Geographic analysis of Red List Rhododendron ( Ericaceae ) taxa by country of origin identifies priorities for ex situ conservation

A Red List assessment is insufficient to determine priorities for ex situ conservation in large genera such as Rhododendron, where there may be hundreds of taxa in any one Red List category. We have utilised an analysis of the geographic origins of 1?215 taxa of Rhododendron (Ericaceae) as a method to prioritise Red List taxa for ex situ conservation. This analysis includes descriptions of distribution and endemism by country of origin, analysis of the incidence of the 715 Red List taxa by country of origin, and determination of the extent to which taxa from each country of origin are in cultivation. We determined that of 30 countries of origin and a ‘Europe’ aggregate, 24 origins contain Red List taxa. Of those 24 origins, 17 origins and ‘Europe’ have greater than 75 % of Red List taxa ‘in cultivation’, as defined in this study, so that Target 8 of the Global Strategy for Plant Conservation has theoretically been met. However, for some of these origins the number of each taxon held ‘in cultivation’ is very low and genetic diversity is likely to be poor. The remaining six countries of origin have less than 75 % of Red List Rhododendron taxa recorded ‘in cultivation’ (Indonesia (28 %), Papua New Guinea (29 %), Malaysia (59 %), China (60 %), Japan (62 %) and Solomon Islands (0 %)). Analysis of a set of Red List factors and ‘not in cultivation’ factors reveals that Red List taxa from Indonesia, China and Papua New Guinea should take priority for ex situ conservation.


InTRoduCTIon
Determination of priorities is one of the key challenges of conservation decision making and underpins successful development of both in situ and ex situ conservation strategies (Oldfield 2010, Ma et al. 2013).Under the Global Strategy for Plant Conservation (GSPC) (Sharrock et al. 2014) priorities are initially driven by the Red List process, conducted as part of Target 2 of the Strategy, whereby taxa under threat are identified and prioritised ahead of remaining taxa.Target 2 then drives Target 8, which requires 75 % of Red List taxa to be in cultivation by 2020 (Sharrock 2012).This process appears to be sufficient for small genera comprised of similar life forms, but does not have a fine enough resolution for large genera of wide distribution and multiple life forms, where there may be hundreds of taxa in any one Red List category.In these genera additional factors such as taxonomic groupings, geographic representation, endemism and centres of diversity (Farnsworth et al. 2006, Castaneda-Alvarez et al. 2015, Cavendar et al. 2015), can be used to identify key groups of taxa.Rhododendron L. (Ericaceae) is one such large genus (of about 1 215 taxa) in which Gibbs et al. (2011) and Argent (2015) Red Listed 715 taxa (MacKay & Gardiner 2017).In an earlier study we focussed on determination of ex situ conservation priorities at the subgenus level 3 (MacKay & Gardiner 2017), thereby laying the foundation for the present study which examines another prioritisation mechanism.Here we examine a series of factors that can be used to identify taxa from geographic origins that should be prioritised for ex situ conservation.
Rhododendron is a useful exemplar of the issues and challenges faced in ex situ conservation.Firstly, it illustrates the 'big genus' (Frodin 2004) characteristics of large size combined with taxonomic complexity and active speciation, which complicate conservation decision making (Ennos et al. 2005, 2012, Goodall-Copestake et al. 2005, Samain et al. 2009, Blackmore et al. 2011), and which prompted our examination of conservation priorities at the subgenus level (MacKay & Gardiner 2017).Secondly, Rhododendron represents a microcosm of conservation issues because it encompasses a wide range of life forms, ecological niches and habitats.The genus is centred in Asia and Southeast Asia and is one of the largest plant genera in China (Lopez-Pujol et al. 2006) and the Southeast Asian floristic region (Van Welzen et al. 2005, Webb & Ree 2012), with only about 40 of 1 215 taxa originating in Europe and North America.The distribution in Asia stretches from Pakistan in the west, through India and the Himalaya (Nepal, Bhutan and Myanmar) to China and eastern Russia, and to the islands of Japan and Taiwan.Moving southeast the distribution encompasses Thailand, Vietnam, the islands of Malaysia, Indonesia, the Philippines and Papua New Guinea as far east as Solomon Islands, and south to Australia.The distribution coincides with biodiversity hotspots in Sikkim (Pradhan et al. 2015), Indo-Burma (Krupnick & Kress 2003), Yunnan, Sichuan and Taiwan (Lopez-Pujol et al. 2006), the mountains of southwest China (Sharrock et al. 2014, Liu et al. 2015), Malaysia (Sharrock et al. 2014, Van der Ent et al. 2015), Sundaland, the Philippines and Wallacea (Krupnick & Kress 2003) and New Guinea (Melick et al. 2012, Sharrock et al. 2014).

Geographic analysis of Red List Rhododendron (Ericaceae) taxa by country of origin identifies priorities for ex situ conservation
Rhododendron inhabits lowland to alpine zones within a range of climate zones, from subarctic tundra to tropical rainforest, to tropical-zone mountaintops.Rhododendrons grow in swamps, meadows, grasslands, on mountainsides, on rocks and cliffs, and in ravines and river valleys (Cox & Hutchison 2008, Gibbs et al. 2011, Argent 2015).Many taxa are forest species, either trees or shrubs, and can be either a major or dominant component of the forest (Maren & Vetaas 2007, Shi & Zhu 2009, Bharali et al. 2011, Paudel et al. 2012).Some species form pure stands (Paudel et al. 2012, Ranjitkar et al. 2014) or are keystone species (Singh et al. 2009, Baird et al. 2014, Bharali et al. 2014).The epiphytic habit is common in some taxonomic groups, particularly in subg.Vireya (Argent 2015).Many are high altitude plants that are key species at, or on either side of, the tree line (Paul et al. 2005, Singh et al. 2009, Paudel et al. 2012).Above the tree line shrubby or creeping rhododendrons occur in shrub associations or in alpine meadows (Cox & Hutchison 2008, Paudel et al. 2012).
Rhododendron taxa have a range of economic and cultural values.Rhododendron niveum Hook.f. is the state tree of Sikkim (Pradhan et al. 2015), while some species are the focus of tourist attractions (Maren & Vetaas 2007, Mao & Gogoi 2012).Various taxa provide ingredients for liquor, fragrance or incense (Singh et al. 2009) and pharmaceutical products (Popescu & Kopp 2013).Many taxa are grown as ornamental plants in their own right (Cox & Cox 1997) or have been hybridised to develop horticultural cultivars (Leslie 2004).Several species support communities through use as firewood (Paul et al. 2005), although overharvesting of firewood is also a threat to survival of some taxa (Singh et al. 2009, Pradhan et al. 2015).Other threats include grazing and cropping, forest clearance, logging, habitat degradation, land development, infrastructure construction, increasing tourism leading to resort development in wilderness areas, illegal collecting, excessive collecting of species used for medicinal purposes, pollution, and climate change (Paul et al. 2005, Maren & Vetaas 2007, Singh et al. 2009, Oldfield 2010, Gibbs et al. 2011, Hird 2012, Ma et al. 2013, Liu et al. 2015, Pradhan et al. 2015, Van der Ent et al. 2015).
Although plant conservation involves a range of initiatives, in this study we focus on ex situ conservation where taxa are held in living collections in botanic gardens.This is an integral component of an overall conservation programme (Heywood 2015) and is fundamental to the achievement of Target 8 of the GSPC (Sharrock et al. 2014).Target 8 can be a challenge for mega-diverse countries such as China (Raven 2011) and India (Jalli et al. 2015) and is made more difficult by a mismatch between location of botanic gardens and areas of greatest diversity (Sharrock et al. 2014, Huang et al. 2015).Other challenges for ex situ conservation include accurate identification of specimens, appropriate documentation of collections, development of collections from wild-source material, acquisition of Red List taxa rather than common taxa, securing sufficient accessions at enough sites, and achieving adequate genetic representation (Lopez-Pujol et al. 2011, Pritchard et al. 2011, Rae 2011, Kozlowski et al. 2012, Cires et al. 2013, Ensslin et al. 2015, Huang et al. 2015).Of particular concern is insufficient capacity to house every rare taxon in a botanical collection (Heywood 2009, Lopez-Pujol et al. 2011, Ma et al. 2014), indicating that robust processes are needed to identify ex situ conservation priorities.
In addition to completion of a Red List assessment, planning for Target 8 requires two key resources; data on the diversity and frequency of taxa already in cultivation (Cires et al. 2013) and, for large genera, additional mechanisms to sort taxa into groups of higher and lower urgency, respectively.In comparison with threatened plants overall (29 % of species in cultivation or seed banks (Sharrock et al. 2014)) Rhododendron is relatively well placed with 70 % of 'all taxa' and 56 % of Red List taxa held in living collections (MacKay & Gardiner 2017)4 ; a small increase from the 67 % and 53 % reported by Botanic Gardens Conservation International (BGCI) in 2012 (BGCI 2012).However, while the overall figure is encouraging, our earlier study determined that no subgenus reached Target 8 and some subgenera are very poorly represented in cultivation (MacKay & Gardiner 2017).
The objective of the present study was to extend the previous study by examining Rhododendron and its Red List taxa in relation to country of origin.Our analysis included: i. description of the origins and endemism of Rhododendron taxa in general and Red List taxa in particular; ii.use of a Red List analysis to identify those countries with the most acute conservation issues; iii.description of the extent to which taxa from each country are in cultivation; iv.application of a 'not in cultivation' analysis to identify those countries whose taxa are poorly represented in cultivation; v. combination of the analyses to identify which countries should take priority for ex situ conservation; and vi.proposal of elements and geographical priorities for subsequent conservation action.

Data-set
Our data-set was originally constructed for a study of conservation priorities in Rhododendron at subgenus level (MacKay & Gardiner 2017) and was re-developed for the present geographical analysis.The data-set comprised 1 215 taxa and their Red List assessments (Gibbs et al. 2011, Argent 2015), with 'Red List taxa' (715 taxa) defined as those in all categories except Least Concern, while the remaining 500 taxa are in the Least Concern category.Taxa included species, subspecies and varieties as listed by Argent (2015) for subg.Vireya and Gibbs et al. (2011) for the other eight subgenera, with taxonomy checked and organised according to Chamberlain et al. (1996), Fang et al. (2005)

Geographical assignment and degree of endemism
For the geographical analysis each taxon was assigned to countries of origin according to Gibbs et al. (2011) and Argent (2015); those taxa with a single country of origin were coded as endemic.All countries were treated as separate data categories, except for those in Europe which were aggregated as 'Europe', as these 24 countries are home to only 10 taxa (eight taxa confined to this region and two circumpolar taxa that are shared with Asia and North America).This approach was taken to simplify the analysis and to reduce the size of tables.
Data were analysed in a number of ways, beginning with patterns of distribution and endemism.Firstly, the numbers of taxa, endemic taxa and non-endemic taxa were determined for each country of origin.

Security 'in cultivation' and derivation of a 'not in cultivation' score'
The fourth analysis examined the extent to which taxa are 'in cultivation' by describing the numbers and percentages of 'all taxa' (Least Concern taxa + Red List taxa), and Red List taxa separately that are 'in cultivation' for each country.The average number of BGCI records per taxon for groups of 'all taxa' and Red List taxa for each country were calculated by dividing the number of records for the group by the number of taxa 'in cultivation' in that group, thereby giving a measure of the security of those taxa that are 'in cultivation'.Any average fewer than three records per taxon is considered a risk threshold (Lowe 1988) below which taxa are not secure in cultivation.
The fifth analysis involved derivation of a 'not in cultivation' score using a series of nine 'not in cultivation' factors, as absence of Red List taxa from collections is also a driver of conservation priorities (Farnsworth et al. 2006, Castaneda-Alvarez et al. 2015).Eight of these factors had a similar form to the Red List analysis and were the numbers and percentages of taxa 'not in cultivation' for Red List taxa, endemic Red List taxa, Data Deficient taxa, and endemic Data Deficient taxa.The ninth factor was a rounded average number of records per Red List taxon on the BGCI database; this was calculated by dividing the number of records for each country by the total number of taxa for that country (not just the number 'in cultivation'), thereby giving a measure of the extent to which that group is represented 'in cultivation'.This ranking had 15 ranks so the highest ranked country scored 15 points, the next ranked country 14 points, and so on.The ranking scores for the nine factors for each country were summed to derive the 'not in cultivation' score (maximum score 135), with the highest scoring countries exhibiting the poorest representation of Red List taxa 'in cultivation' and therefore having the greatest urgency for ex situ conservation.

Development of Total Score and implication for conservation
In the sixth step we ranked origins according to a Total Score in order to identify those origins with the combined problem of an acute conservation issue and poor representation of Red List taxa 'in cultivation'.The analysis involved summing of the Red List score and 'not in cultivation' score to generate a Total Score for each country of origin (maximum score 303 points).The final step was to propose conservation actions and priorities for Rhododendron based on this final ranking.

Patterns of origin and endemism
The greatest number of Rhododendron taxa originate in China (649 taxa) followed by Indonesia ( 229), Myanmar (137), India (107) and Papua New Guinea (100).Seventeen countries and 'Europe' are each the origin of 10 or more taxa, while 13 countries are the origin of fewer than 10 taxa (Table 1).China is also the origin of the greatest number of endemic taxa (442 taxa), followed by Indonesia ( 168), Papua New Guinea (64), Japan (58) and Malaysia (53).Myanmar and India, while having large numbers of taxa, are relatively low in endemics (9 and 14, respectively); the Philippines, the United States of America (USA) and Taiwan have fewer taxa overall but have more endemic taxa (30, 21 and 16, respectively) than the two previous origins.In Australia and Sri Lanka, 100 % of taxa are endemic (a total of three taxa), while high percentages of taxa are endemic in the Philippines (91 %), 'Europe' (80 %), Japan (78 %), Indonesia (73 %) and Taiwan (70 %).China (68 %), Malaysia (64 %) and Papua New Guinea (64 %) are mid-range for this measure.
Table 1 also details the sharing of non-endemic taxa with other countries.For example, more than 50 % of non-endemic Chinese taxa are also found in Myanmar and India, while another 12 countries share 25 % or less of the taxa that originate in China.For Indonesian taxa, about half of the 61 non-endemic taxa are shared with each of Malaysia and Papua New Guinea.
Many countries have single taxa (not always the same taxon) in common with other countries; these tend to be taxa that are widespread in Asia (Rhododendron simsii Planch., R. moul mainense) or the circumpolar taxa (R. lapponicum, R. tomen tosum (Stokes) Harmaja) that are found in the northernmost regions of Asia, Europe and North America.

Red List taxa and endemic Red List taxa
The greatest number of Red List taxa originate in China (447 Red List taxa), followed by Indonesia (111), Myanmar (62), India (45) and Papua New Guinea (34) (    Guinea (29) and Malaysia and Japan (both with 25).For four origins and 'Europe', 100 % of Red List taxa are endemic (the Philippines have 19 such taxa, with only 5 taxa in total from the remaining countries), and Japan has 96 % of Red List taxa endemic.Other countries with high percentages of endemic Red List taxa are Indonesia and Taiwan (91 %) and the USA (88 %).
For the number of Red List taxa rated Data Deficient, China tops the ranking (183 taxa), followed by Indonesia (70) and Papua New Guinea ( 27).The highest ranking for percentage of Red List taxa rated Data Deficient is held by several countries with 100 % for this factor; however, there are only five taxa in total.Next in this ranking are Papua New Guinea (79 %) and Indonesia (63 %).China, although it has 183 Data Deficient taxa, ranks 5th for this factor, as the number of taxa is only 41 % of the Red List taxa from that country.
When number of endemic Data Deficient taxa are considered, China again tops the ranking (166 taxa) and this is 91 % of Data Deficient taxa from that country.The next rankings for number of Data Deficient taxa that are endemic go to Indonesia (64 taxa, 91 % of Data Deficient taxa from that country) and Papua New Guinea (23 taxa, 85 % of Data Deficient taxa from that country).
When the ranking scores (Appendix 1, column 1) are summed for each origin for the eight factors, the Red List score (Fig. 1) is led by China (157 points) followed by Indonesia (154 points), Papua New Guinea (140 points), the Philippines (139 points) and Japan (133 points).

Taxa 'in cultivation'
When the 844 'all taxa' and 400 Red List taxa 'in cultivation' (MacKay & Gardiner 2017) are considered by country of origin, most origins have greater than 75 % of 'all taxa' 'in cultivation' (Table 3, column 4), with the exceptions of China (72 %), Papua New Guinea (64 %), Indonesia (55 %) and the Solomon Islands (50 %).For most countries the average number of BGCI records for 'all taxa' (calculated across only those taxa 'in cultivation') is 10 or more, showing that taxa 'in cultivation' are relatively secure.In contrast, Indonesia, the Philippines, Afghanistan and Pakistan have BGCI averages of three or fewer, indicating that those taxa are not secure 'in cultivation'.
For Red List taxa, 10 countries and 'Europe' have 100 % of taxa 'in cultivation' (

Ranking for 'not in cultivation' factors and 'not in cultivation' score
The rankings for the 'not in cultivation' factors (Appendix 2) demonstrate that China ranks highest for number of Red List taxa 'not in cultivation' (178 taxa), followed by Indonesia (80 taxa) and Papua New Guinea (24).The same three countries, in the same order, also top the ranking for numbers of endemic Red List taxa 'not in cultivation' (168, 75 and 21 taxa, respectively).The ranking for percentage of Red List taxa 'not in cultivation' is led by Solomon Islands (100 % of one taxon), followed by Indonesia (72 %) and Papua New Guinea (71 %).Bhutan ranks highest for percentage of endemic Red List taxa 'not in cultivation' (100 % of one taxon) followed by Indonesia (75 %) and Papua New Guinea (72 %).
China, Indonesia and Papua New Guinea lead the ranking for numbers of Data Deficient taxa 'not in cultivation' (126, 60 and 19 taxa, respectively), and also the ranking for numbers of endemic Data Deficient taxa 'not in cultivation ' (122, 57, 17, respectively).The Solomon Islands (100 % of one taxon) leads the ranking for percentage of Data Deficient taxa 'not in cultivation', followed by Indonesia (86 %), Papua New Guinea (70 %) and China (69 %).For endemic Data Deficient taxa, Bhutan has the highest rank (100 % of one taxon) followed by Indonesia (89 %) and China and Papua New Guinea (74 % each).The ninth factor in the 'not in cultivation' ranking, the average number of records per Red List taxon on the BGCI database (calculated across all taxa in the category, not just those 'in cultivation'), is led by Indonesia and Solomon Islands with a rounded average of zero.Of the nine countries for which the BGCI average is fewer than three records per taxon, four countries (Afghanistan, Pakistan, Solomon Islands and Canada) have relatively few taxa.Brunei has 12 taxa; however, they are all shared with Malaysia, while Indonesia, Malaysia, Papua New Guinea and the Philippines have greater numbers of taxa and thus represent the greater risk.
When the 'not in cultivation' score is calculated (Fig. 2), Red List taxa from Indonesia (127 points), China (119 points) and Papua New Guinea (118 points) have the weakest representation 'in cultivation'.Some countries, such as Myanmar and India, which have relatively high numbers of Red List taxa, are placed in the middle of this ranking because they have greater numbers of their taxa 'in cultivation' and fewer Data Deficient taxa.Countries at the lower end of this ranking (e.g., Laos, Brunei) tend to be those with lower numbers of Red List taxa, with few or none Data Deficient, and with most taxa 'in cultivation'.

Total Score
When countries of origin are ranked for Total Score (the sum of Red List score and 'not in cultivation' score) Indonesia has the highest score (281 points of a maximum of 303) and would therefore be considered the highest priority for ex situ conservation (Fig. 3).In second place is China (276 points), followed by Papua New Guinea (258 points).

Wild-source material
Percentage of wild-source (WS) material 'in cultivation' for each origin varies from 0 % to 100 % (  (Gibbs et al. 2011, Argent 2015) ranked by number of taxa: showing number of taxa, number of endemic taxa, number of Red List taxa, number of endemic Red List taxa for each origin, with the percentage of each category for which there is wild source (WS) material 'in cultivation'.Total number of taxa is 1 215; however, column two will not sum to 1 215 as many taxa have more than one origin.Total number of Red List taxa is 715; however, column six will not sum to 715 as many taxa have more than one origin.

Additional taxa
While undertaking the present analysis we recorded 153 valid taxa that are yet to be Red List evaluated.Although Argent (2015) completed the evaluations for all but four vireya taxa, another 149 taxa from the remaining subgenera were not evaluated by Gibbs et al. (2011).When organised by country of origin (Appendix 3) 96 of the additional taxa originate in China (84 endemic), with 20 from Japan (19 endemic), 11 from Myanmar (2 endemic) and less than nine from each of the other countries.The majority of the additional taxa (100 of 153) are 'in cultivation' (data not shown).

Conservation issues and priorities
In the present study we have analysed the occurrence of the 715 Red List taxa of Rhododendron by country of origin, and determined the incidence of taxa 'in cultivation' and 'not in cultivation' by country of origin, to identify Indonesia and China as priority countries of origin for conservation of this genus.Our results have also highlighted a number of conservation issues.Data on origin and endemism identify countries with shared taxa, where cooperation among countries would be desirable in a conservation programme, e.g., Nepal, India, Bhutan and China.Conversely, countries with high degrees of endemism are also identified, e.g., the Philippines and Japan where only limited inter-country communities of interest will be possible, highlighting the desirability of a strong within-country collections network and effective relationships with international collections.
Data on Red List taxa demonstrate that for some countries endemic taxa are more threatened than non-endemic taxa, while for other countries the reverse is true.Comparison among endemic taxa demonstrates that there is greater conservation urgency for some countries (e.g., endemic taxa from India and Vietnam) than for others (e.g., Myanmar and Bhutan).Prioritising endemic taxa is a recognised conservation approach (Powledge 2011, Cavendar et al. 2015), with our analysis showing that a finer resolution is possible and comparisons can also be made among countries.
The Red List analysis (Fig. 1) identifies China, Indonesia and Papua New Guinea as highest priority countries and, as this ranking excludes any cultivation factors, the priority applies to both in situ and ex situ conservation.While China and Indonesia might be expected to take priority for conservation simply due to high numbers of taxa, they also rank highly because of the extent of endemism and Data Deficiency.The same two factors result in Papua New Guinea and the Philippines ranking above other origins (Myanmar, India) which have more Red List taxa in the first instance.
The Total Score analysis (Fig. 3), where cultivation factors are taken into account, prioritises Indonesia, China and Papua New Guinea for ex situ conservation.While China has a higher Red List score than Indonesia, and China would take priority for in situ conservation, taxa from Indonesia have poorer representation 'in cultivation' and so take priority for ex situ conservation.
(Notably, there is limited wild-source material 'in cultivation' for the countries prioritised by Total Score, exacerbating the ex situ conservation challenge for taxa from those countries.)In the middle of the ranking are countries which may have reasonable numbers of Red List taxa (e.g., Myanmar, Bhutan, Taiwan and the USA); however, factors such as good representation 'in cultivation' or lesser degrees of Data Deficiency place them in the middle of the ranking.At the lower end of the ranking are several countries which tend to have smaller numbers of taxa, fewer endemic taxa, most taxa 'in cultivation' and relatively few Data Deficient taxa.
Data Deficiency is a recurring issue in our analysis.Sixty three percent of Red List Indonesian taxa are rated Data Deficient and Rahman (2008) noted that Data Deficiency was most acute for non-Java species.Data Deficiency is a lesser problem for taxa of Chinese origin, where 41 % of Red List taxa were rated Data Deficient (although the number of taxa is more than twice that of Indonesia), indicating better knowledge of taxa from China and the capacity to assign a Red List category rather than being obliged to use the Data Deficient category.For several countries 100 % of their Data Deficient taxa are endemic, although urgency for conservation action would be greatest for Japan and the Philippines, as they have the greatest numbers of taxa.
The Data Deficient taxa comprise three broad groups.The first group is about 112 taxa for which Gibbs et al. (2011), despite using some 60 references, were unable to provide any information.
The second group is about 90 taxa listed by Gibbs et al. (2011) which are only known from the type specimen or a very limited number of specimens.Often this is a reflection of limited field studies; however, the issue can also be taxonomic.In the third group are about 100 taxa for which Gibbs et al. (2011) indicate that taxonomic status is uncertain or subject to debate; issues include taxa that are poorly described, suspected to be hybrids, or where there is debate about their synonymy with another taxon.These issues often coincide, e.g., few specimens were collected because the taxon is a hybrid.Such difficulties are part of the 'big genus' characteristics of Rhododendron (Frodin 2004) and highlight the importance of using DNA-based methods to enhance our scientific consideration of conservation priorities.
Biogeography is another issue raised by the present results.
Although one of the five top-ranked countries for Total Score is located on mainland Asia (China), the next four are located on nearby island groups in Japan and Southeast Asia (Indonesia, Malaysia, the Philippines and Papua New Guinea).In our study on subg.Vireya (MacKay et al. 2016), we determined that taxa from New Guinea (Indonesia and Papua New Guinea) should be prioritised over taxa from Borneo (Malaysia, Indonesia and Brunei); however, we did not consider the islands that comprise Japan or separate out the many small islands that surround the larger islands of Southeast Asia.The extent to which any individual island is important for Rhododendron conservation is yet to be determined and should be the subject of further research.
Finally, our analysis shows that 17 countries and 'Europe' have met the 75 % of Red List taxa 'in cultivation' required for Target 8, although low average numbers of BGCI records per taxon is a key issue (suggesting that genetic representation is likely to be poor).(Unfortunately this problem is common in plant conservation in general, e.g., 33 % and about 50 % of threatened species were present in only one collection (Pritchard et al. 2011, Cires et al. 2013, Hird & Kramer 2013).)Conversely, six countries do not meet the 75 % requirement for Target 8. Three (China, Japan, Malaysia) are reasonably well placed to achieve 75 % in the near future; however, there is clearly some work to do with respect to Papua New Guinea, Indonesia and the Solomon Islands in terms of both the range of taxa 'in cultivation' and the numbers of each taxon held.

Assumptions and limitations
The key assumption of the present analysis is that the primary drivers for ex situ conservation are threat, endemism, Data Deficiency, and presence 'in cultivation' (Newton & Oldfield 2008, Powledge 2011, Sharrock et al. 2014, Cavendar et al. 2015).
The first limitation of the present study relates to the two key data elements of a geographic analysis for ex situ conservation, an accurate Red List and accurate origin data, and these elements have some unavoidable limitations.Acquiring and organising up-to-date data for a Red List assessment can be difficult in general (Newton & Oldfield 2008, Oldfield 2010, Blackmore et al. 2011, Cires et al. 2013) and this weakness has been noted in particular for Rhododendron (Ma et al. 2014, Rahman & Rozak 2016).Gibbs et al. (2011) recognised this issue and noted the need for additional field research that may modify an assessment.For example, Ma et al. (2014) increased the threat level of two Rhododendron species while Rahman & Rozak (2016) reduced the threat level of two other species on the basis of additional field research.Despite the possible weaknesses in the Rhododendron Red List, conservation planning 'has to start somewhere' and the current Red List is a significant resource that can be used to inform future research.In due course, as new Red List assessments are made or previous assessments updated, these can be used in future conservation planning.The second key element of a geographic analysis is origin data.This is primarily obtained from type specimens, herbarium records and field studies; however, these are not always comprehensive and distribution can sometimes be defined by collecting patterns rather than actual physical distribution.Again, field work can revise distribution data (e.g., Huong & Hiep 2012, Ma et al. 2013, Yang et al. 2015); however, in large genera like Rho dodendron the sheer size of the task is challenging and there can be difficulties obtaining expertise, resources and access for field studies (Gibbs et al. 2011).
The next limitation relates to the 'cultivation' analyses.Because of its size and scope, the BGCI database was used as a primary source of data on taxa in cultivation; subsequently data on collections at the Royal Botanic Gardens Kew and Edinburgh, and in New Zealand, were added as prior studies had shown these to contain significant collections (BGCI 2012, MacKay & Gardiner 2017, MacKay et al. 2017).However, while the aforementioned sources are extensive, and some are readily accessible through online databases, there are other notable Rhododendron collections world-wide.Large private collections in the United Kingdom, the USA and Europe were not included in the BGCI (2012) study, and there were only 11 gardens from China and Asia among the sites considered.Additional research should be undertaken to expand the range of collections used to describe the 'in cultivation' and 'not in cultivation' aspects of any future analyses.Expansion of the range of collections should also overcome the apparent lack of wild-source material for North American taxa.As our main comparison was with British collections, which may not prioritise North American taxa, and the BGCI records do not indicate wild-source, it is likely that the absence of wild-source North American material is an anomaly that will disappear when a wider range of collections is examined.
The final limitation relates to the additional taxa that have not yet been Red List assessed.As only three of the 153 additional taxa come from Indonesia (first ranked in Total Score) and 96 come from China (second ranked for Total Score) with 84 endemic, a re-run of our analysis (when Red List assessments for the additional taxa become available) may create a different ranking.
Similarly, the rankings may be changed by newly discovered species such as R. stanleyi S.James & Argent (James & Argent 2017) and yet to be described species, as indicated by more than 40 wild-collected aff.taxa listed at Edinburgh (Catalogue of the Living collections: http://elmer.rbge.org.uk/bgbase/livcol/bgbaselivcol.php;acc.11 Jan. 2017).

Conservation action
Our analysis shows that Rhododendron taxa from Indonesia and China should take priority for ex situ conservation.This should take place within an overall conservation framework (ex situ and in situ) and in relation to 'all flora' for each country of origin.Indonesia held 21.5 % of Red List flora in general in botanic gardens in 2010 (Purnomo et al. 2010), with the present results showing Red List Rhododendron of Indonesian origin are similarly placed with 28 % 'in cultivation' (but not necessarily in Indonesia).Rhododendron taxa of Chinese origin (69 % of taxa Red Listed) are in a worse position than Chinese flora in general, where 20 % are considered at risk (Huang et al. 2015).Although China has in place a conservation strategy and a range of in situ and ex situ initiatives (Lopez-Pujol et al. 2011, Ma et al. 2013, Huang et al. 2015, 2016), more ex situ collections are needed in general (Raven 2011, Huang 2015), and more field work and ex situ collections are needed for Rho dodendron in particular (Ma et al. 2013(Ma et al. , 2014)).Other countries also have in situ conservation initiatives for 'all flora', e.g., Nepal (IUCN Nepal 2010) and India (Rana & Samant 2010), with initiatives that are particular to Rhododendron (e.g., Singh et al. 2009).Individual countries will wish to determine their own conservation priorities and the place of Rhododendron among those priorities is as yet unknown for many countries.-Individual nations that are a country of origin for Rhododen dron should, where possible and where resources allow: -Primarily focus their ex situ collections on their own endemic taxa, particularly Red List endemic taxa; -Form 'communities of interest' with associated countries in relation to non-endemic taxa.While individual countries may wish to perform their own field studies, research into habitat, taxonomy, and genetic diversity (including DNA-based studies) may be useful areas of cooperation; -Develop a programme of field work to: i. acquire more accessions of taxa that have limited representation in cultivation; ii.investigate relevant aspects of any taxon, with first priority assigned to Critically Endangered taxa; iii.investigate Data Deficient taxa, to clarify their conservation status; and iv.investigate taxa that are taxonomically uncertain; and -Develop additional ex situ collections, preferably in a botanic garden that is compatible with the habitat of origin or within a similar climate zone (while also selecting locations that minimise the risk of hybridisation between plants held in the collection and those in nearby native vegetation).
-Nations that are not a country of origin for Rhododendron should, where possible and where resources allow: -Focus existing ex situ collections on endemic taxa and the most endangered taxa from priority origins, prioritising those origins where there are currently few ex situ collections; -Propagate and disperse existing accessions of priority taxa, to other ex situ collections to guard against loss, particularly taxa of wild-source origin; -Consider becoming the designated primary or duplicate collection for selected groups of taxa, depending on climate zone, resources, and the characteristics of the existing collection; and -Contribute expertise and participate in activities to acquire new ex situ material and develop new collections.
The actions above should be supported by a programme of international coordination and cooperation for matters such as: i. determining where primary and duplicate collections would be located; ii.exchanging wild-source material among collections (where possible); iii.expanding the ex situ analysis to gain a broader coverage of taxa 'in cultivation'; and iv.investigate the diversity of existing collections and develop protocols to manage diversity and avoid genetic drift.
We have already proposed some elements of an international programme of cooperation (MacKay et al. 2016, MacKay & Gardiner 2017), and the results of the present analysis will enable priority countries of origin to be added to the international framework.

ConCLuSIon
The present study has shown that taxa that originate in Indonesia, China and Papua New Guinea are the highest priority for ex situ conservation of Rhododendron, while the priority for in situ conservation should be taxa from China, Indonesia and Papua New Guinea.Target 8 has been achieved for 17 countries and 'Europe' in terms of the percentage of taxa 'in cultivation' although holdings of some taxa are limited and genetic diversity is likely to be low.Compared to plants in general, Rhododendron has a high percentage of taxa 'in cultivation' and conservationists will be relatively well placed to undertake further developments in ex situ conservation.While individual countries will wish to determine their own conservation priorities, and endemic taxa are clearly important, there are also communities of interest where taxa are shared among countries and where joint conservation initiatives may be possible.Given the size of the genus Rhododendron, its taxonomic complexity, and the range of habitats and climate zones it occupies, development of an international and coordinated ex situ strategy is a subject for on-going research.The following countries have nil Red List taxa and therefore score zero for all factors in this ranking: Bangladesh, Cambodia, Greenland, Korea Nth, Korea Sth, Mongolia, Sri Lanka.PNG = Papua New Guinea; Solomons = Solomon Islands; USA = United States of America.

Fig. 1
Fig.1Countries of origin of Rhododendron(Gibbs et al. 2011, Argent 2015) ranked according to Red List score (= sum of ranking scores for eight Red List factors, Appendix 1).Maximum score = 168.The following origins have nil Red List taxa and therefore score zero on this ranking: Bangladesh, Cambodia, Greenland, Korea Nth, Korea Sth, Mongolia, Sri Lanka.

Fig. 2 Fig. 3
Fig.2Countries of origin of Rhododendron(Gibbs et al. 2011, Argent 2015) ranked according to 'not in cultivation' score (= sum of ranking scores for nine 'not in cultivation' factors, Appendix 2).Maximum score = 135.The following origins have nil Red List taxa and therefore score zero on this ranking: Bangladesh, Cambodia, Greenland, Korea Nth, Korea Sth, Mongolia, Sri Lanka.
have nil Red List taxa and therefore score zero on all factors in this ranking: Bangladesh, Cambodia, Greenland, Korea Nth, Korea Sth, Mongolia, Sri Lanka.PNG = Papua New Guinea; Solomons = Solomon Islands; USA = United States of America.

Table 2
).The Red List categories Vulnerable or Data Deficient dominate in most countries; two exceptions are the Philippines and Japan which both have relatively high numbers of Critically Endangered taxa.While China is the origin of the greatest number of Critically Endangered taxa, this comprises only 3 % of Red List taxa overall for that country whereas the Critically Endangered taxa comprise 19 % of the Red List taxa for Japan and 32 %

Table 1
(Gibbs et al. 2011, Argent 2015)ron(Gibbs et al. 2011, Argent 2015)ranked by number of taxa: showing number of taxa, number of endemic taxa, number of non-endemic taxa, and other countries with which the non-endemic taxa are shared.Total number of taxa is 1 215; however, column two will not sum to 1 215 as many taxa have more than one origin.

Table 1
(cont.)Another exception to the general pattern is the reasonably high numbers of taxa in the Near Threatened category for China, Myanmar and India.Data Deficiency is most pronounced for taxa from China, Indonesia and Papua New Guinea.The second entry for each origin in Table2shows the endemic taxa and their spread among the Red List categories.Taxa endemic to China exhibit a similar spread to Red List taxa overall; a pattern which is repeated for Indonesia, Malaysia, Papua New Guinea, Japan and the Philippines.Taxa from India demonstrate a different pattern; the 45 Red List taxa overall are dominated by Vulnerable and Near Threatened listings, while the 11 endemics are distributed almost evenly across the Critically Endangered, Endangered and Vulnerable categories.Although India does not have many endemic taxa they are in a relatively critical position; endemic taxa from Vietnam show a similar pattern.In contrast, the endemic Red List taxa from Myanmar and Bhutan are less threatened compared to Red List taxa overall from those countries.

Table 2
(Gibbs et al. 2011, Argent 2015)Rhododendron(Gibbs et al. 2011, Argent 2015)ranked by number of Red List taxa: showing numbers of taxa in each Red List category (first line), and numbers of endemic taxa in each Red List category (second line).Total number of Red List taxa is 715; however, column two will not sum to 715 as many taxa have more than one origin.

Table 3
(Gibbs et al. 2011, Argent 2015)ron(Gibbs et al. 2011, Argent 2015)ranked by number of taxa: showing numbers and percentages of taxa and Red List taxa 'in cultivation', and average numbers of Botanic Gardens Conservation International (BGCI) records per taxon for 'all taxa' and Red List taxa that are 'in cultivation'.Average numbers of BGCI records = number of records/number of taxa 'in cultivation'.Total number of taxa is 1 215; however, column two will not sum to 1 215 as many taxa have more than one origin.Total number of Red List taxa is 715; however, column six will not sum to 715 as many taxa have more than one origin.

Table 4
Countries of origin of Rhododendron (MacKay et al. 2016)ifically, we have already proposed actions and priorities for ex situ conservation in relation to subgenera (MacKay & Gardiner 2017) and subg.Vireya(MacKay et al. 2016), and many of the underlying principles will also apply to the geographic analysis performed in the present study.The following actions for ex situ conservation are recommended: