The taxonomic quality of biodiversity inventories depends on the accuracy of taxa identification, which ultimately depends on the knowledge and expertise of the identifier44–46. The expertise of the identifier is particularly important for groups of organisms with high taxonomic controversy and difficult species identification, like Hedera. Indeed, we found an average error rate of 47% in the original species identification of herbarium specimens (Table 1). Most of these errors stem from insufficient knowledge of ivies’ diagnostic characters resulting in the absence of an original species identification or in misinterpretations (“misidentifications”: 19% and 18% respectively, Table 1), and concentrates in the regions with more than one ivy species.
Current knowledge on species misidentification rates is limited, but the studies conducted indicate high variability among species and groups of organisms (0–56% for freshwater mussels47; 0,6%-41,1% for European ungulates45; 2,3–5,3% for French plant species46). The high misidentification rates we observed for some of the European ivy species (55% in H. hibernica, 48% in H. azorica or 38% in H. iberica, Table 1) are not surprising, given that even experienced botanists can encounter challenges in identifying ivies due to the difficulty in interpreting leaf trichomes38. Trichomes are small and easily lost during the development of individuals, thus, the first limiting factors for accurate identification of ivies are the magnifying glass (its quality and magnification) and the plant material used. Indeed, half of the misidentified specimens consisted of herbarium sheets containing only reproductive branches (see TaxRev database), a growth phase that rarely retains trichomes on their leaves. However, if these methodological issues were the only obstacle to accurate identification of ivies, the misidentification rate would be comparable for all species while our misidentification rates vary widely between species.
The species-dependent pattern in misidentification rates of ivies is explained both by intrinsic biological causes and human-based cascade errors. Firstly, the two main speciation mechanisms in Hedera (allopolyploidy48 and geographic isolation with slight niche shifts19,29) has rendered complex patterns of species variation in trichomes and leaves21. For example, the stellate-rotate trichomes, found only in H. hibernica, show intermediate characteristics between the other two types of trichomes in Hedera (stellate-multiangulate and scale-like, Fig. 2A; Valcárcel and Vargas 2010). These intermediate features are interpreted as the morphological footprint of the hybrid origin of the tetraploid H. hibernica from two diploid ancestors, one with stellate-multiangulate trichomes (like H. helix) and the other one with scale-like trichomes48,49. Also, the scale-like trichomes of H. iberica show certain characteristics that are more typical of the stellate-rotate trichomes of H. hibernica (small central part and rays of very regular length) than of the typical scale-like trichomes of the remaining Hedera species, which have large central parts and rays of more irregular length21. This morphological similarity between the trichomes of H. iberica and those of H. hibernica may also reflect the evolutionary history of the species, as a recent study has suggested a pattern of nested speciation of H. iberica within H. hibernica29. Interpreting the trichomes of these species can therefore be very difficult, in the case of H. iberica because they can be confused with those of H. hibernica, and in the case of H. hibernica because they can be confused with those of either H. iberica or H. helix23. However, the misidentifications between H. hibernica and H. iberica are rarely recorded in our database. Instead, the most frequent misidentification of the two species is with H. helix (Table 2). These results suggest that the high misidentification rates we observe are not only a direct consequence of the difficulty in ivy species identification.
Consistent patterns of error in species identification (systematic misidentifications) often occur between morphologically similar species, such as H. hibernica and H. helix. However, systematic misidentifications are less likely to occur between morphologically distinct species unless they live in sympatry, which is not the case of H. iberica and H. helix. Systematic misidentifications have a significant impact on the propagation of errors50 that may end up having an amplifying effect on the original misidentification and a collateral effect spreading the misidentification to other species (Cascade errors14). We interpret that the persistent confusion of H. hibernica with H. helix, originally prompted by their morphological resemblance, has contributed significantly to the spread of the notion that all ivies in mainland Europe are the common ivy. A feedback process that has, in turn, contributed to increase the effect of the misidentification of H. hibernica with H. helix while propagating the confusion to the quite distinct H. iberica. Ultimately, we wonder whether the high rate of correct species identification we obtained for H. helix (77%, Table 1) is the result of a greater knowledge about this species, or of a mere chance effect resulting from the widespread delusion that the common ivy is the only Hedera species in the Europe (excluding Macaronesian archipelagos).
Another interesting case of systematic error is the frequent misidentification of H. azorica with the quite distinct H. canariensis. In this case we attribute the systematic confusion to a cascade error originated in a long-standing nomenclatural confusion over the epithet “canariensis”. During the 19th and 20th centuries, several authors used “canariensis” to refer not only to the ivies from the Canary Islands, but also to those from Madeira, the Azores, and even to those from the southwestern Iberian Peninsula with trichomes other than stellate-multiangulate, which we now recognize as H. iberica and H. hibernica36. This is also probably the reason why H. canariensis is the second species with which most west European ivies are confused (after H. helix, Table 2) despite being the most morphologically distinct Hedera species in western Europe because of its entire heart-shaped vegetative leaves.
The systematic misidentifications we observed of all European ivies with H. helix and H. canariensis provides an explanation for the highly discordant results obtained when comparing distribution ranges from online and morphologically reviewed records (Figs. 3 vs. 4). For example, the widespread distribution of H. helix in the Iberian Peninsula according to online records and the contrasting narrow distribution of H. hibernica (Figs. 3 vs. 4), are likely an amplifying effect of the systematic confusion of ivies with H. helix when identifications are done in the field. Unfortunately, the few taxonomic validation systems that have been proposed to deal with species misidentification in biodiversity databases51,52 are not expected to solve the issues with ivies. The use of machine learning for species identification53 can be a powerful tool to validate inaccurate identifications of online photographic records51,52. However, this method may not work for Hedera as the features most frequently presented in photographic records are macromorphological characters of limited diagnostic use in Europe (leaves, flowers or fruits; see Hedera records in GBIF or iNaturalist). Similarly, the use of predictive niche modelling that has proven highly effective to correct inaccurate identifications in online records18 may not yield satisfactory results for European ivies either, as their niches overlap considerably, especially those of the most commonly confused species19,29.
Another notable disagreement between online and morphologically reviewed records is the occurrence in the Iberian Peninsula of H. canariensis and H. maroccana, which are not native in this area. While this can be easily solved by applying a simple geographic data mining, the application of this curation step requires from deep-level knowledge on Hedera to discard a naturalized origin. The naturalization of H. canariensis in the geographical areas indicated (Fig. 4) is unlikely because the climate in these areas is not as warm and humid as the preferred by H. canariensis19. The case of H. maroccana is different as the climate that this species occupies in its native range in Morocco is also present in part of the areas in mainland Europe where it has been recorded19 (Fig. 4). However, we can also exclude a naturalized origin for H. maroccana too, as this species has rarely been found to be naturalized23, despite being frequently used as ornamental plant in gardens in the Iberian Peninsula23 and in fences in southern Europe38. Therefore, the most likely explanation for the European field observations of H. canariensis and H. maroccana is the misidentification of the individuals, most likely caused by cascade errors (as described above).
The high species misidentification rates in the European ivies may have unknown consequences, extending the cascade effects beyond taxonomy14. Indeed, ivies, and particularly H. helix, have been used as role plants to analyse vegetation changes32, monitoring climate change54,55, medicinal uses56,57 and industrial applications58. The conclusions reached in this type of studies ultimately depend on a precise knowledge of the morphological and geographical delimitation of the species. However, our knowledge on ivy species distributions is low, despite the large amount of geographic information of European ivies available22. On the one hand, the morphologically reviewed database we compiled has significant spatial gaps in certain regions of Europe (Canary Islands, the Azores, France, C and E mainland Europe, and to a lesser extent in Great Britain; Table S2, Fig. 3). As spatial gaps have undesirable effects in other dimensions of the species’ knowledge59, the use of this high-quality taxonomic database for fine scale analyses is limited at least in four of the six European ivy species (H. azorica, H. canariensis, H. helix and H. hibernica). On the other hand, we have evidence that the abundant online records that could help to fill these spatial gaps have large taxonomic uncertainty that cannot be fixed with taxonomic validation procedures, thus limiting their use to the regions with low expected taxonomic uncertainty. The MixOcc database compiled here, provides a good balance between taxonomic certainty and geographic coverage. However, there are persistent geographical gaps that need to be properly addressed, particularly in France and Ireland, and to a lesser extent in the UK and several eastern European countries (Fig. 4). This means that despite the perception of public funders that biodiversity inventories are no longer needed, especially in developed and extensively studied parts of the world such as Europe3,10,60, field inventories are still needed, even for highly conspicuous plants such as ivies.
To advance on the challenges ahead with the biodiversity inventory of Hedera in Europe, we propose combining scientific botanical collections of vegetative branches to increase the representation of ivies in European herbaria with the improvement of citizen science procedures for Hedera observations. We encouraged (1) carrying a 10x (preferably 20x) magnifying glass to examine the microscopic but essential features for ivy species diagnosis (trichomes), and (2) always including photographic records. The photographic records must include good pictures of (2a) the trichomes (taken on the underside of a vegetative leaf and using the magnifying glass), (2b) the general aspect of the vegetative branches so that we can examine the phenotypic variation of the leaves, and (2c) a detail of a vegetative leaf representing the shape that predominates in the individual.