Journal of Ecology

doi: 10.1111/j.1365-2745.2011.01829.x

Environmental filtering of dense-wooded species controls above-ground biomass stored in African moist forests Sylvie Gourlet-Fleury1*, Vivien Rossi1, Maxime Rejou-Mechain1, Vincent Freycon1, Adeline Fayolle1, Laurent Saint-Andre´2,3, Guillaume Cornu1, Jean Ge´rard4, Jean-Michel Sarrailh1, Olivier Flores5, Fide`le Baya6, Alain Billand1, Nicolas Fauvet1, Michel Gally7, Matieu Henry2,8,9, Didier Hubert1, Alexandra Pasquier1 and Nicolas Picard10 1 Cirad, UR B&SEF, Biens et Services des Ecosyste`mes Forestiers tropicaux, Campus International de Baillarguet, TA C-105 ⁄ D, F-34398, Montpellier, France; 2UMR Eco&Sols, Ecologie Fonctionnelle & Bioge´ochimie des Sols & Agroe´cosyste`mes, Montpellier SupAgro ⁄ CIRAD ⁄ INRA ⁄ IRD, 2 Place Viala, Baˆt 12, F-34060, Montpellier, France; 3 Inra, UR BEF, Bioge´ochimie des Ecosyste`mes Forestiers, F-54280, Champenoux, France; 4UR Production et valorisation des bois tropicaux et me´diterrane´ens, TA B-40, F-34398, Montpellier, France; 5UMR PVBMT, Universite´ de la Re´union ⁄ CIRAD, 7 chemin de l’IRAT, F-97410, Saint-Pierre, France; 6MEFCP ⁄ ICRA, BP 830, Bangui, Re´publique centrafricaine; 7FRM, 60 Rue Henri Fabre, F-34130 Mauguio, France; 8GEEFT, AgroParisTech, F-34086, Montpellier, France; 9Laboratorio di Ecologia Forestale, Universita` degli Studi della Tuscia, 01100 Viterbo, Italy; and 10 CIRAD, BP 4035, Libreville, Gabon

Summary 1. Regional above-ground biomass estimates for tropical moist forests remain highly inaccurate mostly because they are based on extrapolations from a few plots scattered across a limited range of soils and other environmental conditions. When such conditions impact biomass, the estimation is biased. The effect of soil types on biomass has especially yielded controversial results. 2. We investigated the relationship between above-ground biomass and soil type in undisturbed moist forests in the Central African Republic. We tested the effects of soil texture, as a surrogate for soil resources availability and physical constraints (soil depth and hydromorphy) on biomass. Forest inventory data were collected for trees ‡20 cm stem diameter in 2754 0.5 ha plots scattered over 4888 km2. The plots contained 224 taxons, of which 209 were identified to species. Soil types were characterized from a 1:1 000 000 scale soil map. Species-specific values for wood density were extracted from the CIRAD’s data base of wood technological properties. 3. We found that basal area and biomass differ in their responses to soil type, ranging from 17.8 m2 ha)1 (217.5 t ha)1) to 22.3 m2 ha)1 (273.3 t ha)1). While shallow and hydromorphic soils support forests with both low stem basal area and low biomass, forests on deep resource-poor soils are typically low in basal area but as high in biomass as forests on deep resource-rich soils. We demonstrated that the environmental filtering of slow growing dense-wooded species on resource-poor soils compensates for the low basal area, and we discuss whether this filtering effect is due to low fertility or to low water reserve. 4. Synthesis. We showed that soil physical conditions constrained the amount of biomass stored in tropical moist forests. Contrary to previous reports, our results suggest that biomass is similar on resource-poor and resource-rich soils. This finding highlights both the importance of taking into account soil characteristics and species wood density when trying to predict regional patterns of biomass. Our findings have implications for the evaluation of biomass stocks in tropical forests, in the context of the international negotiations on climate change.

*Correspondence author. E-mail: sylvie.gourlet-fl[email protected]  2011 The Authors. Journal of Ecology  2011 British Ecological Society

2 S. Gourlet-Fleury et al.

Key-words: basal area, Central African Republic, determinants of plant community diversity and structure, life-history strategy, soil fertility, species sorting, vital rates, water reserve, wood density

Introduction Tropical forests play a major role in the global carbon cycle as they contain 45% of terrestrial vegetation stocks and their loss, through deforestation, contributes substantially to anthropogenic emissions of greenhouse gases (Intergovernmental Panel on Climate Change 2000). In the near future, countries with tropical forests could benefit from incentives designed to help reduce emissions from deforestation and degradation (REDD initiative, UN Framework Convention on Climate Change). To benefit from these incentives, they will need to demonstrate that they have reduced carbon emissions, which will require that they have detailed and accurate knowledge of the quantities of carbon stocks in standing forests and soils, and carbon fluxes that result from various land use activities and natural ecosystem processes. Current estimates of above-ground biomass in tropical forests are highly inaccurate at large spatial scales (Houghton 2005) primarily because forest biomass stocks are difficult to quantify (Fearnside 1985; Brown, Gillespie & Lugo 1989; Phillips et al. 1998) and are usually estimated by extrapolating regional patterns from a limited number of plots (Brown et al. 1995; Houghton et al. 2001; Lewis et al. 2009). The extrapolation process raises a number of methodological and technical problems (Chave et al. 2004; Saatchi et al. 2007; Grainger 2008; Nasi et al. 2008). In particular, sampling designs seldom account for the environmental factors underlying biomass variations (Gibbs et al. 2007), e.g. soil fertility and topography, which also partly drive spatial patterns of deforestation (De Castilho et al. 2006; Paoli, Curran & Slik 2008). Extrapolating stocks and losses from deforestation while ignoring these environmental factors results in biases, which underscores the importance of understanding these relationships. Only few studies investigated the relationship between soil fertility and biomass in tropical moist forests and they have yielded conflicting results. At the regional scale (‡104 km2), above-ground biomass of Amazonian old-growth forests was found to be higher in wet regions with infertile soils (Malhi et al. 2006), while in Bornean forests, above-ground biomass was strongly positively correlated with soil fertility (Slik et al. 2010). At the local (W) < 0.001 for 100% of the 1000 bootstrap replications), with highest Gp and AGBp values on deep resource-rich soils (Luvic Phaeozems), and lowest values on hydromorphic soils (Stagnosols) (Table 1 and Fig. 2a). Post hoc comparisons revealed different Gp and AGBp patterns of variation according to soil type. Plot basal area values were the highest on deep resource-rich soils while they were the lowest on deep resource-poor and physically constrained (shallow or hydromorphic) soils (P < 0.05, Table 1). In contrast, AGBp was higher on both the deep resource-rich and resource-poor soils than on the physically constrained soils (P < 0.05; Table 1). When considering deep soils only, higher resources availability did not systematically result in higher biomass. Hence, while deep resource-poor soils (Arenic Acrisols) carry a low basal area, similar to that of physically constrained soils, they carry a high biomass, similar to that of deep resource-rich soils. These results were not challenged by uncertainties associated with AGBp estimates (Table S4).

 2011 The Authors. Journal of Ecology  2011 British Ecological Society, Journal of Ecology

Environmental filtering and tree biomass 5 Table 1. Soil types, number of plots and main structural characteristics of stands (trees ‡20 cm d.b.h.). Significant differences at the a = 0.05 level among soil types are indicated by different lower case letters (Dunnett’s modified Tukey–Kramer post hoc test) Deep resource-rich

Soil type*

Luvic Phaeozems

Deep resource-poor Arenic Ferralsols or Acrisols Nitisols (Valleys)

Acrisols

Number of plots 324 329 489 N (ha)1) 135 ± 5 138 ± 5 129 ± 4 Gp (m2 ha)1) 22.3b ± 0.8 21.3b ± 0.8 21.4b ± 0.6 AGBp† (t ha)1) 273.3c ± 10.5 258.2bc ± 10.2 250.8b ± 7.9

Physically constrained

Arenic Acrisols (Plateaus)

Skeletic Acrisols

Petroplinthic Acrisols

Stagnosols

136 895 356 124 101 133 ± 8 131 ± 3 123 ± 5 121 ± 7 123 ± 9 18.9a ± 1.0 18.3a ± 0.3 18.9a ± 0.6 18.5a ± 1.0 17.8a ± 1.4 258.7bc ± 16.1 252.7b ± 5.1 227.9a ± 7.4 224.7a ± 12.7 217.5a ± 17.6

(b)

180 16

18

20

22

24

Gp (m2 ha–1)

ArAcriV

Acri FerNit

ArAcriP

240

260

280

Deep resource-rich soils Deep resource-poor soils Physically constrained soils

220

SkelAcri

LuPhae

SkelAcri PetAcri Stag

200

PetAcri Stag

Acri FerNit

180

220

240

ArAcriV ArAcriP

AGBp (t ha–1)

260

LuPhae

200

AGBp (t ha–1)

280

300

(a)

300

*Soil types follow the WRB classification (2006) and were based on Boulvert (1983, 1996) descriptions (see Table S1 for details). N, number of trees ‡20 cm d.b.h.; Gp, basal area; AGBp, above-ground biomass. Total number of plots: 2754. Values for N, Gp and AGBp correspond to the mean ± t(n)1, a = 0.05) · SD ⁄ n, scaled to 1 ha, with SD, standard deviation of the sample; and t(n)1, a = 0.05), fractile of the Student distribution. †The Welch anova was first performed on the mean estimates of AGBp for each plot and was then repeated for 1000 samples of 2754 AGBp values (bootstrap, see the Materials and methods section). e[qs] and v[qs] were estimated from the CIRAD’s database on species wood densities.

Deep resource-rich soils Deep resource-poor soils Physically constrained soils

0.50

0.55

0.60

WDp (g cm–3)

Fig. 2. Above-ground biomass (AGBp) against (a) basal area (Gp) and (b) mean wood density (WDp) for each soil type. LuPhae, Luvic Phaeozems; Acri, Acrisols; FerNit, Ferralsols ⁄ Nitisols; ArAcriV, Arenic Acrisols in valleys; ArAcriP, Arenic Acrisols on plateaus; SkelAcri, Skeletic Acrisols; PetAcri, Petroplinthic Acrisols; Stag, Stagnosols. Dots are mean values for AGBp, Gp and WDp by soil type. Error bars correspond to ±t(n)1, a = 0.05) · SD ⁄ n, where n = sample size, SD, standard deviation of the sample, and t(n)1, a = 0.05), fractile of the Student distribution, assuming a normal distribution of errors for all variables.

The compensating effect observed on resource-poor soils results from trees tending to have a higher species wood density on these soils than on any other soil type (Fig. 2b, Welch anova on mean WDp values: W = 330.2, P < 0.001). Disregarding variation in wood density across species by considering a mean value [e(qs) = 0.586, see Table S2] would have resulted in overestimating AGBp by 12% (Stagnosols) to 20% (Ferralsols ⁄ Nitisols) on deep resource-rich and physically constrained soils, and slightly underestimating AGBp on deep resourcepoor soils.

EFFECT OF SOIL TYPE ON FLORISTIC COMPOSITION AND SPECIES STRATEGY

Soil type greatly affected both the floristic and functional composition of tree communities, demonstrating that soil type filters species according to their life-history strategy. We

explored the effect of soil type on floristic composition with a NSCAIV. The first ordination axis (NSCAIV1, 76.3% of explained variance) revealed a marked separation between deep resource-poor soils (Arenic Acrisols on plateaus and, to a lesser extent, in valleys) and all the other soil types (Fig. 3). This separation is mainly due to contrasting distribution patterns across soil types of: (i) species such as Staudtia kamerunensis, Prioria oxyphylla, Entandrophragma cylindricum, Celtis mildbraedii, Pycnanthus angolensis, Lophira alata, Petersianthus macrocarpus, Blighia welwitschii and Irvingia excelsa whose specific basal area (Gs) accounted for 26% and 20% of Gp on Arenic Acrisols on plateaus and in valleys, respectively, and for less than 14% on other soil types and (ii) species such as Triplochiton scleroxylon, Ceiba pentandra, Mansonia altissima, Albizia adianthifolia, Ricinodendron heudelotii, Anonidium mannii, Celtis philippensis, Sterculia tragacantha and Duboscia macrocarpa whose Gs accounted for only 8% and 9% of Gp

 2011 The Authors. Journal of Ecology  2011 British Ecological Society, Journal of Ecology

6 S. Gourlet-Fleury et al. 1.0

wood density and high growth rate tended to be associated with the other soil types (high NSCAIV1 scores). In support of this finding, wood density value and growth rate in M. mabokee¨nsis and T. superba (Table S5) were consistent with their distribution across soil types.

0.0

Acri Stag ArAcriV

PetAcri

Discussion

ArAcriP

In this paper, we disentangled how the observed variation in basal area and above-ground biomass across undisturbed African moist forests is driven by soil type, and we evidenced the effects of soil type on species traits such as wood density.

LuPhae

–1.0

–0.5

NSCAIV2 (12.5%)

0.5

SkelAcri

FerNit Deep resource-rich soils Deep resource-poor soils Physically constrained soils

–1.0

–0.5

AGBP VALUES IN THE STUDY AREA

0.0

0.5

1.0

NSCAIV1 (76.3%) Fig. 3. Correlation circle of the Non-Symmetrical Correspondence Analysis on Instrumental Variable (NSCAIV) analysis (first two axes; percentage of explained variance between brackets) showing the marked opposition between deep nutrient-poor soils and the other soils along the first axis. The Plot · Species table, based on species basal area values, was regressed against the Plot · Soil types table (eight modalities). LuPhae, Luvic Phaeozems; Acri, Acrisols; FerNit, Ferralsols ⁄ Nitisols; ArAcriV, Arenic Acrisols in valleys; ArAcriP, Arenic Acrisols on plateaus; SkelAcri, Skeletic Acrisols; PetAcri, Petroplinthic Acrisols; Stag, Stagnosols. Manilkara mabokee¨nsis and Terminalia superba were omitted from the analysis.

on Arenic Acrisols on plateaus and in valleys, respectively, and for more than 19% of Gp on other soil types. This trend was further strengthened by the spatial pattern of two abundant species that we excluded from the NSCAIV analysis (see Materials and methods). These species clearly discriminated Arenic Acrisols (Manilkara mabokee¨nsis: Gs ‡9% of Gp on Arenic Acrisols and