Field Crops Research 230 (2019) 11–16

Contents lists available at ScienceDirect

Field Crops Research

journal homepage: www.elsevier.com/locate/fcr

Determination of a plant population density threshold for optimizing cotton T

lint yield: A synthesis
⁎
Curtis Adamsa, , Santanu Thapaa, Emi Kimurab
a
Texas A&M AgriLife Research, 11708 Highway 70 South, Vernon, TX, 76384, USA
b
Texas A&M AgriLife Extension, 11708 Highway 70 South, Vernon, TX, 76384, USA

A R T I C LE I N FO A B S T R A C T

Keywords: Technology fees associated with modern cotton cultivars have increased seed costs considerably, giving pro-
Cotton lint ducers the impetus to reduce plant population density, where possible. Most recent studies on cotton population
Population density density, conducted across diverse environments, report similar patterns of crop response: decreases in lint yield
Threshold only at very low densities, with generally consistent yield across all higher densities. But no work had been done
Yield optimization
to bring the literature together, quantitively synthesizing the yield data collectively, to better pinpoint a po-
pulation density threshold. And, notably, little had been reported on the effects of population density in lower-
yielding dryland environments. Quantitatively synthesizing population density datasets from the literature,
including our own dryland data, was the objective of this research. The dryland data showed that lint yield and
biomass partitioning were not affected by population density, similar to higher-yielding environments over the
same range in density. Following normalization of all lint yield data (literature and dryland datasets), a
breakpoint in population density at 35,000 plants ha−1 was identified, which can be interpreted as the minimum
plant density at which yield may be optimized. This rate is lower than the common density recommendation of
81,000 plants ha−1 and substantially lower than the risk-averse rates at which many producers plant (129,000
seeds ha−1 or greater). The analysis showed that yield will decline precipitously below 35,000 plants ha−1,
exposing the enormous risk to cotton producers in approaching this low density, particularly if significant seed
or plant loss is expected. However, the analysis suggests that excessive over-seeding may be occurring in many
cases, resulting in economic losses to producers. The analysis also provides guidance on a threshold for pro-
ducers facing replanting decisions.

1. Introduction
In recent decades, breeding and genetic modification developments
in cotton (Gossypium hirsutum L.) have made substantial progress in
improving the crop, in terms of yield, quality, and ease of management
(USDA-ERS, 2017). These developments have also increased the cost of
seed substantially (Shurley et al., 2010). The National Cotton Council of
America estimated in their 2015–2016 cotton production budgets that
seed costs account for about 16% of the total operating costs for U.S.
cotton producers (NCCA, 2017). High seed costs create an important
impetus for producers to reduce planting population densities, where
possible.
Population density plays a central role in optimizing productivity,
management, and economic return of any field crop, including cotton.
Plant population density in cotton has a substantial impact on plant
architecture and other canopy dynamics, which influences disease and
pest incidence, harvest efficiency and timing, water use, and other
factors (Kaggwa-Asiimwe et al., 2013; Maddonni et al., 2001; Stewart,
2005). Population densities that are too high or too low can have ne-
gative agronomic consequences. Agronomists and Extension specialists
at the University of Florida and North Carolina State University re-
commend achieving final population densities of about 81,000 plants
ha−1 (2.5 plants ft−1 on 40 in. rows) to optimize yield and associated
agronomic factors (Collins, 2015; Donahoe, 2010). A major cotton seed
supplier, DeltaPine, recommended achieving the same rate (DeltaPine,
2017). Donahoe (2010) stated that optimal yields can be achieved at
about 50,000 plants ha−1 (1.5 plants ft−1 on 40 in. rows), but men-
tioned that planting as high as 129,000 plants ha−1 (4.0 plants ft−1 on
40 in. rows) may be necessary if soil or seed conditions are poor (e.g.
soil crusting is likely or the percent viable seed is low). Collins (2015)
stated that over-seeding by about 20% is needed to achieve a desired
density in ideal conditions and that even greater seeding excesses are

⁎
Corresponding author.
E-mail address: curtis.adams@ag.tamu.edu (C. Adams).

https://doi.org/10.1016/j.fcr.2018.10.005
Received 16 August 2018; Received in revised form 9 October 2018; Accepted 9 October 2018
0378-4290/ © 2018 Elsevier B.V. All rights reserved.
C. Adams et al. Field Crops Research 230 (2019) 11–16

Table 1 to 33,000, 51,000, 69,000, 87,000, and 105,000 plants ha−1 in Linqing,
A summary of lint yield in cotton as affected by population density in studies China. In Georgia, USA, Bednarz et al. (2005) found that lint yield was
reported in the literature since year 2000. To be included, the studies con- reduced at 36,000 plants ha−1 relative to yield at 90,000 and 126,000
formed to the criteria given in the methods section. plants ha−1. In Henan, China, yield was reduced at 15,000 plants ha−1
Source Location Year Population Lint relative to peak yields obtained at 51,000 to 87,000 plants ha−1 (Zhi
Density (plants Yield (kg et al., 2016). A two-year study in Texas, USA (Feng et al., 2014) showed
ha−1) ha−1) no difference in lint yield from 75,300 to 226,000 plants ha−1 (a single
Bednarz et al., Georgia, USA 2001– 2002 36,000 1246 b
year’s data from another location showed differences, but the crop was
2005 90,000 1345 a damaged by hail). In a relative outlier, Pettigrew and Johnson (2005)
126,000 1376 a showed that lint yield peaked with plant population densities between
215,000 1363 a 90,000 to 130,000 plants ha−1, while five percent lower yield was
Dai et al., 2015 Linqing, China 2010–2013 15,000 1771 b
observed at 70,000 plants ha−1, in Mississippi, USA.
33,000 2204 a
51,000 2231 a In summary, most recent studies on cotton population density report
69,000 2273 a decreases in lint yield only at very low population densities, with peak
87,000 2292 a yields across broad population ranges, with no decreases in yield evi-
105,000 2277 a
dent even at extremely high densities. A qualitative assessment of the
Feng et al., 2014 Texas, USA 2007 75,300 1278 a
150,600 1294 a
literature suggests that an optimal population density for yield is low,
226,000 1235 a likely far lower than current population density recommendations. No
Feng et al., 2014 Texas, USA 2008 75,300 1633 a work has been done to bring this literature together, quantitatively
150,600 1776 a synthesizing the yield data collectively. And, notably, most recent
226,000 1616 a
studies on population density were conducted in relatively high-
Gwathmey et al., Tennessee, USA 2006–2008 61,000 1752 a
2011 114,000 1780 a yielding rainfed or irrigated conditions and results in lower-yielding
Gwathmey et al., Tennessee, USA 2006–2008 30,000 1459 b dryland conditions are missing. In this study, our first objective was to
2011 63,000 1756 a field test effects of cotton population density in lower-yielding dryland
Gwathmey et al., Tennessee, USA 2006–2008 61,000 1189 a conditions, with a focus on lint yield and plant biomass partitioning.
2011 114,000 1176 a
Gwathmey et al., Tennessee, USA 2006–2008 30,000 1087 b
Our second objective was to combine our lint yield data with compar-
2011 63,000 1234 a able data from other published studies, synthesizing the collective da-
Pettigrew and Mississippi, USA 2001–2004 70,000 1393 b tabase to better understand how yield is affected by population density
Johnson, 90,000 1441 a in cotton.
2005 110,000 1471 a
130,000 1465 a
Wrather et al., Mississippi, USA 2001 23,782 1360 a 2. Materials and methods
2008 33,976 1360 a
67,952 1310 a 2.1. Field study design, management, and analysis
135,904 1250 a
Wrather et al., Mississippi, USA 2002 23,782 1570 a
2008 33,976 1650 a
A two-year field study was conducted at the Texas A&M AgriLife
67,952 1630 a Research Station at Chillicothe, TX during the 2016 and 2017 summer
135,904 1630 a growing seasons on an Abilene clay loam soil in dryland conditions. The
Wrather et al., Mississippi, USA 2003 23,782 850 b area is characterized by a semi-arid climate, with annual average pre-
2008 33,976 1060 a
cipitation of about 635 mm, mostly occurring as spring and fall pre-
67,952 1190 a
135,904 1140 a cipitation with minimal summer rainfall, and high rates of evapo-
Wrather et al., Mississippi, USA 2004 23,782 1340 b transpiration. ‘Phytogen 333 WRF’ cotton was planted in four rows with
2008 33,976 1660 a a row width of 102 cm (40 in.) on 9 June 2016 and 8 June 2017. Plot
67,952 1760 a lengths were 59 m and 40 m in 2016 and 2017, respectively. The study
135,904 1650 a
Zhi et al., 2016 Henan, China 2012 15,000 950 b
was arranged as a randomized complete block design with four re-
51,000 1533 a plications of four cotton population density rate treatments. Population
87,000 1570 a densities, as determined by stand counts taken two weeks after
Zhi et al., 2016 Henan, China 2013 15,000 1313 b planting, were 67,300, 115,300, and 145,400 plants ha−1. All plots
51,000 1546 a
received in-season inputs of herbicide, fertilizer, insecticide, and defo-
87,000 1516 a
liation chemicals according to current recommendations for the region.
Measurements of plant biomass partitioning were taken from 2-m
needed in poor planting conditions. In the semi-arid environments of sections of crop row on three occasions in each year: 48, 70, and 102
central to west Texas, for example, where planting conditions are DAP in 2016 and 47, 77, and 104 DAP in 2017. In analysis and pre-
commonly challenging, regional producers often plant at seeding rates sentation, corresponding sampling times (days) were averaged between
at or in excess of 129,000 seeds ha−1. To put these figures into per- years. After cutting the stem just above the soil level, biomass samples
spective, recent studies provide needed research-based information on were transported to the lab and partitioned into leaf, stem, and re-
how lint yield responds to population density in advanced modern productive biomass components and placed in paper bags.
cotton cultivars (Table 1). Reproductive biomass components included flowers, squares, and bolls.
Wrather et al. (2008) reported that cotton lint yield was reduced at Samples were dried at 60 °C in an air-forced oven. Stand counts were
24,000 plants ha−1 relative to yield at 34,000, 68,000, and 136,000 used to convert aerial biomass measurements into biomass per plant.
plants ha−1 in just two years out of four in a study (no differences Plots were harvested by a stripper on 28 November 2016 and 17
reported in the other two years) conducted in Mississippi, USA. November 2017. Seed cotton samples were ginned to determine lint
Gwathmey et al. (2011) found that lint yield was reduced at 30,000 yield.
plants ha−1 relative to yield from 61,000 plants ha−1 to 114,000 plants In statistical analysis of the field data, population density treatments
ha−1 in Tennessee, USA. In a report with relatively high yields, Dai were considered a fixed effect in the statistical model, while years and
et al. (2015) found lint to be reduced at 15,000 plants ha−1 compared replications (blocks) were considered random effects. The data was
analyzed with SAS 9.3 (SAS Institute, Cary, NC). Differences between

12
C. Adams et al. Field Crops Research 230 (2019) 11–16

treatment yields were determined by ANOVA using Proc Mixed, in-

cluding post-hoc means comparisons adjusted using the Tukey proce-
dure, with a statistical threshold of 0.05. The assumption of normality
was checked for all response variables. Statistical analysis of the bio-
mass partitioning data was done with repeated measures ANOVA in SAS
with the Proc MIXED procedure.

2.2. Literature and field trial population density analysis

Data was mined from the literature (Table 1) for a broad-scale

analysis of the effect of population density on lint yield in modern
cotton varieties. Data were included from studies conducted in a variety
of environments that fit the following predetermined criteria: reports
from year 2000 or later; reported lint yield (i.e. seed cotton was not
used); used fixed row width and intra-row spacing (e.g. no skip row or
hill drop); and presentation of numeric data that was not integrated
across other treatments factors with significant interactions or an un-
clear interaction status. The lint yield data from the currently reported
experiment was also included. Literature yields and plant densities were
reported in many units (e.g. lbs ac−1) and measures (e.g. seeds ft−1),
but were converted to consistent units of kg ha−1 and plants ha−1,
respectively.
The yield data was graphed as a function of plant population density
and we observed that most datasets exhibited similar trends: yield de-
cline only at very low population densities and a plateau in lint yield
across all higher densities, whether the environment was relatively
higher or lower yielding. This was corroborated by the statistical ana-
lysis presented with each dataset (Table 1). As a result, we elected to
apply a normalization procedure to the data, so quantitative analysis
could be performed on the combined dataset. This was done by sum-
marizing the statistical analysis for all studies and calculating an overall
average of datapoints along the yield plateaus (data values statistically
considered peak yield within a given dataset) of all studies. This uni-
versal average was used in making a conversion factor for all datasets,
which was individually multiplied by each dataset value, as follows:
(universal plateau average / average of datapoints within dataset yield
plateau) × dataset value = normalized value. Once normalized, a two-
segment piecewise linear regression model was fitted to the data using
SigmaPlot 13.0 (Systat Software, Inc., San Jose, CA). This procedure is
commonly used to estimate “breakpoints” or trend transitions in data-
sets; it is often applied to determine the point or threshold when a re- Fig. 1. (A) Lint yield response data from the current report (marked by an as-
terisk) and literature reports on tests of population density in cotton; (B) the
sponse variable, such as yield, experiences a shift in response to treat-
same yield response data, normalized to account for differences in yield among
ment factors, similar to plateau models (Adams et al., 2015; Shaver
datasets; (C) and analysis of the normalized data, using a piecewise model, to
et al., 2017; Toms and Lesperance, 2003). identify a breakpoint (Xo) in the data, or the minimum population density at
which yield is expected to be optimized.
3. Results

precipitously at population densities below the 35,000 plants ha−1

3.1. Literature and field trial population density analysis
Across the studies mined from the literature, population density
ranged from 15,000 to about 226,000 plants ha−1 and lint yield ranged
from about 1200 kg ha−1 to about 2300 kg ha−1 (Fig. 1A). Most lit-
erature datasets exhibited a rise in yield from very low plant population
densities—as long as low enough densities were tested—and a yield
plateau across all higher density treatments tested (see statistics in
Table 1). The current study added data representative of dryland cotton
cropping systems in semi-arid environments that are lower-yielding,
which was missing from the collective dataset. In our study, population
density ranged from 67,000 to 145,000 plants ha−1 and lint yield
averaged 800 kg ha−1, with no differences among treatments in lint
yield (Fig. 1A). In Fig. 1A, the dryland data is marked with an asterisk.
A normalization procedure was applied to the datasets to enable a
collective, quantitative analysis of the lint yield response to population
density (Fig. 1B). In the normalized data, a breakpoint was identified at
35,336 plants ha−1 (1.1 plants ft−1 on 40 in. rows) by piecewise linear
regression (Fig. 1C). The normalized data showed that yield will decline
threshold and that yield does not increase above this density threshold.
A lack of data, especially at extremely low population densities (below
about 15,000 plants ha−1), resulted in a lack of detail in the precise
pattern of yield decline below 35,000 plants ha−1. Above 35,000 plants
ha−1, the regression analysis shows that lint yield remains quite stable
even to excessively high population densities.
3.2. Field trial biomass partitioning and production per plant
When considered across the growing season (48, 74 and 103 days
after planting), no differences were found among population density
treatments in the partitioning of biomass to leaves, stems, and re-
productive structures on an aerial basis (Fig. 2). More separation of
treatments, without significance, was evident in leaf biomass (P =
0.2854) and stem biomass (P = 0.3374) than in reproductive (P =
0.9895) biomass. Total biomass productivity (P = 0.8921), as well as
the ratio of reproductive to total biomass productivities (P = 0.8000),
were the same across all population densities tested.

13
C. Adams et al.
Fig. 2. In-season measurements of crop biomass development and partitioning, including leaf, stem, reproductive, and total biomass, and the ratio of reproductive to
total biomass, on aerial and per-plant bases. The error bars represent the standard error of the mean.
Differences in population density resulted in dramatic differences in
biomass production per plant, however, as the plants adapted physio-
logically to their systems and available resources (Fig. 2). Percent dif-
ferences in total biomass production per plant among population den-
sity treatments were minor by 48 days after planting, but quickly
diverged. By 100 days after planting, total biomass per plant was 94%
greater at a population density of 67,300 plants ha−1 than at 145,400
plants ha−1. For the same comparison, leaf biomass was 81% greater,
stem biomass was 76% greater, and reproductive biomass was 106%
greater.
4. Discussion
4.1. Literature and field trial population density analysis
When viewed together, there were interesting trends in lint yield
with population density across literature datasets (Fig. 1A). The tested
population densities ranged about 12-fold and the reported yields, in-
cluding the data from the current dryland study, ranged nearly three-
fold. Most literature datasets exhibited a rise in yield from very low
plant population densities—as long as sufficiently low densities were
14
Field Crops Research 230 (2019) 11–16
tested—and a yield plateau across all other treatments tested, no matter
how high those treatments reached. At high population densities, some
crops exhibit a yield decline associated with a full sigmodal yield re-
sponse curve, while others do not (Edwards and Purcell, 2005; Tetio-
Kagho and Gardner, 1988). The lack of yield decline at high densities in
cotton does not seem to be due to a lack of testing at high-enough plant
population densities: at the highest density tested (226,000 plants
ha−1) among the datasets, there would be a seed every 4.4 cm on rows
spaced at 1 m, an excessive density in practical and physiological
senses. A surprising trend in the data was the similarity in yield re-
sponse to population density among datasets, despite the wide variation
in yield among datasets. Wide variation in yield among the datasets
indicates that non-treatment factors, such as moisture, nutrition, and
genetic yield potential, often limited yield. But to the extent that the
cotton crops reported in literature differed in yield potential and had
other (non-population density) yield-limiting factors, lint yield re-
sponses to population density seem to have occurred independent of
these factors. Pettigrew et al. (2013) reported no population density
interactions with lint yield in cotton genotypes widely differing in yield
potential.
The breakpoint in lint yield with population density identified in
C. Adams et al.
this analysis (Fig. 1C), which can be interpreted as the minimum po-
pulation density at which yield may be optimized, was about 35,000
plants ha−1 (1.1 plant ft−1 on 40 in. rows). This rate is substantially
lower than common density recommendations for cotton. The most
common population density recommendation we found, across Uni-
versity Extension agencies and industry, was 81,000 plants ha−1 (2.5
plants ft−1 on 40 in. rows) (Collins, 2015; Donahoe, 2010; DeltaPine,
2017). This is a recommendation for final population density, which can
be achieved only by over-seeding to account for inevitable, and often
unpredictable, seed and plant loss. Collins (2015) stated that over-
seeding by about 20% is needed to achieve a desire density in ideal
conditions. Donahoe (2010) recommended planting as high as 129,000
plants ha−1 (4.0 plants ft−1 on 40 in. rows) when soil or seed condi-
tions are poor. These conditions would include soil crusting, excessively
hot or cool temperatures, poor seed quality or germination percentage,
and other situations. Over-seeding would also be needed if plant loss,
following emergence, was expected due to pests, disease, high winds/
blowing sand, or other factors. In central and west Texas, where
planting and growing conditions are often challenging, regional pro-
ducers commonly plant at rates at or even in excess of 129,000 seeds
ha−1.
In our analysis, the precipitous decline in yield that was evident at
population densities below 35,000 plants ha−1 shows the enormous
risk to cotton producers in approaching a density this low, particularly
if substantial seed or plant loss is expected. Even in ideal environmental
conditions, seed spacing is usually not perfect in practice and, at this
low population density, any seed gaps would result in yield loss
(Pettigrew et al., 2013). However, the analysis showed that, given that
yield can be optimized at 35,000 plants ha−1 across diverse environ-
ments, excessive over-seeding may be occurring in many cases, re-
sulting in economic losses to producers due to high seed costs. The
analysis also provides guidance or a threshold for producers making
replanting decisions. Following stand loss, if a producer measures a
plant population density at or above 35,000 plants ha−1 without fre-
quent, large gaps between plants, replanting may not be necessary.
While this analysis is focused on the effects of population density on
lint yield, there are additional factors that producers must consider
when making seeding rate decision for cotton, and many of these are
discussed in the following section.
4.2. Non-yield impacts of population density
In addition to lint yield, there are a variety of crop responses, both
positive and negative for cotton production, depending on the pro-
duction system and environment, that result from population density
and may create tradeoffs for cotton producers.
The cotton plant has an extensive ability to adapt its anatomical
structure and physiological functions to accommodate changes in plant
population density. For example, higher population density favors more
fruit production along the primary plant axis, while lower population
density results in increased growth of monopodial and sympodial
branches and fruit set on those branches (Bednarz et al., 2000). This has
implications on crop maturity and harvest efficiency, as secondary and
tertiary bolls that arise with low population density may require ad-
ditional time for full development (Gwathmey et al., 2011; Siebert
et al., 2006). Wide variation in maturity among bolls with low popu-
lation density can diminish fiber quality (Bednarz et al., 2005, 2006;
Jones and Wells, 1998), though fruit retention has been observed to be
higher at lower population densities (Bednarz et al., 2000; Jones and
Wells, 1998; Siebert and Stewart, 2006; Siebert et al., 2006). Increased
lateral branching with low population density can result in shorter
cotton with less vegetative growth, and less vegetative plants require
less plant growth regulator application to control growth, though more
extensive branching and thicker stems on each plant can reduce harvest
efficiency (Larson et al., 2004). In addition to shifts in the position of
fruit-set and fruit maturity, boll size is affected by changes in
15
Field Crops Research 230 (2019) 11–16
population density. Boll size is inversely related to population density
(Bednarz et al., 2000), which can also affect harvest efficiency, as lint
recovery is generally lower from smaller bolls. Planting at higher
densities can delay the initiation of fruiting and increase water use
(Bednarz et al., 2005; Zhi et al., 2016). Planting at high densities has
been shown to improve cotton competitiveness with weeds (Gwathmey
et al., 2011; Street et al., 1981), which may be most important in
modern organic cotton production. There are examples of diseases that
can be controlled, within limits, by population density, such as Verti-
cillium wilt incidence being negatively related to increased seeding rate
(Wheeler et al., 2010).
These factors, and others not discussed here, need to be weighed,
along with yield and expected seed and plant losses, in making planting
population decisions in the field.
4.3. Field trial biomass partitioning and production per plant
Dramatic differences in biomass production per plant across popu-
lation densities in our field test reflect the ability of the cotton plant to
adapt physiologically to its system and available resources (Fig. 2). It is
notable that while leaf (P = 0.2854) and stem (P = 0.3374) biomass
trended numerically lower on an aerial basis in lower population den-
sities than in higher densities, reproductive biomass definitively did not
(P = 0.9895). The ratio of reproductive biomass to total biomass was
stable among population density treatments over time (P = 0.8000)
and was quite high by 102 days after planting. These observations
suggest that canopy size or photosynthetic leaf area was not the limiting
factor in reproductive potential or lint yield, even at the lowest popu-
lation density tested. In a comparison of high- and low-density cotton,
Pettigrew et al. (2013) found that increased light interception by a
higher-density cotton crop was offset by the ability of the leaves of a
lower-density crop to utilize sunlight more efficiently, creating a
counteracting effect that equalized productivity and yield across den-
sity treatments. In a comparison of older and newer cotton cultivars,
Pettigrew et al. (2013) found that newer cultivars have higher lint
percentage and lint index than older cultivars, but with similar amounts
of vegetative biomass.
4.4. Summary
Studies on the effects of population density in cotton have been
carried out around the world in many environments, though no work
had been done to quantitively synthesize the yield data collectively, to
better pinpoint a population density threshold for lint yield. And, no-
tably, little had been reported on the effects of population density in
lower-yielding dryland environments. The dryland data collected in this
study, in a semi-arid environment in Texas, showed that yield was not
affected by population density, similar to higher-yielding environments
over the same range in density. Following normalization of all data,
including literature and dryland datasets, a breakpoint at 35,000 plants
ha−1 was identified, which can be interpreted as the minimum popu-
lation density at which yield may be optimized. This rate is lower than
the common population density recommendation for cotton of about
81,000 plants ha−1 and substantially lower than the extremely high
densities at which many producers plant (129,000 seeds ha−1 or
greater) in challenging growing conditions. The analysis showed that
yield will decline precipitously below 35,000 plants ha−1, exposing the
enormous risk to cotton producers in approaching this low density,
particularly if significant seed or plant loss is expected. However, the
analysis suggests that excessive over-seeding may be occurring in many
cases, resulting in economic losses to producers. The analysis also
provides guidance on a threshold for producers facing replanting de-
cisions.
C. Adams et al.
Acknowledgements
We appreciate the agronomic expertise of Jonathan Ramirez,
Tamara Royer, and Joseph Ramirez and acknowledge their contribu-
tions in carrying out the field operations of this project. This work was
supported byCotton Inc. and the Texas State Support Committee (grant
number 08-293TX). This work was also supported by Texas A&M
AgriLife Research and the USDA National Institute of Food and
Agriculture, Hatch project1011694.
References
Adams, C., Erickson, J., Singh, M., 2015. Investigation and synthesis of sweet sorghum
crop responses to nitrogen and potassium fertilization. Field Crops Res. 178, 1–7.
Bednarz, C.W., Bridges, D.C., Brown, S.M., 2000. Analysis of cotton yield stability across
population densities. Agron. J. 92, 128–135.
Bednarz, C.W., Shurley, W.D., Anthony, W.S., Nichols, R.L., 2005. Yield, quality, and
profitability of cotton produced at varying plant densities. Agron. J. 97, 235–240.
Bednarz, C.W., Nichols, R.L., Brown, S.M., 2006. Plant density modified within-canopy
cotton fiber quality. Crop Sci. 46, 950–956.
Collins, G., 2015. Seeding Rates and Plant Populations. NC State Extension. Published
online:. (Accessed 26 April 2018). https://cotton.ces.ncsu.edu/2015/04/seeding-
rates-and-plant-populations-collins-edmisten/.
Dai, J., Li, W., Tang, W., Zhang, D., Li, Z., Lu, H., Eneji, A.E., Dong, H., 2015.
Manipulation of dry matter accumulation and partitioning with plant density in re-
lation to yield stability of cotton under intensive management. Field Crops Res. 180,
207–215.
DeltaPine, 2017. AgKnowledge Spotlight: Cotton Planting Populations and Row Spacing.
Published online. (Accessed 26 April 2018). http://www.aganytime.com/Pages/
Article.aspx?article=1288.
Donahoe, M., 2010. Calculating Cotton Seeding Rates. UF/IFAS Northwest District Office.
Published online. (Accessed: 26 April 2018). http://lyra.ifas.ufl.edu/LyraServlet?
command=getNewsletter&oid=634479&path=0.0&countyID=district1.ifas.ufl.
edu.
Edwards, J.T., Purcell, L.C., 2005. Soybean yield and biomass responses to increasing
plant population among diverse maturity groups: I. Agronomic characteristics. Crop
Sci. 45, 1770–1777.
Feng, L., Mathis, G., Ritchie, G., Han, Y., Li, Y., Wang, G., Zhi, X., Bednarz, C.W., 2014.
Optimizing irrigation and plant density for improved cotton yield and fiber quality.
Agron. J. 106, 1111–1118.
Gwathmey, C.O., Steckel, L.E., Larson, J.A., Mooney, D.F., 2011. Lower limits of cotton
seeding rates in alternative row widths and patterns. Agron. J. 103, 584–592.
Jones, M.A., Wells, R., 1998. Fiber yield and quality of cotton grown at two divergent
population densities. Crop Sci. 38, 1190–1195.
16
Field Crops Research 230 (2019) 11–16
Kaggwa-Asiimwe, R., Andrade-Sanchez, P., Wang, G., 2013. Plant architecture influences
growth and yield response of upland cotton to population density. Field Crops Res.
145, 52–59.
Larson, J.A., Gwathmey, C.O., Roberts, R.K., Hayes, R.M., 2004. Effects of plant popu-
lation density on net revenues from ultra-narrow-row cotton. J. Cotton Sci. 8, 69–82.
Maddonni, G.A., Otegui, M.E., Cirilo, A.G., 2001. Plant population density, row spacing
and hybrid effects on maize canopy architecture and light attenuation. Field Crops
Res. 71, 183–193.
NCCA, 2017. National Cotton council of America. Cotton Production Costs and Returns:
United States. Published online. http://www.cotton.org/econ/cropinfo/
costsreturns/usa.cfm.
Pettigrew, W.T., Johnson, J.T., 2005. Effects of different seeding rates and plant growth
regulators on early-planted cotton. J. Cotton Sci. 9, 189–198.
Pettigrew, W.T., Meredith, W.R., Zeng, L., 2013. Response of obsolete and modern cotton
genotypes to varying plant densities. J. Cotton Sci. 17, 253–262.
Shaver, B.R., Agudelo, P., Martin, S.B., 2017. Damage functions for sting nematode
(Belonolaimus longicaudatus) on bermudagrass turf. Int. Turfgrass Soc. Res. J. 13,
517–523.
Shurley, W.D., Smith, A.R., Culpepper, A.S., Roberts, P.M., et al., 2010. A budget analysis
of glyphosate-resistant palmer amaranth control and seed technology choices in
Georgia cotton. P. 467-472 Boyd, S. (Ed.), Proceedings of the 2010 Beltwide Cotton
Conference.
Siebert, J.D., Stewart, A.M., 2006. Influence of plant density on cotton response to me-
piquat chloride application. Agron. J. 98, 1634–1639.
Siebert, J.D., Stewart, A.M., Leonard, B.R., 2006. Comparative growth and yield of cotton
planted at various densities and configurations. Agron. J. 98, 562–568.
Stewart, S., 2005. Suggested guidelines for plant growth regulator use on Louisiana
cotton. Louisiana Cooperative Extension Service. Publication Number 2919.
Street, J.E., Buchanan, G.A., Crowley, R.H., McGuire, J.A., 1981. Influence of cotton
(Gossypium hirsutum) densities on competitiveness of pigweed (Amaranthus spp.) and
sicklepod (Cassia obtusifolia). Weed Sci. 29, 253–256.
Tetio-Kagho, F., Gardner, F.P., 1988. Responses of maize to plant population density. II.
Reproductive development, yield, and yield adjustments. Agron. J. 80, 935–940.
Toms, J.D., Lesperance, M.L., 2003. Piecewise regression: a tool for identifying ecological
thresholds. Ecology 84, 2034–2041.
USDA-ERS, 2017. Recent Trends in GE Adoption. Updated in July 12, 2017. Published
online. https://www.ers.usda.gov/data-products/adoption-of-genetically-
engineered-crops-in-the-us/recent-trends-in-ge-adoption/.
Wheeler, T.A., Woodward, J.E., Mullinix, B.G., 2010. Effects of seeding rate on verti-
cillium wilt incidence, yield, and value for three cotton cultivars. J. Cotton Sci. 14,
173–180.
Wrather, J.A., Phipps, B.J., Stevens, W.E., Phillips, A.S., Vories, E.D., 2008. Cotton
planting date and plant population effects on yield and fiber quality in the Mississippi
Delta. J. Cotton Sci. 12, 1–7.
Zhi, X., Han, Y., Li, Y., Wang, G., Du, W., Li, X., Mao, S., Feng, L., 2016. Effects of plant
density on cotton yield components and quality. J. Integr. Agric. 15, 1469–1479.