Gabonius ngouniensis (Pellegr.) Wieringa & Mackinder, 2013

Wieringa, Jan J., Mackinder, Barbara A. & Van Proosdij, André S. J., 2013, Gabonius gen. nov. (Leguminosae, Caesalpinioideae, Detarieae), a distant cousin of Hymenostegia endemic to Gabon, Phytotaxa 142 (1), pp. 15-24 : 17-21

Gabonius ngouniensis (Pellegr.) Wieringa & Mackinder comb. nov. ( Fig. 1 )

Hymenostegia ngouniensis Pellegrin (1942: 247) . Type:— GABON. Ngounié: between Dibwangui and Issala, Le Testu 5284 (lectotype P!, isolectotype: B!, BR, IFAN!, K!), lectotype selected by Léonard (1951: 441).

Shrub, or more commonly a tree 1–20 (–35) m tall, dbh 25 cm (12 m tree)– 40 cm (20 m tree); bark yellowish grey, or greenish brown, peeling (J.J. de Wilde 9328) with lenticels. Slash yellow to ochre, rather fibrous, up to 12 mm thick. Sapwood cream-coloured. Twigs reddish brown to dark brown, sparsely to moderately golden-brown puberulous, hairs hooked (only visible at × 100 or greater magnification), lenticels pale. Stipules in pairs, free but touching at base (intrapetiolar), caducous, triangular to ovate, 1–4.5 × 0.5–1.5 mm, golden-brown to white tomentose, especially over the midvein, lateral areas less dense to glabrous, but margin pubescent again, apex acute. Bud scales absent. Leaves paripinnate, (2–)5–7-jugate, most commonly with 5 or 6 leaflet pairs spaced widely along the rachis, the distance between nodes exceeding the width of the leaflets borne at the nodes, lower pairs smaller than upper pairs, distal or penultimate pair the largest; petiole 0.6–2.1 cm long, rachis 1.7–12.8 cm long, striate, moderately golden-brown puberulous, hairs hooked (only visible at × 100 or greater magnification), leaflets subsessile, narrowly elliptic to elliptic, ovate to sub-rhombate, falcate, upper leaflet pair 3.1–12.2 × 1.4–5.6 cm, adaxial surface glabrous, abaxial surface mostly glabrous or sparsely hairy, patch of hairs sometimes present at the base, margins often ciliate towards the base, mid-vein sub-central, the proximal half of the leaflet slightly larger, distal margin sometimes angular, proximal margin usually rounded, apex acuminate, base asymmetric, proximal half with 0–4 glands, distal half with 4–17 glands, the lowest numbers in the basal leaflets. Inflorescence a lax 7–18-flowered terminal or axillary raceme, sometimes branching once near base or two inflorescence axes arising at the same point, axis 7.5–15 cm long, including peduncle 1.5–5 cm long, moderately golden brown puberulous, hairs hooked (only visible at × 100 or greater magnification), bracts caducous, not seen in herbarium, pale green when young (Wieringa 2413); pedicel 15–27 mm long (at anthesis), moderately puberulous, hairs hooked (only visible at × 100 or greater magnification); bracteoles opposite, persistent, borne at the apex of the pedicel, directly below the hypanthium, petaloid, broadly ovate, 8–20 × 7–13 mm, when mature white or pinkish, deep red or purple spot at base, sometimes red-veined and with red or pale purplish margins, puberulous on both surfaces, more densely in the central area of the abaxial surface, margins ciliate; hypanthium 6–9 mm long, inside glabrous, outside glabrous or with some sparse hairs. Sepals pink, reddish brown, greenish purple outside, brownish green or purple brown inside, reflexed after anthesis, slightly longer than the hypanthium, puberulous towards the base on the inside, otherwise glabrous. Petals 5, glabrous, adaxial and lateral ones similar in size, 10–13 × 5–6 mm, yellow with white lower margins, a small purple basal-central spot and a pale claw when first in flower, turning red with age, claw of lateral petals very narrow, abaxial petals smaller, c. 4 × 1 mm, white. Stamens 10, filaments white or pink, free, anthers deep cream, pale pinkish or purplish brown, connective purple-grey (Wieringa 4493). Ovary 2–3 ovulate, stipitate, the stipe fused along most of its length to the adaxial sidewall of the hypanthium, green, orange-red or brownish with dark red sutures, hairs c. 0.75 mm long, white, sparse on the faces but dense along the margins, the marginal hairs persisting into young fruit and extending along the lower half of the style, stigma peltate, pistil white or greenish-white. Pod compressed, glabrous, dull greyish green or greenish blue outside, light brown to medium brown inside, 23–28 × 6.5–8.8 cm, trapeziform, broadest (“height”) about one-third distance from the apex, lower margin rounded, upper suture not broadened into wings, beak c. 5 mm long on immature pods, remnant of base only seen on mature pods, valves revolute after dehiscence. Seeds 1–2, discoid, c. 3 cm diameter (Wieringa 4493). Seedlings: (based on J.J.F.E. de Wilde 9329), first leaf pair opposite, possibly reduced (see notes), subsequent leaves are alternate, leaflets sessile, glabrous, in 2–4 pairs, largest leaflets at apex of the leaf 60–72 × 18–25 mm, very similar in appearance to adult foliage.

Habitat and Ecology:—Primary and secondary evergreen forest: elevation sea level– 510 m. Populations seem to flower synchronously (van der Burgt 1), flowering recorded from November to July but most commonly in April.

Distribution:—Endemic to Gabon, occurring in the western part of the country from the border with Equatorial Guinea in the north to that with Congo (Brazzaville) in the south (fig. 2).

Since we have chosen to name this genus after Gabon, we decided to test the likelihood that the taxon distribution is indeed limited to Gabon. To assess this we generated a Species Distribution Model (SDM). For generating SDM’s based on presence-only data as in this case, MaxEnt has been documented to outperform other methods ( Elith et al. 2006). As the foreseen distribution is limited to Gabon and possibly to neighbouring countries, we used a 30 arc-sec resolution and applied the analysis to a study area ranging from 8º N to 8º S and from 6 to 24º E, thus roughly encompassing the countries of Gabon, Equatorial Guinea, Cameroon, Democratic Republic of Congo, and Sao Tomé & Principe. The collecting localities of all herbarium vouchers examined in this study were used as input species data. Environmental layers used for model building and projection include 19 BIOCLIM variables ( Hijmans et al. 2005). In addition, the range as well as standard deviation of altitude based on SRTM DEM 90 × 90 m data, obtained through (<srtm.csi.cgiar.org>, accessed 07-10-2011) within each grid cell was calculated and added as variables to act as a proxy for ruggedness of the landscape. Finally, soil parameters of the dominant soil type within each grid cell were extracted from the Harmonised World Soil Database (<http://webarchive.iiasa.ac.at/Research/LUC/ External-World-soil-database/HTML/index.html?sb=1> accessed 16-08-2012). Environmental parameters were checked for multi-collinearity separately for climatic and altitude data as well as for soil data. To avoid overfitting and errors due to multi-collinearity, only uncorrelated parameters were used for model building (Pearson r <0.65 or Spearman Rho <0.65). This resulted in the selection of the following parameters: temperature annual range (BIO7), mean temperature of coldest quarter (BIO11), annual precipitation (BIO12), precipitation seasonality (BIO15), precipitation of warmest quarter (BIO18), precipitation of coldest quarter (BIO19), altitude range (DEM-range), available water capacity range (AWC_CLASS, categorical variable), topsoil bulk density (T_BULK_DENSITY), topsoil calcium carbonate (T_CACO3), topsoil salinity (T_ECE), topsoil sodicity (T_ESP), topsoil gravel content (T_GRAVEL), topsoil organic carbon (T_OC), topsoil pH H2O (T_ PH _H2O), topsoil sand fraction (T_SAND), topsoil base saturation (T_BS) and topsoil cation exchange capacity of the clay fraction (T_CEC_CLAY, categorical variable). Models were built using MaxEnt V.3.3.3.k ( Phillips et al. 2004) applying the default features. In addition a presence / absence distribution map was generated applying a 10 percentile trainings threshold, meaning that 10% of the training locations are allowed to fall outside the predicted distribution, which is assumed to correct for errors in identification and estimation of locations.

The final model was generated using 41 presence records; the result is shown in fig. 3. The Area Under the Curve (AUC) of our model has a value of 0.990, which lies well above the critical threshold of 0.7. As the use of AUC has been highly criticised ( Peterson et al. 2008), we also tested the model against a null model ( Raes & Ter Steege 2007) resulting in a rank number of 100 (out of 100) clearly indicating that the model performs significantly better than random. Based on the predicted potential distribution, Gabonius is nearly endemic for the country of Gabon. The predicted presence on Sao Tomé, as displayed in fig. 3, is not realistic as Sao Tomé is a non-continental island located approx. 250 km off the coastline of Gabon which is impossible to reach for a species with explosive seed dispersal like Gabonius ngouniensis . Moreover, this island is of volcanic origin, with phosphorus-rich soils, which in general are not favourable for Detarieae as discussed above. The only place where Gabonius might occur outside of Gabon is just over the border in Equatorial Guinea; so far the species has not been found in that area.

Conservation:— Gabonius ngouniensis is widespread in Gabon with an A00 of 38137 km 2 (cell size 46 km) and an E00 of 72300 km 2 and occurs within the borders of several National Parks. It is assessed here as Least Concern (LC) according to the criteria of IUCN (2001).

Notes:—Small scars observed on the seedling axis suggest the abscission of a reduced first pair of opposite leaves. At the point of abscission, the epicotyl continues without any signs of transformation into a first stem, suggesting growth is not paused when the reduced first pair of leaves appears. As such this represents a transition between two commonly observed seedling morphologies in Detarieae ( Léonard 1957) . They are (i) seedlings that possess a clear epicotyl and an opposite first leaf pair indicating a pause in growth while this first pair fully develops, for example Talbotiella korupensis Mackinder & Wieringa ( Mackinder et al. 2011: 411) and (ii) seedlings that immediately start producing a shoot with alternative leaves, for example Annea afzelii . This apparent presence of a reduced first leaf pair in Gabonius seedlings represents a (so far) unique morphology in Detarieae . Functionally we consider the seedlings to belong to type (ii) since growth continues beyond the first reduced leaves. We think it unlikely the reduced leaves are functional given they do not persist (see also Mackinder et al. 2013c).

Bark with Green bean-odour (Dibata 132). Freshly dried material has a “nut-like” smell (van der Burgt 1). Several collectors note the flowers are fragrant.

The epithet was written as “ngouniensis” by Pellegrin in 1942, but in a later publication that is much more widely available ( Pellegrin 1949: 90) he spelled it as “ngounyensis”. Since Aubréville (1968a) in his flora treatment also used the later spelling, this spelling currently prevails in publications and databases. Since the epithet refers to a geographical name, its spelling should follow that of the original protologue and is not to be corrected.

In the Flore du Gabon treatment, Aubréville (1968a: 102) lists Le Testu 5890 (P) as the holotype. This is incorrect, since the protologue mentions two collections (Le Testu 5284 & Le Testu 5890) but does not indicate a holotype. Moreover, this cannot be seen as a lectotypification since a lectotype had already been selected by Léonard (1951: 441), who selected 5284.

Figures

FIGURE 1. Flowering branch of Gabonius ngouniensis (photographs by L.J.G. van der Maesen) with close up of inflorescence (upper) to show petal development, shape and colour in an opened flower.