A multifactorial microcosm experiment was established with two different phosphorus concentrations in the topsoil (P+/P−), irrigation (Irr+/Irr−) and presence of continuous vertical pores in the subsoil (CVP+/CVP−) as factors. The experiment was conducted in PVC-pipes with an inner diameter of 19 cm and a total height of 85 cm (Figure 1). Upper segments (height: 20 cm), were filled with topsoil from plots of field experiments with P deficiency (research farm “Dikopshof”, University of Bonn) or elevated P concentrations (experimental station “Campus Klein Altendorf”, University of Bonn). For all treatments, middle and lower segments (height: 45 and 20 cm) were filled with subsoil from the latter site. Both soils have developed from Weichselian loess. Detailed information about soil properties is listed in Table 1. Mean annual temperature and annual precipitation are 9.7˚C and 630 mm at “Dikopshof” and 9.6˚C and 625 mm at “Campus Klein Altendorf”. Topsoil and subsoil were sieved to 5 and 2 mm prior to experimentation. Soil in upper and lower

^aDIK: “Dikopshof”; CKA: “Campus Klein Altendorf”; ^bSoil texture according to: [23] and [24] .

segments was moderately compacted to bulk densities of approximately 1.3 g/cm³. Soil in the middle segment was compacted to approximately 1.6 g/cm³. We used a modified hydraulic device originally designed to lift hey-bales. Soil was compressed with a round “stamp” neatly fitting into the tubes. In order to achieve a more evenly distributed compaction, we pressed each soil column in 3 steps. In half of the middle segments (CVP+) 6 artificial vertical pores were created with an electric drill (diameter: 8 mm).

A gauge was used to make sure that the pores were equally distributed in each of the columns. The density of vertical pores is equivalent to approximately 210 pores m⁻², corresponding to a density of approximately 149 - 256 coarse pores m⁻² previously reported for the soil under study [25] . Artificial pores created with an electric drill can have surface properties different from natural pores. Thus, before cultivation we introduced earthworms into the pores in order to provide more natural pore wall properties. Each pore was incubated with an adult dew worm (Lumbricus terrestris L.) for 4 weeks. Worms were equally distributed over the columns according to their body mass. Each pore opening was closed with a perforated plastic cup to prevent worms from escaping. 0.3 g of dried lucerne-grass cut to 2 mm was placed under each plastic cup as a food source for earthworms. The litter was replaced when approximately half of the material was removed. Overall, up to 1.65 g of lucerne-grass mulch per pore were fed during the incubation. At the end of the incubation period of four weeks all worms were removed from the pores by warming up the column segments in a water bath to approximately 33˚C. Columns without artificial pores were treated in the same way; in particular, they were stored in the same room during incubation and warmed up in a water bath after incubation. Before cultivation, lower column segments were soaked in tap water in order to provide a source of water in the deep subsoil. The top of the lower and middle segments were covered with approximately 3 cm layers of gravel (particle size 7 - 15 mm), to avoid capillary rise. Finally, the column segments were assembled and connected with duct tape.

12 seeds of winter wheat (Triticum aestivum L.) “Graziaro” were sown into each column corresponding to a seeding density of 422 grains m⁻². In total, wheat was cultivated for 16 weeks. During the first 12 weeks, columns were set up in a greenhouse at ambient temperature. At seeding, columns from all treatments were irrigated with 30 ml tap water to allow equal emergence. Thereafter, Irr+ columns were irrigated twice a week with 30 ml tap water, whereas irrigation in the Irr− columns was ceased. During the last 4 weeks of the experiment, drought stress was induced using fan heaters and 400 W mercury-vapor lamps for 12 h/day. The temperature amplitude at plant height was 20˚C - 30˚C. Columns from the Irr+ treatment were daily irrigated with 30 ml tap water for two weeks. Since drought symptoms became visible, irrigation was increased to 100 ml tap water/d for the last two weeks.

Shoot biomass was measured as dry matter with oven drying (65˚C). C/N and P contents of soil and plant samples were measured using dry combustion with a Fisons NA-1500 elemental analyzer and atomic absorption spectrometry (AAS), respectively. Leaf area index was determined with a LI-3100 Area Meter (LI-COR Inc., USA). Middle segments of columns were horizontally dissected into 4 slices, representing soil depths of 20 - 30, 30 - 40, 40 - 50 and 50 - 60 cm. Slices containing artificial pores were sampled as follows.

Each pore was laterally opened using sharp knives. Roots growing inside of the pore lumen were carefully collected with a pair of tweezers. If necessary, roots were separated from the pore wall with a scalpel. The remaining samples were soaked in tap water and washed over a sieve (mesh size: 0.5 mm) in order to separate roots from mineral particles an organic debris. The samples were further sorted to remove any non-root material. Immersed roots were carefully placed on trays in order to avoid overlapping and scanned with 800 dpi using a Perfection V700 Photo scanner (Epson Corp.). Root-length density (cm roots cm⁻³ soil) was determined with the software package WhinRhizoPro 2016 (Regent Instruments Inc.). Soil from columns without artificial pores was directly soaked in tap water. Roots were washed and scanned as described above.

Data were checked for normality with Shapiro-Wilk tests. Since normality was generally given, means were compared without previous transformation by multifactorial ANOVA with continuous vertical pores (CVP), P supply in the topsoil (P) and Irrigation (Irr) as fixed factors and block effect as random factor. Means of shoot parameters were additionally compared on single treatment level by one factorial ANOVA with Tukey tests. All statistical analyses were done with IBM SPSS Statistics Version 25.

Total root length of winter wheat in 20 - 60 cm soil depth was higher when continuous vertical pores were present (Figure 2, Table 2). The magnitude of this

effect gradually increased with increasing soil depth. Irrigation did not change root length significantly, thus Figure 2 shows root length as average over Irr+ and Irr− treatments. Total root length was generally higher throughout all soil depths when winter wheat was cultivated in topsoil containing high concentrations of P (Figure 2, Table 2). This finding is in line with the results from a rhizotron study with wheat [26] showing that P deficiency in the topsoil resulted in decreased biomass of subsoil roots. Plants growing under conditions of low P supply particularly promote topsoil foraging by increased density and elongation of shallow lateral roots [27] .

The positive effect of continuous pores on the root length of wheat in our study was much more evident in treatments with high P concentrations in the topsoil, indicating that suitable growing conditions in the topsoil along with adequate structure of the subsoil are advantageous for extensive exploration of subsoil resources.

Further analysis of the treatments with continuous vertical pores revealed that many wheat roots had reached the lower segment by growing through the large vertical pores in the middle segment. The share of roots growing inside the pore increased with soil depth from 38% in 20 - 30 cm to 64% in 50 - 60 cm (Figure 3). In earlier studies, the proportion of roots in CVPs varied widely from <10% [28] to near 100% [29] , probably depending on soil properties such as penetration resistance and CVP angle [18] . However, previous studies generally revealed an increasing share of roots in CVPs with soil depth in compacted soil layers [28] [30] . In the bulk soil, presence of continuous pores resulted in decreased root length in 20 - 40 cm but increased root length in 40 - 60 cm soil depth (Figure 4). Hence, vertical pores shifted the root system towards deeper soil layers. As mentioned above, water supply did not have a marked effect on the root

length of wheat. Merely in the columns with continuous vertical pores, slightly but in significantly increased root length at drought was observed in the deeper soil layers (data not shown). Higher root length under deficit irrigation was reported previously [31] . Deep roots are considered as beneficial for increasing the amount of subsoil moisture accessible by the root system [32] . The reason for the fact that in our study no pronounced effects were observed may be that crops were exposed to water stress merely during the last stage of the experiment.

Presence of continuous vertical pores generally had a positive effect on shoot dry matter and N uptake of wheat, but the magnitude of the effect was different between treatments (Figure 5(a) and Figure 5(b), Table 2). In columns with low soil moisture in the topsoil and high P concentration in the topsoil, presence of macropores in the subsoil increased shoot dry matter by 66%; in contrast, with irrigation the beneficial effect of macropores on shoot dry matter was only 39% (Table 3). For N uptake, the relations were similar (101% vs. 62%), likewise for

P uptake (83% vs. 33%). Thus, the effect of continuous vertical pores in our experiment can be explained primarily with improved accessibility to water stored in deep soil layers. An earlier study [33] showed that barley plants grew better in pots containing a network of narrow pores than in presence of large artificial pores, which was explained mainly by poor root-soil contact and impeded access to the compacted pore walls. However, the experimental pots did not contain a water saturated soil compartment in the deep soil.

When irrigation was ceased, presence of vertical pores in the middle segment increased P uptake from 8.1 to 14.8 mg/pot (+83%) when the topsoil was rich in nutrients, but from 1.6 to 4.1 mg/pot (+158%) when P concentrations in the topsoil were low (Figure 5(c)). Apparently, improved access to water and phosphorous in the subsoil becomes relatively more important under conditions of P deficiency in the topsoil. Despite the fact that vertical pores enhanced root length in the subsoil and the pores were coated with earthworm feces rich in nutrients, it cannot be clarified whether enhanced P uptake in presence of vertical pores is based on increased P acquisition from the subsoil.

Whereas P uptake from CVP coatings has not yet been quantified, P uptake from the subsoil largely depends on soil properties and growing conditions and varies between 3% - 4% [34] and 85% [35] of total P uptake. Presumably, to some extent the effect of pores can also be based on elevated P uptake from the topsoil as a consequence of additional water uptake from deep soil. In total numbers, however, the effect of vertical pores on P uptake into the shoot was more pronounced when P concentrations in the topsoil were elevated (Figure 5(c)). Our results clearly indicate that favorable topsoil conditions enable crops to draw more benefit from subsoil resources.

It is important to note that the beneficial effect of continuous vertical pores on shoot dry-matter of wheat, as seen in the optimum treatment (Irr+/P+), increased at drought (Irr−/P+) but decreased at P deficiency (Irr+/P−). Hence, the deeper root system in the columns with continuous vertical pores presumably facilitated water uptake from the deep soil, whereas it was less favorable for P uptake, which primarily takes place in the topsoil. In general, wheat cultivated in topsoil with elevated P concentrations yielded approximately twice as much shoot dry matter than wheat cultivated in topsoil with low P concentrations. P uptake into the shoot more than tripled on average with elevated P concentrations in the topsoil.

Water stress during the last stage of the experiment reduced shoot dry matter of wheat by between 13% and 63%, depending on presence of continuous vertical pores and P concentrations in the topsoil (Table 3). Since many leaves already withered at harvest, the response to water stress as measured by leaf area index was much more evident (Figure 5(d)). It is noteworthy that in columns with elevated P concentrations in the topsoil but without vertical pores in the middle segment, irrigation increased shoot dry matter yield from 3.4 to 4.5 g/pot, whereas unwatered columns with vertical pores in the middle segment yielded 5.4 g/pot. Hence, the irrigation in our experiment did not fully equalize the water stress.

Enhanced nutrient concentrations in the topsoil and irrigation increased the shoot: root ratio (Figure 5(e)). Plants typically respond to P deficiency through increased carbon allocation to roots [36] . Preferential partitioning of photosynthates to the roots with decreasing fertilizer concentration has repeatedly been reported [37] [38] and occurs particularly under deficiencies of nutrients mobile in the phloem such as nitrogen and phosphorus [39] . Accordingly, drought can decrease the shoot: root ratio [40] [41] [42] , which can be understood as a shift of allocation of phytosynthates towards water obtaining organs.

Vertical pores in the subsoil generally decreased the ratio of shoot mass to root length (Figure 5(e)). It is important to note that depending on treatment and soil depth, root diameter was either not affected or increased (data not shown). Accordingly, a previous field study [43] showed that increased CVP density rather increased the average diameter of a cereal root system. Hence, in our experiment decreased shoot: root ratio was not caused by a higher share of thin roots but by facilitation of root elongation through vertical pores.

Roots are a major sink for carbon and unnecessarily large root systems can reduce water efficiency and even grain yield [44] . Clearly, modern wheat varieties tend to have higher yield potential, higher harvest index and greater shoot: root ratio than old varieties [45] . On the other hand, low shoot: root ratio can reduce drought stress [46] . Hence, large vertical pores in the subsoil cannot necessarily be considered as beneficial for increasing yield under favorable growing conditions, but they can potentially contribute to drought tolerance.