The Ralston study is located at the long-term research plots located southwest of Ritzville, WA, which was utilized to assess effects of soil conservation and fertilizer efficiencies. The long-term annual precipitation for the study site is 280 mm yr⁻¹. Information about previous crop rotations and site characterization were published by Young et al. (2015). In 2012 and 2013, the study contained four cropping systems, each replicated four times in a randomized complete block design. There were two complete sets of plots with fallow and crops present in any given year. Individual plots were 7.9 × 152 m. Cropping systems included standard-height (i.e., tall) WW (cv. Farnum) following conservation tillage fallow harvested with a cutter bar header, tall WW following no-till chemical fallow harvested with a stripper header, winter triticale (Triticosecale hexaploid L. cv. Trimark 099) following no-till chemical fallow harvested with a cutter bar, and winter triticale following no-till chemical fallow harvested with a stripper header.

A long-term experiment was initiated in 2003 at the Oregon State University Sherman Station near Moro, OR, to evaluate traditional winter wheat-summer fallow cropping system under conservation tillage, intensified, and diversified cropping systems using no-till practices (Machado et al., 2015). Average annual precipitation at the site is 289 mm. The experimental area consisted of 42 plots, each 15 by 105 m, with 14 treatments of eight crop rotations in a randomized complete block design with three replications. In addition to WW-summer fallow, WW-no-till chemical fallow, WW-spring barley (Hordeum vulgare L.)-chemical fallow, WW-winter pea, and continuous WW were evaluated. All phases of the rotations were present every year. Details on management are reported by Machado et al. (2015).

A long term experiment was initiated in 1997 on the Ronald Jirava farm near Ritzville, WA, to evaluate diverse cropping systems using no-till and conservation-till management (Schillinger et al., 2007; Schillinger and Paulitz, 2014). Average annual precipitation at the site is 292 mm. The experiment consisted of 56 plots of 9 by 150 m, six crop rotation treatments with all phases of all rotations present each year in a randomized complete block design with four replications. The traditional rotation was WW-summer fallow. Alternative rotations were WW-spring wheat (SW)-summer fallow, WW-safflower (Carthamus tinctorius L.)-summer fallow, winter triticale -SW-chemical fallow, continuous no-till SW, and no-till SW-spring barley. Further details on treatments and management of the experiment are reported by Schillinger and Paulitz (2014).

Three- and four-year rotational field experiments were initiated at Davenport, WA, in 2011 to evaluate intensive cropping systems under no-till practices diversified with oilseeds and pulses. Average annual precipitation for the site is 353 mm yr⁻¹. Field plots, measuring 3.7 × 15 m, were established in a randomized complete block designed with four replications with each phase of the rotation present each year, with 96 plots and 5 treatments. Data were collected following the initiation of the rotations in 2011.

Three-year field experiments were initiated at Pullman, WA and Davenport, WA, following wheat in 2011 and 2012 (Pan et al., 2016a) to determine the rotational effect of pulses on WW. The long-term average annual precipitation at Pullman is 517 mm yr⁻¹. Field plots, measuring 2 by 15 m, were established in a randomized complete block designed with four replications and seeded to spring canola (Brassica napus L.). In the following spring after canola, plots were split longitudinally (1 by 15 m) and were randomly seeded with inoculated dry field pea (Pisum sativum L.) or SW. In the third year, WW was direct-seeded across the existing split-plot design following spring pea or SW harvest.

In 2013, a subsequent 2-year field experiment was initiated at Pullman, WA and Davenport, WA, to determine the effect of oilseed vs. pulses vs. SW on following WW yields. Field plots, measuring 2 by 15 m, were established in a randomized complete block designed with four replications and seeded to spring canola, spring pea, and spring wheat. In the following year, WW (cv. Madsen at Pullman and Otto at Davenport) was direct-seeded across the existing plot design following spring crops.

A long-term experiment at the University of Idaho Kambitsch farm, north of Genesee, ID, was established in 2000 to study the impact of conservation tillage on WW (Johnson-Maynard et al., 2007). The average annual precipitation at the farm is 695 mm yr⁻¹. Prior to 2000 the land was managed using conventional methods. Tillage treatments (chisel plow and no-till, each with plots 20 × 80 m) were replicated four times across a hillslope running east to west (across the slope). Each tillage treatment was split into three crop zones planted to either pea/chickpea (Ciser arietinum L.), spring barley/wheat, or winter wheat. Tillage plots were 18 × 80 m. Crop subplots are 6 × 80 m (1.2 m alley between tillage strips).

A site-specific, precision agriculture study was initiated in 2011 southwest of Colfax, WA, southeast of Troy, ID, southeast of Genesee, ID and in Leland, ID. The project focused using remote and proximal sensing to characterize and manage field-scale variability in soil water, N, and crop response across the landscape rather than plot-based research. Average annual precipitation is 484 mm yr⁻¹ at Colfax; 602 mm yr⁻¹, Genesee; 675 mm yr⁻¹, Troy; and 721 mm yr⁻¹, Leland. Within each farm, two catchments were selected such that the entire drainage area was captured within a single field. At each farm, one of these catchments were selected for intensive automated and manual monitoring. Each site was equipped with 12 spatially representative subsites which serve as the primary sampling locations within the watershed. A second catchment was reserved for validation purposes and was only monitored over the last 2 years of the project.

A 3-year field study was initiated in Prosser, WA, following wheat 2011 to assess conservation tillage and cover cropping effects on irrigated WW. Average annual precipitation at Prosser is 227 mm yr⁻¹. The cropping sequence was corn (Zea mays)-potatoes (Solanum tuberosum)-winter wheat. Winter wheat was irrigated with 458 mm in the spring and summer. Field plots measuring 4.9 by 15 m, were established in a randomized complete block design with four replicates. The treatments were reduced tillage, reduced tillage winter cover crop, conventional tillage, and conventional tillage winter cover crop. Winter cover following sweet corn was triticale (× Triticolescale) and following potatoes was mustard (Brassica hirta). The conventional tillage treatments were chisel disked prior to planting and post harvesting of every crop. The reduced tillage sequence was only chisel disked prior to planting potatoes and following potato harvest.

At Ritzville and Davenport, WA, composite soil samples from three replicates were taken every 30 cm down to 120 cm (or to an impermeable layer) with a giddings probe after harvest. At Moro, WA, soil samples were collected to a depth of 30 cm and composited. Ammonium and nitrate N were analyzed following KCl extraction and measured colorimetrically with a Quickchem 8000 Series FIA+ system and AutoSampler (Lachat Instruments, Hach Company, Loveland, CO). Total C and N of composted soil samples collected in the first year of each study were measured by combustion in a CN analyzer.

At Prosser, Moro, Ritzville, Ralston, Davenport, Pullman, and Kambitsch, grain was harvested using commercial plot combines. Total above ground biomass was sampled from 1 m² area at maturity and prior to determining harvest index to calculate residue biomass and N yields. Biomass samples were dried at 45–60°C for 48 h, weighed, and threshed with a Vogel Stationary Grain Thresher, from which seeds, chaff, and stems were collected. Seeds were weighed to determine the harvest index. Seeds were ground with a Cyclone Sample Mill (Thomas Scientific, Swedesboro, NJ) for C and N analysis with a C/N autoanalyzer (LECO Corp, St. Joseph, MI) at all sites except for Moro, in which grain protein was measured using the Inframatic 9200 (Perten Instruments, Hägersten, Sweden). Residue yields were calculated from the combined seed yields by applying the harvest index. Residue samples were ground with a Thomas Wiley mill (Thomas Scientific, Swedesboro, NJ) prior to C and N analysis (for all sites except Moro) by combustion using a Truspec Carbon and Nitrogen Analyzer (LECO Corporation, St. Joseph, MI). At Colfax, Genesee, Troy, and Leland locations, total above-ground biomass was determined at each of the 12 locations at harvest by hand-harvesting four 1 m² plots at each site, and total grain yield within each 1 m² locations was determined by threshing. All grain samples were analyzed for protein using Infratec 1241 Grain Analyzer (Foss, Hillerod, Denmark). Total nitrogen in the above ground residue was determined using a Truspec machine (LECO Corporation, St. Joseph, MI).

The mean, standard deviation (SD), and coefficient of variation (CV) were assessed for WW yields within and across all site locations. An analysis was conducted to assess variation in yields from traditional rotations (Table 1), in which fallow periods preceding WW when annual precipitation was <350 mm yr⁻¹, under both conventional and conservation tillage. Annual precipitation was calculated on a crop-year basis from September to August. In a separate analysis, variance was assessed for WW in alternative rotations when annual precipitation was <350 mm yr⁻¹ in which the fallow frequency was reduced to once every 3 years or WW (i.e., no fallow) followed crops. An analysis of variance was conducted with location as the fixed effect and annual precipitation, cropping sequence, and replicates as the random effects using the lme package in R (R Core Team, 2016).

To determine the effect of fallowing prior to WW, the response of WW yields to increasing annual precipitation were best fitted with a quadratic function, and two yield response curves were derived, including WW following crops vs. fallow when annual precipitation was <350 mm yr⁻¹. To determine the efficiency of fertilization across the region in all systems, the relationship between annual precipitation and N fertilization rate, as well as grain N accumulation, was also best fitted with quadratic equations. These relationships were reassessed after annual precipitation was adjusted for the available water holding capacity of the soil, using data obtained from USDA NRCS Web Soil Survey (Soil Survey Staff, Natural Resources Conservation Service and United States Department of Agriculture, 2016). The regression-based predicted yields following fallow vs. recrop, as well as grain N accumulation vs. N fertilizer rate, were assessed by differencing the definite integrals (to calculate the areas underneath the curve) for precipitation ranges of 200–350, 350–500, and 500–650 mm yr⁻¹.

The comparative effects of cropping sequence on WW yields (WW after fallow vs. cereals/pulses/oilseeds), and WW following wheat vs. other cereals/pulses/oilseeds) were assessed using data collected in the iPNW described above relative to data obtained from a literature survey (Appendix 1). The literature survey included, but were not limited to, references provided by Angus et al. (2015) with a particular focus on wheat-fallow. The effect of tillage on wheat yields was also assessed for the iPNW. Winter wheat yields following break crops (i.e., non-wheat cereals, pulses, or oilseeds) were regressed against WW yields after fallow or SW/WW. Break crops included other cereal crops (barley and triticale), oilseeds [including canola, rapeseed, and camelina (Camelina sativa L. Crantz)], and pulse crops [including field pea, lentil (Len culinaris L.), garbanzo beans (Cicer arietinum L.), and lupin (Lupinus L.)]. In addition to the data collected in the present study, WW yields following break crops vs. wheat were compiled from iPNW data presented (Guy, 2013). For tillage comparisons, WW yields under conservation tillage were regressed against WW yields under conventional tillage for selected sites.

Linear regressions were used to assess the relationships among yields, fertilizer N, plant N, and grain N for WW after fallow (only when annual precipitation was <350 mm yr⁻¹) vs. recropped WW for all sites. An analysis of covariance was conducted to determine interaction of cropping sequence with N responses, specifically the relationships between yield vs. fertilizer, grain N vs. fertilizer N, yield vs. grain N, and grain N vs. plant N using the stats package in R (R Core Team, 2016).

For cropping system level comparisons, energy outputs, total fertilizer inputs, and total grain N exports were summed over 3–4 years of observations. Annualized values were calculated to account for differences in the years of observations. Energy was calculated by multiplying grain yields by a conversion factor: legumes and cereals, by 15 MJ kg⁻¹, and oilseed, by 25 MJ kg⁻¹ (Zentner et al., 2004; Farine et al., 2010; Unakitan et al., 2010; Mousavi-Avval et al., 2011; Keshavarz-Afshar and Chen, 2015). The variance of annualized energy outputs, fertilization, and grain N export was compared across increasingly intensified cropping systems at Moro, OR, Ritzville, WA, and Davenport, WA, all of which received <350 mm yr⁻¹ within the study period. Finally, an analysis of covariance was conducted to determine interaction of cropping sequence with rotational N responses, specifically the relationships between total energy output vs. total fertilizer input, total grain N export vs. total fertilizer input, total energy output vs. total grain N, and total grain N export vs. total plant N uptake using the stats package in R (R Core Team, 2016).

At Moro, Ritzville and Davenport, multi-year cropping system N budgets were constructed. Inputs included pre-plant inorganic N summed in the 120-cm soil profile (or 30-cm at Moro), total N fertilizer applied, and estimated N mineralized based on soil organic matter (described by Pan et al., 2016a). Outputs included total crop N uptake and post-harvest inorganic N in the 120-cm soil profile. Nitrogen balance was determined by subtracting N outputs from N inputs. The fertilizer N balance was calculated by subtracting rotational grain N from total fertilizer N inputs.

In the iPNW, water is the most limiting resource in crop production due the region's semi-arid Mediterranean-like climate and steep rainfall gradient within the rain shadow of the Cascade Mountains (Schillinger and Papendick, 2008). As a result, yield potential of WW varies along the regional gradient in annual precipitation (Cook, 1986). Schillinger et al. (2008) determined that modern wheat cultivars can reproduce (e.g., yield grain) with 61 mm of available water in the iPNW, and WW yields improved by almost 20 kg ha⁻¹ for every millimeter gain in available water as available soil water increased from 60 to 350 mm. However, a regional assessment of WW grain yield along the full extent of the annual precipitation gradient in the iPNW is lacking.

We found that WW yields varied considerably across and within the 11 locations (Figure 1A) with annual precipitation ranging from 220 to 650 mm yr⁻¹, with an overall mean of WW in traditional rotations of ~6,000 kg ha⁻¹ and a CV of 31% (Figure 1B, Table 2). The regional variability in WW yields was consistent from 2012 to 2015 (Figure 1D). Annual precipitation and field position (plot or replicate) contributed similarly to the variability in WW within sites (Table 2). Winter wheat yield means were similar to or higher in the traditional fallow rotations than intensified, alternative rotations (Figure 1C), and the stability of WW yields declined, as indicated by the increase in CV, in intensified rotations (Table 2).

Winter wheat yields responded significantly to annual precipitation (Figure 2A). Summer fallowing prior to WW reduced the yield response to annual precipitation in comparison with intensified WW production following cereal, oilseeds, and pulse crops substituting summer fallow. As a result, annual precipitation explained more of the variation in the WW yields when following crops (R² = 0.53) than fallow (R² = 0.31). While the fallow frequency (every other year vs. once every 3 years) did not affect the yield response, foregoing fallow prior to WW resulted in an integrated loss of 263 Mt of WW yield ha⁻¹ when annual precipitation ranged from 200 to 350 mm yr⁻¹. Ultimately, WW production in the fallow cropping system is restricted by stored soil water plus April, May, and June rainfall; overwinter stored soil water dependent upon soil depth. Furthermore, previous studies have shown that the precipitation storage efficiency of fallowing in the iPNW is generally around 30% (Wuest and Schillinger, 2011).

As annual precipitation increases, the topography becomes increasingly sloping, and soil biology changes. Earthworms may be observed above 330–370 mm of mean annual precipitation (Walsh and Johnson-Maynard, 2016), and soil organic matter increases in the topsoil (Morrow et al., 2017). We determined that WW yields reached a maximum of 6,800 kg ha⁻¹ at 520 mm yr⁻¹, within the annual cropping region where precipitation is adequate to economically support crops every year (Figure 2A). The decline in WW yields after reaching the maximum may be due to various factors that contribute to an increase in the actual yield gap under more favorable conditions, such as increasing incidences of pests, disease, crop lodging, and nutrient deficiencies (Cook, 1986). Furthermore, topographic complexity varies across the region leading to differences in soil properties (Horner et al., 1957). As precipitation increases, the presence of finer-textured silty-clay and clay-loam soils with dense (~1.65 Mg m⁻³) argillic and fragipan horizons at depths from 0.2 to 1.2 m below the soil surface can restrict vertical drainage and root penetration leading to the development of seasonal perched water tables (McDaniel et al., 2001; Brooks et al., 2012). These argillic and fragipan soils are generally found where mean annual precipitation exceeds 600 mm and have a limited soil water holding capacity due to the presence of dense subsurface clays (Brooks et al., 2012). In the Paradise Creek Watershed of North Idaho, which receives 650 mm annual precipitation yr⁻¹, precipitation use efficiency is reduced by both restrictive layers, as well as winter runoff when soils saturate and frozen soils reduce infiltration (Brooks et al., 2010). Timely May and June precipitation is particularly critical in wheat production across the iPNW (Schillinger et al., 2008). Therefore, the soil constraints in the highest precipitation zone add additional risk to annual cropping as these shallow argillic and fragipan soils dry out faster and experience more water stress without adequate precipitation in May and June. The constraint of available soil water in the highest precipitation zones is evident when annual precipitation is adjusted for available water holding capacity (Figure 2B), and yields attain a plateau with the range of observed data rather than declining.

The interaction between WW yields and N fertilizer use efficiency has also never been examined across the extent of the iPNW's precipitation gradient. However, the regional soil and climatic properties have also served as the basis for the hypothesized interaction between N use efficiency and annual precipitation outlined by Pan et al. (2007). As precipitation increases in the region, root growth and yields become less limited by drought conditions, and N use efficiency increases as plants recover increasing amounts of fertilizer N from the topsoil. Therefore, fallow periods are practiced to mitigate inefficiencies. A decline in N use efficiency was predicted by Pan et al. (2007) under increasingly high precipitation due to expected N losses in runoff, leaching, and denitrification pathways, in addition to limitations in rooting depth due to the presence of restrictive subsoil clay layers.

Our findings support these hypothesized interactions. First, annual precipitation explained most of the variance in fertilizer N rates (R² = 0.67), and 39% of the variance in the accumulation of grain N (Figure 2C). Grain N increased with annual precipitation, prior to reaching a maximum at 480 mm yr⁻¹ before declining. Similar to yield trends, the decline in grain yield at high levels of precipitation was moderated by adjusting for available soil water holding capacity (Figure 2D). The regression analyses predicted similar amounts of N were applied as fertilizer and exported in grain when annual precipitation ranged from 200 to 350 mm, corresponding with grain N to fertilizer ratios were >1. These high fertilization efficiencies were the result of residual N carryover and the recycling of mineralized N, particularly for WW after fallow. As a result, fallow periods were not only instrumental in maintaining high yields through soil water storage, but also to reducing N deficiencies, as discussed by Pan et al. (2016a). High efficiencies of fertilization were also observed between 350 and 500 mm yr⁻¹, in which the integrated grain N export exceeded fertilizer N by 2,000 kg N ha⁻¹. However, fertilization inefficiencies declined when annual precipitation exceeded 480 mm. The regressions predicted greater amounts of fertilizer were applied than removed in grain, with an integrated fertilizer N deficit of 2,650 kg N ha⁻¹ from 500 to 650 mm yr⁻¹. As fertilization increased beyond this maximum (120–150 kg N ha⁻¹), the grain N to fertilizer ratios were <1. Fall and spring applications of fertilizer are recommended under high precipitation, but current extension guides suggest splitting a predetermined N fertilization rate rather than fine-tuning fertilizer management through tactical in-season applications or variable rates based on crop performance indicators (Mahler, 2007). Furthermore, these findings highlight the opportunity to investigate the utilization of nitrification inhibitors in soils of the inland Pacific Northwest with high risks of nitrate leaching or denitrification (Abalos et al., 2014).

In addition to regional variability, we observed a considerable amount of within-site variability in WW productivity (Table 2), even at a field or plot scale. While a portion of this variability was teased out by differences in year-to-year precipitation, other environmental, genetic, and management factors contribute to within-site variability, such as landscape elevation, slope, aspect (Mulla et al., 1992; Fiez et al., 1994; Yang et al., 1998), subsoil constraints (Robertson et al., 2016), spring rainfall (Schillinger et al., 2008), crop rotation (Hammel, 1995), nutrient availability (Fiez et al., 1995), onset of drought and temperature stress (Gizaw et al., 2016), and crop genetics (Schillinger et al., 2008). Despite the heterogeneity of the landscape, iPNW growers still commonly apply single rates of fertilizer, particularly in the higher precipitation areas (Mahler et al., 2014). The need for site-specific N management across the entire region is warranted, as evidenced by N fertilizer rate explaining only 35% of WW yield (Figure 3A) and 28% of grain N variability (Figure 3C) when WW followed crops, and 21% of the variation in yield and 11% of grain N accumulation when WW followed fallow when precipitation was <350 mm yr⁻¹.

In contrast, our analysis indicated that crop N was a consistent and reliable indicator of yield and grain protein, and may be a useful evaluation tool for site-specific N management in the iPNW. Across all data, grain N accumulation was closely correlated with the synthesis grain biomass (Figure 3B) and total above-ground plant N (Figure 3D) across all data, regardless of location, years, fertilizer rate, and previous crop, with R² values ranging from 0.80 to 0.98. Magney et al. (2016a) demonstrated that in-season canopy monitoring using daily Normalized Difference Vegetation Index (NDVI) data could be used to explain 83% of yield variance and 80% of grain N accumulation. Total N in the biomass and grain has also been shown to be highly correlated to vegetation indices, particularly using red-edge bands, acquired from high resolution (5 × 5 m) satellite imagery (Magney et al., 2016b). Therefore, precision agriculture based on plant-based sensing opportunities may be used in combination with tactical nutrient management to optimize wheat performance. Ultimately, these data can be utilized by decision support tools emerging for the region which allow growers to systematically evaluate their site specific N management practices based on performance criteria, and to diagnose conditions that contribute to suboptimal performance (Brown et al., 2015).

The traditional WW-summer fallow system was developed as an economically feasible strategy to accumulate water and N, manage weeds, and mitigate risk associated in low precipitation areas of the iPNW (Leggett et al., 1974; Bolton and Glenn, 1983). Crop diversity is largely lacking in the iPNW and fallow periods remain common in the low and intermediate precipitation cropping systems, and a decrease in species richness has also been reported along the diminishing mean annual precipitation gradient in other Mediterranean-like, semi-arid regions (Koocheki et al., 2008). The lack of crop diversity may be contrasted with other major wheat belts, such as the dryland systems of Canada and Australia (Cook et al., 2002; Zentner et al., 2002b; Conley et al., 2004; Kirkegaard et al., 2008). For instance, in the semi-arid region of Canada, the prevalence of fallow has decreased from 50 to 15% since the 1970s, coinciding with the development of chemical weed management and the adoption of direct seeding practices (Zentner et al., 2002a). The intensification of crop rotations enabled the expansion of pulse crops in semi-arid Canada (McVicar et al., 2000), which performed well in water-stressed environments of the Canadian prairies dominated by summer rainfall (Angadi et al., 2008; Cutforth et al., 2009; Bueckert and Clarke, 2013). Furthermore, the shallower rooting depth of pulse crops compared to cereals left behind deep residual water to support subsequent cereal and oilseed crops (Gan et al., 2009). By the late 1990s, research on pulse crops highlighted the rotational benefits of pulses to cereals (Miller et al., 2003) as a means to economically intensify crop rotations in Canada's semi-arid zone. These developments provide the rationale for assessing the effects of reduced tillage, crop diversity, and intensified of wheat-based systems in the iPNW.

Despite the yield penalty on WW in continuous cropping systems under low precipitation conditions, continuous cropping may be agronomically feasible on an annualized basis of whole rotations. From 2012 to 2014/2015, intensification of cropping systems through the reduction of fallow frequency or elimination of fallow periods did not affect annualized energy outputs or grain N export (Figure 7), though more intensified systems received more fertilizer inputs at Moro, OR, Ritzville, WA, and Davenport, WA. Alternative cropping systems also did not differ significantly in rotational fertilizer use efficiency patterns from traditional rotations. In particularly, the relationship between fertilizer N and annualized energy production (Figure 8A) and grain N accumulation (Figure 8C) did not differ whether wheat followed fallow or crops when annual precipitation was <350 mm yr⁻¹. Nitrogen fertilization increased energy yields and grain N accumulation, but explained only 13% of the variation in energy outputs (Figure 8A) and 21% of grain N accumulation (Figure 8C) in traditional systems. In comparison, 17% of energy outputs and 25% of grain N variability were explained by N fertilization rates in alternative rotations with the reduction or elimination of fallow. Like WW, plant N was tightly regulated to produce grain (Figure 8B) and synthesis grain protein (Figure 8D) across all data, explaining 82–97%. Importantly, no differences among relationships were observed due to reduction or elimination of fallow when annual precipitation was <350 mm yr⁻¹.

Researchers in other Mediterranean climates have also reported agronomically viable intensified rotations, despite lower wheat yields upon the elimination of fallow (Christiansen et al., 2015). Cropping systems research in Canada have indicated that intensified rotations were agronomically feasible (Zentner et al., 2003), and fallow timing (e.g., wheat following fallow or recrop) predominately influenced annual wheat productivity in intensified rotations even though annualized grain yields increased with cropping intensity (Campbell et al., 2004). Continuous cropping of wheat rotated with legumes reduced the fertilizer requirements of a wheat-fallow rotation, and economics were dependent on commodity prices (Zentner et al., 2001). However, despite greater economic returns, financial risk increased with continuous wheat (Zentner et al., 2006).

In the iPNW, research has indicated that continuous spring cropping may be agronomically feasible (Schillinger et al., 2007; Bewick et al., 2008), even with less precipitation than we observed in our study (Machado et al., 2015). However, in six cropping seasons previously conducted at Ralston, the wheat-fallow system yielded 25% more than continuous spring wheat (Young et al., 2015). Furthermore, observations taken from a only few years, such as in the present study, may not be consistent over decades (Nielsen and Vigil, 2014). Previous research has also found that the profitability diminishes relative to WW-fallow rotations in the low precipitation zone due to high year-to-year yield variability and associated economic risk (Juergens et al., 2004; Schillinger and Young, 2004; Schillinger et al., 2007; Bewick et al., 2008). This finding may be due to differential pricing of alternative spring crops, increased costs associated with N fertilizers, changes in weed pressure (Sullivan et al., 2013), and grain yields in continuous cropping systems that average only 50–60% of WW following fallow.

Lastly, we developed N budgets at Moro, Ritzville, and Davenport to determine the effect of crop intensification in fertilizer N balances and unaccounted for N over multiple years (Table 5). At all locations, the fertilizer N balance was greater (fertilizer N exceeded grain N) for all continuous cropping systems with spring wheat, barley, camelina, or canola implementing conservation practices than the traditional fallow rotations. However, the total N balance had no discernible trends according to intensification. Earlier examination of N fertilizer balances at the Ralston site also revealed that the WW-summer fallow conventional system, when fertilized during the fallow period, had the highest fertilizer N efficiency (grain N/fertilizer N = 76.8%) compared to spring wheat-chemical fallow (64%), due to deeper rooting and efficient soil N extraction by winter wheat down to 150 cm (Pan et al., 2001). Similar to the latter rotation, continuous no-till spring cereal cropping also had a lower fertilizer N efficiency (grain N/fertilizer N = 61.7%) and water extraction, as evidenced by nitrate movement below 90 cm. It is notable that although conventional WW is fertilized during fallow nearly a year before the accelerated N uptake phase of winter wheat, the grain N to fertilizer index is surprisingly high in this low rainfall zone. While mineralized N accumulates during the fallow period in addition to the fertilizer application, soil tests indicate minimal overwinter nitrate leaching beyond the 150 cm root profile. The pre-plant N application during fallow allows growers to spread their workloads, and for the ammonia to nitrify and move into the middle of the winter wheat root zone, at 60–90 cm soil depths. Similarly, high N fertilizer efficiencies of WW-summer fallow grown in the northern iPNW near Okanogan, WA, has been observed, with 79% of fallow-applied ammonia over 20 years accounted for in harvested grain, and the remainder was accountable in an increase in soil organic N over that period (Pan et al., in review). This fertilizer N recovery was higher than the biosolids N recovery (24–37%) WW-summer fallow, when biosolids were applied once every 4 years at 3 different rates. However, biosolids significantly made greater contributions to the build-up of soil organic N, leading to the conclusion that while the synthetic fertilizer mainly fed the wheat grain, the biosolids fed both the wheat grain and the soil organic matter build-up. Nevertheless, the biosolids applications resulted in more net unaccounted for N than the ammonia.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.