Allometric equation for the commonest palm in the Central Congo Peatlands, Raphia laurentii De Wild

The world’s largest tropical peatland lies in the central Congo Basin. Raphia laurentii De Wild. is a palm which forms dominant to mono-dominant stands across ca. 45 % of the peatland area. However, a lack of allometric equation for this canopy-forming trunkless palm with fronds up to 20 m long, means that this it is currently excluded from aboveground biomass (AGB) estimates for these peatland ecosystems. Here we develop an allometric equation for R. laurentii, by destructively sampling 90 R. laurentii stipes (across six mean petiole diameter size classes of: 2-4 cm, 4-5 cm, 5-6 cm, 6-7 cm, >8 cm) in a peat swamp forest, in the Likouala Department, Republic of the Congo. Prior to destructive sampling, the five parameters were measured for each stipe: diameter at stem base, mean diameter of the palm fronds, total diameter of the palm fronds (TDpf), total height and number of palm fronds. After destructive sampling and before weighing each individual was separated into the stem and the following palm frond categories: sheath, petiole, rachis and leaflets. We fitted a linear model relating AGB to each independent predictor variable separately to assess the best variable, finding that palm fronds represented at least 77 % of the total AGB in each diameter classes. We found that TDpf was the best single predictor variable for R. laurentii AGB. We provide a series of equations, based on sampling R. laurentii with fronds ≥2 cm diameter or ≥5 cm diameters. The best allometric equation was: AGB = Exp (−2.691 + 1.425 × ln (TDpf) + 0.695 × ln(H) + 0.395 × ln (WD).. A monodominant 1 ha plot near the harvesting site had a palm AGB of 60.8 Mg ha-1, similar to the AGB of the trees in the same plot, at 86.2 Mg ha-1. Accounting for the AGB of palms is important, and can now be estimated using the allometric equations developed here.


Introduction 4 8
Improved estimates of the carbon stocks and flows in tropical ecosystems is critical The average aboveground biomass of the stem compartment (11.26±2.05 kg) was 2 6 7 lower than that of the palm fronds (44.93±5.36). The allocation of biomass between each between mean petiole diameter classes (Fig 3). Palm frond compartments accounted for more 2 7 0 than 77 % of the aboveground biomass in all mean petiole diameter classes (with the stem 2 7 1 accounting for between 10-23 % of the total biomass). On average the petiole accounted for 2 7 2 the largest proportion of biomass (30.17 %) followed by leaflets (25.17 %), rachis (16 %), increasing mean stipe diameter classes, the proportion of leaflet biomass decreases (Fig 3). sampled.

7 9
Palm frond AGB was significantly higher than the stem AGB in a stipe of R. laurentii 2 8 0 (kruskal.test P-value < 0.001; Fig 4a). AGB increased with increasing mean petiole diameter 2 8 1 class (Fig 4b).  In the analyses where only single predictor variables were used to estimate AGB and 2 8 6 petioles ≥ 2 cm were included in the analysis, TD pf was found to be the best single predictor,  Only WD showed a very weak relationship with AGB. The full list of equations is given in 2 9 2 Table 2. Thus, when only one predictor variable was used, model 1 (m1), using TD pf, was the 2 9 3 best model (Table 2). Yet, the inclusion of other predictor variables (particularly H) 2 9 4 significantly improved the goodness fit of the models, with the exception of N pf or WD ( Table   2  9  5 2). In fact, the combination of TD pf with H increased the R 2 adj and significantly decreased the 2 9 6 AIC, RSE, RMSE and Bias, compared to the models which used TD pf in combination with N pf 2 9 7 (m5) or WD (m6). When three predictor variables were used in the linear model, the fit was 2 9 8 slightly improved. The best R 2 adj , AIC and RSE values included the variables TD pf , H and WD, 2 9 9 m11 in Table 2, in the following formula: When only individuals with palm frond diameters greater than or equal to 5cm were 3 0 2 used for model establishment, the TD pf was the best predictor variable, followed by the N pf , best model using more than one parameter is m18 (Table 2). However, m11 and m18 tend to underestimate AGB for largest R. laurentii individuals (Fig 6). Mean relative error on AGB ( The fastest and most cost-effective field method for estimating the AGB of R. laurentii 3 1 2 stipes is counting palm fronds and using the single parameter equation, N pf , model 5 or model  The contribution of R. laurentii individuals to the total above-ground biomass estimate was 3 1 7 8% and 41% for the tree-dominated EKG-02 and R. laurentii dominated EKG-03 plots, respectively. The percentage difference in biomass estimation between total above-ground 3 1 9 biomass and above-ground biomass of trees was 9% and 71%, for the tree-dominated and R.
18.44 -1.24 of AGB between the stem and the palm fronds of R. laurentii was contrary to that observed Schumach. & Thonn., Areca cotechu L., which also makes it possible to estimate the stem leaflets also varies from one palm species to another. In this study, the biomass of the petioles

Discussion
and leaflets were greater than that of the rachis, which is contrary to the findings of [34] for Elaeis guineensis, where the biomass of the rachis (13.53%) was greater than that of the 3 6 5 petioles (7.95%) and leaflets (6.02%). This difference can be attributed to the length and  We found that the best estimates of R. laurentii AGB were obtained when TD pf , H 3 7 9 and WD were included as variables in the model (m11; We therefore recommend, that when height data is available, m11, which uses TD pf , H and WD as variables, is used to estimate R. laurnentii AGB. Rather than sampling WD in 3 9 1 the field, users of our allometric equation could use the WD values we provide in Table A1.
However, owing to the tendency of m11 to underestimate the AGB of larger R. laurentii 3 9 3 individuals, we recommend that m18, which includes the same variables but includes only 3 9 4 stipes with palm frond diameters ≥ 5cm, is used for larger individuals.

9 5
The increase in AGB estimates for the two forest plots used in this study, particularly 3 9 6 the palm-dominated plot, when estimates of R. laurentii biomass were included using m16, 3 9 7 highlights the importance of including palms in aboveground biomass estimates. All the   V  a  h  e  d  i  A  A  ,  M  a  t  a  j  i  A  ,  B  a  b  a  y  i  -K  a  f  a  k  i  S  ,  E  s  h  a  g  h  i  -R  a  d  J  ,  H  o  d  j  a  t  i  S  M  ,  D  j  o  m  o  A  .  A  l  l  o  m  e  t  r  i  c  e  q  u  a  t  i  o  n  s  5  0  6  f  o  r  p  r  e  d  i  c  t  i  n  g  a  b  o  v  e  g  r  o  u  n  d  b  i  o  m  a  s  s  o  f  b  e  e  c  h  -h  o  r  n  b  e  a  m  s  t  a  n  d  s  i  n  t  h  e  H  y  r  c  a  n  i  a  n  f  o  r  e  s  t  s  o  f  I  r  a  n  .  5  0  7  J  F  o  r  S  c  i  .  2  0  1  4  ;  6  0  :  2  3  6  -2  4  7  .  d  o  i  :  1  0  .  1  7  2  2  1  /  3  9  /  2  0  1  4  -J  F  S  5  0  8   2  8  .  F  o  n  t  o  n  N  H  ,  M  e  d  j  i  b  é  V  ,  D  j  o  m  o  A  ,  K  o  n  d  a  o  u  l  é  J  ,  R  o  s  s  i  V  ,  N  g  o  m  a  n  d  a  A  ,  e  t  a  l  .  A  n  a  l  y  z  i  n  g  A  c  c  u  r  a  c  y  5  0