Dry- season transpiration
of savannah vegetation
Assessment of tree transpiration and
its spatial distribution in Serowe, Botswana
Alejandra Fregoso D.
April, 2002
Dry- season transpiration of savannah vegetation
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
by
Alejandra Fregoso D.
Thesis submitted to the International Institute for Geo-information Science and Earth Observation in
partial fulfilment of the requirements for the degree of Master of Science in Forestry for Sustainable
Development, Natural Resources Management.
Degree Assessment Board
Prof. Dr. Ir. A. de Gier
Dr. Ir. Ersin Seyhan
Dr. M. Lubczynski
Ir. M. Gelens
Dr. Ir. H. Steege
Dr. Y. A. Hussin
Chairman
External Examiner (Free University of Amsterdam)
Second Supervisor and ITC External Member
Second Supervisor/Forestry
Member
Students Advisor Forestry
INTERNATIONAL INSTITUTE FOR GEO-INFORMATION SCIENCE
AND EARTH OBSERVATION
ENSCHEDE, THE NETHERLANDS
Disclaimer
This document describes work undertaken as part of a programme of study at the International
Institute for Geo-Information Science and Earth Obervation. All views and opinions expressed
therein remain the sole responsibility of the author, and do not necessarily represent those of the
institute.
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Abstract
The distribution of people, plants, animals and ecosystems maintenance is closely related to water
availability. In dry lands the most important source of water is groundwater. The growing human
population leads to increasing demand for water, both in terms of quantity and quality, especially for
drinking water. Therefore, water for people and for the environment is an important issue to be
considered in the framework of sustainable use of the resources. The effective management of water
for social and economic development, as well as for protection and maintenance of natural ecosystems
is consequently important in order to meet human water needs.
Dry lands are characterised by a low and erratic rainfall with periodic droughts. Water is generally
scarce and surface water is available only during and shortly after the rainy season. Therefore the main
water supply comes from groundwater reservoirs. Water is a limited resource with differential spatial
and temporal distribution. It is known that vegetation has an influence on the hydrological cycle
through interception and transpiration. Several questions related to the interaction between vegetation
and groundwater are still unknown. The evidences are that the skill of plants to tap groundwater
depends on the depth of the groundwater, soil characteristics as well as the capability of plants to
develop deep roots (Scott et al. 1998). It is known that root systems in arid and semiarid regions often
reach great depths (Cole, 1986). In Botswana Kalahari sands, cases of tree rooting depth have been
found. The most astonishing Kalahari sands record of rooting depth was found in central Botswana.
The roots of most probably Boscia albitrunca were found at 68 m, and water at 141 m below the
surface (Scott et al. 1998).
The main objective of the present research is to assess the contribution of tree transpiration in open
savannah vegetation during dry season, in the east part of Kalahari Desert, Botswana. Specific
questions about transpiration refer to what are the tree savannah water requirements? Are transpiration
rates tree-species specific and or vegetation type specific? Is it possible to assess the spatial
distribution of transpiration through the up scaling of tree sapflow measurements? In order to answer
those questions the vegetation and water interactions are analysed in three different phases of the up
scaling process. It stars with tree stem sapflow measurements (Level 1), continues to plot assessments
(Level 2) and finalises with the estimates of vegetation transpiration (Level 3).
The up scaling approach followed in this study showed to be a very useful technique to measure dryseason transpiration in open savannah vegetation, since it was possible to identify transpiration
contributions by tree species and vegetation class. That is because tree is the main plant attribute
contributing to vegetation transpiration during the dry season. Main difficulties while estimating
vegetation transpiration were linked to two main sources: (1) sapflow estimates have an intrinsic
source of error, which is added when scaling up (e.g. plot, vegetation class) and, (2) underestimation
of the number of individuals per specie that characterize each vegetation class. Farther studies are
needed for a sustainable use of the water resources.
I
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Table of Contents
1.
Introduction .................................................................................................................................... 1
1.1.
Background ............................................................................................................................ 1
1.2.
Water resources...................................................................................................................... 2
1.3.
Water and vegetation interactions.......................................................................................... 4
1.4.
Problem statement.................................................................................................................. 6
1.5.
Research Objective and Questions......................................................................................... 7
1.6.
Conceptual framework........................................................................................................... 7
2. Study Area ...................................................................................................................................... 8
2.1.
Serowe area............................................................................................................................ 8
2.2.
Methods ............................................................................................................................... 11
2.3.
Sub-study area...................................................................................................................... 12
2.4.
Tree level: water usage ........................................................................................................ 13
2.5.
Plot level: transpiration estimations..................................................................................... 19
2.6.
Vegetation class transpiration rate estimations.................................................................... 20
3. Results .......................................................................................................................................... 23
3.1.
Tree level: water usage ........................................................................................................ 23
3.2.
Plot level: transpiration estimations..................................................................................... 32
3.3.
Vegetation level: Scaling plot estimations to vegetation level ............................................ 37
4. Discussion..................................................................................................................................... 40
4.1.
Tree level ............................................................................................................................. 40
4.2.
Plot level .............................................................................................................................. 42
4.3.
Vegetation level ................................................................................................................... 42
5. Conclusions .................................................................................................................................. 44
References ............................................................................................................................................. 45
Appendices ............................................................................................................................................ 48
II
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Figures
Figure 1-1. Trees water usage and water discharge and recharge ........................................................... 2
Figure 1-2. Transpiration up scaling process approach.......................................................................... 7
Figure 2-1. Location of the Serowe area at eastern Botswana. ............................................................... 8
Figure 2-2. Serowe area landscape units. ............................................................................................... 9
Figure 2-3. Conceptual hydro-geological and plant ecological cross-section (source:Lubczynski,
2000).............................................................................................................................................. 10
Figure 2-4. Workflow of the research ................................................................................................... 11
Figure 2-5. Serowe area vegetation and study are location where the work took place…………..…...16
Figure 2-6. Sapflow stations…………….……………………………………………………………..17
Figure 2-7. Setup of TDP sensors………….…………………………………………………………..20
Figure 2-8. ADAS System installation with climatic devices and tree sapflow sensors (source:
modified after Lubczynski, 2000)………………………….…………………………………….20
Figure 2-9. Sapflow sensors installation to the tree and biometric variables measured in
brackets…………………………………………………………………………………………...21
Figure 2-10. Conductive xylem area estimation by tree cutting experiment…………...……………...22
Figure 2-11. Study area of 10 x 10 km covering 4 vegetation types and 25 sample plots systematically
located on a grid of 2 km x 2km……………..………………………………….………………..25
Figure 3-1. Relationship diameter, d.5 and conductive xylem area, Ax for the three species…………29
Figure 3-2. Relationship between conductive xylem area, Ax and Crown area Ca for the three
species………………………………….…………………………………………………………29
Figure 3-3. Relationship between Crown area, Ca and stem diameter, d.5 for the three species……...29
Figure 3-4. Relationship between conductive xylem area Ax and sap velocity v...………….………..30
Figure 3-5. Sap velocity rates v for trees monitored in one day (3/10/2001) at station……….……….31
Figure 3-6. Daily patterns of sap flow of 6 trees monitored during one clear day (3/10/2001) at sapflow
station 4……………………………………………………………………………………...……31
Figure 3-7. Daily pattern of incoming short wave radiation (Kin) in an unclear and clear day at sapflow
station 3 (20 and 21 /09/2001)……………………………………………………………………34
Figure 3-8. Sapflow (Q) and normalized sapflow (Qn) for 5 trees monitored during a clear an unclear
day at sapflow station 3…………………………………………...……………………………...35
Figure 3-9. Daily pattern between PET and Q during the measuring period of sapflow station ……...35
Figure 3-10. Q of tree 1 and PET pattern during measurement s at station 1. The upper figure a) shows
the behavior of each time series, while lower figure b) shows the relation with a lag (30min) of 5
between Q and PT………………………………………………………………………………...36
Figure 3-11. Relationship between Q and PET a) represent regression after finding the 5 lag (1/2h)
between the two time series………………………………………………………………………36
Figure 3-12. Daily Q pattern for sapflow station 3 (monitored) and station 4 (extrapolated)…………37
Figure 3-13. Distribution of transpiration plots………………………………………………………..40
III
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Tables
Table 2-1 Tree species measured and number of sampled individuals in the four sapflow site stations.
....................................................................................................................................................... 14
Table 3-1. Descriptive statistics of the sampled trees for xylem area calculations and sapflow
measurements in the 4-sapflow site stations……………………………………………………...27
Table 3-2. Summary of the relationship between sap velocity rate, v [cm/h], tree diameter d.5 [cm] and
conductive xylem area Ax [cm2] for the different tree species. .................................................... 24
Table 3-3.Water consumption for the monitored trees during their respective period at the sapflow
sites. Biometric characteristics of these trees are presented in Appendix 4-1............................... 28
Table 3-4. Descriptive statistics of water usage for the three species selected. ................................... 28
Table 3-5. Descriptive statistics of A. fleckii, B. albitrunca and L .nelsii trees measured on the 25
sampled plots................................................................................................................................. 33
Table 3-6. Plot transpiration (Tp) and total conductive xylem area (Ax) and daily sapflow (Qp). ....... 36
Table 3-7. Canopy coverage per vegetation classes and the contribution of the three species
considered (A. fleckii, B. albitrunca and L. nelsii). ....................................................................... 37
Table 3-8. Summary of crown area coverage, conductive xylem and transpiration estimations of the
three species selected per vegetation classes................................................................................. 38
Table 3-9. Summary of daily water usage and conductive xylem area per vegetation classes. ............ 39
Table 3-10. Transpiration ranges………………………………………………………………………44
IV
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
1. Introduction
1.1.
Background
The distribution of people, plants, animals and ecosystems maintenance is closely related to water
availability. In dry lands the most important source of water is groundwater. The growing human
population leads to increasing demand for water, both in terms of quantity and quality, especially for
drinking water. Therefore, water for people and for the environment is an important issue to be
considered in the framework of sustainable use of the resources. The effective management of water
for social and economic development, as well as for protection and maintenance of natural ecosystems
is consequently important in order to meet human water needs. It requires the development of reliable
groundwater management strategies that consider groundwater supply and demand. These strategies
should be based on the factors involved in recharge and discharge of groundwater.
Arid and semiarid areas in Africa occupy 50% of the entire continent and support more than 35% of
the total population (CEC, 1986). Dry lands are characterised by a low and erratic rainfall with
periodic droughts. Water is generally scarce and surface water is available only during and shortly
after the rainy season. Therefore the main water supply comes from groundwater reservoirs. Savannah
vegetation, woodlands and sparse forest represent the natural vegetation. The increasing human and
animal populations and its demand for fodder and fuelwood has led to drastic declinations on the
vegetation cover. In (semi) arid regions soil fertility strongly depends upon water availability. Wind
and water erosion as well as surface compaction are the main soil problems (CEC, 1986).
Most of the countries in those regions present high population growth, which in combination with the
peculiar environmental factors as extreme daily temperature and low annual rainfall, lead to social and
ecological problems. In semiarid areas, annual rainfall (350-900 mm), is the limited environmental
factor that allows rain fed agriculture and pastoral practices. In these areas the pressure on the natural
resources is particularly magnified due to the competition for land and water, among the two main
economical activities: agriculture and pastoralism. The intensity of these social activities varies on the
availability of water, where agricultural practices are the main water consumer.
Botswana as a semiarid country in southern Africa, presents most of the problems related to natural
resources management. In this country, land degradation is accelerated by population and livestock
growth. A study of desertification in Botswana by Ringrose (1986), pointed out overgrazing and
overstocking as two of the most relevant factors of land degradation.
Mining of diamonds have also increased the water requirements during the last 30 years (Young et al.
1993). The high level of dependence of this important industry on water resources is sufficient to
justify an efficient and reliable water management for the country.
1
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The main economy inputs (54.5% GDP) in Botswana are mineral exports (especially gemstone
diamonds) and livestock. These two activities contribute to 92.3% of the country exports and 90% of
the country employments. The livestock sector is based on cattle raising to export beef products
mainly to the European Economical Community. The main water supply for these two economical
activities comes from groundwater (Young et al. 1993). Considering that Botswana’s economic is
highly dependent on natural resources, their appropriate management is extremely important.
Nevertheless, as many of the arid and semiarid countries, Botswana presents a lack of information on
natural resources stock.
1.2.
Water resources
Water is a limited resource with differential spatial and temporal distribution in the earth, present in
the 3 material phases (solid, liquid and gas). Water is an essential element for all living organisms.
Water management basically relies on fresh water management that is only around 3% of all the water
over the world. Approximately three-quarters of it are captured in the ice sheets and glaciers, and
about 23% occurs as groundwater, which is potentially accessible for human needs (Dingman, 1994).
Water management requires a better understanding of the complex hydrological cycle. The water
cycle comprehends the interdependence and continuous flows of water, that in a simplified manner
includes the following main phases: water vapour condensate at the atmosphere, and may precipitate;
where the land cover as the vegetation cover intercept part of it. The precipitation that reaches the soil
will be redistributed to the deeper soil layers by percolation, extracted by plant roots and later on
transpired, or evaporated from the soil surface to the atmosphere.
Precipitation
Evaportranspiration
Transpiration
Evaporation from ground
Sapflow
Sapflow in conducting
xilem area
Capillary water
percolation
Capillary fringe
Groundwater
Figure 1-1. Trees water usage and water discharge and recharge
2
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The remaining precipitation could follow different courses (Ward, et al. 1990):
• Remain as surface storage in pools, puddles and eventually evaporated back in to the
atmosphere
• Flow over the ground surface into streams, lakes and eventually evaporated back in to the
atmosphere or seepage towards groundwater or by surface flow into the oceans
• Infiltrate to the groundwater and evaporate back into the atmosphere by
evapotranspiration, or by through flow towards stream channels. Also could be storage in
the underlying groundwater by percolation.
The water flow on the water cycle strongly depends on physical properties of the soil as well as the
vegetation cover (van Dam, 2000). The infiltrated water that reaches the water table and is store as
recharge is considers part of the groundwater reservoir (Dingman, 1994). Ward and Robinson (1990)
define groundwater as the subsurface water in soils and rocks that is fully saturated. Because of its
importance for human purposes (domestic, industry and agriculture) as well as its importance in the
hydrological cycle, many studies on groundwater balance have been done. Groundwater stores are
also dynamic, although, in a continual slow motion. The residence time or the average transit time that
a “parcel” of water spends in the groundwater reservoir could be few years or 1000 of years
(Dingman, 1994). However, the residence time in semi arid and arid areas tends to be the highest due
to the slow pace of groundwater. In general terms groundwater dynamics is affected by several factors
as given in the following equation:
QGin + R = QGout + Eg +/- S +/-Qext
Equation 1
Where Lubczynski (2000) defined that groundwater laterals inflows (QGin) and natural recharge as
precipitation (R), have an input role, while groundwater outflows (QGout) and groundwater
evapotranspiration (Eg) have an output role on water availability. The other factors that have an
input/output role in the groundwater balance are changes of groundwater storage ( S), as well as
external groundwater sink and sources (Qext) as human well exploitation. Lubczynski (2000)
mentioned that importance of groundwater discharge directly to the atmosphere through
evapotranspiration, have been often underestimated in groundwater dynamic models.
Evapotranspiration (ET) is a term that includes all the processes where liquid or solid water becomes
atmospheric water vapour. In most of the cases the term ET includes either the moving upward of
water through soils or transpired by plants to the atmosphere. Actual ET term refers to the water
evaporation process from wet surfaces and transpiration process from plants. The various methods
that provide estimates of ET are commonly based on the physics of the process and in principles of
the conservation of the mass and energy. Actual evapotranspiration estimations consider surface
evaporation (Es), evapotranspiration from the unsaturated zone (Eu), and evapotranspiration from
groundwater (Eg). According to Lubczynski (2000) Actual evapotranspiration is written as:
E = Es + Eu + Eg
Equation 2
Where groundwater evapotranspiration (Eg) is caused mainly by transpiration and not by evaporation.
Since groundwater is stored on an average of 60m depth and in many cases about 800m below the
3
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
earth surface, hence the evaporation process is unfeasible. Estimates of ground water
evapotranspiration (Eg) become complex when the soil is covered by vegetation, due to the ability of
plants to tap water from the underground. Until now there are no direct methods that provide accurate
(Eg) estimates. The distribution of plant roots and the physiological and ecological behaviour of
plants to reach and use groundwater are not well known. This situation becomes more complicated in
arid and semiarid areas, as in southern Africa where the deepest root plants are documented.
A case study of groundwater balance in the Serowe area (eastern Botswana), reported by Lubczynski
(2000) identified as a main problem the determination of the groundwater evapotranspiration
component. Lubczynski states that measured ET is a combination of Eg and Eu and portioning of such
combination is not feasible yet.
The assumption that water loss comes from plants relies under the experience of rooting plant systems
depth for Botswana and South Africa (68 m) Le Maitre et al. 2000. The evapotranspiration
measurements where done during dry season were soil evaporation was almost negligible. Accurate
(Eg) estimates become more important being the connecting link between groundwater balance and
evapotranspiration model.
1.3.
Water and vegetation interactions
Several works in recent years have referred to the importance of vegetation in groundwater dynamics.
However, little is known about this specific relationship that leads to spatial and temporal variations
of groundwater levels (Lubczynski, 2000; Klijn et al. 1995; Timmermans et al. 1999).
It is known that vegetation has an influence on the hydrological cycle through interception and
transpiration. Studies have also shown that plants contribute to infiltration, by improving soil texture
and hydraulic connectivity that leads to impacts on groundwater recharge (Willis, et al, 1987).
However, there is not enough information about the magnitude of the effects of vegetation on
groundwater balance, in terms of discharge and recharge (Lubczynski, 2000). Some factors that affect
evapotranspiration and groundwater balances are: physiological factors (rooting depth and availability
of tap water, deciduous or evergreen species) and abiotic factors (soil texture and roots capacity to
penetrate soils, groundwater table level and the variations through seasons) (Scott et al. 1998; Klijn et
al. 1995).
Several questions related to the interaction between vegetation and groundwater are still unknown.
Some specific questions about transpiration refer to: what are the vegetation water requirements? and
what is the amount of water transpired by vegetation? These questions are needed to be answer in
order to have more accurate knowledge of groundwater balance and a better support for decisionmaking in the management of water, forest and agricultural resources as well as a commercial and
industrial development (Klijn, 1999; Le Maitre et al. 2000; Scott et al. 1998).
Estimates of water use by plants have been done mainly by transpiration studies. The use of sapflow
measurements through stems as assessment of transpiration is a good method to estimate the
magnitude of water used by plants (Hatton, et al. 1994; Granier,1987). In recent years the use of
sapflow measurements has been widely used under the bases of two main approaches: heat pulse
methods and heat balance methods. According to Edwards et al. 1996, the first method uses pulses of
4
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
heat as markers in the sap stream. The second one, measures the components of heat transport from a
continuous heat input.
Estimates of tree water-use have been scaled to stand and vegetation transpiration (Oren et al. 1998;
Vertessy, et al. 1997, Hatton, et al. 1994; Granier, 1987). Estimate transpiration by sapflow
measurements has some advantages over other methods, because the assessed values reflects the
direct water use by plants and are not mixed with evaporation values coming from other sources (e.g.
soils) (Hatton et al. 1994).
Most of the studies related to vegetation and water interactions (transpiration) have been done in
temperate forest and in less extent in semi arid and arid areas. (Kostner et al. 1998; Oren et al.1998;
Vertessy et al.1997; Phillips et al. 1996). The results of an international workshop in water and plants
interaction that took place on South Africa in 1996 strongly recommend the expansion of this type of
studies in order to have base lines for the understanding of semiarid and arid processes (Scott et al.
1998).
Recent studies about the interaction between natural vegetation and groundwater in semiarid
conditions, in South Africa, show that water losses due to evaporation from plant surface
(interception) is around 5%-20% of the total rainfall (Le Maitre, 2000). Other studies have highlighted
interception differences in different vegetation types (Scott et al. 1998). The research of Allen and
Grime (1995) on transpiration measurements by sap flow measurements, showed the role of shrub
transpiration on savannah vegetation type (35% of the total evaporation). The evidences are that the
skill of plants to tap groundwater depends on the depth of the groundwater, soil characteristics as well
as the availability of plants to develop deep roots (Scott et al. 1998).
The root system varies depending on plants biological and physiological characteristics as well as the
environmental conditions where they occur. Until now little information is available on root
architecture, stratification and its vertical and horizontal extent (Haase, 1996). It is known that root
systems in arid and semiarid regions often reach great depths (Cole, 1986). According to Cole (1986)
the Acacia tree species (except Acacia mellifera) in savannah vegetation in Botswana have both, large
taproots and developed lateral roots. Other species present differences in their root system as
contorted roots (Acacia tortillis) or no lateral roots (Boscia albitrunca and Commiphora africana).
These types of root systems enable plants to tap water from groundwater and utilise water contained
on the superficial layers of soil. However, depth of root penetration and root system is not known for
all species.
In Botswana Kalahari sands, cases of tree rooting depth have been found. The most astonishing
Kalahari sands record of rooting depth was found in central Botswana. The roots of most probably
Boscia albitrunca were found at 68 m, and water at 141 m below the surface (Scott et al. 1998). Other
record was found in central Botswana in a mineshaft, where roots of Acacia nigrescens at 50m depths
were evidenced (Cole, 1986). Nevertheless, specific information of the magnitude of water-use by
plants is still unknown. The investigation of the interaction between vegetation and groundwater, as
well as water dynamics is needed for a sustainable natural resources management.
5
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
1.4.
Problem statement
Botswana is included as one of the largely desert countries in Africa with more than 66% of arid area
(CEC, 1986). The sands of Kalahari Desert occupy two-thirds of the country and surface water is
available only during and after rainfall, frequent periods of droughts are present. The groundwater
resources consequently are the main sustains of drinking water for the total population (ca. 1,600 000
inhabitants), for livestock and vegetation. Land degradation throughout the country is the lack of
natural resources management; especially range management (Ringrose, 1986). Changes on
vegetation cover are affected for continuous grazing and human factors. Natural vegetation changes
are cyclical and depend on periods of rainfall and drought. The spatial and temporal distribution of the
vegetation is closely related to the availability of water.
Botswana major constraint for development is threatened by water shortage. The agricultural and
livestock sectors are the major users of water, with more than 30% of the entire demand (ca 137x10 6
m3/annum) by each sector. The domestic demand was about 20% in 1987 (Young et al. 1993).
However the increasing demand for drinking water and the problem of water availability, is an urgent
problem to solve.
Serowe village is the capital of the Central District, one of the ten districts of the country. In Serowe
during the last years, the population has increased rapidly. In 1981 the population was 23,000
inhabitants, while by 1996 the population grew to 35,000 inhabitants. In this situation water demand
has become a problem to solve. During 1988, studies of groundwater evaluation identified and
delineated potential well field areas. As a result of this project 8 boreholes were excavated in 19921994. The production estimated was around 3,000m3/day, however supply was only 1,000 m3/day.
The growing demand for water has reached until now 3,500m3/day, while the estimated offer of
around 3,000m3/day. Nowadays, water supply in Serowe area appeared to be reaching a critical stage
and groundwater strategies have to be urgently developed (WSC, 2000).
Water availability is a regional and local problem in Botswana that urgently needs a water
management strategy. In order to get a better understanding of groundwater dynamics and its
availability is therefore most important to study the factors involved. It is hoped that from this study
relevant information on the vegetation and groundwater interactions will be generated. This
knowledge will help to understand in which way, and to what extent, vegetation through transpiration
is affecting groundwater resources in the Serowe area.
6
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
1.5.
Research Objective and Questions
1.5.1.
Objective
The main objective of the present research is to assess the spatial distribution of trees transpiration in
open savannah vegetation during dry season, in the east part of Kalahari Desert, Botswana.
1.5.2.
Questions
•
•
•
•
1.6.
How much water do trees in savannah vegetation transpire?
Are transpiration rates tree species specific?
Are transpiration rates vegetation type specific?
Is it possible to assess the spatial distribution of transpiration through the up scaling of
tree sapflow measurements?
Conceptual framework
In order to fulfil the objectives, the vegetation and water interactions are analysed in three different
phases of the up scaling process. It stars with tree stem sapflow measurements (Level 1), continues to
plot assessments (Level 2) and finalises with the estimates of vegetation transpiration (Level 3). At
the left side of the figure the methods and tools needed for the up-scaling process are shown. At the
right side the up scaling level is specified (Figure 1-2).
Transpiration by vegetation
Level 3
RS and GIS
Statistics
Transpiration by plot
Level 2
Transpiration by tree
Level 1
by scalar variables
Sap flow meter system
Figure 1-2. Transpiration up scaling process approach
7
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
2. Study Area
2.1.
Serowe area
The observations discussed in this work were performed in the Serowe area (Figure 2.1), located at
the Central District, Botswana; about 275 km NE of Gaborone the Capital city. The Serowe area is
situated at the eastern part of the Kalahari Desert and covers an area of 2,444 km2. The eastern and
western limits are located respectively along longitude 26° 07’ 37.29’’E and 26° 54’ 10.12’’ E. The
northern and southern boundaries are respectively latitude 22° 14’ 10.08’’ S and 22° 30’33.77’’ S.
Figure 2-1. Location of the Serowe area at eastern Botswana.
8
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The climate in Serowe area is semi-arid with highly variable rainfall ranges from 200 to 1100 mm/y.
The two main seasons are cool dry winters and hot moist summers, with a mean annual precipitation
of 447 mm/y (SGS, 1988). The wet season occurs from October until April (summer) with a peak in
January. Rainfall is present as isolated storms where around 100mm in one day can precipitate. It is
important to mention that these isolated events are the main source of groundwater recharge for the
area (Lubczynski, 2000). Evapotranspiration in the area is high during summer (high temperatures
and relative humidity) and low during the dry winter. Soil studies had demonstrated that moisture
during the dry winter season in the first soils layers is neglected (Lubczynski, 2000).
The Serowe area is characterised by a prominent escarpment (90-150 m high) that demarcated the
area in two main terrain units: The western unit is a higher with gentle slopes running towards the
west, and covered by deep thick Kalahari Sands and open savannah vegetation. The eastern lower unit
consist on steeper slopes, especially near the escarpment; the soils composition is divers (arenosols,
luvisols and leptosols) as well as the vegetation cover (riverine woodlands, open savannah and
woodlands) (Lubczynski, 2000) (Figure 2-2).
Figure 2-2. Serowe area landscape units.
The vegetation in the Serowe is characterised by savannah vegetation. The species composition its
distribution and the number of strata above ground vary in the area. The variations along the
landscape units are mainly caused by water and nutrient availability and human activities. Plants from
savannah vegetation have the deepest roots reported, allowing them to reach water from very deep.
According to vegetation studies done by Ecosurv (1998) the savannah vegetation is subdivide on the
bases of the height and degree of canopy cover by the tree layer: woodlands, bushlands and open
savannah (Figure 2-3). Woodlands present 30% of tree canopy and tree height of 5.7m, the main
species are: Croton gratissimus, Acacia erioloba, Terminalia sericea and Peltoforum africanu
(escarpment and western unit) and Acacia tortillis and Acacia nilotica (eastern unit) .The riverine
woodlands are along the alluvial valleys of the escarpment where shallow water is found.
9
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Ecosurve (1998) recognized bushlands as a result of human activities, the species composition is
mainly the same as woodlands however, the tree canopy cover is around 2%, and the bush
(height<4m) canopy cover 25%. The open savannah is characterised by continuous grass layer and
discontinuous spare tree and bush layer. The dominant tree species along the escarpment are:
Terminalia sericea, Securidaca longipedunculata and A. erioloba. In the western unit (sandveld) are:
A. erioloba, A.fleckii, Boscia albitrunca and T. sericea. The bush layer consisted mainly of the same
species but with heights around 3 m and canopy coverage of 14%.
Figure 2-3. Conceptual hydro-geological and plant ecological cross-section (source:Lubczynski, 2000).
showing factors affecting groundwater dynamics in the area, see (Eq. 1).
10
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
2.2.
Methods
Figure 2-4 shows the sequence of the research steps. Once the objective is defined the literature
review of tree water usage and influencing factors supports the elaboration of the conceptual
framework. It is followed by fieldwork preparation and the selection of the study area. In the field,
tree water use and weather factors are measured and remote sensing images are gathered. The data
analysis includes the assessment of the effects of weather in tree transpiration, the evaluation of PET
and transpiration, which together with xylem area experiment allow the calculation of tree, plot and
vegetation transpiration. These results are the input for the development of a transpiration spatial
model of tree savannah vegetation. The methods follow in the present study are described according
to the four major up-scale levels of the research, i.e. tree, plot, vegetation and landscape level.
RESEARCH CONCEPTUALISATION AND DESIGN
LITERATURE REVIEW
DATA
ANALYSIS
FIELD
SURVEY
FIELDWORK PREPARATION AND DATA GATHERING: Study Area Selection
Remote Sensing
Images
Landscape Unit
Selection
Sampling Area
Location
Xylem area
experiment
Sapflow
measurements
Climatic data
Tree Transpiration
Effects of weather in
tree transpiration
Plot Transpiration
PET and
Transpiration
Vegetation
Transpiration
Results
Tree Level
Plot Level
Vegetation Level
DISCUSSION
The up scaling
Process
The Spatial
Transpiration Model
CONCLUSIONS AND RECOMMENDATIONS
Figure 2-4. Workflow of the research
11
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
2.3.
Sub-study area
In order to find a suitable area where to carry the research the Serowe area was examined according to
it 3 main landscape units, with the use of satellite imagery and existing maps. The purpose was to
identify and select, a homogeneous terrain unit in order to try to reduce as much as possible, the site
“environmental noise”. This “environmental noise” or spatial variability of the landscape attributes
(see below) could have an effect in the tree water usage response. The landscape attributes analysed
were:
Hydrology
Geology
Hydro-geological model (Lubczynski 2000).
Soils
Soils map of Botswana (scale 1:1,000,000)
Vegetation
Preliminary vegetation map (Hernandez, 2002)
Data collection was carried out on a study area consisted on a block square of 10 x 10 km, located at
the western landscape unit (sandveld) of the Serowe area, representing the eastern part of the Kalahari
Formation. The research was took place in a homogenous area (sandveld) in terms of its landscape
attributes (soils, vegetation, hydrogeology). The area is characterised by a sandy plateau covered by
open savannah vegetation (2-6.). Within this vegetation type, Ecosurve (1998) already identified three
variations of the open savannah vegetation. The characteristic layers or stratum in terms of its canopy
coverage gives their name: open savannah with trees, open savannah with shrubs and open savannah
with grass (Figure 2-5).
Figure 2-5. Serowe area vegetation and study
area location where the work took place.
The study area is in the western landscape unit
(sandveld). (Source Hernandez, 2002).
12
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
2.4.
Tree level: water usage
Tree selection
The xylem sapflow technique described by Granier (1985, 1987) was used to monitor tree water flow.
This method reflects direct measurements of water transport up wards in to the conductive xylem
tissue of the stem. The conductive xylem is the plant tissue were sap coming from the roots flow
upwards the stem to the leaves. Sapflow measurements were carried out during the sampling period of
11/09 – 4/10/2001. A total of four different located sapflow site stations were installed, with 18
measurements at each. In total 44 savannah trees of three different tree species were studied, with
some measurements doubled at simple trees (Table 2-1) (Figure 2-60). In the present work, a tree is
consider as a single life plant form >2m. The dominant trees that characterised the vegetation of the
area, Acacia fleckii, Boscia albitrunca and Lonchocarpus nelsii of open savannah vegetation were the
target research species. The criteria to select the species were:
•
•
•
•
Tree species characteristic of the open savannah vegetation
High abundance of individuals
Presence of leaves during fieldwork
Feasible to work in terms of tree weight
Figure 2-6. Sapflow stations
13
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The criteria used to select the individual measured trees were:
•
•
•
•
Height >2m and < 7m (weight of the trees, heavy trees would be difficult to manipulate for
the tree cutting experiment)
Stem form and habit (monopodial, or no ramification within 50 cm above ground)
Healthy individuals (avoiding ones with evidence of fire, pruning, lopping, cracks or
parasites)
Identification of the respective crown by each stem.
Table 2-1. Tree species measured and number of sampled individuals in the four sapflow site stations.
ID
AC FLE
BO ALB
LO NEL
Scientific name
Period
Acacia fleckii
Boscia albitrunca
Lonchocarpus nelsii
Total observations
Site 1
11-14 / 09
1 (1)
11 (5)
Stations
Site 2
Site 3
14-19/09
19-29/09
9 (6)
5
1 (1)
7(4)
*
Site 4
29/09-4/10
5 (3)
2(1)
3(2)
**
Observations
19
11
14
44
() number of double measurements per species
* measurements on trees of a specie that is not included in the present work.
Sapflow method
The xylem sapflow method developed by Granier (1985, 1987) is a simple sapflow estimation
technique consisted on a pair of probes inserted in the conductive xylem area of the tree stem. A
distance of 10-12 cm separates probes vertically. The type of sensors used was Dynamax Thermal
Dissipation Probes (TDP). The upper probe is continuously heated by a constant power (0.2W). The
temperature difference between the heated probe and non-heated one is monitored with
thermocouples (Figure 2.7). The method is based on liquid velocity (sap velocity) heat dissipation
theory in the conductive xylem tissue via an empirical calibration made on trunks of several species
(Granier, 1985). Sapflow is calculated by temperature difference between the probes that are affected
by sap velocity. The local sap velocity measured is then extrapolated to total tree (conductive xylem
area, were sap is transported). When sap velocity is minimal or zero, the temperature difference is the
maximum, whereas temperature difference decrease, with sap velocity increase. The expression to
calculate sapflow is the product of sap velocity through temperature difference between
thermocouples (incorporating the empirical calibration factor), extrapolated to total conductive xylem
area. The expression is as follow:
Q = v ∗ Ax
Equation 3
Where sap flow Q [g/h]:
v [cm/h] is the average sap velocity
Ax [cm2] is the conductive xylem at the measuring point
Sap velocity (v) is calculated by the dimensionless empirical factor (K) describe by Granier (1985)
based on the temperature difference measured by the thermocouples, measuring (1 cm2).
14
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
v = 0.0119 ∗ K 1.231 [cm/h]
and
K=
Where:
Equation 4
(∆TM
- ∆T )
∆T
TM, is the maximum T between the probes when flow is zero or close to zero (at night),
and
T, is the temperature difference between the probes when sap flow > 0
The sapflow measurements were carried out on 4 station sites. In each of these, three mobile
Automatic Data Acquisition System (ADAS) where placed. The ADAS systems consist on several
climatic and sapflow devices connected to a Data Logger (Skye datahog). Data loggers recorded
average estimations of climatic and sapflow every 30min (Figure 2-8). The location of the sap flow
stations was based under the following logistical criteria:
•
•
distance between trees was not more than 10 m, due to maximum length of sensor wires
number of selected trees available to measure, sensor systems only can operate when 3 or
6 sensors are set up.
Once the trees were selected, stem diameter d.5 [cm] and crown diameter Cd [m] were measured, as
biometric variables. Xylem sapflow was measured at 0.50m stem height from the ground (d.5). Probes
were inserted in the external xylem area 2cm deep (excluding bark) (Figure 2-9). The probes were
located at a stem height 0.50m (d.5) due to early ramifications of the stem bellow 1.30m (conventional
height). The number of sensors allocated per tree depended on the stem diameter (d.5). Trees with stem
diameter d.5< 7cm had one sensor, and in trees with d.5>7cm, two sensors were placed. This, with the
intention to observe variations in sapflow depending on the position of the probes within the sapwood,
as well as monitor variations due to improper readings of the sensors.
15
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Figure 2-7. Setup of sapflow sensors
Radiation
Air temperature
Wind speed
Sap
Air temperature
Soil temperature
Figure 2-8. ADASystem installation, with climatic devices and tree sapflow sensors (source modified after
Lubczynski, 2000)
16
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Crown diameter Cd
E-W
Trunk diameter d.5
Living xylem
(sapwood)
Trunk diameter, d.5 at
0.50 m
ADAS
SAP
d.5
Bark thickness
Dead xylem
(heartwood & pith)
Data logger
(recorder)
Figure 2-9. Sapflow sensors installation to the tree and biometric variables measured in brackets.
The position of the probes was determined facing the south part of the stem (for southern hemisphere
conditions) in order to avoid direct solar radiation and its effects on thermal gradient influence
between the heated probe and the reference one. A radiation reflective cover was placed to prevent
and protect probes from solar radiation.
Xylem area estimation
Total sapflow xylem estimation (Q) for each tree was calculated as the product of sap velocity (v)
over the conductive xylem area or living xylem area (Ax). The conductive xylem area (Ax) at the
measuring point was estimated for all trees measured at the end of the study by a tree cut experiment.
A less destructive method comprehending the tree boring with use of an increment borer was explored
as an attempt to identify differences in colour tissue between conductive xylem and heartwood.
However, due to the impossibility to differentiate trunk tissues by cores, the tree cut experiment was
finally used. The experiments were carried out under clear days (optimum conditions for high
transpiration) between 09:00 and 16:00hrs. After the sapflow measurements were completed, all trees
were cut at 0.45cm above ground during morning time (09:00-12:00am) and immediately placed on a
container with colour water for at least 3 hours. A water-soluble dye (Eosine) was used to make
evident the difference between the conductive xylem area (also sapwood), from the dead heartwood.
The dying was possible due to the fact that trees still transpire and move water up through the stem
around 3-4hrs after cutting, colouring the conductive xylem tissue (Figure 2-10).
17
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Figure 2-10. Conductive xylem
area estimation by tree cutting
experiment.
After the dying period, a cross sectional cut was done at the same point where sap velocity was
measured (d.5). Following, the stems discs were analysed and biometric variables were measured by
the use of ruler and magnifier lens. The biometric variables were: stem diameter (d.5), bark thickness
and heartwood diameter. Definition of the conductive xylem was possible by following the ring of
hydro-conductive vessels dyed, visible as dots within the tissue and estimating the area, or by
identifying the more inner set of vessels and the ring to estimate heartwood area. From those variables
the conductive xylem area was estimated using the following expression:
Axi = total trunk areai – ((bark areai) – (heartwood areai)) [cm2]
Once all variables where known, total sapflow was estimated for all trees. The following expression
was applied for all trees measured in the 4 sites:
Qi =
n
v ∗ Ax i
Equation 5
i
Tree
i
Qi = i total sap flow [g/h]
vi = i sap flow velocity [cm/h]
Axi = i total live xylem area [cm2] at d.5
Effects of weather in tree transpiration
The weather during fieldwork period changed drastically in terms of air temperature, relative
humidity, wind speed and net radiation. In order to understand and reduce weather effects on
transpiration variation response, sapflow data and weather data per day for all station sites were
analysed. Incoming short wave radiation (Kin), air temperature (Ta), relative humidity (RH), and wind
speed (Ws) were measured at heights of 0.50m and 2.05m above ground and recorded at the ADAS
system with a time resolution of 30 minutes. All these factors were used to calculate potential
evapotranspiration (PET) according to the Penman-Monteith (FAO) (time step method) by AWSET
software. The combine effect of the weather variables in PET was used to analyse the relationship
between PET on tree transpiration response.
18
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The problem of weather effect on sapflow variation was analysed first, by a linear regression model.
Maximum and mean daily values were analysed with maximum and mean daily sapflow values. Later
on, half an hour values of PET and tree sapflow (Q) were analysed. The analysis of variance was used
to test the effect of each variable in tree transpiration variation response. Later on, in order to detect
correlation through time between PET and tree Q, a cross correlation analysis of the two time series
with a successive lag (units of time, in this case 1/2h) was implemented. In this way we obtained the
information about the strength of the relationship between the two series (PET-Q) as well as the lag or
offset in time between them. This analysis was done for all the trees measured in order to find the
parameter that explains transpiration variation. The regression coefficients for each tree were used to
interpolate tree sapflow when the trees were not monitored for the entire campaign period.
Q = bo + b xi
Equation 6
Where:
Q = v ∗ Ax
and the explanatory weather variables tested were:
xi = each of the weather variables:
air temperature, Ta [ºC]
relative humidity, RH [%]
net radiation, Rn [W/m2]
wind speed, Ws [m/s]
soil temperature, Ts [ºC]
potential evapotranspiration PET [mm/h]
Biometric characteristics
The biometric relationships in trees between crown diameter (Cd), stem diameter (d.5) and conductive
xylem area (Ax) are well known however, those figures were unknown for the trees used in this work.
For that propose, a linear regression model between each biometric variable was explored in order to
find the unique species relationships. Since sapflow is related to total conductive xylem area of the
stem, the biometric relationship per species that allow estimating Ax, are therefore relevant in this type
of studies.
The scaling variables tested by the general regression model were:
conductive xylem area (Ax) [cm2]
trunk diameter (d.5) [cm]
crown area (Ca) [m2]
2.5.
Plot level: transpiration estimations
A total of 25 circular plots with 12.62 m in diameter, were systematic located in a grid distance of 2
km by 2 km between each other within the study area of 100 km2 (Figure 2-11) and covering the four
vegetation types described by Hernandez (2002). The systematic sampling design allow us to know
the abundance of the selected tree species in the area, as well as to up-scale tree transpiration to an
area basis (plot), in this case plots cover an area of 500 m2. In every plot biometric data of the tree
species selected were measured and later on its the area coverage calculated.
19
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Biometric data:
a) stem diameter at 0.5m above ground (d.5)
b) crown diameter measured in two directions (N-S and E-W)
To calculate plot sapflow (Qp) quantity per plot, tree species sapflow measurements in a sample of
individual trees were up-scale to plot transpiration (Tp) fluxes estimation. Where sapflow is (Q):
Qp =
n
Qi
Equation 6
i
Where:
i = individual trees
Qp = plot sapflow [litres/d]
Qi = sapflow of individual trees [litres/d]
Plot transpiration fluxes Tp where obtained by dividing Qp over total plot area (Ap):
Tp =
Where:
Tp = plot transpiration [mm/d]
2.6.
Qp
Ap
Qp = plot sapflow [litres/d]
Equation 6
Ap = Total plot area [m2]
Vegetation class transpiration rate estimations
The transpiration rate of open savannah vegetation is estimated by the underlying concept of scalingup tree sapflow measurements from plot level to vegetation class level by the use of the scaling
variable. In this study, vegetation is understood as the vegetative cover characterised by plant species
composition and species abundance of plants growing together under particular environmental
characteristics. The study area was characterised by Hernandez (not published) by the combined use
of remote sensing data (IKONOS, bands 2,3,4; February 2002) images, as well as ground truth data
for classification of the vegetation cover according to the structure classification proposed by the
Kenyan Soil Survey Van Wijngaarden (1985). The study area is represented for four vegetation
classes based on structural characteristics. In this system the classification of the vegetation is based
on canopy coverage by each strata layer (tree, bush, grass). The vegetation classes given by
Hernandez in (2002) are the same recognised in the present work: woodland, dense woodland,
wooded bushland and bushland for more detail see Hernandez (2002).
20
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Figure 2-11. Study area of 10x10 km covering 4 vegetation types and 25 sample plots systematically
located on a grid of 2 km x 2 km.
21
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
In this part of the up scaling process we us as secondary data, the vegetation map done by Hernandez
(2002) as well as the field data of 18 circular plots (500m2), that has the same plot size as the one used
in the present research. A total of 43 plots (25 systematic sampling, 18 secondary data) were analysed
as explain in the level before (plot level), according to the vegetation class that each plots belong
(Figure 2-11). All the plots of the same class were pooled and mean transpiration rates per vegetation
class were estimated.
In this way transpiration of open savannah vegetation class is calculated from extrapolating mean plot
transpiration rates per vegetation class.
Tvi = T pi
Equation 6
Where:
i= vegetation class
Tv = vegetation class transpiration
T p = mean plot transpiration
22
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
3. Results
3.1.
Tree level: water usage
3.1.1.
Biometric characteristics
The biometric characteristics of the measured trees in the 4 station sites show a wide range of
intraspecific and interspecific variability between individuals of the three tree species selected in this
study. These variations are related to different dimensions of the sampled trees and to different
species behaviour or ecological strategies.
Computations for conductive xylem thickness were done and the individuals with conductive xylem
thickness less than 2.4 cm where excluded for the later tree water use analysis. This was because the
sensor length is 2.1 cm, therefore measurements where done at least 2 cm deep in the conductive
xylem. Measurements of trunk diameter and the results from the conductive xylem area experiment
are presented in Appendix 1. A summary of the selected tree species and tree characteristics of the
sample used in the conductive xylem area experiment and sapflow monitoring are shown in Table 3-1.
The tree species Acacia fleckii (A. fleckii) presents the smallest tree dimension between the other two
species. Tree diameter (d.5) varied between 7 and 18 cm and the conductive xylem areas (Ax) varied
between 23 and 163 (cm2). The tree specie Boscia albitrunca (B. albitrunca), presents the biggest
dimensions of the three species. The maximum tree diameter (d.5) and conductive xylem area (Ax)
observed, belongs to B. albitrunca with dimensions of 28 cm and 420 cm2 respectively. The tree
specie Lonchocarpus nelsii (L. nelsii), presents the biggest crown area (Ca) of the trees measured (286
2
103cm ).
Table 3-1. Descriptive statistics of the sampled trees for xylem area calculations and sapflow
measurements in the 4-sapflow site stations.
Species
A. fleckii
No. of
trees
11
B. albitrunca
11
L. nelsii
11
Biometric
characteristics
Range
Minimum
Maximum
Mean
Std.
Deviation
d.5 [cm]
Ax [cm2]
Ca [103 cm2]
11.0
140.3
332.1
7.0
23.3
63.8
18.0
163.6
395.9
11.8
80.7
188.5
3.8
46.1
120.0
d.5 [cm]
Ax [cm2]
Ca [103 cm2]
21.0
464.7
248.1
8.0
34.8
34.6
29.0
499.5
282.7
19.6
252
149.2
7.8
164.2
85.3
d.5 [cm]
Ax [cm2]
Ca [103 cm2]
16.4
335.8
261.5
6.6
24.4
24.6
23.0
360.2
286.1
15.3
159.5
121.0
6.0
114.7
90.0
2
3
2
d.5 [cm], Stem diameter at 0.5 m above ground; Ax [cm ], conductive xylem area; Ca [10 cm ] Crown area.
The biometric relationships between biometric dimensions of the trees as conductive xylem area or
conductive xylem area AxI and stem diameter d.5 (Figure 3-1), crown area Ca and conductive xylem
23
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
area Ax (Figure 3-2) and d.5 and Ca (Figure 3-3) for the three species A. fleckii, B. albitrunca and L.
nelsii were established. Most of them were significant at P 0.05 F test (Appendix 2), except for L.
nelsii where, Ax and d.5 is the strongest relationship. Some of the biometric characteristics do not
differ significantly between species; only stem diameter (d.5) as well as conductive xylem area (Ax)
between A. fleckii and B. albitrunca was significantly different (P<0.001), while biometric
characteristics of L. nelsii did not differ significantly from the other two species.
This is an important finding for farther works on transpiration of these species by the use of the heat
pulse method, since this technique estimate transpiration or water use by an interpolation of the
measured sap velocity to total cross sectional area, or conductive xylem of the stem.
3.1.2.
Sap velocity and conductive xylem area dependence
In order to test if any dependence exists between sap velocity rates and the stem dimensions of the
tree, hence conductive xylem area, a correlation analysis was performed. For this analysis the
conductive xylem area of each tree and mean sap velocity rates were used as input data. The results
are listed in the Table 3-2 and the pattern for each of the tree species shown in Figure 3-4. Occasional
high sap velocity rates for the specific species were measured in individual trees. These high values
(two individuals, one of A. fleckii and one of L.nelsii), however, probably reflect some measurement
error (e.g. a small stem-temperature gradient). Due to the impossibility to relate them to any direct
cause, these high sap velocity rates observations were included in the analysis. The low correlation
and regression coefficient between sap velocity and conductive xylem area, for this analysis were not
significant. As a result from the test of significance of the coefficient b (coefficient of the x variable)
of the general model, we found that sap velocity in the three species is independent from stem
dimensions, inferred in this analysis as conductive xylem area (Appendix 4). Therefore sap velocity
rates for each specie can be utilised as the species-specific coefficient a (intercept coefficient).
Table 3-2. Summary of the relationship between sap velocity rate, v [cm/h], tree diameter d.5 [cm] and
conductive xylem area Ax [cm2] for the different tree species.
Species
A. fleckii
B. albitrunca
L. nelsii
Coefficient
a
2.16
2.4
3.31
Sap velocity – Ax
R
R2
.108
.012
.493
.243
.275
.076
SE
1.77
2.09
3.24
Sap velocity - d.5
R
R2
SE
.154
.024
1.77
.530
.281
2.04
.167
.023
3.32
24
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
A. fleck ii
B. albitrunca
L. nelsii
Figure 3-1. Relationship diameter,
d.5 and conductive xylem area, Ax
for the three species
Conductive xylem, Ax [cm 2]
600
500
400
300
200
100
0
0
5
10
15
20
25
30
Stem diameter, d.5 [cm]
A. fleck ii
B. albitrunca
L. nelsii
Conduactive xylem, Ax [cm 2]
600
Figure 3-2.Relationship between
conductive xylem area, Ax and
Crown area Ca for the three species.
500
400
300
200
100
0
0
100
200
300
400
2
500
3
Crown area, Ca [cm x10 ]
A. fleck ii
B. albitrunca
L. nelsii
Crown area, Ca [cm2x103]
500
Figure 3-3. Relationship between
Crown area, Ca and stem
diameter, d.5 for the three species.
400
300
200
100
0
0
5
10
15
20
25
30
Stem diameter, d. 5 [cm]
25
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
A. fleckii
B. albitrunca
8
v [cm 1/2 h-1]
6
5
4
3
2
12
v [cm 1/2 h-1]
y = 0.0056x + 2.157
R2 = 0.0178
7
y = 0.0074x + 2.4046
R2 = 0.2636
10
8
6
4
2
0
1
0
0
20
40
60
80
100
120
140
160
180
0
200
400
600
Ax [cm2]
Ax [cm2]
L. nelsii
9
8
7
6
5
4
3
2
1
0
y = - 0.0037x + 3.3069
R2 = 0.0357
0
100
200
300
400
Ax [c m2]
Figure 3-4. Relationships between conductive xylem area Ax and sap velocity v, for the three species
3.1.3.
Water usage
Daily water usage per tree varies considerably between trees of the same species. This variation is
related to the large range of tree diameter and hence conductive xylem area, since sap velocity does
not have relation with tree dimensions (see Figure 3-4). The Figure 3-5 shows an example of a daily
pattern of sap velocity (v) rates in 6 trees monitored during one bright day at station 4. These patterns
reflect the differential behaviour of water rates among the tree selected. Where A. fleckii trees
exhibited high mean sap velocity rate v [4.9 cm/h] these due to tree 47 that has different pattern as the
others trees of the same species. Sap velocity of B. albitrunca presented intermediate rates v [2.52
cm/h] and L. nelsii the maximum rates of v [5.57 cm/h]. Sap velocity monitored for these species in
the previous sapflow stations, gave different values of sap velocity for L. nelsii giving the
intermediate v values between the three species and A. fleckii the lowest v rates. Sapflow behaviour
for the same trees is shown in Figure 3-6 water consumption varies significantly up on tree
dimensions. In this case the higher Q is for tree 36 L. nelsii followed by tree 43 B. albitrunca, this
related to its big conductive xylem area and not to v.
26
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
18
Sap vel v [cm 1/2 h-1]
16
14
12
10
8
6
4
2
20 30
16 30
12 30
8 30
4 30
00 30
0
Time
Acfle 38
Acfle 47
Boalb 42
Boalb 43
Lonel 36
Lonel 45
Figure 3-5. Sap velocity rates v for 6 trees monitored in one clear day (3/10/2001) at station 4. Biometric
characteristics of those trees are presented in Appendix 1.
2500
Sapflow Q [g 1/2 h -1]
2000
1500
1000
500
20:30
16:30
12:30
8:30
4:30
0:30
0
Time
Acfle 38
Acfle 47
Boalb 42
Boalb 43
Lonel 36
Lonel 45
Figure 3-6. Daily patterns of sap flow of 6 trees monitored during one clear day (3/10/2001) at sapflow
station 4. Biometric characteristics of the trees are presented in Appendix 1.
Mean daily sap velocity rates v [cm/h], daily sapflow Q [litre/day] and normalized sapflow Qn [litre
103 cm-2 day-1] are listed in Table 3-3. Values of normalized sapflow (Qn) were obtained when
dividing total sapflow (Q) per tree by its projected crown unit area (Ca). In this form we have sapflow
values of each tree expressed in area basis (Qn) [litre 103 cm-2 day-1].
The result of the statistical analysis performed (ANOVA) of mean sap velocity between species,
allowed to differentiate sap velocity rates between A. felckii and B. albitrunca, but not between A.
fleckii – L. nelsii; neither B. albitrunca-L. nelsii (Appendix 4)
27
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Table 3-3. Water consumption for the monitored trees during their respective period at the sapflow sites.
Biometric characteristics of these trees are presented in Appendix 1.
Sapflow
Site
1
2
3
4
Specie
L. nelsii
L. nelsii
L. nelsii
B. albitrunca
L. nelsii
L. nelsii
L. nelsii
A. fleckii
A. fleckii
B. albitrunca
A. fleckii
A. fleckii
A. fleckii
A. fleckii
A. fleckii
B. albitrunca
A. fleckii
B. albitrunca
B. albitrunca
B. albitrunca
A. fleckii
B. albitrunca
B. albitrunca
B. albitrunca
L. nelsii
A. fleckii
A. fleckii
B. albitrunca
B. albitrunca
L. nelsii
L. nelsii
A. fleckii
Tree_ID
V [cm/h]
1
2
3
4
5
6
7
13
14
15
16
20
21
22
23
24
27
29
30
31
32
33
34
35
36
37
38
42
43
44
45
47
1.38
1.08
2.45
9.97
2.46
1.08
1.17
.90
3.30
4.33
3.50
2.81
1.20
1.00
1.40
1.26
1.03
2.76
2.89
2.91
1.47
4.11
4.13
3.38
3.29
1.21
2.80
4.59
1.87
9.99
7.21
6.76
Q [litre/day]
9.25
1.24
8.88
80.03
21.26
7.22
.69
3.03
3.10
33.53
13.70
2.36
2.10
2.10
3.40
2.64
.58
27.89
27.40
2.43
1.66
25.74
49.56
4.97
18.05
.76
6.95
32.43
2.86
30.95
4.25
12.98
Ca
[cm2 x 103]
286.05
59.40
117.02
164.39
120.07
129.46
24.61
395.92
101.79
210.33
373.93
80.42
124.10
172.39
188.57
90.79
63.79
239.75
190.50
34.64
77.93
150.33
282.74
44.30
273.40
70.69
301.91
192.44
41.55
152.05
33.01
202.28
Qn
[litre 103 cm-2 day-1]
.03
.02
.08
.49
.18
.06
.03
.01
.03
.16
.04
.03
.02
.01
.02
.03
.01
.12
.14
.07
.02
.17
.18
.11
.07
.01
.02
.17
.07
.20
.13
.06
A result of various measures of variation of mean sapflow (Q) [litre/day], mean sap velocity (v)
[cm/h] and a normalized sapflow per projected crown area (Qn) [liter 103 cm-2 day-1] for the trees
measured is summarised in the Table 3-3.
Table 3-4. Descriptive statistics of water usage for the three species selected.
Species
A. fleckii
No. of
trees
11
B. albitrunca
11
L. nelsii
11
Sap flux
characteristics
Mean
Std.
Deviation
Range
Minimum
Maximum
Q [liter/day]
V [cm/h]
Qn [liter103 cm-2 day-1]
16.0
5.9
0.05
1.0
0.9
0.01
17.0
6.8
0.06
5.0
2.2
0.02
5.3
1.8
0.02
Q [liter/day]
V [cm/h]
Qn [liter 103 cm-2 day-1]
77.0
8.7
0.47
2.0
1.3
0.01
79.0
10.0
0.48
27.3
3.8
0.15
25.4
2.3
0.13
Q [liter/day]
V [cm/h]
Qn [liter 103 cm-2 day-1]
28.0
9.4
0.17
1.0
0.6
0.02
29.0
10.0
0.19
9.3
2.9
0.08
9.5
3.0
0.06
28
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
3.1.4.
Sapflow variations and weather changes
Weather conditions during the entire fieldwork period were mostly clear, dray and hot, with
occasional cloudy, humid and rainy days due to the starting rainy season.
The incoming short wave radiation (Kin) pattern as a representation of a clear and unclear cloudy day
(Figure 3-7) was used to observe influence of different daily weather conditions and variations on tree
sapflow fluctuations patterns. The figures 3-8 shows variations of sapflow patterns Q and Qn for 5
trees monitored at sapflow station 3 during a clear and unclear day.
Due to different weather conditions from sapflow station to sapflow station and the differential effects
on the sampled trees, each weather factor was analysed independently with tree transpiration response
in their respective sapflow station. As a result of this analysis no good correlation was found between
daily sapflow rates and maximum and mean daily values of the weather data: wind speed (Ws), air
temperature (Ta), relative humidity (RH) and net radiation (Rn) and vapour pressure deficit (D).
When the combined effect of weather factors incorporated in PET was analysed with sapflow
response on half an hour basis, good relation was found for most of the trees. Moreover, when the
cross correlation analysis of two series (PET and Q) was applied, still better results were obtained
(Figure 3-10). From this analysis the lag between PET and sapflow response for each of the trees
monitored was found, and later on used in the linear regression model between PET and Q. The best
relationship found for all trees was the second order polynomial model. Figure 3-9 shows the
correlation pattern between PET with 0 lag and sapflow of one tree (ID 36) measured at sapflow
station 3 Appendix 5.
In c o m in g r a d ia t io n in a n u n c le a r a n d c le a r d a y
20 - 21 sep 2001
800
(Watt/m2)
Shortwave Radiation
1000
600
400
200
18:30
12:30
6:30
0:30
18:30
12:30
6:30
0:30
0
T im e
Figure 3-7. Daily pattern of incoming short wave radiation (Kin) in an unclear and clear day at sapflow
station 3 (20 and 21 / 09/2001).
29
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Sapflow behavior in an unclear and clear day
20 - 21 sep 2001
Q
[g / h]
3500
3000
2500
2000
1500
1000
500
0
0:30
6:30
12:30
18:30
0:30
6:30
12:30
18:30
Time
T_28 (A. fleckii)
T_29 (B. albitrunca)
T_31 (B. albitrunca)
T_32 (A. fleckii)
T_30 (B. albitrunca)
Sapflow behavior in an unclear and clear day
20 - 21 sep 2001
Qn [g 103cm-2 h -1]
20
15
10
5
0
0:30
6:30
12:30
18:30
0:30
6:30
12:30
18:30
Time
T_28 (A. fleckii)
T_31 (B. albitrunca)
T_29 (B. albitrunca)
T_32 (A. fleckii)
T_30 (B. albitrunca)
Figures 3-8. Sapflow (Q) and normalized sapflow (Qn) for 5 trees monitored during a clear an unclear
day at sapflow station 3. Biometric characteristics of these trees are presented in Appendix 1.
T 36
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
2500
2000
1500
1000
30/09/2001
01/10/2001
02/10/2001
20:30
16:30
8:30
12:30
4:30
0:30
20:30
16:30
8:30
12:30
4:30
0:30
20:30
16:30
12:30
8:30
4:30
0:30
20:30
16:30
8:30
12:30
4:30
500
0
PET [ mm / h/2 ]
PET
3500
3000
0:30
Sapflow [ ml / h ]
Sapflow
03/10/2001
Figure 3-9. Daily pattern between PET and Q during the measuring period of sapflow station 3.
Figure 3-9. Daily pattern between PET and Q during the measuring period of sapflow station 3
30
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
However this pattern was not present for all the trees monitored. In those cases the cross correlation
analysis helped to find the best correlation through finding their respective lag (30mi units) between
the two series. In Figure 3-10 representation of the patterns between Q and PET before (a) and after
(b) the cross correlation analysis is shown.
1
0.8
1500
PET [mm/ 1/2h]
Sapflow Q [g/ h]
2000
0.6
1000
0.4
500
0.2
0
0:30
0
6:30 12:30 18:30 0:30
6:30 12:30 18:30
Time
a)
PET
2000
1
1500
0.8
PET [mm/ 1/2h]
Sapflow Q [g/ h]
Sapflow
0.6
1000
0.4
500
0.2
18:30
12:30
6:30
0:30
18:30
6:30
12:30
0
0:30
0
Time
sapflow
b)
PET
Figure 3-10. Q of tree 1 and PET pattern during measurement s at station 1. Figure a) shows the behavior
of each time series, while figure b) shows the relation with a lag (30min) of 5 between Q and PET
The specific coefficients between PET and Q for each tree, where used to interpolate Q for the rest of
the days where sapflow was not measured. In figure 3-11 we observed the correlation pattern between
Q and PET without any cross correlation analysis. The respective coefficients are shown in the figure,
Figure 3-11. b) show the correlation between Q-and PET after finding the 5 lag (30min. unit) between
both Q and PET. Figures 3-10 and 3-11 represent the distribution of Q of tree 1 and PET, before and
after the cross correlation analysis. In this way water usage pattern for each tree during the entire
fieldwork period was estimated by the use of the respective trees coefficients..
PET - Q T1 (Lonel)
2000
2000
y = 717.66x2 + 1022.2x + 123.71
y = -1527.8x2 + 2039.4x + 147.23
R2 = 0.4694
R2 = 0.912
Q [ g/h ]
Q [ g/h ]
PET - Q T1 (Lonel)
1000
0
0
0.2
0.4
0.6
PET [ mm/h ]
0.8
1000
0
1
0
a)
0.1
0.2
0.3
0.4
0.5
PET [ mm/h ]
0.6
0.7
0.8
0.9
b)
Figure 3-11. Relationship between Q and PET. a) represent the relation without any cross correlation
analysis, while b) represent regression after finding the 5 lag (1/2h) between the two time series.
31
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Sapflow interpolationT_30
measured
Extrapolated sapflow
Sapflow
(Boalb)
Site 3
Q [L/day]
4000
Site 4
3000
2000
1000
0:30
12:30
12:30
0:30
0:30
12:30
12:30
0:30
0:30
12:30
0:30
12:30
0:30
12:30
12:30
0:30
0:30
12:30
0:30
12:30
12:30
0:30
0:30
12:30
0:30
12:30
0
Time
Figure 3-12. Daily Q pattern for sapflow station 3 (monitored) and station 4 (extrapolated).
Figure 3-12 shows Q pattern of one tree B. albitrunca, tree ID 30, measured during period of site 3, as
well as the extrapolated pattern of Q for the period of site 4 (not measured for that tree). The equation
used for interpolation was the following:
y = -5308.2x2 + 7917.8x + 182.94 ; r2 = 0.9259
3.2.
Plot level: transpiration estimations
The present results included 43 plots, which 25 were originated from the present work and 18 from
secondary data, by Hernandez (2002). The secondary data was obtained with the same methodological
approach. Plot transpiration depends upon tree density, tree species compositions and tree
characteristics (biometric). Table 4-6 shows a summary of the biometric characteristics per species of
the trees measured during the survey, as well as the number of plots where the species were present.
The species composition of the sampled plots was mainly characterised by Acacia fleckii; present in
28 plots out of 43 growing in clusters. A total of 106 stems were measured (d.5), trunk diameter varied
between 5 and 17 cm, with a mode of 5 cm and conductive xylem area ranging between 3.6 to 138.2
cm2. Individual crown identification was not possible for all the stems; in those cases the cluster
crown area coverage (canopy coverage) was measured. A total of 42 measurements of crown area
coverage were recorded.
B. albitrunca trees where registered in 12 plots out of 43, two of them in combination with A. fleckii
individuals. A total of 21 stems where measured, d.5 ranging from 7 to 38 cm with a mode of 16 cm.
The conductive xylem area was the highest of the three species ranging from 14 to 629 cm2. A total of
22 crown area coverage were recorded.
L. nelsii was the species with less frequency of the three; present in just 4 plots out of 43, two of the
in combination, another with B. albitrunca and one characterised by a dense monospecific cluster of
L. nelsii trees of the same dimensions (d.5 = 6.7 cm). A total of 106 stems were measured ranging from
32
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
6.7 to 15.2 cm with a mode of 6.7 cm, the conductive xylem area varies between 61.2 to 158 cm2.
Only six canopy area coverage were recorded, duet to the cluster behaviour of the species when
present. In 8 plots no presence of the species selected for this work was registered and in some of
them there was no presence of any trees species, just grass and bare soil.
Table 3-5. Descriptive statistics of A. fleckii, B. albitrunca and L .nelsii trees measured on the 25 sampled
plots.
No. of
trees *
106
No. of
Plots
28
B. albitrunca
21
12
L. nelsii
61
4
Species
A. fleckii
Biometric
characteristics
Range Minimum Maximum
d.5 [cm]
Ax [cm2]
12.0
134.6
5.0
3.7
d.5 [cm]
Ax [cm2]
30.7
622.5
d.5 [cm]
Ax [cm2]
8.5
151.4
Std.
Deviation
Mode
Mean
17.0
138.2
5
6.9
35.6
3.3
35.0
7.3
14
38
629.0
16
14.1
126.0
7.9
158.7
6.7
6.8
15.2
158.1
7
10.1
55.7
2.1
33.8
Ax estimations of stems d.5 < 5 cm where not possible to calculate, therefore not included
2
d.5 [cm] stem diameter at 50 cm above ground; Ax [cm ] conductive xylem area
Plot transpiration was obtained by up scaling tree species sapflow measurements in a sample of
individual trees to plot level. From this sample we were able to obtain species-specific information
related to the dimensions of the tree and Ax (biometric characteristics) as well as the sap velocity rate.
Plot sapflow (Qp) is the sum of the trees sapoflow present in the plot. Since sapflow is a product of Ax
times v,first of all we estimate Ax of each tree in the plot by the use of the scaling parameter (d.5 or
Ca). We found that sap velocity (v) was independent of stem diameter (d.5), therefore the values used
to estimate tree sapflow (Q) were the mean values of sap velocity for each species. At follow the
procedure to estimate plot transpiration is exemplified with plot No. 33.
Qp =
n
Tp =
Qi
i
Qp = plot sapflow [litres/d]
Qi= sapflow of individual trees
Qp
Ap
Tp= plot transpiration [mm/d]
Qp= plot sapflow [litres/d]
Ap= plot area [m2]
Q = Ax (v )
Where:
Ax = b0 (Ca ) + b1
and
v = mean _ velocity
33
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
bo
Acacia fleckii
Boscia albitrunca
Lonchocarpus nelsii
Plot
0.39
1.88
0.95
Scientific name
b1
5.27
-27.87
38.63
d.5 [cm]
v [cm/h]
2.62
3.84
2.85
Crown area
103 [cm2]
Plot canopy
coverage
Ax [cm2]
Q (ml/h)
Q [l/d]
Ca /plot
33
33
33
33
33
33
33
33
Acacia fleckii
Acacia fleckii
Acacia fleckii
Acacia fleckii
Acacia fleckii
Acacia fleckii
Acacia fleckii
Acacia fleckii
11.00
5.83
5.85
11.00
13.00
12.00
11.00
12.00
177.21
29.86
31.42
49.09
125.66
59.40
59.40
159.04
3.54
0.60
0.63
0.98
2.51
1.19
1.19
3.18
73.78
16.82
17.42
24.25
53.85
28.24
28.24
66.76
192.97
43.99
45.56
63.43
140.86
73.85
73.85
174.61
4.63
1.06
1.09
1.52
3.38
1.77
1.77
4.19
Total Qp [litres/d]
19.42
Total Tp [mm/d]
0.0388
We found that plot transpiration varies considerably depending on:
•
•
•
density of the trees (number of trees/plot)
dimensions of the trees present (seize, reflected in Ax)
species composition (v species specific)
Table 3-6 shows plot characteristics in terms of species composition, its contribution to the plot total
conductive xylem area (Ax), sapflow (Qp) and plot transpiration (Tp). A full description of plots
characteristics (no. trees, species composition and contributions of Ax and Q are found in Appendix 5.
Plot transpiration of the 43 plots surveyed varies from 0 litres/day where no trees or trees form the
species selected were not present up to 65 litres/day. The maximum plot transpiration value
corresponds to plot No. 19 were 4 trees of B. albitrunca characterised the tree layer and has the
maximum conductive xylem area surveyed (1126.7 cm2). The minimum plot transpiration flux is for
plot No. 20 were 2 trees of A. fleckii were present with small conductive xylem area (7.3 cm2) (Table
3-7). Plot transpiration varies according to tree density, species composition (related to specific sap
velocity, v) and the dimensions of the trees reflected (related to Ax) (Table 3-6). Distributions of the
plots in the study area are presented in Figure 3-13.
34
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
TRANSPIRATION PLOT DISTRIBUTION
N
LEGEND
.000 mm/d
.0187 mm/d
.0191 mm/d
.0242 mm/d
.0669 mm/d
0
2 km
Source: IKONOS image, February 2002
Figure 3-13. Distribution of transpiration plots
35
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Table 3-6. Plot transpiration (Tp) and total conductive xylem area (Ax) and daily sapflow (Qp).
Plot
No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
Species contribution
Total plot
Species
A. fleckii
A. fleckii
B. albitrunca
L. nelsii
A. fleckii
Ax
493.54
156.48
599.79
61.18
337.33
Q
30.98
9.83
55.24
4.18
21.18
Axp {cm2]
493.54
156.48
599.79
61.18
337.33
Qp [ltres/d]
30.98
9.83
55.24
4.18
21.18
Tp [mm/d]
0.0620
0.0197
0.1105
0.0084
0.0424
A. fleckii
A. fleckii
86.65
26.08
5.44
1.64
86.65
26.08
5.44
1.64
0.0109
0.0033
A. fleckii
26.08
1.64
26.08
1.64
0.0033
180.7
611.58
158.14
122.56
14.86
175.01
119.57
578.5
1126.23
7.3
10.95
104.06
118.05
83.58
74.04
282.58
16.64
38.38
10.82
7.7
0.93
10.99
11.01
36.34
103.71
0.46
0.69
6.53
7.41
5.24
4.65
26.02
180.7
769.72
16.64
49.2
0.0333
0.0984
122.56
14.86
294.58
7.7
0.93
22
0.0154
0.0019
0.0440
578.5
1126.23
7.3
10.95
104.06
118.05
83.58
356.62
36.34
103.71
0.46
0.69
6.53
7.41
5.24
30.67
0.0727
0.2074
0.0009
0.0014
0.0131
0.0148
0.0105
0.0613
23.5
1.48
23.5
1.48
0.0030
A. fleckii
A. fleckii
A. fleckii
A. fleckii
A. fleckii
B. albitrunca
A. fleckii
B. albitrunca
A. fleckii
B. albitrunca
B. albitrunca
L. nelsii
B. albitrunca
L. nelsii
A. fleckii
B. albitrunca
214.74
12.1
17.12
309.36
70.93
157.75
48.24
19.2
33.98
92.56
28.21
380.5
8.95
124.82
222.18
11.01
13.48
0.76
1.07
19.41
4.45
14.53
3.02
1.76
2.13
8.52
2.59
26.01
0.82
8.54
13.95
1.01
214.74
12.1
17.12
309.36
70.93
157.75
67.44
13.48
0.76
1.07
19.41
4.45
14.53
4.78
0.0270
0.0015
0.0021
0.0388
0.0089
0.0291
0.0096
126.54
10.65
0.0213
408.71
28.6
0.0572
133.77
9.36
0.0187
233.19
14.96
0.0299
A. fleckii
B. albitrunca
34.45
19.75
2.16
1.81
34.45
19.75
2.16
1.81
0.0043
0.0036
B. albitrunca
A. fleckii
L. nelsii
A. fleckii
A. fleckii
A. fleckii
B. albitrunca
A. fleckii
B. albitrunca
A. fleckii
A. fleckii
A. fleckii
A. fleckii
A. fleckii
A. fleckii
B. albitrunca
A. fleckii
Ax, conductive xylem area; Axp plot total conductive xylem area; Qp, plot sapflow; Tp plot transpiration.
36
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
3.3.
Vegetation level: Scaling plot estimations to vegetation level
Estimations of transpiration at vegetation level were obtained by up scaling plot transpiration
measurements from a sample of plots to vegetation level. The vegetation description and mapping
done by Hernandez (not published) followed the structural approach and was used as an input data.
The vegetation cover in the study area derived from the supervised classification of the IKONOS
image of February 2002 presented canopy coverage of less than 40%. The 4 vegetation classes
defined by Hernandez (2001) are mainly characterised by around 7 plant species per vegetation class.
Hernandez (2001) describe the vegetation by percentage of canopy coverage, as a result from it he
defines dense woodedbushland (BWd) with 38% of canopy coverage, followed by the bushland (B)
with 29% and woodedbushland (BW) with 26%. The class with less canopy coverage is the
grassedbushland (BG) with 13%. A summary of the contribution of the three species selected in this
study (A. fleckii, B. albitrunca and L. nelsii) over the total vegetation cover by each vegetation class is
presented in Table 3-7. From the 43 plots surveyed an estimate of the mean values of crown area Ca,
and conductive xylem area Ax per plot and density average of individuals of A. fleckii, B. albitrunca
and L. nelsii per hectare per vegetation class are presented in Table 3-7.
Table 3-7. Canopy coverage per vegetation classes and the contribution of the three species considered (A.
fleckii, B. albitrunca and L. nelsii).
Vegetation classes*
B
BG
BW
BWd
Total No.
Species
9
6
8
8
Contribution of the spp. selected
% of area covered
by plot
by vegetation cover
3
8
1
8
2
8
12
32
Total vegetation cover
% by plot
29
13
26
38
*Vegetation classes based on the classification of the structure of the woody vegetation (after Weg and Mbuvi, 1975).
B, bushland; BG. grassedbushland; BW. woodedbushland; BWd. dense woodedbushland (secondary data collected by Hernandez,
2001).
Bushland, B.
From the 43 plots surveyed, two were recognised as bushland vegetation class. From this, only one
presented the species selected for this study. The mean estimated contribution of A. fleckii species, in
terms of canopy coverage (Ca) and conductive xylem area (Ax) is the most important for this
vegetation class, with an area of 119 [103 cm2] and 96.77 [cm2] respectively. The estimated daily
amount of water is 6.26 [litres/day]. The estimated contribution of B. albitrunca represents 7 [103 cm2]
for Ca and 56 [cm2] for Ax. In terms of daily water usage, B. albitrunca coverage has an estimate of 5
[litres/day]. The density average of individuals per hectare for A. fleckii was 30 and 10 for B.
albitrunca; this vegetation class presents the lower density of the 4 vegetation classes.
37
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Grassed-bushland, BG
This vegetation class is characterised by the presence of the 3 species selected for this study. The
species with the higher contribution in terms of the mean estimated of canopy coverage (Ca) and
conductive xylem area (Ax) is A. fleckii with 57 [103 cm2] and a daily water usage of 1.88 [litres/day].
In contrast, B. albitrunca and L. nelsii both present similar values of canopy coverage 3 [103 cm2]; with
differential contributions of conductive xylem area 1.5 [cm2] for B. albitrunca and 20 [cm2] for L,
nelsii. The transpiration of B. albitrunca cover presents the smallest amount 0.14 [litres/day] of the
three species and L. nelsii the highest 3.3 [litres/day]. This transpiration contribution per species is
most probably related to the species density for B. albitrunca and L. nelsii with 7 and 20 individuals
per hectare respectively.
Table 3-8. Summary of crown area coverage, conductive xylem and transpiration estimations of the three
species selected per vegetation classes.
Vegetation
type *
B
BG
BW
BWd
No. of
Plots
16
6
16
5
Species
A. fleckii
B. albitrunca
Density
[ha]
30
10
Ca
[103 cm2]
118.79
7.43
Ax
[cm2]
111.09
5.51
Q
[liters/day ]
6.29
4.92
L. nelsii
2
2.4
3.78
0.3
A. fleckii
67
56.61
23.64
1.88
B. albitrunca
L. nelsii
7
20
3.17
3.16
2.98
41.61
0.14
1.42
A. fleckii
24
31.15
15.40
0.30
B. albitrunca
L. nelsii
A. fleckii
B. albitrunca
18
53
140
5
35.40
30.42
609.63
261.38
28.86
34.59
262.05
3.4
2.54
2.33
14.24
25.07
B, bushland; BG. Grassed-bushland; BW. Wooded-bushland; BWd. dense wooded-bushland.
Crown area cover, Ca [103 cm2 .05ha-1], conductive xylem area, Ax [cm 2.05ha-1], and transpiration, Tv [liters/day .05 ha-1].
Wooded-bushland, BW
The canopy coverage of this vegetation class is characterised by the presence of the three species
selected. The main contribution is by the L. nelsii species, with crown area coverage around 30 [103
cm2], conductive xylem area of 34 [cm2]. L. nelsii presented the highest values of transpiration, 2.33
[litres/day] as well as density, 53 individuals / hectare. The B. albitrunca species present mean values
of crown area coverage around 30 [103 cm2] as well as A. fleckii, and conductive xylem areas of 74
and 15 [cm2] respectively. The A. fleckii species contributes with the lowest transpiration value, 0.30
[litres/day] and is also the lowest value of the species in the four vegetation classes. In contrast,
transpiration value of B. albitrunca is 2.5 [litres/day] and is the maximum transpiration value of the
species in the three vegetation classes where presented. The density average of individuals per hectare
of B. albitrunca is 18, presenting the highest species density of the 4 vegetation classes. In contrast, A.
fleckii presented the lowest density of the four classes with 24 indiviuals/hectare.
38
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Dense wooded-bushland, BWd
In this vegetation class, only one species A. fleckii of the three species selected was present. The mean
canopy coverage presented the highest values with an area around 600[103 cm2] and a conductive
xylem area of 222 [cm2]. The density average of individuals per hectare is also the highest of the four
vegetation classes, with 140 individuals/hectare. Therefore, the estimated amount of daily water usage
by this vegetation class resulted to be the highest with 14 [litres/day].
From the mean plot characteristics per vegetation class, transpiration at vegetation was estimated.
Differences of daily water usage between vegetation classes (Table 3-9) are related with the particular
characteristics of the vegetation. The plant species presence or absence, its area cover and density,
define the vegetation and the vegetation transpiration. The dense wooded-bushland (BWd) present the
highest value of daily amount of water transpired, 13.57 [litres/day ha-1], caused mainly by the high
density as well as high canopy area coverage and conductive xylem area. The other 3 vegetation
classes presented lower transpiration values in comparison with BWd class. Variations of
transpiration values among classes, bushland 11 [litres/day], grassed-bushland 9.4 [litres/day],
wooded-bushland 9.4 [litres/day] and dense wooded-bushland 31 [litres/day] are related to different
contributions per species within vegetation classes. However, because of the low number of plots in
each vegetation class little we could say about vegetation class transpiration patterns.
Table 3-9. Summary of daily water usage and conductive xylem area per vegetation classes.
Table of ranges
Ax
[cm2]
B
BG
BW
BWd
16
6
16
5
157.36
61.75
133.80
402.80
Flow, Q [liters / day]
Mean
Q
11.54
9.45
9.39
30.74
Max
Q
12.09
9.37
9.54
33.45
Flux, T [mm / day]
Min
Q
8.58
8.23
8.01
20.59
Mean
T [mm/d]
0.0231
0.0189
0.0188
0.0615
Max
T [mm/d]
0.0242
0.0187
0.0191
0.0669
Min
T [mm/d]
0.0172
0.0165
0.0160
0.0412
B, bushland; BG. Grassed-bushland; BW. Wooded-bushland; BWd. dense wooded-bushland.
conductive xylem area, Ax [cm 2.05ha-1], and sapflow Q [litres/d] and transpiration T [liters/day .05 ha-1].
The spatial distribution of transpiration in the study area followed the vegetation types and represents
the amount of water evaporated from the vegetation cover to the atmosphere in terms of water depth
[mm/d ha-1]. Variations along the fence (light straight line) are observed and could be related as an
effect of the range management that take place on the area. Values of 0 transpiration are assigned to
bare soil, where no vegetative cover exist and evaporation has been defined neglictable (Lubczynski,
2000). Bare soil is related to cattle post and is surrounded by vegetation with low transpiration
amounts.
39
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
4. Discussion
4.1.
Tree level
When analyzing the results from the present research, we have to take in consideration the
environmental characteristics of the area, eastern part of Kalahari Desert: the savanna vegetation have
the deepest roots in the world and have the ability to use water from very deep underground. The
vegetation is under extreme weather conditions: high temperature during the day, high solar radiation,
low relative humidity and no soil moisture.
Sapflow methods are good means to measure and monitor water use by individual trees. The
difficulties on water use measurements of A. fleckii, B. albitrunca and L. nelsii in the eastern part of
the Kalahari Desert, Serowe, are discussed below.
The results shows that daily water use in the three species vary between trees and among species
substantially. Quantity sapflow throughout the study period was lower in A. flecki, than in the other
two species. The small biometric dimensions of the species and its low sap velocity rate support this
behaviour. In opposite way these considerations apply to B. abitrunca that presented the highest
values of both, sap velocity and biometric dimensions (hence sap flow or daily water usage).
However, the statistical analysis (ANOVA) for the difference among species means of sapflow
measurements did not revel significant differences between them. The two extreme high sap velocity
rates observed in A. fleckii and L.nelsii species could have an important influence on the result.
Making not possible the differentiation between A. fleckii and L.nelsii, as well as between L.nelsii and
B. albitrunca water consumption pattern. A larger number of monitored trees would allow to have a
better representation of the tree-to tree variability as well as species-species variations.
Estimates of tree water use by sapflow methods (TDP) rely on integrating sap velocity of the point
estimate (sensors in the stem) in to a stem flow (conductive sapwood area). Some sources of error
when estimating tree water usage by trees are associated with of sap velocity measurement and in
minor magnitude, when estimating conductive xylem area (Hatton, 1990; Phillips et al. 1996). In
order to avoid errors in the present research when calculating sapflow for individual trees, we
investigate the correlation between the two sapflow variables (Ax and v). As the results shown, for
each species we found that sap velocity seems to be no correlated to conductive xylem area. However,
for one of the species B. albitrunca this no correlation is not clearly evident, a larger number of
sample trees could help to revel this relations. With regard to sap velocity species-specific rates, it
was observed that
In the other hand, we investigate for each species the correlation of conductive xylem area with the
other biometrics characteristics. We found consistently strong linear correlation between (Ax, Ca, d.5)
for the three species. These correlations were expected since conductive xylem area is related to stem
diameter, and crown area mainly related to stem diameter. These findings are relevant for future
research, since sap velocity measured at one point in the stem is extrapolated to the entire conductive
xylem area. Therefore, the knowledge of Ax for each tree measured is needed and the present research
contribute to estimate Ax for the three species selected.
Sapflow estimations in the present research were calculated from sap velocity measurements at
approximately 2cm depth in the sapwood of all trees of the three species. Nevertheless, the present
results of tree water consumption pattern, does not consider the possible spatial variation across the
conductive sapwood area. The assumption that sap velocity is constant along the stem may not be
truth for the species monitored. As a result of the conductive xylem area experiments it was observed
that sapflow across the sapwood was not evenly distributed. The heterogeneous pattern of sapflow
may suggest that not all the vessels in the sapwood were actively transporting water. In A. fleckii
species the radial pattern of sapflow was more homogenous than the presented by the other two
40
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
species, whereas B.albitrunca presented the most heterogeneous pattern of the three species. These
observations agree with studies on radial patterns of sap velocity by Hatton (1990), Phillips (1996),
Roupsard et al. (1999), Granier et al. (1996), that demonstrated spatial variability of sap velocity in
different tree species. Granier et al.(1994) present an example of the relevance of spatial variability of
sap velocity in a hardwood species (oak Quercus pertrea), where 80% of sap was flowing at the outer
1 cm of sapwood trunk, with total conductive xylem thickness of 19mm. Köstner et al. (1998)
reported that other source of variation in sapflow for some species is related to infections by fungi. In
the present study it was observed that in A. flecki was common to find the centre of the stem rotten
and in the cavity water was stored. In some cases presence of ants colony inside the cavity (A. fleckii)
was found at the inner part of the bark (B. albitruna). A larger number of monitored trees, than the
one investigated in the present research could help to revel better tree variability of sapflow.
In order to reduce significant errors related to the spatial variation of sap velocity on the scaling up
from point sap velocity measurements to tree and to higher hierarchal levels (plot, vegetation). As
well as to reduce under or over estimation of daily water use by trees it is necessary to assess the
radial sapflow pattern within the conductive xylem of the species selected (Hatton et al. 1990).
However, technical considerations as the low number of trees possible to be sample at a time, the
needs for equipment supervision, time for installation, the field conditions and the high cost of the
specialised equipment have to be consider as well. It’s also important to consider other techniques that
allow sap velocity within the stems as well as conductive xylem area. In the present study tree cutting
was the technique use to estimate Ax. Nevertheless is a destructive technique and more ecological
friendly techniques as tree stem scanning or improvements of tree boring for those species should be
explored.
Weather conditions had a direct influence on sapflow variations. The diurnal pattern of sapflow in this
research approximately followed the course of incoming radiation, net radiation and temperature. A
precise comparison of sapflow with these climatic factors was not possible in basis of maximum daily
values. These finding agree with similar results reported by Köstner et al. (1996), where no
correlation was found between maximum flow and maximum vapor pressure deficit. The authors
mention this response as an effect of not enough deep sap velocity measurement and argued that
direct measurement in deeper conductive xylem than 2 cm, probably would improve the results. This
explanation could be applied in the present study for L. nelsii and B.albitrunca, which are the species
with bigger dimensions. The results of this research are for dry season tree responses however,
variations of sapflow during seasons on the Serowe area are also reported (Masenga, 2001).
Sapflow measurements presented better relation with climatic factors when they were analyzed
together through the PET of a crop model by the FAO Penman-Monteith equation. A comparison
between behavior of PET and sapflow in half an hour values of individual trees, presented a time lag
in most of the trees monitored. PET allowed to interpolate sapflow measurements for the period
where the trees were not measured and to reduce differential variation of tree sapflow by weather
changes along the measuring period (beginning of the rainy season).
PET- FAO Penman-Monteith based on a crop model (high density and homogeneous) under no water
stress conditions, resulted on high evapotranspiration values and an overestimation of actual
evapotranspiration. Comparison of sapflow measurements and PET for Serowe area is presented in
Masenga (2002). In spite that sapflow is a good approximation of transpiration, differences between
both representations of water usage by plants could be related to physiological species response.
Granier et al. (1996) pointed out that an average Scots pine tree (Pinus silvestris) can store ca. 30kg of
water in the stem and use ca. 7% of it before sapflow (at breast height 1.30m) had starts. A relevant
study about phenological patterns and water availability of woody species of Kalahari sand vegetation
by Childes (1988) presented that plants often used water reserves from their vegetative tissue. The
result of water use from the stem tissue during the dry season was observed in stem shrinkage in
Terminalia. sericea and Ochna pulchra, both species common in the Serowe area. These findings
suggest that the use of sapflow measurements (water transport through the stem) as transpiration
(water vapor towards the atmosphere) could reveal an under estimation of actual transpiration.
41
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Nevertheless, sapflow techniques in combination with knowledge of the measured tree species are
good methods to directly estimate tree water consumptions pattern.
4.2.
Plot level
Tree transpiration measurements by sapflow techniques enable estimates of plot transpiration. One of
the advantages of the method relies on the direct tree sapflow measurements, which allow the
investigation transpiration contribution of individual trees within a plot. The difficulties on plot
transpiration estimates presented in this study are discussed.
Plot transpiration was estimated by extrapolation of individual tree sapflow measurements by the
scalar variable (d.5 or Ca). The good relationships between the biometric dimensions and sapwood
area for the three species allow the up scaling process. Plot transpiration depends up on tree species
composition and tree biometric dimensions. As a result of this plot characteristic the results obtained
demonstrate high variation among plots and considerably low transpiration fluxes. Only in two plots
transpiration was > 0.10 mm d-1 and in one plot transpiration > 0.20 mm d-1.
One of the most significant difficulties in the present study was the extrapolation of tree sapflow data
(scale independent) in to plot transpiration fluxes (scale independent). Since plot transpiration depend
on tree density, species composition as well as site environmental factors and is when selection of
sampling strategy has an important effect on plot estimations (Köstner et al. (1996).
The three species selected in this study present a differential distribution pattern among them. This
pattern could be the cause of underestimation of tree sapflow, when calculating plot transpiration. It is
known that plot size has an influence on the probability to record species. This is related to the species
abundance and distribution pattern of the species (scattered or clustered).
A recent study on the issue of impact of plot size on the accuracy of species distribution models by
Pandit (2002) in Namibia, revel the optimum plot size for four species. One of them is B. albitrunca,
species that presents in this study, the higher sapflow measurements as well as the higher sap velocity
rates. The suggested plot size by Pandit (2002) for accurate species distribution is a radius size of
64m. In the present work plot size had a radius of 12.62 m however, tree density was comparable as
the one he reported for this species (2 trees ha-1 and 1 trees ha-1 respectively. The low representation of
L. nelsii in the plot survey, leads to an under estimation of the species transpiration contribution. What
in contrary the A. fleckii distribution pattern (cluster) and most evenly distributed.
The importance of a proper sampling strategy would help to reduce absence of important species that
may have significant contributions on plot transpiration fluxes. Would also help to identify local
differences in plot density.
4.3.
Vegetation level
Tree transpiration fluxes from plots of vegetation classes allow separate the tree (layer) total
transpiration from the transpiration of the vegetation cover (tree, shrub, herb layer). The transpiration
fluxes from the sampled plots of each vegetation class represent the transpiration contribution of the
three species selected from total transpiration. The transpiration of vegetation spatial distribution
analysis of transpiration and the difficulties presented in this study are discussed.
42
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
The vegetation cover was analysed by vegetation classes based on the structure of the plant
components (species). The plant elements (species) characterised the vegetation in terms of
contribution of crown area coverage by each strata layer (tree, shrubs, herbs). The analysis of the three
species selected (A. fleckii, B. albitrunca, L. nelsii) within each vegetation class, allow to characterise
the contribution of those species in each vegetation class (B, BW, BG and BWd) and to differentiate
between the other plant species of the vegetation.
One of the problems we presented in the present study at this level (vegetation) was when we up scale
from plot level to vegetation level. The difficulties are related to: identification of contribution of the
species selected (field and remote sensing data) and characterisation of transpiration fluxes from plot
fluxes (field data). These two variables exhibit high spatial variability, since in open savannah
vegetation despite of its apparent homogeneity; large local differences in density and composition of
the species were found.
In the present study it was observed that the vegetation class BWS (dense wooded bushland),
presented the higher transpiration fluxes (0.0615mm/d) in comparison with the other classes that
present relative similar fluxes. The influence of A. flekii tree density explains the high transpiration of
BWd, this is also consequence of over estimation of that species.
Differentiation of transpiration rates of the four vegetation classes was not clearly observed, the mean
transpiration values obtained are relatively similar between B, BG and BW. When we analysed the
transpiration ranges of each class, it is observed that BG and BW present similar ranges in comparison
with B, making unfeasible the differentiation among them.
43
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
5. Conclusions
This study presents an assessment of tree transpiration and its spatial distribution by up scaling tree
sapflow measurements, in Serowe, Botswana. The up scaling approach followed in this study showed
to be a very useful technique to measure dry-season transpiration in open savannah vegetation, since it
was possible to identify transpiration contributions by tree species and vegetation class. That is
because tree is the main plant attribute contributing to vegetation transpiration during the dry season.
Main conclusions at three different levels can be summarized as follows:
At tree level
• Sapflow measurement is a good technique for assessing transpiration at individual level
(trees) and a suitable tool to explore the species-specific variation and its relation with
weather conditions. However, more research on this issue needs to be carried out for a better
differentiation of tree species transpiration. Further research focusing on this specific topic is
relevant since a transpiration measure at other hierarchical levels (e.g. plot, vegetation class)
is based on sampled tree extrapolation.
At plot level
• One of the most significant difficulties faced during this study was the extrapolation of tree
sap flow data (scale independent) into plot transpiration fluxes (scale independent), since plot
transpiration depends on species composition, tree density and tree biometric dimensions.
Results of this research show that the effect of plot size (probability to record species)
resulted in an underestimation of transpiration contributions of the monitored species, when
scaling up sapflow individual trees.
At vegetation level
Vegetation transpiration depends on plant species composition and density at each vegetation
class. Main difficulties while estimating vegetation transpiration were linked to two main sources:
(1) sapflow estimates have an intrinsic source of error, which is added when scaling up (e.g. plot,
vegetation class) and, (2) underestimation of the number of individuals per specie that
characterize each vegetation class.
44
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
References
Allen, S. J. and V. L. Grime (1995). “Measurements of transpiration from savannah shrubs using sap flow
gauges.” Agricultural and Forest Meteorology. 75: 23-41.
Allen, S.J. and Grime, V.L. 1995. “Measurements of transpiration from savannah shrubs using sap
flow gauges”. Agricultural and Forest Meteorology. Vol 75, pp. 23-41. Elsevier Science B.V.
Atkinson, D. and Fogel, R. 1997. Roots: measurement, function and dry-matter budgets. In: van
Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. “Scalling-up – From Cell to Landscape”.
Society for Experimental Biology Seminar Series 63. Cambridge University Press. United Kingdom.
ermák, J.; Jeník, J.; Ku era, J. and Vladmír, Ž. 1984. “Xylem water flow in a crack willow tree
(Salix fragilis L.) in relation to diurnal changes of environment”. Oecologia. 64: 145-151. SpringVerlag. Berlin.
Childes, S.L. 1989. “Phenology of nine common woody species in semi-arid, deciduous Kalahari
Sand vegetation”. Vegetatio. Vol. 79, pp. 151-163. Kluwer Academic Publishers.
Cole, M.M. 1986. The savannas. Biography and geobotany. Academic Press, London.
Dingman, S.L. 1994. Physical Hydrology. Prentice Hall. USA. p.p.575.
Domingo, F., Villagarcia, L., Brenner, A.J. Puigdefabregas, J. 1999. “Evaporation model for semi-arid
shrub-lands tested against data from SE Spain”. Agricultural and Forest Meteorology. 95: 67-84.
Edwards, W.R.N., Becker, P. and Cermák, J. 1996. “A Unified nomenclature for sap flow measurements”.
Tree Physiology. 17:65-67.
Edwards, W.R.N.; Becker, P. and Èermák, J. 1996. “A unified nomenclature for sap flow
measurements”. Tree Physiology. Vol. 17, 65-67. Heron Publishing – Victoria, Canada.
Environmental Consultants (Ecosurv). 1998. Environmental Report for Central District Land Use
Planning, Unit, Serowe.
Goovaerts, P. 1997. Geostatistics for Natural Resources Evaluation. Oxford University Press, N.Y.
Goulden, M.L. and Field, C.B. 1994. “Three Methods for monitoring the gas exchange of individual
tree canopies: ventilated-chamber, sap-flow and Penman-Monteith measurements on evergreen oaks”.
Technical Report. Functional Ecology. Vol. 8, pp. 125-135.
Granier, A.; Biron, P.; Bréda, N. Pontailler, J.Y and Saugier, B. 1996. “ Transpiration of trees and
forest stands: short and long-term monitoring using sapflow methods”. Global Change Biology. Vol.
2, 265-274. Blackwell Science Ltd.
Granier, A.; Huc, R. and Barigah, S.T. 1996. “Transpiration of natural rain forest and its dependence
on climatic factors”. Agricultural and Forest Meteorology. Vol 78, pp. 19-29. Elsevier Science B.V.
Granier,1987. “Evaluation of transpiration in a Douglas-fir stand by means of sap flow measurements”.
TreePhysiology. 3:309-320.
Grime, J.P.; Thompson, K. and Macgillivray, C.W. 1997. Scaling from plant to community and from
plant to regional flora. In: Van Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. “Scallingup – From Cell to Landscape”. Society for Experimental Biology Seminar Series 63. Cambridge
University Press. United Kingdom.
Harding, R.J.; Blyth, E.M. and Taylor, C.M. 1997. Issues in the aggregation of surface fluxes from a
heterogeneous landscape: from sparse canopies up to the GCM grid scale. In: van Gardingen, P.R.;
Foody, G.M. and Curran, P.J. (eds.) 1997. “Scalling-up – From Cell to Landscape”. Society for
Experimental Biology Seminar Series 63. Cambridge University Press. United Kingdom.
45
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Hasse, P., Pugnaire, F.I., Fernandez, E.M., Puigdefabregas, J. Clark S.C. and Incoll, L.D. 1996. “An
investigation of root depth of the semiarid shrub Retama spheraerocarpa (L.) Bioss. By labelling of
ground water with a chemical tracer” Journal of Hydrology. 177:23-31.
Hatton, T., Moore, S., Reece, P. 1995 “Estimating transpiration in a Eucaliptus populnea woodland with
the heat pulse method: measurements erros and sampling strategies”. Tree Physiology. 15:219-227.
Hatton, T.J.; Catchpole, E.A. and Vertessy, R.A. 1990. “Integration of sapflow velocity to estimate
plant water use”. Tree Physiology. Vol. 6, 201-209. Heron Publishing – Victoria, Canada.
Hernández, A. 2002. “Mapping of woody vegetation in arid zones: a multi-sensor analysis. A case
study in Serowe area, Botswana”. Unpublished MSc-thesis, ITC – International Institute for
Geoinformation Science and Earth Observation, Enschede.
Isaaks, E. H. and R.M. Srivastava. An Introduction to Applied Geostatistics. Oxford University Press,
N.Y.
Jarvis, P.G. 1995. “Scaling processes and problems”. Plant, Cell and Environment. Vol. 18, 10791089. Blackwell Science Ltd.
Katul, G.; Philip, T.; Pataki, D.; Kabala, Z.J. and Oren, R. 1997. “Soil water depletion by oak trees
and the influence of root water uptake on the moisture content spatial statistics”. Water Resources
Research. Vol. 33 (4), pp. 611-623. American Geophysical Union.
Klijn, F. and Witte, J.M. 1995. “Eco-hydrology: Groundwater flow and site factors in plant ecology”.
Hydrology Journal. 7:65-77.
Kostner, B. Granier A. and Cermak, J. 1998. Sapflow measurements in forest stands: methods and
undertainities. Annales des Sciences Forestieres. 55:13-27.
Köstner, B.; Falge, E.M.; Alsheimer, M.; Geyer, R and Tenhunen, J.D. 1998. “Estimating tree canopy
water use via xylem sapflow in an old Norway spruce forest and a comparison with simulation-based
canopy transpiration estimates”. Ann. Sci. For. Vol 55, pp. 125-139. Inra/Elsevier, Paris.
Lammers, R.B.; Band, L.E. and Tague, C.L. 1997. Scaling behaviour of watershed processes. In: van
Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. “Scalling-up – From Cell to Landscape”.
Society for Experimental Biology Seminar Series 63. Cambridge University Press. United Kingdom.
Le Maitere, D. C.; Scott, D. F. and Colvin, C. Information on interactions between groundwater and
vegetation relevant to South African conditions: A review. In: Groundwater: Past achievements and future
challenges. Sililo et al. (eds.) Rótterdam. Balkema: 959-963.
Lubczynski, M. W. 2000. Groundwater evapotranspiration –Underestimated component of the
groundwater balance in a semi-arid environment-Serowe case, Botswana. In: Groundwater: Past
achievements and future challenges. Sililo et al. (eds.) Rótterdam. Balkema: 199-205.
Ludwig, F. 2001. “ Tree-grass Interactions on an East African Savanna: The effects of competition,
facilitation and hydraulic lift”. Documents sur la Gestion des Ressources Tropicales. Tropical
Resource Management Papers. Phd Thesis. Wageningen University. The Netherlands.
Magombedze, L.M. 2002. “Spatial and Temporal Variability of Groundwater Fluxes in a semi-arid
Environment – Serowe (Botswana)”. Unpublished MSc-thesis, ITC – International Institute for
Geoinformation Science and Earth Observation, Enschede.
Oren, R., Philips, N., Katul, G., Ewers, B.E., Pataki, D.E. 1998. “Scaling xylem sap flux and soil water
balance and calculating variance: a method for partitioning water flux in forest”. Annales des Sciences
Forestieres. 55:191-216.
Oren, R.; Zimmermann, R. and Terborgh, J. 1996. “Transpiration in Upper Amazonia Floodplain and
Upland Forests in Response to Drought-Breaking Rains”. Ecology. 77(3), pp. 968-973. Ecological
Society of America.
Pandit, S.N. 2002. “ Impact of plot size on the accuracy of species distribution models”. Unpublished
MSc-thesis, ITC – International Institute for Geoinformation Science and Earth Observation,
Enschede.
Phillips, N, Oren, R. and Zimmermann R. 1996. Radial patterns of xylem sap flow in non-difusse ring
porous tree species. Plan Cell and Environment. 19:983-990.
Roupsard, O.; Ferhi, A.; Granier, A.; Pallo, F.; Depommier, D.; Mallet, B.; Joly, H.I. and Dreyer, E.
1999. “Reverse phenology and dry-season water uptake by Faidherbia albida (Del.) A. Chev. in
46
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
agroforestry parkland of Sudanese west Africa”. Functional Ecology. Vol. 13, 460-472. British
Ecological Society.
Sala, A. and Smith, S.D.1996. “Water use by Tamarix ramosissima and associated phreatophytes in a
Mojave desert floodplain”.Ecological Applications, 6(3), pp. 888-898. Ecological Society of America.
Scott, D. F. and Le Maitre, D. C. (eds.). 1998. The interaction between vegetation and groundwater. CSIR
Report No. ENV/S-C 97161.
Smith, D.M. and Allen, S.J. 1996. “Measurement of sap flow in plant stems”. Journal of Experimental
Botany, Vol. 47 (305), pp. 1833-1844. Oxford University Press.
The Commission of the European Communites (CEC), Bundesministerium für wirtschaftliche
Zusammenarbeit (BMZ), Deutsche Gesellschaft für Technische Zusammenarbait (GTZ). 1986. Towards
control of desertification in Africa drylands. CEC, BMZ, GTZ. Eschborn. p.p. 241.
Timmermans, W. J. and Meijerink, A. M. J. 1999. “Remotely sensed actual evapotranspiration:
implications for groundwater management in Botswana”. International Journal of Applied Earth
Observation and Geoinformation. 1(3/4):222-233.
van Dam, J.C. 2000. “ Field-scale water flow and solute transport – Swap model concepts, parameter
estimation and case studies”. Phd Thesis. Wageningen University. The Netherlands.
van Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. “Scalling-up – From Cell to
Landscape”. Society for Experimental Biology Seminar Series 63. Cambridge University Press.
United Kingdom.
van Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. Science of scaling: a perspective on
future challenges. In: van Gardingen, P.R.; Foody, G.M. and Curran, P.J. (eds.) 1997. “Scalling-up –
From Cell to Landscape”. Society for Experimental Biology Seminar Series 63. Cambridge
University Press. United Kingdom.
Vertessy, R.A., Hatton, T.J., Reece, P., O’Sullivan, S.K., Benyon, R.G., 1997. “Estimating stand water
use of large mountain ash trees and validation of the sap flow measurement technique”. Tree Physiology.
17: 747-756.
WCS. 2000. Serowe well field 2-extension project. Report No. TB10/3/10/95-96.
Werk, K.S.; Oren, R.; Schulze, E.D.; Zimmermann, R. and Meyer, J. 1988. “Performance of two
Picea abies (L.) Karst. Stands at different stages of decline. III Canopy transpiration of green trees”.
Oecologia. 76: 519-524. Spring-Verlag. Berlin.
Willis, R. and Yeh, W. W. 1987. Groundwater systems planning and management. Prentice-Hall. New
Jersey.416 pp.
Young, M.D. and Solgrig, O.T. (eds.) 1993. “The world’s savannas”. Man and the Biosphere series. Vol.
12. UNESCO.
Zimmermann, M.H. and Milburn, J.A. Transport and Storage of Water.
47
DRY-SEASON TRANSPIRATION OF SAVANNAH VEGETATION
Assessment of tree transpiration and its spatial distribution in Serowe, Botswana
Appendices
48