Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)

X Li; J G Prebble; P J de Lange; J I Raine; L Newstrom-Lloyd

doi:10.1371/journal.pone.0269361

. 2022 Jun 3;17(6):e0269361. doi: 10.1371/journal.pone.0269361

Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)

X Li ^1,^*, J G Prebble ¹, P J de Lange ², J I Raine ¹, L Newstrom-Lloyd ³

Editor: Wolfgang Blenau⁴

PMCID: PMC9165797 PMID: 35657968

Abstract

The very similar appearance of pollen of the New Zealand Myrtaceous taxa Leptospermum scoparium s.l. (mānuka) and Kunzea spp. (kānuka) has led palynologists to combine them in paleoecological and melissopalynological studies. This is unfortunate, as differentiation of these taxa would improve understanding of past ecological change and has potential to add value to the New Zealand honey industry, where mānuka honey attracts a premium price. Here, we examine in detail the pollen morphology of the 10 Kunzea species and a number of Leptospermum scoparium morphotypes collected from around New Zealand, using light microscopy, SEM, and Classifynder (an automated palynology system). Our results suggest that at a generic level the New Zealand Leptospermum and Kunzea pollen can be readily differentiated, but the differences between pollen from the morphotypes of Leptospermum or between the species of Kunzea are less discernible. While size is a determinant factor–equatorial diameter of Leptospermum scoparium pollen is 19.08 ± 1.28 μm, compared to 16.30 ± 0.95 μm for Kunzea spp.–other criteria such as surface texture and shape characteristics are also diagnostic. A support vector machine set up to differentiate Leptospermum from Kunzea pollen using images captured by the Classifynder system had a prediction accuracy of ~95%. This study is a step towards future melissopalynological differentiation of mānuka honey using automated pollen image capture and classification approaches.

Introduction

Honey from New Zealand mānuka (Myrtaceae: Leptospermum scoparium J.R.Forst. et G.Forst. agg.) attracts a premium value [1–3], arising from the medical benefits of its non-peroxide antibacterial activity [4]. This has created a demand from industry and regulators for accurate and cost-effective authentication testing of the product [5].

Melissopalynology–the pollen analysis of honey–is widely applied internationally, especially in Europe and America, in combination with chemical, physical, and sensory characters to indicate the approximate contributions of nectar from various plants to the honey [6–9]. However, using melissopalynology techniques to determine the nectar contribution of Leptospermum scoparium in New Zealand honeys has been hampered by the very similar appearance of Leptospermum scoparium pollen to pollen from New Zealand kānuka (Kunzea spp.). Kunzea Rchb., the closest relative to Leptospermum J.R.Forst. et G.Forst. in New Zealand [10], is widely distributed throughout the two main islands of New Zealand [11] and is often found growing with Leptospermum J.R.Forst. et G. Forst. It is also frequently visited by honeybees (Apis mellifera Linnaeus, 1758).

Until the 1980s three species of Leptospermum were regarded as endemic to New Zealand, L. scoparium J.R.Forst. et G.Forst. (“mānuka”), L. ericoides A.Rich. (“kānuka”) and L. sinclairii Kirk [12]. Thompson transferred L. ericoides to Kunzea, treating L. sinclairii as a synonym [13]. After a full taxonomic revision of the Kunzea ericoides complex including morphological, cytological and molecular variation as well as hybridisation experiments, de Lange [11] recognised ten Kunzea species, all endemic to New Zealand. Variation in Leptospermum scoparium is also well known, for example in flower colour, growth habit and yield of the antibacterial precursor chemical DHA (dihydroxyacetone) in nectar [12, 14–17]. Nevertheless, until recently only one species L. scoparium with two varieties, L. scoparium var. incanum Cockayne and L. scoparium var. scoparium have been accepted for New Zealand [12, 16, 18]. In her Australasian revision of the genus Thompson [13] only accepted the one species, L. scoparium for New Zealand but she did not critically examine New Zealand plants. At the time of writing, a New Zealand Department of Conservation funded revision of that nation’s Leptospermum scoparium is underway. Initial results confirm the findings of others, notably Buys et al. [19] that New Zealand populations of L. scoparium are distinct from those populations in Australia and Tasmania attributed to L. scoparium, and so are endemic. Buys et al. [19] also suggested that New Zealand plants are worthy of taxonomic segregation (see [16] and references therein). While the findings from the taxonomic revision of New Zealand L. scoparium are still pending, the morphotype analysed in this paper as L. aff. scoparium (c) “Waikato Peat Bog” has already been established as the new species L. repo de Lange et L.M.H.Schmid, and others are pending [16].

Because of their wide geographical distribution, and wide ecological envelope, pollen records of Leptospermum and Kunzea have received little focus in New Zealand palaeovegetation studies. As a result, limited effort has been assigned to pollen morphological study of the two forms. For example, in his seminal work on melissopalynology of New Zealand honeys Moar [20] referred to both forms simply as Leptospermum (although this was prior to Thompson’s 1989 transfer of L. ericoides to Kunzea), and in vegetation history studies he later classed them together as Leptospermum type [21]. However, differences in pollen morphology (at least on a statistical basis) had previously been observed by Pike [22], McIntyre [23], and Harris et al. [24], and later by Moar himself [25].

The increasing economic importance of mānuka honey has led to new focus on Leptospermum and Kunzea pollen [2]. In 2014, the Ministry for Primary Industries initiated preliminary studies to set up a standard for manuka honey. As part of this, we carried out a pilot study to investigate the possibility of differences between pollen of male and hermaphrodite (bisexual) flowers of New Zealand Leptospermum and Kunzea, both of which are at least in part andromonoecious [11, 26] (pers. obs. Newstrom-Lloyd). It was thought possible that the pollen of male and bisexual flowers may differ slightly in morphology, for instance in size, as has been noted in other andromonoecious plants (e.g., Sagittaria guayanensis Kunth [27]). Based on a small number of plant specimens of Kunzea robusta de Lange et Toelken and Leptospermum scoparium s.l. from the East Cape region we found no discernible difference in pollen of the two flower types within each genus, but a significant difference in average size between pollen of the two genera [28]. This conclusion agreed with the previous results of McIntyre [23] and Harris et al. [24] and was further confirmed by measurements of pollen from a small number of plant samples by Holt & Bebbington [29]. The results of our pilot study are summarised in S1 File.

As a result of these findings, in 2016 we set out to examine in detail the pollen morphology of the 10 New Zealand Kunzea species recently established by de Lange [11], and pollen from Leptospermum scoparium and putative segregates from that species, covering a wide geographic spread of populations throughout the New Zealand range, to determine:

if Leptospermum scoparium s.l. and Kunzea pollen can be differentiated, and
if subpopulations of L. scoparium pollen or interspecific variation in Kunzea exist.

A clear differentiation between New Zealand Leptospermum and Kunzea pollen could be used to establish a pollen-based mānuka honey standard and form the basis for mānuka honey melissopalynological testing. The results of our detailed study are reported here. We have not studied specimens of Australian Leptospermum species which occur in New Zealand as a few small, locally naturalised populations unlikely to be significant sources of honey, except possibly L. laevigatum on Matakana Island [30, 31].

In 2017 the New Zealand Government Ministry for Primary Industries (MPI) established an export standard for mānuka honey based on assay of characteristic chemical marker compounds and the presence of DNA of manuka pollen [1, 32], partly because of the then perceived difficulty in discrimination of Leptospermum and Kunzea pollen. At the present time, all New Zealand export honey passing the MPI tests is accepted as “mānuka” honey, without distinction as to geographic (and thus possibly specific) origin, or whether other nectar sources could be predominant. We contend that melissopalynological analysis remains a useful test because a single analysis can routinely quantify major pollen components (as well as non-pollen entities such as honey-dew fungal spores) and thereby provide a more comprehensive view of the nectar sources of a putative mānuka or kānuka honey. Pollen analysis also requires less specialised equipment and is thus more readily available to individual producers as well as overseas laboratories conversant with applying it as part of Codex Alimentarius requirements for certification of monofloral honeys. It may also be less expensive than the MPI test, especially if automated image recognition is routinely achieved.

Materials and methods

Sample collection

Fifteen geographic/genetic populations from Leptospermum scoparium were selected (see below) across New Zealand, representing 15 or fewer potential species segregates based on growth habit, leaf shape, flower size and colour, capsule shape, size and colour, and chemistry. Additionally, the 10 New Zealand species of Kunzea were selected, along with 10 specimens of putative interspecific Kunzea hybrids. In total, we sampled flowers from 135 herbarium specimens (at least 2 specimens for each Kunzea species and Leptospermum scoparium segregate were taken from both ends of their geographical range) from the Auckland War Memorial Museum herbarium (S1 and S2 Tables). The locations from which the plant specimens were collected are illustrated in Fig 1. Anthers from several fully opened flowers from each herbarium specimen were collected into pre-labelled 15 ml centrifuge tubes. Following the results of our pilot study that demonstrated no difference between the pollen from male and hermaphrodite flowers of Leptospermum scoparium s.l. and Kunzea robusta (S1 File), no attempt was made to separately select male and bisexual flowers in this study.

Fig 1 — Collection localities for (a) *Kunzea* and (b) *Leptospermum scoparium* specimens.

Leptospermum scoparium putative segregates

At the time this study was initiated in 2016 a taxonomic investigation of New Zealand Leptospermum scoparium populations had not been initiated. However, on examination of herbarium specimens (AK) considering the morphological variation evident it was decided to split those collections into putative segregates or “morphotypes”, to see if any could be differentiated by their pollen. While this investigation was in progress, four of the segregates (representing morphotypes “Auckland”, “Waikato Peat Bog”, “East Cape” and “Three Kings” of this paper respectively) were informally published as L. aff. scoparium (a), (c), (d), and (e) as part of a New Zealand-wide conservation assessment of the indigenous vascular flora [33]. In this paper we also examine a further 10 putative segregates (“Surville Cliffs”, “Central Volcanic Plateau”, “Coromandel Swamp”, “Flat Silver”, “Otaipango”, “Papa”, “South Island Mountain”, “Northern South Island”, “Southern South Island” and “Wellington”) and examples of L. scoparium var. incanum with pink-tinged or uniformly pink flowers. Buys et al. [19] included many of these putative segregates in their molecular analysis, finding evidence to merge some of this variation into subclades within a New Zealand (endemic) L. scoparium clade. For the purposes of this study, whilst a taxonomic revision is in progress, we retain usage of the putative segregates as we sampled them for this study, noting that, and as mentioned above, segregate (c) is now formally recognised as L. repo. We also note that putative segregates recognised as “Wellington”, “South Island Mountain”, “Southern South Island”, based on current thinking (P.J. de Lange unpubl.) may be best treated as L. scoparium s.s.–though this needs further study, and so we retain their usage here.

Pollen preparation

Dried herbarium flower samples were rehydrated overnight in deionized water with several drops of filtered detergent. After being vigorously stirred, each of the sample suspensions was sieved through a 90 μm nylon filter cloth to remove non-pollen flower parts. Pollen was then concentrated by centrifugation.

The residual pollen samples were then prepared using Erdtman’s acetolysis method [34]. Each sample was washed with glacial acetic acid, heated in a 9:1 mixture of acetic anhydride and concentrated sulphuric acid for 5 minutes at 95°C in a fume hood, returned to acetic acid, then twice washed in deionized water. Microscope slides were then made using glycerine jelly with safranin stain as the mounting medium.

Light microscopy and manual measurements

All photographs were taken with Zeiss AxioImager microscopes using 100× oil immersion objectives. Pollen grains in polar and equatorial orientations were measured either using an eyepiece graticule or using the measure function in Zen 3.2, an image-processing and analysis software from ZEISS microscopy. Visual measurements were made under the highest magnification available, using a 100× oil immersion objective, 1.6× tube factor, and a 10× eyepiece with an eyepiece graticule of 100 divisions. At this magnification, each division is equivalent to 0.625 μm. Measurements were made to the nearest half division.

Two dimensions which are usually used to characterise angiosperm pollen grain size were measured: polar diameter (Lg max, longitudo: Iversen and Troels-Smith 1950 [35]; P: Erdtman, Nilsson and Praglowski 1992 [34]) and equatorial diameter (Lt+, longitudo transversa: Iversen and Troels-Smith 1950 [35]; E: Erdtman, Nilsson and Praglowski 1992 [34]), as shown in Fig 2. Polar diameter, P, can be measured only in equatorial views of the pollen grain. Equatorial diameter, E, can be measured in suitable equatorial and oblique views, as well as in polar views. As it is constrained to be the maximum dimension along a line at right angles to a line drawn through the grain centre and the midpoint of one of the apertures, it may or may not be equivalent to the maximum Feret diameter measured by Classifynder, depending on the equatorial contour shape (see below). In the oblate pollen grains of Leptospermum scoparium s.l. and Kunzea spp., which tend to lie in polar and oblique views, it is the dimension most readily measured.

Fig 2 — Acetolysed *Leptospermum scoparium* pollen in polar view (left) and equatorial view (right) showing equatorial (E) and polar (P) dimensions.

To avoid unconscious bias in the selection of pollen grains for measurement, every grain encountered in suitable orientation in one or two traverses across the middle part of the area under the microscope slide coverslip was measured. This was continued until totals of 30 equatorial diameters and 10 polar diameters were achieved. Folded or damaged pollen grains were ignored. Specimens of pale and thin-walled pollen grains were ignored as suspected immature pollen grains.

Scanning electron microscopy

Specimens from each of the 10 species of Kunzea spp. and 15 morphotypes of Leptospermum scoparium s.l. were selected, and all samples were prepared as described above, including dehydration, filtration and acetolysis treatment. One drop of pollen suspension from each sample was deposited on a conductive carbon tab attached to an aluminium stub and dried in a fume hood overnight. Specimens were then coated with gold, using a vacuum sputter coater, and examined using a JEOL Neoscope JCM-6000e microscope at 15 kV.

Digital image analysis: Automated capture

The 135 pollen samples were further analysed using a Classifynder version 3a. The Classifynder is an automated palynology system [36]. It includes an automated stage and image capture system, where pollen specimens are digitally photographed under monochromatic dark field illumination using a 40× dry objective lens. Fifty parameters are calculated from each pollen image. These include 7 core parameters that are shape- and size-related: elongation, compactness, convex hull, Heywood circularity, hydraulic radius, maximum Feret diameter, and area [29]. The remaining 43 parameters are mainly texture-related but include additional measurements of shape [36, 37]. The parameters measured were selected empirically during development of the system as those best able to allow discrimination between a range of modern pollen types [37]. Here, we used the Classifynder only for image capture and measurement, but used other tools for classification, described below.

A single microscope slide was examined for each sample. The stage configuration of the Classifynder version 3a restricts analysis to a 10×10 mm square within the 20×20 mm glass cover slips used in this study. For each sample, a 10×10 mm area was scanned by the Classifynder, and digital images of all pollen-like objects were automatically collected. These false-colour images were manually sorted to extract two groups of pollen: specimens orientated in polar view, and those orientated in equatorial view. All images properly presented in these two orientations were included, including the immature pollen grains which were ignored during light microscopic manual measurements. Results reported here are based on these two subsets of images. A median number of 158 pollen grains of either equatorial or polar view was captured from 122 samples (minimum = 16, maximum = 1290 grains). Results from the other 13 samples were discarded due to either scarcity or poor presentation of pollen grains. Although 10 Kunzea hybrids were included in the data processing stage, the results were only used as a reference guide but not incorporated in the data analysis.

Initial inspection of parameter plots revealed a sub-population of specimens with optical parameter values far from the main population, resulting in a bimodal distribution on parameter plots. Re-examination of digital images associated with these outlying specimens revealed various imperfections, mainly poor focus. These specimens were excluded from further analysis, discarding 1190 specimens with compactness parameter of <0.55. This resulted in a total dataset of 17835 Leptospermum and 5867 Kunzea pollen grains, of which approximately two-thirds were images of grains in polar view. The number of observations and summary statistics are reported in S3 and S4 Tables.

Parameter data from the pollen data set was processed using packages from the open-source software R (www.r-project.org). Exploratory data analyses to investigate the extent to which the full 50 Classifynder parameters or the 7 core parameters allowed discrimination between the various Leptospermum scoparium morphotypes and Kunzea species included discriminant function analysis (R package MASS) [38], and principal component analysis (PCA) (R package Vegan) [39]. The distribution of eigenvalues in the PCA results was illustrated using a highest posterior density region, calculated using R package emdbook [40].

Digital image analysis: Machine learning

We tested the skill of a Support Vector Machine predictive model at discriminating between Leptospermum and Kunzea in our 23702 pollen grain dataset [41]. This model, which is optimised to discriminate between two classes (here two pollen types), was tested by 99 iterations of a randomised 80:20 split of the data into training and test sets. We first tested the model on the entire pollen grain dataset, using first all 50 available parameters, then the core 7 parameters. Then, we tested the model on the subset of specimens in polar view, again using all 50 available parameters, then the core 7 parameters.

Results

The detailed pollen morphology descriptions that follow are revised from the original descriptions of McIntyre [23] and Moar [25] and are based on LM and SEM observations.

Leptospermum scoparium s.l. visual observation

Pollen grains are isopolar, oblate, triangular in polar view and flattened oval in equatorial view. In polar view, the side of the amb appears to be often concave and the angles to be extended. Pollen is tricolporate and angulaperturate, very rarely di- or tetracolporate. The ectoapertures are usually narrow, and the endoapertures are narrow but lalongate. Pollen grains are normally syncolpate, but very rarely slightly parasyncolpate due to slight widening of the colpi at the poles (however, polar islands are not present). The exine is always slightly patterned: obscurely flecked under LM, but clearly scabrate under SEM (Figs 3–5). The size of the P axis is 12.98 ± 1.66 μm (n = 188) and E axis 19.08 ± 1.28 μm (n = 490).

Fig 3 — Scale bar = 10 μm. (1a-1e) Surville Cliffs, (2a-2e) Flat Silver, (3a-3e) East Cape, (4a-4e) Waikato Peat Bog, (5a-5e) Coromandel Swamp.

Fig 5 — Scale bar = 10 μm. (1a-1e) Otaipango, (2a-2e) Northern South Island, (3a-3e) Wellington, (4a-4e) Papa, (5a-5e) Southern South Island.

Fig 4 — Scale bar = 10 μm. (1a-1e) South Island Mountain, (2a-2e) Central Volcanic Plateau, (3a-3e) L. *scoparium* var. *incanum*, (4a-4e) Three Kings, (5a-5e, 6a-6e) Auckland.

Pollen of all 15 Leptospermum scoparium s.l. morphotypes were examined under LM and, except for “Otaipango”, under SEM. Acknowledging there is variability within each morphotype, and even within pollen populations collected from individual plants, we make the following observations about pollen morphological characters that assist with discrimination between the morphotypes (Table 1):

Table 1. Pollen morphological characters of Leptospermum scoparium s.l. morphotypes (visual observation).

Morphotype	Size of Pollen Grains				Characters of Pollen Grains
	E (μm)		P (μm)		Sides of Amb		Pattern	Vestibulum	Enlarged Apices	Thickened Arci
	Average	Range	Average	Range	Concave	Straight
Otaipango	19.2	17.0–20.8	14.4	12.1–16		⁺	⁺⁺	⁺⁺	^-	^-
Papa	19	15.9–20.9	11.6	10.6–13	⁺		⁺⁺,⁺	⁺⁺	⁺	⁺
Wellington	17.9	15.7–20.4	11.5	10.3–12.6	⁺		⁺⁺	⁺,^-	^-	^-
Northern South Island	19.1	18.0–20.9	14.2	13.1–15.3	⁺		⁺⁺	^-	⁺	⁺
Southern South Island	19.1	18.4–21.6	14.1	12.9–16.4	⁺		⁺	^-	⁺	^-
Three Kings	20	18.2–21.3	10.3	9.3–12.1		⁺	⁺⁺,⁺	⁺	⁺	^-
East Cape	18.1	16–20.7	11.6	10–13.1	⁺⁺, ⁺		⁺	^-	⁺	⁺
Flat Silver	19	16.9–20.3	13.4	12.7–14.1	⁺⁺, ⁺		⁺⁺,⁺	⁺,^-	⁺,^-	⁺
Surville Cliffs	20.6	18.2–22.1	13.6	13.1–15.1	⁺⁺		⁺	⁺	⁺	^-
Central Volcanic Plateau	18.5	16.6–20.1	13	11.1–14.3	⁺		⁺	^-	⁺	⁺
L. scoparium var. incanum	19.4	17.3–21.9	16	13.5–17.5	⁺		⁺⁺,⁺	⁺	⁺	^-
South Island Mountain	19.8	17.9–22.2	14	12.6–15.8	⁺	⁺	⁺	⁺,^-	⁺	⁺
Waikato Peat Bog	18.6	16.5–20.1	13.4	12.7–14.3	⁺	⁺	⁺⁺,^-	^-	⁺,^-	^-
Auckland	18.8	16–22.4	11.7	10.7–13.1	⁺	⁺	⁺	^-	⁺	^-
Coromandel Swamp	18.3	17.1–20.3	14.3	12.8–16.1	⁺	⁺	⁺	^-	⁺	^-

Open in a new tab

++ denotes feature obviously present

+ denotes feature present

- denotes feature not obviously seen (multiple scores indicate a range of feature states)

Distinctly concave-sided pollen is more prominent in “Surville Cliffs”, “Flat Silver” and “East Cape”, whereas the amb of “Three Kings” and “Otaipango” pollen tends to be relatively straight.
Surface pattern with coarse texture, presented in patches, is especially noticeable in “Otaipango”, “Northern South Island” and “Wellington”, and common in “Papa”, “Three Kings”, L. scoparium var. incanum and “Flat Silver”.
Vestibulum–The ectexine slightly protrudes and is buckled along the sides of the amb and forms a gap with the endexine at the apices of the amb. This feature is usually present in “Otaipango” and “Papa”, and frequently seen in “Three Kings”, “Surville Cliffs” and L. scoparium var. incanum.
Rounded and slightly enlarged apices–this feature is generally distinct in many of the morphotypes except for “Otaipango” and “Wellington”.
Slightly thickened arci, only resolved under SEM, are commonly found in “East Cape”, “Flat Silver”, “Central Volcanic Plateau”, “Papa”, “South Island Mountain” and “Northern South Island” populations.
Both “Waikato Peat Bog” and “Coromandel Swamp” have large variations especially in the shape of their pollen grains, from concave to straighten sides of amb. Tetracolporate pollen grains are frequently observed in these two morphotypes.

Among the specimens examined, “Surville Cliffs” tends to have the largest pollen with average E-axis 20.6 μm, while “Wellington” appears to have the smallest with E-axis 17. 9 μm. A wider range of pollen size is observed within “Auckland” type due to the larger number of specimens examined.

Kunzea visual observations

Pollen grains are isopolar, oblate, triangular in polar view, and flattened oval in equatorial view. In polar view, the side of the amb appears to vary from slightly concave or straight, to slightly convex. Pollen is tricolporate and angulaperturate, very rarely di- or tetracolporate. Ectoapertures are usually narrow, and the endoapertures are narrow but lalongate.

Pollen grains are basically syncolpate, but more often appear to be slightly parasyncolpate, with small apocolpia at the poles. The exine appears uniformly psilate under LM and very obscurely patterned under SEM (Figs 6 and 7). The size of P axis is 11.28 ± 2.33 μm (n = 103), and E axis 16.30 ± 0.95 μm (n = 318).

Fig 6 — Scale bar = 10 μm. (1a-1e) K. *toelkenii*, (2a-2e) K. *sinclairii*, (3a-3e) K. *serotina*, (4a-4e) K. *amathicola*, (5a-5e) K. *ericoides*.

Fig 7 — Scale bar = 10 μm. (1a-1e) K. *linearis*, (2a-2e) K. *tenuicaulis*, (3a-3e) K. *salterae*, (4a-4e) K. *robusta*, (5a-5e) K. *triregensis*.

Of the ten species, K. toelkenii is the only species that could be separated from the others–by size–with mean E-axis of 17.8 μm compared with <17 μm for other Kunzea spp., along with slightly enlarged apices (Table 2, Fig 6). Of the remaining species, we could see no apparent difference in the size or surface texture. We do note some range of amb appearance in polar view, described in Table 2, but we observe considerable variability in all populations as well, such that we regard these observations a tendency, rather than a diagnostic feature. Given we do not observe consistent repeatable differences between the other species of Kunzea, we have not tabulated observations from the small number of hybrid combinations sampled.

Table 2. Pollen morphological characters of Kunzea species (visual observation).

Kunzea species	Size of Pollen Grains				Characters of Pollen Grains
	E (μm)		P (μm)		Sides of Amb	Pattern	Vestibulum	Enlarged Apices	Thickened Arci
	Average	Range	Average	Range	Sides of Amb	Pattern	Vestibulum	Enlarged Apices	Thickened Arci
K. toelkenii	17.8	15.2–20	10	9.4–10.6	slightly concave	^-	^-	⁺	^-
K. linearis	16.9	16.3–17.5	10.5	10–10.6	slightly concave	^-	^-	^-	^-
K. amathicola	16.8	15.6–18.1	9.6	8.8–11.3	slightly concave	^-	^-	^-	^-
K. triregensis	16.7	15.2–17.6	9.1	8.8–10	slightly convex	^-	^-	^-	^-
K. tenuicaulis	16.1	15.0–16.9	9.6	8.8–10.6	straight	^-	^-	^-	^-
K. robusta	16	15.0–17.5	9.9	9.4–10.6	slightly concave	^-	^-	^-	^-
K. ericoides	16.6	14.4–18.1	10.2	9.4–11.3	slightly concave	^-	^-	^-	^-
K. sinclairii	16.2	15.2–17.6	9.7	8.8–10	more concave	^-	^-	^-	^-
K. salterae	15.4	14.4–16.8	9.8	9.4–10.6	straight	^-	^-	^-	^-
K. serotina	15	13.8–16.3	9.4	8.8–10.6	slightly concave	^-	^-	^-	^-

Open in a new tab

+ denotes feature present

- denotes feature not obviously seen

Our size measurements of Kunzea pollen tend to be larger than those previously reported by de Lange [11]. This is likely due to the acetolysis treatment applied in this study, which tends to enlarge pollen grains [42, 43].

Classifynder measurements of Leptospermum scoparium s.l and Kunzea spp.

Inspection of distribution histograms (raw scan data is included in S5 Table) revealed maximum Feret diameter and area are the two parameters (of the 50 parameters described above) that most clearly differentiate Leptospermum scoparium s.l. and Kunzea pollen. This is consistent with the observations of Holt and Bebbington [29], who conducted a similar study on four samples of L. scoparium s.l. and Kunzea. An overlap between the two genera was observed for both parameters (Fig 8).

Fig 8 — *Leptospermum scoparium* (red, n = 11932 specimens) and *Kunzea* (grey, n = 3678). Polygons show the 95% range of the highest posterior density region for each species, calculated using R package coda [44].

Using the Classifynder, the 2.s.d. equatorial diameter of L. scoparium s.l. pollen is 17.93–22.18 μm (n = 11932), while the 2.s.d. equatorial diameter of Kunzea pollen is 14.32–19.00 μm (n = 3678), calculated from the maximum Feret diameter measurements (MFD) of polar view images. While the relative difference between L. scoparium s.l. and Kunzea was similar for both the visual measurements and the Classifynder, maximum Feret diameter from the Classifynder measurements is consistently 0.5–1 μm larger than light-microscope measurements (S1 and S2 Figs). One reason for this discrepancy is that for highly convex grains, the MFD measurement may not necessary be the equatorial diameter as defined in Fig 2, but a capture of a different axis, which is also observed by Holt and Bebbington [29]. This is supported by the observation that the smallest offsets between MFD and visual equatorial measurements are found in the Leptospermum scoparium morphotypes Otaipango, Flat Silver and Surville Cliffs, which have ambs either straight or slightly concave-sided, while relatively larger deviations are observed within convex-sided Kunzea species. We chose to rely on visual measurements of equatorial diameter for the taxonomic descriptions.

Ordination of Classifynder measurements

Although Linear Discriminant Analysis (LDA) seeks to maximize class separation (i.e., between-class variance), LDA of our pollen data produced a very similar result to the ‘unsupervised’ ordination of Principal Components Analysis. Both these analyses show distinct groups of L. scoparium s.l. and Kunzea pollen in ordination space, but with overlap: we report on and illustrate the PCA results here.

For a PCA constrained by the 7 ‘core’ parameters, 85% of the variance is described by the first two axes (Fig 9), while only 66% variance is described by the first two axes for a PCA constrained by all 50 parameters. Although there is overlap between L. scoparium s.l. and Kunzea pollen in the ordination, we see a clear differentiation between the two populations. In the ordination of the 7 ‘core’ parameters, the most important for discriminating between L. scoparium s.l. and Kunzea are maximum Feret diameter, compactness, convex hull, and Heywood Circularity.

We explored the overlap of the morphotypes of L. scoparium, and the species of Kunzea, in subsequent PCA analyses. In both cases, we do not see strong differentiation between L. scoparium s.l. morphotypes (Fig 10), or Kunzea species (Fig 11).

Fig 10 — Polygons show the 95% range highest posterior density region for each sub species. Underlying grey polygon is 95% range of *Kunzea* from Fig 9.

Fig 11 — Polygons show the 95% range highest posterior density region for each sub-species.

Differentiation of Leptospermum scoparium s.l. and Kunzea spp. by Support Vector Machine

The Support Vector Machine trial shows acceptable skill at discrimination between L. scoparium s.l. and Kunzea pollen in our dataset (Table 3). The best results are obtained where all 50 parameters are used to discriminate between specimens in polar view, with a mean prediction accuracy of 97.2%. Where only the seven core parameters are used, and the model is required to discriminate between specimens in both polar view and equatorial view, prediction accuracy drops to 94.7%.

Table 3. Cross-validation results of Support Vector Machine to predict pollen type, based on 99 iterations of random 80:20 splits of data into training:test sets.

Model parameters	precision rate ± s.d.
50 parameters, polar view and equatorial view	96.1% ± 0.3
50 parameters, polar view only	97.2% ± 0.3
7 parameters, polar view and equatorial view	94.7% ± 0.3
7 parameters, polar view only	95.9% ± 0.3

Open in a new tab

Discussion

Generic discrimination

The results of this study confirm observations from previous work based on smaller sample sizes. The large sample size and comprehensive sampling approach in the present study allows differences between the pollen of the New Zealand representatives of Leptospermum and Kunzea to be identified with more confidence. Firstly, there is a significant difference in average size between pollen of the two genera. Our results suggest that it should be possible to estimate the relative proportions of the two genera in a mixed population from a statistical study of equatorial pollen dimensions. Leptospermum scoparium s.l. pollen (18.88–21.50 μm, n = 11932) could be discriminated from Kunzea spp. pollen (15.49–17.83 μm, n = 3768) using the one standard deviation ranges of equatorial diameter. Only one species of Kunzea, K.toelkenii, has a pollen size comparable to that of the smallest pollen of Leptospermum scoparium morphotypes, e.g “Wellington” and part from “Auckland”. However, specimens of this large Kunzea pollen should be easily separated from small Leptospermum scoparium s.l. specimens on their less concave amb, less patterned surface, and presence of apocolpia.

In addition to the size, our observations indicate shape is another characteristic that allows discrimination of the two pollen types. Nearly all Leptospermum scoparium s.l. pollen is clearly concave-sided in polar view and more angular, while Kunzea pollen is generally nearly straight-sided. Pollen of K. sinclairii is slightly concave-sided, but with angles much less extended compared to Leptospermum scoparium s.l.. McIntyre [23] concluded that 55% of Kunzea pollen (as L. ericoides) have convex sides, compared with 4% of Leptospermum scoparium s.l. grains. It is hard to assess the range of the coverage of McIntyre’s specimens, taking the new taxonomic revision of Kunzea into account [11]. Another observation by McIntyre [23], that the colpi are more frequently parasyncolpate in Kunzea than in Leptospermum scoparium s.l., is confirmed by the present study.

The surface sculpture of Australian Leptospermum and Kunzea pollen was discussed by Thornhill et al. [45]. In contrast to Thornhill et al. [45], who noted both scabrate and psilate exine patterning in Australian Leptospermum and Kunzea pollen, our study of New Zealand grains suggests exine pattern can be used to distinguish between the two. For Leptospermum scoparium s.l., the exine tends to be more coarsely sculptured (i.e., scabrate), while for Kunzea pollen, the exine is less patterned and appears to be more psilate. While exine of L. scoparium s.l. exhibits different levels of coarseness, it is a feature that is consistently present through all specimens examined, and is clearly resolved under LM and prominent under SEM. In contrast, the exine of Kunzea pollen is consistently nearly psilate.

Variability within New Zealand Leptospermum and Kunzea

Although we see a clear distinction between pollen of Leptospermum scoparium s.l. and Kunzea spp., differences in pollen between the morphotypes of Leptospermum scoparium s.l. and between species of Kunzea are more subdued. For Kunzea, the average pollen size, combined with concaveness of amb and feature of apices, make it possible to separate K. toelkenii from the others under LM (Figs 6 and 12). Despite these noticeable characteristics, taking intraspecific variation into account, pollen of species within Kunzea spp. are practically indistinguishable from one another. For Leptospermum scoparium s.l., our initial results suggest that some morphological characteristics observed under LM, including size, texture, concaveness of amb, and presence of vestibulum, will be useful to characterise groups of morphotypes. However, Classifynder measurements made in this study do not support differentiation within species of Kunzea or morphotypes of Leptospermum scoparium. This could be partly due to limitations of the apparatus (as discussed below), partly to intraspecific pollen variation within each entity.

Fig 12 — *Leptospermum scoparium*–upper, red; *Kunzea*–lower, grey.

Intraspecific pollen variation is found in many of the Myrtaceae, but separation of ‘true’ variability from confounding issues of experiment design, such as variability in sample processing techniques or environmental morphotypes, remains a major challenge and a matter of debate [46]. Some types of morphological variations, such as the presence of four apertures (especially frequently presented in hybrids), are likely due to natural factors (e.g., genetic variation), as noted in some Australian Myrtaceae species [46]. Although a few aberrations, e.g. where a single flower produces two morphologically different pollen grains, are consistently found in some species of Eucalyptus and occasionally in other Myrtaceae species [46], this is rare or not apparent in morphotypes of Leptospermum scoparium s.l. we have examined. However, other types of morphological variation, for example variability in size, or shape, may reflect a combination of true variability and experimental treatment.

Pollen size can be affected by chemical treatment and mounting medium, maturity of grains, and pollen preservation [43, 47–51]. Although pollen grains shrink under dry and expand under wet conditions, we do not expect freshness of herbarium specimens to be a significant contributor to the intra-specific size variation observed, because shrinkage caused by dehydration is reversible [43]. In addition, because we have applied consistent preparation procedures, chemical treatment and mounting medium are not likely to be the cause of within-group variation in grain size. We do consider it likely that intraspecific size variability could arise from different levels of pollen maturity, along with other factors such as variable sporopollenin content, as discussed by Adeleye et al. [46]. In their study, a small proportion of larger, thinner-walled and convex-sided pollen grains were observed along with “normal” concave-sided grains from the same individual specimen of Leptospermum. This diversity is also observed in some of the samples we analysed and may be partly due to variation in exine development at different levels of maturity of pollen sampled from multiple anthers of the same specimen.

Utility of the Classifynder

The relative ease of collection of a large dataset, and the skill of the Support Vector Machine at discrimination between Leptospermum scoparium s.l. and Kunzea spp. pollen (~95%), shows promise for routine application of the Classifynder or similar device. In some respects we find this level of skill surprising, because some of the characteristics important for a human observer using a 100x oil immersion objective to discriminate between the two types are not available to the lower-resolution Classifynder. In particular, we expect the surface sculptural elements of pollen grains from Leptospermum scoparium s.l. and Kunzea spp., normally less than 1 μm, are of sizes at or below the limit of resolution of the digital imaging used in the Classifynder, which is approximately 1 μm [52]. More generally, we speculate that there may be potential to further improve the skill of a Classifynder-like machine for narrow two-class pollen discrimination problems such as this one. This is because the 50 features captured by the Classifynder were selected during machine design to allow discrimination of a wide range of pollen morphology, with a particular focus on 5 disparate end-member forms [37]. It is possible that a machine optimised for discrimination between a smaller range of pollen forms would be built to capture a different array of features.

Over the past decade, the rapid increase in capability in computational intelligence has resulted in considerable achievements in automatic pollen identification. Deep Learning Convolutional Neural Networks (DLCNN) is one of the most promising applications in the field of image recognition for pollen identification. Compared to a traditional neural network, DLCNN contains many more layers of neurons and is thus appropriate to very large data sets. Based on its capability to process and filter important signals from each layer of complex artificial neurons, DLCNN could extract the features that the network determines to be the most discriminatory aspects of the classes provided in the training sets. This process of feature identification is automatic and free from manual intervention, which obviates the procedure of manually selecting polar and equatorial orientation of pollen grains as practised in this study, while increasing reliability. As a result, over 97% accuracy on a set of 23 different pollen types from Brazil has been achieved by applying Deep Learning Convolutional Neural Networks (DLCNN) to a pollen image dataset composed of 805 microscope images [53]. A further exploration of DLCNN using a larger dataset of over 19,000 pollen images generated using the Classifynder produced even more promising results [54], with a classification success rate up to 98% across 46 different pollen types including New Zealand Leptospermum scoparium and Kunzea ericoides. Further improvements are likely, through increased sample sizes, refined measurement criteria, and improved discrimination networks. The deployment of the Classifynder for image capture and application of DLCNN for pollen classification could be a potential solution.

Conclusion and next steps

We have demonstrated that consistent differentiation between Leptospermum scoparium s.l. and Kunzea spp. pollen collected from New Zealand herbarium specimens is possible, using both visual and Classifynder light microscope measurements. The key characteristics that differentiate L. scoparium and Kunzea pollen are size, shape of amb, and surface texture. A Support Vector Machine trial based on the Classifynder measurements showed acceptable skill at discrimination between L. scoparium and Kunzea pollen. The best results were obtained when all 50 Classifynder parameters were used to discriminate between specimens in polar view, with a mean prediction accuracy of 97.2%.

While there is some indication that differentiation between pollen from some morphotypes of L. scoparium may be possible, further work on a larger sample set (preferably in conjunction with taxonomic studies) would be required to confirm this. An exploration of DLCNN on the larger dataset might have potential to improve the understanding of this issue.

These results could provide a first step to applying melissopalynology techniques to determine the relative contributions of L. scoparium s.l. and Kunzea nectar in manuka honey. Further analysis of honey samples would be required to determine if there are significant correlations between bioactivities and certain types of L. scoparium pollen. Application to paleoecological studies and the geological history of Myrtaceae in New Zealand are other fields yet to be explored.

Supporting information

S1 File. Comparison of pollen dimensions of male and bisexual flowers of Leptospermum scoparium s.l. and Kunzea robusta.

(DOCX)

Click here for additional data file.^{(19.5KB, docx)}

S1 Fig

Comparison for Leptospermum scoparium (red) and Kunzea (grey) of equatorial diameter measured by palynologist using a light microscope, and maximum Feret diameter measured by Classifynder.

(TIF)

Click here for additional data file.^{(1.2MB, tif)}

S2 Fig

Comparison for Leptospermum scoparium (upper) and Kunzea (lower) of equatorial diameter measured by palynologist using a light microscope (denoted by suffix “_h”), and maximum Feret diameter measured by Classifynder (denoted by suffix “_d”).

(TIF)

Click here for additional data file.^{(2.5MB, tif)}

S1 Table. Herbarium specimens studied–Leptospermum.

(XLSX)

Click here for additional data file.^{(16.3KB, xlsx)}

S2 Table. Herbarium specimens studied–Kunzea.

(XLSX)

Click here for additional data file.^{(11.2KB, xlsx)}

S3 Table. Pollen measurements of Leptospermum scoparium s.l.

(XLSX)

Click here for additional data file.^{(12.6KB, xlsx)}

S4 Table. Pollen measurements of Kunzea.

(XLSX)

Click here for additional data file.^{(12.3KB, xlsx)}

S5 Table. Pollen features extracted by Classifynder including 7 core parameters and 43 additional parameters.

(XLSX)

Click here for additional data file.^{(17.7MB, xlsx)}

Acknowledgments

We thank Ewen Cameron of Auckland Museum Herbarium (AK) for permission to destructively sample Kunzea and Leptospermum specimens under his care, and Dallas Mildenhall and Giuseppe Cortese for reviewing the `paper before submission. Andrew Boyes assisted with drafting Fig 1. We also thank two external reviewers for their constructive comments.

Data Availability

All relevant data are within the article and its Supporting Information files.

Funding Statement

An initial study of pollen of male and hermaphrodite flowers by JIR was funded by New Zealand Ministry for Primary Industry. XL received funding [no serial number] from KiwiNet (https://www.mbie.govt.nz/science-and-technology/science-and-innovation/funding-information-and-opportunities/investment-funds/preseed-accelerator-fund/kiwi-innovation-network-limited/) and GNS Science SSIF. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.McDonald CM, Keeling SE, Brewer MJ, Hathaway SC. Using chemical and DNA marker analysis to authenticate a high-value food, manuka honey. NPJ Science of Food. 2018;2(1):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Ministry for Primary Industries [New Zealand]. 2020. Apiculture monitoring report data. Available from: https://www.mpi.govt.nz/dmsdocument/45082-2020-Apiculture-monitoring-report-data
3.Taunton E. Honey exports hit $425m export sweet spot. 2021. Feb 22. Available from: https://www.stuff.co.nz/business/farming/124318150/honey-exports-hit-425m-export-sweet-spot [Google Scholar]
4.Carter DA, Blair SE, Cokcetin NN, Bouzo D, Brooks P, Schothauer R, et al. Therapeutic manuka honey: no longer so alternative. Frontiers in Microbiology. 2016; 7. doi: 10.3389/fmicb.2016.00569 WOS:000374374100002. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Quirke S, Ball R. Counterfeiting in the primary industry sector and the threat to New Zealand’s economy. National Security Journal. 201;1(1):61–74. doi: 10.36878/nsj201901.61 [DOI] [Google Scholar]
6.Codex Alimentarius Commission. Revised codex standard for honey codex Stan 12–1981, Rev. 1 (1987), Rev. 2 (2001). Codex Standard. 1981;12:1–7. [Google Scholar]
7.Sawyer R. Honey Identification. Cardiff, U.K.: Cardiff Academic Press; 1988. [Google Scholar]
8.D’Albore GR. Textbook of melissopalynology: Apimondia Publishing House; 1997. [Google Scholar]
9.Oddo LP, Piro R, Bruneau É, Guyot-Declerck C, Ivanov T, Piskulová J, et al. Main European unifloral honeys: descriptive sheets. Apidologie. 2004;35 (Suppl. 1):S38–S81. [Google Scholar]
10.de Lange PJ, Smissen RD, Wagstaff SJ, Keeling DJ, Murray BG, Toelken HR. A molecular phylogeny and infrageneric classification for Kunzea (Myrtaceae) inferred from rDNA ITS and ETS sequences. Australian Systematic Botany. 2010;23(5):309–19. [Google Scholar]
11.de Lange PJ. A revision of the New Zealand Kunzea ericoides (Myrtaceae) complex. PhytoKeys. 2014;(40):1–185. doi: 10.3897/phytokeys.40.7973 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Allan HH. Flora of New Zealand. Volume I. Christchurch: Manaaki Whenua Press; 1961. [Google Scholar]
13.Redefinitions Thompson J. and nomenclatural changes within the Leptospermum suballiance of Myrtaceae. Telopea. 1983;2(4):379–83. [Google Scholar]
14.Perry NB, Brennan NJ, Van Klink JW, Harris W, Douglas MH, McGimpsey JA, et al. Essential oils from mānuka and kānuka: Chemotaxonomy of Leptospermum. Phytochemistry. 1997;44:1485–1495. [Google Scholar]
15.Porter NG, Smale PE, Nelson MA, Hay AJ, Van Klink JW, Dean CM. Variability in essential oil and plant morphology within a Leptospermum scoparium population. New Zealand Journal of Botany. 1998;36:125–133. [Google Scholar]
16.de Lange P, Schmid LMH. Leptospermum repo (Myrtaceae), a new species from northern Aotearoa/New Zealand peat bog habitats, segregated from Leptospermum scoparium sl. Ukrainian Botanical Journal. 2021;78(4):247–265. doi: 10.15407/ukrbotj78.04.2472021 [DOI] [Google Scholar]
17.Adams CJ, Manley-Harris M, Molan PC. The origin of methylglyoxal in New Zealand mānuka (Leptospermum scoparium) honey. Carbohydrate Research. 2009;344:1050–1053. doi: 10.1016/j.carres.2009.03.020 [DOI] [PubMed] [Google Scholar]
18.Cockayne L. Notes on New Zealand floristic botany, including descriptions of new species &c. (No. 2). Transactions and Proceedings of the New Zealand Institute. 1917;49:56–65. [Google Scholar]
19.Buys MH, Winkworth RC, de Lange PJ, Wilson PG, Mitchell N, Lemmon AR, et al. The phylogenomics of diversification on an island: applying anchored hybrid enrichment to New Zealand Leptospermum scoparium (Myrtaceae). Botanical Journal of the Linnean Society. 2019;191(1):1–17. [Google Scholar]
20.Moar N. Pollen analysis of New Zealand honey. New Zealand Journal of Agricultural Research. 1985;28(1):39–70. [Google Scholar]
21.Moar N, Wilmshurst J, McGlone M. Standardizing names applied to pollen and spores in New Zealand Quaternary palynology. New Zealand Journal of Botany. 2011;49(2):201–29. [Google Scholar]
22.Pike KM. Pollen morphology of Myrtaceae from the south-west Pacific area. Australian Journal of Botany. 1956;4(1):13–53. [Google Scholar]
23.MacIntyre DJ. Pollen morphology of New Zealand species of Myrtaceae. Transactions of the Royal Society of New Zealand. 1963;2(7):83–107. [Google Scholar]
24.Harris W, Porter N, Dawson M. Observations on biosystematic relationships of Kunzea sinclairii and on an intergeneric hybrid Kunzea sinclairii × Leptospermum scoparium. New Zealand Journal of Botany. 1992;30(3):213–30. [Google Scholar]
25.Moar NT. Pollen grains of New Zealand dicotyledonous plants. Christchurch: Manaaki Whenua Press; 1993. [Google Scholar]
26.Primack RB, Lloyd DG. Andromonoecy in the New Zealand montane shrub manuka, Leptospermum scoparium (Myrtaceae). American Journal of Botany. 1980;67(3):361–8. [Google Scholar]
27.Huang SQ. Flower dimorphism and the maintenance of andromonoecy in Sagittaria guyanensis ssp. lappula (Alismataceae). New Phytologist. 2003;157(2):357–64. doi: 10.1046/j.1469-8137.2003.00676.x [DOI] [PubMed] [Google Scholar]
28.Raine JI, Li X. Pollen morphology of mānuka and kānuka. GNS Science Consultancy Report; 2014/118LR. [Google Scholar]
29.Holt KA, Bebbington MS. Separating morphologically similar pollen types using basic shape features from digital images: a preliminary study. Applications in Plant Sciences. 2014;2(8). doi: 10.3732/apps.1400032 [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Ogle CC, de Lange PJ, Cameron EK, Parris BS, Champion PD. Checklist of dicotyledons, gymnosperms and pteridophytes naturalised or casual in New Zealand: Additional records 2007–2019. Perspectives in Biosecurity. 2020;5:45–116. [Google Scholar]
31.Schönberger I, Wilton AD, Boardman KF, Breitwieser I, Cochrane M, de Lange PJ, et al. Checklist of the New Zealand Flora–Seed Plants. Lincoln: Manaaki Whenua Landcare Research; 2022. [Google Scholar]
32.Ministry for Primary Industries [New Zealand]. Criteria for identifying mānuka honey: A summary of the mānuka honey science programme, MPI Technical Paper No:2017/28. ISBN No: 978-1-77665-542-7 (online). Available from: https://www.mpi.govt.nz/dmsdocument/17314-criteria-for-identifying-manuka-honey-summary-report
33.de Lange PJ, Rolfe JR, Barkla JW, Courtney SP, Champion PD, Perrie LR, et al. Conservation status of New Zealand indigenous vascular plants 2017. New Zealand Threat Classification Series 22. Department of Conservation, Wellington. Available from: https://www.doc.govt.nz/documents/science-and-technical/nztcs22entire.pdf [Google Scholar]
34.Erdtman G, Nilsson ST, Praglowski J. Erdtman’s Handbook of Palynology. 2nd ed. Copenhagen: Munksgaard; 1992. [Google Scholar]
35.Iversen J, Troels-Smith J. Pollenmorfologiske definitioner og typer. Danmarks Geologiske Undersøgelse IV Række. 1950;3(8):1–54. [Google Scholar]
36.Holt K, Allen G, Hodgson R, Marsland S, Flenley J. Progress towards an automated trainable pollen location and classifier system for use in the palynology laboratory. Review of Palaeobotany and Palynology. 2011;167(3–4):175–83. [Google Scholar]
37.Zhang Y, Fountain DW, Hodgson RM, Flenley JR, Gunetileke S. Towards automation of palynology 3: Pollen pattern recognition using Gabor transforms and digital moments. Journal of Quaternary Science. 2004;19(8):763–8. [Google Scholar]
38.Ripley B, Venables B, Bates DM, Hornik K, Gebhardt A, Firth D, et al. Package ‘mass’. Cran r. 2013;538:113–20. [Google Scholar]
39.Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara R, et al. Vegan: community ecology package. R package version 2.2–1. 2015. [Google Scholar]
40.Bolker B. Emdbook: support functions and data for “ecological models and data”. R package version. 2016; 1(9). [Google Scholar]
41.Meyer D, Dimitriadou E, Hornik K, Weingessel A, Leisch F, Chang C, et al. Misc. functions of the Department of Statistics, Probability Theory Group (formerly: E1071). Package e1071 TU Wien. 2015. [Google Scholar]
42.Christensen BB. Measurement as a means of identifying fossil pollen. Danmarks Geologiske Undersøgelse IV Række. 1946;3(2):1–22. [Google Scholar]
43.Reitsma T. Size modification of recent pollen grains under different treatments. Review of Palaeobotany and Palynology. 1969;9(3–4):175–202. [Google Scholar]
44.Plummer M, Best N, Cowles K, Vines K. CODA: convergence diagnosis and output analysis for MCMC. R News. 2006;6(1):7–11. [Google Scholar]
45.Thornhill AH, Wilson PG, Drudge J, Barrett MD, Hope GS, Craven LA, et al. Pollen morphology of the Myrtaceae. Part 3: tribes Chamelaucieae, Leptospermeae and Lindsayomyrteae. Australian Journal of Botany. 2012;60:225–259. [Google Scholar]
46.Adeleye MA, Hopf FVL, Haberle SG. Myrtaceae pollen morphology study from Bass Strait islands, Australia, is effective in separating region-specific fossil Myrtaceae pollen types. Review of Palaeobotany and Palynology. 2020;281:104273. [Google Scholar]
47.Andersen ST. Silicone oil as a mounting medium for pollen grains. Danmarks Geologiske Undersøgelse IV Række. 1960;4(1):1–24. [Google Scholar]
48.Fægri K, Deuse P. Size variations in pollen grains with different treatments. Pollen et Spores. 1960;2(2):293. [Google Scholar]
49.Faegri K, Kaland PE, Krzywinski K. Textbook of pollen analysis. John Wiley & Sons Ltd.; 1989. [Google Scholar]
50.Mäkelä EM. Size distinctions between Betula pollen types—a review. Grana. 1996;35(4):248–56. [Google Scholar]
51.Meltsov V, Poska A, Saar M. Pollen size in Carex: the effect of different chemical treatments and mounting media. Grana. 2008;47(3):220–33. [Google Scholar]
52.Allen G. An automated pollen recognition system. M. Sc Thesis, Massey University. 2006. Available from: https://mro.massey.ac.nz/bitstream/handle/10179/613/02whole.pdf
53.Sevillano V, Aznarte JL. Improving classification of pollen grain images of the POLEN23E dataset through three different applications of deep learning convolutional neural networks. PLoS ONE. 2018;13(9). doi: 10.1371/journal.pone.0201807 [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Sevillano V, Holt K, Aznarte JL. Precise automatic classification of 46 different pollen types with convolutional neural networks. PLoS ONE. 2020;15(6). doi: 10.1371/journal.pone.0229751 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0269361.r001

Decision Letter 0

Wolfgang Blenau

8 Mar 2022

PONE-D-21-39233Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)PLOS ONE

Dear Dr. Li,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

The manuscript is rather longish and should therefore be streamlined. Reviewer #2 suggests focusing essentially on the goal, which was evidently to determine whether there are differences between Leptospermum and Kunzea pollen. Subtleties, such as the identification of different morphospecies, could be shifted to the supplement. Furthermore, it should be discussed what advantages the use of pollen analysis could have over DNA sequencing, for example when examining questionable honey samples. You can find further suggestions in the two reports by the Expert Reviewers.

Please submit your revised manuscript by Apr 22 2022 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Wolfgang Blenau

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“Sampling and morphological study was funded by Kiwinet and GNS Science SSIF, while an initial study of pollen of male and hermaphrodite flowers was funded by New Zealand Ministry for Primary Industry.”

Please note that funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“XL received funding [no serial number] from KiwiNet (https://www.mbie.govt.nz/science-and-technology/science-and-innovation/funding-information-and-opportunities/investment-funds/preseed-accelerator-fund/kiwi-innovation-network-limited/) and GNS Science SSIF. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. We note that you have referenced (P.J. de Lange unpubl.) which has currently not yet been accepted for publication. Please remove this from your References and amend this to state in the body of your manuscript: (ie “de Lange et al. [Unpublished]”) as detailed online in our guide for authors

http://journals.plos.org/plosone/s/submission-guidelines#loc-reference-style

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for the opportunity to review this manuscript. The authors’ work into morphological variation between pollen of mānuka and kānuka serves to reinforce previous work which has demonstrated that these two taxa can be separated based on pollen morphology. This has important implications for both quality control in the honey market, as well as in pollen-based vegetation reconstructions in New Zealand.

The paper is well-written and the experimental design sound. I have no hesitation in recommending that it be accepted for publication, subject to a few minor changes/suggestions which I detail below.

Line 16 (Abstract): “..as separation of pollen”. This would be more clearly put as “..as differentiation of these taxa..” or similar. “As separation of pollen” is too general/vague.

Line 40: a close-bracket appears to be missing here.

Line 41: May be useful to specify “New Zealand Government” before Ministry. Although the use of ‘Ministry’ may well imply that to a sufficient degree.

Lines 48 – 49: would be useful to have a reference for these variations in L. scoparium

Lines 50 – 52: this sentence is a bit awkwardly worded and clunky. Probably don’t need both “Nevertheless” and “despite this evidence”. Plus the inclusion of the physical description of one of the varieties in the sentence makes it hard to follow. Suggest reword to “ Nevertheless, until recently only one species, comprising two varieties, has been accepted for New Zealand. These two varieties are …..” etc.

Line 54: add comma after ‘writing’

Line 62: Suggest replace “NZ Quaternary environmental studies” with “NZ palaeovegetation studies”.

Line 69: Should be “Ministry for Primary Industries”.

Lines 86 – 87: More of a comment/suggestion here. Given there is already a standard for manuka honey (cf. lines 40 – 43), is there a need for a pollen-based standard? What value could this bring above the existing test? I think it would add to the impact of the paper if the authors could add a sentence or two which proposes why a standard including pollen morphology/melissopalynology could be beneficial for the industry. I’m aware that there are limitations to the existing MPI standard.

Line 94: I don’t really follow what is meant by “each taxonomic entity geographically apart”. Is it possible to make the meaning clearer here?

Lines 176 – 180: I’m a little bit confused as to how this could happen if the images were manually selected, i.e. I would have assumed that images selected would have been those which were not only properly orientated, but also appeared to be of good quality, i.e., not fuzzy, no debris adhering to the pollen, no hollowing out by aggressive edge detection etc. I’m assuming that some fault in the image is what is meant by “imperfections”.

Perhaps a better/briefer way to capture this aspect of the methodology is to state that only properly oriented images and good quality images (i.e. images without artifacts, etc.) were selected for use.

Alternatively, if the preference is to maintain the text as is, then it would be helpful to know more about the nature of the imperfections in the image quality and why compactness factor was used to define what was retained/discarded.

Line 308/333: a reference for the Coda package should probably be provided, i.e. Plummer et al (2006)

Lines 444 – 457: Yes, I totally agree overall. Yes, the native feature set of the Classifynder is not optimised to distinguishing pairs or sets of very similar types. The DLCNN approach applied in the papers referenced in this section (48 & 49) is probably the most promising avenue, as these NNs essentially define their own features based what they determine to be the most discriminatory aspects of the classes, as provided in the training sets. This aspect might we worth adding to the discussion. Likewise, in [49], the images of manuka and kanuka were randomly oriented, rather than polar & equatorial, and yet the system was still able to differentiate them reasonably well. I know from experience that sifting through 1000s of images for correctly aligned ones is time consuming and would really limit the deployment of the Classifynder in the routine application of a manuka pollen standard. Therefore, the ability to use randomly oriented pollen images is probably necessary here, thus making DLCNNs the best option. Again, I suggest the authors consider rewording the discussion to make reference to these issues.

Reviewer #2: Dear Xun and co-authors,

You have done a thorough job in examining pollen of New Zealand Leptospermum and Kunzea. What I found though was your manuscript lost focus on what your aim was. My understanding of your manuscript is that you want to find pollen characters that can help discriminate between the two NZ Leptospermeae species so that you can identify if the honey is Manuka honey or not. I think you have done that in this paper and it has to do with pollen measurements. But that message gets lost with the deep discussion you make on the differences within each species/morphospecies. I think the differences that you identify within the species are nothing major too. I think that most of the inter-species information could be moved to supplemental data or removed completely. Focus on the differences between the genera.

By focusing on the genera you also need to consider a couple of things.

In your introduction you say "in December 2017 the New Zealand Ministry for Primary Industries (MPI) chose a purity test for manuka honey based on assay of characteristic chemical marker compounds and DNA of manuka pollen". If this is the government policy for determining manuka honey, then why are you trying to identify pollen characters that would help identify manuka honey? Could you not say that DNA analysis of honey would unequivocally tell you if Leptospermum scoparium is in the honey? Why do palynology analyses? Could it be quicker and cheaper than a DNA test. If yes then say so.

This paper suggests that each genus probably has more than one NZ species. I'm not question that here. However, if there is more than one species in Leptospermum then does their pollen all produce manuka honey? If not then which ones. That would make identifying manuka honey using pollen even trickier.

One last consideration is that this study only used native NZ plants from herbarium specimens. You can correct me if I am wrong, but aren't there now Australian species of Leptospermum naturalised in NZ? How would a real world example to identify manuka honey work if you also have Australian Leptospermum pollen muddying the honey. How would you unequivocally say that the naturalised Leptospermum pollen wasn't in the honey?

If you can address these issues in the manuscript then I think it will be more impactful and of more use in honey pollen identification.

Below are comments made while reading the manuscript.

Regards,

Andrew Thornhill

-----------------------------------------------------------------------------

Page 2 Line 14 - I think species needs to be changed to taxa to make grammatical sense. As it reads now it is "Myrtaceous species

Leptospermum scoparium and Kunzea spp (i.e. species)"

Page 2 Line 19 - Is 'a' needed before Classifynder?

Page 2 Line 37 - 'scoparium' isn't italicised

Page 2 Line 38 - Technically this statement is true in that Kunzea is the closest relative to Leptospermum in NZ. I wouldn't say Kunzea is the sister genus of Leptospermum however because phylogenies show them to be separated by at least Asteromyrtus and Neofabricia. See the phylogenies of Maurin et al 2021 and Thornhill et al 2015.

Page 5 line 87 - Do all morphotypes of Leptospermum scoparium form Manuka honey? If they don't, what is the point of looking for different types within Leptospermum scoparium?

Page 12, line 242 - This paper's goal is to determine whether the plants that make manuka honey have different pollen to those that don't. The problem I see here is that if there are 10 species of Leptospermum in NZ - do all of these species form pollen that makes manuka honey. If they don't then which Leptospermum species/morphotypes do? Then the next question is do those Leptospermum species have unique pollen?

Page 12, Line 248 - I would be a bit cautious in using pollen size to help define different species. It looks like the pollen size differences between your different morphospecies is 2um if that. If I was to gather the same pollen and measure I have a feeling my measurements would have a greater error margin than 2um. The ability to recreate your pollen measurements means that everyone would need to use the same magnification, the same eye lens and the same microscope. Given that your size differences between plants is so small I think it is hard to justify the differences.

Page 11 Table 1. How can a feature be obviously seen & not seen at the same time? I'm confused by this table

Page 14, line 286 - This sentence highlights why you need to be cautious using pollen in taxonomy. If you don't use the exact same processing methods and microscopes then you are likely to get a different measurement. The difference in the same species between Pete's observations and the current one is around 3-4 um. This study shows that morphotypes have a difference of 1um or less. If a third study was done they would need to sample all morphotypes to come up with a range of smallest to largest. You wouldn't be able to sample just one morphotype and confidently conclude that you have X morphotype.

Page 15 line 310 - If the goal of the paper is to show the difference between Leptospermum and Kunzea pollen, then I think it appears quite simple. If a pollen is 14.32-17.93 in equatorial diametre then it is Kunzea. If it is 19-22.18 then it is Leptospermum. Anything in between can not be confidently assigned to a genus. The same would apply for measurements in polar view. It's the extreme of each measurement where there is no overlap between the two genera that are the unique pollen feature.

Page 17 line 362. Extra comma before full stop at end of sentence 1.

Page 18 line 381-382. If they can be separated by 1 St. Dev., why would you use another measure that is less accurate?

Page 19 line 396 - I think this is the incorrect citation. It should be this one -

Thornhill, A. H., Wilson, P. G., Drudge, J., Barrett, M. D., Hope, G. S., Craven, L. A., et al. (2012). Pollen morphology of the Myrtaceae. Part 3: tribes Chamelaucieae, Leptospermeae and Lindsayomyrteae. Aust. J. Bot. 60, 225–259. doi:10.1071/BT11176.

Page 19 line 412. Were they able to be separated or not? If there's enough variation within each that they can't be separated, does that not suggest they are a conglomerate?

Page 20 line 424 Is there a range of variation that encompasses this? What is the level of distinction that you feel demonstrates the shift from intraspecific to interspecific variation?

Page 20 line 436 Do your separated populations interbreed at all? Are there other Leptospermum spp. or closely related species?

Page 21 line 455. If Classyfinder has such a high accuracy, then it should have been able to detect any valid differences between the pollen morphologies in the dataset.

Page 21 line 459. Has this been demonstrated effectively? The paper seems to have lost focus on this purpose. The interest in distinguishing the Leptospermum scoparium populations is clear, but the abstract doesn't seem to accurately represent the paper's contents.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Katherine Holt

Reviewer #2: Yes: Andrew Thornhill

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2022 Jun 3;17(6):e0269361. doi: 10.1371/journal.pone.0269361.r002

Author response to Decision Letter 0

22 Apr 2022

Editor’s comments

------Our goal was to explore the range of variation in pollen morphology displayed in New Zealand Leptospermum and Kunzea. To make this useful for melissopalynological characterisation of NZ honeys, this required examination of specimens of all formally and informally established species-level segregates across their geographic range. The results could also be relevant to paleoecological studies, and perhaps to taxonomy.

The text relating to possible differences at intrageneric level, within the Results and Discussion sections of the manuscript, are therefore an essential part of the paper. We have carefully considered if these can be transferred to a supplement, and consider this would be both difficult and a detriment to readership value.

We have revised the Introduction to briefly explain the advantages of pollen analysis. The DNA analysis used in the NZ Government export requirement for honey designated as “manuka monofloral” or “manuka multifloral” is not quantitative (beyond establishing a non-trivial presence of Leptospermum pollen) – the classification is mainly based on the level of a single trace biochemical attributed to Leptospermum. The validity of this test in terms of relative manuka nectar contribution to the honey has been disputed. This is a large topic which we cannot adequately address in our paper, and which needs to be further researched. Our discrimination of NZ Leptospermum and Kunzea pollen provides one method for such research.

We have attended to the reviewer’s questions and comments as noted below. We found these very helpful. Resulting amendments to the text, and inclusion of additional references have tended to lengthen the paper, but we have sought to reduce the length and increase focus by eliminating some sub-sections where there was a degree of duplication or lesser relevance. The main cuts are in the original lines 242-251, 273-296 (Fig 8 moved to Discussion), and 353-373; also Table 2 has been reduced by omitting de Lange’s previous measurements.

The resulting text is in the end shorter than that originally submitted and we believe substantially improved. We hope you will agree.

Journal Requirements:

2. Thank you for stating the following in the Acknowledgments Section of your manuscript:

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

------ We have removed the funding acknowledgments from the manuscript and would like the Funding Statement to be amended to the following:

“An initial study of pollen of male and hermaphrodite flowers by JIR was funded by New Zealand Ministry for Primary Industry. XL received funding [no serial number] from KiwiNet (https://www.mbie.govt.nz/science-and-technology/science-and-innovation/funding-information-and-opportunities/investment-funds/preseed-accelerator-fund/kiwi-innovation-network-limited/) and GNS Science SSIF. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

------This has been corrected.

Review Comments to the Author

The paper is well-written and the experimental design sound. I have no hesitation in recommending that it be accepted for publication, subject to a few minor changes/suggestions which I detail below.

Line 16 (Abstract): “..as separation of pollen”. This would be more clearly put as “..as differentiation of these taxa..” or similar. “As separation of pollen” is too general/vague.

------Accepted and change made

Line 40: a close-bracket appears to be missing here.

------Corrected

Line 41: May be useful to specify “New Zealand Government” before Ministry. Although the use of ‘Ministry’ may well imply that to a sufficient degree.

------Accepted and change made

Lines 48 – 49: would be useful to have a reference for these variations in L. scoparium.

------References inserted

------Reworded

Line 54: add comma after ‘writing’

------comma added

Line 62: Suggest replace “NZ Quaternary environmental studies” with “NZ palaeovegetation studies”.

------Accepted and change made

Line 69: Should be “Ministry for Primary Industries”.

------Corrected

------We have modified the text to include a brief statement of the benefits of melissopalynology. The relationship of the MPI test to the actual manuka contribution to a honey is a complex and we suggest an as yet inadequately studied topic. We felt that to address this issue properly would be beyond the scope of our paper, in which we wished to report mainly basic pollen data which can be built upon in later analyses.

Line 94: I don’t really follow what is meant by “each taxonomic entity geographically apart”. Is it possible to make the meaning clearer here?

------text modified

------We have modified existing text to state that it was mainly poor focus images which caused the outlying results.

Line 308/333: a reference for the Coda package should probably be provided, i.e. Plummer et al (2006)

------Reference provided

------This text has been modified to address the DLCNN advantages mentioned by the reviewer.

Reviewer #2: Dear Xun and co-authors,

------This was our primary purpose, but as a basic pollen morphology study it may have other and perhaps unforeseen applications in future. This is partly why we have included details of the individual Leptospermum morphotypes and Kunzea species.

I think you have done that in this paper and it has to do with pollen measurements. But that message gets lost with the deep discussion you make on the differences within each species/morphospecies. I think the differences that you identify within the species are nothing major too. I think that most of the inter-species information could be moved to supplemental data or removed completely. Focus on the differences between the genera.

------In aiming to test the difference between pollen of mānuka and kānuka, we had to cover the whole range of species and morphotypes, and other morphological features as well as size. If we did not check all of them, how could we conclude that the pollen characteristics of the two genera are persistent within each genus but different between them? This is the body of our research, which we have sought to report accurately. We have considered moving text to supplemental data, but think that this would not fairly represent the work, or be as useful to readers. However, we have attempted to reduce the length of text by removing some duplication or less relevant matter.

By focusing on the genera you also need to consider a couple of things.

------This study was commenced before MPI brought out their requirements for EXPORT manuka honey (note that there is no internal NZ standard, although there are NZ industry recommendations for combined Leptospermum-Kunzea pollen content dating back to 2008). We believe there remain advantages for pollen analysis and have now included a brief explanation in the introduction. See also our response to the editor.

------At the present time, no distinction between possibly different NZ Leptospermum species is made in identifying manuka honey. Such a proposition would probably not be supported by the NZ honey industry.

------We know of 7 casual or fully naturalised Leptospermum species (data from Schönberger et al. (2021) and Ogle et al. (2020), and personal observation of P. de Lange):

L. laevigatum - locally common in northern Bay of Plenty, Matakana Island especially

L. minutifolium - Auckland only - collected once

L. petersonii - Auckland - common garden plant scarce in the wild

L. morrisonii - Auckland / Nelson (Golden Bay) - very uncommon

L. polygalifolium subsp. polygalifolium – locally common in urban Auckland

L. spectabile - Northland - collected once not seen again

L. variabile - Auckland (Western Springs) collected once and not seen again

None of these are sufficiently widespread to be likely to contribute much to the honey industry, although it is possible that hives on Matakana Island will collect some L. laevigatum honey. We have now noted this in the text. Study of the pollen of this species is an omission which we cannot now encompass.

If you can address these issues in the manuscript then I think it will be more impactful and of more use in honey pollen identification.

Below are comments made while reading the manuscript.

Regards,

Andrew Thornhill

-----------------------------------------------------------------------------

Page 2 Line 14 - I think species needs to be changed to taxa to make grammatical sense. As it reads now it is "Myrtaceous species

Leptospermum scoparium and Kunzea spp (i.e. species)"

------Accepted and text changed.

Page 2 Line 19 - Is 'a' needed before Classifynder?

------Text corrected.

Page 2 Line 37 - 'scoparium' isn't italicised

------Text corrected.

------We have modified the text accordingly.

Page 5 line 87 - Do all morphotypes of Leptospermum scoparium form Manuka honey? If they don't, what is the point of looking for different types within Leptospermum scoparium?

------No distinction is at present made between the different possible L. scoparium morphotype sources for mānuka honey.

------As noted above, all NZ Leptospermum plants are at present assumed to be eligible sources. It is known that there may be differences in the chemistry of the resulting honeys (e.g., in the DHA content). In this paper, we demonstrate that there are very limited differences in the pollen morphology between the NZ Leptospermum populations. For NZ honey pollen analysis, this is an important finding of our paper, as also that pollen of all NZ Kunzea pollen can be distinguished from that of all NZ Leptospermum (at least on a reasonably applicable statistical basis).

------This paragraph has been removed to save space. The relationship of our pollen results (which include other morphological characters as well as size) to taxonomy will be better assessed when the taxonomy of NZ Leptospermum is further understood.

Other workers seeking to research pollen morphology of these taxa would have to use a uniform technique, but reproduction of our optics would not be necessary as it is relative sizes and shapes that are important rather than precise measurements.

Page 11 Table 1. How can a feature be obviously seen & not seen at the same time? I'm confused by this table

------Some features are variable, evident in some collections or individual specimens but not so in others. We have scored both states in our table. An explanatory remark has been added to the caption.

------We agree, consistent technique is undoubtedly important. The difference between Peter de Lange’s previous results and our new measurements is principally due to his pollen not being acetolysed. This increases variability due to varying degrees of hydration of the protoplasm, compared to acetolysed pollen where protoplasm is no longer present. We have now removed this paragraph to reduce the length of the paper.

------This may be true for a distinction based solely on pollen size, but separation of the two genera can be improved by including the differences in pollen shape and surface texture. This is evident from the DLCNN results as well as our visual observations.

Page 17 line 362. Extra comma before full stop at end of sentence 1.

------Corrected

Page 18 line 381-382. If they can be separated by 1 St. Dev., why would you use another measure that is less accurate?

------Text amended to remove the 2 std deviation statement.

Page 19 line 396 - I think this is the incorrect citation. It should be this one -

------We have changed the citation as indicated.

Page 19 line 412. Were they able to be separated or not? If there's enough variation within each that they can't be separated, does that not suggest they are a conglomerate?

------We do not understand this remark. Many genera, like Leptospermum and Kunzea, display fairly uniform pollen morphology among well-delineated species. This does not mean that they are not distinct genetic entities, or that some intrageneric groupings may not still be recognised. As noted elsewhere in Discussion, Classifynder measurements may not reveal some of the distinctions possible using visual observation.

Page 20 line 424 Is there a range of variation that encompasses this? What is the level of distinction that you feel demonstrates the shift from intraspecific to interspecific variation?

------The text here is discussing the possible origins of intraspecific variation in size. It discounts freshness of herbarium material, and suggests that variation may result from varied pollen maturity. We cannot quantify this.

Page 20 line 436 Do your separated populations interbreed at all? Are there other Leptospermum spp. or closely related species?

------Kunzea species do interbreed, and their progeny is fully fertile, but hybrids are scarce outside sites of protracted human disturbance (de Lange et al. 2005, de Lange 2014). The situation with Leptospermum is at present uncertain. We have removed this paragraph, as on reconsideration we think that contaminant pollen from nearby genetically different but related populations will not be a numerically significant cause of variation in size measurements.

Page 21 line 455. If Classyfinder has such a high accuracy, then it should have been able to detect any valid differences between the pollen morphologies in the dataset.

------We have expanded discussion of DLCNN following the remarks of Reviewer 1. However, note that Classifynder may have physical or algorithmic deficiencies, e.g., optical resolution of the 40x microscope objective used by Classifynder is less than that of the 100x oil immersion objective used by us for observation of surface texture. See remarks by Reviewer 1 for lines 444 et seq. In general, we have attempted to discover variation within NZ Leptospermum and Kunzea by understandable human observation rather than rely only on a “black box” algorithm.

------We disagree with this assessment. However, we have attempted to improve the focus of the paper by deleting some less relevant text and by improving the introduction.

Attachment

Submitted filename: Response to Reviewers .docx

Click here for additional data file.^{(28.8KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0269361.r003

Decision Letter 1

Wolfgang Blenau

20 May 2022

Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)

PONE-D-21-39233R1

Dear Dr. Li,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Wolfgang Blenau

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

6. Review Comments to the Author

Reviewer #1: (No Response)

Reviewer #2: I think that you have greatly improved the paper by reducing the amount of discussion on the differences between morphospecies. I could only find one correction for you. It is

Line 418 - delete 'of''

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Katherine Holt

Reviewer #2: Yes: Andrew Thornhill

PLoS One. doi: 10.1371/journal.pone.0269361.r004

Acceptance letter

Wolfgang Blenau

25 May 2022

PONE-D-21-39233R1

Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)

Dear Dr. Li:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Wolfgang Blenau

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 File. Comparison of pollen dimensions of male and bisexual flowers of Leptospermum scoparium s.l. and Kunzea robusta.

(DOCX)

Click here for additional data file.^{(19.5KB, docx)}

S1 Fig

Comparison for Leptospermum scoparium (red) and Kunzea (grey) of equatorial diameter measured by palynologist using a light microscope, and maximum Feret diameter measured by Classifynder.

(TIF)

Click here for additional data file.^{(1.2MB, tif)}

S2 Fig

(TIF)

Click here for additional data file.^{(2.5MB, tif)}

S1 Table. Herbarium specimens studied–Leptospermum.

(XLSX)

Click here for additional data file.^{(16.3KB, xlsx)}

S2 Table. Herbarium specimens studied–Kunzea.

(XLSX)

Click here for additional data file.^{(11.2KB, xlsx)}

S3 Table. Pollen measurements of Leptospermum scoparium s.l.

(XLSX)

Click here for additional data file.^{(12.6KB, xlsx)}

S4 Table. Pollen measurements of Kunzea.

(XLSX)

Click here for additional data file.^{(12.3KB, xlsx)}

S5 Table. Pollen features extracted by Classifynder including 7 core parameters and 43 additional parameters.

(XLSX)

Click here for additional data file.^{(17.7MB, xlsx)}

Attachment

Submitted filename: Response to Reviewers .docx

Click here for additional data file.^{(28.8KB, docx)}

Data Availability Statement

All relevant data are within the article and its Supporting Information files.

[pone.0269361.ref001] 1.McDonald CM, Keeling SE, Brewer MJ, Hathaway SC. Using chemical and DNA marker analysis to authenticate a high-value food, manuka honey. NPJ Science of Food. 2018;2(1):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269361.ref002] 2.Ministry for Primary Industries [New Zealand]. 2020. Apiculture monitoring report data. Available from: https://www.mpi.govt.nz/dmsdocument/45082-2020-Apiculture-monitoring-report-data

[pone.0269361.ref003] 3.Taunton E. Honey exports hit $425m export sweet spot. 2021. Feb 22. Available from: https://www.stuff.co.nz/business/farming/124318150/honey-exports-hit-425m-export-sweet-spot [Google Scholar]

[pone.0269361.ref004] 4.Carter DA, Blair SE, Cokcetin NN, Bouzo D, Brooks P, Schothauer R, et al. Therapeutic manuka honey: no longer so alternative. Frontiers in Microbiology. 2016; 7. doi: 10.3389/fmicb.2016.00569 WOS:000374374100002. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269361.ref005] 5.Quirke S, Ball R. Counterfeiting in the primary industry sector and the threat to New Zealand’s economy. National Security Journal. 201;1(1):61–74. doi: 10.36878/nsj201901.61 [DOI] [Google Scholar]

[pone.0269361.ref006] 6.Codex Alimentarius Commission. Revised codex standard for honey codex Stan 12–1981, Rev. 1 (1987), Rev. 2 (2001). Codex Standard. 1981;12:1–7. [Google Scholar]

[pone.0269361.ref007] 7.Sawyer R. Honey Identification. Cardiff, U.K.: Cardiff Academic Press; 1988. [Google Scholar]

[pone.0269361.ref008] 8.D’Albore GR. Textbook of melissopalynology: Apimondia Publishing House; 1997. [Google Scholar]

[pone.0269361.ref009] 9.Oddo LP, Piro R, Bruneau É, Guyot-Declerck C, Ivanov T, Piskulová J, et al. Main European unifloral honeys: descriptive sheets. Apidologie. 2004;35 (Suppl. 1):S38–S81. [Google Scholar]

[pone.0269361.ref010] 10.de Lange PJ, Smissen RD, Wagstaff SJ, Keeling DJ, Murray BG, Toelken HR. A molecular phylogeny and infrageneric classification for Kunzea (Myrtaceae) inferred from rDNA ITS and ETS sequences. Australian Systematic Botany. 2010;23(5):309–19. [Google Scholar]

[pone.0269361.ref011] 11.de Lange PJ. A revision of the New Zealand Kunzea ericoides (Myrtaceae) complex. PhytoKeys. 2014;(40):1–185. doi: 10.3897/phytokeys.40.7973 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269361.ref012] 12.Allan HH. Flora of New Zealand. Volume I. Christchurch: Manaaki Whenua Press; 1961. [Google Scholar]

[pone.0269361.ref013] 13.Redefinitions Thompson J. and nomenclatural changes within the Leptospermum suballiance of Myrtaceae. Telopea. 1983;2(4):379–83. [Google Scholar]

[pone.0269361.ref014] 14.Perry NB, Brennan NJ, Van Klink JW, Harris W, Douglas MH, McGimpsey JA, et al. Essential oils from mānuka and kānuka: Chemotaxonomy of Leptospermum. Phytochemistry. 1997;44:1485–1495. [Google Scholar]

[pone.0269361.ref015] 15.Porter NG, Smale PE, Nelson MA, Hay AJ, Van Klink JW, Dean CM. Variability in essential oil and plant morphology within a Leptospermum scoparium population. New Zealand Journal of Botany. 1998;36:125–133. [Google Scholar]

[pone.0269361.ref016] 16.de Lange P, Schmid LMH. Leptospermum repo (Myrtaceae), a new species from northern Aotearoa/New Zealand peat bog habitats, segregated from Leptospermum scoparium sl. Ukrainian Botanical Journal. 2021;78(4):247–265. doi: 10.15407/ukrbotj78.04.2472021 [DOI] [Google Scholar]

[pone.0269361.ref017] 17.Adams CJ, Manley-Harris M, Molan PC. The origin of methylglyoxal in New Zealand mānuka (Leptospermum scoparium) honey. Carbohydrate Research. 2009;344:1050–1053. doi: 10.1016/j.carres.2009.03.020 [DOI] [PubMed] [Google Scholar]

[pone.0269361.ref018] 18.Cockayne L. Notes on New Zealand floristic botany, including descriptions of new species &c. (No. 2). Transactions and Proceedings of the New Zealand Institute. 1917;49:56–65. [Google Scholar]

[pone.0269361.ref019] 19.Buys MH, Winkworth RC, de Lange PJ, Wilson PG, Mitchell N, Lemmon AR, et al. The phylogenomics of diversification on an island: applying anchored hybrid enrichment to New Zealand Leptospermum scoparium (Myrtaceae). Botanical Journal of the Linnean Society. 2019;191(1):1–17. [Google Scholar]

[pone.0269361.ref020] 20.Moar N. Pollen analysis of New Zealand honey. New Zealand Journal of Agricultural Research. 1985;28(1):39–70. [Google Scholar]

[pone.0269361.ref021] 21.Moar N, Wilmshurst J, McGlone M. Standardizing names applied to pollen and spores in New Zealand Quaternary palynology. New Zealand Journal of Botany. 2011;49(2):201–29. [Google Scholar]

[pone.0269361.ref022] 22.Pike KM. Pollen morphology of Myrtaceae from the south-west Pacific area. Australian Journal of Botany. 1956;4(1):13–53. [Google Scholar]

[pone.0269361.ref023] 23.MacIntyre DJ. Pollen morphology of New Zealand species of Myrtaceae. Transactions of the Royal Society of New Zealand. 1963;2(7):83–107. [Google Scholar]

[pone.0269361.ref024] 24.Harris W, Porter N, Dawson M. Observations on biosystematic relationships of Kunzea sinclairii and on an intergeneric hybrid Kunzea sinclairii × Leptospermum scoparium. New Zealand Journal of Botany. 1992;30(3):213–30. [Google Scholar]

[pone.0269361.ref025] 25.Moar NT. Pollen grains of New Zealand dicotyledonous plants. Christchurch: Manaaki Whenua Press; 1993. [Google Scholar]

[pone.0269361.ref026] 26.Primack RB, Lloyd DG. Andromonoecy in the New Zealand montane shrub manuka, Leptospermum scoparium (Myrtaceae). American Journal of Botany. 1980;67(3):361–8. [Google Scholar]

[pone.0269361.ref027] 27.Huang SQ. Flower dimorphism and the maintenance of andromonoecy in Sagittaria guyanensis ssp. lappula (Alismataceae). New Phytologist. 2003;157(2):357–64. doi: 10.1046/j.1469-8137.2003.00676.x [DOI] [PubMed] [Google Scholar]

[pone.0269361.ref028] 28.Raine JI, Li X. Pollen morphology of mānuka and kānuka. GNS Science Consultancy Report; 2014/118LR. [Google Scholar]

[pone.0269361.ref029] 29.Holt KA, Bebbington MS. Separating morphologically similar pollen types using basic shape features from digital images: a preliminary study. Applications in Plant Sciences. 2014;2(8). doi: 10.3732/apps.1400032 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269361.ref030] 30.Ogle CC, de Lange PJ, Cameron EK, Parris BS, Champion PD. Checklist of dicotyledons, gymnosperms and pteridophytes naturalised or casual in New Zealand: Additional records 2007–2019. Perspectives in Biosecurity. 2020;5:45–116. [Google Scholar]

[pone.0269361.ref031] 31.Schönberger I, Wilton AD, Boardman KF, Breitwieser I, Cochrane M, de Lange PJ, et al. Checklist of the New Zealand Flora–Seed Plants. Lincoln: Manaaki Whenua Landcare Research; 2022. [Google Scholar]

[pone.0269361.ref032] 32.Ministry for Primary Industries [New Zealand]. Criteria for identifying mānuka honey: A summary of the mānuka honey science programme, MPI Technical Paper No:2017/28. ISBN No: 978-1-77665-542-7 (online). Available from: https://www.mpi.govt.nz/dmsdocument/17314-criteria-for-identifying-manuka-honey-summary-report

[pone.0269361.ref033] 33.de Lange PJ, Rolfe JR, Barkla JW, Courtney SP, Champion PD, Perrie LR, et al. Conservation status of New Zealand indigenous vascular plants 2017. New Zealand Threat Classification Series 22. Department of Conservation, Wellington. Available from: https://www.doc.govt.nz/documents/science-and-technical/nztcs22entire.pdf [Google Scholar]

[pone.0269361.ref034] 34.Erdtman G, Nilsson ST, Praglowski J. Erdtman’s Handbook of Palynology. 2nd ed. Copenhagen: Munksgaard; 1992. [Google Scholar]

[pone.0269361.ref035] 35.Iversen J, Troels-Smith J. Pollenmorfologiske definitioner og typer. Danmarks Geologiske Undersøgelse IV Række. 1950;3(8):1–54. [Google Scholar]

[pone.0269361.ref036] 36.Holt K, Allen G, Hodgson R, Marsland S, Flenley J. Progress towards an automated trainable pollen location and classifier system for use in the palynology laboratory. Review of Palaeobotany and Palynology. 2011;167(3–4):175–83. [Google Scholar]

[pone.0269361.ref037] 37.Zhang Y, Fountain DW, Hodgson RM, Flenley JR, Gunetileke S. Towards automation of palynology 3: Pollen pattern recognition using Gabor transforms and digital moments. Journal of Quaternary Science. 2004;19(8):763–8. [Google Scholar]

[pone.0269361.ref038] 38.Ripley B, Venables B, Bates DM, Hornik K, Gebhardt A, Firth D, et al. Package ‘mass’. Cran r. 2013;538:113–20. [Google Scholar]

[pone.0269361.ref039] 39.Oksanen J, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara R, et al. Vegan: community ecology package. R package version 2.2–1. 2015. [Google Scholar]

[pone.0269361.ref040] 40.Bolker B. Emdbook: support functions and data for “ecological models and data”. R package version. 2016; 1(9). [Google Scholar]

[pone.0269361.ref041] 41.Meyer D, Dimitriadou E, Hornik K, Weingessel A, Leisch F, Chang C, et al. Misc. functions of the Department of Statistics, Probability Theory Group (formerly: E1071). Package e1071 TU Wien. 2015. [Google Scholar]

[pone.0269361.ref042] 42.Christensen BB. Measurement as a means of identifying fossil pollen. Danmarks Geologiske Undersøgelse IV Række. 1946;3(2):1–22. [Google Scholar]

[pone.0269361.ref043] 43.Reitsma T. Size modification of recent pollen grains under different treatments. Review of Palaeobotany and Palynology. 1969;9(3–4):175–202. [Google Scholar]

[pone.0269361.ref044] 44.Plummer M, Best N, Cowles K, Vines K. CODA: convergence diagnosis and output analysis for MCMC. R News. 2006;6(1):7–11. [Google Scholar]

[pone.0269361.ref045] 45.Thornhill AH, Wilson PG, Drudge J, Barrett MD, Hope GS, Craven LA, et al. Pollen morphology of the Myrtaceae. Part 3: tribes Chamelaucieae, Leptospermeae and Lindsayomyrteae. Australian Journal of Botany. 2012;60:225–259. [Google Scholar]

[pone.0269361.ref046] 46.Adeleye MA, Hopf FVL, Haberle SG. Myrtaceae pollen morphology study from Bass Strait islands, Australia, is effective in separating region-specific fossil Myrtaceae pollen types. Review of Palaeobotany and Palynology. 2020;281:104273. [Google Scholar]

[pone.0269361.ref047] 47.Andersen ST. Silicone oil as a mounting medium for pollen grains. Danmarks Geologiske Undersøgelse IV Række. 1960;4(1):1–24. [Google Scholar]

[pone.0269361.ref048] 48.Fægri K, Deuse P. Size variations in pollen grains with different treatments. Pollen et Spores. 1960;2(2):293. [Google Scholar]

[pone.0269361.ref049] 49.Faegri K, Kaland PE, Krzywinski K. Textbook of pollen analysis. John Wiley & Sons Ltd.; 1989. [Google Scholar]

[pone.0269361.ref050] 50.Mäkelä EM. Size distinctions between Betula pollen types—a review. Grana. 1996;35(4):248–56. [Google Scholar]

[pone.0269361.ref051] 51.Meltsov V, Poska A, Saar M. Pollen size in Carex: the effect of different chemical treatments and mounting media. Grana. 2008;47(3):220–33. [Google Scholar]

[pone.0269361.ref052] 52.Allen G. An automated pollen recognition system. M. Sc Thesis, Massey University. 2006. Available from: https://mro.massey.ac.nz/bitstream/handle/10179/613/02whole.pdf

[pone.0269361.ref053] 53.Sevillano V, Aznarte JL. Improving classification of pollen grain images of the POLEN23E dataset through three different applications of deep learning convolutional neural networks. PLoS ONE. 2018;13(9). doi: 10.1371/journal.pone.0201807 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0269361.ref054] 54.Sevillano V, Holt K, Aznarte JL. Precise automatic classification of 46 different pollen types with convolutional neural networks. PLoS ONE. 2020;15(6). doi: 10.1371/journal.pone.0229751 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Discrimination of pollen of New Zealand mānuka (Leptospermum scoparium agg.) and kānuka (Kunzea spp.) (Myrtaceae)

X Li

J G Prebble

P J de Lange

J I Raine

L Newstrom-Lloyd

Roles

Abstract

Introduction

Materials and methods

Sample collection

Fig 1.

Leptospermum scoparium putative segregates

Pollen preparation

Light microscopy and manual measurements

Fig 2.

Scanning electron microscopy

Digital image analysis: Automated capture

Digital image analysis: Machine learning

Results

Leptospermum scoparium s.l. visual observation

Fig 3. Polar and equatorial views of Leptospermum scoparium, Part 1.

Fig 5. Polar and equatorial views of Leptospermum scoparium, Part 3.

Fig 4. Polar and equatorial views of Leptospermum scoparium, Part 2.

Table 1. Pollen morphological characters of Leptospermum scoparium s.l. morphotypes (visual observation).

Kunzea visual observations

Fig 6. Polar and equatorial views of Kunzea, Part 1.

Fig 7. Polar and equatorial views of Kunzea, Part 2.

Table 2. Pollen morphological characters of Kunzea species (visual observation).

Classifynder measurements of Leptospermum scoparium s.l and Kunzea spp.

Fig 8. Classifynder measurements of area vs maximum Feret diameter of pollen in polar view.

Ordination of Classifynder measurements

Fig 9. Principal component analysis of the seven core parameters for Leptospermum scoparium and Kunzea pollen in polar view.

Fig 10. Principal component analysis of the seven core parameters for Leptospermum scoparium morphotypes pollen in polar view.

Fig 11. Principal component analysis of the seven core parameters for Kunzea species pollen in polar view.

Differentiation of Leptospermum scoparium s.l. and Kunzea spp. by Support Vector Machine

Table 3. Cross-validation results of Support Vector Machine to predict pollen type, based on 99 iterations of random 80:20 splits of data into training:test sets.

Discussion

Generic discrimination

Variability within New Zealand Leptospermum and Kunzea

Fig 12. Size range of Leptospermum scoparium and Kunzea in equatorial diameter under light microscope.

Utility of the Classifynder

Conclusion and next steps

Supporting information

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Wolfgang Blenau

Roles

Author response to Decision Letter 0

Decision Letter 1

Wolfgang Blenau

Roles

Acceptance letter

Wolfgang Blenau

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases