Science in China Series C: Life Sciences www.scichina.

com
© 2008 SCIENCE IN CHINA PRESS life.scichina.com
www.springerlink.com
Springer

Leaf surface factors of transgenic Bt cotton associated

with the feeding behaviors of cotton aphids: A case
study on non-target effects
XUE Kun, DENG Su, WANG RongJiang, YAN FengMing†, XU ChongRen
College of Life Sciences, Peking University, Beijing 100871, China

The present paper reports case study results of the risk assessment of transgenic Bt cotton on a
non-target pest, cotton aphid, Aphis gossypii. Several types of techniques, i.e., electrical penetration
graph (EPG), light and electron microscopy, bioassays and chemical analysis, were applied to investi-
gate physical and chemical leaf factors of 2 transgenic Bt cotton lines (GK12 and GK19) and their pa-
rental non-Bt cotton line (Simian3) associated with searching and feeding behaviors of cotton aphids
on leaves or leaf extracts of cotton plants. EPG results showed that there were some differences among
behaviors of cotton aphids on 2 Bt cotton and 1 non-Bt cotton lines. Cotton aphids performed similarly
to leaf surface extracts from 3 cotton lines; and leaf surface chemicals, mainly volatiles and waxes,
were almost identical in the components and concentrations among the cotton lines. However, three
cotton lines were quite different from each other in the densities of certain kinds of covering trichomes.
Therefore, the relationships between the physical characteristics and the searching behaviors of cotton
aphids on the three cotton lines were constructed as the regression equations. Glandular trichomes
and covering trichomes with 5 branches influenced the cotton aphids’ searching behaviors effectively;
and other trichomes with other branches affected aphids in varying ways. These results demonstrated
that leaf surface physical factors of transgenic Bt cotton lines different from their parental non-Bt line
could affect the penetration behaviors of non-target cotton aphids. Cotton aphids penetrate and feed
more easily on two Bt cotton lines than on the non-Bt cotton line.

Bt cotton, trichome, leaf surface chemicals, Aphis gossypii, searching behavior, EPG, electrical penetration graph, non-target effect

Bt insecticide toxins can control the lepidopteran insect
pest, e.g. the bollworm Helieoverpa armigera, to reduce
the damage of the crops. Transgenic Bt cotton is the first
transgenic crop planted on a large scale in China and in
the world. Since its release into environment in 1997,
concerns over its ecological risks, including the gene
flow[1] and non-target effects[2], have arisen, and the de-
bates are still continuing. Studies on non-target effects
are essential parts in ecological or risk assessment of
transgenic crops. Many researches focus on the effects
on the other species in the ecosystem except the target
― Corresponding author (email: fmyan@pku.edu.cn)
pest with the materials transgenic Bt cotton[3 8], trans-
―
genic Bt corn[9−12] and transgenic Bt rice[13 15].
Cotton aphid, Aphis gossypii Glover, as one of the
non-target pests in transgenic Bt cotton fields, is a typi-
cal model insect of piercing-sucking homopterans for
risk evaluation of Bt cotton. Field observations on
piercing-sucking insect populations in the transgenic
plant fields have resulted in quite controversial conclu-
sions. The outbreaks of piercing-sucking insects, such as
whiteflies and aphids, were reported in the Bt cotton
―
fields[16 18]. But in the other investigations, the popula-
Received July 10, 2007; accepted November 29, 2007
doi: 10.1007/s11427-008-0028-6
†
Supported by the National Key Basic Research Program (Grant No. G2000046803),
National Natural Science Foundation of China (Grant No. 39970153) and the Public
and Professional Project of Environmental Protection of China (Grant No.
200709047)

Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
tions of these insects in the Bt cotton and the regular
―
cotton fields were found quite similar[19 21]. So carefully
designed experiments on the feeding behaviors of pierc-
ing-sucking insects on transgenic plants in the laboratory
will hopefully provide some evidence to clarify these
confusions on the impacts of transgenic crops on
non-target insects. In the studies on the stylet penetration
behaviors of piercing-sucking insects on their host plants,
electrical penetration graph (EPG) is a powerful tool,
which was first invented by McLean and Kinsey[22] us-
ing an alternate current (AC) circuit with a 106 input
resistor, and modified to a direct current (DC) system
with a 109 input resistor by Tjallingii[23]. EPG wave-
forms correlated with the details of probing behaviors of
insects in certain plant tissues[24] could help to interpret
mechanisms of the interactions between plants and
piercing-sucking insects[25,26], such as the roles insects
play in transmission of non-persistent viruses and other
―
pathogens[27 29], probing behaviors of insects on plants,
and location of the feeding stimulants or antifeedants in
plant tissues[30].
Plant leaves are important organs with vital physio-
logical functions such as photosynthesis, transpiration,
etc[31,32]. Plant leaf surface is the media of insect-plant
interaction, and the substrate characteristics can affect
directly and/or indirectly the three trophic levels. Physi-
cal and chemical factors on the leaf surface influence the
― plant voltage (Wageningen Agricultural University, the
performance of the herbivorous insects[33 38], preda-
tors[39,40] and parasites[41]. Physical and/or chemical char-
acteristics of the leaf surface play key roles in host plant
acceptance by phytophagous insects. Any changes in
transgenic crops in physical and/or chemical characteris-
tics of the leaf surface resulting from insertion of exotic
genes can probably influence searching or acceptance of
host plants by herbivore insects.
In our previous research, the feeding behaviors of the
cotton aphids recorded by EPG on the Bt cotton were
found different from those on the regular cotton in the
searching period probably associated with the leaf sur-
face factors (unpublished data). The present investiga-
tion provides a comprehensive study of the leaf surface
physical and chemical characteristics of two Bt cotton
lines and their parental cotton line, which may be asso-
ciated with feeding or acceptance behaviors of cotton
aphids. Such information would be useful in assessing
the ecological risk, especially non-target effects, of the
transgenic Bt cotton.
146 XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
1 Materials and methods
1.1 Plants and insects
Three cotton lines, i.e., 2 transgenic Bt cotton lines GK12
and GK19 expressing Bt toxic protein of Cry1Ac, de-
veloped by Biotechnology Center of Chinese Academy
of Agricultural Sciences (CAAS), and their parental line
Simian3 (conventional non-Bt cotton line), were used
for experiments. The seeds of these cotton lines, pro-
vided by Institute of Plant Protection, CAAS, were
grown in a standardized soil mixture in plastic pots (10
cm in diameter) and kept in the growth chambers under
conditions of L16:D8 photoperiod, ca. 75% relative hu-
midity and 2700 lx illumination, with temperatures of
28℃ in light and 25℃ in darkness. Plants with 4 fully
extended true leaves were used for experiments.
Cotton aphids, Aphis gossypii Glover, were collected
from cotton fields and reared in a growth chamber at
about 28―25℃ and L16:D8 photoperiod on cotton
plants of Simian3 (the conventional cotton line) in cages.
Kept in the laboratory for more than 20 generations, ap-
terous adult aphids were used for experiments.
1.2 EPG recordings and analysis
The probing behaviors of cotton aphids on test cotton
plants were recorded using the Giga-4 direct current
amplifier with 109 Ω input resistance and an adjustable
Netherlands). The experiments were conducted in a
laboratory under conditions of ca. 25℃ and ca. 600 lx.
Aphids were connected to a ca. 3 cm gold wire (20 μm
in diameter, 99.99% gold, Sigmund Cohn, Mount Ver-
nom, NY, USA) on the pronotum of the insect with a
drop of water-based silver paint. The wired insects were
starved for 1 h prior to recordings. The EPG instrument
was placed in a Faraday cage to prevent electrical noises.
Each of the 4 wired insects was placed near the midrib
on the abaxial surface of the third fully expanded leaf of
a cotton plant. The insects and plants were only used
once. The recordings did not stop until the test aphid
starting the first phloem ingestion. Output signals were
stored onto the hard disk of a PC via an A/D card
(DI-720, Dataq instruments, Inc., USA) with the Win-
Daq Lite acquisition software (Dataq instruments, Inc.,
USA).
Measurement of the EPG waveforms was done with
the software “WinDaq Waveform Browser” (Dataq in-
struments, Inc., USA). The waveform patterns were
visually identified according to Tjallingii[42]. The wave-
forms A, B and C were regarded as pathway waveforms
and labeled as “C” for all these three patterns. Based on
these waveforms and their occurrence sequences, a
number of sequential (Figure 1) and non-sequential pa-
rameters in the EPG recordings representing biological
information were selected to correlate with the aphid
behaviors in searching for the feeding site.
1.3 Leaf surface chemical analysis
The forth fully extended true leaf was excised off the
plant and immediately immersed into 25 mL hexane
(chromatography grade, J & K Chemical Ltd., China)
for 30 s. The solution was concentrated to less than 1
mL with a vacuum pump, then filtered through nylon
membrane (0.45 μm). Hexane was added into the filtrate
to make the final sample volume to 2 mL. For chemical
analysis, 2 μL of samples was injected into a gas chro-
matography GC-14C (Shimadzu, Japan) equipped with a
HP-Innowax fused silica capillary column (30 m×0.25
mm i.d., df=0.25 μm) (J&W Scientific, USA) and an
FID detector. The temperature program was initialized
from 60℃ (held for 8 min) and increased to 180℃ (held
for 5 min) at 3℃/min. The injector and detector tem-
peratures were set as 250℃ and 260℃, respectively.
Nitrogen was used as the carrier gas at the flow rate of
20 mL/min.
1.4 Leaf surface extract bioassay
This bioassay was designed to evaluate the impact of
plant surface extracts on the searching behavior of cot-
ton aphids. A round filter paper of 16 cm in diameter
was cut into four equal parts, and recombined and
placed on the bottom of a petri dish. Then 20 μL of each
leaf surface extract from the three cotton lines and hex-
ane (as the control) was pipetted onto one of the four
filter paper parts, respectively. Ten cotton aphid adults
were released at the center of the filter paper. The petri
dish was kept in the dark. The number of aphids on each
part of the filter paper was counted after 30 min. The
observations were repeated for 15 times.
1.5 Leaf surface trichomes
Immediately after EPG recording ended, two leaf sam-
ples (about 13 mm×8 mm) were excised separately from
the leaf used in the EPG experiment, including the mid-
rib and the region where aphid searched or fed during
EPG recordings. One sample was used for observations
of morphological characteristics of trichomes under an
environmental scanning electron microscope (ESEM)
(Quanta 200, FFEI Company, USA). The other leaf
sample was for counting all types of trichomes under the
stereomicroscope (×40) after the sample became trans-
parent in phenol-lactic acid solution at 100℃ for 1 h .

Figure 1 The sequential parameters of EPG experiments. The numbers in the cycle represent: 1, total duration: from start to the beginning of the first E
waveform; 2, from 1st probe to 1st E; 3, probe before 1st E: the duration of the probe with 1st E waveform; 4, duration of 1st probe; 5, time to 1st probe:
from start to the beginning of the first probe; 6, time from 1st probe to 1st potential drop (pd); 7, time to 1st pd. Np, Non-probe; C, pathway waveform; G,
xylem ingestion; E1, the first stage of phloem ingestion; pd, potential drop in the C waveform.

XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156 147
1.6 Statistical analysis
The EPG results and the quantitative characteristics of
the leaf surface were compared using ANOVA with
PROC GLM[43]. A chi-square test with a null hypothesis
of equal distribution was used to compare the number of
cotton aphids attracted to the leaf surface extracts in
bioassays. The correlation between the aphid behaviors
and the leaf characteristics was constructed using PROC
REG[43]. The significant level was set at P<0.05.
2 Results
2.1 EPG recording and analysis
Among 17 EPG parameters related to the searching be-
haviors of cotton aphids on leaf surface of 2 Bt cotton
lines (GK12 and GK19) and a non-Bt cotton line (Sim-
ian3), 8 parameters were found significantly different,
which reflected different performance of aphids on 3
cotton lines (Table 1).
The parameter “total duration” stands for the time that
aphids spend to find the feeding site from the beginning
of recordings to the 1st E waveform on the host plant. In
our results, the time cotton aphids spent on GK12 and
GK19 was significantly shorter than that on the regular
cotton. The parameters “total duration of C waveform”
and “total duration of non-probe waveform” were simi-
lar to the parameter “total duration”, and the “percentage
of C waveform” performed by the cotton aphids on
Simian3 was significantly lower than that on the two
transgenic cotton lines. “Percentage of non-probe” on
the regular cotton line was higher than those on GK12,
but this parameter on GK19 was statistically equal to
those on GK12 and the regular cotton. “Time to 1st
probe” is the process from start of the recording to the
first probe into the cotton leave surface, and “time to 1st
potential drop(pd)” is the duration from start to the be-
ginning of the first pd. These two parameters reflected
the influence of leaf surface on the searching perform-
ance of cotton aphids. Cotton aphids spent significantly
more time reaching the first probe and get the first pd on
Simian3 than on the two Bt lines, while the duration
from the first probe to 1st pd was statistically similar
among the three cotton lines. Besides, the mean dura-
tions between two potential drops in the probe with E
waveform were longest on the Simian3, and shortest on
GK19, with significant difference.
2.2 Leaf surface extract bioassay
In the 150 cotton aphids, 20 aphids had not walked be-
yond the paper disk which was used for releasing the
test aphids. Among 130 aphids in different filter paper
parts with leaf surface extracts or hexane, 29 cotton
aphids were in the part with the extract from the regular
cotton Simian3, 39 aphids were attracted by the leaf
surface extract from transgenic Bt cotton GK12, 35 aphids
were in the extracts of GK19, while the remaining 27
were found in the control (hexane) part. This showed
that the extracts from those three cotton lines had the
similar effects on the cotton aphids’ searching behaviors
(χ2=2.8< χ 23,0.05=7.81, P>0.05).
2.3 The chemical factors on the leaf surface
Most surface chemicals of cotton leaf could be dissolved
into hexane and be separated by gas chromatography

Table 1 Comparison of the probing behaviors of aphids on the cotton lines in EPG recordings a)
Parameter Parallel plant layer GK12 (n=8) GK19 (n=8) Simian3 (n=8) P value
Total duration (s) surface, epidermis and mesophyll 3612±1647b 3163±909b 10300±2471a 0.0171
Total duration of C (s) surface, epidermis and mesophyll 2220±710b 2474±696b 5744±1619a 0.0608
Percentage of C waveform (%) surface, epidermis and mesophyll 0.787±0.065a 0.824±0.046a 0.588±0.081b 0.0408
Total duration of non-probe (s) surface, epidermis and mesophyll 727±355b 689±231b 3953±1494a 0.026
Percentage of Np (%) surface, epidermis and mesophyll 0.168±0.042b 0.176±0.046ab 0.343±0.079a 0.0754
Time to 1st probe (s) surface 301±177b 40.0±14.4b 2916±1291a 0.0242
Total number of probes surface 5.250±1.436a 8.500±2.739a 8.125±1.737a 0.4828
Total number of Np surface 5.000±1.427a 8.375±2.783a 7.875±1.652a 0.4645
Total number of pds mesophyll and cytoplasm 40.88±11.29a 49.50±14.75a 56.50±11.17a 0.6811
Time to 1st pd (s) surface and mesophyll 371±182b 163±40.4b 2993±1282a 0.0246
Duration of 1st probe (s) epidermis, mesophyll and cytoplasm 271.5±75.5a 259.1±103.5a 1396±1081a 0.3586
Time to 1st pd from1st probe (s) mesophyll 69.58±29.24a 123.83±45.10a 77.00±44.93a 0.5969
Probe duration before 1st E (s) mesophyll 914.1±299.4a 693.3±86.6a 1648.1±585.4a 0.2062
Number of pd in the E probe mesophyll and cytoplasm 18.88±4.94a 18.13±2.64a 20.63±3.65a 0.8959
Time/pd of E probe (s) mesophyll and cytoplasm 45.95±2.34ab 39.72±2.93b 74.82±20.09a 0.1045
a) Mean±SE followed by different letters indicates significant differences at level of P<0.05 (b vs. a) and P<0.01 (c vs. a).

148 XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
(GC). According to the retention times in GC traces,
there were no evident differences in chemical compo-
nents among the transgenic Bt cottons and the control
(Figure 2). To calculate the concentrations of all the
chemicals detected, the largest component, of which
retention time was 32.1 min, was regarded as the inner
marker and its peak area was accounted as 100. The rela-
tive concentrations of the other components were calcu-
lated as the ratios of the peak areas to the largest peak
area. Through the statistic analysis, there was no com-
ponent showing significant difference in the concentra-
tion among the three cotton lines.
2.4 Types and quantitative characteristics of tri-
chomes on the cotton leaf surface
Under the ESEM, two types of trichomes, i.e., glandular
trichomes and covering trichomes, were found on all the
samples of 2 Bt cotton and the non-Bt cotton lines (Fig-
ure 3). The glandular trichomes were distributed on the
abaxial and adaxial leaf surfaces abundantly, while the
covering trichomes, with arm numbers ranging from 1 to
12, existed only along the veins on both sides of leaf
surfaces. The density of the glandular trichome on the
Figure 2 Comparisons of leaf surface chemical profiles among the three cotton lines. There was no significantly different component in the hexane ex-
tracts of the cotton leaf surface (nGK12, GK19, Simian3=7, 6, 6).
XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
regular cotton line Simian3 was median among the three
cotton lines, that on GK12 was the smallest, and that on
GK19 was the largest. The difference between the den-
sities of the two Bt cotton lines was significant, while
the densities of the two densities compared with that on
the regular cotton were not significantly different.
The linear density of the covering trichomes on the
midrib and the density of all the trichomes were calcu-
lated and compared among the three cotton lines (Table
2). The linear density of the covering trichomes on the
midrib was expressed as
Linear density = Number of certain covering trichome/
Length of the midrib.
There were 12 kinds of covering trichomes according
to the number of branches (1 to 12) , but only 9 kinds
were calculated (1 to 9, the covering trichomes with
more than 8 branches were grouped together as “Nine-
branched”). The linear densities of one-branched cover-
ing trichomes of the 2 Bt cotton lines were more than
that of the regular cotton, while the linear densities of
2-, 3-, 8- and 9-branched covering trichomes were
statistically equal among the 3 cotton lines，and the linear
149
Table 2 The quantitative characteristics of the trichomes on leaf samples of 2 Bt and 1 regular cotton lines a)

Linear density of the covering trichomes on the midrib (mm−1) Density of the trichomes on the whole leaf sample (mm−2)
Trichome types
GK12 (n=8) GK19 (n=8) Simian3 (n=8) P value GK12 (n=8) GK19 (n=8) Simian3 (n=8) P value
One-branched 1.041±0.360ab 2.281±0.668a 0.777±0.317b 0.080 0.125±0.034c 0.556±0.093a 0.093±0.029c <0.001
Two-branched 0.645±0.284a 1.259±0.165a 0.844±0.187a 0.153 0.183±0.047a 0.302±0.060a 0.165±0.032a 0.109
Three-branched 0.391±0.084a 0.423±0.219a 0.300±0.064a 0.812 0.067±0.013a 0.103±0.038a 0.114±0.022a 0.427
Four-branched 0.960±0.283a 0.419±0.238c 2.562±0.539a 0.002 0.146±0.030c 0.355±0.064c 0.491±0.101a 0.003
Five-branched 0.278±0.073ab 0.093±0.081b 0.532±0.153a 0.033 0.029±0.008ab 0.021±0.020b 0.069±0.014a 0.078
Six-branched 0.273±0.086c 0.033±0.033c 0.629±0.115a <0.001 0.027±0.010ac 0.009±0.009c 0.057±0.011a 0.011
Seven-branched 0.259±0.129ab 0.016±0.016b 0.597±0.268a 0.079 0.028±0.013a 0.003±0.003a 0.047±0.023a 0.141
Eight-branched 0.3067±0.137a 0.000±0.000a 0.311±0.136a 0.102 0.027±0.013a 0.000±0.000a 0.027±0.011a 0.115
Nine-branched 0.157±0.107a 0.000±0.000a 0.138±0.117a 0.432 0.010±0.007a 0.000±0.000a 0.010±0.009a 0.465
Glandular ― ― ― ― 4.596±0.574b 6.591±0.482a 5.070±0.842ab 0.101
The sum of covering trichomes 4.321±0.949a 4.530±0.809a 6.690±0.910a 0.140 0.642±0.125b 1.129±0.174a 1.072±0.164ab 0.076
The total branches 16.248±3.627c 8.548±2.063c 28.093±4.235a 0.003 2.094±0.454b 2.191±0.529b 4.055±0.558a 0.023

a) Mean±SE followed by different letters indicates significant differences at level of P<0.05 (b vs. a) and P<0.01 (c vs. a).
Figure 3 The morphology of the trichomes on the abaxial surface of the cotton leaf. (a)―(j) One- to ten-branched covering trichome; (k) glandular
trichome is distributed on the vein and lamina; (l) binary glandular trichomes on the lamina; (m) the glandular trichome with a short handle and a multi-cell
capitate.

densities of the other covering trichomes (4-, 5-, 6-, 7-
branched) on 2 Bt cottons were less than those on the
regular cotton. Total linear densities of all kinds of cov-
ering trichomes of the 3 cotton lines were statistically
equal; but when the branches of those covering
trichomes were counted, there were significantly less
branches on both transgenic cottons than on the control.
The sum density of the covering trichomes on the whole
leaf sample of Simian3 was higher than that of GK12,
but lower than that of GK19; like the branch linear den-
sity, the branch density on the lamina of Simian3 was
significantly higher than those of the two transgenic
cotton lines. The density of the glandular trichomes on
the regular cotton was statistically equal to those on the
two transgenic cotton lines, while the density of the
glandular trichome on GK12 was significantly less than

XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156 151
that on GK19.
2.5 The relationship between the searching behav-
iors of cotton aphids and the surface characteristics
In this experiment, the physical characteristics of the
leaf surface included the linear densities of the covering
trichomes on the midrib, the density of the covering
trichomes and the glandular trichome on the lamina. The
parameters of searching behaviors which were related to
the leaf surface factors were chosen. The relationship
among data of cotton aphids on cotton lines GK12, GK-
19 and Simian3 was drawn with the regression method
(Table 3). All the EPG parameters chosen were rela-
tive to the linear density of multi-branched covering
trichomes on the midrib and/or the density of multi-
branched covering trichomes and/or the glandular trich-
ome on the lamina. The few-branched covering trichome
quantitative characteristics affect the searching behav-
iors of the cotton aphids slightly. Lower linear density of
five-branched covering trichome on the midrib, higher
density of four-branched covering trichome and lower
density of glandular trichome would make the total du-
ration of the searching process longer. The parameters
“time to 1st probe” and “time to 1st pd” were so com-
plex that they were correlative to many quantitative
characteristics of a few kinds of covering trichomes. The
other EPG parameters were positively or inversely pro-
portional to certain characteristics of leaf surface.
3 Discussions
From the EPG results, in terms of the parameters “total
duration”, “time to 1st probe” and “time to 1st pd”, the
aphids spent more time in finding the suitable probe site
on regular cotton, which indicated that there must be
some factors on the leaf surface of Simian3 preventing
the aphids’ searching.
Physical and chemical characteristics on the leaf sur-
face could influence the searching behaviors of in-
sects[44]. Although some researchers documented the
volatiles of transgenic plant were not the same[45,46], no
significant difference in the chemical profiles of the
three cotton lines was showed in the present results. The
bioassay results also indicated that cotton aphids per-
formed similarly on the extracts of Bt cotton lines and
the regular cotton line. Therefore, the leaf surface
chemicals should not be responsible for the different
searching and feeding behaviors of cotton aphids on Bt
and non-Bt cotton lines. Physical factors of leaf surface
could be reasonably presumed to be the determinant of
the differences of aphid behaviors due to the different
density of various trichomes found among the 3 cotton
lines.
Physical factors on the leaf surface are mainly density
and types of trichomes. There are two primary types of
epidermal outgrowths, i.e., covering trichomes and
―
capitate or glandular trichomes[47 49]. Trichomes play
important roles in balancing the leaf surface tempera-
―
ture[50 52], keeping the surface clean[53] and defending
the leaf mechanically[54]. The glandular trichomes in
tomato had been found to trap aphids[55]. The hairy va-
rieties of cucurbits were more resistant to aphid than
those of smooth cultivars[56], and in ashgourd, trichome

Table 3 The relationship between the searching behaviors of cotton aphids and the surface characteristics
Regression expression a)
Total duration = 11484 - 6571.9Ld5 + 13442D4 - 1340.8Gd R2=0.5180; P=0.0019
Total duration of C = 8048.9 - 843.25Gd R2=0.2387; P=0.0154
Percentage of C waveform = 0.8565 - 0.4789D4 R2=0.3542; P=0.0022
Total duration of np = 271.79 + 5894.1D4 R2=0.2750; P=0.0085
Percentage of np= 0.1394 + 2.2685D5 R2=0.3441; P=0.0026
Time to 1st probe = -561.53 + 522.7Ld2 - 5361.7Ld5 + 11610Ld6 + 3263.4Ld7 + 29110Ld9 + 45671D5 - 95442D6 - R2=0.9750; P<0.0001
522087D9
Total number of probes = 9.4402 - 7.0999Ld5 R2=0.1828; P=0.0372

Total number of np = 9.2122 - 7.0348Ld5 R2=0.1797; P=0.0390

Time to 1st pd = -3 66.72 + 433.77Ld2 - 5250.2Ld5 + 11241Ld6 + 3415.2Ld7 + 28984Ld9 + 46473D5 - 95501D6 - R2=0.9692; P<0.0001
524025D9
a) Ld1, linear density of one-branched covering trichome; Ld2, linear density of two-branched covering trichome; and so on. D1, the density of
one-branched covering trichome on the lamina; D2, the density of two-branched covering trichome on the lamina; and so on. Gd, the density of the glandu-
lar trichome on the lamina.

152 XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
density had a significant negative influence on the num-
ber of cotton aphids[57]. It can be seen that the more tric-
homes, the more difficulties for aphids to live on the leaf.
Cotton trichomes have been researched extensively,
but no reports have been published on comparisons of
trichomes between Bt and non-Bt cotton lines. In our
present results, all types of covering trichomes were
found on leaves of all the 3 cotton lines, and the densi-
ties of total covering trichomes were equal. However,
when comparing each pair of covering trichomes among
the three cotton lines, densities were quite different. On
Simian3, four-branched covering trichomes had the
largest number, reaching almost 50% of all types of
trichomes, while those on 2 Bt cotton lines were not the
most ones on its leaf surface. We could observe clearly
the trend of the ratio changing in Figure 4. The numbers
of covering trichomes with more than 7 branches in the
three cotton lines were too small and their ratios were
identical, and therefore, it seemed that they had the same
influence on the searching behaviors of cotton aphids.
Although the densities and ratios of one- and two-
branched covering trichomes were different among the
three cotton lines, those two kinds of covering trichomes
affected the searching behaviors of cotton aphids slightly
because of the low densities and the vertical style out-
growing from the lamina. Aphids could walk freely
among them.
We conclude that the leaf surfaces of each cotton line
Figure 4 The ratios of each kind of covering trichomes in the three cotton lines. The ratios of one-branched and four-branched covering trichomes among
the three cotton lines differed significantly.
XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
were covered with a similar number of covering trich-
omes and glandular trichomes, but the total branches of
the covering trichomes on the regular cotton were sig-
nificantly more than those on the two Bt cotton lines.
Therefore, the covering trichomes of regular cotton could
defend the surface effectively, and the aphids had to
spend significantly more time in walking on the leaf
surface to search the appropriate puncturing site on the
non- Bt cotton than on Bt cotton lines.
Inserting bt gene into a plant was aimed at controlling
the lepidopteraans, especially the bollworm Helieoverpa
armigera, and decreasing the loss. The plantation of Bt
cotton on a large scale has been proven to be able to
control the damage of cotton bollworms and has resulted
in economic benefits. However, transgenic processes
could result in certain unintended changes of plant char-
acteristics, such as retarded growth of the plants[45]. The
transgenic Bt cotton GK12 and GK19 were developed
from the regular cotton Simian3 using the T-vector from
Agrobacterium tumefaciens, so the genomes of GK12
and GK19 were almost identical to the Simian3 except
for the transgene. However, under the same growing
conditions, the three cotton lines grew differently in
terms of physical factors of the leaf surface associated
with feeding behaviors of cotton aphids. In cotton, fiber
MYB protein controlled the development of trichome[58].
The possible mechanism was the insertion of bt gene
and its expression affected the expressions of relative
153
genes directly or indirectly. The variations in some cot-
ton characteristics indicated that the transgenic process
may have led to some unintended effects. More well-
designed biochemical and molecular researches might
elucidate mechanisms on the physical changes of the
cotton leaf surface.
The studies on the non-target effects of genetic modi-
fied organisms often use transgenic Bt crops as the objects.
No significant effects on the arthropod community-specific
parameters were due to Bt rice[13], and no marked differ-
ences between Bt and non-Bt rice plots in the population
changes of the important pest planthopper Sogatella
furcifera were found[15]. Bt corn affected the non-target
―
arthropods slightly[10 12], but laboratory feeding tests
showed that Bt corn byproducts reduced growth and
increased mortality of non-target trichopterans[9]. Trans-
genic Bt cottons were studied even earlier. The re-
searches on their non-target effects included the toxi-
cology to certain insects and the investigation of the
―
biodiversity changes in the field[16 21], even the Bt pro-
tein transmission to the higher trophic level[4,5,7]. Fur-
thermore, the methodology of the risk assessment of Bt
cotton was discussed[59]. Several researchers reported
that the structure of the arthropod populations was al-
tered in the Bt cotton fields, and the sucking insects in-
creased evidently, and other gave different results. Ac-
cording to our results, we could presume that the cotton
aphids could more easily insert their stylets into the leaf
surface of Bt cotton lines than the non-Bt cotton line in
1 Messeguer J. Gene flow assessment in transgenic plants. Plant Cell,
Tissue and Organ Culture, 2003, 73: 201―212[DOI]
2 Andow D A, Hilbeck A. Science-based risk assessment for nontarget
effects of transgenic crops. Bioscience, 2004, 54: 637―649[DOI]
3 Sisterson M S, Biggs R W, Manhardt N M, et al. Effects of transgenic
Bt cotton on insecticide use and abundance of two generalist predators.
Entomol Exp Appl, 2007, 124: 305―311[DOI]
4 Sharma H C, Arora R, Pampapathy G. Influence of transgenic cottons
with Bacillus thuringiensis cry1Ac gene on the natural enemies of
Helicoverpa armigera. Biocontrol, 2007, 52: 469―489[DOI]
5 Torres J B, Ruberson J R, Adang M J. Expression of Bacillus
thuringiensis Cry1Ac protein in cotton plants, acquisition by pests and
predators: A tritrophic analysis. Agric Forest Entomol, 2006, 8: 191―
202[DOI]
6 Cattaneo M G, Yafuso C, Schmidt C, et al. Farm-scale evaluation of
the impacts of transgenic cotton on biodiversity, pesticide use, and
yield. Proc Natl Acad Sci USA, 2006, 103: 7571―7576[DOI]
7 Zhang G F, Wan F H, Lovei G L, et al. Transmission of Bt toxin to the
154 XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
terms of searching period of the aphids. Adding the
quick reproduction of aphids and the absence of lepi-
dopteran pests, the exploding of the aphids, possibly of
other piercing- sucking insects, in the Bt cotton fields is
very possible.
For the time being, it is still difficult to draw a general
conclusion or methodology on non-target effects. How-
ever, in studies on non-target effects of transgenic crops,
the following aspects should be considered: (1) as many
as possible transgenic lines and their parental varieties
should be chosen; (2) as many as possible non-target
organisms, including pests, natural enemies and other
organisms, should be used; (3) many techniques should
be used to elucidate the mechanisms of non-target ef-
fects; (4) the laboratory experiments should be com-
bined with field investigations to study not only superfi-
cial phenomena but possible reasons as well; and (5)
techniques of molecular biology, bioinformation and
biotechnology are the most useful tools to elucidate the
mechanisms. Finally, a standard procedure for conduc-
tion of non-target effect studies can be made upon en-
ough research results.
The present research was the first attempt in elucida-
tion of the mechanism on the non-target effects. It is
hopeful that the present case-study results could provide
some useful information for non-target effect assess-
ments of transgenic crops.
We thank W. F. Tjallingii, Wageningen University, the Netherlands, for the
scientific and technical support on DC-EPG applications.
predator Propylaea japonica (Coleoptera: Coccinellidae) through its
aphid prey feeding on transgenic Bt cotton. Environ Entomol, 2006,
35: 143―150
8 Naranjo S E. Long-term assessment of the effects of transgenic Bt
cotton on the abundance of nontarget arthropod natural enemies.
Environ Entomol, 2005, 34: 1193―1210[DOI]
9 Rosi-Marshall E J, Tank J L, Royer T V, et al. Toxins in transgenic
crop byproducts may affect headwater stream ecosystems. Proc Natl
Acad Sci USA, 2007, 104: 16204―16208[DOI]
10 Rose R, Dively G P. Effects of insecticide-treated and lepidopteran-
active Bt Transgenic sweet corn on the abundance and diversity of
arthropods. Environ Entomol, 2007, 36: 1254―1268[DOI]
11 Floate K D, Carcamo H A, Blackshaw R E, et al. Response of ground
beetle (Coleoptera: Carabidae) field populations to four years of
Lepidoptera-Specific Bt corn production. Environ Entomol, 2007, 36:
1269―1274[DOI]
12 Gathmann A, Wirooks L, Hothorn L A, et al. Impact of Bt maize
pollen (MON810) on lepidopteran larvae living on accompanying
 weeds. Mol Ecol, 2006, 15: 2677―2685[DOI]
13 Li F F, Ye G Y, Wu Q, et al. Arthropod abundance and diversity in Bt
and non-Bt rice fields. Environ Entomol, 2007, 36: 646―654[DOI]
14 Bai Y Y, Jiang M X, Cheng J A, et al. Effects of CrylAb toxin on
Propylea japonica (Thunberg) (Coleoptera: Coccinellidae) through its
prey, Nilaparvata lugens St(a)over-circlel (Homoptera: Delpha-
cidae), feeding on Transgenic Bt rice. Environ Entomol, 2006, 35:
1130―1136
15 Chen M, Ye G Y, Liu Z C, et al. Field assessment of the effects of
transgenic rice expressing a fused gene of cry1Ab and cry1Ac from
Bacillus thuringiensis Berliner on nontarget planthopper and
leafhopper populations. Environ Entomol, 2006, 35: 127―134
16 Cui J J, Xia J Y. Effects of Bt (Bacillus thuringiensis) transgenic
cotton on the dynamics of pest population and their enemies. Acta
Phytophylac Sin (in Chinese), 2000, 27: 141―145
17 Reed G L, Andrew S J, Jennifer R, et al. Transgenic Bt potato and
conventional insecticides for Colorado potato beetle management:
Comparative efficacy and non-target impacts. Entomol Exp Appl,
2001, 100: 89―100[DOI]
18 Deng S D, Xu J, Zhang Q W, et al. Effect of transgenic Bt cotton on
population dynamics of the non-target pests and natural enemies of
pests. Acta Entomol Sin (in Chinese), 2003, 46(1): 1―5
19 Wan P, Huang M S, Wu K M, et al. Effects of transgenic Bt cotton on
development and population dynamics of cotton aphid. Sci Agric Sin
(in Chinese), 2003, 36 (12): 1484―1488
20 Wang W G, Wu K M, Liang G M, et al. Occurrence of cotton pests in
the Bt cotton fields and its control strategy. Plant Protection (in Chi-
nese), 1999, 25 (1): 3―5
21 21 Wu K M, Guo YY. Influences of Bacillus thuringiensis Berliner
cotton planting on population dynamics of the cotton aphid, Aphis
gossypii Glover, in Northern China. Environ Entomol, 2003, 32(2):
312―318
22 McLean D L, Kinsey M G. A technique for electrical recording aphid
feeding and salivation. Nature, 1964, 202: 1358―1359
23 Tjallingii W F. Electronic recording of penetration behaviour by
aphids. Entomol Exp Appl, 1978, 24: 521―530
24 Tjallingii W F. Electrical recording of stylet penetration activities. In:
Minks A K, Harrewijn P, eds. Aphid, Their Biology, Nature Enemies
and Control, 1988. 2B, 95―108
25 Ellsbury M M, Backus E A, Ullman D L. History, development, and
application of AC electrical insect feeding monitors. Entomological
Society of America, Lanham, MD: Thomas Say Publications in
Entomology, 1994
26 Walker G P, Backus E A, eds. Principles and applications of electronic
monitoring and other techniques in the study of homopteran feeding
behavior. Entomological Society of America, Lanham, MD: Thomas
Say Publications in Entomology, 2000
27 Martin B, Collar J L, Tjallingii W F, et al. Intracellular ingestion and
salivation by aphids may cause the acquisition and inoculation of
non-persistently transmitted plant viruses. J General Virol, 1997,
XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
78(10): 2701―2705
28 Johnson D D, Walker G P, Creamer R. Stylet penetration behavior
resulting in inoculation of a semipersistently transmitted closterovirus
by the whitefly Bemisia argentifolii. Entomol Exp Appl, 2002, 102(2):
115―123[DOI]
29 Backus E A, Habibij J, Yan F M, et al. Stylet penetration by adult
Homalodisca coagulata on grape: Electrical penetration graph
waveform characterization, tissue correlation, and possible implica-
tions for transmission of Xylella fastidiosa. Ann Entomol Soc Am,
2005, 98(6): 787―813[DOI]
30 Garzo E, Soria C, Gomez-Guillamon M L, et al. Feeding behavior of
Aphis gossypii on resistant accessions of different melon genotypes
(Cucumis melo). Phytoparasitica, 2002, 30(2): 129―140
31 Constable G A, Rawson H M. Carbon production and utilization in
cotton: Inferences from a carbon budget. Austral J Plant Physiol, 1980,
7: 539―553
32 Wullschleger S D, Oosterhuis D M. Photosynthesis, transpiration, and
water-use effieiency of cotton leaves and fruit. Photosynthetica, 1991,
25: 505―515
33 Dimock M B, Kennedy G G. The role of glandular trichomes in the
resistance of Lycopersicon hirsutum f. glabratum to Heliothis zea.
Entomol Exp Appl, 1983, 33: 263―268
34 Yencho G C, Tingey W M. Glandular trichomes of Solanum
berthaultii alter host preference of the Colorado potato beetle,
Leptinotarsa decemlineata. Entomol Exp Appl, 1994, 70: 217 ―
225[DOI]
35 Eigenbrode S D, Espelie K E. Effects of plant epicuticular lipids on
insect herbivores. Ann Rev Entomol, 1995, 40: 171―194[DOI]
36 Eigenbrode S D, Castagnola T, Roux M B, et al. Mobility of three
generalist predators is greater on cabbage with glossy leaf wax than on
cabbage with a wax bloom. Entomol Exp Appl, 1996, 81 (3): 335―
343[DOI]
37 Wilkens R T, Shea G O, Halbreich S, et al. Resource availability and
the trichome defenses of tomato plants. Oecologia, 1996, 106 (2):
181―191[DOI]
38 Chatzivasileiadis E A, Sabelis M W. Toxicity of methyl ketones from
tomato trichomes to Tetranychus urticae Koch. Exp Appl Acarol,
1997, 21 (6-7): 473―484[DOI]
39 Krips O E, Kleijn P W, Willems P E L, et al. Leaf hairs influence
searching efficiency and predation rate of the predatory mite Phyto-
seiulus persimilis (Acari: Phytoseiidae). Exp Appl Acarol, 1999, 23
(2): 119―131[DOI]
40 De Clercq P, Mohaghegh J, Tirry L. Effect of host plant on the
functional response of the predator Podisus nigrispinus (Heteroptera:
Pentatomidae). Biol Contr, 2000, 18 (1): 65―70[DOI]
41 Romeis J, Shanower T G, Zebitz C P W. Physical and chemical plant
characteristics inhibiting the searching behaviour of Trichogramma
chilonis. Entomol Exp Appl, 1998, 87: 275―284[DOI]
42 Tjallingii W F. Continuous recording of stylet penetration activities by
aphids. In: Campbell R K, Eikenbary R D, eds. Aphid-Plant Genotype
155
 Interactions. Amsterdam: Elsevier, 1990, 89―99
43 SAS, User manual. SAS company, 1998
44 Kamel S A, Elkassaby F Y. Relative resistance of cotton varieties to
spider mites, leafhoppers, and aphids. J Econom Entomol, 1965, 58:
209―212
45 Aharoni A, Giri A P, Deuerlein S, et al. Terpenoid metabolism in
wild-type and transgenic arabidopsis plants. Plant Cell, 2003, 15:
2866―2884[DOI]
46 Yan F M, Bengtsson M, Anderson P, et al. Antennal response of cotton
bollworm (Heliocoverpa armigera) to volatiles in transgenic Bt
cotton. J Appl Entomol, 2004, 128(5): 354―357[DOI]
47 Webber I E. Anatomy of the leaf and stem of Gossypium. J Agric Res,
1938, 57: 269―286
48 Charles T B, McCarty J C, Jenkins J N, et al. Frequency of pigment
glands and capitate and covering trichomes in nascent leaves of
selected cottons. Crop Sci, 1983, 23: 369―371
49 Bondada B R, Oosterhuis D M. Comparative epidermal ultrastructure
of cotton (Gossypium hirsutum L.) leaf, bract and capsule wall. Ann
Bot, 2000, 86 (6): 1143―1152[DOI]
50 Ehleringer J. Leaf absorptances of Mojave and Sonoran desert plants.
Oecologia, 1981, 49: 366―370[DOI]
51 Baldocchi D, Verma S B, Rosenberg N J, et al. Leaf pubescence
effects on the mass and energy exchange between soybean canopies
156 XUE Kun et al. Sci China Ser C-Life Sci | Feb. 2008 | vol. 51 | no. 2 | 145-156
and the atmosphere. Agron J, 1983, 75: 537―542
52 Melcher P J, Goldstein G, Meinzer F C, et al. Determinants of thermal
balance in the Hawaiian giant rosette plant Argyroxiphium sand-
wicense. Oecologia, 1994, 98: 412―418[DOI]
53 Brewer C A, Smith W K. Patterns of leaf surface wetness for montane
and subalpine plants. Plant, Cell and Environment, 1997, 20: 1―
11[DOI]
54 Levin D A. The role of trichomes in plant defense. Q Rev Biol, 1973,
48: 3―15
55 Johnson B. The influence on aphids of the glandular hairs on tomato
plants. Plant Pathol, 1956, 5: 130―132
56 Horn D J. Ecological Approach to Pest Management. London: the
Guilford Press, 1988
57 Khan M M H, Kundu R, Alam M Z. Impact of trichome density on the
infestation of Aphis gossypii Glover and incidence of virus disease in
ashgourd Benincasa hispida (Thunb.) Cogn. Int J Pest Manage, 2000,
46 (3): 201―204[DOI]
58 Wang S, Wang J W, Yu N, et al. Control of plant trichome develop-
ment by a cotton fiber MYB gene. Plant Cell, 2004, 16: 2323―
2334[DOI]
59 Liu X D, Zhai B P, Zhang X X, et al. Impact of transgenic cotton
plants on a non-target pest, Aphis gossypii Glover. Ecol Entomol,
2005, 30: 307―315[DOI]