Functional diversity quantifies the diversity of species’ traits in biological communities, and this metric is regarded as key to understanding ecosystem processes as well as responses to environmental stress or disturbance (Laliberté & Legendre, 2010). Higher functional diversity signifies greater differences among species’ trait values, more distinct ecological functions, and thus potentially enhanced functional stability against perturbations caused by anthropogenic or environmental stresses (Cardinale et al., 2012). The ability of plants to respond phenotypically to changing environmental conditions provides a reliable tool for the assessment of environment-induced changes in species community composition. Environmental factors can determine which species within the regional pool can thrive in a given community, based on their suitability to local environmental conditions (Cornwell et al., 2009). Previous studies have suggested that directional changes in functional diversity along environmental gradients can be interpreted as evidence of habitat-filtering processes in trait space (e.g., Rolo et al., 2016; Laughlin et al., 2018; Mitchell et al., 2018). However, once species have passed through environmental filters, biotic interactions may also shape the functional structure of communities (Bello et al., 2015). Community traits may converge towards certain values if highly competitive species tend to displace weak competitors (Mayfield & Levine, 2010). On the other hand, functional diversity may increase if minimizing trait similarity is the dominant process. For example, the species regeneration of early successional stages progressed from functionally narrow (i.e., with similar trait values; trait convergence) towards an increasingly wide functional trait space at late successional stages (i.e., trait divergence; Buzzard et al., 2016). Therefore, communities are expected to be shaped by the joint effects of abiotic filtering and biotic interactions.
Limestone hills are elements of karst landscapes, which are characterized by distinctive topography that develops as water dissolves soluble calcareous rock. Tropical forests on limestone hills are viewed as relict communities, with many endemic and rare species (Lynch & Zomlefer, 2016). In Thailand, many studies have demonstrated (e.g., Latinne et al., 2013; Phutthai & Hughes, 2017) that limestone hills provide a critical habitat for endemic and rare species of plants and animals. However, natural disturbances and human activities have complex negative effects on the biodiversity of these limestone hill ecosystems. Therefore, limestone forest restoration is critical, and it is necessary to consider the environmental factors of limestone hill forests when selecting suitable species. Despite such drastic threats to this unique ecosystem, the ecological processes of forest growth on limestone hills in Thailand remain poorly understood (Clark et al., 2014). Therefore, the goal of the present study is to assess the responses of functional traits to environmental conditions that may considerably differ with respect to elevation. Our primary objective was to elucidate key ecological processes to inform necessary improvements in limestone forest management. We focused on determining the environmental factors that most strongly affect tree functional traits in limestone hill forests of Thailand.
The study area is located on a limestone hill in Phrae Province, northern Thailand ca. 550 km from Bangkok (17°70′–18°84′N, 99°58′–100°32′E) (Figure 1). The site spans elevations (Elv) of 320–460 m a.s.l. The mean annual temperature and rainfall are 32 °C and 189.6 mm, respectively. The region experiences two main seasons: (i) the wet season (May–October; mean rainfall and temperature of 159 mm and 32.6 °C, respectively) and (ii) the dry season (November–April; mean rainfall and temperature of 23 mm and 25.6 °C, respectively). The dry season is subdivided into cool-dry (November–January) and hot-dry (February–April) sub-seasons (Meteorological Department, 2019).
Field measurements were conducted from October 2015 until December 2016. Three forest sites along an altitudinal gradient on a limestone hill were selected for study thereafter referred as: a lower forest site (LFS), a middle forest site (MFS), and an upper forest site (UFS). The forest sites were differentiated by the Elv and geomorphology along the limestone hill (Table 1). We established a 20 × 20 m permanent plot in each forest site, thereby generating a total of 54 plots for a total area of 2.16 ha (Table 1). We measured the diameters at breast height (DBH at 1.30 m above ground level) for all trees with DBH ≥ 4.5 cm in each established plot. All trees were tagged and identified to the species level. In addition, leaf specimens were collected to verify species identification by comparing them to specimens in the herbarium at the Department of National Park, Wildlife and Plant Conservation. Nomenclature followed the Flora of Thailand (Smitinand, 2014).
Study area map depicting the three forest site plot numbers. Site selection was based on elevation and the geomorphological features of the hill.
Forest site | Descriptions | Number of plots |
---|---|---|
Lower forest site (LFS) | A forest community located at the base of the limestone hill, with rocky outcroppings covering >50% of the area. The site spanned elevations of 320–360 m a.s.l. | 18 |
Middle forest site (MFS) | A forest stand located on a long narrow ridge on the high, steep slope of the limestone hill. The site spanned elevations of 370–400 m a.s.l. | 22 |
Upper forest site (UFS) | A forest stand established on a rocky outcropping located at the summit of the limestone hill. The site spanned elevations of 410–460 m a.s.l. | 14 |
In addition to elevation we also estimated the percentage of rocky outcropping cover (%RC), percentages of nitrogen (%N), phosphorus (%P), and potassium (%K) in the soil and measured photosynthetically active radiation (PAR), and soil moisture content (SMC) for each of the 54 plots.
To measure SMC and soil nutrients (N, P, K), we collected 100 cm3 soil samples from the topsoil layer (0–15 cm) in October 2016 using a soil core sampler by taking sub-samples from the center and at each of the four corners of each plot (five points per plot). Two sets of soil samples per plot were collected: the first for analysis of SMC (%), determined as the ratio of fresh weight to dry weight, and the second for soil nutrient analyses (%N, %P, and %K). Measurements of N, P and K were determined using the Kjeldahl, Blay II, and Flame Photometric methods, respectively, and analyses were conducted at the soil laboratory of Chiang Mai University. All samples were collected on the same day, 10 days after the last rainfall and close to the onset of the dry season, when rain was infrequent.
Photosynthetically active radiation (µmol photons m−2 s−1) is a measure of the photon flux within the 400–700 nm spectral band of solar radiation. PAR was measured at the center and at each of the four corners of each plot using a Spectrum 3415FX LightScout External Sensor (Spectrum Technologies, Plainfield, IL, USA). PAR measurements were taken at 1.3 m above ground level on a sunny day (08:00–10:00) in November 2016. The average PAR for each plot was calculated as the arithmetic average of five measurements. The average Elv (m) was measured at the center and at each of the four corners of each plot (five points per plot) using a Sun Company Altimeter 202 (Sun Company, Wheat Ridge, CO, USA). The %RC was estimated by counting the number of grid cells wherein rock appeared, which was obtained by dividing each 20 × 20 m plot into 100 2 × 2 m grid cells. These environmental factor measurements were used to represent environmental conditions in our analyses of woody tree abundance and functional traits at each site.
We selected functional traits that were reflective of different ecological strategies for plants and were known to be sensitive to community assembly processes, following Pérez-Harguindeguy
One individual of each target species was randomly selected per plot to quantify the functional structure of the sampled community following Rolo
We compared environmental factors among sites (LFS, MFS, and UFS) using an analysis of variance (ANOVA). We calculated the means of measurements made within plots and identified significant differences among sites. For each site, we calculated the following: species richness, stem density (stem ha−1), stem basal area (m2 ha−1), species frequency (%, the ratio of the number of plots in which species occurs to the total of all sampling plots), relative stem density (%), relative basal area (%), and relative species frequency (%). From these results, we derived the importance value index (IVI) as the sum of the relative stem densities, relative basal areas, and relative species frequencies for each species in each forest site. We used the IVI to identify the prevalent species in each site. IVI is an excellent indication of the vegetational importance of a species within a site, since it is sensitive to such variables as apparent contagion or exceptional basal area (Curtis & McIntosh, 1951). In addition, the Shannon–Wiener index (
Community-level means were assessed within each plot to analyze the functional trait diversity of woody trees. We calculated community-level weighted mean (CWM) values for single traits and determined functional dispersion (FDis) and functional richness (FRic) values for multivariate traits (Mouchet
To determine which environmental variables (PAR, %RC, Elv, SMC, %N, %P, and %K) were related to functional trait diversity (single and multivariate traits) of trees in each site, we applied generalized linear mixed models, using the lme4 package (Bates
Environmental variables (PAR, %RC, Elv, SMC, %N, %P, and %K) differed significantly among the LFS, MFS, and UFS along the elevational gradient (Table 2). Elv significantly differed among the three stands (
Measurements of the environmental factors at each forest site. Measurements were taken at the three study sites, i.e., lower forest site (LFS), middle forest site (MFS), and upper forest site (UFS), along an altitudinal gradient on a limestone hill. Different superscript letters within a row indicate significantly different means (ANOVA test,
Factors | LFS | MFS | UFS | |
---|---|---|---|---|
Elevation (m) | 73.88 ± 13.53c | 145.90 ± 14.76b | 222 ± 10.80a | 0.0004 |
PAR (μmol photons m−2 s−1) | 37.83 ± 6.78b | 109.17 ± 44.38b | 349 ± 73.81a | 0.0032 |
Rocky outcroppings (%) | 75.33 ± 1.86b | 80.00 ± 1.08b | 87.57 ± 1.11a | 0.0041 |
Soil moisture content (%) | 3.60 ± 0.30a | 2.90 ± 0.24ab | 2.60 ± 0.95b | 0.0025 |
Nitrogen (%) | 60.63 ± 16.11ns | 61.29 ± 6.96ns | 59.62 ± 2.71ns | 0.1241 |
Phosphorus (%) | 0.25 ± 0.01ns | 0.24 ± 0.01ns | 0.29 ± 0.02ns | 0.0723 |
Potassium (%) | 0.42 ± 0.04a | 0.19 ± 0.02b | 0.019 ± 0.02b | 0.0251 |
We identified a total of 21 species of woody trees (Appendix A), from 17 genera and 15 families. The LFS and MFS exhibited higher species richness and diversity than UFS. Both stand basal area (18.9 m2 ha−1) and stem density (2175 stems ha−1) were highest in the UFS, followed by the MFS and LFS (Table 3). The five most important value indices of tree species within each forest site are shown in Table 4.
Ecological characteristics of the three forest sites growing along an altitudinal gradient on a limestone hill in Phrae Province, northern Thailand. The three study sites were lower forest sites (LFS), middle forest sites (MFS), and upper forest sites (UFS).
Ecological characteristics | Forest sites | ||
---|---|---|---|
LFS | MFS | UFS | |
1. Number of species | 21 | 21 | 16 |
2. Species diversity index (H′) | 2.04 | 2.43 | 1.40 |
3. Stand basal area (m2 ha−1) | 13.2 | 15.9 | 18.9 |
4. Stem density (stems ha−1) | 1556 | 1616 | 2175 |
Basal area (Ba, m2 ha−1) and stem density (D, stem ha−1) of the five prevalent species in each of the three forest sites growing along an altitudinal gradient on a limestone hill, ranked with the importance value index (IVI, sum of the relative basal area [RDo, %], relative density [RD, %], and relative frequency [RF, %]) of each species.
Forest site | Species | Ba | D | RDo | RD | RF | IVI |
---|---|---|---|---|---|---|---|
Lower forest site | |||||||
1 | 2.69 | 738.89 | 20.36 | 47.5 | 14.4 | 82.26 | |
2 | 2.11 | 211.11 | 15.96 | 13.57 | 12 | 41.53 | |
3 | 1.36 | 61.11 | 10.27 | 3.93 | 8 | 22.2 | |
4 | 0.95 | 44.44 | 7.2 | 2.86 | 8.8 | 18.86 | |
5 | 1.03 | 61.11 | 7.79 | 3.93 | 4.8 | 16.52 | |
Middle forest site | |||||||
1 | 3.18 | 381.82 | 19.89 | 23.63 | 13.16 | 56.68 | |
2 | 2.02 | 202.27 | 12.64 | 12.52 | 9.87 | 35.02 | |
3 | 1.62 | 200.00 | 10.12 | 12.38 | 10.53 | 33.02 | |
4 | 2.52 | 81.82 | 15.76 | 5.06 | 10.53 | 31.35 | |
5 | 1.37 | 168.18 | 8.59 | 10.41 | 7.89 | 26.89 | |
Upper forest site | |||||||
1 | 10.51 | 1457.14 | 55.49 | 67 | 16.87 | 139.36 | |
2 | 2.78 | 114.29 | 14.69 | 5.25 | 13.25 | 33.2 | |
3 | 1.09 | 107.14 | 5.73 | 4.93 | 10.84 | 21.5 | |
4 | 0.93 | 121.43 | 4.93 | 5.58 | 9.64 | 20.15 | |
5 | 0.66 | 107.14 | 3.47 | 4.93 | 9.64 | 18.03 |
The PCA applied on the functional traits among species explained 63.48% of the total variation. The PCA ordination plot graphically depicted the relationship between tree species and their functional traits across the three forest sites (Figure 2). Several dominant species displayed distinguishing characteristics, such as a high Hmax in
The single traits of woody trees differed in the CWM among forest sites (Table 5). The CWM of LS was higher in the MFS and UFS than in the LFS (
Community-level weighted means (CWM) of leaf size (LS), specific leaf area (SLA), leaf thickness (LT), wood density (WD), maximum plant height (Hmax), functional richness (FRic), and functional dispersion (FDis) of woody tree species measured in the different forest sites, i.e., lower forest site (LFS), middle forest site (MFS), and upper forest site (UFS), growing along an altitudinal gradient on a limestone hill.
Functional trait | LFS | MFS | UFS | Sig. |
---|---|---|---|---|
CWM-LS | 133.80 ± 54.85b | 299.23 ± 52.07a | 253.42 ± 81.26a | ** |
CWM-SLA | 16.34 ± 8.16b | 24.27 ± 11.22a | 14.88 ± 5.84b | ** |
CWM-LT | 0.23 ± 0.02c | 0.26 ± 0.04b | 0.35 ± 0.03a | *** |
CWM-WD | 1.88 ± 0.16a | 1.62 ± 0.39ab | 1.50 ± 0.22b | *** |
CWM-Hmax | 15.84 ± 1.23a | 11.29 ± 21.79ab | 8.32 ± 0.48b | ** |
FRic | 3.16 ± 2.72b | 6.66 ± 3.06a | 5.66 ± 3.98a | ** |
FDis | 1.32 ± 0.32b | 2.12 ± 0.41a | 1.86 ± 0.44a | *** |
Different lowercase letters within a row indicate significantly different means (ANOVA test,
For the multivariate traits, the FDis and FRic values were positively related to Elv (
Selected factors from the generalized linear mixed models evaluating the community functional values (FD) of woody plants relating to the environmental variables.
FD | Environmental variables | ||||||
---|---|---|---|---|---|---|---|
Elv | RC | PAR | SMC | N | P | K | |
FDis | 0.026** | ||||||
FRic | 0.021** | ||||||
CWM | |||||||
LS | 2.302*** | 7.856*** | 1.508*** | −3.137*** | −3.767*** | 2.535*** | −1.705*** |
SLA | 1.372*** | 0.031*** | 0.001*** | −0.002*** | 0.853 | ||
LT | 0.461*** | 1.003*** | 2.014*** | ||||
WD | −1.241*** | −0.127*** | |||||
Hmax | −0.006*** | −0.546*** | 0.151*** |
Only the estimated value of significant factors, selected to minimize Akaike’s information criterion, are shown.
Plant functional traits can drive the growth and establishment potential of species in various forest communities. These traits are often used as proxies to determine whether species utilize different ecological strategies for reproduction and resource capture (McGill
High functional trait diversity (i.e., high variation in functional traits within a community) is expected to result in enhanced community-level plant productivity and resource-use efficiency (Laureto
We found that the CWMs of leaf traits (LS, SLA, and LT) were highest in the MFS and UFS; however, the CWMs of WD and Hmax were highest in the LFS. Facilitative interactions may also occur in woody plant communities of limestone hills at medium and high elevations, as these habitats are associated with faster growing species compared with the species typically found at lower elevation sites (Wright
We showed that high Elv enhanced the functional trait space (i.e., divergent traits) in MFS and UFS, while the LFS exhibited narrower functional trait diversity (i.e., trait convergence). Within this limestone hill study area, sites with high Elv, %RC, and sunlight exhibited increased CWMs of LS, SLA, and LT, indicating that these environmental factors were associated with light-demanding (faster growing) tree species. However, in areas with low values of these factors, we observed increased CWMs of WD and Hmax, indicating associations with shade-tolerant (slower growing) species. Restoration practices within degraded areas on limestone hills must consider the extreme environments created by large areas of bare rock, shallow soil, and high light. To achieve successful restoration in these areas, managers should plant species with large leaves, high SLAs, and/or thicker leaves, as species with these traits are associated with extremely degraded environments.