An Observatory Plot System for Grain Production in the Neotropical Savannas

The neotropical savanna is the second largest biome in South America, with significant potential for agricultural development. In Colombia, this biome is experiencing rapid land-use change leading to the conversion of seminatural landscapes into to intensive agricultural systems. Our Dataset Paper documents the emerging intensive grain production systems. Between 2011 and 2013, we established 336 observatory plots within farmer’s maize, rice, and soybean fields along a 200Km transect from Puerto Lopez (Meta) to Viento (Vichada). From each of these plots, we submit 184 descriptors or variables capturing their location, rotation history, management, and environment. Our specific objective in collecting the data was to identify key factors explaining yield variation, with emphasis on interactions between management and environmental factors potentially informing the development of site-specific management protocols. Beyond this objective, the dataset submitted here is intended to support additional inquiries contributing to the sustainable development of agriculture in the neotropical savannas.


Introduction
The neotropical savanna is the second largest biome in South America, occupying about 250 million hectares of land [1][2][3]. Its soils are notorious for their high acidity and aluminum levels toxic to most crops. Still, Nobel Prize Laureate Norman Borlaug called it "the last agricultural frontier in the world" [4]. Indeed, between 1955 and 2005 the Brazilian savannas experienced an extraordinary frontier expansion leading to the cultivation of over 40 million hectares of land previously considered infertile [5]. Reflecting on this achievement, Borlaug envisioned a similar transformation for the savannas in Colombia, Venezuela, and Southern Africa [5]. Contributing to his vision, our Dataset Paper documents the ongoing landuse conversion of seminatural savannas into intensive grain production systems in Colombia.
Our objective in collecting the data was to help identify key factors explaining yield variation in maize, rice, and soybeans on farmers' fields. From an operational standpoint, we recognized that these factors were of two types: those that easily lend themselves to agronomic manipulation and those that do not. The first group includes variables like soil pH, which can be adjusted relatively easily by liming. We call this group management factors. The second group includes variables like soil texture, which cannot be changed. We call this group environmental (or zoning) factors. We further recognized that the influence of some management factors on yield may depend on one or more environmental factors. For example, the same amount of irrigated water may cause yield improvements in sandy soils and may cause the root to rot in clay soils. Characterizing these types of interactions between management and environmental factors is the foundation of site-specific agriculture. Aware of this potential, our study was designed to provide a stepping stone for the development of site-specific grain agriculture in the Colombian savannas.
Encouraged by multiple requests of our dataset to address additional research questions, we are pleased to formally present it to our community of interest in this Dataset Paper. The objective of submitting this dataset, therefore, is to encourage and support diverse research inquiries contributing to the sustainable development of agriculture in the neotropical savannas.

Methodology
The study was conducted in the Colombian savannas, locally known as "Llanos Orientales, " a region that extends from the Meta Department to the Venezuelan border (Figure 1(c)). Its climate is characterized by a wet season that begins in March and a dry season that begins in December, with an average annual temperature around 26 ∘ C [6,7]. The length of the wet season accommodates two planting seasons for grain crops, one around April and another around September. Soils are mainly Oxisols with low fertility and high acidity and Al saturation [6,8]. A good ecological characterization of the region is provided by Blydenstein [9]. Our method involved the establishment of observatory plots within farmer's maize, rice, and soybean fields along a 200 Km transect from Puerto Lopez to Viento from 2011 to 2013. We call these plots "EGM, " after the Spanish acronym for Georeferenced Sampling Station (Figures 2(a) and 2(b)).
EGMs were 20 m 2 , a size we chose because it facilitated intensive sampling and matched the experimental field size used by the Colombian Ministry of Agriculture to evaluate and register new cultivars. A series of farm visits, involving unstructured interviews with farm managers and guided field inspections, helped us survey the variability between and within fields with respect to topography, soil texture, rotation history, and yield history. During the inspections, we consultatively established two or more EGMs per field, in such way that captured the greatest perceived variability with respect to the above-mentioned factors. Our sampling within fields was therefore not random but was designed to increase statistical variance with respect to yield and a few of its potentially important environmental determinants.
We relied on three sources of data: farm records, direct measurements, and geographic information systems (GIS) databases. Farm records helped us capture rotation history, crop cultivar, and planting dates. Direct measurements helped us capture soil parameters, plant density, and yield. Immediately before the planting season, we collected three soil subsamples (Figure 2(c)) along a diagonal transect across the EGM, at depths of 0-10 cm and 10-20 cm, and bulked them into a single sample per depth profile. These samples were submitted for chemical analyses to the soil laboratory at the International Center for Tropical Agriculture (CIAT). In addition, core samples of 100 cm 3 volume (Figure 2(d)) were taken from near the center of the EGM, at depths of 0-10 cm and 10-20 cm, and submitted for physical analyses to the Soil Laboratory at the Colombian Corporation of Agricultural Research (Corpoica). Plant density, yield, and grain moisture were measured within two weeks of the field's intended harvest date. We harvested the EGM manually to measure yield and grain moisture content ( Figure 2(b)). We used these two values to adjust yield based on the moisture content desired for storage (i.e., dry yield), which is 14.2% for rice and maize and 12.2% for soybeans. Finally, EGMs were georeferenced using geographic positioning system receivers (GPSMap 76CSx; Garmin, Olathe, Kansas, USA), and the coordinates were used to retrieve 250 m normalized difference vegetation index (NDVI) data from the Moderate Resolution Imaging Spectroradiometer (MODIS, [10]), precipitation data from the Tropical Rainfall Measuring Mission (TRMM; [11]), and interpolated climate data from WorldClim [12].

Dataset Description
The dataset associated with this Dataset Paper consists of 2 items which are described as follows.

Concluding Remarks
This comprehensive Dataset Paper is submitted to support research leading to the sustainable agricultural development of the neotropical savannas. Its specific design, however, responds to our interest in identifying management by environment interactions characterizing the potential for site-specific grain agriculture in the region. Our approach is informed by the rapidly growing literature demonstrating the promise of ecoinformatics approaches to streamline agricultural research [13][14][15][16][17][18]. Based on these experiences, we believe our Dataset Paper holds significant potential to facilitate a quantum leap in agricultural research for the development of the Colombian savannas.

Dataset Availability
The dataset associated with this Dataset Paper is dedicated to the public domain using the CC0 waiver and is available at http://dx.doi.org/10.1155/2015/625846/dataset.