WorldForestID: Addressing the need for standardized wood reference collections to support authentication analysis technologies; a way forward for checking the origin and identity of traded timber

needed to support or refute origin and species claims in traded products. We the building of a geo-referenced wood reference collection (xylarium) supported by herbarium voucher specimens. The WorldForest ID program, hereinafter referred to as WFID (www.world fores tid.org), is embarking on large-scale field collections of wood samples suitable for science-based authentication technologies.


| INTRODUC TI ON
Forest products are humankind's most used inedible renewable resource, but supplies are not inexhaustible. Deforestation, either for utilization or to clear land for other purposes, has far-reaching environmental effects including reducing biodiversity, pushing individual species toward extinction, destroying habitats, and changing the climate and soil. The full economic value of timber is often unrecognized, and there is a very high volume of illegal trade in wood products that can only be quantified when confiscations have taken place (e.g., rosewood comprises 35% of the monetary value of confiscated illegal wildlife trade, UNODC, 2016a). Supply chains are complicated, often covering long distances and lead times: trees are routinely harvested at one location, transported through one, two or more countries before being processed into manufactured products such as furniture, flooring, or paper. End products are subsequently exported to consumers in distant countries. It may be the case for some supply chains that verification of origin is, for all practical purposes, impossible within existing certification and documentation methods. This means that the truth of two fundamental authentication attributes of timber-harvest origin and species-may remain uncertain or invisible, even when complex investigatory resources or due diligence systems are applied.
Although there is legislation in many countries aimed at regulating trade in timber and its products, there remains a high level of global illegal activity. In tropical forests, 50-90 percent of forestry activities is illegal in sourcing or trade practices ( Van der Werf et al., 2009). There are also problems with illegal extraction of temperate species such as larch (Larix spp., Blanc-Jolivet et al., 2018) and oak (Quercus spp., Schroeder et al., 2016). It is impossible to regulate trade without the ability to assess whether origin and identity claims are correct. Currently, investigative tools focus on documentation control, such as the veracity of the chain of custody, forest permits, and certifications. The vast array of legitimate timber traders has conventionally relied on certification and audit as the best means for validating the claims made about the origin and species of traded timber.
Until recently little use has been made of scientific evidence to support due diligence with respect to traded timber, although over the last 5 years there have been concerted efforts to change this (Dormontt et al., 2015;Koch et al., 2015;Schmitz et al., 2020;Schmitz, Beeckman, et al., 2019;Schmitz, Blanc-Jolivet, et al., 2019;UNODC, 2016b). In this article, we describe the development of the World Forest ID program (WFID), a working prototype for building geo-referenced, wood, and foliage reference collections. WFID has embarked on developing protocols for large scale field collections, comprehensive processing, and sample curation with the aim of providing wood samples suitable for an increasing range of science-based authentication technologies capable of validating origin and species. WFID focused technologies are those that have been shown to identify species and/or provenance of test samples effectively and reliably when compared to reference data. These include wood anatomy, stable isotope analysis, Direct Analysis in Real Time Time-of-Flight Mass Spectrometry (DART-TOFMS), and genetic approaches. These methods have been described in the UNODC Best Practices Guide for Forensic Timber Identification (2016) and most recently in the GTTN data analysis guide (Schmitz et al., 2020).
Importantly, in addition to supporting the needs of timber trade due diligence requirements, WFID sampling and handling protocols have been designed in collaboration with the US Department of Justice and UK Competent Authorities to be legally robust, meeting the needs of regulatory and enforcement authorities in the preparation and delivery of court action in producer and consumer countries.
WFID collections begin with the development and approval of certified plans prior to teams extracting geo-referenced timber samples in the field. After collection, geo-referenced samples are sent to Kew where they are quarantined, curated, and then submitted to WFID partner laboratories for analysis with results available to enforcement and other end-users via the WFID database housed at the University of Connecticut. In addition to field-collected samples, we outline how improving wood reference collections will underpin and enhance the usability of scientific data derived from them. Finally, although there has been encouraging progress, we emphasize that the wide range of active participants in trading timber must be encouraged and supported to adopt science-based wood authentication. that the material collected is suitable for current and future scientific analysis. We describe the process of collection and validation from field to laboratory and the advantages and disadvantages of the main techniques used to ascertain/verify identity and provenance. Ultimately, we envisage the day that scientific methods will be used routinely and successfully by timber traders, manufacturers, retailers, and law enforcement to accept or reject identity and provenance claims on internationally traded timber and forest products and, where necessary, to support prosecutions when laws such as EU Timber Regulations, Lacey Act and CITES are infringed.

K E Y W O R D S
DART-TOFMS, illegal logging, international trade, provenance, SIRA (stable isotope ratio analysis), wood anatomy, wood identification, xylarium Better communication and understanding will be needed between traders, legislators, scientists-and the wider timber consumer public-before accurate labeling of timber is achieved and the trade-in products resulting from illegal logging is effectively addressed.

| THE NEED FOR ACCUR ATE PROVENAN CE AND IDENTIT Y
Without knowledge of the identity and provenance of the woods in a product, it is impossible to know whether they are derived from legitimate legal and sustainable sources. Laws and regulations in many countries require timber importers to have completed and made available documents demonstrating that appropriate due diligence has been carried out describing the chain of custody of the traded timber. Different regulatory systems approach due diligence in different ways, for example, which participants in a supply chain must be identified as part of the process; and when in the supply chain due diligence regulations apply. For example, EU Timber Regulations (EUTR)-UKTR post-Brexit-require that due diligence declarations apply only when and if a product is first placed on a market. Other regulations, such as the Lacey Act in the USA require importers to correctly declare origin and species at the point of entry of their imported timber consignments. This is not as straightforward as it looks given that participants further up supply chains may be more or less anonymous, making the declaration task of US importers more challenging. Many taxa, such as oak (Quercus) and rosewood (mainly Dalbergia, also Pterocarpus, Guibourtia), have very wide geographic distributions and origin cannot be assigned from knowing the identity of species alone. Also with regard to species, precise identification can be difficult or impossible especially without expert knowledge (Gasson, 2011). Further complications include the need for taxonomic revision of many, especially tropical genera traded in the international market. The genus Dipteryx, with well-known wood anatomy (Gasson, 1999) is a good example where identification of species is problematic. Dipteryx odorata, one of 13 or 14 species in the genus (Carvalho et al., 2020; Plants of the World Online) has been claimed to be exported from Peru, but morphology and molecular evidence suggest that this species is not actually present there Honorio Coronado et al., 2020).
Incorporating these taxonomic refinements into trade and legislation on a large scale for many tree taxa is daunting. Determination of origin and species is thus a hard task for many traders, especially for those in non-integrated supply chains who rely on fallible paperwork alone to fulfill their harvest-origin and species declarations. For regulatory authorities engaged with enforcing the relevant legislation the task is harder still.

| IDENTIT Y: WHAT IS IT ?
This is the first question most people ask about a wooden item and it can be very difficult to answer. In many cases, an answer such as beech (Fagus), oak (Quercus), ash (Fraxinus), or teak (Tectona) will suffice, although botanically these are genera and not species.
These common names are mainly unambiguous, but even names such as teak and mahogany can relate to several genera, some of which include species that are protected by regulations under the Convention on International Trade in Endangered Species (CITES) and others not. The use of scientific names is essential. For example, Quercus (oak) and Fraxinus (ash) have very wide geographic distributions in the temperate northern hemisphere and comprise many species (c. 530 and 43, respectively), and pantropical genera such as Dalbergia (rosewood) are often very large (c. 250 species). Not only are these genera widespread, but the woods within each genus look very similar to each other and often to other genera, and many cannot be separated to species level by eye or under the microscope.
Trade-in timbers such as ash (Fraxinus) can also be a plant health issue when moved geographically (Spence et al., 2020).

| PROVENAN CE: WHERE DOE S IT COME FROM?
This is usually a more difficult question to answer than the identity of species or genus. Commercial forestry expertise extends across the globe, with the same species being grown in more than one country, often on more than one continent. The products of harvested timber, sawn timber, and logs can be traded to a second country, part processed, and transhipped to a third country where finished products are manufactured before re-exporting to markets in the west. Illegal logging and export of timber are a problem in many countries (EIA, 2012; including Peru), and there is compelling evidence that oak furniture and flooring imported into the USA and Europe often comprises several species of white oak from North America and the Russian Far East (EIA, 2013). It cannot be assumed that products made in a country come from trees that grew there.

| OUR APPROACH TO AN SWERING THE S E T WO QUE S TI ON S
Scientific methods are needed that will allow the comparison of suspect wood with reference wood samples of precisely known geographic origin and identity. These comparisons will only be possible when there is sufficient reference material (i.e., correctly named wood samples) to meet this need and scientific techniques that are reliable enough to answer the two questions. The passage of all samples from forest to Kew and beyond is being recorded on a smartphone application (https://app.world fores tid.org/) developed by colleagues at the University of Tennessee. This app, based on TreeSnap (Crocker et al., 2019) has been designed to capture field data including precise location coordinates and fraud prevention measures and meets the chain of custody needs of law enforcement.

| Sampling in forest
A key requirement is to collect samples suitable for any type of wood identification analysis method and to ship them from remote forests under conditions that preclude degradation and the growth of molds. WFID has developed collection protocols for two types of sample: sawn discs obtained from tree trunks during or close to the time of harvest and wood cores obtained from a mechanical device, the "Pickering Punch," used to bore fresh material from standing trees. The "Pickering Punch" comprises a hardened sharp metal tube that is driven into a tree at waist height. A core 18 mm × 125 mm is collected in the device and transferred to a cardboard storage tube ( Figure 1a Logistics and moisture control are essential facets of reference sample collection: relevant permissions, export, and import certification requirements are needed, including CITES permits when necessary. Voucher material in the form of leaves in silica gel (for future DNA analysis) and, where possible, herbarium specimens comprising foliage, flowers, and fruits, are also collected. It is required that voucher specimens are obtained directly from the sampled tree and not from foliage found around it. Green timber has a high moisture content (typically >80%) making sample spoilage post-harvest a risk.
Moisture is controlled by local drying and the use of rechargeable silica gel pouches that are placed alongside the sample immediately after collection and during shipment to Kew.
Samples are received at Kew's Plant Quarantine Unit where they are inspected for contamination (i.e., pests and pathogens). All incoming material is frozen at −40°C for 72 hr to kill any invertebrates that may be present. Only when necessary, samples are autoclaved to kill bacteria, yeasts, and fungi and subsequently put in an incubator to dry. Contaminated packaging is autoclaved prior to disposal.
Field collection protocols are constantly being refined to eliminate mold issues caused entirely by inadequate drying prior to shipping.
Following processing in Quarantine, the samples are checked for identity using wood anatomy (by PEG, IMB) and morphological examination of foliage, flowers, or fruits (by Kew Herbarium colleagues).
Provided there are no identification issues, wood subsamples of the entire collection are cut for shipping to analytical laboratories for further vetting. The methods used, DART-TOFMS and IRMS, efficiently screen the selection for outliers that can then be substantiated by the more time-consuming anatomical methods. By cross-checking the identification results of each method prior to incorporation in the final database, a robust collection is constructed.

| ME THODS US ED FOR IDENTIFI C ATI ON
These can be broadly separated into observational (macroscopic and microscopic), chemical (various chromatographic, mass spectroscopy techniques), and DNA analysis. These are discussed below. Paula Filho et al., 2014;Ravindran et al., 2020;Souza et al., 2020;Wang et al., 2013;Yusof et al., 2013). This is clearly a rapidly moving field of study. For automated machine learning to achieve its potential, adequate training datasets are required. Two issues that remain to be solved are the availability of enough reliably identified replicates of wood of a given species taking into account variations in radial growth rate and ensuring that the image analysis programs are not biased from lack of samples.

| Microscopic examination of thin sections on microscope slides
Microscopic analysis is the traditional method used by most wood anatomists and where many anatomical features can be seen with thin sections that cannot be seen on a solid piece of wood at low magnification (Gasson, 2011;IAWA Committee, 1989, 2004Koch et al., 2011Koch et al., , 2015. Wood needs to be boiled or at least softened with warm water before thin (15-40 μm) sections are cut using either a microtome or single-sided razorblade.
The wood is sectioned in three planes (Transverse Section (TS),

Tangential Longitudinal Section (TLS), and Radial Longitudinal
Section (RLS)) and compared with reference microscope slides under a compound microscope at magnifications mainly between ×40 and ×400, and less often up to ×1,000 for very fine anatomical features. Even if a good reference microscope slide collection is available, none are comprehensive and online databases and multiple-entry keys should be employed for comparison such as InsideWood (2004) (https://insid ewood.lib.ncsu.edu/search), and Richter and Dallwitz (2000).
Microscopy permits the examination of many features that cannot be seen by eye or with a hand lens and some characters are very good at narrowing down the number of possible identities of a piece of wood (IAWA Committee, 1989, 2004

| Chemical techniques
Chemometrics include the search for chemicals that are unique or re-

| Gene sequences-DNA
DNA is most easily extracted from green tissue such as leaves. It can be extracted from the inner bark (i.e., living phloem), the cambium, and the sapwood, especially if the sapwood has plenty of living axial parenchyma or ray tissue, and this will depend on the taxon and its anatomy. DNA is much more difficult to obtain from heartwood, very old timber samples in xylaria and antique furniture (Abe et al., 2011)  Although it may be possible to match DNA sequences from wood with existing known sequences, and thereby identify a genus or species, this will not necessarily pinpoint geographic origin. Lowe and Cross (2011)

| ME THODS US ED FOR PROVENAN CE
The most widely used method that can reliably provide information on provenance is stable isotope ratio analysis (SIRA). As discussed above, it has advantages over DNA analysis for provenance, but cannot be used at all for identification.

| Stable isotope ratio analysis for timber provenance verification
Stable isotope ratio analysis (SIRA) is the technique with the most promise for verifying the provenance of timber. All living organisms sequester elements during their lives, and the stable isotope patterns of H, O, C, N, S, and other elements will differ depending on where the organism lived. It is unlikely that the pattern exhibited by these elements in combination will be the same in distant places.
SIRA involves the use of an Isotope Ratio Mass Spectrometer (IRMS), a specialized device developed by Nier (1940), to measure the relative isotopic abundance of elements within a given sample. It is now a widely established analytical discipline, proving popular with scientists wanting to establish the geographic, chemical, and biological origins of organic substances. Since the turn of the century, a growing number of laboratories around the world have adopted the technology as a means of verifying the origin of food and drink (Boner & Förstel, 2004;Heaton et al., 2008;Kelly et al., 2007;Kelly & Rhodes, 2002;Li et al., 2014;Pilgrim et al., 2010). The same principles used to authenticate food were later applied to timber provenance research (Boner et al., 2007;Gori et al., 2013Gori et al., , 2018Horacek et al., 2009;Kagawa et al., 2008;Kagawa & Leavitt, 2010;Keppler et al., 2007;Rees, 2015). Stable isotope signatures have also been used to assess how environmental and physiological effects define the isotopic composition of C, O, and N in the wood of tropical trees (Sleen et al., 2017).
Verifying the precise geographic origin of timber typically depends on comparing a sample with a claimed origin against an authentic reference database covering the same origin. The technique is used routinely to assess legality, compliance with labeling legislation, validation of claims in forest certification schemes, and its use  Watkinson et al., 2020) and reports on Gabonese and Solomon Islands samples will soon be available on the WFID website.
The samples collected by WFID will provide abundant material for research on stable isotope signatures. This includes investigating how good the geographic resolution is for a particular species and whether adjacent tree species with different phenologies sequester stable isotopes in the same proportions. Ultimately, WFID consortium partners would like to be able to tell precisely where a tree grew and whether a piece of timber came from the concession it is said to come from, or from an adjacent protected area such as a national park. Once the WFID xylarium is fully established the consortium partners will be in a stronger position to answer such questions.

| FUTURE-PROOFING REFEREN CE COLLEC TIONS -WorldFores t ID
To ensure that wood reference samples are suitable for all current and future techniques, well-preserved and curated, uncontaminated, reliably identified, and accurately provenanced wood samples are needed in our xylaria. The aim of the WFID program is that specimens collected from now on will meet all these requirements. A particular need is that all the geographic provenance information is accurate and cannot be interfered with. We are aware that samples could be supplied with fraudulent geo-locations in an attempt to mislead, by, for example, stating that timber comes from a concession rather than a nearby protected area. For this reason, a high level of security will be needed to access full details of each tree collected, and WFID collectors will be unable to change or delete information that was automatically captured by the WFID smartphone app at the time of collection. We are currently obtaining our samples from trusted FSC concessions and will have rigorous selection criteria for future providers of wood samples. All field expeditions have included a WFID Advisory Board member who provides the training for and supervision of field crews. As WFID grows the number of trusted partners will increase.
The WFID smartphone application (https://app.world fores tid. org/) is being used to record the precise geographic location of F I G U R E 3 Geographic locations of trees collected for WFID from Peru. The yellow spots show the number of trees collected in that location. Clicking on the spot will lead to individual tree data. The detail available will depend on the user's level of access each tree, photos of the tree, its identity, and details of the extent of "voucher" material (leaves in silica for future DNA sequencing and enough material to be verified/identified by a herbarium taxonomist). Figure 3 shows the locations of trees collected in Peru.

| Reconciling the requirements of legislation and what science can provide
Legislation requires yes/no black or white answers. Biological science can rarely provide such unequivocal results. Plants are given scientific names by taxonomists who often differ in their opinions and their classifications of related taxa. A knowledge of nomenclature and scientific names is beyond the interest of many timber users but does go some way in explaining why there are so many discrepancies between botanical nomenclature and names used by industry.
The situation is compounded with common/local/vernacular names where the same name can be used for several often botanically unrelated taxa or many different names are used for the same species.
Timber and timber products in trade are accompanied by documents stating the identity of the woods involved, but these are often inaccurate, either inadvertently or deliberately. Unraveling common names and matching them to scientific names can be complicated, but resources such as woodid.info and Miller & Ilic (common name database online) will often help if searched with care in the context of the information available. Another complicating issue is that timber products such as furniture often consist of several different woods manufactured in one place but with different geographic origins. The inescapable solution is that the documents associated with timber and its products must use scientific names because common or trade names are too imprecise and open to misinterpretation.

| CON CLUS IONS
Ascertaining the identity and provenance of woods in trade can be challenging and is increasingly required in efforts to monitor, regulate, and enforce laws governing the international timber trade. In addition to global policy changes for more transparent supply chains, the best way forward is to build comprehensive collections of reliably identified vouchered wood samples of the relevant timber species with known precise provenance, and to submit these samples to observational and chemical analysis, producing reference data that can be used for comparison with traded timber. Cross-disciplinary research is undergoing that focuses on complementary identification methods rather than competition between them. A combination of two or three techniques, including anatomy, stable isotope ratio analysis, and DART-TOFMS will provide the answers much better than one technique alone, and they can be used to test the accuracy of permits accompanying timber shipments. A prerequisite for all samples in the reference xylarium is that they have been received from trusted sources, are accurately named and reliably geo-referenced for provenance. Ultimately, when reference collections of traded taxa are sufficiently comprehensive it will be possible to assess the accuracy of claims of any shipment of timber, raw, or manufactured.

AUTH O R ' S CO NTR I B UTI O N
All authors are active participants in WFID, responsible for various aspects of the program described in this paper. PEG conceived and coordinated his co-authors contributions as follows: RY, RPG, MPF, PEG, and ETL on the organization and aims of WFID, SR, IAMB, PEG, RPG, and RY on the stages between field to Kew and beyond, PEG on anatomy, CAL on chemistry, RY and GR on SIRA, and PEG and MPF on genetics.