9. Dietary niches of Andrena species in north-western Africa

Wood (2023) presented pollen load data for understudied West Mediterranean species of Andrena, including data from Morocco. Supplementary data are presented here (Table 1) for species with distributions broadly restricted to North Africa, and which are hence not present in the Iberian Peninsula, with some taxa extending to the Canary Islands, Sicily, and the Levant. A total of 294 pollen loads from 36 species from six countries was analysed. The majority of species for which novel data are presented are oligolectic, or suspected to be oligolectic based on their phylogenetic position and the known dietary specialisation of closely related species. For example, though relatively few pollen loads are presented for species of Chrysandrena, Suandrena Warncke, 1968, Taeniandrena, and Ulandrena Warncke, 1968 in the present work, members of these subgenera are well known to be strongly associated with the botanical families Asteraceae, Brassicaceae, Fabaceae, and Asteraceae, respectively. The proportion of oligolectic or probably oligolectic species is 80.6% (29/36 species), which is extremely high, though this is of course biased because more widespread polylectic species are not considered here; the overall trends in the pollen preferences of Moroccan Andrena at an individual species and at a community level will be analysed in detail in a subsequent work.

However, it is important to note here some interesting findings within the pollen preferences of North African Andrena . Andrena mediovittata Pérez, 1895 (Fig. 37A; Canary Islands, Morocco, and Algeria) is a rare species in collections, and appears to be found in marginal desert edge environments in Morocco and Algeria flying early in the year (March–April), which may explain why it is so infrequently collected. The subspecies A. m. arvensis Warncke, 1968 is found on the Canary Islands. Analysis of pollen showed that A. mediovittata sensu lato collects predominantly from Asteraceae, but also from Brassicaceae in much smaller quantities. Interestingly, of the 83.6% of pollen collected from Asteraceae, three subfamilies were used: Asteroideae (25.6%), Carduoideae (18.9%), and Cichorioideae (39.1%). This use of multiple Asteraceae subfamilies is notable in an Andrena species which collects such a large proportion of its pollen from Asteraceae; most Andrena species specialised on Asteraceae utilise a single subfamily. For example, Andrena isis Schmiedeknecht, 1900 (Fig. 37B; Canary Islands and North Africa to the Levant) collects exclusively from the Asteraceae subfamily Cichorioideae, as is typical for the subgenus Chlorandrena . However, as for the Iberian fauna (Wood 2023), several North African Chlorandrena specialise on the Asteraceae subfamily Asteroideae, here represented by A. boyerella Dours, 1872 (Fig. 37C; north-western Africa and Sicily) and A. sinuata Pérez, 1895 (North Africa). The same division can be seen in the subgenus Ulandrena, where A. speciosa Friese, 1899 (North Africa to the Levant) and A. tadorna Warncke, 1974 (North Africa to the Levant and the Arabian Peninsula) appear to exclusively utilise Asteroideae and Cichorioideae, respectively.

Andrena spolata (Fig. 37D; Canary Islands and North Africa to the Levant) was unsurprisingly found to be oligolectic on Brassicaceae . To date, all known Aciandrena species for which pollen preferences have been assessed have been found to be specialised on this botanical family (e.g., Wood 2023), though A. judaea Scheuchl & Pisanty, 2016 is strongly suspected of being oligolectic on Sedum ( Crassulaceae) (G. Pisanty, in litt.). Given the taxonomic complexity involved in both the classification of the subgenus and the delineation of its constituent species, broad generalisations are limited, but it is here hypothesised that all true Aciandrena species will be found to be specialised on Brassicaceae . Their use of such a common and abundant resource combined with their pattern of local or regional endemism suggests close adaptation to local climatic conditions, or perhaps repeated expansion and shrinking of distributions; further study of this phenomenon is warranted. Another lineage that extensively uses Brassicaceae is the subgenus Truncandrena Warncke, 1968 . Not all members of this subgenus are specialised on Brassicaceae (e.g., some are specialised on Cistaceae; Wood 2023), but it is the most commonly used botanical family; this is well illustrated by A. rufescens Pérez, 1895 (Fig. 37E; Morocco and Algeria), which is found early in the season (March–April) on ruderal species of Brassicaceae .

Andrena marsae Schmiedeknecht, 1900 (Fig. 37F; Morocco, Algeria, and Tunisia) is a poorlyknown species that was recently placed in the small subgenus Ovandrena Wood, 2023 . Surprisingly, A. marsae was found to be polylectic; two other species of Ovandrena, A. oviventris Pérez, 1895 (West Mediterranean) and A. farinosa Pérez, 1895 (Spain only), were found to be oligolectic on Reseda and Fabaceae, respectively. Though A. marsae does collect from Fabaceae (Fig. 37F), it does not appear to be restricted to this botanical family. The dietary niche of the fourth Ovandrena species, A. farinosoides Wood, 2020 (Morocco only), which is known from semi-desert habitats in north-eastern Morocco, remains unclear; a single female was collected during 2022 on Peganum harmala ( Nitrariaceae), but had her scopae empty.

Andrena caesia Warncke, 1974 (Fig. 37G; Morocco, Algeria, and Tunisia) was found to be oligolectic on Fabaceae, as is expected for members of Taeniandrena (Praz et al. 2022). In desert edge habitats it foraged from a variety of herbaceous and shrubby Fabaceae, including Astragalus, Lygos, Ononis, and other Genistae. This broad use of multiple subfamilies within Fabaceae suggests, as for A. prazi sp. nov., that the niche of A. caesia may be temporal, showing adaptation to local environmental conditions rather than to specific Fabaceae resources, within the context of being a lineage of Andrena specialised on this family.

Finally, though most of the North African species studied here were found to be oligolectic, some lineages did display polylecty. This included Andrena succinea Dours, 1872 (Fig. 37H; North Africa to the Middle East), which was recently returned to species status and placed in the newly erected subgenus Pruinosandrena Wood, 2023 (Wood 2023). Nothing has been published on the pollen preferences of members of this subgenus. The limited number of samples available for A. succinea suggest that it is polylectic, though more work is needed to cement this assessment, as well as to determine the niche of the other species of Pruinosandrena .