FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2900377459> ?p ?o ?g. }

• W2900377459 endingPage "1092" @default.
• W2900377459 startingPage "1079" @default.
• W2900377459 abstract "Síntesis: Nuestros resultados aumentan nuestra comprensión del espectro económico rápido–lento en plantas debido a que, tanto los cambios ambientales a lo largo de gradientes, así como procesos locales pueden simultáneamente promover diferentes estrategias de adquisición de recursos por debajo y por encima del suelo en ambientes extremadamente pobres. Palabras clave: Gradiente de aridez, Desierto de Atacama, estrategias ecológicas, rasgos funcionales, hoja, raíz, comunidades de arbustos, tallo, variación funcional de rasgos. The fast–slow plant economics spectrum (fast–slow PES hereafter) is a whole-plant ecological strategy framework that postulates that along gradients of resource availability such as water, carbon, and nutrients, the trade-off between acquisition and conservation of resources translates into different successful ecological strategies for plant communities at different points along a gradient (de la Riva, Tosto, et al., 2016; Pérez-Ramos et al., 2012; Reich, 2014). Accordingly, along gradients of water availability, plants with traits that allow them to acquire and use water quickly (i.e., a fast strategy) are only successful in environments with high-water input; in contrast, having slow water-acquisition traits (i.e., a slow strategy) is advantageous in low-water input environments because it promotes water conservation, which in turn enhances plant survival (Carvajal, Loayza, Rios, Gianoli, & Squeo, 2017; Chapin, 1991; Reich, 2014). However, some studies have suggested that when the environment becomes extremely arid and highly seasonal, a slow strategy (for leaf traits at least) may not be beneficial because maintaining slow traits for long time periods involves a greater energetic expenditure than maintaining fast traits, ultimately decreasing plant survival (Mooney & Dunn, 1970; Nilsen & Muller, 1981). Consequently, having traits for fast acquisition and use of water in these environments may allow plants to take advantage of the periods of high water availability (Ackerly, 2004; Mooney, 1982; Nilsen & Muller, 1981). A prerequisite for a fast–slow PES to occur is that traits of the different plant organs (e.g., leaf, stems, and roots) must be coordinated because of evolutionary and biophysical constraints (Freschet, Cornelissen, Logtestijn, & Aerts, 2010; Reich, 2014; Reich et al., 1999). That is, to acquire resources efficiently, being fast or slow at the leaf level, also involves being fast or slow at the stem and root levels; otherwise, a plant may not be able to establish in a given locality (Reich, 2014). While some studies bear out this expectation (e.g., de la Riva, Tosto, et al., 2016; Freschet et al., 2010; Liu et al., 2010; Méndez-Alonzo, Paz, Zuluaga, Rosell, & Olson, 2012), others do not (e.g., Baraloto et al., 2010; Butterfield, Bradford, Munson, & Gremer, 2017; Fortunel, Fine, & Baraloto, 2012; Kramer-Walter et al., 2016). For example, studying Amazonian trees, Baraloto et al. (2010) found orthogonal leaf and stem traits; that is, leaf and stem traits represented independent strategies of resource acquisition. The level of coordination between root and leaf traits can be multidimensional, meaning that certain root traits are coordinated with leaf traits, while others vary independently in determining resource acquisition strategies (Butterfield et al., 2017; Fort, Jouany, & Cruz, 2013; Kramer-Walter et al., 2016). Weak or lack of coordination among traits can occur because different plant organs respond to different selection pressures, respond to the same selection pressures but in different ways, or because of different constraints on different plant organs (Bergmann, Ryo, Prati, Hempel, & Rillig, 2017; Kembel & Cahill, 2011; Weemstra et al., 2016; Withington, Reich, Oleksyn, & Eissenstat, 2006). All of these alternatives result in specific trait combinations that allow plants to enhance their fitness in a given locality (Messier, Lechowicz, McGill, Violle, & Enquist, 2017; Valverde-Barrantes & Blackwood, 2016). In hyper-arid environments, however, plant traits are expected to be highly coordinated because the strong selection pressure imposed by aridity will restrict the suits of trait attributes that allow plants to cope with local environmental conditions (Chapin, 1991; Reich, 2014). Different plant species can have different ecological strategies within a single point along a resource gradient (i.e., within particular communities) because factors such as plant–plant interactions, microenvironmental variability, or perturbation, among others, can promote trait variation (Ackerly & Cornwell, 2007; Adler, Fajardo, Kleinhesselink, & Kraft, 2013; Butterfield & Briggs, 2011; Butterfield et al., 2017; Grime, 2006; Maestre, Callaway, Valladares, & Lortie, 2009; Morales, Squeo, Tracol, Armas, & Gutierrez, 2015; Stubbs & Wilson, 2004). For example, Ackerly and Cornwell (2007) found that in Mediterranean woody plant communities of Coastal California, the effect of perturbation, as well as below- and above-ground partitioning of resources on trait variation among coexisting species, promoted different ecological strategies. From a geographical perspective, low diversity of ecological strategies along a resource gradient is expected to occur in sites where there is low input of water because as the harshness of the environment increases, above-ground competition among coexisting species decreases (e.g., Coyle et al., 2014; Grime et al., 1997; Swenson & Enquist, 2007; but see Butterfield et al., 2017). Therefore, a shift from high to low trait variation along aridity gradients should occur as the environment becomes drier. Woody shrub communities of the Coastal Atacama Desert constitute an ideal system to test the fast–slow economics spectrum hypothesis because water is considered the main resource limiting plant abundance and distribution (Squeo et al., 1998). This desert, which constitutes a hyper-arid environment according to De Martonne's aridity index (De Martonne, 1926), exhibits a north to south (between 23°S and 30°S) increase in rainfall from <1–82 mm/year (Rundel et al., 1991; Squeo et al., 2006); therefore, within this hyper-arid desert, there is a marked north to south aridity gradient biologically relevant at both the community and intraspecific levels (e.g., Carvajal et al., 2017; López, Squeo, Armas, Kelt, & Gutiérrez, 2016; Squeo, Ehleringer, Olivares, & Arancio, 1994; Sotomayor & Gutiérrez, 2015). In this study, we used leaf, stem, and root traits that have functional significance for woody shrub species, to test the hypothesis that as aridity increases in a hyper-arid environment, selection pressures act simultaneously on trait attributes, variation, and coordination to promote an inverse pattern in the fast–slow PES. Specifically, we predict that in the most arid end of a hyper-arid aridity gradient, (a) shrub assemblages will exhibit trait values that reflect a water acquisition strategy at the whole plant level; (b) traits will be highly coordinated across different plant organs; and (c) low variation in ecological strategies will predominate among shrub assemblages. We conducted this study during 2015 in four sites (listed from north to south) along the Coastal Atacama Desert: Quebrada El León (QL), Norte de Llanos de Challe (LLCHA), Chañaral de Aceituno (CHA), and Romeral (ROM) (Supporting Information Figure S1). Mean annual precipitation at these sites ranges between 14 and 80 mm, mean annual temperature is relatively constant along the gradient and fluctuates between 15 and 17°C (Table 1). Most of this rainfall falls in very few pulses interspersed with long periods of drought, which can range from 10 months in the wettest sites (Carvajal, Loayza, López, Toro, & Squeo, 2014), to many years in the driest sites (Dirección General de aguas, Chile—http://snia.dga.cl/BNAConsultas/reportes). The area is also affected by El Niño Southern Oscillation cycle (ENSO), which promotes high inter-annual variability of rainfall (Houston, 2006) that increases towards the north (Carvajal et al., 2017; Carvajal, Loayza, & Squeo, 2015). For each site, we calculated De Martonne's aridity index (De Martonne, 1926) (), where and represent mean annual precipitation and temperature, respectively. A decrease in values of DEMAI is indicative of an increase in aridity and values below 5 indicate hyper-aridity. Accordingly, all sites studied are in a hyper-arid region (Table 1). To avoid variation caused by factors other than the one produced by aridity, all sites were located on sandy soils (stabilized dunes) of west-facing slopes of less than 5%. At each of the four sites, we installed twenty 50 × 2 m plots separated by at least 100 m (N = 20/site, except QL that had 22 plots), which we used to identify the dominant shrub species. We recorded all species present at each plot, as well as the number of individuals per species. Shrub richness decreased from 19 at Romeral (least arid site) to 10 at Quebrada el León (most arid site; Supporting Information Table S1). Species diversity followed the same pattern, being highest at the least dry site and gradually decreasing as aridity increased (Supporting Information Table S1). To quantify the functional structure (i.e., trait composition at the community level) of the shrub assemblages, we selected those species that collectively represented approximately 90% of the total abundance at each site (see Supporting Information Table S1 for the list of species selected and their relative abundance per site). The number of individuals of the selected species was used to estimate relative species abundance at each plot. We measured a set of leaf, stem, as well as fine and coarse root traits (see Supporting Information Table S2 for a description of the functional significance of traits) that have been previously linked with plant ecological strategies following standardized protocols (Pérez-Harguindeguy et al., 2013). At each site, we haphazardly selected 50 individuals to measure above-ground traits, except for one species at QL whose abundance was very low (10 sampled individuals). For stem and leaf chemistry traits, we sampled five individuals per species per site. Within 1–2 hr after harvest, all fully expanded leaves per plant were scanned (Scanner HP Scanjet 200; maximum resolution of 2,400 × 4,800 dpi) and their fresh weight (g) was measured using an analytical balance (ADAM PGL 203). Leaves were then oven-dried at 60°C (Binder FED 53–720) for 48 hr to measure dry weight. We measured leaf area (LA [cm2]) using ImageJ (Schneider, Rasband, & Eliceiri, 2012) and calculated specific leaf area (SLA; the ratio of leaf area to leaf mass [cm2/g]) and leaf dry matter content (LDMC; the ratio of leaf dry mass to fresh mass [mg/g]). We used the water displacement method, to estimate wood density of stem (WDs; the ratio of oven dry mass to green volume [g/cm3]) in a short section (i.e., 5–7 cm) of stem before removing its bark (Osazuwa-Peters, Zanne, & PrometheusWiki Contributors, 2011). Because root trait quantification involves a destructive procedure, for coarse-root traits (roots with diameter ≥2 mm) we only measured three individuals per species per site, and five individuals for fine-root traits (roots with diameter <2 mm). Root depth distribution, described by the beta index (β), was estimated by excavating adjacent to each shrub, a trench 1 m deep that extended outwards 1.6 m from the plant. Then, starting at the centre of each shrub, we collected a total of 100 20 × 20 × 10 cm blocks of soil samples from the trench walls (Supporting Information Figure S2). To determine root biomass, we separated roots according to their diameter into fine, medium and coarse roots (Ø < 1, 1–2 and >2 mm, respectively), and dried them in an oven at 60°C in order to stabilize their weight and be able to weigh them. Total root mass of each shrub was estimated by considering the root mass within the trench to be a proportion (~4%) of the total root biomass in a cylinder (1.6 m radius and 1 m depth; Morales et al., 2015). β was estimated from the asymptotic equation: Y = 1 – βd, where Y is the accumulated proportion of root biomass from the soil surface down to d depth (Gale & Grigal, 1987). Higher values of β indicate a greater proportion of roots deep in the soil (Gale & Grigal, 1987). We used roots of about 5 mm in diameter to estimate wood density of root (WDr; the ratio of oven-dried mass to green volume [g/cm3]) using the same methodology as for stems. Within 2 hr after harvesting, fine roots from each individual were cleaned with distilled water, scanned (Scanner HP Scanjet 200) and their fresh weight (g) recorded using an analytical balance (ADAM PGL 203). Roots were then oven-dried at 60°C (Binder FED 53–720) for 48 hr to record their dry weight. We estimated root dry matter content (RDMC; the ratio of root dry mass to fresh mass [mg/g]) and specific root length (SRL; the ratio of root length to dry mass [m/g]). From root images, we measured total root length using ImageJ (Schneider et al., 2012). Finally, leaf nitrogen concentration (LNC [%]), foliar carbon isotope ratio (δ13C [‰]) and root nitrogen concentration (RNC [%]) was quantified using an isotope ratio mass spectrometer at Laboratorio de Biogeoquímica e Isotopos Estables Aplicados (LABASI) at the Pontificia Universidad Católica de Chile. To examine shift and variation in fast–slow ecological strategies at the community level, we calculated two community-level metrics. First, we estimated community-weighted mean traits (Garnier et al., 2004) as , where pik is the relative abundance of species i at site k and xik is the trait value of species i at site k. CWM was estimated for each trait, within each plot of every site because it represents the dominant trait value of plants in a given community and thus allows us to identify how plants use and acquire resources under different aridity regimes. Second, we quantified the community-weighted trait variance (Sonnier, Shipley, & Navas, 2010) as , where pik is the relative abundance of species i at site k, xij is the trait value j of species i, and CWMjk is the community weighted mean of trait j at site k. CWV quantifies the variability of trait attributes around the mean trait value of the community. CWV was estimated for each trait, within each plot of every site and used as indicative of the variation in ecological strategies. To evaluate the shift in the functional structure of shrub assemblages in different plant organs along the aridity gradient, we performed separate principal component analyses (PCAs) for leaf and root traits. We used PCAs because the first axis explains a high proportion of the CWM variation, thus the scores of these axis can be used as a proxy for the fast–slow PES given that they represent gradients of trait variation across sites (de la Riva, Tosto, et al., 2016; Freschet et al., 2010). Because WDs was the only stem trait, we conducted a linear regression between DEMAI and the CWM of WDs to assess how WDs values change along the aridity gradient. Moreover, to assess how CWM values of each leaf and root trait change along the aridity gradient, we conducted separate linear regressions with quadratic components (Yi = β0 + β1Xi + β2Xi2 + εi), using DEMAI as an independent variable (Xi) and CWM values of each root and leaf traits as response variables (Yi). To test the effects of aridity on CWM, we performed a non-parametric permutational multivariate analysis of variance (PERMANOVA; Anderson & Walsh, 2013) using sites as independent factors. To quantify whether variation in functional structure of shrub assemblages along the gradient was mediated by changes in species occurrence or by changes in species abundance, we decomposed species turnover by estimating the relative importance of species occurrence and species abundance following de la Riva, Perez-Ramos, et al. (2016). Briefly, for each plant organ, we calculated three parameters: (a) a “fixed” trait value, which indicates the effect of species turnover, estimated as the CWM trait values averaged across all sites where a species was found (i.e., site-independent trait value); (b) an “unweighted” trait value across all sites where a species was found, which indicates the effect of species occurrence and; (c) species-abundance as the difference between “fixed” and “unweighted” values and reveals the pure effects of species abundance. Next, following Lepš, de Bello, Smilauer, and Dolezal (2011), we conducted an individual PERMANOVA for each parameter (i.e., fixed, unweighted and species abundance) using sites as independent factors and extracted the sum of squares (SS) from each model (SSfixed, SSunweighted, and SSspecies abundance, respectively). SSfixed represents total trait variation due to species turnover, SSunweighted and SSspecies abundance represent the contributions of species occurrence and species abundance to species turnover, respectively. Because species occurrence and species abundances could be responding to environmental factors in the same (= positive covariation) or opposite directions (= negative covariation), we also calculated the covariation component as SScov = SSfixed−SSunweighted–SSspecies abundance. To assess whether pairs of traits were perfectly coordinated (i.e., the slope of the relationship equal to one) at different spatial scales, we performed standardized major axis (SMA) regressions (Warton, Duursma, Falster, & Taskinen, 2012) at regional (i.e., considering all sites) and local scales (i.e., within each site). Given that the only plant organ that showed a clear differentiation within the fast–slow PES along the first PCA axis were leaves (leaf traits hereafter), we used the scores of this axis to regress against the CWM of WDs and each of the CWM of root traits. Coordination between stem and root traits was assessed by conducting SMA regressions between the CWM of WDs and each of the CWM of root traits. We used SMA regressions because these are useful when the aim is not to predict a dependent variable from an independent one, but rather to summarize the relationship between two variables (Warton, Wright, Falster, & Westoby, 2006). Finally, to examine the effects of aridity on trait variation (CWV), we performed a series (one per trait) of general linear models (GLMs; Crawley, 2007), using a Gaussian error distribution (link function “identity”). For all GLMs, we considered aridity as a main factor and the CWV of each trait as the response variable. All statistical analyses were performed using the r statistical environment (R Core Development Core, 2014). Aridity had a significant effect on the CWM of leaf traits (PERMANOVA: F3,78 = 64.73, R2 = 0.71, p < 0.0001). The first PCA axis accounted for 49.5% of the total variation across sites (Supporting Information Table S3). Specifically, the three drier sites had high values (positive PCA scores) of SLA and low values of LDMC and LNC (Figures 1a and 2, Supporting Information Table S4), suggesting that for morphological traits at least, ecological strategies shift from slow to fast as aridity increases. The second PCA axis was related to high values of δ13C and low values (negative PCA scores) of LA, which were associated with QL and CHA respectively (Figures 1a and 2, Supporting Information Table S4). The CWM of root traits was also significantly affected by aridity (PERMANOVA: F3,78 = 18.37, R2 = 0.41, p < 0.0001). The first PCA axis, which accounted for 36.5% of the total CWM variation across sites (Supporting Information Table S3) revealed that the two least arid sites (ROM and CHA) were associated with high values of SRL and RDMC, and with low WDr and RNC (Figure 1b and 2, Supporting Information Table S4). The second PCA axis, which accounted for 25% of the total CWM variation across sites, showed high values of β associated with QL and ROM (Figures 1b and 2, Supporting Information Table S4). These results reveal that the pattern of variation in root traits along an aridity gradient is complex and cannot be associated to a particular strategy within the fast–slow PES. WDs showed a significant negative relationship with aridity (Figure 1c), suggesting a change from slow (high WDs values) to fast (low WDs values) resource acquisition as aridity increases. The decomposition of species turnover into its components revealed that among-site averaged leaf trait values varied mainly due to changes in species occurrence (93.3%), rather than by changes in species abundance (10.1%) (Figure 3). In addition, only a fraction of the variation in averaged leaf traits values was because of a decrease in abundances resulting from changes in species occurrence (negative covariation of −4.1%; Figure 3). Thus, changes in SLA, LNC, LA, δ13C, and LDMC results mainly from changes in occurrence across sites. The response of WDs to aridity is also mainly explained by changes in species occurrence (69.5%), rather than by changes in species abundance (2.9%). In this case, however, there was a positive covariation (27.6%) between these two components of species turnover (Figure 3), suggesting that the decrease in the averaged WDs values associated to changes in occurrence along the gradient is in part also explained by a correlated decrease in the abundance of species with high values of WDs. The change in averaged root traits values along the gradient was explained both by changes in species occurrence (33.1%) and abundance (22.6%), and the change was intensified by the joint positive effect of components (positive covariation of 44.3%; Figure 3), suggesting that changes in mean trait values of SRL, RDMC, β, RNC, and WDr are influenced by both the contribution of species occurrence and species abundance. The SMA regressions between the first PCA axis of leaf traits and CWM values of SRL, RDMC, β, and WDs was negative at the regional scale (Figure 4, Supporting Information Table S5). At local scale, the relationship between the first PCA axis of leaf traits with the above-mentioned root traits was negative in some sites, but unrelated in others. Specifically, there was a relationship between leaf traits and SRL only at LLCHA (Figure 4, Supporting Information Table S5), between leaf traits and RDMC only at ROM and CHA (the least arid sites), and between leaf traits and β and WDs at QL and LLCHA (the most arid sites; Figure 4, Supporting Information Table S5). There was no relationship between leaf traits and WDr and RNC at the regional scale (Figure 4, Supporting Information Table S5). However, at a local scale, the relationship between leaf traits and WDr was negative in the most arid sites (QL and LLCHA; Figure 4, Supporting Information Table S5). The relationship between leaf traits and RNC was negative in the most arid site (QL) and positive in the least arid ones (ROM and CHA). The slope of the above relationships, except SRL in CHA and LLCHA at local scale, was statistically different from – 1 or +1 (Supporting Information Table S5), revealing that these organs are not perfectly coordinated. At the regional scale, the SMA regression revealed a positive relationship between CWM values of WDs and CWM values of RDMC and β, and a negative relationship between WDr and RNC (Figure 5, Supporting Information Table S6). Moreover, there was no relationship between WDs and SRL (Figure 5, Supporting Information Table S6). At a local scale, the relationship between WDs with RDMC was positive only in QL (the most arid site) and unrelated in the other sites (Figure 5, Supporting Information Table S6). The relationship between WDs with β was positive in LLCHA, CHA, and ROM; there was no relationship between these variables in QL. The relationship between WDs and WDr was positive in LLCHA, negative in CHA, and non-significant in the other sites (Figure 5, Supporting Information Table S6); with RNC, the relationship was positive in ROM (least arid sites) and non-significant in the other sites. The relationship between WDs and SRL was a negative only at ROM (Figure 5, Supporting Information Table S6). The slope of the relationships between WDs and root traits was statistically different from –1 or +1 (Supporting Information Table S6) in all sites, except for RNC in LLCHA and ROM and WDr at CHA. Again, these results suggest that WDs and other root traits are not perfectly coordinated. Aridity had significant effects on the variation of all CWV trait values, except LNC (Table 2). RNC, WDr, β, and WDs were more variable at the most arid site (QL) than in the other sites (Figure 6). Conversely, SRL had a higher variability at the least arid site (ROM). Variation of LDMC and δ13C tended to be higher in the three drier sites than in the least arid one (ROM; Figure 6). Variation of LA and SLA was lowest at LLCHA and CHA, respectively than in the other sites (Figure 6). RDMC was more variable at LLCHA than in either end of the aridity gradient (Figure 6). Our results reveal three important findings. First, that most of the leaf traits, as well as WDs, responded to aridity by shifting from a slow to a fast resource acquisition strategy as aridity increased; that is, shrub assemblages along the Coastal Atacama Desert adopt a faster above-ground strategy as the environment turns drier. In contrast, below-ground traits showed a more complex pattern of shift along the aridity gradient that cannot be associated to a particular strategy within the fast–slow PES. Second, there was some level of coordination among traits from different plant organs. Third, except for SRL, trait variation did not decrease as aridity increased. Together, these results partially support the hypothesis that as aridity increases in a hyper-arid environment, selection pressures act simultaneously on trait attributes, variation and coordination to promote an inverse pattern in the fast–slow PES. Although several studies have shown shifts in the fast–slow PES at both the species and community levels (e.g., de la Riva, Tosto, et al., 2016; Pérez-Ramos et al., 2012; Reich, 2014), to our knowledge this is the first study that reports an inverse pattern of the spectrum. Our prediction that in the most arid end of a hyper-arid aridity gradient shrub assemblages will exhibit trait values that reflect a water acquisition strategy at the whole plant level was partially supported. Changes in both leaf traits and WDs values reflect a shift from a slow to a fast strategy as aridity increased; this shift was determined mainly by changes in species occurrence across sites. Shifts in the functional structure of plant assemblages associated with changes in species occurrence have been related to habitat filtering because those species having sets of traits unsuitable to cope with particular environmental conditions tend to be eliminated (de la Riva, Perez-Ramos, et al., 2016; Keddy, 1992). We propose that selection pressures imposed by the hyper-aridity of this system, in conjunction with the short growing season, which increases from south to north (Rundel et al., 1991), lead to an inverse pattern of the fast–slow PES spectrum. Under these extreme conditions, a fast strategy may be beneficial because it allows plants to take advantage of the short periods of high resource availability (Mooney & Dunn, 1970; Nilsen & Muller, 1981), thus increasing their probability of survival (Ackerly, 2004; Mooney, 1982; Nilsen & Muller, 1981). The only above-ground trait that departed from a fast strategy was LNC; this trait depends on soil nitrogen availability, which in arid ecosystems is not only low but also highly dependent on soil moisture (Ward, 2009; Yu, Fan, Harris, & Li, 2017). Consequently, low levels of LNC may not have been determined by the net nitrogen acquisition rate, rather than by soil nitrogen availability. Conversely, below-ground traits did not exhibit a clear pattern within the fast–slow PES along the aridity gradient. Moreover, below-ground mean trait variation across sites was determined by changes in species occurrence and abundance, suggesting that both habitat filtering and local biotic processes determine the functional structure of root traits. Therefore, the apparent lack of functional adjustment to aridity at the root level may result to some extent from the hyper-aridity of this system, but also from local biotic processes, such as competition and/or facilitation (Cahill, 1999; López et al., 2016), which increase trait dissimilarity and allow plant coexistence by way of niche differentiation (Bernard-Verdier et al., 2012; de la Riva, Perez-Ramos, et al., 2016). At the regional scale, leaf traits showed a negative relationship with WDs, indicating that fast resource use by leaves is accompanied by fast water conduction by the stem; therefore, above-ground traits respond concordantly to aridity promoting rapid resource acquisition. Our results are in line with other studies that show some degree of coordination between leaf and stem traits (Freschet et al., 2010; Méndez-Alonzo et al., 2012). With respect to the coordination between leaf and root traits, we found negative relationships between leaf traits and RDMC, SRL, and β. In other words, fast resource use by leaves was associated with shallow root systems (low β values), low root length investment per root mass (low SRL values), and roots with low tissue investment in fine roots (low RDMC values). WDs was positively related to RDMC and β, meaning that fast water conduction by the stem is accompanied by a shallow root system and low tissue investment. In water-limited ecosystems, shallow root systems may be more relevant than deep root systems because: (a) both water and nitrogen availability concentrate in shallow soil layers, which are available only during brief time periods, followed by long periods of low resource supply (Jobbágy & Jackson, 2001; Noy-Meir, 1973; Reynolds, Kemp, Ogle, 2004, & Fernandez, 2004) and (b) shallow root systems are cheaper to construct and maintain than deep root systems (Adiku, Rose, Braddock, & Ozier-Lafontaine, 2000; Schenk & Jackson, 2002). Therefore, in arid and semiarid environments, shallow root systems have been indicated as a resource acquisition trait (Fort et al., 2013) because after small rainfall events, they enable plants to use water from shallow soil layers faster than deep root systems (Moreno-Gutiérrez, Dawson, Nicolás, & Querejeta, 2012). Additionally, it has been suggested that roots with low values of RDMC and SRL have an increased life span and therefore they increase the duration of the growing season (Weemstra et al., 2016; Withington et al., 2006; Zhou, Bai, Zhang, & Zhang, 2018). On the other hand, WDs was negatively with WDr and RNC. The negative relationship with RNC indicates that fast water transport by the stem is accompanied by fast fine root metabolism because high nitrogen content increases respiration rates (Guo et al., 2008), which in turn suggests that these traits are coordinated to have a fast rate of resource acquisition. The relationship between WDs and WDr is opposite from what is expected by theory (see Fortunel et al., 2012); that is, we showed that as WDs increases (fast resource acquisition) WDr decreases (slow resource acquisition). High WD is associated with a high resistance to mechanical damage (Van Gelder, Poorter, & Sterck, 2006) and low tissue mortality (King, Davies, Tan, & Noor, 2006); therefore, a decrease in WDr may promote plant persistence by maintaining the viability of coarse root tissue. Finally, the absence of a relationship between WDs and SRL, and between leaf traits with RNC and WDr indicated that these pair of traits are orthogonal, probably because of different selection pressures or constraints on these particular traits (Kembel & Cahill, 2011; Weemstra et al., 2016). In summary, at the regional scale most traits tend to show some degree of coordination, but the functional significance of this coordination differs from what is expected under the fast–slow PES framework. At local scales, coordination among traits was found in some sites, but not others. This may result from the occurrence of local processes, such as environmental heterogeneity and biotic interactions that promote a broader range of ecological strategies. For example, Kembel & Cahil (2011) suggested that environmental heterogeneity within communities may promote trait coordination because it leads to the coexistence of species with different trait values. Conversely, the lack of coordination suggests that above- and below-ground traits represent different ecological strategies that promote independent axes of niche differentiation for plant organs (Ackerly, 2004; Ackerly & Cornwell, 2007). The pattern of among-site trait variation reveals a mixture of different ecological strategies within each shrub assemblage along the aridity gradient, yet these are not segregated according to the gradient; rather, they showed an idiosyncratic pattern of variation. Our results differ from those of previous empirical studies in which trait variation decreased in poor-resource environments (Coyle et al., 2014; Grime et al., 1997; Swenson & Enquist, 2007) or alternatively, in which variability increased with aridity (e.g., Butterfield & Callaway, 2013; Butterfield et al., 2017). The pattern of trait variation we observed here could have resulted if different assembly processes at each site determine different traits. In this light, although several processes could explain our results, local biotic interactions, as well as environmental heterogeneity are the most plausible ones for two reasons. First, because López et al. (2016) found support for the stress gradient hypothesis in the Coastal Atacama Desert, meaning that facilitation increased and competition decreased as the environment become drier. Both facilitation and competition can increase functional trait variation in plant communities (Butterfield & Briggs, 2011; Butterfield & Callaway, 2013; Kraft, Valencia, & Ackerly, 2008) and thus could have promoted a broader range of ecological strategies. Second, spatial and temporal environmental-resource heterogeneity has been suggested as a promoter of high trait diversity because it reduces interspecific niche overlap (Adler et al., 2013; Chesson, 2000; Chesson et al., 2004). The few studies that have assessed how traits vary with resource heterogeneity have found a positive relationship between both variables (Butterfield et al., 2017; Price et al., 2017). For example, Price et al. (2017) studying grassland communities in Europe, found that high variation in leaf area, LDMC and SRL were related to heterogeneity in soil depth, a variable that has been related to spatial heterogeneity of soil resources (Gazol et al., 2012). Therefore, trait variability at different sites could be determined both by plant–plant interactions and within-site heterogeneity in soil resource availability. The main conclusion of this study is that, contrary to what is predicted by the fast-slow PES framework, a strong environmental filter, such as aridity, does not necessarily lead to an integrated whole-plant economics spectrum. Biotic processes can also be important drivers in determining resource acquisition strategies, particularly for below-ground traits. Given that the relative importance of assembly processes driving trait variation differs between above and below-ground traits, the interaction between habitat filtering and biotic processes can thus result in different above- and below-ground strategies, which change the functional significance of trait coordination. We thank all those who assisted with the field data collection, particularly Claus Westphal, Paulina Vera, and Natalio Roque. We also thank Víctor M. Escobedo, James F. Cahill, Jr., and two anonymous reviewers for suggestions that improved an earlier version of this manuscript. This research was supported by grants from FONDECYT Regular 1151020 and CONICYT PIA APOYO CCTE AFB170008. D.E.C. and C.A.D. were supported by CONICYT doctoral fellowships (21140050 and 21150334, respectively). A.P.L. was supported by FONDECYT initiation grant (11140400). R.S.R. was awarded a Rufford Small Grant 10015-1 and a DIULS Regular 2015. None of the authors declare conflicts of interest. D.E.C., A.P.L., and F.A.S. designed the study. D.E.C and C.A.D collected data. D.E.C. analysed all data and was the primary writer of the manuscript. R.S.R., A.P.L, and F.A.S contributed to writing and revising the manuscript. Data associated with this manuscript are deposited in the Dryad Digital Repository: at https://doi.org/10.5061/dryad.p9b92jh (Carvajal, Loayza, Rios, Delpiano, & Squeo, 2018). Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article." @default.
• W2900377459 created "2018-11-16" @default.
• W2900377459 creator A5055451223 @default.
• W2900377459 creator A5062739356 @default.
• W2900377459 creator A5063926212 @default.
• W2900377459 creator A5083146310 @default.
• W2900377459 creator A5091813676 @default.
• W2900377459 date "2018-12-10" @default.
• W2900377459 modified "2023-10-17" @default.
• W2900377459 title "A hyper‐arid environment shapes an inverse pattern of the fast–slow plant economics spectrum for above‐, but not below‐ground resource acquisition strategies" @default.
• W2900377459 cites W1541683308 @default.
• W2900377459 cites W1549764193 @default.
• W2900377459 cites W1553682677 @default.
• W2900377459 cites W1767254696 @default.
• W2900377459 cites W1915702182 @default.
• W2900377459 cites W1970001131 @default.
• W2900377459 cites W1975290893 @default.
• W2900377459 cites W1997052720 @default.
• W2900377459 cites W2001937735 @default.
• W2900377459 cites W2021125901 @default.
• W2900377459 cites W2023815040 @default.
• W2900377459 cites W2024926484 @default.
• W2900377459 cites W2025930408 @default.
• W2900377459 cites W2027704510 @default.
• W2900377459 cites W2034293169 @default.
• W2900377459 cites W2049774065 @default.
• W2900377459 cites W2050364851 @default.
• W2900377459 cites W2056581871 @default.
• W2900377459 cites W2060098961 @default.
• W2900377459 cites W2060230965 @default.
• W2900377459 cites W2077637812 @default.
• W2900377459 cites W2083044213 @default.
• W2900377459 cites W2088164236 @default.
• W2900377459 cites W2094635217 @default.
• W2900377459 cites W2096212927 @default.
• W2900377459 cites W2096985849 @default.
• W2900377459 cites W2099540110 @default.
• W2900377459 cites W2101020813 @default.
• W2900377459 cites W2109859491 @default.
• W2900377459 cites W2111568029 @default.
• W2900377459 cites W2115841368 @default.
• W2900377459 cites W2116299300 @default.
• W2900377459 cites W2116344164 @default.
• W2900377459 cites W2117029130 @default.
• W2900377459 cites W2117354222 @default.
• W2900377459 cites W2119567434 @default.
• W2900377459 cites W2126051177 @default.
• W2900377459 cites W2126535795 @default.
• W2900377459 cites W2127136425 @default.
• W2900377459 cites W2127474289 @default.
• W2900377459 cites W2127928904 @default.
• W2900377459 cites W2128723120 @default.
• W2900377459 cites W2130373714 @default.
• W2900377459 cites W2131942074 @default.
• W2900377459 cites W2145051394 @default.
• W2900377459 cites W2145711663 @default.
• W2900377459 cites W2147841106 @default.
• W2900377459 cites W2150016885 @default.
• W2900377459 cites W2150597302 @default.
• W2900377459 cites W2152987645 @default.
• W2900377459 cites W2154948142 @default.
• W2900377459 cites W2157701121 @default.
• W2900377459 cites W2160264708 @default.
• W2900377459 cites W2163083294 @default.
• W2900377459 cites W2163688790 @default.
• W2900377459 cites W2166015583 @default.
• W2900377459 cites W2172956642 @default.
• W2900377459 cites W2270063699 @default.
• W2900377459 cites W2284861330 @default.
• W2900377459 cites W2388386570 @default.
• W2900377459 cites W2462899698 @default.
• W2900377459 cites W2517920139 @default.
• W2900377459 cites W2556017858 @default.
• W2900377459 cites W2570098172 @default.
• W2900377459 cites W2586740126 @default.
• W2900377459 cites W2600569241 @default.
• W2900377459 cites W2755758409 @default.
• W2900377459 cites W2765868792 @default.
• W2900377459 cites W2791775578 @default.
• W2900377459 cites W3019724458 @default.
• W2900377459 cites W4229591924 @default.
• W2900377459 cites W4249959076 @default.
• W2900377459 cites W4379372972 @default.
• W2900377459 doi "https://doi.org/10.1111/1365-2745.13092" @default.
• W2900377459 hasPublicationYear "2018" @default.
• W2900377459 type Work @default.
• W2900377459 sameAs 2900377459 @default.
• W2900377459 citedByCount "38" @default.
• W2900377459 countsByYear W29003774592020 @default.
• W2900377459 countsByYear W29003774592021 @default.
• W2900377459 countsByYear W29003774592022 @default.
• W2900377459 countsByYear W29003774592023 @default.
• W2900377459 crossrefType "journal-article" @default.
• W2900377459 hasAuthorship W2900377459A5055451223 @default.
• W2900377459 hasAuthorship W2900377459A5062739356 @default.
• W2900377459 hasAuthorship W2900377459A5063926212 @default.
• W2900377459 hasAuthorship W2900377459A5083146310 @default.
• W2900377459 hasAuthorship W2900377459A5091813676 @default.

• next