The history of P research in the WANA region broadly parallels that in Europe and North America, but only beginning later and with considerably less fertilizer being applied. In essence, it had its beginnings in the 1970s with the advent of chemical fertilization (Mataret al.,1992; Ryan, 1983). Since that time our knowledge of both fundamental and applied aspects of P followed a predictable course. In charting the evaluation of P research, we have arbitrarily described it in two categories: (1) research at the Interna- tional Center for Agricultural Research in the Dry Areas (ICARDA), one of the main centers of the global international agricultural research network (Deaneet al., 2010), primarily in Syria as proxy for soil and climatic condi- tions and cropping systems in the Mediterranean region as a whole and (2) research conducted in countries of the region, often in collaboration with ICARDA. The review does not purport to exhaustively reflect the entire body of P research in the entire WANA region, but rather seeks to broadly reflect the progression of P research that typifies the region as a whole.
5.1. ICARDA’s research in Syria
While there was soil limited soil research in Syria prior to the establishment of ICARDA in 1977 (Ryan, 2002), the bulk of P research has occurred since then, especially in the following two decades. As most of the work necessarily of an applied nature and conducted with crops in the field, experimental stations and on farms, the research efforts were backstopped by observations under field conditions and by laboratory soil incubation studies. The applied research program involving nutrition of field crops was enhanced by the establishment of field stations that represented the range of rainfall conditions typical of both Syrian and Mediterranean rainfed crop- ping conditions (Ryan et al., 1997). While the available soil P levels were initially low, and therefore, suitable to fertilizer P response trials, a gradual buildup of available from routine fertilization in many of the stations variety trails invalidated on-station trials involving P, except in limited sections of the station. The ensuing research was validated by on-farm trials.
A chronology of development in P research is pertinent to this review.
5.1.1. Field responses to phosphorus
Following the early highlighting of the potential importance of P for cropping systems involving cereals and legumes (Saxana, 1979), while early observations of Harmsen et al. (1983) indicated that responses to N were highest at the higher rainfall site (480 mm) and responses to P were highest at the low rainfall site (270 mm), thus suggesting an interaction of P with soil moisture. Research at the experimental stations not only showed the value of P fertilization for yields, but also for increasing root growth and water-use efficiency (Brown et al., 1987); notably, this work showed no differential response of cereal varieties to P fertilizer. Focusing on the drier sites,Gregoryet al. (1984)showed the importance of P fertilization for early root growth and water-use efficiency. Similarly, data from a 4-year study (Cooper et al., 1987b) showed that both N and P increased water-use efficiency from 16.4 to 33.5 kg ha1 mm1 in 1980/81 (299 mm) and from 7.8 to 12.0 kg ha1mm1in 1983/84 (204 mm), thus highlighting the overall importance of balanced fertilization for crop yields and efficient water use.
On-farm trials gave a considerable boost to the on-station P research.
A major 4-year study involving 75 trials in the mainly barley-growing area (200–300 mm) of northern Syria (with yearly trial site rainfall data of 136–568 mm yr1) showed that virtually all sites responded to fertilization, again noting the relative importance of P in lower rainfall sites (Jones and Wahabi, 1992). These studies underpinned a change in Government policy toward allowing fertilizer allocation in drier areas (Mazid and Bailey, 1992). A similar extensive series of on-farm trials for 4-years in the favorable rainfall zones (300–600 mm yr1), mainly with wheat, underlined the importance of rainfall in dictating crop yields, with fertilization having a secondary influence (Palaet al., 1996); in contrast to the barley study, responses to P were low, probably reflecting higher soil P values from fertilizer use in more favorable areas. In a study of vetch/barley mixture, Wahbi et al. (1993) indicated another dimension of fertilizer use, with P showing a residual effect in the second year but not for N. Given the importance of forage legumes, P was also shown to enhance forage yields and quality (Osmanet al., 1993).
While most trials involved field crops, the potential of P fertilization on rangeland and nature pastures was investigated byOsman and Cocks (1992).
At a P-deficient, shallow stony site, a small application of P (11 kg ha1), broadcasted as superphosphate, had a significant impact on forage growth as well as on yield parameters from sheep (meat, wool, milk); improved forage quality was evident even 7 years after P application. Stemming from the studies at ICARDA, a more recent study of P fertilization of rangeland was conducted in neighboring Lebanon in an attempt to reverse degradation that is so commonly associated with public areas designated for communal grazing (Haddadet al., 2010). Significant increases occurred in forage yields as well as in increases in the proportions of native legume species, as noted in
the studies from Syria. Despite such positive responses to P fertilization of rangeland, most of which are inherently low in available P, adoption of the practice of fertilizing such areas is unlikely because of the large extent of rangeland and common ownership at the community or tribal level.
The 1990s witnessed considerable research momentum to improve ani- mal production in the WANA region by the introduction and development of forages cultivated in association with cereals. Particular emphasis was given to the self-regenerating medic pasture species. Despite the positive advantages of medic in cereal rotations in terms of enhancing cereal yields (Ryanet al.,2010) as well as SOM and available soil N (Ryanet al., 2008b), the effort to introduce medics in the Mediterranean region agriculture was to fizzle out for various technical and economic reasons (Christiansen et al., 2000). Nevertheless, P fertilization aspect of these forages was addressed in parallel studies in the 1990s.
In a greenhouse study with and without nodulation, and with Zn, all four medic species that were assessed responded to P and Zn (Materon and Ryan, 1995). In the follow-up field trial (Materon and Ryan, 1996), vetch (Vicia sativa) and medic, but not chickling, also known as grasspea (Lathyrus sativus), responded to P at a site where available Olsen P was low (2.1 mg P kg1) but not where P was higher (4.2 mg P kg1). The effect for Zn was positive for all species, but only where the test level (DTPA) was low (0.6 mg Zn kg1). This was one of the first trials that indicated the importance of Zn and P nutrition, as well as the significance of soil test levels. A response to P occurred only when the soil test level for Zn was adequate (i.e., above 0.5 mg kg1) while responses to applied Zn were observed when soil P test levels were adequate. There was no indication of any negative effect of P fertilizers on soil Zn, an interaction that was prominent in the early soils’ literature.
5.1.2. Soil-related studies
Parallel to crop response studies, much of the research sought to relate such responses to soil P and to interpret field observations by soil studies. The basic element in any soil-testing program is to identify a suitable extractant and procedure to reflect the pool of available P for crop uptake, thus serving as a guide to the needs for fertilization followed by appropriate calibration of application rates in field studies. Confirming previous works on calcareous soils in Lebanon (Ryan and Ayubi, 1981),Mataret al. (1988) showed that the well-established Olsen procedure was appropriate for Syria (as well as countries of the region); universal soil tests such as NH4HCO3-DTPA (Soltanpour, 1985) were also considered appropriate. The focus of such efforts involved defining a “critical” soil test level below which P is deficient and above which P is not needed; in essence, such tests reflected the degree of deficiency/sufficiency and thus provided a basis for rationale fertilizer application rates.
The soil test calibration program in Syria embraced countries of the region, with common protocols, workshops, and related reports and pro- ceedings (Matar et al., 1988a; Ryan, 1997; Ryan and Matar, 1990, 1992;
Soltanpour, 1986). Notwithstanding differences between crops and soil types, a critical Olsen P range of 5–7 mg kg1 was generally observed.
Where P levels were less than 5 mg kg1application was recommended at 25 kg P ha1, while between 5 and 10 mg kg1, the amount of P was reduced to 10–18 kg P; above 10 mg kg1, P fertilizer was not advised. The numerous studies that emerged from ICARDA’s soil test calibration pro- gram highlighted the importance of various factors in relation to crop P responses: moisture, temperature, soil type, and crop species. A recent study (Ryanet al., 2008c) indicated the importance of soil depth—and therefore root volume—in controlling P fertilizer response.
An important factor related to P fertilizer use efficiency is the application method. With developments in machinery and the combine drill, banding or placement of P fertilizer is now increasingly adopted and is highly efficient, especially where soil P test levels are low (Ryan, 2002). While broadcasting was—and still is—the norm in Syria,Matar and Brown (1989a)compared broadcasting and banding at ICARDA’s stations using soil and crop growth and uptake measurements: banding at 17.5 kg P ha1 was as effective as 52.5 kg P ha1broadcasted in terms of early growth stimulations and yield increases. In addition, root growth was greater in the banded zone and the enhanced uptake was attributed to better root-fertilizer contact in the banded zone. However, soil measurements showed that P availability decreased with time and was independent of application method (Matar and Brown, 1989b).
Using fertilized soil in metal cylinders embedded in the field for 10 months, Matar (1990)found a predictable pattern of rapid initial reversion followed by a slower phase as observed in similar laboratory studies in Lebanon (Ryan et al., 1985a,b).Matar (1990)calculated “half lives” for soil-applied P, being 5 months at the wetter site (Jindiress, 480 mm) and 10 months at the other two sites, Tel Hadya (340 mm) and Breda (270 mm), with predicted values being close to actual observations. However, one has to bear in mind the discrepancy between such experimental conditions with controlled undis- turbed soil volumes and actual conditions under field cropping.
FollowingMatar and Brown’s (1989b)field study on P transformation, a related laboratory study of 19 soils considered changes in P availability up to 240 days at P rates of 20–500 mg P ha1 (Afif et al., 1993). After the anticipated pattern of an initial rapid removal of soluble P from solution, there was no change in the availability index after 60 days’ incubation. The change in the index was negatively related to Fe oxides and cation exchange capacity. However, the reaction at 500 mg kg1was correlated only with CaCO3. The study again questioned the value of such long-term incubation studies to simulate what happens in the field; clearly, any controlled study of this nature has its shortcomings in that regard. A similar laboratory study
(Bakheit and Dakermanji, 1993) showed adsorption to be related to clay content, while desorption was a high as 25% of adsorbed P.
While many soil and environmental factors influence soil P reactions in the short and long term, one incubation study examined the effect of organic matter on P chemistry by the addition of manure (% volume) in a 19-month incubation study conducted out-of-doors in pots (Habibet al., 1994). The rate of reversion of soluble P was slower in the presence of the added organic matter, which also reduced P sorption and increased desorp- tion in laboratory studies of the treated soil samples. However, in an incubation study, samples from a long-term rotation trial where SOM had been increased by legumes in the cereal-based rotation (1.10–1.32%), there was no differential effect of the SOM on P behavior (Kabengiet al., 2003), nor was there any marked influence of varying temperature, moisture, or wetting-drying to simulate field conditions; in that respect because of its reactivity, P behaves differently to N in response to such factors.
The many studies involving P in Syria invariably involved conventional sources such as single superphosphate or triple superphosphate in green- house and laboratory studies. Though newer and potentially more efficient sources have been considered (Chien et al., 2009), work on alternative nonconventional P sources in WANA was limited to a study of rock-P which was shown to be ineffective (Habibet al., 1999) a result consistent with the abundance of CaCO3 in such soils. Clearly, soil pH has an important bearing on the effectiveness of rock phosphate, as well as the crop type. In a study (Bekele and Ho¨fner, 1993) of a soil with pH 5.9 from Ethopia, rape (Brassicia napus L.) was able to effectively use P from both
“reactive” and “nonreactive” rock-P sources; however, barley was unable to use P from the nonreactive rock-P source.
Earlier work on acid-generating P fertilizer in Lebanon had implications for irrigated cropping. Ryan and Tabbara (1989) had shown that an experimental material, urea-phosphate, could increase infiltration in sodium-affected soils in the laboratory. Related laboratory studies showed considerable P movement from the point of emitters with solutions of urea- phosphate and phosphoric acid (Ryanet al., 1988a), but urea phosphate had little impact in eliminating Fe chlorosis in soybeans (Glycine max) grown in highly calcareous soils. While acid-producing materials are unlikely to be used in dryland cropping, they have potential to be used in pressurized irrigation systems, both as a source of P and a means of removing precipi- tates in drip lines (Ryan, 2000). Other diverse sources of P, which have, however, considerable use potential include municipal wastewater. While this water is increasing and is a source for irrigation, it is also a valuable source of nutrients, including P (Ryan et al. 2006b). Nevertheless, the dangers of using untreated sewage water cannot be overlooked; the argu- ment of wastewater treatment plants is compelling from the environmental and natural resource standpoints.
While the quest to improve the efficiency of applied fertilizer P mainly centered an application method, some limited attention was given to indirect influence of soil mycorrhizae on P nutrition of crops. Thus, a pot experiment showed that inoculation of chickpea with vescicular–arbuscular mycorrizal fungi (VAM) improved plant growth and doubled P uptake at low and intermediate (3, 6, 12 mg kg1) P application rates in sterilized P- deficient soil (Weberet al., 1992). However, at the higher P application rate (24 mg kg1), there was no significant difference between inoculated and non-inoculated plants. At all P rates, mycorrhizal plants had higher shoot P concentrations than non-inoculated plants. However, mychorrizal infec- tion was not significantly affected by P application. On the contrary, the companion field trial showed that neither P nor mycorrhizal inoculation influenced any plant parameter; infection rate was high at flowering stage irrespective of treatment. This may have been adequate to ensure efficient P uptake under field conditions. The conclusions from these experiments are consistent with research elsewhere; effects are only observed in sterile conditions, with little encouraging results from field trials and little success with inoculation.
Another factor that cannot be overlooked in order to improve P fertil- izer efficiency is the actual spatial variability that exists in farmers’ fields.
Uniformity in terms of soil nutrients and other soil properties is rare under cropping conditions in the WANA region (Abdel Monem et al., 1989).
Variability in the field is attributed to inherent soil differences as well as erosion and management factors such as animal dropping, selective grazing of plant species, land application of fertilizers and manures, uneven crop stands, etc. (Ryanet al., 1998). Applying a standard rate of P to a variable field is inefficient, with too much applied where soil P levels are high and too little in parts of the field where levels are low.
The concept of variable fertilizer application is unlikely to find any application in the region given the small field sizes and the general absence of soil test information; such approaches are now widespread in the US and elsewhere. Spatial variability at experiment stations (Ryan et al., 2011), especially for P, has implications for field plot research and is not easily overcome by standard cropping or “homogenization.” Another dimension of P variability was illustrated in the station survey ofRyanet al. (1997)in terms of depth-wise distribution. While the classical decrease was apparent, but such profiles varied significantly not only in the stations across the rainfall transect but also within the stations. This was one of the few cases where total P was considered, in addition to the available P form.
Most studies of soil and plant P in Syria involved conventional approaches. Notwithstanding its advantages, the use of 32P in P studies is rare in most developing countries, largely because of costs, logistic and safety considerations, and its limited half-life. However, one study of a typical soil in northern Syria (Calcixerollic Xerochrept) used32P to study the fate of P
applied as superphosphate in a greenhouse study (Asfary et al., 2004).
Isotopic measurements showed that about 50% of the applied P was non- exchangeable within 2 days after application. However, this rapid transfor- mation is at variance with most other studies that show that much of the added P is plant-available over time (Syers et al., 2008). Thus, caution should be exercised in extrapolating the results of brief studies under controlled conditions to field studies. Applied 32P was recovered in the various inorganic fractions, with Ca–P being dominant (68%) followed by occluded Fe–P (48%). While such sequential fractionation schemes indicate the dominance of CaCO3on soil P forms, there are limitations of chemical extractants to describing discrete P fractions.
5.1.3. Buildup of residual soil phosphorus
As in any area of scientific endeavor, the acquisition of new knowledge inevitably leads to a revision of old concepts. Nowhere is this more in evidence than in the case of P fertilizer reactions in soils. The prevailing thinking that emerged from the early laboratory and greenhouse and short- term field P studies was that added P was irreversibly fixed by soil compo- nents and that its efficiency of crop uptake was low, typically 5–15% of the P applied. However, with longer-term field studies, the concept of soil P buildup emerged, and along with it the value of residual soil P for crops in subsequent years following the initial season (Aulakhet al., 2003; Barrow, 1980). The values of P from previous years of fertilization were clearly shown by Johnston (2000) with observations from the long-term agro- nomic trials at Rothamsted. We now recognize that such P residues can even pose a threat of environmental pollution (Tunneyet al., 1997). The current concept of residual soil P has been revisited by Syers et al. (2008) following the reassessment of data from long-term P fertilization studies.
The authors used the “balance method” for calculating recovery and effi- ciency of fertilizer P as it takes residual P into account. While recognizing the equilibria that exist between various P pools, as illustrated byJohnston (2000), Syers et al.(2008) arrived at P-use efficiencies as high as 90% and concluded that P is not irreversibly fixed in soils as was previously thought.
By comparison with other regions of the world, such as in the tropics where soils are highly weathered, acidic, and high in Fe and Al, the mainly calcareous soils of the WANA region are not high “P-fixing” (Delgado and Scalenghe, 2008). However, reaction of such soils with soluble P fertilizer does result in reduced solubility and plant availability (Delgadoet al., 2002a).
Consequently, the observations of a buildup of P experiment stations (Ryan et al., 1997) and in farmer’s fields even from modest fertilizer P applications (Wahbiet al., 1994) were not surprising.
Such observations led to the establishment of a 10-year field trial that assessed current and residual P under the prevailing cereal/legumes rotation system at ICARDA’s three experimental stations across a rainfall range