Observations of a foraging trail of A. lobicornis showed that one group of workers was collecting pieces of dicotyledonous plants whilst another group was cutting and carrying fungal fruit-bodies to the nest (Figure 1). These basidiocarps had grown on the surface of several pats of cow dung (Figure 2) located 50–70 m away from the nest. During a 5-min period, 10 ants entered their nest carrying entire basidiocarps or parts of them. Both immature and fully expanded basidiocarps were collected (see Additional file 1). Further documentation is available from the corresponding author upon request.
Basidiocarps were 10–30 mm long, and fully expanded caps measured 6–10 mm diam. Basidiospores were produced abundantly by mature basidiocarps. They were thick-walled, brown, flattened, somewhat angular in outline, and possessed a basal scar and an apical germ pore (Figure 3). They measured 11.0-13.0 × 7.8-9.1 × 6.6-8.0 μm. Keys of coprophilous fungi (Watling and Gregory 1987Richardson and Watling 1997) permitted identification of the coprophilous fungus as Psilocybe coprophila (Bull.: Fr.) Kumm. [syn. Deconica coprophila (Bull.: Fr.) Karst.]. The ITS rDNA sequence, deposited in GenBank as accession number JX235960, confirmed P. coprophila (accession AJ519795) to be the closest available match, showing a sequence identity at 591 out of 595 nt overlap.
Although A. lobicornis is known to harvest a variety of plants (Franzel and Farji-Brener 2000), fungi have not been described as being part of its collections before. Indeed, to the best of our knowledge the study by Lechner and Josens (2012) is the only previous observation of any leaf-cutting ant species collecting fungal basidiocarps. In that study, fruit bodies of Agrocybe cylindracea (DC) Maire that had grown on the surface of Populus bark in Buenos Aires (Argentina) were collected by Acromyrmex lundii Guérin-Méneville. Further, Lechner and Josens (2012) were able to demonstrate that A. lundii incorporated A. cylindracea basidiocarp material into its fungus gardens under laboratory conditions.
It is not known why Acromyrmex ants should forage on P. coprophila or A. cylindracea, given that they cultivate their Leucoagaricus diet in their nest. Further, although fungal mycelium is a suitable food source in being rich in carbohydrates and proteins (Mueller et al. 2001), few non-leaf-cutting ants seem to have exploited this. Euprenolepis procera Emery from South-East Asian rainforests is the only known ant species specializing in the collection of fungal fruit bodies as the main diet (Witte and Maschwitz 2008).
The nutritional interactions between Leucoagaricus and attine ants are largely unknown, although enzymatic contributions by both symbiotic partners to the degradation and processing of the collected plant material are beginning to be revealed (Richard et al. 2005Silva et al. 2006). In addition, fungus gardens are subject to contamination by microbes originating from soil and plant material (Carreiro et al. 1997Pagnocca et al. 2012). These may be controlled by grooming behaviour and by antibiotics produced by bacteria (Pseudonocardia spp.) colonising the cuticle of attine ants Poulsen and Currie (2010). However, except for the laboratory study by Lechner and Josens (2012) there is no record of the presence of mushroom-type basidiomycetes in fungus gardens other than Leucoagaricus itself.
Our observation of basidiocarp collections by attine ants raises obvious questions relating to the origin of fungiculture. The ‘Consumption First’ model (Weber 1972) postulates that a fungus species which was at first collected and directly consumed by ants might have become a mutualistic symbiont over time, once the ants had become capable of cultivating it and transmitting it to their offspring. More detailed field observations should be conducted to assess the frequency of basidiocarp collection by A. lobicornis in nature. Fungus gardens of basidiocarp-collecting colonies should be analysed for the presence of these basidiomycetes using microbiological or molecular biological methods.