Study area
Water consumption of P. juliflora (hereafter Prosopis) was measured in the Amibara District (Fig. 1) of the Afar Region, Ethiopia (9.16° to 9.21° N and 40.08° to 40.12° E at 740 m a.s.l.). The study area is located in the Awash River Basin and includes both the floodplains of the Awash River and the adjacent dryland. Although relatively water scarce, the Awash River Basin is the most developed and utilized river basin in Ethiopia27. The human and livestock population in this basin are estimated at 18.6 and 34.4 million, respectively, and nearly 70% of large irrigation schemes in Ethiopia are located in the Awash Basin28. The mean annual river flow/discharge of the basin at the terminal Lake Abe is estimated at about 4.6 billion m3 of water although exposed to evaporation27. Water scarcity, particularly in the lower parts of the river, is the major limiting factor for irrigation development, particularly during the low flow season27.
The Afar Region has a mean annual rainfall of about 560 mm29. The region is the hottest part of Ethiopia, with a mean annual temperature of 31 °C. The mean maximum temperature reaches up to 41 °C in June, and the mean minimum temperature ranges from 21 to 22 °C between November and December30. The biome can be described as semi-arid to semi-desert. The natural vegetation consists of scattered dry shrubs, woodland comprising different Vachellia (Acacia) species, bushland, grassland and wooded grassland18. The area has different soil types, including silt fertile soils, sandy soils, heavy clays and rocky outcrops, and a wide range of altitudes ranging from 175 m below sea level to 2,992 m a.s.l. Shiferaw et al.30 found that Prosopis has primarily invaded areas ranging from rangelands to farmland. Main sources of livelihood are pastoralism and some agro-pastoralism around small rural towns30. The main crops grown in the floodplains of Awash River are cotton and sugarcane.
Experimental design
To investigate the temporal and spatial variation in water use by Prosopis across the landscape, data collection was done in the two most heavily invaded habitat types in the Afar Region. These include the floodplains of Awash River and the adjacent non-riparian drylands. Sap flow monitoring stations were established at two sites in the floodplains of the Awash River and at two sites in the dryland area where soil moisture levels were low (Fig. 1). The study areas were representative of other parts of the Afar Region that are invaded by Prosopis.
Site 1 was located near Worer Agricultural Research Center some 200 m away from Awash River and about 30 m from a nearby irrigation canal, and was considered a floodplain site. The area was used for crop production until 2012 after which it was abandoned due to shallow water inducing soil salinity problems. Soon after abandonment, Prosopis invaded the area and established dense stands with 100% canopy cover in most places comprising trees up to 6.5 m height. Soil moisture was relatively high at Site 1 due to proximity to the river. The area invaded by Prosopis was about 6 ha and the soils were temporarily flooded loam and clay soils. Except for annual grasses in some open spaces, there was no undergrowth vegetation, probably due to the dense Prosopis cover.
Site 2 was in the floodplains of a tributary of Awash River and located near Berta locality (Fig. 1). There is a continuous flow of water from a pumped well to less than 10 m from the site where sap flow was measured. The soil is a sandy loam formation, which maintains relatively high moisture content. The Prosopis stand was about 3 ha in size, had a closed canopy and comprised trees of more than 5 m height.
Site 3 was on dryland in the former rangeland of Hallaideghe locality (Fig. 1), characterized by sandy loam soil formation. There was no surface water source except from rainfall and seasonal flooding from the West Harerghe highlands. This area is now invaded by Prosopis with a closed canopy and tree heights of close to 5 m. The size of the invaded area was about 100 ha.
Site 4 was in the drylands of Berta locality (Fig. 1). The soil is sandy with a high rocky outcrop. The dominant indigenous vegetation around this site consisted of a few Senegalia senegal (L.) Britton (or Acacia senegal (L.) Willd) and other small shrubs and grasses. The area is now dominated by Prosopis stands with a closed canopy with trees reaching up to 4 m height. At this site, a meteorological station was installed next to the sap flow monitoring equipment.
The sap flow data in this study therefore provide insight into how tree water use varied across the landscape including (a) floodplains, (b) dryland areas, and (c) across these two most heavily invaded habitats in the Afar Region. All experimental sites were fenced and protected from animal and human interference. Moreover, safety boxes were also made to protect the equipment from weather and other damages.
Tree and stand water use measurements
The amount of water used by individual Prosopis trees was determined using the heat ratio method (HRM) of monitoring tree sap flow31. This technique was selected because it is suitable for measuring low and reverse sap flows which are likely present in desert-adapted species such as Prosopis14,32. In total, four sap flow stations were established (Sites 1 to 4) with three trees instrumented per station. Trees with different stem diameters were selected to capture the variation in transpiration rates in the study region. Stem diameter of the instrumented trees was measured just below the branching at about 60 cm above the ground at all sites. Each sap flow station comprised a CR1000 data logger and an AM16/32B multiplexer, as specified by Campbell Scientific, Inc., Logan UT, USA. Each system was powered by a 70 Ah (12 V) rechargeable battery using 50 W solar panels. Four sets of heaters applied heat to each tree for 0.5 s every hour through a custom-made relay control module. Moreover, a pair of equally placed (0.5 cm) T-Type thermocouples was installed on either side of the heater to measure the sapwood temperature before and after pulsing the heat. With a precision drilling rig, two 2.0 mm diameter holes were carefully made for the thermocouples to minimize errors due to probe misalignments. Heater holes were about 1.8 mm diameter to ensure a tight fit to facilitate heat transfer to the wood during pulsing.
In this sap flow monitoring technique, the heat pulse velocity (Vh, cm/h) is logarithmically related to the ratio of temperature increases upstream and downstream from a heater (v1/v2) as shown in Eq. (1), Burgess et al31:
$$Vh=left(frac{k}{x}right)(mathrm{ln}left(frac{v1}{v2}right))times 3600$$
(1)
where Vh is heat velocity cm/hour, k is the thermal diffusivity which was assigned a nominal value of 2.5 × 10–3 cm2/s for wood, x is the distance (cm) between the heater and either temperature probe (~ 0.5 cm), and v1 and v2 are increases in temperature before and after pulsing31.
The thermocouples were installed in the sapwood at depths ranging from 0.8 to 1.1 cm under the bark to capture the radial changes in sap velocity. Wounding corrections were applied according to the method described by Swanson and Whitfield57. The depth of the sapwood was determined visually as it was possible to distinguish between the sap wood and heartwood boundaries from the changes in the color of the wood. The individual tree sap flow volume in liters per hour were converted to stand level transpiration (in mm per hour) using the approach described by Dzikiti et al14 in which the instrumented trees were assigned to a particular stem size class. The stand level transpiration was then calculated as a weighted sum of the transpiration rates by the trees in each stem size class with the proportion of trees in each size class as the weights. The volumetric soil water content in the root-zone of the trees were measured at each site using a single soil water content reflectometer probe (Model CS616: Campbell Scientific, Inc., Logan UT, USA) installed horizontally at a depth of 50 cm. Sap flow and soil moisture data were measured for 15 months (from November 2016 to January 2018) while evapotranspiration was measured for 11 months (from January to November 2017).
To study the dynamics of total actual evapotranspiration (ETa) from Prosopis stands, an open path eddy covariance (EC) system was installed on Site 1 and data were collected for 11 months from January to November 2017 as it was not possible to continue measuring for more periods due to equipment limitations. The EC was borrowed from Addis Ababa University only for one year so ETa from the other sites could not be measured. The EC system was the IRGASON system which comprised a sonic anemometer (Model: CSAT3A Campbell Scientific Inc., Logan UT, USA) that measured the wind speed in 3-D at 10 Hz frequency. The H2O/CO2 concentrations of the atmosphere were measured using an Infrared Gas Analyzers (Model: EC150, Campbell Scientific, Inc., Logan UT, USA). The collected data was stored by a data logger (Model: CR3000: Campbell Scientific, Inc., Logan UT, USA) on a Compact Flash card module (NL115 or CFM100). To quantify the changes in the energy balance of the study site, two other components of the surface energy balance were measured. These include the net radiation, which was measured using a single component net radiometer (Model: NR-LITE2: Manufacturer: Kipp & Zonnen, Delft, The Netherlands) that was mounted at the top of the tower (~ 7.5 m above the ground). The IRGASON sensor was installed outside the surface roughness layer of the canopy at an average height of about one meter above the Prosopis tree canopy. This ensured a uniform fetch around the tower with a flux foot print of about 100 m radius.
Air temperature and humidity were measured at high frequency using a temperature and humidity Probe (Model HMP155A-L, Campbell Scientific, 2013). The high frequency data were further corrected for 1) lack of sensor levelness (coordinate rotation), 2) sensor time lags, and 3) fluctuations in the air density using the EddyPro version 6.0 software (Li-COR, Nebraska, USA). Sensor separation corrections were not necessary as the IRGA and sonic are a single unit.
Allometric characteristics are one of the major biological factors affecting the eco-physiology of plant species. For example, sapwood area is usually correlated with stem diameter14. The sapwood area estimated from the stem diameter measurements was used to calculate the sap flow volumes from the sap velocity measured by the HRM system.
Weather and soil water dynamics
To measure solar irradiance, precipitation, air temperature, relative humidity and air pressure, an automatic weather station was set up at Site 4, which was located within 7 km from the other three sites. The solar radiation sensor was installed on a horizontal leveling fixture mounted on a south facing cross bar to avoid self-shading errors. A wind sentry was used to measure the wind speed and direction (Model 03,001, R.M. Young; Campbell Scientific, Inc., Logan UT, USA). Rainfall was monitored using a tipping bucket rain gauge (Model TE525-L, Campbell Scientific, Inc., Logan UT, USA). The weather station comprised an Em50 (a 5-channel data logger) and ECH2O utility software from Decagon, USA.
Wind speed was obtained from the weather station located at Worer Agricultural Research Center, which was about 500 m away from Site 1. The weather station had a temperature and humidity probe (Model CS500, Vaisala, Finland) installed at a height of about 2.0 m above ground and the station also measured wind speed using a cup or rotational anemometer installed at 2 m high.
The energy transferred into and out of the ground was measured using clusters of soil heat flux plates (Model: HFP01SC-L, Delft, The Netherlands), while soil temperature was recorded using soil averaging thermocouples (Model: TCAV-L: Campbell Scientific, Inc., Logan UT, USA). The soil heat flux plates were installed at 8 cm depth and the soil averaging and soil moisture data measured with the soil water content reflectometers (Model: CS616-L: Campbell Scientific, Inc., Logan UT, USA) were used to correct the soil heat flux for the energy stored by the soil layer above them. At all sites, the sensors were connected to a data logger (Model CR1000, Campbell Scientific, Inc., Logan UT, USA) programmed with a scan interval of 90 s, and data were stored at hourly intervals over the 11 months study period. All data were downloaded every 21 days from data loggers.
Drivers of water use by the invasive Prosopis
To identify the main drivers of water use by Prosopis invasions in the Awash River basin of the Afar Region correlations were sought between the various water use variables (transpiration and evapotranspiration) as dependent variables and microclimate factors, i.e. solar radiation, wind speed, vapor pressure deficit of the air (VPD), soil moisture, and ET0 as explanatory variables.
Upscaling Prosopis water use moderation to the Afar Regional level
To upscale the Prosopis transpiration and ET from the individual study sites to the regional scale, a regression equation was developed using the fractional vegetation cover information mapped by Shiferaw et al26. This mapping quantified the Prosopis distribution and cover in the Afar Region at a 15 × 15 m spatial resolution based on explanatory variables including Landsat panchromatic images, other biophysical parameters and field observations. The fractional cover map was generated using a robust modelling approach, Random Forest Algorithm, with a large amount of field observations (> 3000 plots) and seventeen explanatory variables33. Then, we estimated the amount of water used by Prosopis stands at each of the four sites in mm/day per pixel with 100% cover. This was extrapolated to all fractional cover levels per pixel indicated in the fractional cover map for Prosopis in the study area. Moreover, canopy cover from the experimental plots of Prosopis trees was estimated. Then, the relationship between sap flow and fractional cover as well as between ET and fractional cover (Fci) over the invaded area was developed as shown in Eq. (2):
$${text{f }}left( {text{x}} right) = sum ({text{Fci}}({text{Wi}})S)$$
(2)
where f(x) is either total water use (sap flow) or total ET in mm/day over the whole study area; Fci is fractional cover at pixel level i, Wi is water use either from sap flow or stand ET at 100% canopy cover, and S is a pixel size of 225 m2.
Finally, we estimated the financial costs incurred from the loss of water through Prosopis transpiration and ET. This was done by taking the water charge of payment for ecosystems services by investors to Awash Basin Organization which was set at US$ 0.00015 per m3 according to Ayana et al34. Also, we estimated the market price and the net benefits of cotton35,58 and sugarcane59, which are major crops grown in the study area, which could be grown with the amount of water used by Prosopis.
Data reduction and statistical analyses
LoggerNet 4.1 (Campbell Scientific, Inc, Logan UT, USA) was used for downloading sap flow data from data loggers to the laptop and for converting the data to 30 min interval values. EddyPro 6.0 (Licor Nebraska, Lincoln, USA) was employed for processing the high frequency EC data used for calculating ET. Sap flow rates were calculated following Burgess et al31. The FAO Penman–Monteith Eq. (3) was used to calculate the hourly and daily reference evapotranspiration (ETo) using the weather data. Multiple linear regressions were carried out using either sap flow or ET as response variable and solar radiation, soil moisture, wind speed, vapor pressure deficit and potential ET as explanatory variables at a time in an open source R software version 3.3.360. Maps were made using an open source Quantum GIS (QGIS3.8.3) software61.
Source: Ecology - nature.com
