Plant Biodiversity Drivers in Brazilian Campos Rupestres: Insights from Phylogenetic Structure

Abstract

Old, climate-buffered infertile landscapes (Ocbils) have attracted increasing levels of interest in recent years because of their exceptionally diverse plant communities. Brazil's campos rupestres (rupestrian grasslands) are home to almost 15% of Brazil's native flora in less than 0.8% of Brazil's territory: an ideal study system for exploring variation in floristic diversity and phylogenetic structure in sites differing in geology and phytophysiognomy. We found significant differences in floristic diversity and phylogenetic structure across a range of study sites encompassing open vegetation and forest on quartzite (FQ) and on ironstone substrates, commonly termed canga. Substrate and physiognomy were key in structuring floristic diversity in the Espinhaço and physiognomy was more important than substrate in structuring phylogenetic diversity, with neither substrate nor its interaction with physiognomy accounting for significant variation in phylogenetic structure. Phylogenetic clustering was significant in open vegetation on both canga and quartzite, reflecting the potential role of environmental filtering in these exposed montane communities adapted to multiple environmental stressors. In forest communities, phylogenetic clustering was significant only at relatively deep nodes of the phylogeny in FQ while no significant phylogenetic clustering was detected across forest on canga (FC), which may be attributable to proximity to the megadiverse Atlantic forest biome and/or comparatively benign environmental conditions in FC with relatively deep, nutrient-rich soils and access to edaphic water reliable in comparison to those for open vegetation on canga and open or forest communities on quartzite. Clades representing relatively old lineages are significantly over-represented in campos rupestres on quartzite, consistent with the Gondwanan Heritage Hypothesis of Ocbil theory. In contrast, forested sites on canga are recognized as Yodfels. To be effective, conservation measures must take account of the distinct communities which are encompassed within the broad term campos rupestres, and the differing vulnerabilities of Ocbils and Yodfels.

Keywords: campo rupestre; canga; eastern Brazil; phylogenetic clustering; rupestrian grasslands; substrate; vegetation types.

FIGURE 1

Campos rupestres with different substrates in Minas Gerais. (a) Landscape of the Serra do Cipó, Mun. Santana de Pirapama, showing different open quartzite (OQ) vegetation types; (b) OQ - close-up of ‘cerrado-rupestre’ with Eriocaulaceae in the foreground at the Serra do Cipó; (c) forest on quartzite (FQ) - view of a forest grove in a matrix of open vegetation, Serra do Cipó; (d) FQ - Inside a riverine forest, Serra do Cipó; (e) FC - forest formation on canga at the Serra da Gandarela, Mun. Rio Acima; (f) open vegetation on canga (OC) - view of canga field at the Serra de Capanema, Mun. Catas Altas (photos a–d William Milliken; e–f Pedro L. Viana).

FIGURE 2

Map showing the geographical location of the sites included in our analysis. Each site is classified according to the substrate and physiognomy of the vegetation.

FIGURE 3

The “Espinhaço megatree” indicating vegetation and substrate affinities for the 2920 species. Plant families mentioned in the results and discussion are highlighted.

FIGURE 4

Dendrogram showing results of Ward hierarchical clustering of 47 sites in the Espinhaço range.

FIGURE 5

Non-metric multidimensional scaling (NMDS) ordination of 47 sites in the Espinhaco range. Bray–Curtis distance, final stress = 0.23. Ellipses show 95% confidence limits for delimitation of each group.

FIGURE 6

Values of MPD, MNTD and their standardized metrics comparing the observed values against a null model: the NRI and NTI, respectively. Individual sites (dots in the boxplots) with values of NRI and NTI above +1.96 (gray lines) are statistically phylogenetically clustered.

FIGURE 7

Phylogenetic representations of the “Espinhaço megatree” summarizing over-represented and under-represented clades for each of the four vegetation:substrate types. Clades over-represented in a particular habitat type are shown in red, while under-represented clades are highlighted in green. Images of selected taxa for over-represented (framed in red) and under-represented (framed in green) clades are placed adjacent to each phylogeny as follows: forest on canga – Chamaecostus subsessilis (top left, photo by B. Klitgaard), Solanum didymum (bottom left, photo by W. Milliken), Rhynchospora exaltata (top right, photo by W. Milliken), Serjania paradoxa (middle right, photo by D. Zappi), Myrcia venulosa (bottom right, photo by W. Milliken); open on canga – Paepalanthus erectifolius (top left, photo by D. Zappi), Eremanthus incanus (bottom left, photo by P. L. Viana); forest on quartzite – Paepalanthus planifolius (top left, photo by W. Milliken), Vellozia glabra (bottom left, photo by W. Milliken), Panicum sellowii (top right, photo by W. Milliken), Trembleya laniflora (bottom right, photo by D. Zappi); open on quartzite – Cephalostemon riedelianum (left, photo by W. Milliken), Paepalanthus comans (top right, photo by D. Zappi), Banisteriopsis malifolia (bottom right, photo by W. Milliken).