Desert ecosystems are characterised by a patchy distribution of resources. Nutrient sinks associated with landscape modulators (trees) differ markedly from the resource-poor interpatch matrix. Fauna can also act as landscape modulators, modifying patch dynamics by redistributing resources via 'ecosystem engineering’. In semi-arid woodlands, malleefowl (Leipoa ocellata: Megapodiidae) reconfigure surface characteristics by scavenging leaf litter to construct large incubation mounds. The extent to which this movement of resources creates a novel patch and alters extant patches is largely unknown. Ecosystem engineering effects by megapodes have been little studied, but are potentially great, particularly in drylands, where mammalian engineers are known to enhance ecosystem function and drive restoration.

We measured vegetation, ground cover, and soil chemistry at malleefowl mounds and four extant microsites (trees and open areas close to, and far from, mounds) and predicted that: 1) malleefowl mounds would represent enriched, yet novel, microsites; 2) the characteristics of tree and open patches close to the mounds would differ from those away from the mounds, because of the diminishing intensity of disturbance; and 3) effects at tree and open patches close to the mound would be shorter-term, compared to the more substantial high-resource patch formation occurring at the mound, but we expected all effects would diminish with time since malleefowl activity.

We found that: 1) malleefowl mounds were novel microsites with soil chemistry more similar to tree-modulated patches, and groundcover and vegetation variables more similar to the open, interpatch matrix; 2) effects extended to tree and open patches near the mound, but most effects were short lived; and 3) some novel mound attributes (e.g., soil pH, phosphorus, nitrogen, carbon) were greater at mounds, irrespective of their age, while less plant cover and richness on young mounds dissipated with age.

Synthesis: Mound-building megapodes can modulate the distribution of locally derived resources and create a novel microsite. Engineering effects can enhance spatial heterogeneity and ecosystem function over broad spatial and temporal scales, and may assist with ecological restoration, particularly in depauperate, arid systems.

Study system

The study was conducted on Calperum Station in the Murray mallee region, near Renmark, South Australia (-33.659°, 140.563°). Calperum was formerly a pastoral property, but livestock were removed in 1994. The area is on the boundary between arid and semi-arid, rainfall is highly variable and averages 256 mm. The landscape is characterised by a broad sand sheet (the Woorinen Formation; Lawrence et al. 1988), upon which are superimposed low west-east trending sand dunes dominated by mallee (Eucalyptus spp.) woodlands and shrublands communities. The soils are dominated by calcareous earths (Isbell 2021), with abundant carbonate in the profile and therefore high pH levels (Hutton and Dixon 1981). Slopes are generally level to slightly undulating (< 2%).

The dominant overstorey tree species are Eucalyptus socialis, E. oleosa, E. dumosa and E. gracilis. Understorey ranges from sparse Zygophyllum spp. forbs to extensive areas of spinifex (Triodia scariosa) and native shrubs (Senna artemisiodes ssp., Eremophila scoparia and Beyeria opaca). Mallee trees are the dominant biotic patch type in these woodlands and function as resource modulators (Travers and Eldridge 2012). Trees accumulate considerable litter beneath their canopies, providing important habitat for a range of invertebrates (Noble et al. 1996). The accumulated litter layer around the trees acts as ground fuel and facilitates frequent fires in mallee landscapes (Haslem et al. 2011). The density of mallee trees varies markedly, and projected foliage cover of trees and their understorey litter vary with time since last fire. Typically, cover increases rapidly up to about 30% and stabilises about 30 years after fire (Haslem et al. 2011).

Malleefowl mound selection

Malleefowl mounds have been monitored annually at Calperum since 2012 within three adjacent 2 km x 2 km grids (National Malleefowl Monitoring Grids; National Malleefowl Recovery Team, 2016). All mounds within each grid are visited during the breeding season and recorded as active or inactive. Calperum is at the arid end of the range of malleefowl (Jones and Goth 2008). The average density of active and inactive malleefowl mounds at Calperum is approximately five mounds per km² (Heather Neilly, unpublished data, 2018). In August 2018, when malleefowl were preparing their mounds, we selected 12 mounds for study across the three monitoring grids. Four mounds were identified as active within the past 3 years, and four mounds active 3-6 years previously. An additional four mounds older than 6 years were selected, randomly, from a larger pool of mounds known to be more than 6 years old. To sub-sample mounds older than 6 years, any mound with an average rim height of < 10 cm was excluded from selection. Mounds of this height are classified as ‘long unused’ and ‘very degraded’ and generally have no history of ever being active during the time they have been monitored (National Malleefowl Recovery Team 2016).

Microsites, and plant and soil measurements

At each of the 12 mound locations we selected five microsites (Appendix S1, Fig. 1) to test the effects of mounds on biotic and abiotic attributes within the vicinity of the mounds. These microsites were: 1) the mound (Fig. 2a), 2) a Eucalyptus tree closest and within 15 m of the mound (hereafter ‘Tree near’), 3) an open area adjacent to the nearest tree and within 15 m of the mound (‘Open near’), 4) a Eucalyptus tree 50 m from the mound (‘Tree far’), and 5) an open area adjacent to the distant tree (‘Open far’). All measurements were taken from within a 5 m by 5 m quadrat centred upon each microsite. The tree and open microsites near the mound represent the zones of potential malleefowl leaf-raking activity, which has been known to extend to 15 m beyond the mound (Priddel and Wheeler 2003). The distant open and tree microsites would be expected to be outside the zone of malleefowl activity.  

Within each quadrat we assessed groundstorey plant cover (visual estimate from single observer), abundance (number of individuals) and species richness (number of species), and the cover of litter, biocrusts (soil lichens and bryophytes) and bare soil. Ten samples of the surface 10 cm of the soil were collected, mixed, and a subsample taken for chemical analyses after air drying and sieving (< 2 mm). Soils were analysed for total organic carbon (C; using H₂SO₄ added to soil wetted by dichromate solution Cr₂O₇^2-), available nitrogen (NO₃, using 2M KCl extraction, N), available phosphorus (P; Colwell, 1:100 soil:extract) and pH (1:5 soil:water extract).