Abstract
In fall 2000, an on-farm sustainable agricultural research project was established for cotton (Gossypium hirstum L.) in Tift County, Georgia. Twenty fields that were to be planted to cotton in 2001 were identified which were approximately 5 to 10 acres in size. Four randomly selected fields were assigned to each of five cover crops: 1) cereal rye (Secale cereale L.); 2) crimson clover (Trifolium incarnatum L.); 3) legume mixture of balansa clover (T. michelianum Savi), crimson clover, and hairy vetch (Vicia villosa Roth); 4) previous legume mixture plus cereal rye; and 5) no cover crop (fallow) in conventionally tilled fields. Cotton was planted in two rows (36 in apart) on six foot beds. A strip was burned out in each row in the four cover treatments using paraquat so cotton could be planted. In the spring, insect counts were determined using sweep nets in the covers and when the cotton was small. Cotton plants from emergence to four weeks old were not sampled since the sweep net could break the fragile cotton plants. Five more weeks of sweep net data were collected from cotton. Insect samples after this involved whole plants since the cotton was too big for the sweep net to be effective. Each field was divided into 24 x 24 foot sample areas beginning at the center of the field. Each week for 14 or 15 weeks, 21 samples were obtained from each field. Comprising the 21 random samples were one sample from the four center plots, one sample from each of the four sides, and four samples from each of the four quadrants. Thus throughout the season, a five acre field could have most of the plots sampled at least once. Larger fields saw a smaller percentage of all the plots sampled. No interior plot excluding the center and edges was sampled a second time until every plot had been sampled once. A response surface was fitted for the weekly data for each field. As would be expected, high densities of insects resulted in a significant fit.
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Recommended Citation
Mullinix, Benjamin G. and Tillman, Glynn
(2006).
"USING RANDOM SAMPLING TO ESTIMATE INSECT COUNTS AS RESPONSE SURFACES INVOLVING SPACE AND TIME,"
Conference on Applied Statistics in Agriculture.
https://doi.org/10.4148/2475-7772.1123
USING RANDOM SAMPLING TO ESTIMATE INSECT COUNTS AS RESPONSE SURFACES INVOLVING SPACE AND TIME
In fall 2000, an on-farm sustainable agricultural research project was established for cotton (Gossypium hirstum L.) in Tift County, Georgia. Twenty fields that were to be planted to cotton in 2001 were identified which were approximately 5 to 10 acres in size. Four randomly selected fields were assigned to each of five cover crops: 1) cereal rye (Secale cereale L.); 2) crimson clover (Trifolium incarnatum L.); 3) legume mixture of balansa clover (T. michelianum Savi), crimson clover, and hairy vetch (Vicia villosa Roth); 4) previous legume mixture plus cereal rye; and 5) no cover crop (fallow) in conventionally tilled fields. Cotton was planted in two rows (36 in apart) on six foot beds. A strip was burned out in each row in the four cover treatments using paraquat so cotton could be planted. In the spring, insect counts were determined using sweep nets in the covers and when the cotton was small. Cotton plants from emergence to four weeks old were not sampled since the sweep net could break the fragile cotton plants. Five more weeks of sweep net data were collected from cotton. Insect samples after this involved whole plants since the cotton was too big for the sweep net to be effective. Each field was divided into 24 x 24 foot sample areas beginning at the center of the field. Each week for 14 or 15 weeks, 21 samples were obtained from each field. Comprising the 21 random samples were one sample from the four center plots, one sample from each of the four sides, and four samples from each of the four quadrants. Thus throughout the season, a five acre field could have most of the plots sampled at least once. Larger fields saw a smaller percentage of all the plots sampled. No interior plot excluding the center and edges was sampled a second time until every plot had been sampled once. A response surface was fitted for the weekly data for each field. As would be expected, high densities of insects resulted in a significant fit.
