Investigations on the Identification of Long Term Input of Phosphorus from Organic Sources by Standard Soil P Analysis

Untersuchungen zur Identifizierung von langfristigen Phosphoreinträgen aus organischen Materialien mittels standardisierter Boden-P-Analytik

2 Material and methods

I- 78 Samples from a pot experiment investigating the performance of rendering by-products as P fertilizers, conducted by the Institute of Crop and Soil Science (JKI), Braunschweig, Germany in 2010 and 2011. 39 Pots were filled with 6 kg of soil (classified as Dystic Cambisol/Orthic Luvisol) which was collected in April 2010 and sieved to < 1 cm before filling it into the 10L-Mitscherlich-pots. Four kinds of fertilizers were used: Meat and bone meal (org-MBM), bone meal (BM), rock phosphate (RP) and monocalciumphosphate (MCP). The pots were arranged randomly with three replicates for each treatment. Additionally 3 pots were used as control with no P application (zero P). Two crops were used in this pot experiment: in the first year ryegrass (Lolium perenne L) and in the second year maize (Zea mays L). The chemical characteristics of the soil are summarized in Tab. 1. The fertilizer application rates are given in Tab. 2.

II- 20 soil samples from a long term field experiment in Braunschweig (Julius Kühn-Institute), Germany (10° 27’ E; 52° 18’ N), investigating the influence of fertilization with meat and bone meal on various soil microbiological parameters (experimental site “FV 10”). The soil type is a Cambisol with a loamy sand soil texture (< 6.5% clay; > 47% sand). The samples were collected within 5 years since 2008 (4 samples/year). The field was fertilized either with MBM or with calcium ammonium nitrate and triplesuperphosphate.

III- 77 samples were collected in 2011 from 4 different regions in Germany during a field sampling campaign within the framework of the EU project “Baltic Manure” (14 fields). These regions differ in the dominating soils and dominant parent material. The sampled fields were either fertilized with manure or with soluble mineral fertilizers. Hence, the soils vary significantly in their characteristics such as the total and available content of P, soil texture and pH. The properties of these soil samples are compiled in Tab. 3.

I- Calcium acetate lactate method (CAL), according to Schüller (1969), P was extracted with 0.05 M Ca-acetate (Ca(C₂H₃O₂)₂) + 0.05 M Ca-lactate (Ca[CH₃CH(OH)COO]₂) + 0.03 M acetic acid (CH₃COOH)

II- Sodium bicarbonate extraction method (Olsen) (Olsen et al., 1954), 100 ml of solution (0.5 M NaHCO₃) buffered to pH 8.5 with 1 M NaOH/1 L water

III- Water extraction method (van der Paauw et al., 1971)

IV- So-called total P was determined in an aqua regia (AR) digest according to AbfKlärV (1992).: the solution was composed of 3:1 (v/v) HCl: HNO₃. AR is considered as a digestion method, but for the easier way of expression it was included here when the term “extraction methods” is used.

3 Results and discussion

The absolute amounts of extracted inorganic P (P-Col.) tended to be higher in CAL than in Olsen in all four groups (org-M, MBM, BM and min); however, no significant difference between these two methods was found for the minerally fertilized plots. Also, no difference between CAL and Olsen was found for the zero-P plots (Tab. 5).

Some scientists argued about the low ability of Olsen to extract P from organically fertilized soils in comparison to acidic reagents and proved in their studies that the absolute amount of extractable P increased whenever the acidity of the reagent increased (Sims, 2000; Ige et al., 2006). This is explained by the fact that the low pH of the reagent will dissolve more phosphate associated with Ca (Pizzeghello et al., 2011). However, in this study, CAL and Olsen extracted almost the same amounts of P from mineral fertilizers. Some of the minerally fertilized plots had received RP as a fertilizer, which also contains poorly soluble Ca-P, so it was checked if the different chemistry of RP would be reflected in the P extractability of the respective samples, particularly with regard to the Olsen extract. Since for the minerally fertilized samples from groups II and III the exact fertilizer type was not known, only samples from group I (comprising RP and MCP in the category “mineral”) were included in this test (Tab. 6). However, there was still no significant difference in extractable P between CAL and Olsen, respectively, neither for the rock phosphates, nor for the highly soluble MCP, so obviously the Olsen extract was fully capable of dissolving the same amount of Ca-phosphate as the CAL extract. Hence, there must be another reason why Olsen performed worse than CAL in extracting available P from the two organic fertilizers and from BM.

Nevertheless, this differentiation brings up some useful information: As could be expected, samples from “RP-plots” displayed the lowest solubilities in CAL, Olsen and water. This observation agrees with Jeng et al. (2006), who also report RP-fertilized plots to display lower P-solubilities than those fertilized with MBM. This can be explained by the fact that the Ca-P in MBM, as well as in BM and manure, is not completely derived from apatite, but also to a considerable amount from different chemical structures, which are obviously easier to break up than the very sparingly soluble apatite in RP (Bationo et al., 1991; Menon and Chien, 1995). MCP-fertilized plots, on the other hand, display the highest solubility in the Olsen-extract. Interestingly, however, their solubilities in water are on the same low level as those for the org-M-, MBM- and BM-plots.

Evaluation of these results must be done with care, however, since the plots are not homogeneous with regard to their soil properties such as soil pH and content of organic matter/organic carbon, which will have a considerable influence on the transformation of the various fertilizers and their different P-forms in the soils (see below). Thus, these observations can only describe visible trends.

Main reason for this difference is the fact that the colorimetric method basically measures inorganic P, i.e., orthophosphate ions, while ICP determines total dissolved P including organic and inorganic P (Matula, 2010). There is much evidence in the literature confirming this finding (Nathan and Sun, 1998; Elliason et al., 2001; Nathan et al., 2002; Mallarino, 2003; Pittman et al., 2005; Jacoby, 2005; Ziadi et al., 2009; Matula, 2010, 2011). A highly significant correlation was noticed between P-ICP and P-Col. for all extraction methods. Fig. 1 presents the regression plots of the two analytical methods showing high R² values for all extractants, and the regression slopes display that ICP values were higher than colorimetric ones with about 3, 1, and 2% for CAL, Olsen, and water. In the case of AR, the regression slope is below 1, how­ever, from the large intercept on the y-axis it can be seen clearly that, at least within the range of values occurring here, ICP-derived values exceed those determined by colorimetry in this case as well.

Fig. 1. Relationship between P measured by ICP-OES and P measured colorimetrically in four extraction me­thods.

For the two extraction methods Olsen and AR, the ratio of P-ICP/P-Col. was nonlinearly related to P-Col., i.e. the difference between ICP and colorimetry was proportionally larger where soils contained low amounts of extractable P (Fig. 2). These results matched with the reports by Mallarino (2003), Pittman et al. (2005) and Adesanwo et al. (2013). Both Mallarino (2003) and Pittman et al. (2005) doubted that this relation could be affected by variation in soil pH and organic C. According to Adesanwo et al. (2013), one explanation might be the difference in sensitivity of these two methods, with ICP being the more sensitive one, however, it is surprising then that this phenomenon was not observed in the case of CAL and water extraction.

Fig. 2. Relationship between P-Col. and the ratio between P-ICP and P-Col. for (A) CAL, (B) Olsen, (C) water, and (D) AR methods.

The plot for CAL (Fig. 2A) shows clearly that there are two subsets of samples, so samples were divided into two groups according to the values of the ratio between P-ICP and P-Col., where the first group ranged from 0.60 to 1.25 and the second one between 1.26–1.70. Yet, as can be seen in Fig. 3, the same results were obtained; i.e. the difference between ICP and colorimetry was not larger where soils contained low amounts of extractable P.

Fig. 3. Relationship between P-Col. and the ratio of P-ICP and P-Col. in the CAL extract for two subsets of samples: A) ratio between 0.60–1.25 and B) ratio between 1.26–1.70.

Fig. 4. Comparisons between the relative shares of orga­nic P estimated by four different extraction me­thods (n = 175). Different letters indicate significant differences between cases at the level p < 0.05.

The low share of organic P estimated for the AR extract may be explained by the fact that the strong acid reagent obviously breaks up most of the organic P compounds into ortho-P, which will then be measured as inorganic P. The relatively low share of estimated organic P in the water extract indicates that water will only extract orthophosphate to a large extent, while it comprises – if at all – only a very small amount of dissolved organic P.

In the next step, the relative share of dissolved organic P in the four soil extractants was compared for different fertilization regimes. Although one may speculate that long-term organic fertilization (org-M, org-MBM) will be displayed by a higher share of organic P in the soil extracts than mineral fertilization, this was clearly not the case for the investigated samples (see Fig. 5). On the contrary, when extracted by CAL, organically fertilized plots displayed even higher shares of inorganic P than minerally fertilized plots (Fig. 5A). This, however, agrees well with the results of various researchers who reported that more inorganic than organic P was found in soil fertilized with manure for a long time in comparison to soils which had received mineral fertilizer: For example, Sharpley et al. (1984) found that there was an increase in the proportion of total P as inorganic P from 34 to 71% with larger manure applications. One reason for this can be seen in the fact
that in the first place, between 60–90% of P in manure is inorganic (Barnett, 1994; Leinweber, 1996; Sharpley and Moyer, 2000). In addition to this, the long-term application of manure promotes the activity of microbes and enzymes, thus inducing mineralization of organic P, which further increases the share of inorganic (and soluble) P (Mozaffari and Sims, 1994; Motavalli and Miles, 2002). Depending on the soil pH, chemical hydrolysis of organic P may also play a role in further reducing the organic traces of fertilization (see for example Baldwin, 1998 and Matula, 2010). Similar findings were reported by Adesanwo et al. (2013), who compared STP-analysis by colorimetry and ICP-OES in 30 manured and 30 non-manured soil samples, applying water, CaCl₂- and Mehlich 3 extraction methods. They also speculated that chemical transformations of organic P in manured soils might have leveled out differences in organic P contents between manured and non-manured soils, thus accounting for the lack of any effect of manure history on differences between ICP-P and colorimetric P.

Fig. 5. Comparison between the estimated relative share of organic P, differentiated by fertilization regime, in A) CAL, B) Olsen, C) water and D) AR extract.

For the other three extraction methods, Olsen, water and AR, the estimated organic P decreased in the following order: org-M > min > org-MBM, however, with no significant differences between org-M and min for the Olsen extractant (Tab. 9). The low amount of estimated organic P in org-MBM plots could be the result of a transformation of MBM-P into a more soluble and thus plant-available form after 2 years of experiment, which is compa­rable to what Ylivainio et al. (2008) reported about their experiment, which was carried out in non-calcareous soils, i.e. under conditions similar to this study.

Interestingly, a high estimated share of organic P (mean values) was found in the zero P plots, in the case of the Olsen extract these plots even showed the highest share of organic P compared with the other fertilization regimes (25.4% of total extracted P) (Tab. 9). One possible reason could be that mineralization of organic P in P deficient soils may be rather slow since nutrient defi­ciency will generally lead to a slow growth of roots which in turn decreases the availability of oxygen, one of the important factors that affect the mineralization: according to Harrison (1982), only 1% of organic P is mineralized per year in P deficient soils.

The one-way ANOVA with Tukey post-hoc test used to compare the shares of organic P estimated by the four extraction methods (CAL, Olsen, water and AR) showed no significant differences between water and AR for most fertilizer regimes, except for org-M and zero-P. Between CAL and Olsen, significant differences in the estimated share of organic P were found for minerally fertilized plots and for zero-P (Tab. 9, Fig. 5A–D).

In Tab. 11 the number of soil samples falling into each of the STP classes, differentiated by fertilization regime is shown.

If the initial speculation about long-term organic fer­tilization leaving traces in the form of higher shares of extractable organic P had fallen true, this should have shown most clearly in the plots that had received large amounts of organic fertilizers (some of them over a long period of time), i.e. classes D and E of org-MBM or org-M. However, the results displayed the opposite in all four extracts: The average estimated share of organic P in samples fertilized with org-M (for example) in class E was 2.4, 7.9, 9.6 and 8.4% of total extracted P by CAL, Olsen water and AR extracts, respectively. In class A, these averages were 18.1, 39.2, 24.2 and 16.7% (Fig. 6). Comparable relations were found for org-MBM (Fig. 6). Obviously, the received organic P in class E (and D) was not enriched in the soluble organic substance of the soil, but was either transferred quickly into inorganic P forms by microbial soil processes stimulated by the large addition of organic matter from MBM/manure (see Mondini et al., 2008), or it was turned into stable organic P compounds which were not solubilized by the tested extractants. Several researchers recorded a reduction in the amount of organic P in soil even after long application of manure, as mentioned before. Yet, no similar observation has been reported for MBM, since long-term experiments dealing with MBM have been rare, as Bøen and Haraldsen (2013) stated, probably because the restrictions on using MBM as a fertilizer in EU countries were only lifted since 2006.

Fig. 6. Comparison between the relative share of organic P estimated for different fertilization types for CAL, Olsen, water and AR extracts, differentiated by STP class. Different letters indicate significant differen­ces between cases (within one STP class) at the level p < 0.05. In class D, zero plots were not included into the ANOVA because n = 1; in class E, a t-test between min and org-M was performed.

However, the larger difference between P-ICP and P-Col. in class A may at least in parts be a result of reasons not related to organic P but rather to other factors such as the differences in the sensitivity of these two methods, which are most clearly observed when soil test P is low (see above, chapter 3.2).

In comparison to those plots fertilized with mineral fertilizers (min), the manured plots (org-M) generally did not reveal a higher estimated share of organic P in the four extractants in class A (lowest P supply), confirming that the soluble organic P coming in with the manure transferred quickly into inorganic P by hydrolysis and other processes.

For class E (highest P supply), the estimated share of organic P was significantly higher in the minerally fer­tilized plots than in the ones fertilized with org-M only in CAL and Olsen, while in water and AR there was a sig­nificantly higher share of organic P in the plots fertilized with org-M (Fig. 6). Maybe this could be explained by the fact that at high organic P loads, organic P is no longer hydrolyzed completely, but some of it remains as dissolved organic P (water-soluble) or stable organic P (which can only be digested by AR). The reason why not everything is hydrolyzed in this case may be the increase in soil pH coming along with high amounts of organic fertilizer (manure).

Fig. 7. Scatterplot and coefficient of determination for the relationship between soil pH and relative share of or­ganic P estimated by four extraction methods.

The correlation between soil-pH and estimated share of organic P was also checked for each fertilizer regime separately (Tab. 12).

For the CAL extract, negative and linear correlations existed between soil pH and the relative share of organic P in all five fertilizer regimes, however, the strength of this relationship was different, with the strongest correlation found for zero P fertilization (r = –0.93**) followed by org-MBM, BM and min with moderate correlations, while the weakest correlation was found for org-M (r = –0.57). For the Olsen extract there were negative and moderate correlations only for organically fertilized soils, i.e. the variants org-MBM and org-M (r = –0.64**, –0.57** respectively). A similar relationship was found in the investigations by Turner et al. (2003) and Turner and Haygarth (2003): The latter study, which was done for a range of UK pasture soils, confirmed that the concentration of organic P estimated by bicarbonate extractant was controlled by soil pH and this concentration increased when soil pH < 5. Similar correlations were reported by Mallarino (2003), Pittman et al. (2005) and Adesanwo et al. (2013). A negative relationship between estimated share of organic P (which is based on the difference in ICP-P and P-col.) on the one hand and soil pH on the other hand indicates that the estimate increases as soil pH decreases. As pointed out by Adesanwo et al. (2013) (see also: Eckert and Watson (1996), Elliason et al. (2001)), the exact role that pH plays in this respect is so far not clear. The authors speculate that some indirect influence of other parameters (such as the presence of suspended soil particles in the extract) which were not measured here might have an influence.

However, this was not confirmed by the results of this study. While C_org was not determined here, correlation tests for C_total and estimated share of organic P should give a first indication. When looking at all samples as one group, no visible relationship between the amount of C_total and on the amount of estimated relative share of organic P was observed. A very weak but still significant trend towards a positive effect of total C on the estimated relative share of organic P was observed in Olsen, water and AR extracts (r = 0.22***, 0.21** and 0.23** respectively). In contrast, there was a very weak negative effect for CAL, which was, however, not significant (r = –0.08) (Fig. 8).

Fig. 8. Scatterplots for the relationship between C% and relative share of organic P estimated by four extraction methods, looking at all samples as one group (n = 175).

The correlation between total C and the estimated share of organic P was also examined for each fertilizer regime separately in order to check whether a relationship would be visible for the organically fertilized plots at least (Tab. 13).

Only the org-MBM plots showed significant positive correlations of moderate to high strength between the total C and estimated organic P for all extraction methods, while no significant correlations were found for the manured plots (org-M). Significant and moderately strong correlations were also found for the zero plots, with regard to the estimates from CAL and AR extracts, however, in the case of CAL, the relationship was negative.

However, as the presented results demonstrated, long-term organic fertilization is not necessarily reflected in a higher share of soluble organic P than mineral fertilization. On the contrary, when extracted by CAL, organi­cally fertilized plots displayed even higher shares of inorganic P than minerally fertilized plots. This agrees well with reports of other researchers in the literature. One reason for this can be seen in the fact that between 60–90% of P in manure is inorganic in the first place. In addition, long-term application of manure promotes the activity of microbes and enzymes, thus inducing mineralization of organic P, which further increases the share of inorganic (and soluble) P. Depending on the soil pH, chemical hydrolysis of organic P may also play a role in further reducing the organic traces of fertilization. Furthermore, chemical transformations of organic P in manured soils might level out differences in organic P between manured and non-manured soils, thus accounting for the lack of any effect of manure history on differences between ICP-P and colorimetric P.

A comparably high estimated share of organic P was found in the zero P plots. This was probably due to a rather slow mineralization of organic P in P deficient soils associated with nutrient deficiency, which generally leads to a slow growth of roots, thus decreasing the availability of oxygen, which is an important factor affecting mineralization rates.

If the initial speculation about long-term organic fer­tilization leaving traces in the form of higher shares of extractable organic P had fallen true, this should have shown most clearly in the plots that had received large amounts of organic fertilizers, thus ranging in classes D and E. However, the results displayed the opposite in all four extracts. Obviously, the received organic P in class E (and D) was not enriched in the soluble organic substance of the soil, but was either transferred quickly into inorganic P forms by microbial soil processes stimulated by the large addition of organic matter from MBM/manure, or it was turned into stable organic P compounds which were not solubilized by the tested extractants. For class E, the estimated share of organic P was significantly higher in the minerally fertilized plots than in the ones fertilized with org-M only in CAL and Olsen, while in water and AR there was a significantly higher share of organic P in the plots fertilized with org-M. Maybe this could be explained by the fact that at high organic P loads, organic P is no longer hydrolyzed completely, but some of it remains as dissolved organic P (water-soluble) or stable organic P (which can only be digested by AR). The reason why not everything is hydrolyzed in this case may be the increase in soil pH coming along with high amounts of organic fertilizer (manure).

Inconsistent results were recorded when the influence of soil pH on the estimated share of organic P was tested. As already pointed out by other researchers, the exact role that pH plays in this respect is still unclear and needs further investigation. Some authors speculate that some indirect influence of other parameters (such as the presence of suspended soil particles in the extract) which were not measured here might have an influence. Similarly, the role of organic carbon with regard to relative shares of soluble organic P could not be clarified in our study and warrants further research.