Plant functional trait data and reflectance spectra for 22 palmiet wetland species

We provide reflectance spectra for 22 South African palmiet wetland species collected in spring 2015 from three wetlands throughout the Cape Floristic Region. In addition, we provide summarized plant functional trait data, as well as supporting and meta-data. Reflectance spectra were collected with a portable ASD Fieldspec Pro using standard methods. The 14 plant functional traits were measured on 10 replicates of each species, following standard protocols. We provide tables detailing these standard methods, as well a table with hypotheses on how these 14 continuous traits, as well as an additional 9 categorical traits, may affect ecosystem service provision. In addition, tables are attached which detail which functional and spectral groups these species belong to, according to the data. Finally, we include a photographic plate of the species data are provide for. We make these data available in an effort to assist in research on the understanding of how traits affect ecosystem service provision in wetlands, and particularly of whether remote sensing can be used to map these traits in wetlands.


a b s t r a c t
We provide reflectance spectra for 22 South African palmiet wetland species collected in spring 2015 from three wetlands throughout the Cape Floristic Region. In addition, we provide summarized plant functional trait data, as well as supporting and meta-data. Reflectance spectra were collected with a portable ASD Fieldspec Pro using standard methods. The 14 plant functional traits were measured on 10 replicates of each species, following standard protocols. We provide tables detailing these standard methods, as well a table with hypotheses on how these 14 continuous traits, as well as an additional 9 categorical traits, may affect ecosystem service provision. In addition, tables are attached which detail which functional and spectral groups these species belong to, according to the data. Finally, we include a photographic plate of the species data are provide for. We make these data available in an effort to assist in research on the understanding of how traits affect ecosystem service provision in wetlands, and particularly of whether remote sensing can be used to map these traits in wetlands.

Value of the data
The reflectance spectra could be used to form spectral libraries for these South African wetland species, and used in future hyperspectral remote sensing exercises (e.g. spectral unmixing).
These spectra could additionally be used with other traits collected for these species to take the analysis further.
The trait summary data could be used to augment meta-analysis; or international wetland studies.

Data
The dataset of this article provides reflectance spectra for wetland species as well as associated plant functional trait data [1]. The raw reflectance spectra for the 22 palmiet wetland species are included as an excel file (Appendix A). Meta-data about these measurements can be found in Table 1. Hypotheses about how each of the plant functional traits measured in this study may relate to ecosystem services is shown in Table 2. Table 3 gives details about the measurement (standard protocol) relating to each of the plant functional traits measured. Table 4 gives a summary of the data for each trait (for all 22 species). Tables 5 and 6 give additional output from analyses; the former simple regression analyses, the latter with partial least squares regression (PLSR). We performed PLSR using the 'pls' package [2] and 'autopls' code [3] in R to determine which PFTs could be predicted from the reflectance spectra. Table 7 details  functional groupings of the 22 species and average trait values per group, whereas Table 8 does the same, but for spectral groups. Fig. 1 shows pictures of each of the 22 species.

Experimental design, materials, and methods
These data form part of the Supplementary material of a publication in Remote Sensing of Environment [1]. Relevant sections from the methods have been extracted from this publication. Table 2 Hypotheses of how the selected plant functional traits would be expected to link to Ecosystem Service provision (based on expert opinion). ↑ symbolizes a possible positive correlation, ↓ a negative correlation,a non-directional relationship, andsignifies no relationship. Italicized traits are categorical.  Table 1 Species list of the 22 dominant plant species in South African palmiet wetlands and the wetlands they were recorded as being dominant in (from data recorded in plots) as well as the wetland the specimens for the reflectance measurements were collected from. Letters correspond to the photographs in Plate S1.  Table 3 The 23 functional traits collected for the 22 species used in this study. All methods were based on the standardised protocol of Pérez-Harguindeguy et al. [4]. For categorical traits the codes assigned are shown in brackets.  (3), Annual (4), Tuft (tussock) (5), Rhizome (6), Stolon (7), Suffrutex

Study design
Species composition data were obtained from 39 plots in the three different palmiet wetlands. Plots were arranged on seven transects (100-200 m) along cross sections through the wetlands, with six plots (3 Â 3 m) placed between 20-50 m apart, yielding a total of 36 plots. In the Goukou wetland, three extra plots were added to fully capture variation in plant communities. Species and their relative abundances were recorded in each plot, using the Braun-Blanquet Scale [5]. Dominant species were defined as those making up more than 25% cover in any plot. The resultant 22 species are listed in Table 1, Fig. 1. Ten mature specimens from each dominant species were collected from their wetland of origin for measurement of PFTs at the respective field station or in the lab (depending on the trait). Traits were collected once for each species from random specimens in the field (maximum abundance approach, Carmona et al. [6]). Extra specimens were collected from one of the three sites for each species (Table 1).

Plant functional traits
We measured 23 PFTs, each selected as they were predicted to have a link to at least one wetland ecosystem service (Table 2). Definitions and methods for the measurements of each PFT are given in Table 3; and for all commonly used PFTs we used the standardized protocol for measurements [7]. Of the PFTs measured, 16 were morphological/anatomical, and seven were biochemical in nature (Table 3). For biochemical traits, samples were cleaned, dried at 70°C for 48 h, ground and homogenised using a mill to 0.5 mm particles. Total carbon and total nitrogen were determined by total  Table 5 The relationship between average reflectance over the four averaged sections of the spectrum and plant functional traits for five key traits. Both variables (average reflectance) and the plant functional trait were logged(10) in each regression.  Table 7 Functional groups of 22 dominant South African wetland species based on cluster analysis with 23 functional traits. The top 10 predictors (traits) driving the separation of groups are shown as average values per functional group. The numbers in brackets indicate the importance of each predictor in driving the grouping. For categorical traits the number given is not an average but the mode (most common form of the trait). Corresponding categories for these codes can be found in Table 3.

Species Functional Group
Cellulose  combustion of 5 mg of each sample on a Flash 2000 CN-analyzer (Thermo Fisher Scientific). To determine plant silicon content, we used a procedure for extracting biogenic silica (Schoelynck et al. 2010), which involved incubating a 25 mg sample of dried plant material in a 0.1 m Na 2 CO 3 mixture which was placed in a water bath at 80°C for 4 h. This dissolved biogenic silica was then spectrophotometrically analysed on a Thermo IRIS inductively coupled plasmaspectrophotometer  (ICP; Thermo Fisher, Franklin, MA, USA). Plant lignin and cellulose content were measured using the Van Soest method [8]. Summary statistics are shown for each of the continuous PFTs in Table 4.

Reflectance measurements
Plant canopy spectra were measured in the field in November 2015 (spring) under clear sky conditions within two hours of local solar noon. Phenology has been shown to be valuable in discriminating wetland species (e.g. reed beds) and spring is the season in which interspecific phenological distinctions are generally at their greatest [9,10]. All reflectance measurements were taken with a portable ASD Fieldspec Pro (ASD Inc., Boulder, USA). The probe was held at a constant distance of 60 cm above the surface (25°FOV; diameter 26.59 cm), keeping the sensor perpendicular to the angle of the sun. Live (wet) specimens from each species were arranged on a large matt black (non-reflective: uniform o 5% reflectance across the 350-2500 nm range) surface (1.5 Â 2 m), with leaves facing upwards (adaxial surface up) where possible. This measurement set-up allowed us to measure the reflectance of individual plant species without background contamination originating from soil or other plant species. This set-up thus allowed us to make a one-on-one comparison between reflectance and PFTs. It is acknowledged that the spectral effects of 3D canopy structure (i.e. volume scattering effects) were not fully captured with this set-up. Since this study focussed primarily on leaf traits, this is not expected to present any problems.
Twenty spectral signatures were collected for each species. There were two cases where data had to be excluded due to equipment problems (see Table 1 for details). Between readings for each species, the ASD was optimised using a spectralon (Spectralon s , Labsphere, North Sutton, USA) and white reference measurements were captured. Spectra were collected over the range of 350-2500 nm with 1 nm intervals. ASD binary files were first converted to ASCII reflectance files using ViewSpecPro and subsequently post-processed to remove data in the water absorption bands at 1350-1460 nm and 1790-2000 nm as well as noise at 2350-2500 nm.