Plant rupicolous. Stem aerial, abbreviated, erect to slightly creeping, brown, internodes short, 0.1 - 0.5 (- 1.0) cm long, 2.0 - 5.0 diam., roots densely covering stem, numerous, white to pale green, thick and succulent, up to 0.8 cm diam. Prophylls and mesophylls 2.0 - 7.0 cm long, ca. 2.0 cm wide, membranaceous, vinaceous green, greyish brown to brown when dried, persistent after decomposing to brown fibres. Leaves ± rosulate; petiole (1.6 -) 2.9 - 17.4 (- 38.9) cm long, 0.2 - 0.9 cm diam., 0.2 - 1.3 times longer than leaf blade, usually deeper than wide, sometimes terete, adaxial side rarely planar, usually canaliculate with channel shallow to deep and broad to narrow, with margins acute to obtuse to narrowly keeled, abaxial side rounded, usually not keeled, basal sheath short, (1.0 -) 1.5 – 3 (- 3.5) cm long; leaf blade (7.0 -) 10.4 - 25.5 (- 75.5) cm long, (2.5 -) 3.5 - 7.4 (- 13.7) cm wide, 2.1 - 4.8 times longer than wide, widest point (= centre of gravity) at (2.6 -) 4.0 - 12.6 (- 18.9) cm from blade base, narrowly oblong, elliptic, ovate or lanceolate, curved to planar in each half, when fresh rigidly coriaceous and brittle and U-shaped to slightly arched in cross-section, base usually obtuse to rounded, sometimes acute, cuneate, truncate or emarginate, apex acute to rounded, sometimes shortly acuminate, or retuse, mucronate, margins planar, adaxial surface dull green, abaxial surface paler and without punctiform glands, finer venation in dried leaves strongly and reticulately prominent but in living plants somewhat obscured and inconspicuous and only slightly impressed on adaxial surface; midvein (primary vein) prominent on both surfaces, adaxially obtuse to rounded and yellowish, abaxially more prominent, ± triangular in cross-section, centrally obtuse to acutely keeled, secondary veins 11 - 16 pairs, not prominent on either surface, poorly differentiated from tertiaries and quaternaries, joining into an inframarginal collective vein 0.1 - 1.3 cm from the blade margin. Inflorescence solitary per leaf, usually overtopping leaves; peduncle (12.4 -) 19.0 - 39.3 (- 58.5) cm long, 0.2 - 0.8 cm diam., green to brown with white to vinaceous points at the apex, terete or with two opposite keels; spathe (1.9 -) 2.8 - 5.3 (- 10.1) cm long, (0.75 -) 1.08 - 1.70 (- 3.6) cm wide, 0.6 - 1.2 times longer than spadix, 1.9 - 4.7 times longer than wide, at anthesis held at 45º - 90º, later usually becoming erect, sometimes clasping the spadix, persistent to fruiting stage, ovate, elliptic, oblong or lanceolate, cymbiform to recurved at acuminate apex, base rounded to emarginate, thickly coriaceous, margin slightly convolute to planar, ventrally purplish brown, vinaceous to yellowish rose, dorsally dark green tinged purplish brown, spathe decurrency acute, obtuse to cuneate, extending for 0.2 - 1.6 cm along peduncle; spadix (2.2 -) 2.8 - 6.1 (- 11.6) cm long, (0.4 -) 0.53 - 0.85 (- 1.4) cm diam., 4.3 - 11.3 times longer than wide, cylindric to conic or rarely clavate, dark, dull purple at anthesis becoming brown afterwards, sessile or with green stipe 0.1 - 2 (- 4.0) cm long. Flowers 0.33 - 0.49 cm long, 0.36 - 0.42 cm wide, relative flower size 2.3 - 3.6 (ratio of spadix and flower width), anthers cream becoming brown when dried, style slightly prolonged, stigma subglobose, ca. 0.1 cm diam., projecting beyond tepals for 0.05 cm. Fruit a berry 0.4 - 1.0 cm long, 0.2 - 0.8 cm wide, broadly obovoid, rounded at apex, green to dark green, whitish at base; seed 0.5 cm long, 0.35 cm wide, oblong-ovate, elliptic to plano-convex in transverse section, yellowish.
Conservation assessment by Andrade et al. (2019):
A spatial analysis was carried out with the GeoCat online tool (Bachman et al. 2011) using 32 collections from 24 different localities. The preliminary conservation status assessments of extent of occurrence (EOO) and area of occupancy (AOO) were based on the geographical coordinates of the specimen records. We used latitude and longitude coordinates from the material given in ESM 2.
When the IUCN default cell width (2 km) was used, A. harleyi was classified as Vulnerable (VU) based on Extent of Occurrence (EOO = 7,074 sq.km) and Endangered (EN) based on Area of Occupancy (AOO = 72 sq.km). With the auto-value cell width (16 km) the classification was Vulnerable (VU) based on Extent of Occurrence (EOO = 7,074 sq.km) and Not Threatened (NT) based on Area of Occupancy (AOO = 2,530 sq.km).
Among the species of sect. Urospadix occurring in the Chapada Diamantina, Anthurium harleyi is most similar to A. erskinei, sharing the following characters: rupicolous habit with the stem wedged in rock crevices, leaf blades simple, thick, rigid, brittle and usually attenuate to rounded at the base, spathes patent to erect. Anthurium harleyi differs most clearly from A. erskinei in the following characters: inflorescence projecting more strongly from the foliage (greater relative inflorescence length), shorter and narrower leaf blade, lower leaf blade centre of gravity, shorter spathe, and relatively larger flowers (fewer flowers per spadix width). Additional distinctive characters of A. harleyi are: midrib yellowish on adaxial side, spathe more strongly patent at anthesis with recurved apex, later becoming erect, spadix dark purple at anthesis, and more northerly distribution within the Chapada Diamantina, suggesting an allopatric pattern in relation to A. erskinei.
Temponi (2006), in an extensive study of Brazilian Anthurium species, provided molecular phylogenetic evidence for the view that the closest relatives of species of sect. Urospadix in the semi-arid interior of Bahia are to be found in the humid Atlantic forest, a suggestion made earlier by Mayo (1978, 1990) on grounds of morphological similarity. Anthurium erskinei and A. cleistanthum G.M.Barroso from Espirito Santo were paired in Temponi's total evidence tree, implying that A. erskinei, and by extension the related A. harleyi, belong to a lineage whose members occur mainly in humid forest. This prompts the conjecture that the evolution of rupicolous species of Anthurium in the Chapada Diamantina adapted to seasonal drought was influenced by cyclic episodes of forest advance and retreat in the interior of northeast Brazil during and after the Pleistocene. Andrade et al. (2007) and Andrade & Mayo (2010) invoked a similar cause to account for genetic patterns in Monstera species in Northeast Brazil.