International Journal of Horticulture, 2017, Vol.7, No. 12, 95-105
97
For plants with a first interim cut i.e., detachment of yam at the time of first interim growth record, no further cut
is made; however a second interim growth record is taken on the plants alive, before final harvest. For plants with
yams attached to stems, experiment continued after growth recording till the end of plant lifetime. Stress is severe
on plants under yam detachment.
It may be mentioned that in a similar set-up under plant stress of no weeding, little fertilizer, and little irrigation
before first interim reading on yam, regular plant care started only after first interim reading till the end of
experiment; in that case the strategy of yam detachment made around middle of experiment is seen to be a
superior strategy to maximise total yield with 650 g of seed weight, among seed weights 500 g, 650 g, 800 g; see
Dasgupta (2016).
The present experiment is conducted with a little vermicompost given in the pits at start, and a few irrigations
given only in extreme summer. Here the plant stress is extreme in harsh agro-climatic environment.
Individual growth curves are drawn with linear interpolation in between, from four possible readings on
underground yam per plant viz., one in the beginning, two interim readings during experiment, and one at the end
of plant lifetime. Overall growth curves in each categories are drawn by computing the average response (mean /
median) at specific time points where at least one yam observation is available in individual growth curves and
then using nonparametric lowess / spline regression on averaged points, see Dasgupta (2015a), Dasgupta (2015b).
The growth curves help to understand underground yam deposition process over plant life time.
Our procedure of comparison of two strategies of yam harvest is mainly based on growth curve analysis. In
addition, asymptotic normal test and Welch
t
test to compare harvesting strategies of yam are investigated.
Quantification of the advantage in yam yield, when two strategies are compared, is made by nonparametric
Hodges-Lehmann estimator, see Hodges and Lehmann (1963).
3 Results
3.1 Parametric and nonparametric techniques to compare different harvesting procedures
Consider the approximate normal test
z = (
x -
y) /
{(s
2 x
/ m) + (s
2 y
/ n)} =(1.3215 - 1.06033) /
(0.183/10+0.009276/3) =1.785632 for equality
of means of total yield when first and second interim cut is considered for seed weight 800 g over two distinct set
of yam plants; z is significant at 5% level,
p
value in one sided alternative that cut in second interim produces
higher yield turns out to be
p
=0.0371. When the same comparison of first and second interim cut is made for seed
weight 650 g, z = 2.446258; z is highly significant,
p
value in one sided alternative being
p
=0.0072.
If we compare total yield with second interim cut for seed weight 800 g being better than that for 650 g, then z =
2.042643, and z is significant at 5% level,
p
value in one sided alternative being
p
=0.0205.
For seed weight 500 g, growth curves of plants show downward trend till first intervention, and then these remain
parallel to
x
axis showing a stalled growth, this indicates 500 g is minimal seed weight for plant survival in harsh
agro-climatic condition of conducted experiment.
The value of z may be compared with a
t
variable, referred as Welch's
t
statistic with degrees of freedom (
d.f.
)
computed as {(s
2 x
/ m) + (s
2 y
/ n)}
2
/ {(s
2 x
/m)
2
/ (m-1) + (s
2 y
/n)
2
/ (n-1)}. The true distribution of the test statistic
depends slightly on the two unknown population variances and this is an approximate test. In the above three
cases of comparison with approximate normal test, the
d.f.
of Welch's
t
statistic are approximately 11, 10 and 17,
respectively. The
p
values are low for one-sided alternatives; the values in three cases are 0.050864, 0.017241 and
0.028458, respectively.
As already mentioned, in a similar set-up under severe plant stress of no weeding, little fertilizer, little irrigation
before first interim reading on yam; detachment around middle of experiment is seen as a superior strategy to
maximise total yield with 650 g seed weight, among seed weights 500 g, 650 g, 800 g; see Dasgupta (2016).
