pd (Rosado and de Mattos 2010).
On the other hand, Eugenia, Erythroxylum and Ocotea responded strongly to NRD and/or VPD (Table IV). The negative relationship between NRD and Ψ pd for Erythroxylum may indicate a lower capacity for xylem refi lling during prolonged rainless periods because of the species' high wood density (Donovan et al. 2001, Rosado andde Mattos 2010). Interestingly, predawn and midday F v /F m showed significant negative relationships with VPD and NRD, indicating that chlorophyll fl uorescence may be more descriptive of the response of a species to water availability than Ψ L .
Plants growing on sandy soils have a higher Ψ pd than those growing on clay soils . However, this is true only when the former have access to deeper soil water profi les and/or soil moisture is frequently replenished by rainfall . The Ψ pd of restinga species growing where mean annual rainfall was 1100-1300 mm was higher than for other species occurring in sandy environments where the mean annual rainfall was around 450 mm (-0.1 to -1.70 MPa compared with -1.0 to -5.0 MPa, respectively) (Zunzunegui et al. 2005. In comparison with other seasonal environments, the variation in Ψ pd observed in restinga plants, -0.1 to -0.43 MPa in the wet season and -0.19 to -1.70 MPa during the dry season, was not as marked. In Mediterranean plants, Ψ pd varies from -0.1 to -0.86 MPa during spring and from -1.45 to -2.28 MPa during summer (Mediavilla andEscudero 2004, David et al. 2007). In the Brazilian savannas, woody species show a significant decline in Ψ pd , from values close to 0 MPa in the wet season to around -0.5 to -1.0 MPa during the dry season (Bucci et al. 2005). It must be remembered, however, that it is not only soil properties, root access to the water table and/or rainfall patterns that affect Ψ pd . Differences in Ψ pd among species may also be related to processes such as hydraulic conductance and nocturnal transpiration (Donovan et al. 2001, Dawson et al. 2007. For Clusia, the CAM behavior of nocturnal stomatal opening, as also observed in C 3 plants that show nocturnal transpiration, could explain the low Ψ pd , (Donovan et al. 2001). On the other hand, in Erythroxylum, the only brevi-deciduous species in this study, a decrease in the rate of water delivery to the leaves because of the species' high wood density (~0.77 g cm -3 ; Rosado and de Mattos 2010) could promote more negative Ψ pd and Ψ md , as described in other studies (Donovan et al. 2001, Meinzer 2003. In conclusion, temporal variation in Ψ L and chlorophyll fluorescence was found useful for differentiating between groups of sandy soil species that are responsive or unresponsive to water availability. However, chlorophyll fluorescence appeared to be a more sensitive descriptor of their seasonal and short-term responses to NRD. The use of plant traits as proxies for the responses of communities to environmental change is gaining importance (Sutherland et al. 2013). In this study we have demonstrated that a better comprehension of how tropical species from drought-prone habitats cope with changes in water availability can be based on measurements of seasonal variation in Ψ L and chlorophyll fl uorescence. These fi ndings reinforce the importance of ecophysiological knowledge when selecting functional traits for predicting responses to climate change (Rosado et al. 2013), and open important avenues for models specifi cally concerning species from dry environments. In addition, because we focused on species from the Atlantic rain forest complex that are able to occupy coastal sandy plains environments, our results contribute to the understanding of the environ-mental fi ltering that affects community assembly in resource-poor habitats.

ACKNOWLEDGMENTS
We are very grateful to P. Cavalin and M.I. Braz for their invaluable help with fi eldwork. We thank the staff of NUPEM-UFRJ and the National Park for logistic support. We deeply thank both reviewers for good comments and suggestions that helped us improve the manuscript. The authors were supported by grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) and the Programa de Pesquisas Ecológicas de Longa Duração no Brasil (PELD-CNPq).