06-results.md

Results

Our dataset extracted from IPNI comprised $nomenclatural_act_count$ nomenclatural act records published in $publications_total$ different publications between $year_min$ and $year_max$. $dois_from_ipni$ of these had DOIs recorded in IPNI, $dois_from_col$ were backfilled with DOIs recorded in the Catalogue of Life. We found that each year, the IPNI dataset include acts published in an average of $publications_annual_mean$ different publications (n=$number_of_years$). $nomenclatural_act_proportion_in_serial$% of names are published in serials. The spike in numbers of nomenclatural acts published for 2018 was caused by the large volume of combinations ($nomenclatural_act_glovap_count$) published in GLOVAP [@christenhusz_global_2018]. On average each year, less than one quarter ($oa_annual_pc_avg$%) of nomenclatural acts are published open access. We did not observe a trend towards open access throughout the time period analysed (Figure @fig:fig2 (a)). When looking at the kinds of open access used (Figure @fig:fig2 (b)), we find that gold dominates. Here we do find a small difference through time, with green (self-archiving) showing a slightly greater representation in the earlier years in the study. The distribution of nomenclatural acts in different publication titles follows a "long tail" (leptokurtic) pattern. We have visualised the number of nomenclatural acts per publication titles and their open / closed statuses using a more restricted time scale (2019-2021), to better reflect current working practices (Figure @fig:fig3). For legibility, we cut the "long tail" at 80% of the dataset. When analysing the IPNI data, we find that many journals active in the publication of nomenclatural acts are undiscoverable - the containing bibliographic work is not labelled with a DOI (Figure @fig:fig3) - or the acts are labelled with a DOI which is now unresolvable ($dois_unresolvable$ in total, in $publications_w_unresolvable_dois$ different publications).

{#fig:fig2}

{#fig:fig3 height=90% }

Open access takeup by distribution

Africa, South America and South Asia, some of the botanically most diverse areas, have the greatest proportion of undiscoverable nomenclatural acts (Figure @fig:fig4 (i)), whereas Europe, North America and Australia have the lowest proportion. When considering only the discoverable literature using DOIs (Figure @fig:fig4 (ii)), South America and South Asia have the lowest proportion of open publication of nomenclatural acts, though Africa has a higher proportion.

{#fig:fig4}

Availability of digitised type specimen metadata

When considering the taxa with digitised type specimen metadata available online in GBIF, $taxa2gbiftypeavailability.taxa_with_types_available_pc$% of taxa in the WCVP taxonomy have type material available ($taxa2gbiftypeavailability.taxa_with_types_available_count$ of $taxa2gbiftypeavailability.taxon_count$) in the dataset downloaded. Examination of the intersection of the location of the GBIF data provider with the native range of the taxon from the WCVP distribution dataset shows that $taxa2nativerangetypeavailability.continent_code_l1.taxon_represented_pc$% are represented by type material mobilised from within the continent of the native range of the taxon (the lowest level of geographical precision) and $taxa2nativerangetypeavailability.area_code_l3.taxon_represented_pc$% are mobilised to GBIF from within a country of the native range of the taxon (the highest level of precision for the distribution dataset). This is an example of the most species-rich areas being often the most resource-poor [@meyer_multidimensional_2016].