American Journal of Botany 101(8): 1286–1292, 2014.
DRIVERS OF A RIPARIAN FOREST SPECIALIST (CAREX REMOTA, CYPERACEAE): IT IS NOT ONLY A MATTER OF SOIL MOISTURE1 JAIME URIA-DIEZ2,4, ANTONIO GAZOL3, AND RICARDO IBÁÑEZ2 2Department
of Environmental Biology, University of Navarra, Irunlarrea s/n 31008 Pamplona, Spain; and 3Institute of Ecology and Earth Sciences, University of Tartu, Lai 40 51005 Tartu, Estonia
• Premise of the study: Plants respond to the prevailing conditions in the surrounding environment, but since they are dynamic systems this response may vary during their life. Thus, the identification of key aspects for the maintenance of plant populations requires the consideration of plant performance across environmental gradients and along life stages. This study examines how abiotic conditions and biotic interactions and processes determine the spatial distribution of two life-story stages that play a key role in the functioning of a representative population of Carex remota. • Methods: We used structural equation modeling (SEM) to test for direct and indirect influences of abiotic and biotic factors on seedlings and adults of Carex remota. The variables used in the analysis were number of seedlings, cover of adults, soil moisture, leaf litter cover, relative light, and topographic position. • Key results: Population patterns partially depend on direct and indirect effects of abiotic conditions. Whereas adult individuals were only affected by topsoil moisture, seedling emergence was largely affected by multiple environmental conditions. The number of seedlings increased with high topsoil moisture, low leaf-litter values, high light values as well as in low parts of the study area. The importance of adult individuals in determining seedling success is also highlighted: higher abundance provides seed rain in the surroundings and modifies the microenvironmental conditions favoring high seedling establishment. • Conclusions: As hypothesized, adults and seedlings responded to the environmental conditions differently. Seedling emergence was a critical aspect in C. remota performance, and abrupt changes in the environment during this stage may strongly influence population performance. Key words: Carex remota; Cyperaceae; forest streams; seedling establishment; soil moisture; structural equation modeling; temperate deciduous forest.
Understanding the mechanisms responsible for the maintenance of plant populations, i.e., individuals of one species that inhabit a certain place, has been a principal goal of ecologists for centuries (Begon et al., 1996). Current patterns in plant populations are the result of the mixed action of abiotic and biotic forces to which their members, and in some cases their ancestors, have been exposed in the past (Crawley, 1997). According to niche theory, physical environmental conditions represent a primal filter since plants specialize in particular abiotic conditions (Silvertown and Charlesworth, 2001), but intra- and interspecific interactions among coexisting individuals within a local community also determine plant population patterns (Chesson, 2000). The physical environment (e.g., temperature, soil pH, water, radiation) may exert direct or indirect effects on plant species distribution across gradients (Hagan et al., 2006), and they can determine the outcome of biotic interactions among coexisting plants by modifying their strength and direction, altering the likelihood for a species to persist (García-Cervigón et al., 2013; Kardol et al., 2013). In addition, the response of plant individuals can vary depending on their ontogeny (Huber et al., 1999). The success of a certain plant population may depend on how individuals respond to the physical environmental 1 Manuscript received 6 January 2013; revision accepted 9 July 2014. The authors thank the Departmento de Desarrollo Rural y Medio Ambiente del Gobierno de Navarra and the Parque Natural Señorío de Bertiz for allowing the research. The research was supported by funding from Fundación Caja Navarra and Fundación Universitaria de Navarra. 4 Author for correspondence (e-mail: [email protected])
doi:10.3732/ajb.1300443
conditions and how they interact with their neighbors, and this may be influenced by the variation in the abiotic and biotic conditions and the ontogeny of each considered individual. The identification of key aspects for the maintenance of plant populations can help us understand natural systems and develop suitable conservation and management plans. Broad-leaved forests and their understory plants are the climax vegetation in boreal and temperate regions of Europe (Paillet et al., 2010). Despite the low biomass, the understory plays an important role in the functioning of forests because it accounts for the largest number of species (Gilliam and Roberts, 2003; Whigham, 2004). Specific studies have been carried out to understand the factors maintaining the populations of herbaceous species in the forest understory (for a review, see Gilliam, 2007; Whigham, 2004). It is well documented that most herb species are long-lived perennials characterized by low dispersibility and clonal growth (Gilliam, 2007) so we would expect their lifehistory parameters to influence the spatial structure of neighboring individuals and populations. Within the forest understory, the variability in structural factors favors the coexistence of species with different abiotic and biotic requirements. For example, riparian areas, especially low-order streams, provide favorable conditions for populations of species that are not present in neighboring upslope areas (D’Souza et al., 2012). The individuals of these forest-stream specialists are strongly linked to the physical environmental conditions prevailing near the stream (wet places and shady areas) and highly depend on the spatial extent of the water flow (Uria-Diez and Ibáñez, 2014). Riparian forest herb species have localized dispersal, but seeds can be efficiently transported by water over long distances (Araujo Calçada et al., 2013). Therefore, determining the factors that influence
American Journal of Botany 101(8): 1286–1292, 2014; http://www.amjbot.org/ © 2014 Botanical Society of America
1286
August 2014]
URIA-DIEZ—DRIVERS OF A RIPARIAN FOREST SPECIALIST
the spatial patterns of individuals at local scales is crucial for populations of species inhabiting forest streams. Microhabitat heterogeneity has long been recognized as a factor determining plant distribution patterns in the understory of deciduous forests (Beatty, 1984; Burton et al., 2011). Small topographic variations near forest streams such as small dips created by the water flow modify the prevailing abiotic conditions favoring the presence of plant individuals (Pabst and Spies, 1999). In addition, topographic variations can determine patterns of plant individuals indirectly by modifying abiotic factors such as light values (Vockenhuber et al., 2011), topsoil moisture (Hagan et al., 2006), and soil nutrient content (Tateno and Takeda, 2003). For example, Uria-Diez and Ibáñez (2014) showed that the distribution of the specialist riparian sedge Carex remota L. (Cyperaceae) was strongly linked to high topsoil moisture values near the stream that were mainly determined by small differences in elevation from the stream bank. The abiotic and biotic factors that influence plant populations can vary depending on the ontogeny of individuals (Huber et al., 1999). For example, in the germination and establishment of herbaceous species, one of these filters is the physical barrier imposed by the tree leaf litter (Facelli and Pickett, 1991; Vellend et al., 2000). The germination and establishment of forest herbs tend to be strongly sensitive to variations in topography, temperature, and light (Jacquemyn et al., 2010; Laskurain et al., 2004; Zukowski et al., 2010). In addition, establishment is a phase of the plant life cycle strongly determined by the biotic interactions among seedlings and with adult conspecifics (Leishman, 1999; De la Cruz et al., 2008). In the case of forest herbs, short-distance dispersal can concentrate new establishments close to old establishments, hence facilitating patchiness within populations in homogeneous environments (Baeten et al., 2009). Adult individuals can influence seedlings directly by competitive hierarchy and also indirectly by modifying the abiotic conditions (Cavieres et al., 2007). The outcome will depend on the relative scales of the seed shadow, the influence of competition and the environmental heterogeneity. In this study, we sample a representative population of the riparian forest specialist Carex remota and use it to understand the response of individuals to the physical environmental conditions and the presence of their neighbors across environmental gradients and along life stages. Specifically, we are interested in determining the main factors influencing the distribution patterns of C. remota when it grows on low-order streams, and we aim to disentangle (1) how abiotic conditions determine the distribution patterns of adults, and (2) how abiotic filters, biotic interactions, and ecological processes determine the emergence of seedlings in low order stream banks.
1287
Carex remota is a forest specialist sedge and is distributed over Europe, North Africa, and West Asia, being frequent in the northern Iberian Peninsula (Luceño, 2008). The species is shade-tolerant and occurs in moist soils of beech and riparian forests (Uria-Diez, 2011). It is an herbaceous perennial sedge, presumably long-living, with tufted shoots and without the capability to form stems that grows horizontally along the ground (stolons) (Brändel and Schütz, 2005). Stems and leaves are 25–50 cm long on average (Luceño, 2008). Carex remota flowers in spring and seeds ripen in summer (Hegi, 1980). Average seed size is 3–3.2 × 1–1.3 mm (Luceño, 2008) and average mass is 0.5 mg (Brändel, 2005). Brändel and Schütz (2005) have extensively studied the germination ecology of this species. Seeds are conditionally dormant at maturity and after-ripen at low temperatures. Germination is greater at high temperatures, high fluctuations in temperatures, and in light. Burial experiments have demonstrated the capability of the species to build up a long-term, persistent seed bank. A population of C. remota was selected in a study area of 256 m2 based on previous studies of the watershed (Gazol and Ibáñez, 2010). The area was covered by a uniform F. sylvatica overstory, and C. remota was the only herbaceous species in the understory. This simple biotic system, without any within-guild, interspecific interactions, simplified the study and allowed us to focus on the role of abiotic factors and intraspecific interactions. We also considered this population representative of the watershed and the region because of its high number of individuals, size, growth form, and ability to produce viable seeds. Study design—The 256-m2 area was subdivided into 256 nonoverlapping, 1-m2 plots (Fig. 1). In each 1-m2 plot, we measured the cover of adults and number of seedlings of the C. remota population. Adults were defined as those possessing reproductive structures. Seedlings were defined as new individuals from seeds that germinated in the previous spring and having a small shoot with up to 3 or 4 leaves. The study area includes a segment of a first-order stream and two moist zones. We also measured several abiotic variables quantified throughout the study area (Table 1). Different grids were used for sampling those variables, according to their heterogeneity, although we averaged or interpolated values to define all
MATERIALS AND METHODS Study site and species—The study was carried out in the Suspiro watershed, Señorío de Bertiz Natural Park, northern Spain. The area is on a north-oriented slope, 350 m above sea level. The climate is oceanic, with a mean annual precipitation of 1600 mm and a drier period in summer, although water runs almost continuously throughout the year. Temperatures range from 7.2°C in the coldest month, January, to 22.4°C in the warmest, July, with an annual mean of 13.4°C (Gobierno de Navarra, 2009; period 1992–2009). Parent rock materials are predominantly schists (Gobierno de Navarra, 1995). Vegetation consists of an oldgrowth deciduous forest dominated by beech (Fagus sylvatica L.), accompanied by some scattered species including pedunculate oak (Quercus robur L.) and sessile oak [Quercus petraea (Matt.) Liebl.].
Fig. 1. Study area with 256 nonoverlapping 1-m2 plots in Señorío de Bertiz Natural Park (Spain). In each 1-m2 plot, cover of Carex remota adults is represented by black circles, and number of seedlings is represented by white numbers. Soil moisture values are also gray.
1288
[Vol. 101
AMERICAN JOURNAL OF BOTANY
TABLE 1.
Summary of the environmental variables quantified throughout the study area in Señorío de Bertiz Natural Park (Spain) and used for the structural equation model.
Variable
Mean
SD
Range
Soil moisture (vol. H2O/vol. soil) Leaf litter cover (%) Relative light (W/m2) Topographic position (m)
30.25 52.07 0.264 3.30
20.69 25.05 0.041 1.40
5.5–82.9 0–100 0.182–0.355 0–6.4
variables in each of the 256 plots. Top soil moisture was measured in a regular grid of 1024 sampling points due to its high variability. Measures were taken twice, in August 2009 and February 2010, with a soil moisture sensor SM200 (Delta-T device, Cambridge, UK). We estimated the cover and the depth of the leaf litter layer in each 1 m2 plot. Solar radiation indices were quantified using hemispherical photography at 64 points in a regular grid. Photographs were taken at 80–100 cm above ground with a digital camera (EOS 50D, Canon, Tokyo, Japan) equipped with a fish-eye lens (4.5 mm F2.8 EX DC, Sigma, Tokyo, Japan), with a 180° field of view. Images were analyzed using Hemiview canopy analysis software version 2.1 (Delta-T Devices, Cambridge, UK). We calculated the indirect, direct, and global site factor (ISF, DSF, and GSF), which are the proportions of indirect, direct and global solar radiation, respectively, that reach a given location over a year. In 256 points, we also measured the height, relative to the lowest point in the area, to create a microtopographic map of the study area. Statistical analyses and hypothetical model development—We used structural equation modeling (SEM; Bollen, 1989) to determine the influence that abiotic and biotic conditions have on the distribution of the C. remota population. SEM is a statistical method that provides a direct linkage between the hypothesis of ideas and statistical testing and has been widely used in ecology during the last years (Grace, 2006). In addition, SEM allows testing direct and indirect (i.e., through the modification of a third factor) relationships among the variables studied. To discover which factors influence the distribution of the C. remota population, we developed a hypothetical model based on previous knowledge of the system and ecological theory. In our model, we considered abiotic variables that have been found to have a significant influence on C. remota distribution, such as soil moisture, and others that can have potential influences on the distribution of nemoral species such as leaf-litter coverage and topographic position. Similarly, and because we were interested in discovering the factors influencing C. remota establishment, we considered light conditions as an additional abiotic factor. Our hypothetical model included direct and indirect relationships among the variables listed above (Fig. 2). We hypothesize that the distribution of C. remota adult individuals can be directly and positively influenced by the soil moisture conditions, since this species strongly depends on water availability. Similarly,
Fig. 2. Hypothetical structural equation model established to study the direct and indirect influence of the different abiotic and biotic conditions on the distribution of adults and seedlings of Carex remota. Singleheaded arrows indicate direct causal relationships among variables; double-headed arrows indicate nonstudied relationships. Hypothesized positive relationships are indicated with solid lines; negative ones are indicated with dashed lines.
we expect an indirect negative influence of topographic position (relative height) on the distribution of C. remota adults, since soil water retention capacity is influenced by the topographic conditions. We also expect that the distribution of C. remota adults can modify the abiotic conditions, mainly by its negative influence on leaf-litter coverage. Regarding the distribution of C. remota seedlings, we expect that they will be directly and positively influenced by the soil water content. Similarly, we expect a direct positive influence of light conditions and negative of leaf-litter coverage on seedling distribution due to the importance of these two factors on seed germination and seedling establishment. Finally, we expect that seedling distribution can be directly influenced by the distribution of adult individuals as a consequence of propagule pressure and indirectly through the influence of adult individuals on leaf-litter coverage. Structural equation models were calculated using the IBM-SPSS AMOS version 19 (IBM, Armonk, New York, USA) statistical software. Before the analyses, we studied whether the different variables were normally distributed by visual and numerical analyses of their skewness and kurtosis. Variables with a significant departure from normality were log-transformed [x′ = log (x + 1)]. Specifically, we transformed cover of C. remota adults and number of C. remota seedlings. Models were estimated using maximum likelihood procedures. The χ2 statistic and its associated probability (p) were used to evaluate the overall model fit. Values of p > 0.05 indicate a statistically acceptable model fit (Grace, 2006). Nevertheless, the test of SEM validity is a process that combines statistical testing and theoretical justification (Grace et al., 2010). In case the proposed model failed to match the system under study, we considered the removal and addition of pathways taking into account the absolute model fit, a theoretical justification, and the principle of parsimony. Specifically, we studied the removal of unimportant relationships and the inclusion of novel causal relationships that will improve the model fit while keeping the biological meaning of the system. The procedure was stopped when a model yielded satisfactory statistical results (χ2 associated p > 0.05). A critical issue is that the spatial distribution of biological data can be influenced by stochastic processes that can create spatial autocorrelation and violate the assumptions of parametric statistics (Peres-Neto and Legendre, 2010). The influence of the spatial autocorrelation can have important implications on the calculation of model parameters and the residuals (Dormann et al., 2007; Beguería and Pueyo, 2009). To control for the influence of spatial autocorrelation, we studied the presence of spatial autocorrelation in the model residuals of the two main variables: the adult abundance and the seedling number of C. remota. We used Mantel tests (Legendre and Legendre, 2012) to test whether the distance between pairs of residual values was correlated with geographic distance. These analyses were performed using the R statistical software (R Development Core Team, 2014) using the vegan library (2.0-7; Oksanen et al., 2008).
RESULTS A total of 275 C. remota adult individuals and 1619 seedlings were found in the area studied. Sixty-six of the 256 plots were occupied by both adults and seedlings, 19 only by adults, and 39 only by seedlings. The spatial distribution of topsoil moisture is shown in Fig. 1 and its values summarized in Table 1. Similarly, the measurement of leaf-litter coverage, topography and light conditions indicated variation in these two variables (Table 1). The calculation of the hypothesized SEM yielded nonsatisfactory statistics (χ2 = 21.20; df = 6; p = 0.002). Therefore, we considered the inclusion of different direct and indirect relationships among the variables to improve model fit while considering the biological meaning of the potential relationships. First, we removed the relationship between topographic position and leaf-litter coverage because it was unimportant (r = 0.008; p = 0.88). Since the results still yielded nonsatisfactory results (χ2 = 21.20; df = 7; p = 0.003), we developed a new model including a causal relationship between light and soil moisture (i.e., indicating that light conditions can modify soil moisture). The inclusion of this pathway increased the model fit substantially, but the results were still not satisfactory (χ2 = 15.53; df = 6; p = 0.016). Thus, we included a new causal relationship between topographic position and C. remota seedlings (Fig. 3). The new SEM yielded satisfactory statistical results
August 2014]
URIA-DIEZ—DRIVERS OF A RIPARIAN FOREST SPECIALIST
1289
Fig. 3. Structural equation model studying the direct and indirect influence of the different abiotic and biotic conditions on the distribution of adults and seedlings of Carex remota (compare with the model in Fig. 2). Single-headed arrows indicate direct causal relationships among variables; double-headed arrows indicate nonstudied relationships. Hypothesized positive relationships are indicated with solid lines; negative ones are indicated with dashed lines. The coefficients associated with the different relationships are listed in Table 2. Lines are proportional to the standardized effect of one variable on each other. The R2 values on top of each box represent the fraction of variance explained of each variable.
(χ2 = 8.27; df = 5; p = 0.123) and kept the biological meaning of the system. The residuals of the model showed no evidence of spatial autocorrelation as C. remota adult abundance residuals (r = 0.013; p = 0.27) and C. remota seedlings residuals (r = 0.001; p = 0.46) were not related to geographical distance. As was suggested in our model, soil moisture content has a positive direct influence on C. remota adult distribution (Fig. 4; Table 2). The model also shows an indirect positive influence of topographic position on C. remota adult distribution, since topographic position had a positive influence on soil moisture. Seedling distribution was the result of multiple complex effects of most of the variables considered, as was suggested in our model (Fig. 3; Table 3). In line with our expectations, C. remota seedlings were positively influenced by soil moisture, the distribution of adult individuals and light conditions. Similarly, seedling distribution was negatively influenced by leaf-litter coverage as we hypothesized in our hypothetical model (Fig. 2). The proposed model accounted for 64% of the variation in C. remota adult distribution and 37% of the variation in the distribution of seedlings (Fig. 5). DISCUSSION Our results demonstrate that the distribution patterns of the forest specialist Carex remota partially depend on direct and indirect effects of abiotic conditions. Our findings confirm that seedling establishment, a crucial phase in the life cycle of the plant, is strongly dependent on multiple abiotic conditions. As hypothesized, the number of seedlings increases in favorable microsites characterized by high topsoil moisture, low leaf-litter values, and high light values. These results are in accordance with our previous knowledge about the germination of forest specialist herbs. Furthermore, our results highlight the importance of the presence of adult individuals and their size in determining seedling success. Larger individuals ensure seed production favoring seed rain in their surroundings. In addition,
Fig. 4. Linear relationship between the distribution of adult individuals of Carex remota and soil moisture content in the Bertiz forest. The scatter diagrams show the relationship between percentage-scale cover of adult individuals (log-transformed) and soil moisture. Zeroes have been removed from the graphs. The red line indicates the linear relationship between the two variables considering zeroes.
these individuals can modify the microenvironmental conditions favoring the establishment of more seedlings. Finally, the results show that adult individuals are mainly determined by topsoil moisture levels indicating that the establishment of seedlings is the crucial stage for population development. A substantial fraction of variation remains unexplained and points to a possible impact of factors not considered in this study, such as past events and genetic population structure on the distribution of C. remota. Overall, our study suggests that abrupt changes in moisture and light conditions during the germination phase or the removal of adult individuals may influence the establishment of new C. remota individuals triggering changes seedling recruitment and adult survival. The availability of safe sites for germination may be considered a key aspect for the performance of C. remota populations. The abundance of C. remota strongly depends on water availability, as the cover of adults increases with the increase of topsoil water content, which is partially determined by the topographic heterogeneity. In addition, the results of our model confirm that C. remota is shade tolerant since the distribution of adult individuals was independent of light conditions. Furthermore, C. remota is a specialized sedge that inhabits constantly moist or damp, water saturated, and badly aerated sites (Grime et al., 2007; Uria-Diez et al., 2013) and is able to tolerate the deep shade of forest interiors. In contrast to the results obtained for adult individuals, the distribution patterns of the seedlings of C. remota depend on multiple direct and indirect abiotic conditions, as well as their positive interaction with adult conspecifics. These differences between individuals at different stages
1290
[Vol. 101
AMERICAN JOURNAL OF BOTANY
TABLE 2.
Path coefficients for the structural equation model of Carex remota. For each of the paths proposed in Fig. 3, the estimated coefficient, standard error, its standardized estimate, and associated P value are shown.
Path
Estimate
SE
Standardized estimate
P
Topographic position → Soil moisture Topographic position → C. remota seedlings Relative light → Soil moisture Relative light → C. remota seedlings Soil moisture → C. remota adult Soil moisture → C. remota seedlings C. remota adult → leaf-litter coverage C. remota adult → C. remota seedlings leaf-litter coverage → C. remota seedlings Topographic position ←→ Relative light
2.41 −0.062 −65.508 3.422 0.038 0.011 −11.529 0.183 −0.002 0.031
0.779 0.024 27.376 0.823 0.002 0.003 1.986 0.065 0.001 0.004
0.223 −0.161 −2.393 0.252 0.798 0.314 −0.342 0.243 −0.110 0.528
0.002 0.008 0.017
DRIVERS OF A RIPARIAN FOREST SPECIALIST (CAREX REMOTA, CYPERACEAE): IT IS NOT ONLY A MATTER OF SOIL MOISTURE1 JAIME URIA-DIEZ2,4, ANTONIO GAZOL3, AND RICARDO IBÁÑEZ2 2Department
of Environmental Biology, University of Navarra, Irunlarrea s/n 31008 Pamplona, Spain; and 3Institute of Ecology and Earth Sciences, University of Tartu, Lai 40 51005 Tartu, Estonia
• Premise of the study: Plants respond to the prevailing conditions in the surrounding environment, but since they are dynamic systems this response may vary during their life. Thus, the identification of key aspects for the maintenance of plant populations requires the consideration of plant performance across environmental gradients and along life stages. This study examines how abiotic conditions and biotic interactions and processes determine the spatial distribution of two life-story stages that play a key role in the functioning of a representative population of Carex remota. • Methods: We used structural equation modeling (SEM) to test for direct and indirect influences of abiotic and biotic factors on seedlings and adults of Carex remota. The variables used in the analysis were number of seedlings, cover of adults, soil moisture, leaf litter cover, relative light, and topographic position. • Key results: Population patterns partially depend on direct and indirect effects of abiotic conditions. Whereas adult individuals were only affected by topsoil moisture, seedling emergence was largely affected by multiple environmental conditions. The number of seedlings increased with high topsoil moisture, low leaf-litter values, high light values as well as in low parts of the study area. The importance of adult individuals in determining seedling success is also highlighted: higher abundance provides seed rain in the surroundings and modifies the microenvironmental conditions favoring high seedling establishment. • Conclusions: As hypothesized, adults and seedlings responded to the environmental conditions differently. Seedling emergence was a critical aspect in C. remota performance, and abrupt changes in the environment during this stage may strongly influence population performance. Key words: Carex remota; Cyperaceae; forest streams; seedling establishment; soil moisture; structural equation modeling; temperate deciduous forest.
Understanding the mechanisms responsible for the maintenance of plant populations, i.e., individuals of one species that inhabit a certain place, has been a principal goal of ecologists for centuries (Begon et al., 1996). Current patterns in plant populations are the result of the mixed action of abiotic and biotic forces to which their members, and in some cases their ancestors, have been exposed in the past (Crawley, 1997). According to niche theory, physical environmental conditions represent a primal filter since plants specialize in particular abiotic conditions (Silvertown and Charlesworth, 2001), but intra- and interspecific interactions among coexisting individuals within a local community also determine plant population patterns (Chesson, 2000). The physical environment (e.g., temperature, soil pH, water, radiation) may exert direct or indirect effects on plant species distribution across gradients (Hagan et al., 2006), and they can determine the outcome of biotic interactions among coexisting plants by modifying their strength and direction, altering the likelihood for a species to persist (García-Cervigón et al., 2013; Kardol et al., 2013). In addition, the response of plant individuals can vary depending on their ontogeny (Huber et al., 1999). The success of a certain plant population may depend on how individuals respond to the physical environmental 1 Manuscript received 6 January 2013; revision accepted 9 July 2014. The authors thank the Departmento de Desarrollo Rural y Medio Ambiente del Gobierno de Navarra and the Parque Natural Señorío de Bertiz for allowing the research. The research was supported by funding from Fundación Caja Navarra and Fundación Universitaria de Navarra. 4 Author for correspondence (e-mail: [email protected])
doi:10.3732/ajb.1300443
conditions and how they interact with their neighbors, and this may be influenced by the variation in the abiotic and biotic conditions and the ontogeny of each considered individual. The identification of key aspects for the maintenance of plant populations can help us understand natural systems and develop suitable conservation and management plans. Broad-leaved forests and their understory plants are the climax vegetation in boreal and temperate regions of Europe (Paillet et al., 2010). Despite the low biomass, the understory plays an important role in the functioning of forests because it accounts for the largest number of species (Gilliam and Roberts, 2003; Whigham, 2004). Specific studies have been carried out to understand the factors maintaining the populations of herbaceous species in the forest understory (for a review, see Gilliam, 2007; Whigham, 2004). It is well documented that most herb species are long-lived perennials characterized by low dispersibility and clonal growth (Gilliam, 2007) so we would expect their lifehistory parameters to influence the spatial structure of neighboring individuals and populations. Within the forest understory, the variability in structural factors favors the coexistence of species with different abiotic and biotic requirements. For example, riparian areas, especially low-order streams, provide favorable conditions for populations of species that are not present in neighboring upslope areas (D’Souza et al., 2012). The individuals of these forest-stream specialists are strongly linked to the physical environmental conditions prevailing near the stream (wet places and shady areas) and highly depend on the spatial extent of the water flow (Uria-Diez and Ibáñez, 2014). Riparian forest herb species have localized dispersal, but seeds can be efficiently transported by water over long distances (Araujo Calçada et al., 2013). Therefore, determining the factors that influence
American Journal of Botany 101(8): 1286–1292, 2014; http://www.amjbot.org/ © 2014 Botanical Society of America
1286
August 2014]
URIA-DIEZ—DRIVERS OF A RIPARIAN FOREST SPECIALIST
the spatial patterns of individuals at local scales is crucial for populations of species inhabiting forest streams. Microhabitat heterogeneity has long been recognized as a factor determining plant distribution patterns in the understory of deciduous forests (Beatty, 1984; Burton et al., 2011). Small topographic variations near forest streams such as small dips created by the water flow modify the prevailing abiotic conditions favoring the presence of plant individuals (Pabst and Spies, 1999). In addition, topographic variations can determine patterns of plant individuals indirectly by modifying abiotic factors such as light values (Vockenhuber et al., 2011), topsoil moisture (Hagan et al., 2006), and soil nutrient content (Tateno and Takeda, 2003). For example, Uria-Diez and Ibáñez (2014) showed that the distribution of the specialist riparian sedge Carex remota L. (Cyperaceae) was strongly linked to high topsoil moisture values near the stream that were mainly determined by small differences in elevation from the stream bank. The abiotic and biotic factors that influence plant populations can vary depending on the ontogeny of individuals (Huber et al., 1999). For example, in the germination and establishment of herbaceous species, one of these filters is the physical barrier imposed by the tree leaf litter (Facelli and Pickett, 1991; Vellend et al., 2000). The germination and establishment of forest herbs tend to be strongly sensitive to variations in topography, temperature, and light (Jacquemyn et al., 2010; Laskurain et al., 2004; Zukowski et al., 2010). In addition, establishment is a phase of the plant life cycle strongly determined by the biotic interactions among seedlings and with adult conspecifics (Leishman, 1999; De la Cruz et al., 2008). In the case of forest herbs, short-distance dispersal can concentrate new establishments close to old establishments, hence facilitating patchiness within populations in homogeneous environments (Baeten et al., 2009). Adult individuals can influence seedlings directly by competitive hierarchy and also indirectly by modifying the abiotic conditions (Cavieres et al., 2007). The outcome will depend on the relative scales of the seed shadow, the influence of competition and the environmental heterogeneity. In this study, we sample a representative population of the riparian forest specialist Carex remota and use it to understand the response of individuals to the physical environmental conditions and the presence of their neighbors across environmental gradients and along life stages. Specifically, we are interested in determining the main factors influencing the distribution patterns of C. remota when it grows on low-order streams, and we aim to disentangle (1) how abiotic conditions determine the distribution patterns of adults, and (2) how abiotic filters, biotic interactions, and ecological processes determine the emergence of seedlings in low order stream banks.
1287
Carex remota is a forest specialist sedge and is distributed over Europe, North Africa, and West Asia, being frequent in the northern Iberian Peninsula (Luceño, 2008). The species is shade-tolerant and occurs in moist soils of beech and riparian forests (Uria-Diez, 2011). It is an herbaceous perennial sedge, presumably long-living, with tufted shoots and without the capability to form stems that grows horizontally along the ground (stolons) (Brändel and Schütz, 2005). Stems and leaves are 25–50 cm long on average (Luceño, 2008). Carex remota flowers in spring and seeds ripen in summer (Hegi, 1980). Average seed size is 3–3.2 × 1–1.3 mm (Luceño, 2008) and average mass is 0.5 mg (Brändel, 2005). Brändel and Schütz (2005) have extensively studied the germination ecology of this species. Seeds are conditionally dormant at maturity and after-ripen at low temperatures. Germination is greater at high temperatures, high fluctuations in temperatures, and in light. Burial experiments have demonstrated the capability of the species to build up a long-term, persistent seed bank. A population of C. remota was selected in a study area of 256 m2 based on previous studies of the watershed (Gazol and Ibáñez, 2010). The area was covered by a uniform F. sylvatica overstory, and C. remota was the only herbaceous species in the understory. This simple biotic system, without any within-guild, interspecific interactions, simplified the study and allowed us to focus on the role of abiotic factors and intraspecific interactions. We also considered this population representative of the watershed and the region because of its high number of individuals, size, growth form, and ability to produce viable seeds. Study design—The 256-m2 area was subdivided into 256 nonoverlapping, 1-m2 plots (Fig. 1). In each 1-m2 plot, we measured the cover of adults and number of seedlings of the C. remota population. Adults were defined as those possessing reproductive structures. Seedlings were defined as new individuals from seeds that germinated in the previous spring and having a small shoot with up to 3 or 4 leaves. The study area includes a segment of a first-order stream and two moist zones. We also measured several abiotic variables quantified throughout the study area (Table 1). Different grids were used for sampling those variables, according to their heterogeneity, although we averaged or interpolated values to define all
MATERIALS AND METHODS Study site and species—The study was carried out in the Suspiro watershed, Señorío de Bertiz Natural Park, northern Spain. The area is on a north-oriented slope, 350 m above sea level. The climate is oceanic, with a mean annual precipitation of 1600 mm and a drier period in summer, although water runs almost continuously throughout the year. Temperatures range from 7.2°C in the coldest month, January, to 22.4°C in the warmest, July, with an annual mean of 13.4°C (Gobierno de Navarra, 2009; period 1992–2009). Parent rock materials are predominantly schists (Gobierno de Navarra, 1995). Vegetation consists of an oldgrowth deciduous forest dominated by beech (Fagus sylvatica L.), accompanied by some scattered species including pedunculate oak (Quercus robur L.) and sessile oak [Quercus petraea (Matt.) Liebl.].
Fig. 1. Study area with 256 nonoverlapping 1-m2 plots in Señorío de Bertiz Natural Park (Spain). In each 1-m2 plot, cover of Carex remota adults is represented by black circles, and number of seedlings is represented by white numbers. Soil moisture values are also gray.
1288
[Vol. 101
AMERICAN JOURNAL OF BOTANY
TABLE 1.
Summary of the environmental variables quantified throughout the study area in Señorío de Bertiz Natural Park (Spain) and used for the structural equation model.
Variable
Mean
SD
Range
Soil moisture (vol. H2O/vol. soil) Leaf litter cover (%) Relative light (W/m2) Topographic position (m)
30.25 52.07 0.264 3.30
20.69 25.05 0.041 1.40
5.5–82.9 0–100 0.182–0.355 0–6.4
variables in each of the 256 plots. Top soil moisture was measured in a regular grid of 1024 sampling points due to its high variability. Measures were taken twice, in August 2009 and February 2010, with a soil moisture sensor SM200 (Delta-T device, Cambridge, UK). We estimated the cover and the depth of the leaf litter layer in each 1 m2 plot. Solar radiation indices were quantified using hemispherical photography at 64 points in a regular grid. Photographs were taken at 80–100 cm above ground with a digital camera (EOS 50D, Canon, Tokyo, Japan) equipped with a fish-eye lens (4.5 mm F2.8 EX DC, Sigma, Tokyo, Japan), with a 180° field of view. Images were analyzed using Hemiview canopy analysis software version 2.1 (Delta-T Devices, Cambridge, UK). We calculated the indirect, direct, and global site factor (ISF, DSF, and GSF), which are the proportions of indirect, direct and global solar radiation, respectively, that reach a given location over a year. In 256 points, we also measured the height, relative to the lowest point in the area, to create a microtopographic map of the study area. Statistical analyses and hypothetical model development—We used structural equation modeling (SEM; Bollen, 1989) to determine the influence that abiotic and biotic conditions have on the distribution of the C. remota population. SEM is a statistical method that provides a direct linkage between the hypothesis of ideas and statistical testing and has been widely used in ecology during the last years (Grace, 2006). In addition, SEM allows testing direct and indirect (i.e., through the modification of a third factor) relationships among the variables studied. To discover which factors influence the distribution of the C. remota population, we developed a hypothetical model based on previous knowledge of the system and ecological theory. In our model, we considered abiotic variables that have been found to have a significant influence on C. remota distribution, such as soil moisture, and others that can have potential influences on the distribution of nemoral species such as leaf-litter coverage and topographic position. Similarly, and because we were interested in discovering the factors influencing C. remota establishment, we considered light conditions as an additional abiotic factor. Our hypothetical model included direct and indirect relationships among the variables listed above (Fig. 2). We hypothesize that the distribution of C. remota adult individuals can be directly and positively influenced by the soil moisture conditions, since this species strongly depends on water availability. Similarly,
Fig. 2. Hypothetical structural equation model established to study the direct and indirect influence of the different abiotic and biotic conditions on the distribution of adults and seedlings of Carex remota. Singleheaded arrows indicate direct causal relationships among variables; double-headed arrows indicate nonstudied relationships. Hypothesized positive relationships are indicated with solid lines; negative ones are indicated with dashed lines.
we expect an indirect negative influence of topographic position (relative height) on the distribution of C. remota adults, since soil water retention capacity is influenced by the topographic conditions. We also expect that the distribution of C. remota adults can modify the abiotic conditions, mainly by its negative influence on leaf-litter coverage. Regarding the distribution of C. remota seedlings, we expect that they will be directly and positively influenced by the soil water content. Similarly, we expect a direct positive influence of light conditions and negative of leaf-litter coverage on seedling distribution due to the importance of these two factors on seed germination and seedling establishment. Finally, we expect that seedling distribution can be directly influenced by the distribution of adult individuals as a consequence of propagule pressure and indirectly through the influence of adult individuals on leaf-litter coverage. Structural equation models were calculated using the IBM-SPSS AMOS version 19 (IBM, Armonk, New York, USA) statistical software. Before the analyses, we studied whether the different variables were normally distributed by visual and numerical analyses of their skewness and kurtosis. Variables with a significant departure from normality were log-transformed [x′ = log (x + 1)]. Specifically, we transformed cover of C. remota adults and number of C. remota seedlings. Models were estimated using maximum likelihood procedures. The χ2 statistic and its associated probability (p) were used to evaluate the overall model fit. Values of p > 0.05 indicate a statistically acceptable model fit (Grace, 2006). Nevertheless, the test of SEM validity is a process that combines statistical testing and theoretical justification (Grace et al., 2010). In case the proposed model failed to match the system under study, we considered the removal and addition of pathways taking into account the absolute model fit, a theoretical justification, and the principle of parsimony. Specifically, we studied the removal of unimportant relationships and the inclusion of novel causal relationships that will improve the model fit while keeping the biological meaning of the system. The procedure was stopped when a model yielded satisfactory statistical results (χ2 associated p > 0.05). A critical issue is that the spatial distribution of biological data can be influenced by stochastic processes that can create spatial autocorrelation and violate the assumptions of parametric statistics (Peres-Neto and Legendre, 2010). The influence of the spatial autocorrelation can have important implications on the calculation of model parameters and the residuals (Dormann et al., 2007; Beguería and Pueyo, 2009). To control for the influence of spatial autocorrelation, we studied the presence of spatial autocorrelation in the model residuals of the two main variables: the adult abundance and the seedling number of C. remota. We used Mantel tests (Legendre and Legendre, 2012) to test whether the distance between pairs of residual values was correlated with geographic distance. These analyses were performed using the R statistical software (R Development Core Team, 2014) using the vegan library (2.0-7; Oksanen et al., 2008).
RESULTS A total of 275 C. remota adult individuals and 1619 seedlings were found in the area studied. Sixty-six of the 256 plots were occupied by both adults and seedlings, 19 only by adults, and 39 only by seedlings. The spatial distribution of topsoil moisture is shown in Fig. 1 and its values summarized in Table 1. Similarly, the measurement of leaf-litter coverage, topography and light conditions indicated variation in these two variables (Table 1). The calculation of the hypothesized SEM yielded nonsatisfactory statistics (χ2 = 21.20; df = 6; p = 0.002). Therefore, we considered the inclusion of different direct and indirect relationships among the variables to improve model fit while considering the biological meaning of the potential relationships. First, we removed the relationship between topographic position and leaf-litter coverage because it was unimportant (r = 0.008; p = 0.88). Since the results still yielded nonsatisfactory results (χ2 = 21.20; df = 7; p = 0.003), we developed a new model including a causal relationship between light and soil moisture (i.e., indicating that light conditions can modify soil moisture). The inclusion of this pathway increased the model fit substantially, but the results were still not satisfactory (χ2 = 15.53; df = 6; p = 0.016). Thus, we included a new causal relationship between topographic position and C. remota seedlings (Fig. 3). The new SEM yielded satisfactory statistical results
August 2014]
URIA-DIEZ—DRIVERS OF A RIPARIAN FOREST SPECIALIST
1289
Fig. 3. Structural equation model studying the direct and indirect influence of the different abiotic and biotic conditions on the distribution of adults and seedlings of Carex remota (compare with the model in Fig. 2). Single-headed arrows indicate direct causal relationships among variables; double-headed arrows indicate nonstudied relationships. Hypothesized positive relationships are indicated with solid lines; negative ones are indicated with dashed lines. The coefficients associated with the different relationships are listed in Table 2. Lines are proportional to the standardized effect of one variable on each other. The R2 values on top of each box represent the fraction of variance explained of each variable.
(χ2 = 8.27; df = 5; p = 0.123) and kept the biological meaning of the system. The residuals of the model showed no evidence of spatial autocorrelation as C. remota adult abundance residuals (r = 0.013; p = 0.27) and C. remota seedlings residuals (r = 0.001; p = 0.46) were not related to geographical distance. As was suggested in our model, soil moisture content has a positive direct influence on C. remota adult distribution (Fig. 4; Table 2). The model also shows an indirect positive influence of topographic position on C. remota adult distribution, since topographic position had a positive influence on soil moisture. Seedling distribution was the result of multiple complex effects of most of the variables considered, as was suggested in our model (Fig. 3; Table 3). In line with our expectations, C. remota seedlings were positively influenced by soil moisture, the distribution of adult individuals and light conditions. Similarly, seedling distribution was negatively influenced by leaf-litter coverage as we hypothesized in our hypothetical model (Fig. 2). The proposed model accounted for 64% of the variation in C. remota adult distribution and 37% of the variation in the distribution of seedlings (Fig. 5). DISCUSSION Our results demonstrate that the distribution patterns of the forest specialist Carex remota partially depend on direct and indirect effects of abiotic conditions. Our findings confirm that seedling establishment, a crucial phase in the life cycle of the plant, is strongly dependent on multiple abiotic conditions. As hypothesized, the number of seedlings increases in favorable microsites characterized by high topsoil moisture, low leaf-litter values, and high light values. These results are in accordance with our previous knowledge about the germination of forest specialist herbs. Furthermore, our results highlight the importance of the presence of adult individuals and their size in determining seedling success. Larger individuals ensure seed production favoring seed rain in their surroundings. In addition,
Fig. 4. Linear relationship between the distribution of adult individuals of Carex remota and soil moisture content in the Bertiz forest. The scatter diagrams show the relationship between percentage-scale cover of adult individuals (log-transformed) and soil moisture. Zeroes have been removed from the graphs. The red line indicates the linear relationship between the two variables considering zeroes.
these individuals can modify the microenvironmental conditions favoring the establishment of more seedlings. Finally, the results show that adult individuals are mainly determined by topsoil moisture levels indicating that the establishment of seedlings is the crucial stage for population development. A substantial fraction of variation remains unexplained and points to a possible impact of factors not considered in this study, such as past events and genetic population structure on the distribution of C. remota. Overall, our study suggests that abrupt changes in moisture and light conditions during the germination phase or the removal of adult individuals may influence the establishment of new C. remota individuals triggering changes seedling recruitment and adult survival. The availability of safe sites for germination may be considered a key aspect for the performance of C. remota populations. The abundance of C. remota strongly depends on water availability, as the cover of adults increases with the increase of topsoil water content, which is partially determined by the topographic heterogeneity. In addition, the results of our model confirm that C. remota is shade tolerant since the distribution of adult individuals was independent of light conditions. Furthermore, C. remota is a specialized sedge that inhabits constantly moist or damp, water saturated, and badly aerated sites (Grime et al., 2007; Uria-Diez et al., 2013) and is able to tolerate the deep shade of forest interiors. In contrast to the results obtained for adult individuals, the distribution patterns of the seedlings of C. remota depend on multiple direct and indirect abiotic conditions, as well as their positive interaction with adult conspecifics. These differences between individuals at different stages
1290
[Vol. 101
AMERICAN JOURNAL OF BOTANY
TABLE 2.
Path coefficients for the structural equation model of Carex remota. For each of the paths proposed in Fig. 3, the estimated coefficient, standard error, its standardized estimate, and associated P value are shown.
Path
Estimate
SE
Standardized estimate
P
Topographic position → Soil moisture Topographic position → C. remota seedlings Relative light → Soil moisture Relative light → C. remota seedlings Soil moisture → C. remota adult Soil moisture → C. remota seedlings C. remota adult → leaf-litter coverage C. remota adult → C. remota seedlings leaf-litter coverage → C. remota seedlings Topographic position ←→ Relative light
2.41 −0.062 −65.508 3.422 0.038 0.011 −11.529 0.183 −0.002 0.031
0.779 0.024 27.376 0.823 0.002 0.003 1.986 0.065 0.001 0.004
0.223 −0.161 −2.393 0.252 0.798 0.314 −0.342 0.243 −0.110 0.528
0.002 0.008 0.017
