^{1}

^{*}

^{2}

^{2}

^{1}

^{1}

^{3}

Knowledge of the quantity of woody biomass of poplar short-rotation coppice (SRC) on agricultural land is a basic tool for management decisions like rotation length, volume production and the financial balance sheet of economic activities. The expansion of SRC requires a fast, reliable, easily applicable and cheap method for estimating the biomass yield, but existing methods are based on labour-demanding and lengthy measurements of all shoots per tree. The objective of this study was to verify a novel rapid biomass estimation method that uses averaged attributes of only a few largest shoots as a predictor variable for woody biomass in a poplar SRC, hybrid clone J-105 (

Growing trees on agricultural land in innovative production systems like a short-rotation coppice (SRC) or agroforestry systems has been recommended by many experts as one of the most promising methods for not only meeting production goals, but also tackling the main problems of current intensive agriculture and consequences of climate change-soil erosion and biodiversity loss (

The main advantage of SRCs is that they combine sustainable production of renewable biomass with multiple ecosystem services, such as erosion control, increased biodiversity, carbon sequestration and local cooling. The organic cover (leaf fall) in the young plantation (third-to-fourth year) also has a positive effect on the invertebrate community (

Non-destructive methods of biomass yield prediction are important either for preparing the appropriate logistic chains for harvest or deciding the date of harvest. Well managed logistics saves time and money. Allometric models with easily measurable tree variables are the most commonly used tool for the estimation of total above-ground biomass in SRCs (

Clone J-105 is a hybrid between the European black poplar and the East Asian Japanese poplar (

To tackle this issue, we aimed (i) to verify the accuracy of the estimation of woody biomass based on allometric equations using the averaged attributes of a few largest shoots as a predictor variable at the SRC plantations, and (ii) to find the combination of the most accurately predicting shoot parameter and the lowest number of shoots on which such a parameter needs to be measured for biomass estimation.

The study was performed at two existing SRC plantations in the Czech Republic (

Nineteen poplar stumps with shoots (hereinafter referred to as sample trees) with 98 shoots were selected and destructively sampled at Sobechleby in January 2020 and 20 sample trees with 89 shoots at Vícenice in January 2021. The number of sample trees with shoots was chosen based on the data published in other studies dealing with biomass estimation related to SRC of poplar,

All results were analysed and presented as fresh mass of the dendromass (hereafter referred to as biomass).

We measured the stand structure (

Inventory of plantations and sampling of sample trees took place on one day in the week when the plantations were harvested. The wood chips were taken to the heating plant within 14 days after harvest. This short logistics chain was chosen to minimize the variability of changes in the moisture content of the shoots and wood chips.

Samples of wood were taken from ten sample trees to determine the moisture content at both plantations. The wood samples (stem disk about 2 cm thick) were taken from the base, the middle and top parts of the dominant shoots of these trees. The sample tree selection was carried out proportionally to the shoot diameter frequency distribution at both plantations. The wood samples were weighed in the field with the P 221 pocket digital scale, with an accuracy of 0.01 g (Sounon, Wanjiang, China). All wood samples were labelled, transported in paper bags to the laboratory and dried at 80°C until reaching a constant weight and then weighed. Subsequently, the water content of the wood was calculated and expressed as a percentage.

To determine the change in moisture in the wood chips during storage at the depot (where the wood chips are stored until they are taken to the customer), 2 samples of wood chips were taken on the 7^{th} and 14^{th} day of deposition of the wood chips at the depot. Mixed samples of wood chips were taken and weighed on-site.

For each sample tree, the shoots were categorized from the smallest to the largest (thickest) by DBH for the measured shoot parameter (BD, DBH and L). We recorded the value of each parameter of the largest shoot, and calculated the average value of the corresponding parameters of the two largest shoots, then of the three largest shoots, and so forth, until we averaged the values of the parameters of the five largest shoots (hereafter referred to as L_{avg}, DBH_{avg} and BD_{avg}).

The averaged parameter values (L_{avg}, in m; DBH_{avg}, in cm; Bd_{avg}, in cm) were used as independent variables in non-linear models (_{stump}, kg). The exponential and power-law functional forms were used to fit the data. The absolute growth rate is constant for linear models and increases monotonically for exponential and power-law models. The exponential models are frequently used to analyse growth data and may be suitable in the initial growth stage (

where _{stump}), _{avg}, DBH_{avg} and BD_{avg}) for the given number of the largest shoots (1-5) on a stump, while ^{2}, based on the likelihood-ratio (^{2}_{adj} - _{st}, kg), plot- or stand-level (biomass ha^{-1}) estimates (hereafter referred to as stand-level biomass estimates - Biomass_{stand}, in t), we simply summed all the Biomass_{stump} and compared these sums with the sums of the weighed shoot biomasses per sample tree (Biomass_{weighed}).

The differences between the total-model biomass and the total-weighed biomass of all sample trees were determined by calculating the relative differences in the data. The total biomass of the SRC per unit area (1 ha) was calculated by the sum of the individual mean fresh mass weights of all stumps of the inventory plots and recalculated to the total biomass of the inventory plots to a unit area of 1 ha. All results were analysed and presented as biomass.

Analyses were performed in R (

Biomass_{stump} was better estimated by the power-law models than by the exponential models for most of the predictors (_{weighed }(_{stump }estimates based on them varied among the number of averaged largest shoots (_{avg} and L_{avg}, DBH_{avg} was a better predictor due to higher pseudo-R^{2}_{adj} and lower RMSE values (_{avg} increased with each shoot added to DBH_{avg} (^{2}_{adj} was decreasing from the largest to the smallest shoot (^{2}_{adj} values increase after adding the fourth largest averaged shoot in the models with DBH_{avg} alike, as in the models with BD_{avg} and L_{avg }(_{stump }in Sobechleby. Nevertheless, the prediction based on the DBH_{avg} of only the two largest shoots also had a very high value of pseudo-R^{2}_{adj} = 0.95, slightly lower than when the 3 largest shoots were used (pseudo-R^{2}_{adj} = 0.97). The minimum deviations in accuracy were found after adding the second and the third largest averaged shoots in the models in Vícenice. Adding the fourth and fifth largest averaged shoot in the models did not lead to an increase in the accuracy of predictions.

The contribution of each individual shoot (Biomass_{indiv}) to the total shoot biomass per stump decreased exponentially,

The predicted total (summed) biomass of Biomass_{st} differed from the total weighed biomass only by ≤ 1.2% in the prediction using BD, and by ≤ 0.06% in the prediction using DBH at both sites (

The biomass at the stand level was estimated by measuring the forest structure on the inventory plots at both plantations and by calculating the wood biomass using the allometric equations. The mean DBH, measured on the three largest shoots was used as a predictor in the allometric equations based on the results above (

We demonstrated that woody biomass of multi-shoot poplar SRC can be well estimated using the parameters of only one shoot (or two/three when higher accuracy is required), depending on whether the estimation is for the stump or stand level. For sufficiently accurate total biomass estimates, it is enough to measure just the largest shoot for the stump level and the three largest shoots for the stand level. It can be assumed that the number of shoots for the calculation of the total woody biomass will also depend on the number of dominant shoots and the number of dead shoots on the stump.

It is generally concluded that the power-law model estimates biomass better than the exponential model; however, in most studies on short rotation crops, the exponential model is usually applied (

DBH was a better predictor than BD and L, due to higher pseudo-R_{adj}^{2} and lower RMSE values. The parameter L from the shoots proved to be least accurate as a biomass predictor. This is likely because the shoots may be broken or their tips killed by frost, fungi, herbivores, of infection; therefore, shoot length may not be representative of the whole shoot (

The DBH of the three largest shoots is sufficient for nearly perfect estimates with negligible errors (≤ 0.06 %) at the level of all sample trees. Using the parameters of more than the three largest shoots does not lead to more precise estimates and may even make the estimates less accurate for both the stump and stand level. This results correspond with the findings of

A number of methods for indirect woody biomass estimation in coppices have been published (

We observed that the predicted total (summed) sample tree biomass was most accurate when using the DBH of the three largest shoots, with a deviation from the total weighed biomass of ≤ 0.06% only. This result corresponds with the finding of

Estimates of biomass at the stand level were compared with the real amount of biomass, which had been harvested and chipped on the plantations and transported to the heating plant where it was weighed. The difference between the observed amount of biomass and its prediction was calculated as a percentage deviation from the actual amount of biomass. However, a part of biomass was lost during the harvesting operations and remained in the field. The comparison did not take into account these losses during the transport of and manipulation with the wood chips, which can amount up to 5% of the total volume of the harvested biomass (

The reason for the relatively higher underestimation of biomass at the Vícenice plantation is not clear. It could have been caused by the stratified random selection of sampling plots within the plantation during the inventory. In specific cases, this may lead to the underrepresention of microsites where the amount of biomass on the stump may be significantly different (greater or smaller) from the rest of the plantation. The random sampling probably failed to fully capture more biomass produced on the stumps in microsites with greater nutrient or water availability. Therefore, the application of a regular network of inventory plots instead of stratified random sampling may be a more appropriate approach for biomass sampling in SRCs. This could minimise the omission of some types of microsite in large heterogeneous plantations.

Our results confirmed that the methodology applied for the estimation of biomass yield using allometric equations with averaged attributes of a few largest shoots, can be conveniently applied to poplar hybrid clone J-105. For a sufficient accuracy of biomass yield estimates in multi-stemmed trees of poplar SRC, we suggest to measure the DBH of the 3 largest shoots of each stump. Moreover, we illustrated how easily and quickly biomass yield can be estimated by measuring the DBH, and demonstrated that multi-stemmed trees can be simply included into ecosystem studies of woody vegetation with less time-consuming measurements.

We thank Verbava a.s. for their cooperation in selecting poplar SRC plantations and for providing information. We thank Michaela Kruttová for help with data collection. This work was supported by the Specific University Research Fund of the FFWT Mendel University in Brno (LDF_VP_2019031).

The authors declare no conflict of interests.

^{nd}edn) (Kenneth DRA, Burnham P eds). Springer, New York, USA, pp. 488.

^{2}measures based on wald and likelihood ratio joint significance tests. The American Statistician 44: 250-253.

_{2}-enriched sweetgum stand. New Phytologist 150: 477-487.

^{st}World Conference on Biomass for Energy and Industry”. Sevilla (Spain) 5-9 June 2000. NERF, Petten, Netherlands, pp. 536-539.

The relationship between the pseudo-R^{2}_{adj} and RMSE values of the power-law models for shoot biomass (per stump) estimations and the largest shoots averaged for the given parameter (BD_{avg}, DBH_{avg} or L_{avg}) used as an estimator in the models. (a, b): Sobechleby; (c, d): Vícenice.

Prediction of the sum of the biomasses of all the shoots per stump based on averaged DBHs of 1, 2, 3, 4 and 5 largest shoots per stump. (a): Sobechleby; (b): Vícenice.

Mean relative contribution of the individual largest shoots per stump to the total shoot biomass ranked according to their size per that stump,

Relative differences between the total biomass of all sample trees (Biomass_{st}) and total weighed biomass (shown as bars) of all measured shoots in relation to the number of the averaged largest shoots and the relevant parameter used for estimation. For the averaged basal diameter (BD_{avg}) and the diameter at breast height (DBH_{avg}), the 1-5 largest shoots were used and, for the averaged length (L_{avg}), the 1-5 largest shoots within each stump were used. (a) Sobechleby; (b) Vícenice.

Main characteristics of the two poplar SRC sites analyzed in this study.

Characteristics | SRC Site | |
---|---|---|

Sobechleby | Vícenice | |

Area (ha) | 17 | 11.5 |

Clone | J-105 | J-105 |

Age (yrs) | 4 | 3 |

Rotation | 2 | 2 |

Planting density (stump ha^{-1}) |
7692 | 6540 |

Coordinates | 49°28′ 00.139″ N17°38′ 50.755″ E | 49°05′ 13.164″ N15°47′ 58.714″ E |

Altitude (m a.s.l.) | 300 | 459 |

Poplar short-rotation coppice structure. (BA): stand basal area; (WM): wood moisture.

Variable | SRC Site | |
---|---|---|

Sobechleby | Vícenice | |

DBH (cm) | 3.4 ± 2.1 | 2.8 ± 2.3 |

Height(m) | 5.7 ± 3.1 | 4.9 ± 2.5 |

BA (m^{2} ha^{-1}) |
14.7 | 14.4 |

No. of stumps (pcs ha^{-1}) |
7.692 | 6.540 |

No. of stems per stump (pcs) | 4.6 ± 0.8 | 4.5 ± 1.1 |

No. of shoots (pcs ha^{-1}) |
35.383 | 29.430 |

WM (%) | 55.9 | 55.5 |

Pseudo-R^{2}adj values and the AIC of the generalized power-law and exponential models for estimation of prediction of the total shoot biomass per stump based on basal diameter (BD), DBH, and shoot length (L) for the given number of the averaged largest shoots (ALS, 1-5). The first value in each cell represents Sobechleby, the second Vícenice.

ALS | Parameter | Power-law model | Exponential model | ||
---|---|---|---|---|---|

AIC | R^{2} |
AIC | R^{2} |
||

1 | BD | 136.4/118.5 | 0.64/0.94 | 141.2/127.1 | 0.54/0.91 |

DBH | 124.9/119.6 | 0.81/0.94 | 129.7/135.9 | 0.75/0.87 | |

L | 123.1/65.4 | 0.81/0.92 | 125.9/66.9 | 0.79/0.90 | |

2 | BD | 104.9/125.4 | 0.93/0.94 | 110.3/125.5 | 0.91/0.89 |

DBH | 98.1/111.2 | 0.95/0.94 | 107.4/127.2 | 0.92/0.88 | |

L | 113.4/73.9 | 0.89/0.80 | 117.9/72.8 | 0.86/0.82 | |

3 | BD | 100.9/122.6 | 0.94/0.93 | 118.1/122.6 | 0.86/0.83 |

DBH | 91.3/118.7 | 0.97/0.93 | 107.4/139.9 | 0.87/0.77 | |

L | 120.4/67.1 | 0.85/0.67 | 129.8/68.5 | 0.75/0.61 | |

4 | BD | 106.7/78.9 | 0.87/0.94 | 109.9/96.4 | 0.85/0.78 |

DBH | 100.4/83.6 | 0.91/0.92 | 104.2/99.7 | 0.89/0.72 | |

L | 122.1/60.9 | 0.67/0.68 | 122.9/63.9 | 0.68/0.54 | |

5 | BD | 67.7/40.7 | 0.87/0.92 | 68.2/40.7 | 0.86/0.91 |

DBH | 66.9/67.8 | 0.88/0.92 | 67.1/79.6 | 0.88/0.73 | |

L | 79.7/20.1 | 0.62/0.88 | 79.9/20.6 | 0.61/0.86 |

Differences between estimated and real biomass of poplar SRC. (DEAB): deviation of the estimate from the real amount of biomass.

Variable | Parameter | Site | |
---|---|---|---|

Sobechleby | Vícenice | ||

Biomass(t ha^{-1}) |
Estimate for the largest shoots | 130.6 | 140.1 |

Estimate for the 2 largest shoots | 101.9 | 124.1 | |

Estimate for the 3 largest shoots | 89.2 | 113.7 | |

DEAB(%) | The largest shoots | 36.5 | 33.9 |

The 2 largest shoots | 6.8 | 18.6 | |

The 3 largest shoots | -6.5 | 14.2 |