Improving Checklists of Alien Plants with Digitised Herbarium Collections: The Example of D.R. Congo (Africa)


 The Democratic Republic of the Congo (D.R. Congo) represents a striking gap of knowledge on alien plant species. In this paper, we use digitised herbarium collections to assemble a checklist of alien plant species in D.R. Congo and to examine patterns in the alien flora. The new checklist comprises 426 alien species i.e., 182 (42.5%) casuals, 244 (57.5%) naturalised of which 80 (19% of aliens) are invasive. Discrepancies with previous databases are discussed. For many species in previous databases, we failed to find evidence for occurrence outside cultivation. A total of 158 taxa were not included in previous lists, 44 of which are new to D.R. Congo. Considering the size of the country and its rich native flora, the alien flora of D.R. Congo does not appear to be species rich. The alien flora is particularly rich in Fabaceae (15%) and in annual species (36%). America is by far the most important source continent (65%) and the proportion of annuals of American origin is particularly large among the most widespread species. Invasive success is discussed in terms of residence time. The very low number of new species records after 1960 is most likely accounted for by decreasing sampling effort. The results illustrate how herbarium collections can be used to critically revise existing checklists of alien species in tropical Africa. Field work is urgently needed to improve coverage of recent introductions and to monitor the status of alien species, especially in protected areas and around botanic gardens.

In this paper, we use digitised herbarium collections to assemble a new checklist of alien plant species in D.R. Congo. First, based on herbarium specimens, all records in GLoNAF and GRIIS lists are checked for effective presence outside cultivation in D.R. Congo, geographic distribution, and date of rst record.
Secondly, based on extensive data mining from digitised collections, we systematically tracked species not recorded in either list. Based on the new list, we provide the rst comprehensive overview of the alien ora of D.R. Congo, in terms of phytogeographic origin, taxonomic assemblage, life forms, date of rst record and occurrence.

Study area
The Democratic Republic of the Congo covers 2,345,409 km 2 in Central Africa, spanning from 13°S to 5°N (altitude range: 0-5110 m). Its population is ca. 86 million. D.R. Congo was governed by the king of Belgium since 1885 and was a Belgian colony from 1909 to 1960. D.R. Congo has proportionally fewer paved highways than any country in Africa and its nominal GDP per capita is one of the lowest in the world (Herderschee et al. 2012). Agriculture is the largest sector in economy with 10 million ha cultivated (FAO 2013). D.R. Congo comprises 18% of the world's tropical forests, but the Congo Basin is subjected to steadily increasing human in uence due to deforestation and urbanisation (Anonymous 2012;Potapov et al. 2013), which could favour the expansion of non-native species

2007)
i.e., tropical rain forest (Af), tropical monsoon (Am), tropical wet and dry (Aw), temperate with dry winter and hot summer (Cwa), temperate with dry winter and warm summer (Cwb). The vegetation of D.R. Congo is highly diversi ed depending on climate and phytogeographic context, including Guineo-Congolian rainforest, Zambezian dry tropical woodlands and grasslands, afromontane and afroalpine communities (White 1983). Robyns (1948)  All entries, both from GRIIS and GLoNAF lists, and additional species from all other sources, were checked for effective presence in D.R. Congo. Effective presence is testi ed by a specimen in a collection. The most important collection for D.R. Congo is BR (www.botanicalcollections.be). Specimens in other collections were retrieved by GBIF. Literature records not supported by material evidence cannot be checked for taxonomic correctness and were therefore excluded. Very few exceptions were accepted i.e., when an herbarium specimen not seen by us is cited in a taxonomic publication, veri ed by an expert. The accepted names follow the APD, or POWO for taxa not covered in the former reference.

Presence outside cultivation
Herbarium labels have been systematically screened for locality and habitat information. First, presence outside cultivation has been veri ed by exhaustive screening of collecting information recorded.
Specimens collected from cultivated plants have been discarded (herbarium label explicitly mentioning "cultivé/cultivated", "planté/planted", "introduit/introduced"). Collections made in botanical gardens (Kisantu, Eala, University of Kisangani), arboreta (L'Etoile at Lubumbashi), and agronomic research stations of INEAC (Yangambi, etc.) have also been discarded, unless the collecting information on the label clearly indicated that the specimen was taken from a plant escaped from cultivation ("subspontané/ spontaneous", "échappé/escaped", "naturalisé/naturalised", "envahissant/invasive"). Any species reported by FAC as "subspontané" (i.e., escaped) or naturalised has been included. We have used different sources of information to assess the status of species in D.R. Congo, i.e., POWO, APD, FAC, Flora Zambesiaca and Flora of Tropical east Africa. We have generally accepted the status proposed by the majority of these sources. When the status was unclear e.g., when different sources give contradictory information (disregarding obvious material errors), the species has been included in the list as "cryptogenic" (Pagad et al. 2018).

Naturalisation and invasiveness
A preliminary attempt was made to assign each species to a particular stage in the invasion process. Naturalised species are those that form self-replacing populations independently of direct human intervention (Richardson et al. 2000(Richardson et al. , 2011Blackburn et al. 2011). First, we have considered as naturalised, any species previously reported as such in FAC, or other published oristic and vegetation studies, acknowledging that naturalisation can be effective only locally. Second, we have accepted as being naturalised species ful lling at least one of the following criteria: i) represented by large numbers of specimens in collections, ii) not restricted to anthropogenic habitats, iii) collected repeatedly from the same locality or the same region over an extended period.
Other aliens are referred to as "casuals". These comprise both deliberately introduced species occasionally escaping from cultivation but not maintaining self-sustainable populations and unintentionally introduced species observed as isolated specimen mostly in anthropogenic habitats.
Naturalised species have been assessed for invasiveness, based on Richardson et al. (2000) criteria i.e., "invasive species are a subset of naturalised species that produce reproductive offspring often in very large numbers at considerable distances from the parents and/or site of introduction, and have the potential to spread over long distances". Our assessment is based on previous reports in the literature, and on the number of specimens in collection. However, this criterion must be considered with caution because some species which have experienced recent and/or regional expansion are often underrepresented in collections.

Life form
Life form, (i.e., annual herbaceous, perennial herbaceous, tree, shrub, climber and aquatic), and region of origin were obtained from oras and POWO.  Table 1. The precise identity of a few taxa is uncertain (Bellucia cf. pentamera, Cecropia cf. pachystachya, Gnaphalium cf. pensylvanicum, Taraxacum sp., Vicia sativa s.l.)) and further taxonomic work is needed. Table 2 compares the revised checklist with previous databases. Only 194 out of 397 species in GRIIS list and 254 of 522 species in GLoNAF list are con rmed as alien species occurring outside cultivation in D.R.
Congo. Many species in previous databases have been excluded for different reasons. First, for a total of 168 species (149 species in GRIIS and 159 species in GLoNAF), all specimens in collections appear to have been collected in botanic gardens, agronomic experimental stations, amenity gardens or otherwise cultivated specimens (Supplementary Information S2). BR collections comprise quite many specimens collected in the botanic gardens of Kisantu and Eala. For a few of these, however, collecting information indicates that the species has escaped from cultivation in the garden, and these have been accepted in our list. Secondly, a total of 88 species in previous databases are native to D.R. Congo (28 in GRIIS, 83 in GloNAF) ( Supplementary Information S3). For 12 species, the native range is not known with certainty and these have been accepted in the introduced list as "cryptogenic" (Supplementary Information S1).

Phytogeographic assemblage
The Americas are by far the most frequent source continent (65%), far ahead of all other continents (Fig. 3). The proportion of naturalised and invasive species varies among source continents (χ 2 obs = 38.73, d.f. = 10, P < 0.001). 27% of American aliens in D.R. Congo are invasive i.e., a much larger proportion than for all other origins (5%). Seventy-ve invasive species in D.R. Congo are from the Americas vs. only 8 invasive aliens from other regions.

Number of specimens and earliest record date
The number of specimens in collection ranges from 1 to 553 (Ageratum conyzaeoides). The earliest collection date is 1869 (Schwenckia americana) and the most recent new record is 2010 (Stachytarpheta cayennensis) (Supplementary Information S1). The number of specimens increases signi cantly with time since rst record (Fig. 4). When controlling for time, invasive species tend to have larger numbers of specimens compared to the rest of aliens.
The cumulated number of species has increased regularly in the rst half of the 20th century, reaching a plateau after 1960 (Fig. 5). As few as 31 new species have been added to the alien ora after 1960.

Distribution within D.R. Congo
Specimens in collections provide the only reliable, veri able evidence for effective presence and status (cultivated/wild) in D.R. Congo. A substantial number of records in previous lists are not supported by voucher materials. A few of these are material errors, including specimens collected in neighbouring countries (Burundi, Rwanda). Due to specimen mislabelling, several species have long been included in the ora of D.R. Congo while they were in fact collected in China (Robbrecht et al. 2021). Specimens in collection allowed us to detect several identi cation errors and wrong geolocalisation ( Supplementary  Information S4). Quite a few native species were also included in error in previous lists (Supplementary Information S3).
Our new list comprises many species that were not included in previous databases. Not all of these, however, represent additions to the alien ora of D.R. Congo, because they were already recorded as alien in D.R. Congo in Plants of the World Online. Extensive data mining from collections guided by literature search has allowed us to uncover forty-four species that had never been reported for D.R. Congo hitherto ( Supplementary Information S5).

Alien species richness
Based on the 522 naturalised species in GloNAF Pyšek et al. (2017) estimated the proportion of alien species in the ora of D.R. Congo to be 4.5%. Based on our revised checklist, the proportion of naturalised species is 2.3% (3.9% when including casuals). The naturalised ora of D.R. Congo appears to be relatively species-poor, in comparison to global patterns. Based on the species richness -area relationship in Pyšek et al. (2017), the expected number of naturalised species in D.R. Congo is ca. 1000 species i.e., four-fold the actual number (two-fold when including casuals). Second, based on the correlation between species richness in the alien and the native ora , the expected number of naturalised species is 457 species (for 11,000 native species), higher than the actual number (244), even when including casuals (425). D.R. Congo is obviously not a hotspot of alien species richness. These gures t in well with the generally low alien species richness in tropical regions (Fine 2002). Higher intrinsic resistance of tropical ecosystems could be due to "fewer available free ecological niches, faster recovery of vegetation after disturbance or a lower introduction rate" (van Kleunen et al. However, our alien list is likely conservative for different reasons. First, some species, represented in collections only as specimens from cultivation, may actually also exist as garden escapes. Botanic gardens potentially represent important sources of alien plant introduction, especially in the tropics (Dawson et al. 2008). Naturalisation from botanic gardens in D.R. Congo has been circumstantially reported (Kembelo 1996;Binggeli 2011). Field observations in Congolese botanic gardens and neighbouring areas could reveal many more naturalised species than accepted here. Second, herbarium collections have poor coverage of recent decades, a well-known sampling bias in central Africa (Meyer et al. 2016). The cumulated number of alien species shows a plateau in the last three decades (Fig. 5), with very few additions after 1960, most likely due to decreasing sampling effort. The same pattern of saturation was previously found for Ghana and Zimbabwe

Native vs. alien; naturalisation and invasiveness
Our assessment of naturalisation is likely conservative. Evidence for naturalisation cannot be derived from herbarium specimens. Specimen number is not always a reliable indicator of naturalisation success, because a species can be naturalised only locally. Several alien species appear to have naturalised in botanic gardens, from which they have expanded to neighbouring regions (Musa acuminata, Cecropia sp., Rivina humilis, Petiveria alliacea …). On the other hand, some species represented by large number of specimens in collections are still not considered as naturalised by recent oras e.g., Solanum aethiopicum (Bikandu et al. 2020). Interestingly, a few species that were explicitly reported as being naturalised locally have not been collected for a very long time (e.g., Musa acuminata in Yangambi), suggesting that such species have gone extinct or have failed to spread from their naturalised population. On the contrary, other species naturalised in botanic gardens, have spread quickly to neighbouring regions (e.g., Cecropia cf. pachystachya in the region of Eala (Hauman 1948)). The (past and present) role of botanic gardens in the introduction of alien species in D.R. Congo deserves further investigation.
In contrast to the relatively low number of alien species, the number of invasive species (80) appears to be quite high in comparison to expectations based on the number of naturalised species (21). This could be due to our too inclusive criterion used in this work. However, our estimation is remarkably close to the 84 invasive species in Zimbabwe (Maroyi 2012). Interestingly, most of the species considered here as invasive were already recorded in D.R. Congo over a century ago. This suggests that long residence time is a most important determinant of invasiveness (Ahern et al. 2010;Philips et al. 2010). However, when controlling for time since rst record, invasive species tend to have more specimens in collections. This could indicate that these species indeed have a higher spread rate, or, alternatively, that they were introduced long before the rst specimen was collected.  Table 3). The alien ora is richer in Fabaceae (17%) and poorer in Poaceae (9%) than global patterns. A prominent contribution of Fabaceae was also found in Ghana by Ansong et al. (2019). The large contribution of Solanaceae (ranking 3d, 9% of aliens) is a striking feature of the alien ora. Many Solanaceae species have been introduced to D.R. Congo as edible plants (leaves and/or fruits) and have often escaped from cultivation (Bikandu et al. 2020). However, in terms of invasive species, Asteraceae stand out with 16 species, most likely due to their relatively high dispersal capacity (anemochory).
Concerning life forms, the contribution of annuals (38%) is markedly higher than global patterns (23%) and very similar to Zimbabwe (37%); in Ghana, the bias towards to annuals is much less marked (Table 3). Trees and shrubs comprise 22% of aliens, very similar to Zimbabwe. The large contribution of annuals in Zimbabwe and D.R. Congo could be accounted for by the long dry season in the Zambezian ecoregion (southern D.R. Congo). The bias towards annuals is even more striking for invasive species, with 42 invasive annuals vs. only 15 invasive trees and shrubs. However, as pointed out by Bordbar and Meerts (2020), life history patterns are strongly family-speci c in the alien ora of D.R. Congo, with trees and shrubs much overrepresented among alien Fabaceae, and annuals much overrepresented among alien Asteraceae, re ecting contrasting introduction pathways (i.e., deliberate introduction for forestry purposes for Fabaceae, vs. accidental introduction of weeds for Asteraceae).
Concerning the region of origin, the overwhelming contribution of the Americas (65%), much larger compared to the global scale (30%) is in line with Ghana and Zimbabwe. This is accounted for by a long history of transatlantic exchanges of crops (and contaminant seed) between Africa and the Americas, dating back to the 16th century (Wild 1978) and possibly even earlier (Gallagher 2016). The Americas are even more overrepresented among the most widespread species (Table 1) and among invasive species (Fig. 3), as found in Ghana.
Considering the history of European colonisation, the virtual lack of European species is striking, as previously found in Ghana (Ansong et al. 2019). Climatic mismatch is likely the cause. This hypothesis is supported by the observation that quite a few cryptogenic species with a mostly temperate distribution, occur in D.R. Congo only at high elevation (Afromontane belt in the Albertine Rift.). Whether such populations are native or were introduced by humans is di cult to ascertain without evidence from molecular markers. Some of these mostly high-altitude species in D.R. Congo also have ruderal populations at lower elevation (e.g., Poa annua), certainly of recent anthropogenic origin. Such species are therefore accepted as aliens, as recommended by Pagad

Conclusions
Massive digitisation and online release of images of herbarium specimens offers excellent opportunities to improve checklists of alien species in tropical Africa. For D.R. Congo, many previous records are not con rmed by voucher materials collected outside cultivation and should be deleted from existing checklists. Conversely, extensive search in online collections made it possible to discover many previously unknown species for the alien ora of D.R. Congo.
The alien ora of D.R. Congo is relatively species-poor considering the large area of the country and its rich native ora. The overwhelming contribution of American species, of annual Asteraceae, woody Fabaceae, Solanaceae are striking features of the alien ora.
Field work is urgently needed to assess the status of many alien species, and to ensure better coverage of recent introductions. Botanical gardens and protected nature areas are priority targets in this research agenda.

Declarations Funding
No funding was received for conducting this study

Con icts of interest/Competing interests
The authors have no con icts of interest to declare that are relevant to the content of this article Availability of data and material All data are available in Supplementary