*3.1.4. Experiment 3. Establisment of Arachis pintoi accessions using seed pods*

This experiment compared the establishment of three *A. pintoi* accessions using seed pods: CIAT 17434 (cv. Pinto peanut or Amarillo), 18744 and 18748. Seed germination was assessed in the laboratory at room temperature; using 125 seeds per accession. Petri dishes, bottomlined with filter paper, were used and were watered twice daily. The seeding rate was equivalent to 10 kg of germinable seed pods per hectare, based on quadruplicate germina‐ tion tests. The experimental plots (10 m x 5 m; ten 5 m length rows/plot) were established within a grazing experiment where milk production from native pastures and native pas‐ tures associated with *A. pintoi* was to be compared. Three replicates were established in one paddock and three in another. Each replicate had three plots, with an accession each. Plots were excluded from grazing for the 12 weeks of the establishment period. A 2% aqueous sol‐ ution of glyphosate was applied on a 0.30 m wide strip 15 days before planting to eliminate competition from existing vegetation. Distance between rows and planting positions was 1.0 m and 0.5 m, respectively. Seed pods were placed in a 5 cm deep hole made with pointed wooden stick, and lightly covered with soil by the planter's foot. Three replicates were planted on August 2 and three on September 3, 1996. Fertiliser was not applied.

#### **3.2. Measurements and statistical analyses**

a similar number of planting positions per sub-plot and two sampling hills/m2

*3.1.3. Experiment 2. Type of control of native pasture growth, with or without P fertiliser*

*tortus*) native pastures, in Australia.

0.07 m depth hole adjacent to the planting position.

*3.1.4. Experiment 3. Establisment of Arachis pintoi accessions using seed pods*

cide application.

56 Soil Fertility

pling quadrat, regardless of type of tillage. Fertiliser was broadcast 30 days after planting.

This experiment tested the combined effect of the type of pasture vegetation control: herbi‐ cide (glyphosate) or slashing (by machete) with or without burning of dead vegetation, and with or without localised P-fertilisation which resulted in eight treatment combinations. The choice of treatments attempted to reduce competition to *A*. *pintoi* from existing native pas‐ ture vegetation, enhance legume establishment and early growth, following the approach described by [25] for the establishment of legumes into existing Speargrass (*Heteropogon con‐*

Slashing was done by machete and burning was carried out between 1-5 days after slashing. A 2% aqueous solution of glyphosate (480 g of isopropyl amine salt of glyphosate/l) was ap‐ plied on a 0.25 m wide strip 15 days before planting; burning was done 15 days after herbi‐

The planting legume was done between June 28 and July 3. Application of herbicide and herbicide plus burning, and slashing or slashing plus burning, were applied 15-16 days and 3-5 days earlier, respectively. Vegetative material, 0.25 m length stolons with eight nodes, was used for planting. This material was inoculated just prior to planting with a specific *Bradyrhizobium* culture obtained by suspension of 1 kg of profusely nodulated *A. pintoi* ground roots in a solution 7.5 litres of water and 1.5 litres of sugarcane molasses. Three sto‐ lons per planting position were put in a hole and covered with soil, allowing about 1/3 of the stolon to remain above ground. Distances among rows and planting position were 1.0 m and 0.5 m, respectively. The sub plot (10.0 m x 6.5 m) had 10 rows with 14 planting posi‐ tions/row. Two sampling quadrats (2 m x 1 m) each with 4 planting positions were random‐ ly allocated per sub plot. Single super phosphate (30 kg of P/ha) was applied at planting in a

This experiment compared the establishment of three *A. pintoi* accessions using seed pods: CIAT 17434 (cv. Pinto peanut or Amarillo), 18744 and 18748. Seed germination was assessed in the laboratory at room temperature; using 125 seeds per accession. Petri dishes, bottomlined with filter paper, were used and were watered twice daily. The seeding rate was equivalent to 10 kg of germinable seed pods per hectare, based on quadruplicate germina‐ tion tests. The experimental plots (10 m x 5 m; ten 5 m length rows/plot) were established within a grazing experiment where milk production from native pastures and native pas‐ tures associated with *A. pintoi* was to be compared. Three replicates were established in one paddock and three in another. Each replicate had three plots, with an accession each. Plots were excluded from grazing for the 12 weeks of the establishment period. A 2% aqueous sol‐ ution of glyphosate was applied on a 0.30 m wide strip 15 days before planting to eliminate competition from existing vegetation. Distance between rows and planting positions was 1.0

in each sam‐

The response variables were: 1) plant number (PN, plants/m2 ) by counting; 2) plant height (PH, cm), on each plant within the sampling quadrat, measured with a ruler from the soil surface to the uppermost part of the plant; and 3) soil covered by the legume or cover (COV, % of quadrat area covered by the legume) measured with the aid of a 1 m2 quadrat, divided into 25 squares, which was placed over the row. These measurements were done on weeks 4, 8 and 12 after planting [26]. In experiment 1, PH was not measured, but COV was meas‐ ured again at 24 weeks after planting.

In experiment 1, there were no field replications, since it was perceived that treatments ap‐ plied in larger areas would have a closer resemblance to that of farmers' fields. Also, if sev‐ eral sampling quadrats were used within each treatment plot, this would yield information as useful as that obtained from randomised complete block designs. In experiments 2 and 3, the design was a randomised complete block design with 3 blocks as replicates. The treat‐ ment arrangement was a split-plot in experiment 2, where the main plot was the combina‐ tion of type of pasture vegetation control (slashing and herbicide), while the combinations of burning (with and without) and P application (with and without) were the sub-plots; addi‐ tionally the effect of time after planting was considered a sub-sub-plot. The treatment ar‐ rangement of the third experiment was a split plot, in which the main factor was the combination of month of planting by accession and time after planting the sub-plot. Here, number of plants was expressed as "plants/50 m2 ", in order to be clearer and avoid fractions of plant/m2 . Analyses of variance were done with linear additive models in accordance to the experimental design [27]. The natural log transformation of the response variable was used if its response to time was exponential. If necessary, linear or exponential relationships provided rates of increase with time in the measured variables. Also, means comparisons using Tukey's test were done when was necessary.
