the effect of genotype, environment and time of harvest on
TRANSCRIPT
The effect of genotype, environment and time of harvest on
sugarcane yields in Florida, USA
R.A. Gilbert a,*, J.M. Shine Jr.b, J.D. Miller c, R.W. Rice b, C.R. Rainbolt a
aUniversity of Florida, EREC, 3200 E. Palm Beach Road, Belle Glade, FL 33430, USAb Sugar Cane Growers Cooperative of Florida, P.O. Box 666, Belle Glade, FL 33430, USAcUSDA-ARS, Sugarcane Field Station, 12990 US Hwy. 441, Canal Point, FL 33438, USA
Received 11 February 2005; received in revised form 11 February 2005; accepted 12 February 2005
Abstract
Sugarcane (Saccharum spp.) is grown across different production environments and is harvested over a 5-month (mid-
October–mid-March) period in Florida. While many studies have examined genotype � environment interactions and their
implications for breeding program design, knowledge is limited regarding interactions of genotype, environment and time of
harvest and their implications for growers. Three non-confounded data sets (‘‘case studies’’) were analyzed to determine the
effects of these three factors on kilograms of sugar per ton (KST), tons of cane per hectare (TCH) and tons of sugar per hectare
(TSH) for recently released cultivars in south Florida. Cultivar (genotype), environment, time of harvest and their interactions
had significant effects on KST, TCH and TSH. Sugarcane KST and TSH were reduced by 28 and 29%, respectively, when
harvested in mid-October compared to optimum harvest dates in February. TSH varied from 2 to 46% across environments. The
Lakeview ‘‘warmland’’ site near Lake Okeechobee recorded significantly higher TCH and TSH than other sites, and cultivars
CP88-1508 and CP88-1834 recorded higher relative yields at Lakeview. Cultivar TSH varied up to 51% across the case studies.
CP89-2143 had significantly higher KST than other cultivars in all 21 pairwise comparisons in the three case studies, and a
remarkably high, stable KST ranking across environments. Growers in the Everglades Agricultural Area interested in improving
sugarcane crop sucrose concentration should plant CP89-2143. However, significant genotype � environment interactions for
other cultivars support continued multi-locational evaluation of sugarcane germplasm both during the breeding program and
following cultivar release.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Sugarcane; Genotype; Environment; Time of harvest
www.elsevier.com/locate/fcr
Field Crops Research 95 (2006) 156–170
* Corresponding author. Tel.: +1 561 993 1535;
fax: +1 561 993 1582.
E-mail address: [email protected] (R.A. Gilbert).
0378-4290/$ – see front matter # 2005 Elsevier B.V. All rights reserved
doi:10.1016/j.fcr.2005.02.006
1. Introduction
Sugarcane (Saccharum spp.) is harvested over a
5-month period (mid-October–mid-March), across
differing agroecologies in the Everglades Agricultural
Area (EAA) of south Florida. Both environment
.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 157
and time of harvest influence sugarcane sucrose
accumulation. Growers must factor in sugarcane
genotype and environment when scheduling their
harvests.
Environmental effects on sugarcane yields may be
due to differing nutrient deficiencies (Anderson et al.,
1995), disease pressures (Magarey and Mewing, 1994)
or climatic differences between locations. However,
the vast majority of studies addressing the effects of
location on sugarcane yields have focused on the
interaction of genotype � environment. Although
numerous studies have reported significant G � E
interactions and recommended sugarcane selection in
differing environments (Arceneaux and Hebert, 1943;
Glaz et al., 1985; Milligan et al., 1990; Bull et al.,
1992; Mirzawan et al., 1994; Bissessur et al., 2000),
other studies have concluded that the number of
environments in sugarcane breeding programs could
be reduced (Gravois and Milligan, 1992; Milligan,
1994; Jackson and McRae, 1998; De Sousa-Vieira and
Milligan, 1999).
For example, Jackson and McRae (1998) con-
cluded that selecting families based on broad rather
than specific environmental adaptation was a more
productive breeding strategy in Australia. Milligan
(1994) reported that the number of final testing sites in
the Louisiana breeding program could be reduced
from 13 to 10 and still maintain 96% sucrose yield
repeatability. De Sousa-Vieira and Milligan (1999)
found that family � environment (environment-years)
variances had a minor effect on gains through
selection. However, Bull et al. (1992) concluded that
G � E interaction was large enough to reduce gains
made from selection at a central experiment station.
Mirzawan et al. (1994) tested G � E interaction
repeatability in Australia, and found that each
environment generated a different pattern of discri-
mination among the cultivars.
While G � E interactions have been studied
primarily as a decision aid for the design of sugarcane
breeding programs, less information has been pub-
lished in the scientific literature on G � E interactions
of released commercial sugarcane germplasm. In
Mauritius, Julien and Delaveau (1977) examined
sugarcane dry matter and sucrose content at three
dates of harvest for three varieties in four locations.
Partitioning of dry matter was affected more by date of
harvest and cultivar than by location. Chang (1996)
studied brix, purity and sugar content for six cultivars
grown at six locations for three crop years in Taiwan.
The G � E and G � E � year interactions were
significant for brix and sugar content, but not for
purity.
In Australia, Ellis et al. (2001) used district level
mill production data to examine sugarcane productiv-
ity trends over a 38-year period. The G � E interaction
term accounted for 4% of total variation in TCH, 5%
of KST and 7% of TSH. Cultivars performed well in
the first few years following release but poorly in late
ratoons. The authors noted that aggregated production
data confounded variety performance with other
factors. Ellis et al. (2004) also compared variety
trials to commercial production data in Australia.
Anomalies in ranking between data sets were
attributed to uneven deployment of clones in
commercial fields. Released variety trials have been
conducted in South Africa since 1966 (Redshaw,
2000) to produce varietal recommendation domains
for growers.
The examination of the significance of environment,
genotype and G � E interaction in recently released
germplasm is important both for grower choice of
cultivars and for verification of breeding program
results. In addition, since breeding programs often lack
the resources to allow replanting of the same cultivars at
the same location (Brown and Glaz, 2001), data sets in
which environment, crop age and year are not
confounded are often limited or simply not available.
Since the south Florida harvest season extends over
a 5-month period, it is also important to examine
trends of sucrose accumulation and yield over time as
well as space, as sugarcane harvested prior to
physiological maturity will not have reached peak
sucrose content (Miller and James, 1977; Gilbert et al.,
2004).
The objective of this experiment was to determine
the effect of genotype, environment, time of harvest
and their interactions on sucrose content and yield in
commercial sugarcane germplasm. Three non-con-
founded data sets (termed ‘‘case studies’’) were
examined separately to strengthen the inference base
for the analysis. While agronomic studies usually
report the results of a single experiment, the
magnitude of data collected in this project allowed
for the analysis of a total of three separate experi-
mental data sets. Each data set was non-confounded
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170158
with respect to environment (different locations
within any given case study employ the same
year, crop age, cultivars, and biweekly harvest
periods). We chose to use all available data (three
data sets rather than one) to strengthen the inference
base for the analysis. This approach allowed us to
determine if the environmental effect on sugarcane
growth and its interaction with genotype and time of
harvest were consistent across different experimental
conditions.
2. Materials and methods
The trial was conducted in Florida at five locations
(some locations were repeated in different years)
across two different cropping seasons (Table 1). The
data analyzed for this experiment was part of a larger
cultivar � time sucrose accumulation trial (Gilbert
et al., 2004). From the larger data set, it was possible to
construct a total of three unique data sets (case studies)
to examine the non-confounded effect of environment
in each case study. The same year, crop age, cultivars
and sampling times were used for the different
environments in each case study. The effects of year
and crop age were thus not included in the statistical
analysis. The EREC environment was located at
268390N, 808380W, with a Lauderhill muck (euic,
hyperthermic Lithic Haplosaprist) soil type. The
Hillsboro environment was located at 268310N,
808280W, also with a Lauderhill muck soil. The
Lakeview environment was located at 268480N,
808410W, with a Torry muck (euic, hyperthermic,
Typic Haplosaprist) soil type. This environment is
Table 1
The three case studies
Case study Planting date Locations Year Cr
1 9 September 1997 Lakeview 1999/2000 Fir
21 November 1997 Hundley
8 January 1998 EREC
2 19 December 2000 EREC 2001/2002 Pla
31 October 2000 Hillsboro
3 16 November 1999 Hundley 2001/2002 Fir
31 August 1999 Lakeview
21 December 1999 Sundance
Within a given case study, calendar year, crop age, cultivar and sampling
regarded as ‘‘warmland muck’’ due to its proximity to
Lake Okeechobee. The lake effect reduces the risk of
freezes at this site. None of the cropping seasons
included in this study included a freeze event. The soil
at the Lakeview site also has a higher mineral content
(68–75%) than the other sites (20–48%). The Hundley
site was located at 268410N, 808270W, with a Lauderhill
muck soil. The Sundance site was located at 268360N,
808520W, also with a Lauderhill muck soil. The
Sundance environment is a ‘‘transitional area’’ with
higher sand content than the other Histosols included in
this study.
Cultivars included in this study, from the ‘‘CP’’
breeding program at Canal Point, FL, were released
from 1977 to 1996. They comprised > 78% of the
Florida sugarcane acreage in 2003 (Table 2), and were
thus representative of the range of commercial
germplasm used in Florida.
All experiments were planted in a randomized
block design using three or four replications with
cultivar as the main treatment. Table 1 lists the
planting date for each site. All sites were mechanically
harvested in mid-March of each year. Each plot was 4
rows wide by 10.7 m long with 1.5 m row spacing.
Plant populations were determined by whole-plot stalk
counts performed in July or August of each season.
Millable stalks with significant diameter at observer
breast height are normally counted during July and
August in Florida when stalk population has
stabilized, and lodging from tropical storms has not
yet occurred. Five-stalk plant samples were harvested
from each plot on 11 occasions roughly at 2-week
intervals starting from October 14. Sampling dates
were grouped by 2-week intervals (biweeks). Biweek
op age Cultivars
st ratoon CP 70-1133, CP 72-1210, CP 72-2086, CP 80-1743,
CP 80-1827, CP 88-1508, CP 88-1762, CP 89-2143
nt CP 72-2086, CP 78-1628, CP 80-1743, CP 84-1198,
CP 85-1382, CP 88-1762, CP 89-2143, CP 89-2377
st ratoon CP 72-2086, CP 80-1743, CP 84-1198, CP 85-1382,
CP 88-1762, CP 88-1834, CP 89-2143, CP 89-2377
dates were consistent across different environments (locations).
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 159
Table 2
Commercial cultivars included in the three case studies including year of commercial release, Crop Science registration reference, percent of
Florida 2003 acreage and sucrose yield as percent of check when released
Cultivar Year released Crop Sci. Reference 2003 FL areaa (%) TSH yield (%) (check)
CP 70-1133 1977 1978, 18:526 1.1 136 (CP 63-588)
CP 72-1210 1980 1981, 21:797 <1.0 125 (CP 63-588)
CP 72-2086 1982 1984, 24:210 9.1 113 (CP 63-588)
CP 78-1628 1989 1991, 31:236 12.3 91 (CP 70-1133)
CP 80-1743 1989 1991, 31:235 28.7 88 (CP 70-1133)
CP 80-1827 1987 1990, 30:232 <1.0 106 (CP 70-1133)
CP 84-1198 1992 1994, 34:1404 4.8 100 (CP 70-1133)
CP 85-1382 1993 1995, 35:1214 <1.0 109 (CP 70-1133)
CP 88-1508 1995 1997, 37:1387 <1.0 106 (CP 70-1133)
CP 88-1762 1995 1997, 37:1388 11.4 100 (CP 70-1133)
CP 88-1834 1996 2000, 40:576 <1.0 105 (CP 70-1133)
CP 89-2143 1996 2000, 40:577 10.7 117 (CP 70-1133)
CP 89-2377 1996 2000, 40:577 <1.0 107 (CP 70-1133)
a Glaz and Vonderwell (2004).
1 was October 14–October 27, biweek 2 (October 28–
November 10), biweek 3 (November 11–November
24), biweek 4 (November 25–December 8), biweek 5
(December 9–December 22), biweek 6 (December
23–January 6), biweek 7 (January 7–January 20),
biweek 8 (January 21–February 3), biweek 9
(February 4–February 17), biweek 10 (February 18–
March 3) and biweek 11 (March 4–March 17). For
Case Study 2, inclement weather and logistical
difficulties prevented sampling events during biweeks
1, 4 and 5.
Harvested sugarcane samples were topped in the
field, and millable fresh stalk weight was recorded.
Plant fresh weights were used to determine individual
stalk weight (kg stalk�1), and tons of cane per hectare
(TCH, t ha�1) were calculated as the product of stalk
number and stalk weight. To determine sucrose
concentration (KST, kg sucrose t�1), the harvest
samples were ground and the crusher juice analyzed
for Brix and pol. Brix, a measure of % soluble solids,
was measured using a refractometer which automa-
tically corrected for temperature (Meade, 1963). Pol, a
unitless measure of the polarization of the sugar
solution, was measured using a saccharimeter. Sucrose
yield (TSH, t sucrose ha�1) was calculated as the
product of TCH and KST (divided by 1000 to convert
kg sucrose to metric tons). A theoretical economic
index (EI, US$ ha�1) was also calculated which takes
into account harvesting, transport and milling costs
associated with TCH (Deren et al., 1995). This index is
not an exact measure of grower profits but rather a
method to rank genotypes based on economic criteria.
Repeated measures analysis of variance (Littell et al.,
2002) using SAS1 was performed to determine the
significance of the main effects of genotype, environ-
ment, and time, and the interaction of environ-
ment � time and genotype � environment. The proc
mixed procedure in SAS was used to perform the
repeated measures ANOVA. The covariance model for
each analysis was chosen based on procedures
outlined by Littell et al. (2002, pp. 281–285). The
interaction of genotype � time was analyzed sepa-
rately (Gilbert et al., 2004).
Paired t-tests were performed using least square
means estimates in SAS1 (Littell et al., 2002) to
contrast each pair of treatment means individually for
all effects. Due to space limitations, these pairwise
contrasts are discussed in the text, but are not
presented in tables. Instead, the least significant
difference at P = 0.05 (LSD.05) value is presented.
Cultivar KST, TCH and TSH rankings between
environments were correlated for the genoty-
pe � environment interaction means using Spear-
man’s rank correlation coefficient (rs):
rs ¼ 1 � 6X
d2� �
=ðn3 � nÞ (1)
where d is the difference in the rank of a given cultivar
between two environments, n the number of cultivars
in common between two environments.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170160
Where noted, statistical significance refers to
P < 0.05 (*), P < 0.01 (**) and P < 0.001 (***) levels.
3. Results
3.1. Environment
The effect of environment was significant
(P < 0.01) on KST, TCH and TSH in all three Case
Studies (Table 3). Due to the extremely large F-test
values (Table 3) for the main effects of environment
(F-test range 26–104; mean 58.2), time (5.4–409;
111), and genotype (0.6–74; 19.6) relative to the
interaction terms of environment � time (1.04–38;
10.6) and genotype � environment (1.97–19.7; 6.77)
both main effects and interaction terms are discussed
when statistically significant. Three-way interactions
Table 3
Repeated measures ANOVA results for KST, TCH and TSHa for the thre
Case Study 1 C
DF F-test D
KST
Environment (E) 2 26.1***
Time (T) 10 311***
Genotype (G) 7 21.7***
Rep 2 2.65 NS
E � T 20 7.45***
G � E 14 4.31***
G � E � T 140 1.02 NS 4
TCH
Environment (E) 2 45.4***
Time (T) 10 13.8***
Genotype (G) 7 0.70 NS
Rep 2 7.06**
E � T 20 1.42 NS
G � E 14 2.57*
G � E � T 140 1.15 NS 4
TSH
Environment (E) 2 48.8***
Time (T) 10 90.3***
Genotype (G) 7 0.64 NS
Rep 2 4.67*
E � T 20 2.51***
G � E 14 1.97 NS
G � E � T 140 1.12 NS 4
a KST: kg sucrose t�1; TCH: t cane ha�1; TSH: t sucrose ha�1.* Indicates F-test significance at the P < 0.05 level.
** Indicates F-test significance at the P < 0.01 level.*** Indicates F-test significance at the P < 0.001 level.
of genotype � environment � time were not signifi-
cant (F-test range 0.69–1.54; mean 1.12) in the
majority of cases and are not discussed. In Case Study
1, the EREC site had significantly lower KST, TCH
and TSH compared to the Hundley and Lakeview sites
(Table 4). However, in Case Study 2, the EREC site
had significantly higher KST, TCH and TSH than the
Hillsboro site. The ranking of sites in Case Study 3 for
mean KST, TCH and TSH was Lakeview > Hundley
> Sundance.
In addition to higher soil mineral content, the
Lakeview site is distinguished by its location in the
‘‘warmland’’ muck area <1 km from Lake Okeecho-
bee. Air temperature data was not collected at all the
study sites, however 1999 and 2001 weather station
data from the University of Florida EREC (0.5 km
from the EREC test site) and the USDA-ARS
sugarcane field station at Canal Point (also <1 km
e case studies
ase Study 2 Case Study 3
F F-test DF F-test
1 53.9*** 2 104***
7 51.4*** 10 409***
7 26.1*** 7 74.1***
3 1.25 NS 3 6.17**
7 2.56* 20 6.35***
7 4.58** 14 19.7***
9 1.54* 140 1.27*
1 10.0** 2 91.0***
7 8.63*** 10 5.37***
7 16.2*** 7 4.23**
3 3.37* 3 0.90 NS
7 38.1*** 20 1.04 NS
7 7.92*** 14 4.65***
9 0.69 NS 140 1.11 NS
1 41.3*** 2 104***
7 24.0*** 10 89.6***
7 20.9*** 7 7.99***
3 4.11* 3 0.50 NS
7 33.1*** 20 2.91***
7 11.4*** 14 3.88***
9 0.98 NS 140 1.24*
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 161
Table 4
Sugarcane sucrose content, biomass yield, sucrose yield and economic index for the three case studies averaged by environment (location)
Location KST (kg sucrose t�1) TCH (t cane ha�1) TSH (t sucrose ha�1) EI (US$ ha�1)
Case Study 1
EREC 128 ca 87 b 11.2 b 1940 b
Hundley 136 a 124 a 16.9 a 3380 a
Lakeview 132 b 125 a 16.5 a 3230 a
Case Study 2
EREC 132 a 133 a 17.7 a 3510 a
Hillsboro 125 b 125 b 15.8 b 2930 b
Case Study 3
Hundley 131 b 136 b 18.0 b 3560 b
Lakeview 134 a 151 a 20.2 a 4140 a
Sundance 125 c 87 c 11.0 c 1900 c
a Means within a case study followed by the same letters are not significantly different (P < 0.05).
from Lake Okeechobee), document a 2.3–2.8%
warmer average annual temperature at Canal Point,
and a 7.0–8.8% higher average annual cumulative
thermal time (using a base temperature of 10 8C) at
Canal Point. Sugarcane leaf development and tonnage
is strongly influenced by temperature (Inman-Bamber,
1994; Bonnett, 1998; Sinclair et al., 2004) thus the
documented temperature differences are consistent
with the higher yields measured at Lakeview. High
yields at Lakeview also translated into higher
economic indices at this site (Table 4).
3.2. Time of harvest
Time of harvest is an important consideration in
commercial use of varieties. Given present milling
capacity in south Florida, a full 5 months are required
Table 5
Sucrose content by time of harvest (biweek) for all three case studies
Biweek Period Case Study 1 (kg sucrose t�1)
1 October 14–October 27 103
2 October 28–November 10 116
3 November 11–November 24 122
4 November 25–December 8 130
5 December 9–December 22 132
6 December 23–January 6 137
7 January 7–January 20 141
8 January 21–February 3 140
9 February 4–February 17 143
10 February 18–March 3 144
11 March 4–March 17 144
LSD.05 1.51
to process the 450,000 acres planted to sugarcane
(Saccharum spp.). Some sugarcane must be harvested
before achieving physiological maturity to sustain early
season (October–November) milling operations.
‘‘Early maturing’’ cultivars are preferentially harvested
during this time, recognizing that they may not have
reached their peak sucrose content, but may have higher
sucrose content than other later-maturing cultivars at
the onset of mill operations (Gilbert et al., 2004).
Time of harvest had a highly significant (P < 0.001)
effect on KST, TCH and TSH in all three case studies
(Table 3). KSTincreased significantly with each biweek
up to biweek 6 in Case Study 1 (Table 5). In Case Study
2, biweek 2 (the first sampling date for this case study)
and 3 had significantly lower KST than later harvest
periods. For Case Study 3, KSTwas significantly lower
in biweeks 1–5 than subsequent dates in 38 of 40
Case Study 2 (kg sucrose t�1) Case Study 3 (kg sucrose t�1)
– 100
117 113
124 125
– 130
– 135
132 136
131 137
134 140
133 141
132 140
127 136
1.82 1.34
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170162
Table 6
Sucrose yield and economic index by time of harvest (biweek) for all three case studies
Biweek Period Case Study 1
(t sucrose ha�1)
Case Study 2
(t sucrose ha�1)
Case Study 3
(t sucrose ha�1)
Case Study 1
(US$ ha�1)
Case Study 2
(US$ ha�1)
Case Study 3
(US$ ha�1)
1 October 14–October 27 10.5 – 12.2 1390 – 1630
2 October 28–November 10 12.4 14.5 13.5 1940 2470 2180
3 November 11–November 24 13.2 15.8 15.2 2300 2890 2810
4 November 25–December 8 14.7 – 16.3 2770 – 3170
5 December 9–December 22 15.6 – 17.1 3020 – 3430
6 December 23–January 6 15.7 17.8 17.4 3110 3520 3570
7 January 7–January 20 15.9 16.6 17.7 3250 3230 3640
8 January 21–February 3 15.2 17.1 17.5 3070 3410 3640
9 February 4–February 17 16.6 17.6 18.1 3450 3510 3790
10 February 18–March 3 16.9 16.9 18.0 3540 3300 3760
11 March 4–March 17 16.8 17.7 17.7 3510 3420 3590
LSD.05 0.62 0.52 0.67 190 140 170
pairwise mean comparisons (Table 5). Sugarcane
harvested in mid-October (biweek 1) had 29% lower
sucrose concentration (100–103 kg sucrose t�1) than
optimal harvest dates (141–144 kg sucrose t�1).
The trend of TSH over time followed a similar trend
to KST. In Case Study 1, sugarcane harvested from
biweeks 1–4 had significantly lower TSH (10.5–
14.7 t sucrose ha�1) in 33 of 34 pairwise contrasts
relative to later biweeks (15.2–16.9 t sucrose ha�1;
Table 6). Similarly, sugarcane harvested during biweeks
2 and 3 in Case Study 2 (14.5–15.8 t sucrose ha�1) and
biweeks 1–4 in Case Study 3 (12.2–16.3 t sucrose ha�1)
had significantly lower TSH than all subsequent
sampling dates (Table 6). Compared to time of optimum
TSH, harvesting sugarcane in mid- to late-October
resulted in yield reductions of 37% (Case Study 1), 18%
(Case Study 2) and 33% (Case Study 3). Economic
indices in mid-October were only 40% those in mid-
February (Table 6).
3.3. Genotype
The effect of genotype was highly significant
(P < 0.001) on KST in all three case studies (Table 3),
and significant (P < 0.01) on TCH and TSH in Case
Studies 2 and 3.
Across all three case studies, CP89-2143 consis-
tently averaged significantly greater KST (139–
144 kg sucrose t�1) than any other cultivar. Compared
to other cultivars, CP89-2143 KST was 7–13% greater
in Case Study 1, 2–12% greater in Case Study 2 and 5–
16% greater in Case Study 3.
In the two case studies including CP85-1382, the
cultivar consistently rankedat ornear the bottomofKST
responses. Although CP89-2143 KST was greatest
across all cultivars and case studies, CP88-1762 KST
was also favorably high in Case Study 1 (132 kg
sucrose t�1) and Case Study 3 (135 kg sucrose t�1).
CP72-2086 KST profiles were consistently high across
all three case studies, ranging from 133 to 136 kg
sucrose t�1.
TCH values were similar across cultivars in Case
Study 1. CP89-2377 and CP80-1743 had the highest
TCH levels that exceeded 140 t cane ha�1 in Case
Study 2. CP89-2377 is a high tonnage, low KST cultivar
(Gilbert et al., 2004). CP88-1762 and CP89-2143 were
the only cultivars approaching 140 t cane ha�1 in Case
Study 3, while CP85-1382 was notable for its poor TCH
(109–114 t cane ha�1) relative to other cultivars in both
Case Studies 2 and 3.
Although TSH for CP 89-2143 numerically
exceeded other cultivars, there were no significant
differences in pairwise comparisons of cultivar TSH in
Case Study 1 (Table 7). CP78-1628, CP89-2143 and
CP89-2377 all had TSH > 18.0 t sucrose ha�1 in Case
Study 2, which was significantly greater than CP85-
1382, CP72-2086 and CP84-1198 (Table 7). CP85-
1382 was notable for having the lowest TSH (13.4 t
sucrose ha�1) in Case Study 2. CP85-1382 (14.4 t
sucrose ha�1) and CP88-1834 (13.0 t sucrose ha�1)
had the lowest TSH in Case Study 3 (Table 7). In a
related study assessing cultivar sucrose accumulation
over time, CP85-1382 also performed poorly (Gilbert
et al., 2004). Economic indices of CP85-1382 were
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 163
Table 7
Sucrose yield and economic index by genotype for all three case studies
Cultivar Case Study 1
(t sucrose ha�1)
Case Study 2
(t sucrose ha�1)
Case Study 3
(t sucrose ha�1)
Case Study 1
(US$ ha�1)
Case Study 2
(US$ ha�1)
Case Study 3
(US$ ha�1)
CP 70-1133 15.1 – – 2840 – –
CP 72-1210 14.0 – – 2660 – –
CP 72-2086 15.0 14.7 15.8 2930 2900 3080
CP 78-1628 – 18.5 – – 3740 –
CP 80-1743 15.0 17.6 16.4 2710 3300 3240
CP 80-1827 14.6 – – 2780 – –
CP 84-1198 – 15.3 16.6 – 2780 3210
CP 85-1382 – 13.4 14.4 – 2370 2630
CP 88-1508 14.4 – – 2710 – –
CP 88-1762 14.7 17.8 18.9 2830 3420 3860
CP 88-1834 – – 13.0 – – 2290
CP 89-2143 16.0 18.4 19.7 3350 3820 4180
CP 89-2377 – 18.2 16.5 – 3430 3130
LSD.05 NS 0.88 1.55 380 190 360
only 63% those of CP89-2143, implying a severe
penalty to growers when using the wrong cultivar.
Recently released CP cultivars varied widely in
sucrose accumulation and yield. Superior sucrose
concentration is a favorable trait of CP89-2143, and
has been the main factor for the increase in CP89-2143
acreage in the EAA over the last few years (Glaz and
Vonderwell, 2004). Given the clear economic super-
iority of CP89-2143 to other cultivars (Table 7) its
acreage is likely to continue to increase. In the case of
CP89-2143 the pre-release breeding program data
(Table 2), variety trial data from this study (Table 7),
and commercial census data (Glaz and Vonderwell,
2004) are all indicative of its profitability. In contrast,
CP80-1743 is a profitable cultivar ranked #1 in Florida
sugarcane acreage, yet it performs poorly in small
plots (Tables 2 and 7).
3.4. Environment � time
The interaction of environment � time of harvest
was significant for KST and TSH in all three case
studies, but was not significant for TCH in Case
Studies 1 and 3 (Table 3).
The EREC site had particularly low KST early in
the season compared to other sites in Case Study 1.
Early season KST differences in Case Study 3 were
not significant in biweeks 1–2 between the Lakeview
and Sundance environments. However, Sundance KST
was significantly lower than both Lakeview and
Hundley at every sampling date beginning with
biweek 3.
TCH differences between environments were gen-
erally consistent across sampling times. Differences in
TCH between EREC and both Hundley and Lakeview
were highly significant on every sampling date in Case
Study 1, while Hundley and Lakeview TCH were never
significantly different in this case study. The EREC and
Hillsboro environment TCH were not significantly
different on 10 of 11 dates in the Case Study 2. The
Sundance environment had highly significant reduc-
tions in TCH compared to both Hundley and Lakeview
environments at every sampling date in Case Study 3.
TSH differences between sites also were generally
maintained across sampling times. EREC TSH was
significantly lower than Hundley and Lakeview
throughout the harvest season in Case Study 1. The
Sundance site in Case Study 3 had significantly lower
TSH than Hundley and Lakeview at every sampling
date. The Lakeview site was superior to Hundley on 10
of 11 dates in this case study.
3.5. Genotype � environment
The genotype � environment (G � E) interaction
term was significant for KST and TCH in all three case
studies, and for TSH in Case Studies 2 and 3 (Table 3).
Four of eight cultivars had non-significant differ-
ences in KST between environments in Case Study 1
(Fig. 1A). However, CP80-1743 and CP72-1210 had
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170164
Fig. 1. Genotype (Cultivar) � environment sucrose content (KST) means for (A) Case Study 1, (B) Case Study 2, and (C) Case Study 3.yDifferent letter indicates significant difference between locations for a given cultivar at P < 0.05 level. zNumber indicates cultivar rank at a
given location.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 165
Fig. 2. Genotype (Cultivar) � environment biomass yield (TCH) means for (A) Case Study 1, (B) Case Study 2, and (C) Case Study 3. yDifferent
letter indicates significant difference between locations for a given cultivar at P < 0.05 level. zNumber indicates cultivar rank at a given location.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170166
Fig. 3. Genotype (Cultivar) � environment sucrose yield (TSH) means for (A) Case Study 1, (B) Case Study 2, and (C) Case Study 3. yDifferent
letter indicates significant difference between locations for a given cultivar at P < 0.05 level. zNumber indicates cultivar rank at a given location.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 167
highly significant differences in KST between
Hundley and EREC environments in this case study.
CP72-2086 and CP89-2143 had highly significant
differences in KST between environments in Case
Study 2 (Fig. 1B). CP84-1198 also had highly
significant differences in KST between Lakeview
and both Hundley and Sundance in Case Study 3
(Fig. 1C). CP72-2086 and CP88-1762 had stable KST
across environments in this case study.
The significant interaction terms for genotype �environment implies changes in cultivar ranking for
mean KST across environments. In Case Study 1,
CP72-1210 was ranked seventh in KST at EREC but
second in KST at Lakeview (Fig. 1A). Large shifts in
cultivar KST rank between environments also
occurred for CP80-1743 (Fig. 1B) and CP84-1198
(Fig. 1C).
In contrast, CP89-2143 maintained a remarkably
high, stable KST ranking across all three case studies.
CP89-2143 was ranked first in KST in 7 of the 8
environment-years, and second in one environment-
year (Fig. 1). CP89-2143 has clearly superior KST
compared to recently released CP cultivars across a
wide range of environments in the EAA.
The genotype � environment interaction term for
TCH was significant in all case studies (Table 3).
There were changes in cultivar TCH rank between
EREC, Hundley and Lakeview in Case Study 1
(Fig. 2A). For example, across all cultivars in Case
Study 1, CP88-1508 recorded both the lowest mean
TCH values (EREC) and highest mean TCH values
(Hundley and Lakeview; Fig. 2A). CP70-1133 and
CP80-1743 reflected the greatest stability (no sig-
nificant differences across sites) in TCH trends across
Case Study 1 environments. Although CP72-2086,
CP80-1743 and CP89-2143 recorded significantly
greater TCH at EREC than Hillsboro (Case Study 2;
Fig. 2B), the exact opposite trend was observed for
CP88-1762. In Case Study 3, CP88-1834 was notable
for its high TCH at Lakeview compared to Sundance,
although all cultivars recorded significantly lower
TCH values at Sundance compared to the other
environments in Case Study 3 (Fig. 2C).
Genotype � environment TSH was not significant
(P = 0.073; Table 3) for Case Study 1 (Fig. 3A). CP70-
1133 was the only cultivar not to have significant
differences in TSH between environments in this data
set. CP88-1508 and CP89-2143 had highly significant
reductions in TSH at EREC compared to the other
sites in Case Study 1. The genotype � environment
interaction term was significant for Case Studies 2 and
3 (Table 3). CP72-2086 and CP89-2143 had highly
significant increases in TSH at EREC compared to
Hillsboro (Fig. 3B). In contrast, CP88-1762 had
significantly greater TSH at Hillsboro (ranked first)
than EREC (ranked sixth). In Case Study 3 (Fig. 3C),
the three most recently released cultivars (CP88-1834,
CP89-2143 and CP89-2377) had the three highest
TSH ranks at the Lakeview site. CP88-1834 relative
performance was markedly better at Lakeview (ranked
third) than Hillsboro or Sundance (ranked last).
Spearman rank correlation coefficients (rs) were
significant (>0.83) between Hundley and Sundance
environments for KST and TSH in Case Study 3,
whereas rs was not significant (�0.17 to 0.28) in any
contrast with Lakeview in this case study. Even though
mean yield was closer between the Hundley and
Lakeview sites (Table 4), relative cultivar performance
was closer between Hundley and Sundance. A
significant negative correlation (�0.85) was found
between cultivar TSH rank at EREC versus Lakeview
in Case Study 1. This indicates that Lakeview is
located in a different agroecology than the other sites
included in this study.
4. Discussion
Our results highlight the influence of environment
on sugarcane yields in a visually homogenous region.
The EAA sugarcane production area is characterized
by flat basin topography, well-drained organic soils
with high N mineralization rates, and high to very high
organic matter contents (Bottcher and Izuno, 1994).
Unlike other sugarcane production areas in the world,
rainfall is not considered a limiting factor to sugarcane
production in the EAA due to the excellent water-
holding capacity of the organic soils and abundant
water supply from Lake Okeechobee (Alvarez et al.,
1982). TSH yields averaged over the same cultivars,
growing season, crop age and time of harvest varied
greatly from 2 to 46% among environments. In
contrast to the results of Julien and Delaveau (1977) in
Mauritius, this study supports arguments for multi-
locational evaluation of sugarcane germplasm in
Florida both during the breeding program and
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170168
following cultivar release. South Africa has had a
released variety trial program in place since 1966
(Redshaw, 2000) to recommend cultivars to growers in
different agroclimatic zones. Released variety trials
make inherent sense in S. Africa where 23 bioresource
regions have been identified in Kwa-Zulu Natal
province, and sugarcane production areas range from
loamy sandy soils in warm coastal areas to clayey soils
in cooler highlands. This study indicates that a similar
approach to released variety trials may be useful in
more homogenous regions.
Temporal trends in KST and TSH were consistent
with other investigations on older germplasm (Miller
and James, 1977; Rice, 1974) which found that
sugarcane harvested prior to mid-December in the
EAA will possess sub-optimal sucrose content.
Grower harvest strategies should include harvesting
cultivars with high ‘‘early season’’ sucrose levels
during October and November (Gilbert et al., 2004).
Cox et al. (1998) advocated a similar strategy to
increase grower gross returns in Australia. Both
selection and post-release trials should be harvested at
different stages of the season to identify such cultivars.
Although time of harvest had a significant effect on
TCH, there were no clear trends. One reason for this
complexity is that fresh weights, not dry weights, are
customarily measured for sugarcane. Crop fresh
weight may vary both over the season as plant relative
water content changes, and diurnally due to plant
stomatal activity (Liu and Helyar, 2003). Thus both
time of season and time of day affect fresh weight at
harvest. In addition, there is a direct relation in
repeated measures experiments between size of
individual samples and total plot size required. The
five-stalk samples used had higher variability in TCH
than KST, which may have been ameliorated by
increased number of stalks per sample. However
increasing sample size would have greatly increased
the resources required for land and labor to implement
these experiments.
It is interesting to note the extent of variation in
harvest traits of commercially released germplasm.
With the notable exception of CP89-2143, CP
cultivars have generally recorded yields within 10%
of the breeding program check in pre-release variety
trials (Table 2). However, across the case studies, KST
varied 13–18%, TCH 11-36%, and TSH 14–51%. Part
of this variability is due to genetic gain over time.
Edme et al. (2005) estimated that commercial sucrose
yield in the EAA had improved at the rate of
0.10 t sucrose ha�1 year�1 over a 33-yr period, 69%
of which was attributed to the CP breeding program.
However our current study indicates overall gain in
sugarcane yields will not translate into consistently
improved performance from each new cultivar.
Sucrose yields recorded by CP85-1382 and CP88-
1834 were inferior to earlier germplasm.
Sugarcane germplasm is released after numerous
years of replicated on-farm trials, yet considerable
variation in cultivar relative performance may be
expected following cultivar release. Many breeding
programs do not have the resources available to assess
cultivar performance following release. Relative
performance of new cultivars compared to industry
standards is often obtained ad hoc from mill managers
and industry professionals without replicated tests.
Ellis et al. (2004) compared variety trials to
commercial production in Australia, and reported
that differences in cultivar ranking between data sets
were due to uneven deployment of cultivars in
commercial fields. They concluded that variety trials
could not be enhanced to evaluate uneven deployment
effects. However, in S. Africa (Redshaw, 2000) post-
release variety trials have been used to recommend
varieties to growers. Our study indicates that a
systematic agronomic evaluation of released germ-
plasm is valuable in determining relative cultivar
performance and recommendation domains.
The environment � time of harvest interactions
indicate that relative differences in TSH between
environments at the beginning of the harvest season
are unlikely to change, whereas relative differences in
KST between environments may shift during the first
4–6 weeks of the harvest season. KST appears to
fluctuate both over time and space in the first 2 months
of the harvest season. Changes in KST should be
monitored across environments during the early
harvest period to ensure that the most profitable
environments are harvested during that time.
Significant G � E interactions indicated that the
Lakeview site was located in a different agroecolo-
gical zone than the other sites. Differences in soil
depth, mineral content and air temperature may
contribute to G � E interactions in the EAA. Lake-
view is <1 km from Lake Okeechobee in a ‘‘warm-
land’’ area, with soils containing appreciably greater
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170 169
mineral content than the other sites included in this
study. Early breeding strategies in the EAA recog-
nized the importance of selection for both ‘‘warm-
land’’ sites and ‘‘coldland’’ sites further from Lake
Okeechobee (Bourne, 1972). Cultivars F31-962, F31-
436 and CL41-223 occupied over 50% of the EAA
acreage in the 1940–1960s, but faded from promi-
nence as sugarcane acreage spread further from the
lake. Rates of leaf appearance versus thermal time
differ among sugarcane cultivars (Bonnett, 1998;
Sinclair et al., 2004). Differing cultivar growth rates at
different cumulative thermal time may be part of the
mechanism involved in the G � E interaction of
‘‘warmland’’ versus ‘‘coldland’’ sites. Although the
CP program breeds new cultivars in a ‘‘warmland’’
environment adjacent to Lake Okeechobee, all
cultivars are tested in multiple ‘‘coldland’’ areas
and one ‘‘warmland’’ environment before cultivar
release. This data set indicates that a significant G � E
interaction still exists in many recently released
cultivars, with the recommendation domain of CP88-
1508 and CP88-1834 closer to Lake Okeechobee than
CP72-2086 or CP80-1743.
The vast majority of the scientific literature has
focused on the significance of G � E interactions for
breeding program design (Kang and Miller, 1984;
Glaz et al., 1985; Milligan et al., 1990; Bull et al.,
1992; Milligan, 1994; Mirzawan et al., 1994; Jackson
and McRae, 1998; Bissessur et al., 2000; Brown and
Glaz, 2001; Kimbeng et al., 2002). This study
indicates that significant G � E interactions persist
beyond cultivar release in commercially grown
germplasm. The high variation in sugarcane harvest
traits encountered in ‘‘elite’’ germplasm point to the
value of systematic agronomic monitoring following
cultivar release.
Acknowledgements
The authors gratefully acknowledge the assistance
of Mr. Robert Taylor, Mr. Matthew Duchrow, Mr.
Vincent Sampson and Mr. Henry Westcarth in sample
collection and processing. This research was sup-
ported by the Florida Agricultural Experiment Station
and a grant from the Sugar Cane Growers Cooperative
of Florida and approved for publication as Journal
Series No. R-10014.
References
Alvarez, J.A., Crane, D.R., Spreen, T.H., Kidder, G., 1982. A yield
prediction model for Florida sugarcane. Agric. Syst. 9, 161–179.
Anderson, D.L., de Boer, H.G., Portier, K.M., 1995. Identification of
nutritional and environmental factors affecting sugarcane produc-
tion in Barbados. Commun. Soil Sci. Plant Anal. 26, 2887–2901.
Arceneaux, G., Hebert, L.P., 1943. A statistical analysis of varietal
yields of sugarcane obtained over a period of years. Agron. J. 35,
148–160.
Bissessur, D., Tilney-Bassett, R.A.E., Lim Shin Chong, L.C.Y.,
Domaingue, R., Julien, M.H.R., 2000. Family � environment
and genotype � environment interactions for sugarcane across
two contrasting marginal environments in Mauritius. Exp. Agric.
36, 101–114.
Bonnett, G.D., 1998. Rate of leaf appearance in sugarcane, including
a comparison of a range of varieties. Aust. J. Plant Physiol. 25,
829–834.
Bottcher, A.B., Izuno, F.T., 1994. Everglades Agricultural Area
(EAA): Water, Soil, Crop and Environmental Management.
University Press of Florida, Gainesville, FL.
Bourne, B.A., 1972. Significant developments in the early phases of
the Florida cane sugar industry. Sugar Azucar 67, 19–23.
Brown, J.S., Glaz, B., 2001. Analysis of resource allocation in final
stage sugarcane cultivar selection. Crop Sci. 41, 57–62.
Bull, J.K., Hogarth, D.M., Basford, K.E., 1992. Impact of genoty-
pe � environment interaction on response to selection in sugar-
cane. Aust. J. Exp. Agric. 32, 731–737.
Chang, Y.S., 1996. Estimating heritability of and correlations among
brix, purity, and sugar content in sugarcane using balanced
multiple environment and year data. Taiwan Sugar Res. Inst.
Rep. 151, 1–10.
Cox, M.C., Ridge, D.R., Hussey, B., 1998. Optimum time of harvest
for high early CCS sugar varieties. Proc. Aust. Soc. Sugar Cane
Technol. 20, 218–223.
Deren, C.W., Alvarez, J., Glaz, B., 1995. Use of economic criteria
for selecting clones in a sugarcane breeding program. Proc. Int.
Soc. Sugar Cane Technol. 21, 437–447.
De Sousa-Vieira, O., Milligan, S.B., 1999. Intrarow plant spacing
and family � environment interaction effects on sugarcane
family evaluation. Crop Sci. 39, 358–364.
Edme, S.J., Miller, J.D., Glaz, B., Tai, P.Y.P., Comstock, J.C., 2005.
Genetic contributions to yield gains in the Florida sugarcane
industry across 33 years. Crop Sci. 45, 92–97.
Ellis, R.N., Basford, K.E., Cooper, M., Leslie, J.K., Blyth, D.E., 2001.
A methodology for analysis of sugarcane productivity trends. I.
Analysis across districts. Aust. J. Agric. Res. 52, 1001–1009.
Ellis, R.N., Basford, K.E., Leslie, J.K., Hogarth, D.M., Cooper, M.,
2004. A methodology for analysis of sugarcane productivity
trends. 2. Comparing variety trials with commercial productiv-
ity. Aust. J. Agric. Res. 55, 109–116.
Gilbert, R.A., Shine Jr., J.M., Miller, J.D., Rice, R.W., 2004. Sucrose
accumulation and harvest schedule recommendations for CP
sugarcane cultivars. Online. Crop Management. doi: 10.1094/
CM-2004-0402-01-RS.
Glaz, B., Vonderwell, J., 2004. Sugarcane variety census: Florida
2003. Sugar J. 67, 11–19.
R.A. Gilbert et al. / Field Crops Research 95 (2006) 156–170170
Glaz, B., Miller, J.D., Kang, M.S., 1985. Evaluation of cultivar-testing
environments in sugarcane. Theor. Appl. Genet. 71, 22–25.
Gravois, K.A., Milligan, S.B., 1992. Genetic relationships between
fiber and sugarcane yield components. Crop Sci. 32, 62–67.
Inman-Bamber, N.G., 1994. Temperature and seasonal effects on
canopy development and light interception of sugarcane. Field
Crops Res. 36, 41–51.
Jackson, P.A., McRae, T.A., 1998. Gains from selection of broadly
adapted and specifically adapted sugarcane families. Field Crops
Res. 59, 151–162.
Julien, M.H.R., Delaveau, P., 1977. The effects of time of harvest on
the partitioning of dry matter in three sugarcane varieties grown
in contrasting environments. Proc. Int. Soc. Sugar Cane Technol.
16, 1755–1769.
Kang, M.S., Miller, J.D., 1984. Genotype � environment interac-
tions for cane and sugar yield and their implications in sugarcane
breeding. Crop Sci. 24, 435–440.
Kimbeng, C.A., Rattey, A.R., Hetherington, M., 2002. Interpretation
and implications of genotype by environment interactions in
advanced stage sugarcane selection trials in central Queensland.
Aust. J. Agric. Res. 53, 1035–1045.
Littell, R.C., Stroup, W.W., Freund, R.J., 2002. SAS1 for Linear
Models, 4th ed. SAS Institute Inc., Cary, NC, 466 pp.
Liu, D.L., Helyar, K.R., 2003. Simulation of seasonal stalk water
content and fresh weight yield of sugarcane. Field Crops Res. 82,
59–73.
Magarey, R.C., Mewing, C.M., 1994. Effect of sugarcane cultivars
and environment on inoculum density of Pachymetra chaunor-
hiza in Queensland. Plant Dis. 78, 1193–1196.
Meade, G.P., 1963. Spencer-Meade Cane Sugar Handbook. Wiley,
New York.
Miller, J.D., James, N.I., 1977. Maturity of six sugarcane varieties in
Florida. Proc. Am. Soc. Sugar Cane Technol. 7, 107–111.
Milligan, S.B., 1994. Test site allocation within and among stages of
a sugarcane breeding program. Crop Sci. 34, 1184–1190.
Milligan, S.B., Gravois, K.A., Bischoff, K.P., Martin, F.A., 1990.
Crop effects on broad-sense heritabilities and genetic variances
of sugarcane yield components. Crop Sci. 30, 344–349.
Mirzawan, P.D.N., Cooper, M., DeLacy, I.H., Hogarth, D.M., 1994.
Retrospective analysis of the relationships among the test envir-
onments of the southern Queensland sugarcane breeding pro-
gramme. Theor. Appl. Genet. 88, 707–716.
Redshaw, K., 2000. Agronomic evaluation of released varieties in
South Africa. Intl. Soc. Sugar Cane Technol. Agron. Workshop,
2–6 December 2000. FL, USA. Online abstract. http://issct.int-
net.mu/agroabs.htm#4.
Rice, E., 1974. Maturity studies of sugarcane varieties in Florida.
Proc. Am. Soc. Sugar Cane Technol. 4, 33–35.
Sinclair, T.R., Gilbert, R.A., Perdomo, R.E., Shine Jr., J.M., Powell,
G., Montes, G., 2004. Sugarcane leaf area development under
field conditions in Florida, USA. Field Crops Res. 88, 171–
178.