Genome Size and Descriptors of Leaf Morphology as Indicators of Hybridyzation in Colombian cultigens of coca Erythroxylum spp

Genome Size and Descriptors of Leaf Morphology as Indicators of Hybridization in Colombian cultigens of coca Erythroxylum spp

A Thesis by Fausto Villafrade Rodriguez Zapata

Submitted to the Department of Biological Sciences of the Universidad de los Andes in partial fulfillment for the requirements for the Degree of Master in Biological Sciences

Approved by:

Committee Members: Prof Santiago Madrinan,˜ Ph.D. Prof Juan Armando Sanchez, Ph.D. Carlos Galeano, Ph.D. Carlos Guarnizo, Ph. D.

November 27 2015, Bogota D.C., Colombia. Acknowledgments

I would like to give special thanks to the following people:

To Coronel Miguel Tunjano and Agente Alberto Cepeda, Direccion´ Nacional Antinarcoticos´ Policia Nacional de Colombia, for their collaboration and guidance at the Pijaos coca experimental farm.

To Dr. John Mario Gonzalez´ and Natalia Bolanos,˜ Facultad de Medicina Universidad de Los Andes, for their essential help with the flow cytometric measurements without which it would have been impossible to complete this work.

To Prof. Rocio Cortes,´ Universidad Distrital, for her experience and forethought during collection trips.

To Adolfo Jara, Universidad de los Andes Laboratorio de Botanica´ y Sistematica,´ and Fabio Avila, Jard´ın Botanico´ de Bogota,´ for their taxonomic prowess and help determining the identity of material, and for their help during field trips.

To Natalia Contreras, Universidad de los Andes, Laboratorio de Botanica´ y Sistematica;´ Fanny Gonzalez, Ivan Calixto, Luisa Duenas,˜ Universidad de los Andes, BIOMMAR, for their invalu- able help during field trips.

ii iii

To Daniel Franco, undergraduate Biology student, Universidad de Los Andes, for his impressive enthusiasm and dedication in mounting coca leaves for image analysis, and for his time lapse photographic adventures showing coca leaf and flower develop- ment.

To Dr. Mathias Lorieux, Edgar A Torres, Lady Johanna Arbelaez,´ Proyecto de Arroz, International Center for Tropical Agriculture, CIAT, for kindly providing the rice seeds used as standard for flow cytometry measurement of nuclear DNA content.

To Vicerrectoria de Investigaciones, Universidad de Los Andes, and Proyecto Semilla, Departamento de Ciencias Biologicas,´ for project funding.

To my old friend Humberto Pardo for generously providing his services as driver and amateur photographer.

To Aleyda Reinoso, for her immense support during the last 2 years.

To my family, Luciana, Lina, Andrea, Ana, and most specially to my parents Luz Mar´ıa Zapata and Lino Rodr´ıguez for still believing in me. Abstract

Recent biochemical and morphological data suggests that coca crops in Colombia are increasingly composed of hybrids coming from traditionally cultivated varieties of and Erythroxylum novogranatense. Previous studies can not discrimi- nate these cultigens on the basis of linear measurements of leaf morphology. The current study aims at measuring quantifiable differences in Fourier Elliptic Descriptors of leaf morphology and genome sizes among collected Colombian cultigens. Our current hypothesis is that hybrids must show intermediate morphologies and genome sizes of the parental varieties, and that these vari- ables could be correlated. Given this scenario genome size and leaf morphology could be used as indicators of hybridization in cultivated coca populations, and used for the identification of further collected material. However no informative differences of leaf morphology or genome size were found between studied cultigens.

Key words: Erythroxylum; Genome Size;Flow cytometry; Leaf Geo- metric Morphology; Elliptic Fourier Descriptors.

iv Contents

Acknowledgments ii Abstract iv ListofTables vi ListofFigures vii 1Background 2 1.1. Hybridization and the cultivated varieties of Erythroxylum spp. 2 1.2. Leaf morphology in cultivated cocas 3 1.3. Fourier Elliptic Descriptors (FED) analysis of leaf morphology 4 1.4. Erythroxylum genome size and hybridization 4 2 Methods 6 2.1. Plant Material 6 2.2. Flow cytometry measurement of genome size 10 2.3. Elliptic Fourier Descriptors (EFD) analysis of leaf morphology 11 3ResultsandDiscussion 13 3.1. Differences in genome size of the two cultivated coca species and presumptive hybrids 13

v 3.2. The finding of an anomalous genome sized cultigen: Possible tetraploid? 20 3.3. Elliptic descriptors of leaf morphology: differ- ences between Erythroxylum coca and E. novo- granatense. 21 3.4. Conclusions 26 AFlowCytometryProcedures 28 A.1. Sample Filtration 29 A.2. Gating 30 A.3. Experimental Reproducibility 31 A.4. Nuclear DNA content correction for -80oC frozen tissue samples 32 A.5. Differences in genome size among cultigens 34 BLeafMorphology 35 B.1. Mounted specimen for morphometrics analysis 36 B.2. Linear measurements of leaf morphology 37 B.3. Mean leaf shape by cultigen 38 Bibliography 39

List of Tables

2.1. Erythroxylum at DIRAN experimental farm Arboretum 7 2.2. Taxonomic determination of the plant material and postulation of hybrids 8

vi 2.3. Herbarium material used for morphometric analy- sis 8

3.1. Nuclear DNA Content by Cultivated Erythroxylum sp. Cultigen and Species Group. 14

A.1. -80oC frozen and fresh samples by cultigen 32 A.2. Tuckey’s honest significant differences in genome size among cultigens 34

B.1. Linear measurements of leaf morphology 37

List of Figures

2.1. Geographic distribution of coca plant material avail- able from Pijaos Experimental Farm Arboretum collection, DIRAN. August 2015 9

3.1. Genome size by Erythroxylum sp. cultigen. 16 3.2. Tukey's honest significance differences among culti- gen genome sizes. 17 3.3. Differences in genome size between propagation methods by species group 19 3.4. Flow cytogram showing the anomalous nuclear DNA content of 'Boliviana Negra' 20 3.5. EFD Harmonic Contribution to Erythroxylum spp. cultigen mean leaf shape. 22

vii List of Figures 1

3.6. Morphospace of Erythroxylum spp. cultigen mean leaf shape. 23 3.7. Leaf morphology relationships among Colombian coca cultigens. 25 3.8. Relationship Between Leaf Morphology and Genome Size in Cultivated Erythroxylum spp. 26

A.1. Homogenate filtration contraption and procedure 29 A.2. Top: Plant nuclei, typically 1-3% of all particles, lie on a straight line in the FITC-H vs Comp-PE-A fluorescence scatterplot, the gate was defined for a sample transferred to other samples automatically in FlowJo, and adjusted manually.Bottom: From the gate defined at top nuclei populations were gated manually, and the mean fluorescence and coefficient of variation from these population gates were reported 30 A.3. Reproducibility of cytometric measurements 31 A.4. Nuclear DNA content correction for -80oC frozen tissue samples 33

B.1. Mounted leaves example. Specimen of Erythroxy- lum coca 'Boliviana Negra' 36 B.2. Mean leaf shape by cultigen 38 Chapter 1

Background

1.1. Hybridization and the cultivated varieties of Erythroxylum spp.

The presence of hybrids in illicit coca, Erythroxylum spp., fields seems to be increasing in Colombia. A rising fraction of the ma- terial seized by the Colombian authorities in the last 15 years corresponds to presumptive hybrids between the two cultivated species, Erythroxylum coca and Erythroxylum novogranatense, for which there is evidence from morphological [Galindo & Fernan- dez, 2010], genetic [Johnson et al., 2005, Johnson et al., 2003], and biochemical studies [Casale et al., 2014]. Economically important concentrations of cocaine (in the or- der of 1% of dry leaf weight) have been recorded just for the two cultivated species, and more recently also in Erythroxylum laetevirens [Bieri et al., 2006]. Both Erythroxylum coca and E. novo- granatense are closely related as shown by AFLP [Emche et al., 2011] and nuclear gene markers (ITS,idhB) [Islam, 2011]. Al- though there's no certainty as if the cultivated species are recip- rocal sister groups, the evidence points that way [Emche et al., 2011]. Ever since the time of the Tim Plowman's thorough studies [Plowman, 1979, Plowman, 1982, Plowman, 1984], the cultivated species were further separated into four varieties, two for each 2 1.2 Leaf morphology in cultivated cocas 3 species, with almost allopatric distributions and distinct charac- ters(see appendix 1). In the time of Plowman's work, the most widely illegally cultivated variety was E. coca var. coca, while E. coca var. ipadu was consumed by indigenous people as powder around the amazon basin; E. novogranatense var. novogranatense was present in Colombia and Venezuela and mainly cultivated in the Caribean coast for indigenous consumption and as an or- namental plant in house yards. Although E. novogranatense var. truxillense growth seemed to be restricted to the dry lowlands of Peru´ its special aromatics made it very appealing for conoisseur coca chewers, and for the same reason, after removal of cocaine, was the cultivar used for the Coca-Cola recipe [Plowman, 1979]. Furthermore, the separation around these clearly bounded popu- lations appears to be not only geographical, and enviromental, but is reflected in incipient reproductive barriers as demonstrated by artificial crosses [Bohm et al., 1982], and in a the genetic structure of the populations as shown by molecular DNA markers [Johnson et al., 2003, Johnson et al., 2005, Emche et al., 2011]. From these cultivated material present in Colombia, Johnson [Johnson et al., 2005, Johnson et al., 2003] has proposed that due to pressures from the authorities the coca cultivators have been selecting hybrids from E. coca var. coca × E. novogranatense var. novogranatense and E. coca var. coca × E. coca var. ipadu populations that can be better adjusted to local conditions.

1.2. Leaf morphology in cultivated cocas

As a leaf crop, the morphological, anatomical and develop- mental details of Erythroxylum leaves should be highly relevant for understanding cultivated coca diversity. Leaf morphology was part of the differences that allowed Morris to describe, and subsequently segregate E. novogranatense from E. coca. According to [Rury, 1981] leaves of E. coca, are usualy broadly elliptic to broadly lanceolate, often much larger and more broadly elliptic in Amazonian coca E. coca var. ipadu. Leaves of E. novogranatense are mostly oblong to oblong-elliptic less commonly oblong lanceolate 4 Background or obovate. Leaves of E. novogranatense var. truxillense are nar- rowly elliptic to elliptic lanceolate, and rarely obovate. However coca leaves vary inter and interspecifically showing an intergrad- ing series of shapes. Interspecific hybrids show intermediate phenotypes regarding branch architecture and leaf morphology [Johnson et al., 2003], being more elliptic in shape than either of the parents.

1.3. Fourier Elliptic Descriptors (FED) analysis of leaf morphology

FED analysis, coupled with PCA has been widely used as a technique for quantifying leaf form in many species [?, Neto et al., 2006, Strauss, 2010]. These descriptors have been shown to correlate with genetic factors during tomato leaf development [Chitwood et al., 2014]. This analysis is very similar to traditional Fourier analysis in that it decomposes a contour as a lineal combi- nation of sine, and cosine functions projected around an elliptic. The base function coefficients, serve as descriptors of the shape representing the charge of each of the harmonic base functions [Strauss, 2010]. The main advantage of this analisys, is that co- efficients are normalized so the numbers are scale invariant. In a similar manner these coefficients are invariant with respect to form orientation, so different leaves of different sizes and posi- tions can be processed by this algorithm to render comparable results. The mathematical structure of the output data, a vector of coefficients, lends itself as input for multivariate analysis like PCA, an for the construction of classificators via machine learning [Neto et al., 2006].

1.4. Erythroxylum genome size and hybridization

It is well known that the cultivated cocas are heterostylous, present self-incompatibility and thus have an outcrossing sys- tem of reproduction [Bohm et al., 1982, Ganders, 1979]. This reproductive systems has allowed farmers to seemingly increase 1.4 Erythroxylum genome size and hybridization 5 hybridization in cultivated coca [Casale et al., 2014, Galindo & Fernandez, 2010, Johnson et al., 2005]. However a rigorous study on the leaf morphology and the presumptive hybridization is still lacking. As a simpler way to identifying hybrids rather than more time consuming laboratory procedures leaf morphology is of special interest for plant identification, not just for botanists but for Colombian authorities. However, the morphology alone must be correlated with genetic evidence of hybridization in order to increase it's value as a proxy measurement. The flow cytometry based estimation of genome size has been used in natural and synthetic hybrids where hybrid intermediate genome sizes has been shown to be the case [Greilhuber & Leitch, 2013], alongside morphometric evidence. As part of the evidence for hybridiza- tion by Colombian cocaleros comes from the difficult taxonomic placement of specimens through morphological assessment, it follows naturally that a morphological phenotypical measure has to be taken in conjunction with the c-value data, leaf morphology was the obvious candidate. In order to answer this question we chose to use Fourier Elliptic Descriptors (EFD) [Strauss, 2010] of leaf morphology, as previous linear measurements [Galindo & Fernandez, 2010] have not been useful for discriminating between different cultivated populations in spite that there is biochemi- cal [Casale et al., 2014], and genetic evidence of hybridization [Johnson et al., 2005]. Chapter 2

Methods

2.1. Plant Material

An estimate of the the C-value around 0.5 and 1.5 pg was made from a bootstrap 95% confidence interval of the known dis- tribution of genome sizes for the order, as no genome size determination was available either for nor it's sister group Rhizophoraceae (Kew C-value database). Oryza sativa 'Nipponbare' with a 1C value of 0.818 pg (394.6 ± 0.33 Mb) [Ohmido et al., 2000] was the standard of choice for genome size estimation. Seeds were kindly provided by the International Cen- ter for Tropical Agriculture (CIAT) in Palmira, Valle del Cauca, Colombia, and germinated in Petri dishes over damp towel papers for 3 days in the dark before sowing the seedlings to 3L plant pots with black earth. Rice germination and growth were carried out in a room at 28oC with 12 hour light / 12 hours dark photoperiod. Taxonomic determination was by Fabio Avila, Orlando Jara, and the author, following diagnostic charachters according to [Rury, 1981] and [Bohm et al., 1982], principally the detantion/crenation of staminal tube in order to dertermine species and variety. How- ever given the taxonomic uncertainty for this study the I did not use infraspecific classification. Hybrid plants were considered as the cultigens identified in this manner by [Galindo & Fernandez, 2010] and where species determination seemingly contraticted

6 2.1 Plant Material 7

Table 2.1: Erythroxylum plants at DIRAN experimental farm Arboretum

Species Cultigen Propagation Locality Department Collection method year Amarga cuttings Valle del Guamuez Putumayo 2007 Amazonas unknown La Pedrera Amazonas 2009 Boliviana Blanca unknown Valle del Guamuez Putumayo 2007 Boliviana Negra cuttings Valle del Guamuez Putumayo 2007 Boliviana Roja seed Valle del Guamuez Putumayo 2007 Caturra cuttings Milan Caqueta 2011 Chirosa unknown Valle del Guamuez Putumayo 2007 Crespa seed Valle del Guamuez Putumayo 2007 Cuarentana unknown Santa Rosa Bolivar 2011 Dulce unknown San Jose Guaviare 2009 Gigante cuttings Valle del Guamuez Putumayo 2007 Millonaria unknown Patiroja cuttings Valle del Guamuez Putumayo 2007 Peruana Blanca cuttings Taraza Antioquia 2010 Peruana Negra unknown Pomarosa cuttings Valle del Guamuez Putumayo 2007 Tingo seed San Pablo Bolivar 2012 Tingo Macho seed San Pablo Bolivar 2012 Tingo Maria seed San Pablo Bolivar 2012 Tingo Negra cuttings San Pablo Bolivar 2012 Tingo Pajarita seed Valle del Guamuez Putumayo 2007 Hybrid Tingo Peruana cuttings Valle del Guamuez Putumayo 2007

that work. It is clear that full genetic studies are required to have the certainty of the hybrid origin of the material, and its classification as hybrids should be taken as a hypothesis given the contradictory morphological evidence. For morphometric analysis online herbarium paterial was used as reference Table Erythroxylum samples were taken from the arboretum collec- tion of 22 coca cultigens in the Pijaos experimental farm of the Diereccion´ Antinarcoticos´ de la Policia Nacional de Colombia (DIRAN), near El Espinal, Tolima (600 masl) [Regalado, 2013]. Plants have been collected from around the country during an- tidrug operations for the last 9 years (Colonel Tunjano, personal communication), mostly from the Colombian South, Putumayo, Caqueta,´ and Caribbean Mountains and raised in the dry forest of the Magdalena Valley. The plants are organized in a 24 plot of ˜3x4m each, are attended with fertilizer, drip and sprinkler irrigation for optimal growth given the local conditions. 98 indi- 8 Methods

Table 2.2: Taxonomic determination of the plant material and postulation of hybrids

Galindo & Bonilla , 2010 Avila, Jara 2015 Species group Amarga coca ipadu coca Amazonas ipadu coca Boliviana Negra cf. coca ipadu coca Boliviana Roja coca ipadu coca Caturra coca coca Chirosa coca ipadu coca Cuarentana coca coca Dulce ipadu ipadu coca Patiroja ipadu ipadu coca Peruana Blanca coca coca Peruana Negra coca coca Pomarosa coca coca coca Tingo Macho coca coca Boliviana Blanca E. coca X E. novogranatense (?) ipadu hybrid Gigante E. coca X E. novogranatense (?) coca hybrid Millonaria undetermined hybrid Tingo Maria novogranantense coca hybrid Tingo Peruana E. coca X E. novogranatense (?) coca hybrid Tingo Negra truxillense coca hybrid Crespa truxillense truxillense novogranatense Tingo truxillense novogranatense Tingo Pajarita novogranatense novogranatense novogranatense

Table 2.3: Herbarium material used for morphometric analysis

Collector Determined by Herbarium Variety Notes Wade Davis 19 Tim Plowman F ipadu JGSA¡nchez˜ Vega 232 Adollfo jara F truxillense Plowman T 6663 Tim Plowman US ipadu isotype Purdie Tim Plowman K novogranatense Lectoparatype viduals, with between 3 to 6 individuals per cultigen, were tagged with a metal plate and voucher specimens (RC1966-RC3005 Rocio Cortes, FRZ1-FRZ87 Fausto RodrAguez)˜ for each individual with geographic origin notes has been deposited in the Universidad de Los Andes Herbarium (ANDES). In addition to the herbarium specimens, whenever possible three young branches with three young twigs were collected from each tagged individual for geo- metric morphology measurements, both kind of specimens were pressed and dried in an oven at 65oC for 24 hours following stan- dard herbarium procedures. Additionally 2 - 3 small twigs with apical buds and young leaves were collected in 2 x 4 inches paper 2.1 Plant Material 9 bags labeled and placed over a thick plastic grid separating the bags from damp paper towel in the bottom of a plastic recipient with lid. The recipient was maintained on ice until transferred to 4oC up to one week before flow cytometry measurements or discarding the leaf samples.

Cuarentana( Tingo( Peruana(Blanca( Tingo(Macho( !( Tingo(Maria( Amarga( !( Tingo(Negra( Boliviana(Negra( !( Boliviana(Blanca( Boliviana(Roja( Chirosa( Dulce( Crespa( !(

Gigante( !( Pa;roja( !( Pomarosa( !( Amazonas( Tingo(Pajarita( Caturra( Tingo(Peruana(

Figure 2.1: Geographic distribution of coca plant material avail- able from Pijaos Experimental Farm Arboretum collection, DI- RAN. August 2015 10 Methods

2.2. Flow cytometry measurement of genome size

2.2.1. Plant processing Plant material was prepared according to the Galbraith's pro- cedure [Galbraith & Lambert, 2011] with the following modifica- tions. Approximately 100 mg of fresh or frozen coca leaves, or 0.5 cm2 for the nipponbare standard, were immersed in 1000 uL Galbraith’s Buffer plus PVP and betamercaptoethanol (45 mM MgCl2, 20 mM MOPS, 30 mM sodium citrate, 0.1 % (vol/vol) Triton X-100, 2% PVP, and 1µL/mL betamercaptoethanol) in a 60 mm plastic petri dish on ice. The tissue was chopped with a new sharp razor blade for 30 to 60 seconds. The nuclei suspension was filtered through a 30 um Millipore nylon membrane. The filtrate (∼500 uL) was centrifuged at 5000g during 5 min, the su- pernatant discarded, and the nuclei pellet resuspended in 50 uL of Galbraith’s Buffer with propidium iodide added to a final con- centration of 50 µg/mL simultaneously with RNase at 50 µg /mL. A pooled sample with equal volumes of (50 µL) of sample and standard filtrate was prepared. The samples where incubated in ice protected from light at least 30 min before cytometry readings.

2.2.2. Flow cytometric meassurements BD FACSCanto II cytometer pre-run preparation was carried on according to manufacturer specifications, including fluorescent bead validation for FITC and PE fluorophores. Data acquisition was done with BD FACSDiva Software version 6.1.3. The measure- ment parameters specified where FSC, SSC for light scatter and FITC (525 nm), PE (575 nm) fluorescence emission resulting from excitation with a 20 mW laser at 488 mW. A logarithmic scatter plot of FITC vs PE, was used for gating data points along the di- agonal corresponding to nuclei signal as described by [Galbraith & Lambert, 2011]. From the resulting FCS files mean and CV of the PE signal was calculated from G0/G1 subpopulations selected with FlowJo X 10.0.7r2. 2.3 Elliptic Fourier Descriptors (EFD) analysis of leaf morphology 11 2.3. Elliptic Fourier Descriptors (EFD) analysis of leaf morphology

2.3.1. Specimen mounting and image acquisition The leaves from dry specimens were mounted with transpar- ent adhesive tape in standard letter sized white bond paper with the addaxial surface up and leaf tips pointing to the same long side of the paper, taking care all the leaves would have the most straight orientation. Branches were consequently mounted in the same paper or on an separate one whenever mounted leaves did not allowed the necessary space. Each leaf was marked with a sequential number and the number recorded in the correspond- ing node at the mounted branch. After this, the specimen was scanned with a hpScanjet 1320 scanner at 300 dpi resolution to a 24bit color jpeg image.

2.3.2. Contour acquisition A custom made script was written in Python using the OpenCV API [Bradski, 2000] for contour acquisition. This script loads the image file, desaturizes and binarizes the image. Where multi- ple leaflets were arranged in the image, the script identified and tagged individual leaves, then proceeded to align the major axis of the leaflet with the horizontal axis of the image. Finally the script does the contour detection, measurement of some - metric descriptors (width, length, area, aspect ratio, perimeter, eccentricity), and chain encoding of the contour. The output is a .chc file similar to the SHAPES [?] output, and compatible to be loaded by the package Momocs in R [Bonhomme & Picq, 2013].

2.3.3. Geometric Morphology with Momocs The first part of the analysis with Momocs corresponded to preprocessing and cleaning of the contours. After having acquired the contours for more than 3000 leaves, the contours were loaded to a Momocs Out object where they were aligned, scaled and centered, thus normalizing the ss for subsequent analysis. Fol- lowing this the EFDs coefficient matrix was calculated for the 12 Methods leaf shapes with efourier function, with 11 harmonics explain- ing 99% of the variance in the shapes, afterwards a PCA and a hierarchical clustering was made over this morphometric ma- trix. A panel of contours was useful to identify the shape outliers and artifacts of scanning/contour acquisition, and were trimmed from the set identifying clusters of non leaf shapes in the den- drogram of hierarchical clustering using the package APE and Phylo. Further analysis of the symmetric Fourier (symm function) components allowed for discarding bent, incomplete or abnormal leaves. The depurated dataset with 2715 leaves was subjected to PCA, MANOVA for for differences among cultigens and differ- ences among species groups. As the analysis of variance showed difference between cultigens but the PCA was not clear on this due to the sheer amount of data points a simplified PCA was made with the cultigen mean shapes in order to ilustrate the morphometric space and the difference between populations. Chapter 3

Results and Discussion

3.1. Differences in genome size of the two cultivated coca species and presumptive hybrids

The quality of flow cytometric measurements showed that reproducibility of the method for absolute genome size measure- ments in fresh and frozen coca leaves was acceptable. After 130 measurements the genome size internal standard Oryza sativa 'Nipponbare' (1C = 0.818 pg) [Ohmido et al., 2000] had a flu- orescence coefficient of variation (CV) of 5.7%, with a mean of 26443 and a standard deviation of 1498 arbitrary fluorescence units. Similarly, a single individual from greenhouse seed prop- agated 'Tingo Maria' was used as control for all experimental sessions, and after 20 measurements it showed a genome size CV of 3.3% (Table 3.1). Coca plants resulted somewhat difficult for cytometric measurements given the tendency of the leaves to oxidize as they dehydrated. This oxidation results in DNA degra- dation, as evidenced in DNA extraction. In cases where no fresh tissue was available, quick frozen tissue in liquid nitrogen and stored at -80oC was used with the caveat that these sample prob- ably underperformed in accuracy and precision [Dolezel et al., 2003]. Although coefficients of variation of frozen samples were significantly greater than the values obtained with fresh samples 13 14 Results and Discussion

Table 3.1: Nuclear DNA Content by Cultivated Erythroxylum sp. Cultigen and Species Group.

2C 1C Plant n m Fold Mean Mean CV Change (pg) (Mb) E. coca 2.65 ± 0.50 1294 ± 243 18.8% 52 62 1.07 E. coca without Boliviana Negra 2.65 ± 0.07 1294 ± 34 2.6% 46 52 1.57 Amarga 2.60 ± 0.04 1273 ± 20 1.5% 4 5 Amazonas 2.70 ± 0.07 1320 ± 33 2.5% 3 3 Boliviana Negra 3.97 ± 0.03 1943 ± 13 0.7% 6 10 Boliviana Roja 2.63 ± 0.07 1287 ± 34 2.6% 5 7 Caturra 2.69 ± 0.03 1315 ± 14 1.1% 5 5 Chirosa 2.53 ± 0.12 1236 ± 57 4.6% 3 3 Cuarentana 2.60 1270 1 1 Dulce 2.67 ± 0.02 1306 ± 12 0.9% 4 4 Patiroja 2.61 ± 0.05 1277 ± 23 1.8% 3 5 Pomarosa 2.59 ± 0.05 1267 ± 23 1.8% 3 3 Peruana Blanca 2.68 ± 0.05 1312 ± 23 1.8% 4 4 Peruana Negra 2.67 ± 0.06 1303 ± 29 2.2% 6 7 Tingo Macho 2.69 ± 0.04 1313 ± 20 1.5% 5 5 Hybrid 2.63 ± 0.10 1288 ± 48 3.7% 22 50 1.08 Boliviana Blanca 2.54 ± 0.02 1242 ± 11 0.9% 3 3 Gigante 2.51 ± 0.04 1229 ± 19 1.6% 4 9 Millonaria 2.71 ± 0.04 1325 ± 21 1.6% 2 3 Tingo Maria 2.68 ± 0.06 1311 ± 27 2.1% 5 5 Tingo Negra 2.71 ± 0.03 1323 ± 15 1.1% 4 6 Tingo Peruana 2.66 ± 0.04 1298 ± 19 1.5% 4 4 E. novogranatense 2.54 ± 0.12 1243 ± 60 4.8% 14 18 1.10 Crespa 2.46 ± 0.07 1201 ± 32 2.7% 4 6 Tingo 2.69 ± 0.04 1316 ± 22 1.7% 5 5 Tingo Pajarita 2.46 ± 0.02 1203 ± 9 0.7% 5 7

Total 2.72 ± 0.35 1329 ± 173 13.0% 88 130 1.62 Total without Boliviana Negra 2.63 ± 0.09 1284 ± 45 3.5% 82 120 1.10

1 pg DNA = 978 Mb, according to [Dolezel et al., 2003]. n number of different plants. m number of different flow cytometry measurements. Some plants were measured more than once. See Appendix for full dataset 3.1 Differences in genome size of the two cultivated coca species and presumptive hybrids 15 (Appendix A2) most of them were under the guideline of 5%. Regarding acuracy, paired measurements for the same plant of cultigen were used as calibration for estimating the degree of subestimation through interpolation (Appendix 3). In this man- ner 22 samples were corrected according to the calibration found. Results described hereon are accordingly based on this correction. Successful cytometric genome size measurements were taken for 89 of the 98 tagged plants, spanning 21 cultigens in the arboretum (Figure B.2 Table 2.1) through a combination of fresh and frozen samples. There is a significant difference between the nuclear DNA content of E. novogranatense and E. coca 2.54 ± 0.12 2.65 ± 0.50 pg (nested ANOVA p < 1e16), even when the anomalous result of textscflTingo Maria' is taken into account, the discussio on this particular case is left for another section an the current discussion does not include this anomaly. Even when there was an appre- ciable difference between coca species no significant difference was found between E. coca and the presumptive hybrids (nested ANOVA p = 0.08). It is interesting to see the low variability of genome size measurements in the coca poulations, interespecifi- cally they differ by 1.1 fold while intraspefically the average vari- ation was bellow this point, and is comparable to the observed 1.1 fold variation in accessions of Arabidopsis thaliana[Schmuths et al., 2004]. The 'Tingo' cultigen, determined as E. novogranatense, not only has a grater 2C DNA content than a presumptive hybrid, 'Chirosa' but differs the most, 0.16 pg, with their conspecific rep- resentatives 'Tingo pajarita' and 'Crespa' E. coca Figure A.3. In contrast 'Gigante' has a lower genome size in the range and dif- fers significatively from other presumptive hybrids and E. coca representatives. 16 Results and Discussion

Genome Size by Cultigen (n = 110)

Amarga ●●● ● Amazonas ● ● ● Boliviana Negra ●● ●●●● Boliviana Roja ● ● ●●●● ● Caturra ●●●● Chirosa ● ●● Cuarentana ● Dulce ●●●● Patiroja ● ●● ●● Peruana Blanca ● ● ●● Peruana Negra ● ●●●●● ● Pomarosa ● ● ● Tingo Macho ● ●● ● Boliviana Blanca ●●● Gigante ●● ●●●● ● Millonaria ● ● Tingo Maria ● ●●● ● Tingo Peruana ●●● ● Tingo Negra ●●●●● ● coca Crespa ● ● ● ● hybrid Tingo ●● ●● ● novogranatense Tingo Pajarita ●●●●●● Control range

2.0 2.5 3.0 3.5 4.0

2C Nuclear DNA Content (pg)

Figure 3.1: Genome size by Erythroxylum sp. cultigen. Each point represents a different plant. Some of the plants were measured more than once, either from fresh or frozen tissue, according to availability, thus the point represents the mean with correction for frozen tissue. Black bar shows the range of variation of measure- ments for a single plant of 'Tingo Maria' that was used as fresh tissue control for 20 measurement sessions. The reproducibility of the technique is evidenced in that this control plant had a 2C- value coefficient of variation of 3.3%. By the same way it can be observed that differences between cultigens are about the same magnitude than the resolution of the technique. 3.1 Differences in genome size of the two cultivated coca species and presumptive hybrids 17

Significant Differences in Genome Size among Cultigens

Peruana Negra Millonaria Tingo Peruana Blanca Caturra

Amazonas

Dulce

Tingo Maria

Tingo Negra

Tingo Macho

Tingo Pajarita

Boliviana Roja

Crespa Tingo Peruana Chirosa Boliviana Blanca Gigante

Figure 3.2: Tukey's honest significance differences among cultigen genome sizes resulting from an ANOVA with nested cultigens in each species group, adjusted p < 0.05. Line width proportional to the negative logarithm of p-value. Thinner line p ∼ 10−1. Thicker line p ∼ 10−6. Arrow heads point to the minimum number in the pairwise comparison. Colors according to species group: blue E. coca, orange hybrid, red E. coca. 18 Results and Discussion

Life history traits [Greilhuber & Leitch, 2013] like genera- tion time and invasiveness, and altitude and ecosystem [Marhold, 2010] are known covariates of genome size in angiosperms. Given the variation in mode of reproduction present in the cocas, it was pertinent to know if the observed differences cuould be due to wether they were mainly propagated by seed or vegetatively by cuttings. It is well known for farmers that E. novogranatense is mainly propagated by seeds, while E. coca 'Ipadu' is mainly prop- agated by cuttings for production in the amazonian basin. As in other crops, like Manihot esculenta, the two types of reproduction are not mutually exclusive, none the less one is often preferred by cultivators. Accordingly a two way nested ANOVA of cultigens by species was carried out. One could naively expect that clonally propagated lines would show lower variation than sexually repro- duced populations. However he observed difference in nuclear DNA content was not found to be statistically significant among propagation type groups, the only significant difference found was the relatively notable difference between species, i.e. the E. novogranatense vs E. coca comparison. Genome size in interbreeding congeneric species varies above the range observed for Erythroxylum in the present work. In the case of the Helianthus complex H. anomalous has a 2C content of 11.5 pg while H petiolaries has genome size around 6.8 pg, a difference of 1.7 fold [Baack et al., 2004]. In Malus the observed variation among diploid species and their hydrids is of 1.245 and 1.65 pg, a 1.3 fold increase [Hofer, 2001]. In a natural hybrid zone of Cercium including 12 diploid taxa and 12 hybrids observed genome sizes went from 2.14 to 3.60 pg [Bures et al., 2004]. The small variation in genome size observed among coca culti- gens reinforces the hypothesis that cultivated cocas might have diverged recently. For example traditional genetic markers as ITS an rbcL fail to discern relationships between these species [Islam, 2011], and the basal polytomies among the Archeoerythroxylum section were only resolved when more specific markers idhB by Islam and AFLP by [Emche et al., 2011]. 3.1 Differences in genome size of the two cultivated coca species and presumptive hybrids 19

Genome size by Species and Propagation Mode

n

coca ● ●●● ● ● ●● ●● ● 12

hybrid ● ● ● ● ● ● 6

novogranatense * ●● ● 3

unknown ● ● ●● ● 5 seed ● ● coca 2 cuttings ● ● ● ●● 5 unknown ● ● 2 seed ● hybrid 1 cuttings ● ● ● 3 unknown 0 seed * ●● ● novo 3 cuttings 0

2.4 2.5 2.6 2.7 2.8

2C Nuclear DNA Content (pg)

Figure 3.3: Differences in genome size between propagation meth- ods by species group. Upper panel: species differences found by an ANOVA of nested cultigens. Lower panel: two way nested ANOVA. In both cases significant differences (p < 1.0E − 5) were found between E. novogranatense and the other two species groups (*). No significant difference between propagation methods was found, neither among hybrids and E. coca. n = number of cultigens by species. Cultigen nuclear DNA content was measured for multiple plants as described in Table 3.1. 20 Results and Discussion

3.2. The finding of an anomalous genome sized cultigen: Possible tetraploid?

120 2.438 pg E. novogranatense

G0/G1 100

80

0.818 pg 60 Oryza sativa

Counts 'Nipponbare' 3.903 pg

G0/G1 E. coca 'Boliviana Negra' G0/G1 40

4.931 pg 20 E. novogranatense G2

0 0 50000 100000 150000 200000 250000

Comp−PE−A

Figure 3.4: Flow cytogram showing the anomalous nuclear DNA content of 'Boliviana Negra'. Although it is 1.6 fold the diploid E. novogranatense 'Tingo Pajarita' content it does not reach the expected tetraploid size, represented here by the G2 peak of E. novogranatense. Comp-PE-A: FACSCanto II PE (phycoerythrin) channel compen- sated fluorescence pulse area in arbitray units. PE Filter: 530nm, 30nm bandwidth. 3.3 Elliptic descriptors of leaf morphology: differences between Erythroxylum coca and E. novogranatense. 21 There is an unexpected difference in genome size of Boliviana Negra (2C = 3.97pg, 1943 Mb) with respect to other presumptive E. coca and E. novogranatense cultigens (Students t test p = 1.E−16), and its ratio is 1.62, more than the previously reported intraspe- cific variation of 1.2 fold in Microsperis douglassi [Greilhuber & Leitch, 2013], but not unusual at all compared to Helianthus or Malus, and below the known intraspecific variabilty for grasses, like rice, corn, maize or Brachypodium [Garcia et al., 2014]. Al- though it does not seem a duplication when compared to other E. coca, its exceptionally high DNA content deserves further investi- gation, a karyotype is required to fully discard this possibility.

3.3. Elliptic descriptors of leaf morphology: differences between Erythroxylum coca and E. novogranatense.

A total of 2291 leaves, corresponding to 54 plants from 21 cultigens were left after mounting, scanning and image process- ing. These numbers include a lectotype for E. novogranatense, and an isotype for 'Ipadu' and two other herbarium specimens one each for 'coca' and textscflTruxillense'. On average each cultigen was represented by 2 individuas with 40 leafs each, the greatest number of leafs per individual were recorded for 'Crespa' a plant characterized by small leaves which are retained during flower- ing, a typical trait of E. novogranatense [Bohm et al., 1982, Rury, 1981]. In contrast the least numbers of leaves per individual were recorded for E. coca and putative hybrid specimens. Plants of E. coca usually are shed at the flowering nodes, and are retained only at the growing tips, and so so a typical leaf remains attached attached to the plant for no more than one growth season. There seems o be a compromise in coca leaf size and morphology. While plants from the Amazon basin have bigger, less coriaceous more easily dehydrated leaves (easier to become oxidized, an thus more difficult for flow cytometry methods, smaller more coriaceous 22 Results and Discussion

Figure 3.5: EFD Harmonic Contribution to Leaf Shape of Erythrox- ylum spp. The first 6 harmonics are shown for brevity. The greatest weight in the morphospace PCA comes from the first 2 harmonics, which are shown to relate to base shape (decurrent-rounded) and leaf blade form (elliptic-obovate). leaves that are more resistant to dehydration were more suitable to flow cytometry. The Fourier Elliptic Analysis of leaf mean shape by cultigen, assuming 11 harmonics which explained 99% of shape variance, shows that the first two harmonics seem to correlate strongly with leaf base and leaf overall shape Figure 3.5. The greater the ampli- tude of the first harmonic, the more rounded the base and more widely elliptic the leaf blade. In contrast, the second harmonic varies with lesser amplitude towards a more elliptic shape and rounded base, whereas the greater its amplitude the more obovate the leaf blade, simultaneously with a more recurrent leaf base. 3.3 Elliptic descriptors of leaf morphology: differences between Erythroxylum coca and E. novogranatense. 23 PC2 7.05%

trxl BlvB Crsp

coca TngP Crnt ipad nvgr PC1 TngN Dulc TngM PrnB Ting AmrgChrs 89.7% Ctrr BlvN Mlln PrnN BlvR Pmrs Amzn Ptrj Ggnt

Eigenvalues Coca Leaf Morphospace

Figure 3.6: Morphospace of Erythroxylum spp. cultigen mean leaf shape represented as the first two components of the EFD normalized PCA. Note the divergence of 'Tingo Pajarita', with a greater aspect ratio than the bulk of other coca leaves (n=21 cultigens from m=2291 leaves).

These first two harmonics will have the greatest weight in the corresponding first 2 principal components of the morphometric space defined by EFDs.

A MANOVA of the morphometric space established by the Fourier Elliptic Descriptors showed significant difference (p = 2.5 E-7) between E. coca and E. novogranatense, but no between puta- tive hybrids and E. coca. Again, the behaviour of textscflTingo' differs from the other E. novogranatense representatives, showing no statistically significant difference with E. coca cultigens and the hybrid textscflTingo Negra'. This causes a certain overlap in leaf morphology among species, which makes species identifica- tion from single leaves very difficult. The observed distribution in the morphospace seems to be congruent with the hypothesis that hybrid progeny has an intermediate phenotype with respect to the parents. At least for these samples, the herbarium spec- imens tend to be located in the periphery of the morphospace, while presumptive hybrids tend to congregate in a central core of the distribution that coincides with the more elliptical inter- mediate leaf phenotype. For the mean leaves representing the population we can see that PCA1 accounts for 90% of the observed 24 Results and Discussion variance in the leaf shape. This principal component seems to be describing an elongation/roundness axis, where E. novogranatense. leaves tend to be longer than wider, and E. coca leaves tend to be rounder. This change in the aspect ratio is also correlated with another aspect of the first harmonic of the Fourier transform that seems to influence the base of the leaf to become more decurrent when leaves are longer, while rounder leaves tend to have a more cuneate base. of This result coincides with the general perception that E. novogranatense. leaves are oblong to oblanceolate while E. coca leaves are elliptical to obate. This distances through this mor- phological space can be used to represent the similarities among leave through a dendrogram. Very quatitatively there seems to be two far phenotypes, where the herbarium material of coca/ipau are clusterd together with other E. coca plants, while at the other extreme 'Tingo Pajarita' clusters naturally with the mean leaf of the E. novogranatense. lectotype. there is a middle cluster wher most of the coca and hybrids are found and are accompaniedby the 'Crespa' as the nearest representative to the herbarium mate- rial of 'Truxillense' c and of the but seems to contradict previous analysis of lineal morphological data that found no taxonomically significant clustering at all [Galindo & Fernandez, 2010]. 3.3 Elliptic descriptors of leaf morphology: differences between Erythroxylum coca and E. novogranatense. 25

Millonaria Amazonas C Cuarentana ipadu coca Chirosa Amarga Boliviana Negra Tingo Macho Dulce Pomarosa Boliviana Roja Gigante Peruana Blanca truxillense Crespa H Boliviana Blanca Tingo Negra Tingo Peruana Negra Caturra Patiroja N Tingo Pajarita novogranatense

0.20 0.15 0.10 0.05 0.00

Figure 3.7: Morphological relationships among Colombian coca cultigens. The dendrogram results from a Euclidean distance analysis of morphological PCAs. And identifies three branches of morphology N = novogranatense type mormhology, grater aspect ratio. C = coca type morphology ovate to obovate leaves with loweaspect ratio. H = hybrid type morphology, mostly elliptic. 26 Results and Discussion

Figure 3.8: Relationship between leaf morphology and genome size in cultivated Erythroxylum spp.. PCA1 comes from the princi- pal component analysis of EFDs. of mean leaf shape by cultigen, genome size (Cvalue) as 2C (pg). No significant correlation was found as denoted from R2 and p-value.

3.4. Conclusions

The analysis of the nuclear DNA content is a precise technique that given this sample size is powerful enough to detect a dif- ference of 100 pg in 2C, equivalent to near 50 Mb in 1200 Mb of 1C genome size. As a matter of fact the technique allows for an uncertainty in the genome size variation that is less than half the observed intraspecific variation in the plant diploid genome 3.4 Conclusions 27 size. But the fact that the resolution of the technique is above the threshold of detectability does not make the tool robust or practical for the work of routinely detecting hybrid plants coming from the field. The inconvenience of keeping intact nuclei for the experiment makes this technique really hard to implement as a hybrid detecting method. On the other hand there is still the question of whether the plants used in the current as presump- tive hybrids were really hybrids. Baack and collaborators [Baack et al., 2004] found that in the Helianthus complex it is common in synthetic hybrids to have the genome size of one of the parentals not an intermediate one. The alternative scenario of intermediate phenotype was more frequently found by [Bures et al., 2004]in the forest Compositae Cirsium. Given this, the presumptive hy- brids observed here adjust more to the first scenario than the second. In fact the observe pattern of morphological variation is as well consistent with the model that if there is indeed hybrid material in the studied samples, most probably those presumptive hybrid samples have phenotypes that are nearer one of the par- ents. In this case it seems that the more divergent cultigen from the rest of the population is 'Tingo Pajarita' representing the E. novogranatense species. The additional detail that well identified herbarium material shows itself remarkably at the extreme of the leaf phenotype, that it makes me wonder if there has been a detectable change between in the phenotype between populations at the time of collection for herbarium an now. Although in this sense the cytometric and morphpmetric evidence points to the same direction, it does not imply any correlation, in fact I found no correlation beteen the genome size and the principal compo- nent morphological axis (Figure 3.8). I think the real scenario must be of permanent introgression by the donor parent, in this case E. novogranatense., as it is of lesser economic value, and there- fore is grown in lesser proportion with respect to the other more crystallizable alkaloid productive cultigens. A dedicated genetic study of this population by means of RADseq, would determine the structure of the populations in these samples, and help in resolving the nature of the anomalous genome size of 'Boliviana negra'. Appendix A

Flow Cytometry Procedures

28 A.1 Sample Filtration 29

A.1. Sample Filtration

Flow Cytometry Filtration - Allison Church Bird, 2005

Home-made filtration device for flow cytometry sample preparation Load Sample Mesh (nylon screening) should range between 20-53 µm pore sizes – available from Small Parts Inc., Miami Lakes, FL www.smallparts.com

Gently force sample through filter by inserting P1000 into upper tip and applying pressure to enact filtration and collect sample PROCEDURE •Cut ends off 2 blue 1 ml pipette tips (black) •Press upper tip into mesh (red) placed on top of lower tube •Secure upper tip in lower tip by friction •Place assembly into collection tube (green) and load sample and filter •Filter after labeling and just before cytometry as cells, antibodies and proteins such as BSA in the flow buffer can precipitate during storage

Figure A.1: Homogenate filtration contraption and procedure 30 Flow Cytometry Procedures

A.2. Gating

Figure A.2: Top: Plant nuclei, typically 1-3% of all particles, lie on a straight line in the FITC-H vs Comp-PE-A fluorescence scat- terplot, the gate was defined for a sample transferred to other samples automatically in FlowJo, and adjusted manually.Bottom: From the gate defined at top nuclei populations were gated manu- ally, and the mean fluorescence and coefficient of variation from these population gates were reported A.3 Experimental Reproducibility 31

A.3. Experimental Reproducibility

Internal Standard Individual Control Individual Control 40 3.0 6

2.5 5 30

Erythroxylum sp. hybrid Oryza sativa 2.0 4 Erythroxylum sp. hybrid 'Tingo Maria' 'Tingo Maria' 'Nipponbare' n = 20 n = 20 n = 130 mean = 8661 mean = 2.691 SD = 7385 SD = 0.089 20 mean = 26484 1.5 CV = 8.5% 3 CV = 3.3% SD = 1424

Frequency CV = 5.4% Frequency Frequency 1.0 2 10

0.5 1

0 0.0 0

20000 25000 30000 35000 40000 6e+04 7e+04 8e+04 9e+04 1e+05 2.0 2.5 3.0 3.5 4.0

Comp−PE−A fluorescence Comp−PE−A fluorescence Nuclear DNA content, 2C (pg)

Reproducibility by tissue type 7 6 ●

5 n= 22 ● 4

3 n= 108 2 Comp−PE−A Fluorescence Coefficient of Variation

−80 fresh

Figure A.3: Reproducibility of cytometric measurements. Top: Distribution of fluorescence and calculated nuclear DNA content for standard and control plants. Bottom: Difference in Coefficient of Variation in nuclei measurements among frozen and fresh samples. Altough -80oC samples have a significatn higher CV than fresh samples most samples remain below the 5% gideline for reliable measurements. 32 Flow Cytometry Procedures

A.4. Nuclear DNA content correction for -80oC frozen tissue samples

Table A.1: -80oC frozen and fresh samples by cultigen

Measurements m Plants n) Cultigen -80 fresh total -80 fresh total Amarga 1 4 5 1 3 4 Amazonas 0 3 3 0 3 3 Boliviana Blanca 0 3 3 0 3 3 Boliviana Negra 1 9 10 1 6 6 Boliviana Roja 3 4 7 3 4 5 Caturra 0 5 5 0 5 5 Chirosa 1 2 3 1 2 3 Crespa 0 6 6 0 4 4 Cuarentana 1 0 1 1 0 1 Dulce 1 3 4 1 3 4 Gigante 4 5 9 4 4 4 Millonaria 0 3 3 0 2 2 Patiroja 3 2 5 3 2 3 Peruana Blanca 0 4 4 0 4 4 Peruana Negra 0 7 7 0 6 6 Pomarosa 3 0 3 3 0 3 Tingo 0 5 5 0 5 5 Tingo Macho 1 4 5 1 4 5 Tingo Maria 1 4 5 1 4 5 Tingo Maria Control 0 20 20 0 1 1 Tingo Negra 0 6 6 0 4 4 Tingo Pajarita 0 7 7 0 5 5 Tingo Peruana 2 2 4 2 2 4

Total 22 108 130 22 76 89 A.4 Nuclear DNA content correction for -80oC frozen tissue samples 33

● Fresh => Frozen Frozen => Corrected 4.0 Identity

Corrected = 1.07 Frozen − 0.12 3.5 3.0

● ●● 2C Nuclear DNA Content (pg) ● ●● ● ● ● ●● ● ● 2.5 ● ● ●

Frozen = 0.11 + 0.94 Fresh R2 = 0.92 2.0 2.0 2.5 3.0 3.5 4.0

2C Nuclear DNA Content (pg)

Figure A.4: Nuclear DNA content correction for -80oC frozen tissue samples. 22 plants have at least of frozen measurement and 4(3 Pomarosa, 1 Cuarentana) have no fresh cultigen counterpart. With these 18 points the regression line was calculated. The correction line is just the inverse function. 34 Flow Cytometry Procedures

A.5. Differences in genome size among cultigens

Table A.2: Tuckey’s honest significant differences in genome size among cultigens

max min diff lwr upr p.adj -log(p) Tingo Negra Crespa 0.25 0.10 0.40 0.000004 5 Tingo Crespa 0.23 0.09 0.38 0.000005 5 Tingo Macho Crespa 0.23 0.08 0.37 0.000010 5 Tingo Maria Crespa 0.22 0.08 0.37 0.000017 5 Peruana Negra Crespa 0.21 0.07 0.35 0.000038 4 Peruana Blanca Crespa 0.23 0.07 0.38 0.000040 4 Dulce Crespa 0.21 0.06 0.37 0.000173 4 Millonaria Tingo Pajarita 0.25 0.43 0.07 0.000259 4 Millonaria Crespa 0.25 0.06 0.44 0.000382 3 Tingo Peruana Tingo Pajarita 0.19 0.05 0.34 0.000433 3 Tingo Peruana Crespa 0.20 0.04 0.35 0.000832 3 Boliviana Roja Tingo Pajarita 0.17 0.31 0.03 0.001601 3 Tingo Negra Gigante 0.19 0.04 0.35 0.001639 3 Tingo Gigante 0.18 0.03 0.32 0.002804 3 Boliviana Roja Crespa 0.18 0.32 0.03 0.003070 3 Caturra Gigante 0.18 0.32 0.03 0.003170 2 Tingo Macho Gigante 0.17 0.02 0.32 0.005484 2 Tingo Maria Gigante 0.17 0.02 0.31 0.008759 2 Amazonas Gigante 0.19 0.35 0.02 0.011261 2 Peruana Blanca Gigante 0.17 0.02 0.32 0.013280 2 Tingo Negra Chirosa 0.18 0.01 0.34 0.020185 2 Peruana Negra Gigante 0.15 0.01 0.29 0.020480 2 Millonaria Gigante 0.19 0.01 0.38 0.033752 1 Tingo Chirosa 0.16 0.00 0.32 0.036568 1 Caturra Chirosa 0.16 0.32 0.00 0.040242 1 Dulce Gigante 0.16 0.31 0.00 0.043138 1 Tingo Negra Boliviana Blanca 0.17 0.00 0.33 0.048572 1 Appendix B

Leaf Morphology

35 36 Leaf Morphology

B.1. Mounted specimen for morphometrics analysis

Figure B.1: Mounted leaves example. Specimen of Erythroxylum coca 'Boliviana Negra' B.2 Linear measurements of leaf morphology 37

B.2. Linear measurements of leaf morphology

Table B.1: Linear measurements of leaf morphology

Aspect Width Length Perimerter Area Ratio Cultigen Leaves n mean sd mean sd mean sd mean sd mean sd Amarga 67 2 207 63 436 111 2.2 0.3 1135 294 67616 35205 Amazonas 100 3 238 68 464 130 2.0 0.3 1244 336 84312 45604 Boliviana Blanca 122 3 222 63 478 127 2.2 0.2 1230 325 78186 40370 Boliviana Negra 50 3 227 69 480 137 2.2 0.4 1247 347 81612 44872 Boliviana Roja 86 2 223 52 454 110 2.0 0.2 1187 270 73734 31146 Caturra 47 2 256 92 562 191 2.3 0.3 1442 478 111254 80393 Chirosa 63 1 225 72 459 127 2.1 0.2 1199 330 77264 41276 coca 18 1 144 41 242 68 1.7 0.3 671 179 25895 14739 Crespa 393 3 145 46 289 84 2.0 0.2 762 219 31379 17913 Cuarentana 49 1 175 40 340 74 2.0 0.2 916 200 43724 17581 Dulce 20 1 211 52 444 113 2.1 0.2 1148 278 68408 30860 Gigante 168 4 254 64 515 120 2.1 0.2 1340 305 95273 41624 ipadu 55 3 245 68 401 117 1.7 0.3 1125 307 74573 42462 Millonaria 68 2 283 92 542 166 2.0 0.3 1448 431 118325 67309 novogranatense 26 1 127 46 313 105 2.5 0.3 779 256 31315 19836 Patiroja 17 1 221 68 494 122 2.3 0.2 1272 330 80669 38549 Peruana Blanca 74 2 214 60 418 118 2.0 0.3 1124 287 66055 34029 Peruana Negra 151 2 181 56 398 104 2.3 0.3 1035 269 52626 28989 Pomarosa 148 3 244 67 497 134 2.1 0.2 1297 333 90357 44070 Tingo 153 4 190 61 418 139 2.2 0.3 1072 336 60257 36421 Tingo Macho 139 4 253 69 526 143 2.1 0.3 1366 347 99120 49877 Tingo Negra 223 4 172 59 373 125 2.2 0.3 966 314 48714 32313 Tingo Pajarita 42 1 129 33 347 70 2.7 0.3 845 174 30816 13338 truxillense 12 1 126 20 267 44 2.1 0.2 686 104 23322 7073 Total 2291 54 38 Leaf Morphology

B.3. Mean leaf shape by cultigen

Amarga Amazonas Boliviana Blanca Boliviana Negra Boliviana Roja

Caturra Chirosa coca Crespa Cuarentana

Dulce Gigante ipadu Millonaria novogranatense

Patiroja Peruana Blanca Peruana Negra Pomarosa Tingo

Tingo Macho Tingo Negra Tingo Pajarita truxillense

Figure B.2: Mean leaf shape by cultigen. Colored by species group, red E. novogranatense, E. coca, yellow presumptive hybrids Bibliography

[Baack et al., 2004] Baack, E. J., Whitney, K. D., Rieseberg, L. H., & Baack, E. (2004). Hybridization and genome size evolution : timing and magnitude of nuclear DNA content increases in Helianthus homoploid hybrid species.

[Bieri et al., 2006] Bieri, S., Brachet, A., Veuthey, J.-L., & Christen, P. (2006). Cocaine distribution in wild Erythroxylum species. Journal of ethnopharmacology, 103(3), 439–47.

[Bohm et al., 1982] Bohm, B. A., Ganders, F. R., & Plowman, T. (1982). Biosystematics and Evolution of Cultivated Coca (Ery- throxylaceae). Systematic Botany, 7(2), 121.

[Bonhomme & Picq, 2013] Bonhomme, V. & Picq, S. (2013). Mo- mocs: outline analysis using R. Journal of Statistical . . . , 56(13).

[Bradski, 2000] Bradski, G. (2000). The OpenCV Library. Dr Dobbs Journal of Software Tools, 25, 120–125.

[Bures et al., 2004] Bures, P., Wang, Y.-F., Horova, L., & Suda, J. (2004). Genome size variation in Central European species of Cirsium (Compositae) and their natural hybrids. Annals of botany, 94(3), 353–63.

39 40 Bibliography

[Casale et al., 2014] Casale, J. F., Mallette, J. R., & Jones, L. M. (2014). Chemosystematic identification of fifteen new cocaine- bearing Erythroxylum cultigens grown in Colombia for illicit cocaine production. Forensic science international, 237, 30–9. [Chitwood et al., 2014] Chitwood, D. H., Ranjan, A., Kumar, R., Ichihashi, Y., Zumstein, K., Headland, L. R., Ostria-Gallardo, E., Aguilar-Martinez, J. a., Bush, S., Carriedo, L., Fulop, D., Martinez, C. C., Peng, J., Maloof, J. N., & Sinha, N. R. (2014). Resolving Distinct Genetic Regulators of Tomato Leaf Shape within a Heteroblastic and Ontogenetic Context. The Plant Cell. [Dolezel et al., 2003] Dolezel, J., Bartos, J., Voglmayr, H., & Greil- huber, J. (2003). Nuclear DNA content and genome size of trout and human. Cytometry. Part A : the journal of the International Society for Analytical Cytology, 51(2), 127–8; author reply 129. [Emche et al., 2011] Emche, S. D., Zhang, D., Islam, M. B., Bai- ley, B. a., & Meinhardt, L. W. (2011). AFLP Phylogeny of 36 Erythroxylum Species. Tropical Plant Biology, 4(2), 126–133. [Galbraith & Lambert, 2011] Galbraith, D. W. & Lambert, G. M. (2011). Using the BD Accuri C6 Flow Cytometer for Rapid and Accurate Analysis of the Nuclear DNA Contents of Flowering Plants. (August). [Galindo & Fernandez, 2010] Galindo, A. & Fernandez, J. L. (2010). Especies cultivadas del genero´ Erythroxylum P. Browne. Revision´ del tema desde la perspectiva forense. Revista de la Academia Colombiana de Ciencias Exactas, F´ısicas y Naturales, 34(133), 455–465. [Ganders, 1979] Ganders, F. R. (1979). The biology of heterostyly. New Zealand Journal of Botany, 17(4), 607–635. [Garcia et al., 2014] Garcia, S., Leitch, I. J., Anadon-Rosell, A., Canela, M. A., Galvez,´ F., Garnatje, T., Gras, A., Hidalgo, O., Johnston, E., Mas de Xaxars, G., Pellicer, J., Siljak-Yakovlev, S., Valles,` J., Vitales, D., & Bennett, M. D. (2014). Recent updates and developments to plant genome size databases. Nucleic acids research, 42(Database issue), D1159–66. Bibliography 41

[Greilhuber & Leitch, 2013] Greilhuber, J. & Leitch, I. (2013). Genome Size and the Phenotype. In J. Greilhuber, J. Dolezel, & J. F. Wendel (Eds.), Plant Genome Diversity Volume 2 Physical Structure, Behaviour and Evolution of Plant Genomes (pp. 323– 334). Vienna: Springer Vienna.

[Hofer, 2001] Hofer, J. (2001). Genetic Control of Leaf Morphol- ogy: A Partial View. Annals of Botany, 88(6), 1129–1139.

[Islam, 2011] Islam, M. (2011). Tracing the evolutionary history of coca (Erythroxylum). PhD thesis, UNIVERSITY OF COLORADO AT BOULDER.

[Johnson et al., 2003] Johnson, E. L., Saunders, J. a., Mischke, S., Helling, C. S., & Emche, S. D. (2003). Identification of Ery- throxylum taxa by AFLP DNA analysis. Phytochemistry, 64(1), 187–197.

[Johnson et al., 2005] Johnson, E. L., Zhang, D., & Emche, S. D. (2005). Inter- and intra-specific variation among five Erythroxy- lum taxa assessed by AFLP. Annals of botany, 95(4), 601–8.

[Marhold, 2010] Marhold, K. (2010). Genome size correlates with growth form , habitat and phylogeny in the Andean La- siocephalus ( Asteraceae ). Preslia. (November 2015).

[Neto et al., 2006] Neto, J. a. C., Meyer, G. E., Jones, D. D., & Samal, A. K. (2006). Plant species identification using Ellip- tic Fourier leaf shape analysis. Computers and Electronics in Agriculture, 50(2), 121–134.

[Ohmido et al., 2000] Ohmido, N., Kijima, K., & Akiyama, Y. (2000). Quantification of total genomic DNA and selected repet- itive sequences reveals concurrent changes in different DNA families in indica and japonica rice. Molecular and General . . . , 263(3), 388–394.

[Plowman, 1979] Plowman, T. (1979). THE IDENTITY OF AMA- ZONIAN AND TRUJILLO COCA. Botanical Museum Leaflets, Harvard University, 27(1/2), 45–68. 42 Bibliography

[Plowman, 1982] Plowman, T. (1982). The identification of coca (Erythroxylum species): 1860-1910. Botanical Journal of the Linnean Society, 84(4), 329–353.

[Plowman, 1984] Plowman, T. (1984). New taxa of erythroxylum from the amazon basin. Supl. ACTA AMAZONIZA, 14((1/2)), 117:143.

[Regalado, 2013] Regalado, A. (2013). Coca science seeks an an- swer in kilos. Science (New York, N.Y.), 341(6145), 453.

[Rury, 1981] Rury, P. M. (1981). Systematic anatomy of Erythroxy- lum P. Browne: Practical and evolutionary implications for the cultivated cocas. Journal of Ethnopharmacology, 3(2-3), 229–263.

[Schmuths et al., 2004] Schmuths, H., Meister, A., Horres, R., & Bachmann, K. (2004). Genome size variation among accessions of Arabidopsis thaliana. Annals of botany, 93(3), 317–21.

[Strauss, 2010] Strauss, R. E. (2010). Morphometrics for Nonmor- phometricians. 124.