Aquatic Botany 131 (2016) 51–56
Contents lists available at ScienceDirect
Aquatic Botany
jou rnal homepage: www.elsevier.com/locate/aquabot
Short communication
Morphological study of Lemna minuta Kunth, an alien species often
mistaken for the native L. minor L. (Araceae)
∗
Simona Ceschin , Ilaria Leacche, Stefano Pascucci, Silverio Abati
Roma Tre University, Department of Science, Viale G. Marconi 446, Rome 00146, Italy
a r t i c l e i n f o a b s t r a c t
Article history: Lemna minuta Kunth and Lemna minor L. are two small aquatic floating plants easily mistakable for similar
Received 16 April 2015
morphology. The need to distinguish with certainty these two species is a relevant issue, not only from a
Received in revised form 20 January 2016
floristic viewpoint, but also for establishing management plans in wetlands where L. minuta is an invasive
Accepted 30 January 2016
alien species as in Europe. The vein number and frond length are considered by most authors as the main
Available online 28 February 2016
morphological characters discriminating the two species. However, in this study the use of these two
characters has not been effective for the determination of 43 specimens out of 248 living specimens of
Keywords:
Lemna collected from 15 different wetlands of the Central Italy. Statistical analyses of the dataset made it
Alien species
possible to better define the variability of the morphological characters of these two species and to identify
Floating plant
additional diagnostic characters to use for improving the morphological discrimination between them.
Congeneric species
Decision Tree Among these characters, width, apex and shape of the frond, were the most helpful. Decision Trees were
Duckweed elaborated for differentiating L. minor from L. minuta with greater efficiency both in the laboratory and in
Morphological characterization the field. The increase in percentage of correct determination of Lemna specimens through measurement
and parallel utilization of the emerged morphological characters will facilitate the botanists’ activities,
but also will have practical implications, such as the ability to contribute better to the decision-making
system in drawing up plans for plant biodiversity protection and/or eradication of the alien species.
© 2016 Elsevier B.V. All rights reserved.
1. Introduction five aliens: L. minuta, L. aequinoctialis Welw., L. perpusilla Torrey, L.
turionifera Landolt and L. valdiviana Phil. (DAISIE 2008).
Aquatic ecosystems are among the most threatened habitats The Lemna’s vegetative body (frond) is a complex of tissues
worldwide (Millenium Ecosystem Assessment, 2005) and among with few differentiations, reduced morphological features, and
the most vulnerable to biological invasions (Vitousek et al., 1997; frequently lacking of reproductive structures. This scarcity of char-
Capers et al., 2007). For a better conservation of their native com- acters for distinguishing the species of Lemna, makes the taxonomic
munities, it becomes necessary to correctly verify the presence of discrimination between them difficult (Landolt 1986; Donald et al.,
invasive alien species and then to undertake management actions 2002; Wang et al., 2010), and in particular between L. minor and L.
that can limit the spread and potential impact of these species in minuta, which exhibit very similar morphological characters (Lucey
the invaded ecosystem. 2003; Mifsud 2010). Thus, these two Lemna species are easily mis-
One invasive plant species widespread in European standing takable between them.
inland waters is Lemna minuta Kunth (Verloove 2006; Branquart In literature, for the morphological discrimination between
et al. 2007; Thiébaut 2007; Celesti-Grapow et al., 2009; Lukács et al., these two species there is no complete concordance between
2014), which is among the smallest and fastest growing flower- authors due to different and sometimes conflicting set of pheno-
ing plants (Landolt 1986). In Europe, the Lemna genus (duckweeds; typic characters to use (e.g., Landolt 1992; Wolff and Landolt, 1994;
Araceae) is represented by three native species: Lemna minor L., the Stace 2010; Eckehart 2011). For example the number of fronds
most common European duckweed, L. gibba L. and L. trisulca L., and is not considered by Stace (2010) and Wolff and Landolt (1994)
as a character for discriminating L. minuta from L. minor, while
other authors recognize this character as diagnostic although they
disagree among themselves about the numerical range associated ∗
Corresponding author.
with L. minor and L. minuta. However, most authors agree that
E-mail addresses: [email protected] (S. Ceschin), [email protected]
primarily vein number and secondarily frond length are the best
(I. Leacche), [email protected] (S. Pascucci),
[email protected] (S. Abati). characters for differentiating the two species (Landolt 1992; Wolff
http://dx.doi.org/10.1016/j.aquabot.2016.01.005
0304-3770/© 2016 Elsevier B.V. All rights reserved.
52 S. Ceschin et al. / Aquatic Botany 131 (2016) 51–56
Table 1
and Landolt, 1994; Galán and Castroviejo 2008; Stace 2010). Par-
Repartition of the Lemna specimens based on vein number and length frond.
ticularly, L. minor is usually described by 3–5 veins, frond 1–8 (10)
mm long, while L. minuta by 0–1 vein and frond 0.8–2.0(4.0) mm Length frond (mm)
long. <2 2–3 >3
However, both characters, although necessary to distinguish L. Vein n 0–1 58 (23%) (A) 33 (13%) (B) 10 (4%) (C)
minor from L. minuta, show some limits. Vein number is a character >1 11 (4%) (D) 38 (15%) (E) 98 (40%) (F)
often scarcely observable either in the field or in the laboratory,
even on samples treated properly (Landolt 1992); frond length
U-CMAD3). To better observe some qualitative characters, such
is a character partially overlapping between the two species (e.g.
as vein number and aerenchyma extension, the fronds were dis-
Landolt 1992; Wolff and Landolt, 1994; Stace 2010).
◦
coloured with ethanol 95 for at least 24 h following Landolt (1992)
Therefore it is necessary to identify other morphological char-
and subsequently observed under a stereoscope with a light source
acters, which combined with those mentioned above, can better
from below.
distinguish L. minor from L. minuta. This need is even more relevant
For the analysis of quantitative characters, digital photos were
in Europe since L. minor is a native species while L. minuta is an
processed with the program ImageJ vers. 1.47. The measurements
invasive alien species which often shares the same habitats with L.
were carried out only on the major frond, if the specimen consisted
minor, threatening its presence (Landolt 1986; Iberite et al., 2011;
of contiguous fronds.
Mazzini et al., 2014).
The present study aimed to better define the ranges of the mor-
phological characters usually used for discriminating L. minor from 2.2. Statistical analysis
L. minuta, and to identify additional morphological characters for
better differentiating these two species. Decision Trees were devel- Firstly, the Lemna specimens were characterized on the basis
oped for differentiating the two species directly in the field or in the of the vein number and frond length, the two main mor-
laboratory with the minimum number of morphological characters phological characters recognized in literature as discriminating
and the lowest degree of error. between the two species. In particular, the specimens were
assigned to one of two Lemna species on the basis of non-
overlapping ranges of these two characters to allow a certain
2. Materials and methods
assignment.
On these specimens certainly assigned to L. minuta or L. minor
2.1. Collection and analysis of Lemna specimens
and belonging to populations sampled at different sites, the mor-
phological variability was tested (inter-population variability) by
A morphological comparison between L. minuta and L. minor was
applying the non-parametric Kruskal–Wallis test (H) on the medi-
carried out. This analysis is based on observations and measure-
ans of the measured values for the frond length.
ments of morphological characters made on 248 living specimens
Subsequently, the whole dataset was processed through
of Lemna collected in 2013–2014 from 15 wetlands of the Central
Non-Metric Dimensional Scaling (NMDS), for evaluating the con-
Italy, between May and September.
tribution of each variable (character) for discriminating the two
Only the vegetative characters were chosen for the investigation
species. Gower index for quantitative variables and Hamming index
because no specimen was observed with flowering structures or
for qualitative variables were used as measures of similarity dis-
fruits, as indeed happens in most of specimens occurring in Europe,
tance and a biplot was obtained correlating the scores of the two
especially as regards L. minuta (Landolt 1980; Bramley 1997). In
first ordination axes with each morphological character (Kendall
addition, the characters related to the roots (total length, root cap)
concordance coefficient, ).
were not considered because their correct measurement would
Exclusively for quantitative characters, a Spearman correlation
not have been possible in some specimens, since the roots of both
was performed for identifying any redundant characters and for
species are very fragile and break easily.
highlighting potential differences in the morphological plastic-
A total of fourteen morphological characters, including both
ity of the two species. To check significant differences between
qualitative (*) and quantitative ones, were selected after con-
the two species on the basis of each morphological character
sulting the literature describing the two investigated Lemna
considered, Mann–Whitney tests (U) on quantitative characters,
species. 2
and test on qualitative ones, were performed. Non-parametric
tests were chosen for a small number of each population’s
(1) frond shape (elliptic-obovate)*
specimens (Kruskal–Wallis test) or for non-normal distribution
(2) frond symmetry (symmetric-asymmetric)*
of some variables, tested by Shapiro–Wilk test (Mann–Whitney
(3) frond colour (light green-dark green)* test).
(4) frond border (hyaline all around-exclusively at frond-base) *
Finally, for producing statistically significant dichotomous keys,
(5) frond apex (rounded-acuminate)*
built with the smallest number of dichotomies and with the best
(6) aerenchyma (extended up apex-extended up ¾ of the frond
discriminating variables, a recursive binary partitioning with a
area)*
permutation test for producing Decision Tree, was used (party R
(7) frond length (mm)
package). Four different Decision Trees (DT) were drawn up, tak-
(8) frond width (mm)
ing into account if the assignment of specimens to one or the other
(9) number of contiguous fronds (n)
species of Lemna was done directly in the field or in the laboratory.
2
(10) frond area (mm )
The DT f (Decision Tree field) considers that in the field it is very
(11) frond length/frond width
difficult to observe some characters (i.e. vein number, hyaline edge,
(12) distance between frond-base and root-attachment (mm)
aerenchyma) or to obtain a high precision in measurement of others
(13) base-root attachment distance/frond length
(i.e. frond length, width, area). For these reasons, such characters
(14) number of veins (n)
are not considered diagnostic in the field or their measurement
is approximated to mm. DT fqual (Decision Tree field qualitative)
The specimens of both species were observed under a stereo- was drawn up for use in the field, on the basis of only qualitative
scope (Olympus SZX16) equipped with digital camera (Olympus characters. DT l (Decision Tree laboratory) was drawn up for the
S. Ceschin et al. / Aquatic Botany 131 (2016) 51–56 53
Fig. 1. NMDS based on 14 morphological characters.
Fig. 2. Morphological features of L. minor and L. minuta.
◦ ◦
For the qualitative characters, the frequency values are showed in brackets, while for the quantitative ones, the 5 and 95 percentile values are reported, with maximum
and minimum values in brackets. Proportion and overlap percentages between the two species were reported.
laboratory, where it is possible to establish the vein number. Also All statistical analyses were performed with R (R Core Team,
a DT lwv (Decision Tree laboratory without veins) was produced 2014), except NMDS that was performed with PAST package ver.
for cases when the observation of veins is impossible or at least 1.94b (Hammer et al., 2001).
uncertain, even in cases where the correct discoloration procedure
of specimens has been adopted.
54 S. Ceschin et al. / Aquatic Botany 131 (2016) 51–56
Fig. 3. Decision Trees for discriminating Lemna minor from Lemna minuta. DT l: Decision Tree laboratory starting from the non-observation of the vein number; DTwv: Decision
Tree laboratory starting without the observation of the vein number; DT f: Decision Tree field; DT fqual: Decision Tree field utilizing only qualitative characters. Specimens’
number (n) and p value are reported.
L. minuta and L. minor proportions are displayed in dark grey and light gray, respectively.
3. Results and discussion of Lemna (17% of all collected specimens) would be of uncertain
determination, using only these two morphological characters.
3.1. Lemna specimens’ analysis These results highlight limitations in the use of these two char-
acters for discriminating between the two species. Such limitations
The morphological analysis of the 248 living specimens, based come down to two main problems. The first is linked to the fact that
on vein number and frond length, has identified six separated cases the veins are sometimes scarcely visible (Landolt 1992; Lucey 2003)
(Table 1). Case A, represented by specimens with fronds character- and the observation of a single vein does not necessarily exclude
ized by 0–1 vein and length <2 mm, which, following literature, the presence of others not clearly observable. The second prob-
can be assigned with certainty to L. minuta; case F with fronds hav- lem is related to the partial overlap of the frond length values (e.g.,
ing a vein number >1 and length >3 mm, which can be determined Wolff and Landolt, 1994; Lucey 2003; Mifsud, 2010), which can be
with certainty as L. minor. The specimens included in cases D and of 2–3 mm for both species. Thus, vein number and frond length
E, although representing fronds with reduced length, are charac- are good characters for discriminating the two species but only for
terized by a vein number >1. This second character leads these those Lemna specimens with visible veins, possibly with more than
specimens being attributed to L. minor, giving more weight to the 1 vein, and frond length less than 2 or greater than 3 mm.
vein number than to frond length. Case B includes fronds with 0–1 As regards the morphology variability between populations
vein, but with an intermediate length of 2–3 mm; case C, fronds belonging to the group A for L. minuta and the groups D, E and F for
with 0–1 vein, but length >3 mm. On the basis of these values, the L. minor, it is high and statistically significant (H = 15.74, p < 0.01 for
specimens constituting cases B and C cannot be assigned with cer- L. minuta; H = 75.35, p < 0.001 for L. minor) in both species, indicat-
tainty to one of the two species. In particular, a total of 43 specimens ing a wide variability among specimens belonging to populations
sampled in different sites.
S. Ceschin et al. / Aquatic Botany 131 (2016) 51–56 55
3.2. Identification of additional morphological characters for Tree (Fig. 1, cases B and C). The ameliorative contribution of this
distinguishing L. minuta from L. minor DT is due to a better definition of the range for the frond length
character. In DT lwv, frond width and length are the best characters,
The ordination biplot (NMDS) shows that the two groups with an 8% of probability of arriving at an incorrect determination.
assigned with certainty (A and D + E + F) differ mainly along the first As regards the DT f, it is significant that frond width and length,
axis (Fig. 1). Most of the vectors of the quantitative characters show measured without decimals, are the two characters most effective
a spatial direction towards the dataset attributed to the L. minor in discriminating the two species in the field. Although these two
populations, underlining the larger size of this species compared characters are the same of DT lwv, the case frequency of incorrect
to L. minuta. Among quantitative characters, the most correlated determination increases to 11% on the total cases. This percentage
with the first axis are frond length, width and area, while among increases to 20%, if the determination in the field is made using only
qualitative ones are frond shape, apex and symmetry. qualitative characters (DT lqual). This highlights once again that the
Applying Mann–Whitney test on data related to quantita- diagnostic role of the quantitative characters is greater than that of
tive characters, frond length, width, area and vein number are the qualitative ones.
mainly confirmed as good and significantly discriminating char-
acters (U < 400, p < 0.001), since the two species show low values
4. Conclusions
of overlapping for these characters. However, among these char-
acters, some are strongly correlated with each other, a fact that
This study highlighted that the exclusive use of the morphologi-
makes it possible to consider only some of them in order to avoid
cal characters more commonly used in literature for discriminating
redundancies. In particular, considering the entire dataset, length,
L. minor from L. minuta (i.e. length frond and vein number), leads
width, frond area and distance-based node are strongly corre-
to 17% of non-resolved cases within our dataset. A better definition
lated (rs > 0.84). However, it should be noted that, if the correlation
of the ranges of the characters describing the morphology of these
between the characters is calculated by keeping the two species’
duckweeds was useful for better differentiating the two species.
subsets separate, frond width remains well correlated with length
Moreover, the identification of other useful diagnostic characters,
in the case of L. minor (rs = 0.84), while the correlation is lower for L.
such as frond width, apex and secondarily frond shape, made it pos-
minuta (rs = 0.72). This suggests that frond shape of L. minor is more
sible to decrease the margin of identification error. The drawing up
stable (i.e. frond length and width vary in a strictly proportional
of Decision Trees has made it possible to improve the determination
way) respect to L. minuta that instead shows a more variable frond
of the two species in the laboratory and in the field.
shape.
However, only the measurement and parallel observation of the
Among qualitative characters, frond apex was the best char-
selected morphological characters can increase the percentage of
acter, followed by frond shape. Aerenchyma, colour and frond
correct determination of the sampled Lemna specimens.
symmetry do not significantly discriminate the two species, show-
Minimizing the error of determination between L. minor and L.
2
ing an almost complete overlapping ( test; p > 0.05). The result on
minuta is of great importance not only from floristic viewpoint but
frond symmetry does not match with the findings of other authors,
also for conservation and management of wetlands colonized by
who instead consider such character as diagnostic and highly dis-
these species (Iberite et al., 2011). Indeed, this becomes a require-
criminating regarding the two species (Wolff and Landolt, 1994;
ment in Europe since the alien invasive L. minuta can be easily
Eckehart 2011).
confused with the native L. minor (Lucey 2003; Mifsud 2010), with
Based on our results, it has been possible to better define vari-
consequential negative effects linked to its undisturbed spread,
ability degree or ranges of each characters analysed in the two
such as the direct competition with native aquatic flora (Janes et al.,
Lemna species (Fig. 2). Particularly noteworthy is the range more
1996; Iamonico et al., 2010), and, more generally, the alteration
accurately defined by frond length character in both species. Con-
to the structure and function of the invaded aquatic ecosystem
sidering the relative proportions and overlap percentages between
(Dussart et al., 1993; Marrone and Naselli-Flores, 2011).
the two species calculated for qualitative and quantitative charac-
ters, respectively (Fig. 2), it is emphasised that quantitative ones,
Acknowledgements
especially vein number, length, width and frond area, are the
best discriminating characters, showing low overlap percentages
The authors are grateful to Dr. Ilaria Leacche and Dr. Fabrizio
between the two species, i.e. <13%. This means that the use of one
Piccari, for their help during fieldwork and for the documenta-
of these characters, and the defined thresholds for them, made it
tion provided. They also thank Dr. Martin Bennett for the linguistic
possible to discriminate between the two species in about 90% of
review of this manuscript.
cases, while only 10% of the assignments to one or the other species
is arbitrary. The approach of considering more characters than one
greatly reduces the error percentage and thus increases the possi- Appendix A. Supplementary data
bility of correctly assigning the Lemna specimens. The assignment
error will decrease further if the measurement of the best quantita- Supplementary data associated with this article can be found, in
tive characters is combined with the evaluation of those qualitative the online version, at http://dx.doi.org/10.1016/j.aquabot.2016.01.
characters which to be proved the most discriminating, such as 005.
apex and frond shape.
References
3.3. Decision Tree (DT)
Bramley, J., 1997. Flowering in British Lemna: a rare, cyclic or simply overlooked
On the basis of our findings, the whole dataset was used for phenomenon? Freshwater Forum 7, 2–6.
Branquart, E., Stiers, I., Triest, L., Vanderhoeven, S., Landuyt, W., Van Rossum, F.,
drawing up four Decision Trees (Fig. 3). The DT l is, as expected, the
Van Verloove, F., 2007. Harmonia database, Available from: http://ias.
most able to discriminate the two species, based on the observation biodiversity.be.
of vein number and frond length measured below the millimetre. Capers, R.S., Selsky, R., Bugbee, G.J., White, J.C., 2007. Aquatic plant community
invasibility and scale-dependent patterns in native and invasive species
The use of this DT leads to an incorrect determination only in 3% of
richness. Ecology 88, 3135–3143.
cases, thus improving the incorrect assignment of the 17% obtained
DAISIE, 2008. European invasive alien species gateway. http://www.europe-aliens.
considering the same characters but without using such Decision org/.
56 S. Ceschin et al. / Aquatic Botany 131 (2016) 51–56
Celesti-Grapow, L., Alessandrini, A., Arrigoni, P.V., Banfi, E., Bernardo, L., Bovio, M., Lukács, B.A., Mesterházy, A., Vidéki, R., Király, G., 2014. Alien aquatic vascular
Brundu, G., Cagiotti, M.R., Camarda, I., Carli, E., et al., 2009. Inventory of the plants in Hungary (Pannonian ecoregion): historical aspects, data set and
non-native flora of Italy. Plant Biosyst. 143 (2), 386–430. trends. Plant Biosyst., http://dx.doi.org/10.1080/11263504.2014.987846.
Donald, H.L., Crawford, D.J., Landolt, E., Gabel, J.D., Kimball, R.T., 2002. Phylogeny Marrone, F., Naselli-Flores, L., 2011. Primo reperto di una lenticchia d’acqua
and systematics of Lemnaceae, Duckweed Family. Syst. Bot. 27, 221–240. alloctona in Sicilia: Lemna minuta Kunth (Araceae, Lemnoideae). Nat. Sic. 35
Dussart, G., Robertson, J., Bramley, J., 1993. Death of a lake. Biol. Sc. Rev. 5 (5), 8–10. (2), 179–235.
Eckehart, J.J., 2011. Rothmaler Exkursionsflora von Deutschland. Gefäßpflanzen: Mazzini, I., Ceschin, S., Abati, S., Piccari, F., Rossi, A., Faranda, C., Gliozzi, E., 2014.
Grundband., Auflage, 20. Spektrum Akademischer Verlag, pp. 133–134. Ostracod communities associated to aquatic macrophytes in an urban park:
Galán, A., Castroviejo, S., 2008. Lemnaceae. In: Castroviejo, S., Colleagues (Eds.), the example of the Caffarella Valley (Park of the Appia Antica, Rome,Italy).
Flora iberica, 18. CSIC Real Jardín Botánico, Madrid http://www.floraiberica.es/. Inter. Rev. Hydrobiol. 99, 1–11.
Hammer, Ø., Harper, D.A.T., Ryan, P.D., 2001. PAST: paleontological statistics Millenium Ecosystem Assessment, 2005. Ecosystems and human Well-being:
software package for education and data analysis. Palaeontol. Electron. 4 (1), Wetlands and water synthesis. World Resources Institute, Washington.
1–9. Mifsud, S., 2010. First occurrences of Lemna minuta Kunth (Fam: Lemnaceae) in the
Iamonico, D., Abati, S., Iberite, M., 2010. Lemna minuta Kunth (Araceae) nel Lazio Maltese Islands. C Medit. Nat. 5 (2), 1–4.
(Italia centrale): note morfologiche e osservazioni sui caratteri d’invasività. In: R Core Team, 2014. R: A language and environment for statistical computing. R
Proc. 18th Meeting Forum Natura Mediterraneo on “Le specie aliene nel Foundation for Statistical Computing, Vienna, Austria http://www.R-project.
Mediterraneo”, 2010 March 20–21; Paliano, Italy, Available from: http://www. org/.
naturamediterraneo.com/primoconvegnoNM/. Stace, C., 2010. New Flora of the British Isles. Cambridge University, Cambridge.
Iberite, M., Iamonico, D., Abati, S., Abate, G., 2011. Lemna valdiviana Phil. (Araceae) Thiébaut, G., 2007. Non-indigenous aquatic and semiaquatic plant species in
as a potential invasive species in Italy and Europe: taxonomic study and first France. In: Gherardi, F. (Ed.), Biological Invaders in Inland Waters: Profiles,
observations on its ecology and distribution. Plant Biosyst. 145, 751–755. Distribution and Threats. Springer, pp. 209–229.
Janes, A.R., Eaton, W.J., Hardwick, K., 1996. The effects of floating mats of Azolla Verloove, F., 2006. Catalogue of the Neophytes in Belgium (1800–2005). Scrip. Bot.
filiculoides Lam and Lemna minuta Kunth on the growth of submerged Belg. 39, 89.
macrophytes. Hydrobiologia 340, 23–26. Vitousek, P.M., D’Antonio, C.M., Loope, L.L., Rejmanek, M., Westbrooks, R., 1997.
Landolt, E., 1980. Key to the determination of taxa within the family of Lemnaceae. Introduced species: a significant component of human-caused global change.
Veroff. Geobot. 70, 13–21. New Zealand J. Ecol 21, 1–16.
Landolt, E., 1986. The family of Lemnaceae–A monographyc study. 2. Veröff. Wang, W., Wu, Y., Yan, Y., Ermakova, M., Kerstetter, R., Messing, J., 2010. DNA
Geobot. Inst., ETH, Stifung rübel, Zurich. barcoding of the Lemnaceae, a family of aquatic monocots. Plant Biol. 10, 205.
Landolt, E., 1992. Lemnaceae: a duckweed family. J. Arizona Nevada Acad. Sci. 26, Wolff, P., Landolt, E., 1994. Spread of Lemna turionifera (Lemnaceae), the red
10–14. duckweed, in Poland. Fragm. Florist. Geobot. 39 (2), 439–451.
Lucey, J., 2003. Lemna minuta Kunth (least duckweed) F. Cork (V. C. H5) Irish Bot.
news 13, 5–8.