Diseño de muestreo para BGBD

1. Sampling Design for the BGBD Inventory

The sampling was carried out in each one of three selected windows and in each one of four land use types for each window (Rain Forest, Agroforestry, Maize and Pasture). The three windows in Los Tuxtlas region are associated with three common property communities or ´ejidos´: Ejido Adolfo López Mateos, Ejido Venustiano Carranza and Ejido San Fernando.

The number of selected observation points was eight for each soil use (32 sampling points per window) for a total of 96 points for three windows.

2. Macrofauna Sampling methodology

The sampling was carried out in November – December 2003. In each sampling point one monolith (25x25cm x30 cm) (Anderson and Ingram, 1993) was divided in 4 parts: the litter and the three strata 0-10, 10-20 and 20- 30cm. Each part was manually revised and all fauna > 2mm was collected. The earthworms were preserved in formaldehyde at 4% and the other organisms, first they were deposited in a “Pampel” solution (alcohol/acetic acid /formaldehyde) and then in alcohol at 70%.

3. Soil sampling for microorganisms and soil physicochemical parameters

In each Sampling point a composite soil sample was done to do the inventory of the microorganisms (nitrogen fixing bacteria (NTB), mycorriza, phyto-pathogens fungi and nematodes) and to measure the physicochemical parameters of the soil (pH, C, N, mineral N, P, CIC).

This composite sample was made with a core sampler having a size of 5cm diameter and 20 cm height. The sub samples (12-20) were taken at 0-20 cm depth without litter but keeping the roots. The cores were collected around 2 circles of 3 and 6 m radius, the monolith (for macro fauna sampling) being the center of the circles as recommended by Susilo ( 2004). The soil cores were put in 2 new large plastic bags in the field and then in the station we put together the content of the 2 bags and we mixed and homogenized the soil with a small shovel, taking off stones and breaking the large aggregates. This composite sample per sampling point weighted approximately 4.5 kg. Once the soil homogenization was made the correspondent soil quantity was deposited in a clean plastic bag to send it to the corresponding specialist for its study (NTB: 200g, mycorriza: 2000 g, phyto-pathogens fungi: 600g; nematodes 500g; physicochemical parameters: 1100g), the samples that needed to be refrigerated were kept in ice boxes while they reached the respective laboratory for their analysis.

4. Experimental Design for Analysis

The initial experimental design for analysis was a two-factorial design with n = 8±4 replicates (sampling points in each treatment combination). It was assumed that each “ejido” has a different level of deforestation (Table 1).

Tabla 1. Experimental design

5. Variance Components

Between landscapes with different deforestation level

n = 32 d.f. = 2

Between intensity of land use

d.f. = 3

Rain Forest

Agroforestry

Pasture

Maize

Residual d.f. = 84

6. A priory analysis (ANOVA)

Table 2. Variance analysis with interaction

This design is valid if it is assumed that there is total independence between replicates (Initially, a replicate was considered as such if the sampling points were separated 150 m one of each other, in order to prevent autocorrelation and trying to keep a grid design) and that the windows were homogeneous. However, the windows have different conditions, e.g. deforestation level, altitude, soil type, etc., and they would be considered as blocks. In this case, to test the block effect (window) an assumption is that there is not interaction (because the block has a restriction error) and the design become as:

Table 3. Variance analysis without interaction

Recently, after preliminary analysis, we concluded that diversity patterns were confused and that it was necessary add new sampling points in order to clarify the picture about those patterns. On this way, new points were added in each windows and news criterion about the true replication become present. On this way, the initial sampling design becomes different and unbalanced due to both, to the new sampling points and to the new replication criterion (table 4). It was considered that not all points were true replicas because they were in the same plot. At the present, a new possibility for analysis is that a parcel be considered as a replicate independently if two o more sampling points are present. Thus we would have only 49 true replicas.

Table 4. New experimental design

If the new replication criterion is applied and the new sampling points are considered, the ANOVA model becomes as:

Table 5. Variance analysis considering the new sampling points, new replication criterion and interaction

In a similar way, if the windows are considered different and they are functioning as blocks, then the model become as:

Table 6. Variance analysis considering the new sampling points, new replication criterion whit out interaction.