Both fresh and preserved (herbarium) material were used in this project. Dried material can be sectioned after rehydrating in a 5% solution of Contrad 70® (Decon Labs, King of Prussia, PA, U.S.A.) for a period of 24 hours (Tomlinson et al. 2011), though better results are obtained with fresh material.

The living material used in this study came mainly from the collections at Montgomery Botanical Center (MBC, Miami, FL) and the Jardim Botânico Plantarum (Nova Odessa, São Paulo, Brazil). A few were collected from Fairchild Tropical Botanic Garden (Coral Gables, FL). The dried material was often from air-dried specimens made while doing fieldwork, from the herbarium at Jardim Botânico Plantarum (HPL, Novo Odessa, São Paulo, Brazil), and from the Fairchild Tropical Botanic Garden herbarium (FTG, Miami, FL). A few specimens were from the following herbaria: G, IBGE, IPA, K, MO, NY and US.

There are a number of ways to hand section leaflet margins and most of these are covered in Noblick (2013). This type of research neither requires expensive hardware or use of chemicals and dyes. To start with, the leaflet sampled should always be collected from the same place on the plant. In this study, leaflets were sampled from the middle of a central leaflet. The following equipment was used: a hand microtome, a sharp knife, a straight razor, a double sided razor blade, a small artists brush (one of the smallest ones), a dropper bottle of water, a watch glass, a stone or plate to sharpen the straight razor blade, and a carrot (Figure 1A). The hand microtome was purchased from a home schooling site (Homesciencetools.com, Billings, MT, U.S.A.) for about $45. The traditional straight razor I use can be purchased from an online shaving store, though the one I purchased from no longer exists. One of the cheapest pre-sharpened stainless steel straight razors for beginners will probably do fine, and they can be purchased for as little as $14 from Jet.com (Hoboken, NJ, U.S.A.). I use a Dovo hollow ground stainless steel straight razor, which is available online for about $80. I prefer the traditional straight razor blade over those with replacement blades. Finally, to keep a razor-sharp edge, one needs a sharpening stone. The Dia-sharp 3 micron 8000 mesh (DMT D8EE 8” Extra Extra Fine Diamond Stone) (DMTsharp.com or Diamond Machining Technology, Marlbourough, MA, U.S.A) was found to be an excellent choice. Using it frequently between sectionings will maintain the sharpness required for making clean, thin sections.

Using a small sharp knife, a piece of carrot is cut into a small cube that will fit in the hand microtome (Figure 1B, C). A perpendicular, vertical slit is cut in the top of the cube with the double-sided razor blade, but not all the way through. A piece of the palm leaf margin is placed into the vertical carrot slit and the carrot cube is clamped into the hand microtome. The carrot cube is adjusted down until it is barely showing above the microtome plate. The first cut is made across the carrot to cut off the excess leaf material and carrot. The blade is re-sharpened using the Diamond Stone and water (Figure 1D). Lubricate the surface of the carrot with a drop of water and slide the straight razor across the microtome in a horizontal slicing movement, while pressing the side of the blade firmly against the plate (Figure 1E). If at first, it does not cut anything, adjust the hand microtome up about a ¼ of a turn and try again. Always keep the specimen lubricated with water using the eye dropper. After obtaining a section, tease the carrot away from the section or drop the section with attached carrot into a watch glass with water and tease the carrot from the section using the narrow artist brush, leaving the section in the watch glass (Figure 1F). After several sections have been successfully made, examine the sections in the watch glass under a dissecting microscope.

To prepare the slide, some glass slides, glass cover slips, artist brush, a bottle of 1:1 glycerin and water solution, and a dissecting needle are needed. Glass slides with frosted glass along one side to write on are preferred. The glass slide and glass cover slip should always be cleaned with distilled water or 70 % alcohol before using. Label the frosted portion of the slide and spread a drop or two of the 1:1 glycerin and water solution on the slide. While looking through the dissecting microscope, select the thinnest sections from the watch glass with the narrow artist brush and place them into the 1:1 glycerin droplet on the slide. After placing a number of the sections on the slide (ca. 6), then cover the sections with the glass cover slip. The best way to insure that no bubbles are left behind under the cover slip is to use a dissecting needle. Place one edge of the cover slip at the edge of the glycerin droplet with the sections and begin gently lowering it into place over the sections. By placing the dissecting needle tip on its side under the other edge of the cover slip, while slowly pulling out the needle as the cover slip lowers into place, most of the air bubbles should exit from under the cover slip on the side of the exiting needle.

The glass slide is now placed under a compound light microscope and photographed under the 10× objective (100× magnification). Images were taken with a Nikon Coolpix 4500 digital camera with a Leitz periplan 10 x/18 eye piece adapter. The camera was powered by a Powerline Universal AC adapter so there was no need to rely on battery power. The camera was set to P mode, white balance was adjusted to the light source of the microscope, and the lens was set to Fisheye 2. The images were cleaned of background spots, adjusted for brightness and contrast, and sharpened if necessary using Adobe Photoshop. Two adjacent images near the tip were photo merged using Adobe Photoshop. A stage micrometer was used to apply a scale to each image and replaced by a scale bar to aid in assembling these into a plate, while maintaining the scale.

This paper is focused mainly on characters of the more easily sectioned leaflet margin and not on the more difficult to section midrib. Epidermis and dark staining idioblasts (tanniniferous cells) were also not examined. Characters assessed during this study follow some of Glassman’s and Tomlinson’s characters as listed above in the introduction. Figure 2 clarifies much of the terminology for characters used in this paper. To be clear, in each leaflet cross-section the upper or superior side of the lamina is called the adaxial surface. The lower or inferior side is called the abaxial (Dransfield et al. 2008, Esau 1977).

The outer most layer of the leaf is the cuticle (Figure 2, c), a non-cellular waxy layer produced by the epidermis (Dransfield et al. 2008). The cuticle is followed by the epidermis, “outer skin” (Figure 2, e), followed by the hypodermis, “under skin” (Figure 2, h), which is finally followed by the mesophyll or “middle leaf” region (Figure 2, m). The mesophyll can be undifferentiated (m) or differentiated into a distinct palisade mesophyll (mp) and spongy mesophyll (ms). The palisade mesophyll near the adaxial surface is composed of vertically linear cells; the spongy mesophyll near the abaxial surface is composed of round cells and intercellular spaces of various sizes. Within the mesophyll are veins of various sizes, which are also known as vascular bundles or fibro vascular bundles (Tomlinson et al. 2011). In this paper they will be referred to as primary veins (Figure 2, pv), secondary veins (Figure 2, sv), and minor veins, which, depending on their location, are called mesophyll veins (Figure 2, mv), adaxial minor veins (Figure 2, adv) or abaxial minor veins (Figure 2, abv). Some primary and secondary veins are often attached to the adaxial hypodermis and sometimes to both adaxial and abaxial surfaces by fibrous sheath extensions. If the attachment extends to both surfaces via a narrow fibrous sheath extension, the vein appears girder-like and is indeed referred to as a girder (Tomlinson et al. 2011) (Figure 2D, g). In some veins the fibrous sheath becomes so enlarged with fibers that they are referred to as veins with exaggerated fibrous sheaths (Tomlinson et al. 2011) (Figure 2B, C, D, vex). Some veins have exaggerated sheaths both adaxially and abaxially as in S. guimaraesensis (Figure 5) or S. pimentae and S. procumbens (Figure 7). In addition to the veins, the laminal tissues are supported by nonvascular fibers or fiber bundles of various sizes. Some have large fiber bundles adjacent to or near their margins (Figure 2C, fb). Many fiber bundles are adaxial and may reach close to 1/3 to ½ or more across the mesophyll (Figure 2A, C, adf). Minor, intermediate and major (large) fiber bundles can be found adaxially (Figure 2A, C, D, E, adf). Many minor fiber bundles are mainly abundant abaxially (Figure 2A, C, D, E, abf). Occasionally minor fiber bundles are scattered throughout the mesophyll (Figure 2A, mf).

Mirrored anatomy—anatomy in which the abaxial surface is identical or similar to the abaxial surface. One surface appears to be reflected in the other. In this study, this term is not used exactly in the classical sense, where adaxial veins are exactly opposite other abaxial veins or adaxial fiber bundles are opposite abaxial fiber bundles (again, one surface reflected in the opposite surface). Here the term is used to describe the situation where structures are lined up perfectly opposite each other as if in a reflection. For example, adaxial fiber bundles may lie exactly opposite from abaxial veins or adaxial veins are opposite abaxial fiber bundles as in S. mendanhensis (Figure 6) or S. pleioclada (Figure 7). Again it is a reflection, but not necessarily of the same structures. This arrangement is not common in Syagrus.

Dorsiventral anatomy — anatomy in which the adaxial surface is very different from the abaxial surface. This is the anatomy most commonly seen in most Syagrus species.

The following characters were examined and used in this key:

Veins

• Presence/absence of adaxial minor veins

• Presence/absence of abaxial minor veins

• Presence of both adaxial and abaxial minor veins

• Quantity of adaxial minor veins (many, few)

• Adaxial minor veins (paired or opposite) abaxial minor veins

• Paired adaxial and abaxial minor veins (nearly touching or not)

• Presence/absence of abaxial vein with an exaggerated fibrous sheath

• Quantity of abaxial minor veins (many, few)

• Presence/absence of large marginal vein with exaggerated fibrous sheath

• Presence/absence of large marginal vein with exaggerated fibrous sheath on both adaxial and abaxial side

• Presence/absence of mesophyll minor veins

• Most mesophyll veins with or lacking abaxially thickened fibrous sheaths

• Location of mesophyll veins (middle, upper half, lower half)

• Presence/absence of primary or secondary vein (near or not near) the tip of the margin

• Primary, secondary or minor veins (attached, unattached)

• Surface to which the veins are attached (adaxial side only, both sides)

• Manner in which veins are attached (broadly attached, by one narrow fibrous extension, by two narrow fibrous extensions or girder-like)

• Order of arrangement of veins near the margin (secondary vein – minor veins –primary vein, primary vein – minor veins)

• Abaxial minor veins alternating with abaxial fiber bundles (or no such pattern).

Fibers

• Presence/absence of adaxial fiber bundles

• Size of the adaxial fiber bundles (reaching 1/3 to ½ across mesophyll, reaching less than 1/3, reaching less than 1/5 across mesophyll)

• Shape of adaxial fiber bundles (long and fat, long and skinny, elliptical, oblong, wedge-shaped, fat or skinny icicle-shaped, short and fat, irregular, rounded)

• First adaxial fiber bundle (the largest, ca. same size as others)

• Quantity of adaxial fiber bundles (many, occasional to none or few)

• Alignment of adaxial fiber bundles across from abaxial minor veins (aligned and opposite or not aligned)

• Mirrored versus dorsiventral anatomy (as defined above)

• Small adaxial fiber bundles alternating with larger adaxial fiber bundles

• Adaxial fiber bundles opposite abaxial veins (opposite each other, no such arrangement)

• Presence/absence of large to extra-large fiber bundle at or near the margin

• Presence/absence of abaxial fiber bundles

• Quantity of abaxial fiber bundles (many, few to none)

• Presence/absence of marginal minor fiber bundles (several, few to absent)

• Presence/absence of minor mesophyll fiber bundles

• Location of minor mesophyll fiber bundles (upper half, lower half, scattered throughout)

Other non-fiber, non-vein characters

• Differentiated or undifferentiated mesophyll

• If differentiated mesophyll, then the number of palisade layers (2 or 3 versus one)

• Cuticle layer (thick, thin)

• Adaxial hypodermis layer (sclerenchymous with thickened walls, near continuous one layer thickness of fibers, discontinuous layer of fibers, round bubble-like shaped cells, small rectangular shaped cells).

• Presence/absence of tiny fibers in the abaxial hypodermis layer

• Presence/absence of trichomes on the exterior abaxial surface

• Lamina thickness (<0.25 mm or 0.25 mm or more)

Non-anatomical characters used in the key

• Stem size (short underground [acaulescent], taller aerial stem)

• Stem number (solitary, in pairs or clustering)

• Leaf orientation (procumbent, spreading)

• Abaxial pubescence, tomentum or trichome color (brown, silvery) on the leaflets

• Leaf color (silver blue, green to dark blue-green)

• Middle leaflet measurements

• Leaf rachis length measurements

This key was designed for use in the field using simple tools and simple methods, which means using minimal equipment, no staining, low magnification (no higher than 100×), and the use of simple characters. Refer to the characters in the methods and Figure 2 for clarification. By using the methods listed above and following many of the simple techniques mentioned by Tomlinson et al. (2011), rapid results can be achieved in a laboratory provided with only the simplest of equipment.

I have examined several specimens for each species. However, I make no claim that this has been an exhaustive study. In making use of this key the user must allow for slight variation due to environmental and population differences from the published images. As more specimens were sampled, I expected to see more variation, but was surprised to see how many species stayed true to their basic arrangement of veins and fiber bundles, as well as agreed with previous work done by Glassman (1972, 1987). Nonetheless, a few do vary and for this reason more than one image is used to represent some species. Examples of this are seen in Syagrus cerqueirana (Figure 3) where the marginal vein with the exaggerated fibrous sheath varies in size along with the number of adaxial and abaxial minor veins and fiber bundles. Another example is S. cocoides (Figure 3), where the marginal tip shape varies, along with the quantity and frequency of adaxial fiber bundles. Syagrus glazioviana (Figure 4, 5) is an extreme example of variation. The variation seen in this species either represents true variation within a single species or it is revealing a complex of several closely related species. Environment definitely plays a part in this variation as seen in S. harleyi (Figure 5) where the low elevation form has a sclerenchymous hypodermis on the margins and along the adaxial surface making it more drought resistant, while the high elevation form has a sclerenchymous margin, but has less sclerenchymous tissue along the adaxial surface. The variation seen in S. hoehnei (Figure 5) represents two different populations (Harri Lorenzi, pers. comm.). Syagrus minor also represents two specimens separated by over 100 km with one having adaxial veins, which are nearly absent in the other. The variation seen in Syagrus vagans (Figure 8) clearly demonstrates the futility of trying to use marginal shape for species determination, at least in Syagrus.

Although I was not attempting to show close relationships in his key, nevertheless, certain relationships are suggested based on how these species resolved in the key. Such relationships can only be confirmed further by an in depth phylogenetic molecular analysis of the genus.

An outline of the key is presented (Figure 9, 10) to allow one to more clearly visualize the branches of the key and is not meant to depict a phylogenetic analysis. This is not a cladogram. An interesting, but unintentional consequence of producing this key has been to discover how species might be related based on their anatomical phenetics, i.e. possible relationships based on their overall similarity in anatomy or the organization of their anatomical characters. It must be emphasized that this is not an actual phenetic analysis using some kind of distance coefficient. In some cases, branches of the key contain species that had been previously shown to be related by molecular analyses (Meerow et al. 2009, 2014). As demonstrated in Table 1, species of the former genus Lytocaryum all resolve in the same branch of the key (Figure 10, branch 53) as they did in the molecular analysis of those analyzed. Species of the strongly supported Rain Forest clade (Meerow 2009, 2014) all emerge together in a branch of the key containing mostly Amazonian, Caribbean, Andean and few Atlantic forest species (Figure 10, branch 60). The Eastern Brazilian species of the molecular analyses are not as neatly grouped, but of the few species analyzed, they emerge in two portions of the key, but usually among other Eastern or Central-Western Brazilian species (Table 1, Figure 9, branches 18 and 43’). Cluster stemmed species mostly emerge in the same branch of the key (Table 1, Figure 9, branch 3’).

Frequently, neighboring species key out in the same branch (Figure 9, 10). Branch 2 contains species primarily from the Central-West region of Brazil. Syagrus allagopteroides and S. minor are from adjacent areas of Goiás and Minas Gerais (branch 4). All of the three subspecies of S. graminifolia (as defined by Noblick, 2017) fall out together in the key (branch 15’) and while they are from two different geographic regions of Brazil (Southeast and Central-west), they are really only from the neighboring states of Goiás and Minas Gerais, which just happen to form the boundary between these two regions.

Branch 19 contains species that can all be found in the state of Minas Gerais, Brazil. Syagrus mendanhensis and S. pleioclada (branch 20) are from the neighboring areas of the Serra Diamantina and Serra do Cipó in Minas Gerais as are S. gouveiana, S. duartei, and S. evansiana (branch 19). In that same branch, Syagrus kellyana grows in the neighboring Serra do Mar region of Minas Gerais closer to the coast, while S. coronata can be found growing in the northeastern part of the same state adjacent to its Bahian distribution.

Species of branch 19’, with one exception, come from the Central-West or adjacent Southeastern region of Brazil. Syagrus glazioviana, S. rupicola and S. longipedunculata (branch 29’) can all be found in the state of Goiás with S. rupicola and S. longipedunculata restricted to the Serra dos Veadeiros in the northern part of Goiás while S. glazioviana is more widespread.

Several Bahian species emerge in branch 18’. Syagrus microphylla and S. werdermannii grow in the mountainous interior of Bahia. Syagrus vagans tolerates the dry caatinga habitat in and between the mountains and the coast of Bahia. Finally, the rare, recently discovered S. santosii and more common S. schizophylla flourish near the coast.

With few exceptions most of the species in branch 37 grow in Brazil’s Central-West region, Mato Grosso, Mato Grosso do Sul, and Goiás (S. procumbens, S. pimentae, S. lilliputiana, S. guimaraesensis, S. pompeoi, S. pleiocladoides, S. oleracea). However, the S. oleracea distribution overlaps with S. cearensis from Brazil’s Northeast region and at one time they were considered to be the same species (Glassman 1987, Noblick 2004, 2017).

Branch 46 contains cerrado species from the adjacent Central-West and Southeast regions of Brazil. Syagrus emasensis and S. menzeliana are neighboring species (branch 49’) with the first growing in the southwestern corner of Goiás in the Parque Nacional das Emas and the other growing just south of the park.

It is not surprising to see both forms of S. harleyi key out in branch 51. However, the high elevation form also may key out earlier in branch 19 as some of the abaxial minor veins can be interpreted as being mesophyll veins.

All of the former Lytocaryum species (branch 53) grow in Atlantic Forest of Eastern Brazil from Brazil’s Northeast region and the eastern hills of Bahia (S. itapebiensis), through Brazil’s Southeast region including the Serra do Mar mountains of Espirito Santo (S. insignis), Rio de Janeiro (S. weddelliana), and into the state of São Paulo (S. hoehnei).

Most of the species found in branch 57’ are from the Caribbean, northern part of South America, the Amazon or the Andes. Syagrus amara from the Caribbean logically falls out in branch 62 next to S. stenopetala (Venezuela) and S. orinocensis (Venezuela, Colombia), the next closest Syagrus species. In branch 60, S. stratincola and S. inajai co-exist in the rain forests of Suriname and French Guyana of northern South America. Syagrus inajai, which is found throughout much of the Amazon basin also emerges near S. botryophora of the Atlantic Forest, an Amazonian connection that was also resolved in the molecular analyses (Meerow 2009, 2014). A similar Amazonian/Atlantic Forest connection pops up in branch 67’ between the pre-Amazonian species of S. vermicularis and the Atlantic Forest species, S. pseudococos. This Amazonian connection shown both here and in the molecular analyses offers further evidence that the two forests were connected at one time. Finally, several Andean species emerge in branch 60’ with S. smithii (Colombia, Ecuador, and Peru), S. sancona (Venezuela to Bolivia), and S. cardenasii (Bolivia).