Non-destructive leaf area estimation in Myrtus
communis plants
G. Gugliuzza, G. Fascella, M.M. Mammano and M. Militello
Consiglio per la Ricerca e la Sperimentazione in Agricoltura, Unità di ricerca per il recupero e la valorizzazione
delle specie floricole mediterranee (CRA SFM), Bagheria (Palermo), Italy.
Abstract
Leaf area (LA) is an important biometrical parameter recorded for evaluating
plant growth in field and greenhouse experiments. In the present study a LA
estimation model was developed for myrtle (Myrtus communis L.) pot plants under
different water regimes, using linear measurements of leaf length (L) and width (W).
The aim of the work was to establish if a non-destructive LA model could be used to
estimate plant leaf area and, if this model could be used as predictor of plant suboptimal water availability condition. A different approach from those reported in
literature was used. The total aerial part (all leaves) of two-year old potted plants of
myrtle, were used to develop the model. LA and leaf dimensions were measured with
an area meter. Frequency distributions of leaf dimensions were created through the
L/W ratio. Frequency distributions provided evidence of leaf-shape polymorphism
that greatly influenced the power of the model. Regression analyses of LA versus L and
W showed several models that could be used for estimating the area of individual
myrtle leaves, with significant differences among the models. A quadratic model
having L as the independent variables (y=aL2+bL+c) provided the most accurate
estimate (R2=0.92) of leaf area. Validation of the model was performed with the LA
data measured using plants grown under different water regimes to evaluate the
possibility of using the model to evaluate different treatment effects. Results showed
that the correlation between measured and predicted areas was high (R2=0.92±0.1) in
every case. The quadratic model developed in this study did not show evidence of the
effect of different water regime treatments on total plant leaf area.
Keywords: allometric model, regression curve, myrtle
INTRODUCTION
Many authors report the importance of plant leaf area (LA) as a key variable for most
agronomic and physiological studies involving light interception, transpiration,
photosynthetic efficiency, plant growth and responses to fertilizers and irrigation (Kumar,
2009; Rouphael et al., 2010). The close relation between LA, plant growth and productivity
make this parameter a fundamental component of crop growth models when it is not
possible to proceed with a destructive sampling. A large number of methods, either
destructive or not, have been developed to measure or estimate plant leaf area. The most
common approach is to develop ratios and regression estimators by using simple measured
leaf parameters such as length (L) and width (W). This method does not require leaves to be
detached, permitting a continuous monitoring during the plants growth period, reducing the
variability associated with destructive sampling procedures (Kumar, 2009). Non-destructive
methods, based on linear measurements, are quicker and easier than destructive ones to
execute, and are characterized by high precision and accuracy as demonstrated for crops,
like chestnut (Serdar and Demirsoy, 2006), faba bean (Peksen, 2007), saffron (Kumar, 2009),
red pepper (De Swart et al., 2004), Euphorbia × lomi (Fascella et al., 2009) and sunflower
(Rouphael et al., 2007).
In some plant species, the presence of abiotic stress can induce morphological
variations in leaves (Akinci and Losel, 2012). Many authors reported the effects of different
kinds of stress on leaf morphology of several species, both spontaneous and cultivated (Fini
Acta Hortic. 1104. ISHS 2015. DOI 10.17660/ActaHortic.2015.1104.14
XXIX IHC – Proc. Int. Symp. on Ornamental Horticulture in the Global Greenhouse
Ed.: R.A. Criley
89
and Ferrini, 2010). Moreover, a relation between leaf morphology of Myrtus communis L.
(myrtle) and solar radiation was evidenced by other authors (Mendes et al., 2001; Tattini et
al., 2006). M. communis is an evergreen sclerophyll shrub that grows naturally in the
Mediterranean basin, but also extends to eastern Iran and Afghanistan. It occurs in
woodlands, maquis and garrigues, but also in coastal areas where drought stress and high
solar radiation are very frequent.
The aim of this study was to develop a model that could be able to predict LA of myrtle
plants from single parameters (L and/or W) and, if this model could be used as predictor of
plant sub-optimal water availability condition.
MATERIALS AND METHODS
The experiment was conducted at the experimental farm of the Agricultural Research
Council, Research Unit for Mediterranean Flower Species of Palermo (38°5’N, -13°30’E, 23 m
a.s.l.), Italy, during the 2012 growing season.
Local myrtle genotypes, from spontaneous Sicilian maquis, were used to develop the
leaf area prediction model. For model validation two-year-old myrtle plants grown outdoors
were used. Plants were grown in plastic pots containing 3 L of peat and perlite (2:1, v/v), in
single rows, at a plant density of 5 plants m-2.
Plants were daily irrigated with a drip-irrigation system by using two different water
volumes: “T1” plants received 100% of the daily evapo-transpiration (ETe), “T2” plants
received 66% of ETe. Evapo-traspiration was estimated by a gravimetric method.
All plant leaves were collected in June (end of the blooming time) from six plants per
treatment.
Leaf length was measured from the lamina tip to the point of intersection of the lamina
and stem, while leaf width was measured as the maximum perpendicular dimension of the
length. LA and leaf dimensions were measured using a leaf area meter (“WinDIAS 3” Image
Analysis System for Leaves, DELTA-T Devices Ltd., Cambridge, England) calibrated to
0.01 cm2.
For model design, a double approach was carried out. A first series of model was
designed (with an upper-case letter followed by lower-case m, e.g., Am) on a sample of
mature leaves, from the middle part of the stem, with a L/W ratio between 1.8 and 2.0. A
second series of model was designed (upper-case letter followed by lower-case t, e.g., At)
with total plant leaves.
A sample of 216 healthy leaves were analyzed for building of the first model series,
while for the construction of second model series, 3215 healthy leaves were analyzed.
For both model building, LA was regressed with L, W, L2, W2, and the product L×W and
2
2
L ×W . Model equation (coefficients (b) and constants (a)) and mean square error (MSE),
means area and coefficient of determination (R2) were also reported.
Model validation was performed with the LA data measured in plants irrigated with
different water regimes. A sample of 200 leaves was made by random selection from the
total leaf area of six plants per treatment. The LA of individual leaves was predicted using
the best model selected, and was compared with the measured LA.
RESULTS AND DISCUSSION
Regression analyses of LA versus L and W revealed several models that could be used
for estimating the area of individual myrtle leaves (Tables 1 and 2).
The regression estimation model, developed on a sample of mature leaves, resulted in
many equations having a coefficient of determination (R2) over 0.9 (Table 1). Amongst the
single parameter equations based on leaf length, the quadratic equation indicated as “Bm”
was selected (Figure 1) for achieving the same estimated mean area as compared to that
measured (2.744 vs. 2.744 cm2), and for the high coefficient of determination (R2=0.92). This
model can also easily be used in physiological and agronomic studies.
90
Table 1. Regression estimation model of leaf area on mature leaves of Myrtus communis.
Statistical population: 216 units. Coefficient of determination (R2) and mean square
errors (MSE). Measured mean area: 2.744 (cm2).
Model
Am
Bm
Cm
Gm
Hm
Im
Lm
Mm
Nm
Parameter
Equation model
Length
y=2.1489x-2.5454
Length
y=0.3745x2+0.1578x-0.029
Length
y=0.4828x1.8759
Length×width
y=0.5118x+0.1425
Length×width y=-0.0008x2+0.5227x+0.1124
Length×width
y=0.5907x0.9435
2
2
Length ×width
y=0.035x+1.6432
Length2×width2 y=-0.0001x2+0.0571x+1.2417
Length2×width2
y=0.5907x0.4718
Mean area
2.744
2.744
2.725
2.744
2.744
2.725
2.743
2.827
2.725
MSE
0.0192
0.0174
0.0173
0.0167
0.0167
0.0166
0.0246
0.0213
0.0166
R2
0.897
0.916
0.928
0.921
0.921
0.931
0.850
0.909
0.931
Table 2. Regression estimating model of leaf area on total plant leaves in Myrtus communis.
Statistical population: 742 units. Coefficient of determination (R2) and mean square
errors (MSE). Measured mean area: 2.818 (cm2).
Model
At
Bt
Ct
Dt
Et
Ft
Gt
Ht
It
Lt
Mt
Nt
Parameter
Length
Length
Length
Width
Width
Width
Length×width
Length×width
Length×width
Length2×width2
Length2×width2
Lengtht2×width2
Equation model
y=0.4338x+1.2944
y=-0.0341x2+0.6874x+0.9129
y=1.5265x0.5057
y=0.3022x+0.9284
y=-0.0261x2+0.4963x+0.6365
y=1.0698x0.5055
y=0.5153x+0.3808
y=-0.0027x2+0.5497x+0.2955
y=0.729x0.8708
y=0.0376x+1.7538
y=-0.0002x2+0.0605x+1.4003
y=0.729x0.4354
Mean area
2.386
2.414
2.417
1.466
1.431
1.418
2.818
2.819
2.783
2.817
2.749
2.783
MSE
0.0294
0.0280
0.0523
0.0944
0.0958
0.0965
0.0261
0.0261
0.0260
0.0323
0.0292
0.0260
R2
0.895
0.920
0.921
0.670
0.693
0.695
0.862
0.862
0.881
0.786
0.840
0.881
Figure 1. Regression model validation. On left side, Bm model based on leaf length (L) of
mature leaves, on the basis of L/W ratio. On right side, Bt model based on leaf L
using total plant leaves.
Generally regression estimation models developed on all plant leaves are less accurate
with respect to the previous model based on mature leave (Table 2), due to the leaf-shape
polymorphism noted. A frequency distribution of the leaf dimensions of L/W ratio showed
91
that young leaves are more circular (L/W=1.4) than adult leaves, which were more stretched
(L/W>1.8) (Figure 2). Models based on both dimensional parameters, L and W, had a lower
coefficient of determination as compared with models with only L, but the mean areas
estimated were closer to those measured (Figure 1). Amongst the single parameter
equations based on leaf length, the quadratic equation indicated as “Bt” was selected for
achieving a high coefficient of determination (R2) for the estimated mean leaf area and low
MSE value (Table 2). Regression models evidenced as the equations based on L and W are
more accurate, in terms of mean areas estimation, than equations based on one parameter
because the sample of measured leaves included leaves with different L and same W
(Figure 3).
Figure 2. Frequency distribution of L/W ratio of myrtle leaves. On the left side: distribution
using plants cultivated under optimal water availability. On the right side:
distribution using plants under water stress.
Figure 3. Regression model validation on total plant leaves. Square symbols represent
model Bt using L/W ratio. Circle symbols represent model Gt using leaf L.
These results are in agreement with those reported by De Swart et al. (2004) using red
pepper. The analysis of leaf shape distribution (different L/W ratios in total plant leaves)
possibly reveals the presence of new leaves emerging. This possibility is linked to the
presence of different elongations of leaf lamina with the same W.
In addition, a close relation is known between drought stress and leaf emergence
during the growth period (Xu et al., 2009). The application of the regression model on plants
92
subjected to different irrigation regimes, was not able to evidence the presence of stress
conditions during the growth period (Figure 4), as reported on other species (Blanco and
Folegatti, 2005). In contrast, a significant difference was recorded in the frequency
distribution of L/W ratio between well watered and stressed plants, confirming the
possibility to use this method to evaluate the presence of drought stress.
Figure 4. Regression model estimation used for evaluating the effect of water stress on
potted plants of Myrtus. Square symbols represent model Gt, circle symbols
represent model Bt using total plant leaves; rhombus symbols represent model
Bm using mature leaves.
CONCLUSIONS
Myrtle plants generally show leaf shape polymorphism. The model developed using
mature leaves (indicated as Bm), showed that the correlation between measured and
predicted areas was high. The models using the single parameter L (Equations Bt and Bm)
were more acceptable for estimating LA, due to their low R2 and MSE values. Equation G
based on L and W measurements, could also estimate LA accurately, but doubled the time
required for leaf measurement. The quadratic model developed in this study did not detect
evidence of the effect of different water availability, but the study of frequency distribution
of L/W ratio can be used to evaluate the presence of drought stress during the growth
period.
ACKNOWLEDGEMENTS
The present work was supported by the Italian Ministry of Education, University and
Research with the project PON “Sustainable production of pot plant in Mediterranean
Environment” - National Operational Programme “Research and Competitiveness” 20072013 PON “R & C”.
93
Literature cited
Akinci, S., and Losel, D.M. (2012). Plant water-stress response mechanisms. In Plant Water-Stress Response
Mechanisms, Water Stress, P.I.M.M. Rahman, ed. (InTech), p.15–42.
Blanco, F.F., and Folegatti, M.V. (2005). Estimation of leaf area for greenhouse cucumber by linear measurements
under salinity and grafting. Sci. Agric. 62, 305–309 http://dx.doi.org/10.1590/S0103-90162005000400001.
De Swart, E., Groenwold, R., Kanne, H.J., Stam, P., Marcelis, L.F.M., and Voorrips, R.E. (2004). Non-destructive
estimation of leaf area for different plant ages and accessions of Capsicum annuum L. J. Hortic. Sci. Biotechnol. 79,
764–770.
Fascella, G., Maggiore, P., Zizzo, G.V., Colla, G., and Rouphael, Y. (2009). A simple and low-cost method for leaf area
measurement in Euphorbia × lomi Thai hybrids. Advances in Horticultural Science 23, 57–60.
Fini, A., and Ferrini, F. (2010). Growth, leaf gas exchange and leaf anatomy of three ornamental shrubs grown
under different light intensities. Europ. J. Hortic. Sci. 75, 111–117.
Kumar, R. (2009). Calibration and validation of regression model for non-destructive leaf area estimation of
saffron (Crocus sativus L.). Sci. Hortic. (Amsterdam) 122, 142–145 http://dx.doi.org/10.1016/j.scienta.
2009.03.019.
Mendes, M.M., Gazarini, L.C., and Rodrigues, M.L. (2001). Acclimation of Myrtus communis to contrasting
Mediterranean light environments - effects on structure and chemical composition of foliage and plant water
relations. Environ. Exp. Bot. 45, 165–178 http://dx.doi.org/10.1016/S0098-8472(01)00073-9. PubMed
Peksen, E. (2007). Non-destructive leaf area estimation model for faba bean (Vicia faba L.). Sci. Hortic.
(Amsterdam) 113, 322–328 http://dx.doi.org/10.1016/j.scienta.2007.04.003.
Rouphael, Y., Colla, G., Fanasca, S., and Karam, F. (2007). Leaf area estimation of sunflower leaves from simple
linear measurements. Photosynthetica 45, 306–308 http://dx.doi.org/10.1007/s11099-007-0051-z.
Rouphael, Y., Mouneimne, A.H., Ismail, A., Mendoza-De Gyves, E., Rivera, C.M., and Colla, G. (2010). Modeling
individual leaf area of rose (Rosa hybrida L.) based on leaf length and width measurement. Photosynthetica 48, 9–
15 http://dx.doi.org/10.1007/s11099-010-0003-x.
Serdar, U., and Demirsoy, H. (2006). Non-destructive leaf area estimation in chestnut. Sci. Hortic. (Amsterdam)
108, 227–230 http://dx.doi.org/10.1016/j.scienta.2006.01.025.
Tattini, M., Remorini, D., Pinelli, P., Agati, G., Saracini, E., Traversi, M.L., and Massai, R. (2006). Morpho-anatomical,
physiological and biochemical adjustments in response to root zone salinity stress and high solar radiation in two
Mediterranean evergreen shrubs, Myrtus communis and Pistacia lentiscus. New Phytol. 170, 779–794
http://dx.doi.org/10.1111/j.1469-8137.2006.01723.x. PubMed
Xu, F., Guo, W., Xu, W., Wei, Y., and Wang, R. (2009). Leaf morphology correlates with water and light availability:
what consequences for simple and compound leaves? Prog. Nat. Sci. 19, 1789–1798 http://dx.doi.org/10.1016/
j.pnsc.2009.10.001.
94