Tree growth and the tree-site complex
Tree growth is dependent of numerous factors which, combined in the tree-site complex, can have a limiting effect on tree growth. According to the law of limiting factors, a biological process like growth, cannot proceed faster than is allowed by the most limiting one. Degree and duration of a limiting effect varies along time, and if a factor changes so that it is no longer limiting, the rate of the process will increase until some other factor becomes limiting. Annual growth appears as the final result of the effects along time of the different limiting factors on the different biological processes involved in tree growth
Water and mineral uptakes as well as gas exchanges are the most evident biological processes from which tree-growth is depending on. The main external limiting factors of these biological processes have to be identified in the climate system and its two main components precipitation and energy flux. Seasonal variation in precipitation and energy flux are the motor of year to year growth variations.
In the tree site complex, the characteristics of the site (topography, substratum nature, soil components and structure, surrounding vegetation…) play the role of modulators of the limiting factors considered as inputs. They can amplify or reduce the contribution of some limiting factors on a homogeneous climatic area. Site characteristics variability introduces the spatial variation in tree growth rates.
Other limiting factors, like those associated to competition processes, forest management, fire, parasitism, insect diseases, damages from animals, can be considered as disturbances. They can progressively or abruptly change tree growth rate.
Tree radial growth as a record of environmental impacts on tree-growth.
Of course, the real result of tree growth processes along the year is the total annual biomass increase. But, only radial-growth is available as a record. One can assume (except for the juvenile period) that tree rings are good estimators of annual wood production over the life of the tree. Tree ring series can be considered as the results, for centuries, of year by year experimentations of tree ring to environment relationships.
As an example, a section of Pinus sylvestris allows to follow the changes in radial growth. Most of these changes are due to interannual climate variability, but also to the impact of traumatic events, or to the progressive evolution of the environment which can modify annual growth.
One can also see on that section that not only ring-width is variable but also ring structure. Some years are characterised by few latewood (1933-1942, 1958-1960..), others by an equal proportion of latewood and earlywood (1958-1963, 1974..). A more precise observation allows to detail how wood structure changes from one year to the next and even along the year according to the seasonal ( In non-tropical countries) characteristics of tree-growth. Here is the basic concept on which dendrochronology is built : One year = One ring.
Quantitative data associated to ring structure (resinuous species)
Climate and time control.
Each year, growth processes, and particularly cambial activity, which is directly involved in ring-width construction, are significantly influenced by climate. But numerous other factors, biotic or abiotic, can modify the rate of tree growth and consequently tree-ring features. Each tree, according to site characteristics ( the tree-site complex) reacts differently to weather fluctuations. Despite these variations, if one compares along time ring-width fluctuations on large area submitted to the same climatic variations, it appears, most of the time, strong correlation between the different trees. As an example, in the whole French Mediterranean area, the 1956 ring of Pinus halepensis is always a very narrow ring and sometimes, in part of the circumference, the ring is missing. The common situation, on such areas, is that ring-width series fluctuate synchronously from tree to tree. In such case tree can be considered as a recorder of climate interannual variability; here is the basis of Dendrochronology: Because of these synchronous variations one can crossdate ring chronologies of different trees and so it is possible to attribute surely the correct year to each ring. Dendrochronological science starts after this preliminary which conditions most of the technical operations involved in tree ring analysis, from field data collect to statistical treatments.
Field operations in Dendrochronology.
Introduction :
Field collect is conditioned by the problematic involving tree ring series analysis. Sometimes, dendrochronology is only used for dating purpose: archaeologists use dendrochronology to date wood artefacts or to extract some information about man behaviour in forests, origin of wood artefacts, foresters use also dendrochronology to know exactly age structure of the forest, geologists take advantages of trunks in sediments or at contact with rivers or glaciers to date events in sedimentation or erosion..etc…This is the restrictive meaning of Dendrochronology. But Dendrochronology is mainly involved in two other fields: ecology and climatology where it brings a time scale inaccessible to direct scientific observation and experimentation and more closely adapted to the phenomenons involved . We will expose here, from the site selection to the core collect, the methods associated to these fields. Discrimination between these two fields is only dictated by the problematic at the start of the research.
Site selection :
For climatological purpose (for example to reconstruct climate from tree rings), the objective in site selection is to obtain the best climatic signal recorded in tree ring series. Such a situation occurs when the trees of the sample are affected by the same set of growth-limiting factors all along their life. It means that the trees selected acts, for precise periods in the year, just as recorders of precipitation or temperature or other meteorological parameter. The sample has to be designed in order to avoid non-climatic factors and to maintain them more or less constant. This is the reason why dendroclimatological studies are mainly involving trees in the most extreme environment where, for example, drought or very low temperature limit growth. In regions with less extreme conditions, like temperate or moist climate, extraction of the climatic signal supposed to be registered by tree rings became much more difficult. The constraint of climate is weaker and, year by year, the combination of limiting factors can change. These two situations make the contrast between sensitive and complacent trees.
For ecological purpose (for example to precise autecology of a tree species), the strategy in site selection is much more controlled by the ecological knowledge of the species behaviour. Of course, one can expect that the situations in extreme conditions will bring probably the most important information on tree behaviour. But It is necessary to collect trees from populations representative of the whole ecological area covered by the species. Foresters for example want to know the conditions associated to the best productive sites. The multiple combinations of genetic, site conditions, human impact and spatial climate variability change tree behaviour and modify growth patterns. This is the reason why it cannot be reasonable to infer ecological information from few populations of a species covering a large geographical area including different topographies, substratum, and climate.
Trees selection:
Wood samples acquisition
Technical operations preceding measurements
For ring-width measurements only
For densitometric analysis
Then, using a double saw, a small lath of wood, perfectly calibrated in thickness ( 0.5 to 1 mm) is obtained. For the measurement, in order to follow at best the cellular files, the direction of sawing ( angle with the core axis) is specified on the sample under the magnification of a binocular.
Lath of wood are exposed to xray in order to obtain a slide on which densitometric measurement are possible using an optical micro-densitometer. Calibration of real wood density with optical density ( level of greys) is obtained using a calibration wedge of cellulose acetate.
Measurement of ring width:
Densitometry:
In both cases measurements are recorded on a computer.
Dating control on numerical data.
Measurement of ring width:
Densitometry:
Data analysis
Data analysis includes computing operations from chronology standardisation processes to tree growth modelling. As far as standardisation is concerned, after visual and statistical checking of the agreement between series, a common process is applied in order to obtain an homogeneous data set prior to calculating of tree-ring to climate relationships. Standardisation processes
Ring width:
Tree-growth modelling
- Statistical model (based on response functions) using linear regression after principal component analysis of meteorological data without physiological effect of CO2 (Keller,1997; 2000).
- Statistical model using linear regression after principal component analysis of meteorological data and taking into account fertilising effect of CO2 (Rathgeber et al., 2000). To take into account CO2 effect a b factor is calculated on the basis of 20th century net primary production (NPP) trend well evidenced (Badeau et al., 1995 ; Becker et al., 1994, 1995). This approach uses both response function and b factor in a predictive way. But this method based on classical dendrochronological principles assumes that environmental factors in one hand and the tree-ring to climate relationships in the other hand are stable over time. This assumption could be erroneous in the context of global changes because environmental factors are changing.
- Application of a Biogeochemical model : BIOME3.5C (Rathgeber, 2000) in order to provide a more realistic evaluation of global change impacts on forest stands production. BIOME3.5C which was developed to work on climatic means and on spatial variability in ecosystem distribution is modified to work in a chronological mode on a yearly time step. The approach tested on Pinus halepensis series includes three stages (Nicault et al., 2000) :
- Statistical model using non linear regression with Bayes formula and Metropolis-Hastings algorithm (Boreux et al., 2000).
This selection depends on soil water content and more precisely on interaction between soil water deficit and temperature. Such a selection needs a very good knowledge on ecophysiology of the different species used in Format. But also, the first approach using response functions can provide indications on the periods on which the parameters are operational.
Predicted tree-growth from models
Both for ecological or climatological purpose, perfect dating of annual ring is a drastic preliminary condition for the success of inferences to be extracted from tree ring series. This dating control is done by means of crossdating. For living trees, the counting of rings from the first ring under the bark to the pith does not avoid errors: taking as a true annual ring an intra-annual growth band inside a ring (false ring) or missing a ring which does not appear in the section examined (missing ring). The only way to avoid such problem is to practice crossdating on a large number of trees growing in the selected site. Moreover the average of replicated measurements allows to emphasis the common signal, (i.e. the climatic signal) registered in tree rings. Averaging processes minimises the effect of non-climatic factors which differ among individuals. Consequently, on each site, 10 to 15 trees are selected. They are supposed to grow in homogeneous ecological conditions and so to record the same climatic signal which is the expression of the climatic limiting factors through the local environment:, what we have called the tree-site complex. This tree site complex is representative of (for the different trees) a particular and homogeneous combination of limiting factors ( climatic, genetic, ecological…). If, it can be easy to avoid from the choice individuals too much outside of the mean standard, it is sometimes difficult to obtain a good representation of the tree population. In the most precise studies it may be necessary to make sub-samples according for example to the tree status ( dominant or non-dominant) or the age of the trees. In larger approaches sampling is focused on dominant trees, but it is necessary to keep in mind that this status can have changed over time.
Crossdating on cores refreshed by razor blade
Coring living trees
Coring nedds care :
- the borer must be inserted in the trunk as orthogonally as possible
- any visible traumatism must be avoided
- in the first coring geometrical center is targeted. Looking at first core adjustments are made for the next corings
The best way to have a good surface on which to observe tree ring series is of course to fall down the tree in order to obtain a stem cross section. The method can be sometimes used in co-operation with foresters when timbering is programmed. In most of the cases, cross dating and then measurements of ring-width as well as densitometric analysis are performed with small cores 4mm in diameter extracted from the tree by an increment borer. In order to avoid dissymetry in radial tree-growth measurement 2 or 3 cores are extracted on each tree. Consequently, on each sampling site 20 to 45 cores are collected and brought back to the laboratory. Precise cares have to be taken in coring, particularly when densitometric analysis will be performed. The most important is the position of the borer on the trunk . In order to obtain lately an observation surface the most perfectly perpendicular to the long axis of tracheids and fibers, the borer has to be also positioned perpendicularly to trunk axis.
Direct observation of rings on cores extracted from the borer is rarely possible and moreover ring width measurement quite impossible. Good observation and measurement need a perfect transverse section. After a correct reorientation of the core as the piece of wood was in the trunk, such a section is obtained either by refreshing with a razor blade or polishing the surface selected in order to obtain a plane surface allowing to access to the cell structure.
Densitometric analysis needs a more complex technical manipulation. At first, resins and sugar are extracted using a soxhlet.
Before the extraction of the lath from the wood support
Preparation of the cores for densitometry
Determination of the cutting angle: the core has to be cut in order follow at best the cells files along a radius.
Cutting of the cores for densitometry
Using a double saw the cylindrical core is transformed into a small lath. The thickness is perfectly controled (0.5 to 1 mm) all along the length of the lath.
Cutting the lath off the support
Measurement of ring width is performed under a binocular using a mechanical device which makes the core moving along a direction parallel to the cells files. The record of each ring measurement is controled by the operator who detects visually the end of each ring.
Crossdating and dating control
The microdensitometer designed for dendrochronology allows to a precise measurement of intra and interannual density fluctuations. The possibility to drive the sensors on the screen allows to follow precisely a cellular file and so, to accurately record the limit between each ring starting with low density values and finishing with the highest values. Such a control is also necessary to avoid false values linked to anatomical features like resin ducts or artefacts like failures.
Curves obtained from ring width series are visually compared over a light table and the statistical agreement between the series is checked by means of either parametric (correlation coefficients, Belfast cross t test) or non parametric tests ( Gleichlaugfigkeit).
Curves obtained from densities series are also visually controlled on the monitor. Characteristics of the annual ring structure is expressed by the densitometric profile.
In order to remove non climatic frequency from the ring-width and density series, two procedures are carried out. One consists in calculating ring areas series in which geometrical and age trends are reduced. Sometimes surfaces series do not need any other filtering or transformation process but it happens this procedure does not remove all mid and low-frequency variance. Therefore, another procedure is used which consists in transforming the nonstationary raw data series into a new series of stationary tree-ring indices. This is accomplished by dividing each measured ring value by the expected growth approximated by a 10-year digital filter.
Surfaces are calculated by using trunk circumference for individual series. Elementary ring-width series are standardised by means of a 10-year filter ; then mean individual and mean site chronologies are built-up.
Densitometry:
Densitometry series are limited to the only last 100 years. Elementary density series are standardised using a 10-year filter ; then mean individual and mean site chronologies are built-up.
Four approaches are in progress :
1) First stage consists in estimating (NPP) on the basis of dendrochronological variables (tree-ring width, density). The aim of this stage is to link ecosystem production and tree radial growth measurements.
2) Then a global biogeochemical model is fitted and validated on local dendrochronological data .
3) Afterwards the biogeochemical model is used to provide simulations of the NPP variations under a 2 CO2 scenario.
Model is as follows :
Tree ring index = g CO2 x [a0 + a1 f1 Bioclim1 +a2f2Bioclim2 +….]
- Functions g, f1, f2,… are defined from biological background like energetic flux and water supply.
- All the model parameters are uncertain posterior probability density functions (Bayesian approach)
Parameters involved in the Biome 3.5C model are used. The current selected parameters that must be non correlated are as follows :
- Soil moisture (mm)
- GDD5 : Sum of temperature above five degrees (°C).
MTCO : Temperature of the coldest month
A modification of Biome3.5C makes it possible to select the period on which the parameters will be calculated, according to the species and the area.
For example : for Pinus halepensis in south east France, selected bioclimatic parameters are :
- GDD5 from March to April
- SM from May to July.
The model is running but its improvement needs to improve the use of appropriate bioclimatic parameters. So, discussion on these bioclimatic parameters and their appropriate function is still open.
Predictive models involve tree-ring data, climate data, and output provided by regionalised climate models (ARPEGE, Météo-France). But the anomalies predicted by ARPEGE (Météo-France) are provided on a 50km x 50 km scale grid-points while climate and tree ring data are available on physical sites. Moreover, the location of the meteorological station is more or less remote from the tree site. Such a discrepancy can bias growth assessment from climatic data. Solving of these problems is attempted by some preliminary data treatments.
At first, simulation available for the recent years and for the radiative effect of 2 CO2 are adapted from the closest model grid-point to the tree sites using interpolation methods taking into account altitude and topography.
Second, meteorological data are also transposed to the tree site location by the same interpolation methods.
These two treatments will be used in order to produce modified data to be included in the data base and directly available for the ultimate stages of the calculation of predicted growth associated to climate change.