Edaphic and light conditions of sympatric plant morphotypes in western Amazonia

Abstract Here I present a dataset of edaphic and light conditions associated with the occurrence of sympatric morphotypes of Geonoma macrostachys (Arecaceae/Palmae), a candidate case study from Amazonia hypothesized to have evolved under ecological speciation. Transects were established in three lowland rainforests in Peru, and the abundance of each local morphotype of this species was recorded in a total area of 4.95 hectares. Composite soil samples and hemispherical photographs were taken along the transects were the species occurred to obtain information on soil nutrients, soil texture, and indirect measurements of light availability. The raw and summary tables disclose the characteristics of each study site and habitats within them, which could be useful to soil scientists, ecologists, and conservationists engaged in similar research activities or meta-analyses in Amazonia.


Introduction
It is well known that soil chemistry, texture, and topography can determine the plant community composition and species richness at different spatial scales (e.g. Gentry 1981, Eiserhardt et al. 2011. For example, the turnover of community species composition along a soil fertility gradient has been documented at local and regional scales (e.g. Poulsen et al. 2006, Andersen et al. 2010, Guèze et al. 2013. Plant species grow preferentially under different soil nutrient concentrations and textures (e.g. John et al. 2007, Baribault et al. 2012). Flooding versus good drainage also affects plant distribution (e.g. Silvertown et al. 1999, Duque et al. 2002. Soil texture is related to drainage, and it characterizes the bulk density, surface area, and air space in between soil particles, affecting the water-holding capacity and hydraulic conductivity of soils (Rawls et al. 1982, Sollins 1998, Palm et al. 2007). Topography also influences species distributions through its interaction with other environmental factors such as soil nutrients, hydrology, wind exposure, temperature and even biotic factors (Trichon 1997, Pausas and Austin 2001, Klinger and Rejmánek 2010. Its effect on plant performance is thus indirect, difficult to interpret and often site specific (Vormisto et al. 2004). Although less studied, the distributions of many plant species show strong associations with light availability (e.g. Terborgh and Mathews 1999). The vertical distribution of foliage in a forest allows light to penetrate the understory through vertical and lateral gaps of different sizes, creating a vertical and horizontal light heterogeneity in the forest understory (Oberbauer et al. 1989, Montgomery 2004) that could allow resource partitioning among species. These plant responses to abiotic conditions suggest an important role for habitat heterogeneity not only as a mechanism that facilitates the coexistance of high species diversity, but also as a speciation driver (e.g. Gentry 1989, Haffer 1997, Nosil 2012. Documentation of habitat heterogeneity should thus be an important component in biodiversity studies. Nosil (2012) defined ecological speciation as the process by which barriers to gene flow evolve between populations as a result of ecologically based divergent selection between environments. The interaction of individuals with their environment is thus a key agent of selection under this mode of speciation, making the documentation of habitat preferences between populations an important observation (yet not the only one) to empirically distinguish ecological speciation. The palm species complex, Geonoma macrostachys Mart. (Arecaceae), is a potential case study of ecological speciation in western Amazonia. Local morphotypes of this lowland forest palm differ in leaf shape, show a strong habitat differentiation, are reproductively isolated by differences in pollinator guild and flower phenology while genetic data suggest an independent evolution of the morphotypes in each forest site (Listabarth 1993, Roncal 2005, Roncal 2006, Roncal et al. 2007).
Here, I present a dataset of edaphic and light properties that were used to determine the presence and degree of habitat differentiation between local morphotypes of G. macrostachys in three lowland moist forests in Peru (Roncal 2005, Roncal 2006). These publications did not make the raw data available. Following Svenning (1999), I define habitat as the environmental conditions occurring at the scale of a floodplain or terra firme (i.e. more than one km ). I refer to microhabitat as those characteristics within major habitat types that change at scales less than 10 m (Svenning 1999). This information could complement similar environmental studies spanning the distribution range of this palm species in order to test more rigorously the ecological speciation hypothesis in Amazonian plants. Finally, the environmental data available here could be useful to soil scientists, ecologists, and conservationists who seek detailed environmental information at the habitat and microhabitat scales for this part of the Amazon basin.

Project description
Title: Habitat differentiation of sympatric Geonoma macrostachys (Arecaceae) morphotypes in Peruvian lowland forests Personel: Julissa Roncal Study area description: Fieldwork was carried out at three sites. The Amazon Conservatory of Tropical Studies (ACTS) is situated adjacent to the Sucusari, a small tributary to the Napo River in northeast Peru. ACTS is located within the Explornapo Reserve, a 1,725 ha of mostly primary forest, property of Explorama Tours (Vasquez 1997). Soils in the reserve belong to the Pebas formation, which dates back to the Middle Miocene (Hoorn 1994), and gave rise to clay and silty clay soils with a higher than average nutrient content (Vasquez 1997, Vormisto et al. 2004). Most of the reserve is covered by terra firme forest but the area adjacent to the Sucusari was classified as Igapo or floodplain. For a detailed description of the floristic composition of the area see (Vasquez 1997). The Loma Linda Native Reserve (LLNR) is a 332.16 ha protected area located adjacent to the Palcazu River in central Peru. No information on the geology or soil type of the reserve has been published. Two main habitat types were visually recognized in the field: a topographically irregular red-soil habitat, and a flat white-soil habitat. Finally, the 1,000 ha study area of Cocha Cashu biological station (EBCC) is located within the lowlands of the 1,532,000 ha of Manu National Park in southeastern Peru (Terborgh 1990). Soils at EBCC within the 6 km-wide meander belt of the Manu River (floodplain forest) are composed of young alluvial silt and clay carried from the Andes. Soils in the uplands (terra firme) of EBCC, dissected by numerous streams, are sandy (Terborgh 1990). Foster (1990) described the floristic composition of the Manu river floodplain forests.

Sampling methods
Sampling description: At each site, transects of 10 m wide and 290 m long were established on each main habitat described in the 'study site' section, and separated from one another by at least 200 m. Eleven, twelve, and fourteen transects were established at EBCC, LLNR, and ACTS, respectively. Transects were divided into plots of 10 m × 10 m and all G. macrostachys adult individuals having the minimum reproductive height were recorded in every other plot to avoid spatial autocorrelation (Suppl. material 1). The position of transects are disclosed in Table 2. The total area sampled in this study was 4.95 hectares. A map of the trail system at ACTS can be found in Suppl. material 2, and a LANDSAT map, as well as the trail system at EBCC can be found in http:// cochacashu.sandiegozooglobal.org/researchers/maps/. Map of the three study sites in Peru where soil and light conditions were measured. Locality acronyms are the same as in Table 1.  The inclination of every other plot along each transect was measured with a clinometer (PM5/360PC, Suunto®, Finland) in the middle of the plot. Soil samples for laboratory analyses were taken from 78, 76, and 87 plots from ACTS, LLNR, and EBCC, respectively (241 soil samples in total). Plots were randomly chosen along transects so that at least 40 soil samples per morphotype at each site were collected with no more than nine soil samples per transect. Since at EBCC fewer than 40 plots were recorded to have the acaulis morphotype, 17 additional soil samples were collected from haphazard acaulis individuals in the forest. For the same reason, nine soil samples from haphazardly chosen large morphotype individuals were collected at LLNR. At each plot, the top 20 cm of soil profile (Ah horizon) was sampled at three points within a 0.5 m radius of the palm(s), using a 2.5 cm diameter × 30 cm high metallic cylinder, and mixed to obtain a composite soil sample. This procedure was also followed for plots where the two varieties were found, collecting only one composite sample.
Soil texture was quantified using a hydrometer, which calculates the proportional distribution of sand (particle size of 0.05 mm and larger), silt (0.002-0.05 mm) and clay (<0.002 mm) in the soil through the application of the Stoke's law of mineral particle separation by size, based on the settling rate in suspension (Thein and Graveel 2002). Soils were further assigned to one of the 12 textural classes using the United States Department of Agriculture (USDA) textural triangle (Thein and Graveel 2002). Soil chemical analyses included pH using an electrode in a 1:1 solution of soil and water, and the following extractable cations: Ca, Mg, P, K, Zn, Mn, Cu, B, and Na, using the Mehlich 1 extractant and an Inductively Coupled Plasma (TJA 61E, Thermo Electron Corporation, Florida). These analyses were conducted at the Agricultural Service Laboratory of Clemson University. Suppl. material 3 presents the raw data. Table 3 is a summary table showing mean values and standard deviations for each main habitat within the study sites. Table 4 is another summary showing only the significantly different edaphic variables between morphotypes. Soil textural classes were also different between habitats at each site (Fig.  2). Clay and clay loam soils characterize the floodplain of EBCC and ACTS, while sandy soils characterize the terra firme at these sites. The white soil habitat at LLNR presents sand, loamy sand, and sandy loam, while the red soil habitat is mostly composed of sandy clay loam, clay loam and clay Fig. 2   Hemispherical photographs were used to obtain an indirect measure of light availability for 40 palm individuals of each morphotype at each study site. Hemispherical photography is a technique used to estimate forest light conditions in the subcanopy and understory since light measurements obtained from this method correlated highly with direct measurements of photosynthetic photon flux density (Chazdon and Field 1987, Roxburgh and Kelly 1995, Machado and Reich 1999, Engelbrecht and Herz 2001. Individuals selected for this purpose were the same as those selected for soil analyses. I used a Nikon 8 mm fisheye lens (180°field of view) mounted on a Nikon COOLPIX 995 digital camera. Photographs were taken under uniformly overcast conditions (usually at dawn) to avoid reflection. The camera was oriented with a hand-held compass to ensure that a light emitting diode attached to the fisheye lens pointed the north, the camera was also leveled in a tripod Distribution of soil textural classes at the three study sites following the USDA textural triangle system (Thein and Graveel 2002). Plots sampled from the main habitat types are distinguished on each case. Data used for these figures were obtained from Suppl. material 3. before each photograph. Hemispheric photographs were analyzed with Gap Light Analyzer (GLA) software version 2.0 (Frazer et al. 1999, http://www.rem.sfu.ca/forestry/gla/), which calculates the proportions of direct and diffuse radiation beneath the canopy relative to those above the canopy. The output of GLA includes the following light variables (definitions taken from software manual, Frazer et al. 1999): "Percentage of canopy openness is the percentage of open sky seen from beneath a forest canopy. This measure is computed from the hemispherical photograph only, and does not take into account the influence of the surrounding topography" "Leaf area index 4Ring is the effective leaf area index integrated over the zenith angles 0 to 60°" "Leaf area index 5Ring is the effective leaf area index integrated over the zenith angles 0 to 75°" "Transmitted direct is the amount of direct solar radiation transmitted by the canopy in mol m d " "Transmitted diffuse is the amount of diffuse solar radiation transmitted by the canopy in mol m d " "Transmitted total is the sum of transmitted direct and transmitted diffuse" "Percentage transmitted direct is the ratio of transmitted direct to above direct mask (defined as the amount of direct radiation incident on a horizontal or tilted surface) multiplied by 100%" "Percentage transmitted diffuse is the ratio of transmitted diffuse to above diffuse mask (defined as the amount of diffuse radiation incident on a horizontal or tilted surface) multiplied by 100%" "Percentage transmitted total is the ratio transmitted total to above total mask (defined as the sum of above direct mask and above diffuse mask) multiplied by 100%" Photographs were analyzed twice so that threshold values were averaged before running the program. To document the light environment of the forest, 40 photographs were taken at random points on each habitat type at each site, these represent the control points in Suppl. material 4. Random numbers were used to select the location along the trail systems and the camera was located at the average G. macrostachys crown height (approximately 90 cm). Control points were not taken at LLNR since the lack of a trail system made this task impractical. Suppl. material 4 presents the raw data, while Table 5