Bottom-up processing of nanobiomaterials enables the creation of a variety of macroscopic structures in natural systems. Here, we use optical means to produce macroscopic-assembled structures of nanoparticles (NPs) from protein molecules by using light-induced bubble (LIB) generation under asymmetric pressure-driven flow in a microchannel. The broadband optical response of assembled NPs facilitates the application of photon pressure and photothermal convection when irradiated by using an infrared laser. The presence of a large amount of protein allows the generation of a vast number of stable LIBs from optically assembled metallic NP-fixed beads (MNFBs). In the case of more diluted albumin solutions, the shrinking of a single LIB can cause the aggregation of MNFBs via fg-level albumin (3.4 fg in the observation region), like a microscale bubblegum. The size of the resulting aggregate can be controlled by changing the concentration of protein. These findings can be used to devise production methods not only for broadband optical nanocomposites but also for label-free methods to detect an extremely small amount of protein.

Natural biological systems exhibit various unique structures and patterns at the microscopic and macroscopic scales (including Turing patterns1 and biological photonic crystals2) arising from the self-assembly of molecules or nanoscale composites in an evolutionary process under environmental fluctuations and external stimuli. Many functional systems consist of nanoscale biomolecules, such as light-harvesting antennae for collecting solar energy,3,4 neurons for detecting external signals,5,6 and vesicles for metabolism and material transport.7–9 Recent developments in nanotechnology mediated by biological molecules allow us to artificially produce a variety of structures by self-assembly of DNA-modified nanoparticles (NPs) and target DNA as the binder molecule10–12 and by programmable nanoscale patterns with DNA origami techniques.13 Such assembly processes can also be used for highly sensitive detection of a small amount of DNA and label-free electric detection of single nucleotide polymorphism.14,15 Furthermore, proteins are crucial nanobiomaterial building blocks for a variety of vital systems,16–18 wherein the constituent amino acids form long chains and lead to complicated biological interactions. If a protein can be used to remotely control the bottom-up assembly of non-biological nanomaterials with external fields, the production of macroscopic structures and techniques to detect biomolecules could be dramatically widened in scope.

With these dynamic approaches in mind, there are a number of available methods for transporting proteins toward the observation region, such as the use of fluidic pressure electrophoresis and magnetic fields.19–24 Although these methods can provide the necessary force for transporting targeted proteins over substantial distances, the bottom-up assembly of non-biological nanomaterials can be greatly developed if multiple local driving forces are combined. Optical trapping based on light-induced electromagnetic force (LIEF) was developed for trapping micron-sized dielectric objects25,26 and to measure weak forces between biomolecules.27,28 Recent progress in this technique has enabled the trapping and fixation of nanoscale molecules, as well as crystallization.29–31 Conversely, the photothermal effects of metallic NPs arising from localized surface-plasmon resonances (LSPRs) have attracted the attention of many researchers as a therapeutic agent (using hyperthermia).32–34 Higher laser intensity vaporizes the solvent and generates nanobubbles and microbubbles.35–38 Moreover, the collective LSPR phenomenon leads to significant spectral broadening and a redshift,39 and high-density assembled structures of metallic NPs absorb the infrared (IR) photons strongly.40 Thus, the photothermal effects of high-density metallic NPs under IR laser irradiation induce dramatic liquid-gas phase transitions, leading to the generation of submillimeter bubbles in droplets and local assembling of floating objects via the light-induced convection force (LICF).41,42 By carefully adjusting the laser conditions, the DNA double-strand formation can be accelerated at the air-liquid interface under LIEF and LICF,43 allowing a small amount of DNA to be detected via submillimeter assembly formation. However, in these previous studies, assembly by LIEF and LICF was studied in a simple droplet as an open system under more-symmetric hydrodynamic conditions. If the dynamics of target objects would be restricted in a narrow space, such as a microchannel, and if the target objects were effectively transported via fluidic flow, light-induced assembly would exhibit unconventional aspects in a closed system under an asymmetric pressure-driven flow. For example, in such a restricted geometry and under the one-dimensional controlled flow, the efficiency of detecting nanoscale biomolecules would be enhanced via the assembling process.

Here, we clarify the assembly process of NPs in a closed microchannel by changing the concentration (amount) of protein. This approach is based on the synergetic effects of LIEF and LICF and with the help of a directional pressure-driven flow (Fig. 1). In particular, our aim is to use a confined geometry in the microchannel to increase the probability of samples passing through the laser spot in conjunction with the steady flow generated by using a syringe pump (see Fig. S1 of the supplementary material), thereby realizing efficient light-induced acceleration of the bottom-up process via nanoscale biomolecules.

FIG. 1.

Schematic drawings of accelerated protein-assisted assembly of Au nanoparticle-fixed beads (AuNP-FBs) by laser irradiation in a microchannel. (a) Generation of multiple light-induced bubbles (LIBs) in the case of a large amount of protein (high concentration). (b) Shrinkage of a single LIB in the case of a small amount of protein (low concentration).

FIG. 1.

Schematic drawings of accelerated protein-assisted assembly of Au nanoparticle-fixed beads (AuNP-FBs) by laser irradiation in a microchannel. (a) Generation of multiple light-induced bubbles (LIBs) in the case of a large amount of protein (high concentration). (b) Shrinkage of a single LIB in the case of a small amount of protein (low concentration).

Close modal

To first determine the optimal conditions for assembly by LIEF and LICF, gold NP-fixed beads (AuNP-FBs) were selected since the dissipative force as a component of the LIEF arising from photon momentum transfer (which is proportional to the extinction spectrum) is greatly enhanced in the IR region because of the collective phenomena of LSPR via the electromagnetic field of the light,40 as shown in Fig. 2 [refer also the experimentally observed extinction spectrum in Fig. S2(a) and the dissipative force in Fig. S2(b) of the supplementary material]. Considering about 600 AuNPs on a bead, the theoretical magnitude of the dissipative force in Fig. 2(b) is consistent with the experimental values given in Fig. S2(b). Moving the plane containing the laser spot (focal plane) along the height direction (z direction) confirms that the generation efficiency of light-induced bubbles (LIBs) is high near the ceiling because of the strong dissipative force pushing AuNP-FBs toward the upper region for the case of static flow in the absence of pressure-driven flow (injection by capillary force with a micropipette) and of dynamic flow in the presence of pressure-driven flow (injection by using a syringe pump) at 4.76 × 102µm/s (refer Methods in the supplementary material).

FIG. 2.

Spectra of dissipative force on an AuNP-FB and experimental configuration. (a) Model for the calculation. (b) Theoretically determined dissipative force on AuNP-FBs with different AuNP densities. The vertical axis is normalized by the input laser power with 2.5 µm spot diameter. “dp” is the average inter-surface distance between AuNPs for the calculation. Panel (b) shows the spectra of dissipative force on a single AuNP. Schematic images for the case of (c) the laser spot near the microchannel ceiling and (d) of the laser spot far from the microchannel ceiling. The orange horizontal lines in panels (c) and (d) indicate the focal plane.

FIG. 2.

Spectra of dissipative force on an AuNP-FB and experimental configuration. (a) Model for the calculation. (b) Theoretically determined dissipative force on AuNP-FBs with different AuNP densities. The vertical axis is normalized by the input laser power with 2.5 µm spot diameter. “dp” is the average inter-surface distance between AuNPs for the calculation. Panel (b) shows the spectra of dissipative force on a single AuNP. Schematic images for the case of (c) the laser spot near the microchannel ceiling and (d) of the laser spot far from the microchannel ceiling. The orange horizontal lines in panels (c) and (d) indicate the focal plane.

Close modal

The loosely focused IR laser on the ceiling of the microchannel with a low numerical aperture objective lens (×40) increases the local concentration of AuNP-FBs because the strong dissipative force exceeds the gradient force [Fig. 2(c)], whereas the AuNP-FBs can escape when focusing the laser near the bottom of the microchannel [Fig. 2(d)]. To attract AuNP-FBs toward the high-intensity region by using the gradient force and avoiding an excessively strong photothermal effect (which would hinder the controllability), 1064 nm was selected as the irradiation wavelength because it is red-detuned from a collective LSPR mode. As a preliminary experiment, the time from just after laser irradiation up to bubble generation was measured in comparison with the bubble-generation efficiency (i.e., efficiency of assembling AuNP-FBs) at each height. The assembly efficiency of AuNP-FBs can be enhanced near the ceiling due to the upward photon pressure and horizontal gradient force toward the laser spot, which leads to a high probability of bubble generation in that area. Meanwhile, the laser is also focused into the sample, which is under dynamic flow, and the bubble-generation efficiency is determined in a similar manner by changing the focal-plane height. The bubble-generation efficiency in this region depends on height, which is related to the parabolic velocity profile in the microchannel [see Fig. S1(b) in the supplementary material] that forms because of the shear stress near the walls of the microchannel. At the center of the microchannel in the z direction, the velocity of pressure-driven flow is too high to induce sufficient LIEF and LICF for the trapping and assembling AuNP-FBs. The distribution of the pressure-driven flow enhances the assembling efficiency of targets through the dissipative force just at the ceiling, whereas the AuNP-FBs are pushed away by fast flow near the center of the microchannel in the z direction. The common feature in both cases (i.e., static and dynamic flows) is that LIBs are easily generated near the ceiling of the microchannel.

Next, egg albumin was used as the protein for binding AuNP-FBs because it foams as a result of air denaturation (where it is stabilized by adsorption due to hydrophobic interaction) and heat denaturation (thermal solidification),44 which facilitates the assembly of AuNP-FBs. Moreover, to investigate the assembly of AuNP-FBs in the presence of different protein concentrations in the microchannel, we added albumin solutions with differing concentrations (50 or 500 µg/mL) to the AuNP-FBs colloidal solution. The amount of albumin in a selected region (300 µm × 225 µm × 100 µm) was estimated to be 3.4 × 102 pg or 3.4 ng. As shown in Fig. 3(a), many AuNP-fixed beads are transported by the pressure-driven flow toward the laser spot near the ceiling, and LIBs are generated in the presence of egg albumin. A stagnation area (zero velocity) appears between the ceiling and large LIBs arising from the AuNP-fixed beads and gathered by the dissipative force exerted by the IR laser. In both cases [3.4 × 102 pg and 3.4 ng albumin; Figs. 3(b) and 3(c), respectively], a large number of LIBs are produced and are stabilized for several minutes (see Movie S1 of the supplementary material). In particular, in Fig. 3(c), increasing the concentration of albumin increases the number of LIBs generated. In the absence of albumin, even when more than two bubbles are generated, the smaller bubble tends to always coalesce with the larger bubble into a single bubble. This tendency is attributed to air denaturation and the foaming properties of albumin. When the surface of each LIB is covered with albumin, the coalescence of bubbles is blocked, leading to the production of many steady foams.

FIG. 3.

Multibubbles are generated in the microchannel in the case of high albumin concentrations. (a) Schematic image of multibubble generation by the synergetic effects of light pressure and convection under pressure-driven flow: (b) after addition of 50 µg/mL albumin and (c) after addition of 500 µg/mL albumin (see Movie S1 of the supplementary material).

FIG. 3.

Multibubbles are generated in the microchannel in the case of high albumin concentrations. (a) Schematic image of multibubble generation by the synergetic effects of light pressure and convection under pressure-driven flow: (b) after addition of 50 µg/mL albumin and (c) after addition of 500 µg/mL albumin (see Movie S1 of the supplementary material).

Close modal

Figures 4(a) and 4(b) show the results when a smaller amount of albumin is added, where 3.4 fg or 3.4 pg of albumin are present in the observation region of 300 µm × 225 µm × 100 µm (concentrations of 0.5 ng/mL or 0.5 µg/mL, respectively). The flow velocity was set to a high value of 2.38 × 104µm/s (0.05 mL/min) to effectively accumulate the albumins onto the surface of LIB. After 6 s of laser irradiation, a single LIB was generated with 3.4 fg albumin, as shown in Fig. 4(a). After continuing the laser irradiation for a further 10 s, the irradiation was stopped, and the LIB began to shrink (see Movie S2 of the supplementary material). The LIB took 3 min to finish shrinking, leaving behind a black aggregate (we also confirmed that a generated LIB shrank sooner in the absence of protein). Similar experiments were performed with 3.4 pg albumin [see Fig. 4(b)], and larger aggregates were obtained and LIB shrinkage occurred over more than 10 min (such a stability is attributed to the interface energy of the surrounding proteins). AuNP-FBs and albumins are considered to be transported by optically induced flow and pressure-driven flow toward the surface of a LIB, where the adsorbed AuNP-FBs and albumins remain aggregated after shrinkage, analogous to bubblegum. We assume a model in which a part of the bubble on the ceiling is covered with AuNP-FBs, as shown in the inset of Fig. 4(c). For each amount of added albumin, we then calculate from the observed bubble radii the cross-sectional area of an aggregate of AuNP-FBs after the LIB shrinks. The experimental values plotted as a function of protein mass are consistent with these calculated values. Based on these calculations, the aggregate is composed of AuNP-FBs assembled via albumins as binder molecules adsorbed on the surface of LIBs. Due to hydrophobic interactions between proteins and AuNP-FBs, aggregates adsorb strongly onto the gaseous bubble, but no aggregates emerge in the absence of albumin. As shown in these experiments, with a small amount of albumin, the generated LIB is unstable, unlike in the multibubble experiment shown in Fig. 3 (the large error bar between 104 and 105 fg indicates the boundary between the multibubble region and the shrinkage region). In this case, LIBs could not be covered uniformly with small amounts of AuNP-FBs via albumins at the gas-liquid interface of the LIB [see the model in the inset of Fig. 4(c)], where the gas (water vapor) in the bubble dissolves in the water and the LIB shrinks. The observed cross-sectional area of an aggregate correlates with the mass of albumin, which implies that the protein concentration in solution may be quantitatively evaluated based on aggregate size. These results indicate that pressure-driven flow can accelerate the light-induced production of assembled NP-protein structures and provide an unconventional method to rapidly detect fg-level proteins through LIB shrinkage if a narrow space could be made in order to increase the probability of interaction between laser irradiation area and target samples.

FIG. 4.

Aggregates remaining after LIB shrinkage, where aggregates are composed of AuNP-FBs and albumin. The LIB and aggregate were observed after shrinkage at various concentrations. Optical transmission images at the equatorial plane (a) at 6 s (bubble is generated) and at 186 s (bubble vanishes) after the onset of laser irradiation with 3.4 fg albumin in the observation region (0.5 ng/mL) (see Movie S2 of the supplementary material) and (b) at 12 s (bubble is generated) and at 635 s (bubble vanishes) after the onset of laser irradiation with 3.4 pg albumin in the observation region (0.5 µg/mL). (c) Experimental values indicate that the aggregate cross-sectional area depends logarithmically on the mass of albumin in the observation region (n = 3 for each protein mass). The calculated values are obtained from the observed bubble diameters based on the model shown in the inset of panel (c).

FIG. 4.

Aggregates remaining after LIB shrinkage, where aggregates are composed of AuNP-FBs and albumin. The LIB and aggregate were observed after shrinkage at various concentrations. Optical transmission images at the equatorial plane (a) at 6 s (bubble is generated) and at 186 s (bubble vanishes) after the onset of laser irradiation with 3.4 fg albumin in the observation region (0.5 ng/mL) (see Movie S2 of the supplementary material) and (b) at 12 s (bubble is generated) and at 635 s (bubble vanishes) after the onset of laser irradiation with 3.4 pg albumin in the observation region (0.5 µg/mL). (c) Experimental values indicate that the aggregate cross-sectional area depends logarithmically on the mass of albumin in the observation region (n = 3 for each protein mass). The calculated values are obtained from the observed bubble diameters based on the model shown in the inset of panel (c).

Close modal

In conclusion, we reveal herein a guiding principle whereby laser-induced electromagnetic interactions (photon pressure) and hydrodynamic effects (convection and bubble generation) rapidly produce an assembled structure of NPs with a broadband optical spectrum (for example, extinction, absorption, and scattering). The assembled structure can be produced by using proteins confined in the narrow space of a microchannel and with the aid of pressure-driven flow. The form of the assembled structure thus depends strongly on the protein concentration. For high protein concentration (>3.4 × 102 pg), multiple stable LIBs are generated because of the protein properties. For low protein concentration, however, unstable LIBs are generated and a size-controllable aggregate of adsorbed NPs may develop depending on the amount of protein present (3.4 fg–34 pg).

From the viewpoint of biosensing, fg-level protein concentrations can be detected based on LIB shrinkage within only a few minutes. This expected sensitivity is three orders of magnitude greater than that of commercial kits for protein detection, which often require hours of observation. These findings thus open an avenue for the unconventional bottom-up production of assembled structures of broadband optical nanostructures, such as white-light harvesting systems [even with other types of metallic nanoparticle-fixed beads (MNFBs)] using nanobiomaterials, or for the ultrafast detection of an extremely small amount of nanobiomaterial if the guiding principle is exploited in a narrow space and if the effect of pressure-driven flow is analyzed in detail.

See supplementary material for more details on the methods, movie legends, and supplementary figures including the experimental setup.

The authors would like to thank GreenChem, Inc., for advice on sample preparation and Sysmex Corporation for advice on the biological samples. We also thank Mr. Y. Yamamoto, Professor I. Nakase, Professor K. Imura, Professor H. Shiigi, Professor T. Nagaoka, and Professor H. Ishihara for their advice and support. A major part of this work was supported by the Grant-in-Aid for Scientific Research (A) (No. 17H00856), Grant-in-Aid for Scientific Research (B) (Nos. 26286029, 15H03010, and 16H03827), Grant-in-Aid for Scientific Research on Innovative Areas (Nos. 16H06507 and 16H06505) from JSPS, SENTAN-JST, The Canon Foundation, and The Nakatani Foundation, and the Key Project Grant Program of Osaka Prefecture University.

1.
T.
Sekimura
,
S.
Noji
,
N.
Ueno
, and
P.
Maini
,
Morphogenesis and Pattern Formation in Biological Systems: Experiments and Models
(
Springer
,
Japan
,
2013
).
2.
J.
Wang
,
Y.
Zhang
,
S.
Wang
,
Y.
Song
, and
L.
Jiang
,
Acc. Chem. Res.
44
,
405
(
2011
).
3.
N. Y.
Kiang
,
J. L.
Siefert
,
G.
Govindjee
, and
R. E.
Blankenship
,
Astrobiology
7
,
222
(
2007
).
4.
S.
Scheuring
and
J. N.
Sturgis
,
Science
309
,
484
(
2005
).
5.
S.
Szobota
and
E. Y.
Isacoff
,
Annu. Rev. Biophys.
39
,
329
(
2010
).
6.
J. B.
Bryson
,
C. B.
Machado
,
M.
Crossley
,
D.
Stevenson
,
V.
Bros-Facer
,
J.
Burrone
,
L.
Greensmith
, and
I.
Lieberam
,
Science
344
,
94
(
2014
).
7.
M.
Rosoff
,
Vesicles
(
Marcel Dekker, Inc.
,
NY
,
1996
).
8.
T.
Harada
and
D. E.
Discher
,
Nature
471
,
172
(
2011
).
9.
H. C.
Shum
,
Y.
Zhao
,
S. H.
Kim
, and
D. A.
Weitz
,
Angew. Chem., Int. Ed.
50
,
1648
(
2011
).
10.
A. P.
Alivisatos
 et al,
Nature
382
,
609
(
1996
).
11.
C. A.
Mirkin
,
R. L.
Letsinger
,
R. C.
Mucic
, and
J. J.
Storhoff
,
Nature
382
,
607
(
1996
).
12.
P.
Cigler
,
A. K. R.
Lytton-Jean
,
D. G.
Anderson
,
M. G.
Finn
, and
S. Y.
Park
,
Nat. Mater.
9
,
918
(
2010
).
13.
P. W.
Rothemund
,
Nature
440
,
297
(
2006
).
14.
G. P.
Acuna
 et al,
Science
338
,
506
(
2012
).
15.
S.
Tokonami
,
H.
Shiigi
, and
T.
Nagaoka
,
Anal. Chem.
80
,
8071
(
2008
).
16.
H.
Hartley
,
Nature
168
,
244
(
1951
).
17.
M. R.
O’Connell
,
R.
Gamsjaeger
, and
J. P.
Mackay
,
Proteomics
9
,
5224
(
2009
).
18.
S.
Dey
,
D. W.
Ritchie
, and
E. D.
Levy
,
Nat. Methods
15
,
67
(
2018
).
19.
C.
Escobedo
,
A. G.
Brolo
,
R.
Gordon
, and
D.
Sinton
,
Nano Lett.
12
,
1592
(
2012
).
20.
F.-K.
Liu
and
G.-T.
Wei
,
Anal. Chim. Acta
510
,
77
(
2004
).
21.
W. N.
Burnette
,
Anal. Biochem.
112
,
195
(
1981
).
22.
J. R.
Whiteaker
,
L.
Zhao
,
H. Y.
Zhang
,
L.-C.
Feng
,
B. D.
Piening
,
L.
Anderson
, and
A. G.
Paulovich
,
Anal. Biochem.
362
,
44
(
2007
).
23.
J. W.
Choi
,
K. W.
Oh
,
J. H.
Thomas
,
W. R.
Heineman
,
H. B.
Halsall
,
J. H.
Nevin
,
A. J.
Helmicki
,
H. T.
Henderson
, and
C. H.
Ahn
,
Lab Chip
2
,
27
(
2002
).
24.
P.
Yao
,
Z.
Liu
,
S.
Tung
,
Z.
Dong
, and
L.
Liu
,
J. Lab. Autom.
21
,
387
(
2016
).
25.
A.
Ashkin
,
Phys. Rev. Lett.
24
,
156
(
1970
).
26.
A.
Ashkin
,
J. M.
Dziedzic
,
J. E.
Bjorkholm
, and
S.
Chu
,
Opt. Lett.
11
,
288
(
1986
).
27.
J. T.
Finer
,
R. M.
Simmons
, and
J. A.
Spudich
,
Nature
368
,
113
(
1994
).
28.
J. C.
Meiners
and
S. R.
Quake
,
Phys. Rev. Lett.
84
,
5014
(
2000
).
29.
S.
Ito
,
H.
Yoshikawa
, and
H.
Masuhara
,
Appl. Phys. Lett.
78
,
4046
(
2001
).
30.
Y.
Pang
and
R.
Gordon
,
Nano Lett.
12
,
402
(
2012
).
31.
T.
Sugiyama
,
K.
Yuyama
, and
H.
Masuhara
,
Acc. Chem. Res.
45
,
1946
(
2012
).
32.
V. P.
Zharov
,
J.-W.
Kim
,
D. T.
Curiel
, and
M.
Everts
,
Nanomedicine
1
,
326
(
2005
).
33.
Q.
Zou
,
M.
Abbas
,
L.
Zhao
,
S.
Li
,
G.
Shen
, and
X.
Yan
,
J. Am. Chem. Soc.
139
,
1921
(
2017
).
34.
C.
Kojima
,
Y.
Watanabe
,
H.
Hattori
, and
T.
Iida
,
J. Phys. Chem. C
115
,
19091
(
2011
).
35.
G.
Baffou
,
J.
Polleux
,
H.
Rigneault
, and
S.
Monneret
,
J. Phys. Chem. C
118
,
4890
(
2014
).
36.
E.
Lukianova-Hleb
,
Y.
Hu
,
L.
Latterini
,
L.
Tarpani
,
S.
Lee
,
A.
Rebekah
,
J.
Drezek
,
H.
Hafner
, and
D. O.
Lapotko
,
ACS Nano
4
,
2109
(
2010
).
37.
Z.
Fang
,
Y. –R.
Zhen
,
O.
Neumann
,
A.
Polman
,
F. J. G.
de Abajo
,
P.
Nordlander
, and
N. J.
Halas
,
Nano Lett.
13
,
1736
(
2013
).
38.
K.
Setoura
,
S.
Ito
, and
H.
Miyasaka
,
Nanoscale
9
,
719
(
2017
).
39.
T.
Iida
,
J. Phys. Chem. Lett.
3
,
332
(
2012
).
40.
S.
Tokonami
,
S.
Hidaka
,
K.
Nishida
,
Y.
Yamamoto
,
H.
Nakao
, and
T.
Iida
,
J. Phys. Chem. C
117
,
15247
(
2013
).
41.
Y.
Nishimura
,
K.
Nishida
,
Y.
Yamamoto
,
S.
Ito
,
S.
Tokonami
, and
T.
Iida
,
J. Phys. Chem. C
118
,
18799
(
2014
).
42.
Y.
Yamamoto
,
E.
Shimizu
,
Y.
Nishimura
,
T.
Iida
, and
S.
Tokonami
,
Opt. Mater. Express
6
,
1280
(
2016
).
43.
T.
Iida
,
Y.
Nishimura
,
M.
Tamura
,
K.
Nishida
,
S.
Ito
, and
S.
Tokonami
,
Sci. Rep.
6
,
37768
(
2016
).
44.
A. C. C.
Alleoni
,
Sci. Agric.
63
,
291
(
2006
).

Supplementary Material