The study site is located near the village of Sardinilla, Central Panama (9°19′ N, 79°38′ W), which is approximately 50 km north of Panama City. The elevation of the site is 70 m with a slightly undulating topography. Mean annual precipitation at nearby Buena Vista is 2350 mm, with only 25–50 mm of monthly rainfall during peak dry season (January‒March), 120–130 mm of monthly rainfall during the two transition periods one in April and one December, and 250 mm of monthly rainfall during the rainy season (May to November) [
46]. The mean annual temperature of the region is 26.2 °C. The soils are derived from Tertiary limestone and other sedimentary rocks resulting in a clayey Luvisol (pH 5.5; C 5%; N 0.4%) [
52]. The original forest vegetation before clear cutting at the Sardinilla site was probably a tropical moist lowland forest, with similar species composition and forest structure to that on Barro Colorado National Monument [
53], 40 km west of Sardinilla (9°09′ N, 79°51′ W). About the land use history of the site, it is known that the forest was clear-cut in 1952 and 1953 and small scale agriculture was established. A common practice in the area was to convert the agricultural fields after two years with crops (commonly used crops are corn, plantain and yucca) into pasture by seeding a high yielding C4 African grass,
Hyparrhenia rufa [
54].
In July 2001, the site was converted into an experimental tree plantation with the aim to investigate tree mixture effects (Sardinilla Project). On a total area of 9 ha, 24 different plots were established using altogether six tree species native to the natural tropical moist lowland forest (
Table 1). The plantation was set up with plots of one, three and six tree species from three different ecological groups [
55,
56]. Two light demanding, early successional tree species (
Luehea seemannii Triana & Planch and
Cordia alliodora (Ruiz & Pavon) Oken), two light-intermediate tree species (
Anacardium excelsum (Bert. & Balb. ex Kunth) Skeels and
Hura crepitans L.) and two shade tolerant, late successional tree species (
Cedrela odorata L. and
Tabebuia rosea (Bertol.) DC.) were chosen for the establishment of the experimental plantation. The six tree species and three ecological groups were selected based on relative growth rates (RGR), measured on Barro Colorado Island (9.1% and 7.0%; 5.9% and 4.9%; 2.3% and 3.4%, respectively [
55]). Under natural conditions on BCI, the tree species
Hura crepitans, Anacardium excelsum and
Cedrela odorata were classified as “brevi-deciduous”.
Cordia alliodora and
Tabebuia rosea were considered as “deciduous” and
Luehea seemannii as “facultatively deciduous” [
57]. Further, these species differ in their specific leave area and intercept properties [
58]. However, Meinzer
et al. [
57] also state that under the seasonally dry condition of the study region, leaf phenology seems to be a continuum with considerable overlap among categories. Interestingly, the classification into phenology groups was quite different for the different species compared to the observations from natural forests in the area [
57]. Those differences might be due to methodological differences; e.g., the better access to the canopy for the observation of leaf cover in the tree plantation and the far higher number of replicates per species in the artificial system than in the highly diverse natural forest from Barro Colorado Island. The intense observation in the plantation further allows a more detailed classification of the species regarding to their leave phenology. Despite this, the initially hypothesized differences in species functionality were proven manifold by various studies conducted at the site and were found to be true for a wide variety of ecosystem functions and processes [
37]. For example, after a couple of years of growth, the initially hypothesized canopy stratification by trees with different growth rates became visible. Accordingly, the upper canopy was formed by the fast-growing light demanding species and the lower canopy by the late successional shade tolerant species [
37]. Besides the ecological functioning of the tree species,
Anacardium excelsum (wild cashew or espavé),
Cordia alliodora (jennywood or bocote),
Cedrela odorata (Spanish cedar or cedro) and
Tabebuia rosea (roble) are valuable timber species with excellent wood properties. Many parts of
Hura crepitans (sandbox tree, possumwood or jabillo) can be used for medical purposes and
Luehea seemannii (this species was used by Erwin [
59] for the estimation of the worldwide number of arthropod species) is commonly used as fire wood.
Each of the established plots (
n = 24) has a size of 45 m by 45 m and tree saplings were planted with 3 m spacing between trees (225 trees per plot), which is the common planting design for industrial tree plantations in Central America [
60]. Each of the plots is divided into four subplots 22.5 m by 22.5 m in size. Seeds were collected from forests within the Barro Colorado National Monument and germinated on-site at Sardinilla. Saplings of approximately 50 cm height were planted. The plantation is tended following standard forestry practices in Panama, the understory being cut three times a year. Survival of saplings was very high (>90%), however,
Cordia alliodora did not establish in monoculture probably due to missing mycorrizha in the bare pasture soil [
60]. Therefore, this species was omitted in most of the investigations. Tree growth (tree height and diameter) was inventoried on an annual basis since 2001.