Paper Reading: (2021) Integrating terrestrial laser scanning with functional–structural plant models to investigate ecological and evolutionary processes of forest communities-1

11 minute read

Published:

Original paper: O’Sullivan, Hannah, et al. “Integrating terrestrial laser scanning with functional–structural plant models to investigate ecological and evolutionary processes of forest communities.” Annals of Botany 128.6 (2021): 663-684. https://doi.org/10.1093/aob/mcab120

This is a record of the reading process, aimed at personal learning, reading and writing practicing.

Abstract

Background

  • Computational constraints and data deficiency is the limitation of FSPM.
  • TLS provides the data.

Scope

  • The paper provides the guidelines for for incorporating TLS data into the FSPM workflow.

Conclusions

  • Using TLS + FSPM to virtually explore the spatial and temporal dynamics of forest communities is feasible.

1 Introduction

Now there is a potential game changer: a combination of terrestrial laser scanning (TLS) and highly detailed functional–structural plant models (FSPMs) that mimic actual plants (e.g. Louarn and Song, 2020). Together, they provide a pair of tools that fulfil many prerequisites for addressing challenges with research questions related to the structural dynamics of forests.

Terrestrial laser scanning, optionally complemented with scanning from unmanned aerial vehicles, is a method that enables capturing the full 3-D structure of a forest stand with close to centimetre-level detail in a short amount of time (Liang et al., 2016; Calders et al., 2020), including the identification of individual trees and their species (Xi et al., 2020), the ground vegetation (Muumbe et al., 2021) and the topography of the terrain (Diamond et al., 2020). Methods for collecting additional spectral information to estimate the state of various physiological processes within trees along with the structural scanning are also being developed (Junttila et al., 2021).

  • TLS

The applicability of FSPMs has so far been limited, much due to the amount of required data and high computational demands of the models; hence the existing models typically cover the spatial extent of just a few tree individuals (Sievänen et  al., 2008; Zhang and DeAngelis 2020). On the other hand, one of the former limitations in the development of FSPMs, the availability of structural data for the construction and validation of models, is now changing as FSPMs are suited to directly utilize data provided by TLS, and use it to develop models that explore long-term dynamic processes within trees and stands (Sievänen et al., 2018).

  • FSPM

In this review, we will focus on recent developments of TLS and FSPMs, evaluate their current capabilities in data collection and modelling capacity, and evaluate the current and anticipated future applicability of these techniques in solving ecological and evolutionary questions when used in combination.

  • In the first of the following three sections we will address the impact of plant architecture on vegetation modelling from the individual plant level to the forest level and the impacts of forest evolution.
  • In the second section we will explore FSPMs and their three different sub-models: (1) physiological, (2) environmental and (3) architectural.
  • In the third section we will evaluate the realm of TLS and the benefits that this field has to scale up FSPMs from an individual-level modelling paradigm[n. 范式] to a community-level modelling paradigm.
  • In the third section we also propose a potential TLS to FSPM workflow, evaluating the advantages and disadvantages of data collection and software involved.
  • Lastly, we discuss the current and future synergies between TLS and FSPM for modelling forest communities.

2 Models of indivisual woody plant structure and function

Tree form and function have been studied using various hypotheses on how the extension growth, space-filling by the crown and thickening of already existing woody axes and root growth and the assimilation yield are related (West et al., 1999; Pałubicki et al., 2019).

Regular scaling between woody axes and growing shoots or leaves has been observed and explained by several theories/models (Shinozaki et al., 1964; West et al., 2009).

  • Shinozaki, Kichiro, et al. “A quantitative analysis of plant form-the pipe model theory: I. Basic analyses.” Japanese Journal of ecology 14.3 (1964): 97-105. https://doi.org/10.18960/seitai.14.3_97

The parameter values of the scaling relationships vary with growing conditions and with tree crowns (Bentley et al., 2013).

Approaches have also been used in the investigation of woody plant form and function:

  • A carbon–nitrogen balance (Valentine and Mäkelä, 2012).
  • Photosynthesis (Peltoniemi et al., 2012).
  • Transport of carbohydrates within the crown (Nikinmaa et al., 2013).
  • Rules can be defined by self-organization in branch growth (Pałubicki et al., 2009).
  • The crown development actually follows resource optimization (Givnish, 1984; Mäkelä et al., 2008; Sterck and Schieving, 2011; Franklin et al., 2020).
  • May need to be known.

One important area of active research is the accurate estimation of above-ground biomass (AGB) in forests globally, especially in the context of rapidly changing environments (Brienen et al., 2015).

Formalized tree allometric models have been widely used to estimate carbon stocks and balance in forests (Chave et al., 2005, 2014).

  • AGB is important.

2.1 Models of forest communities

Individual woody plant structure determines forest community structure, but this is itself determined by the structure of neighbouring woody plants.

Models without 3-D structural variation will not capture within- and between-plant microenvironments, therefore losing information regarding fine-scale patterns of form and function in heterogeneous[adj. 异构的] environments.

It is important to investigate and evaluate the relative 3-D structural contributions of individuals in forest communities and how these ultimately impact vegetation assembly and dynamics (Zhang and DeAngelis, 2020).

2.2 Evolutionary models

3 General features of FSPMs

A good sample of previous research into FSPMs and related topics can be found in special issues of

  • New Phytologist 166 (3), 2005;
  • Functional Plant Biology 35, 2008;
  • Annals of Botany 101 (8), 2008, 107 (5), 2011, 108 (6), 2011, 114 (4), 2014; 121 (5), 2018 and 126 (4), 2020;
  • Ecological Modelling 290 (1–2), 2014.

Table 1 shows a sample of FSPMs applied for simulation of growth of woody plants.

Some models include roots but only as a passive part of the plant. There are numerous functional–structural models of root systems (Barczi et al., 2018; Schnepf et al., 2018) but none of them have been linked to the models listed here.

Notable examples of such software include

  • AmapSim (Barczi et al., 2018),
  • GroIMP (Hemmerling et al., 2008),
  • Virtual Laboratory/L-studio (http://www.algorithmicbotany.org/virtual_laboratory/; Prusinkiewicz, 2000)
  • OpenAlea/OpenAleaLab (https://team.inria.fr/virtualplants/software/; Pradal et al., 2008).

Table 1. An overview of FSPMs that simulate dynamic growth of woody plant species. In addition to the publications, information about the models on the site www.quantitative-plant.org has been utilized. N.i., no information available; UML, Unified modelling language.

ModelSpeciesTime step/time rangeParameter inputMulti-plant simulation?Programming language/softwareOpen-sourcePurpose/commentCitation
ECOPHYSPoplarHour/several yearsBiologically derivedYesC++N.i.Disentangling growth factors(Host et al., 1996, 2008)
GreenLabGeneric/beech, Mongolian pineCan choose/decadesBiologically derived/Stat. fittingYesMatlab, Java, Cpp, ScilabUpon request, freewareResource capture, plant architecture(Cournède et al., 2008; Letort et al., 2008; Wang et al., 2012)
GroIMP (Beech)Generic/beechYear/decadesBiologically derivedYesGroIMP, XL (Kniemeyer and Kurth, 2008)YesDemonstration of software(Hemmerling et al., 2008)
IMappleApple treeHour/decadesBiologically derivedYesC#N.i.Horticulture(Kang et al., 2016)
LIGNUMPine, maple, poplar/genericYear or growth cycle/decadesBiologically derivedYesC++, L+Cc (Vos et al., 2007)Upon request, GitHubStudy of crown development(Perttunen 2001; Sievänen et al., 2008; Lu et al., 2011)
L-PeachPeach tree~1 d/several yearsBiologically derivedFocus is one treeL-studio (Prusinkiewicz, 2000)Upon requestHorticulture(Allen et al., 2005; Lopez et al., 2008; Da Silva et al., 2014)
Macadamia modelMacadamia treeDay/growing seasonBiologically derivedNoL-studio (Prusinkiewicz, 2000), L + C (Vos et al., 2007)N.i.Horticulture, proof of concept(Auzmendi and Hanan 2020)
MAppleTApple tree~1 d/several yearsBiologically derivedFocus is one treeL-Py (Boudon et al., 2012), OpenAlea (Pradal et al., 2008)YesHorticulture(Costes et al., 2008)
QualiTreePeach (fruit trees)DayBiologically derivedYes (canopies as ellipsoids)UML (object technologies)N.i.Horticulture, fruit development(Lescourret et al., 2011; Mirás-Avalos et al., 2011)
Radiata pinePinus radiataMonth/several yearsBiologically derivedNoBasicN.i.Silvicultural decisions(Fernández et al., 2011)
Sterck modelGeneric deciduousDay/decadesBiologically derivedNoN.i.N.i.Ecological–evolutionary problems(Sterck et al., 2005; Sterck and Schieving 2007)
V-MangoMango treeDay/several yearsStat. fitting/Biologically derivedNoL-Py (Boudon et al., 2012), OpenAlea (Pradal et al., 2008)N.i.Horticulture(Boudon et al., 2020)

FSPMs always feature three main components or sub-models:

  • a physiological model,
  • an environmental model,
  • a structural model.

Each sub-model (environmental, structural and physiological) must be evaluated in turn to ascertain trade-offs between reality, generality and simplicity at larger spatial and temporal scales.

3.1 Physiological sub-models

The growth and development of woody plants can be split into two categories of physiological processes:

  • the ‘fast processes’ of photosynthesis and respiration,
  • the ‘slow process’ of carbon allocation. Carbon is assimilated through photosynthesis and allocated either towards new growth or maintenance, whilst carbon losses occur through respiration and senescence.

Photosynthesis represented in FSPMs:

  • implicit models (Mäkelä and Hari, 1986),
  • empirical models (Le Roux et  al., 2001),
  • biochemical models, notably the Farquhar–von Caemmerer–Berry (FvCB) model for C3 photosynthesis (Farquhar et al., 1980; von Caemmerer and Farquhar, 1981), which has been scaled up for whole canopies in Medlyn et al. (2003).
  • Most FSPMs include a carbon allocation model based on the pipe model theory (PMT) proposed by Shinozaki et  al. (1964) over half a century ago (Godin, 2000; Le Roux et al., 2001; Lehnebach et al., 2018).
  • Many of its assumptions are outdated.
  • As an alternative, the authors suggest an additional model step not included in the PMT, which robustly quantifies foliage and sapwood partitioning regardless of species, age or stand condition (Lehnebach et al, 2018).

3.2 Environmental sub-models

Light conditions are fundamental to understanding how individuals optimize form and function within their respective communities.

Most methods of light simulation involve computing how much photosynthetically active radiation (PAR) is intercepted or absorbed by plant structural elements. To speed up radiation calculations, spatial discretization method has been used, such as

  • the method of voxel space (Li et al., 2018),
  • techniques based on Monte Carlo sampling in path tracing (Cieslak et al., 2008),
  • shadow propagation (Pałubicki et al., 2009).

it has become possible to model detailed absorption, transmission and reflectance for the full light spectrum using ray-tracing models (Henke and Buck-Sorlin, 2017).

The idea of reverse ray-tracing is to change the point of view from the radiation source to the entities themselves. Thus, the rays from every element are traced to the source, which ensures that each element is sampled properly and the errors are bounded. This approach has been used in a number of static FSPMs, including

  • Helios (Bailey, 2018, 2019),
  • and the FSPM generation platform GroIMP (Kniemeyer and Kurth, 2008; Hemmerling et  al., 2008).

3.3 Architectural sub-models

One of the most widely known and commonly used methods of representing the 3-D structure of woody plants is through Lindenmayer systems, otherwise known as L-systems (Prusinkiewicz, 2000).

Parameterizing L-systems in an informative way requires many structural parameters, which were difficult, if not impossible, to attain until the recent application of TLS technology to forest communities.