|Main authors:||Fernando Teixeira and Gottlieb Basch
|iSQAPERiS editor:||Jane Brandt|
|Source document:||Teixeira, F. and Basch, G. (2019) Performance of promising land managment practices to populate recommendations of SQAPP. iSQAPER Project Deliverable 6.1, 45 pp|
In the spring/summer 2016, all 132 innovative AMPs and respective control fields/plots were subjected to a visual soil assessment (VSA). VSA indicators observed at that time were 2 baseline indicators, 6 soil indicators and measurement of 4 soil properties.
In the mid-spring/summer 2018, a new VSA campaign, comprising more VSA soil indicators, and measurements of an extensive range soil properties were conducted by the case study sites on the fields under selected AMPs and respective controls (a total of 20 pairs of AMP/control).
The correlations between VSA soil indicators and measured soil properties for both campaigns are reported here and the results are discussed.
|1. Results of the 2016 campaign|
|2. Results of the 2018 campaign|
|3. Discussion of results|
The number of pairs AMP/control surveyed in 2016 was 132 (n=264). Because some case study sites did not (or only partially) measured the proposed soil properties, the number of observations for those properties are lower.
Infiltration rate, pH and labile organic carbon (LOC)
The sample size for infiltration rate is n=264, for pH n=256 and for LOC n=230. VSA indicators, in broad terms, showed weak to low moderate positive correlations with infiltration rate and pH, and, for those VSA that showed a correlation with LOC, it was weak and negative. Detailed description below (also Figure 3).
VSA structure shows a weak, positive Spearman correlation with pH (rs=0.16) and infiltration rate (rs=0.21), both statistically significant for α=0.05 (Figure 3 top left). Correlation with LOC is non-existing (< |0.1|).
VSA porosity shows a moderate correlation with pH (rs=0.31) and a weak correlation with infiltration rate (rs=0.14), both statistically significant for α=0.05. Correlation with LOC is non-existing.
VSA stability (slaking test) shows a positive weak correlation with pH (rs=0.21), statistically significant. Correlations with infiltration rate and LOC are non-existing.
VSA tillage pan shows a weak, positive Spearman correlation with pH (rs=0.26) and infiltration rate (rs=0.18), and a weak negative correlation with LOC (rs=-0.15), all statistically significant.
VSA soil colour shows a moderate correlation with pH (rs=0.33) and a weak negative correlation with LOC (rs=-0.13), both statistically significant for α=0.05. Correlation with infiltration rate is non-existing.
VSA biodiversity (earthworm count) only shows a weak negative correlation with LOC (rs=-0.24).
VSA susceptibility to wind and water erosion only shows a positive weak correlation with infiltration rate (rs=0.19) and pH (rs=0.12), but only the correlation with infiltration rate is statistically significant.
VSA surface ponding shows a moderate correlation with pH (rs=0.30) and a weak correlation with infiltration rate (rs=0.19), both statistically significant for α=0.05. Correlation with LOC is non-existing.
The sample size for Spearman’s correlation studies is n=106, from 5 case study sites.
VSA tillage pan has a negative weak correlation with organic matter (rs=-0.15);
VSA biodiversity (earthworm count) has a weak positive correlation (rs=0.17);
and both VSA baseline indicators show no correlation.
The other VSA indicator have positive, statistically significant correlations with organic matter: weak correlation for VSA structure (rs=0.24), and moderate correlations with VSA soil colour (rs=0.38), VSA stability (rs=0.40) and VSA porosity (rs=0.34) (see Figure 4).
The number of pairs AMP/control surveyed in 2018 was 20 (n=40). Because some soil properties were not measured by some case study sites, n for some properties is lower. Although presented in the text below, correlations with soil texture and particles (% of sand, silt and clay) are not further discussed in this Deliverable (it will be part of Deliverable 6.2).
VSA structure shows a positive and moderate Spearman’s rank correlation coefficient with SOC, Ntot and macrofauna, respectively rs=0.45, 0.42 and 0.48, and a negative correlation with sand content, all statistically significant for α=0.05. Bordering statistical significance, a moderate negative correlation with stone content can also be observed (Figure 6, blue bars). All other measured soil properties showed no correlations, or very weak, not statistically significant.
Comparing the above correlations results with correlations using ranking after classification of measured properties (orange bars), coefficients drop for SOC (0.45 to 0.40), Ntot (0.42 to 0.30) and stone content (absolute coefficient, because of reverse ordering, from 0.36 to 0.25), but only Ntot changed (lost) statistical significance.
VSA porosity shows a positive moderate Spearman’s rank correlation coefficient with SOC, Ntot and Silt, respectively rs=0.37, 0.40 and 0.38, and a negative correlation with sand content (rs=0.41), all statistically significant. Macrofauna groups and microorganism C also show a moderate positive correlation with porosity (see Figure 7, blue bars).
Differences with results of correlations using ranking after classification of the above soil properties are slight, except for Ntot (orange bars). However, for clay (data not depicted) there is an important difference between correlation coefficients, passing from rs=-0.05 to -0.28 (it will not be discussed).
VSA stability (slaking) shows positive Spearman’s rank correlation coefficients with Ntot and SOC, respectively rs=0.36 and 0.34, statistically significant for α=0.05. And moderate/ weak correlations with microorganism C and electrical conductivity (EC).
Main differences with results of correlations using ranking after classification of soil properties are slight. There’s also a decrease in correlation with EC from 0.28 to 0.04 that will be discussed in the next section.
VSA subsoil compaction (formation of hardpans) shows a positive Spearman’s rank correlation coefficient with microorganism C, number of macrofauna groups, SOC, Ntot, texture (from coarse to medium fine textures), silt of respectively rs=0.53, 0.41, 0.44, 0.37, 0.37 and 0.40, and a negative correlation with sand and available P, respectively rs=-0.54 and -0.33, all statistically significant for α=0.05. The Spearman’s correlation between subsoil compaction and susceptibility to compaction was weak and negative and almost none existing when using only bulk density.
Using ranking after classification, we observe slight differences in the correlations although it only resulted in a change in statistical significance with Ntot (0.37 to 0.31).
VSA number and colour of mottles shows weak Spearman’s correlations with all properties but number of macrofauna groups, where a moderate positive correlation exists, although not statistically significant.
Differences in the correlations resulting from using ranking after classification, are minor and with no apparent relevance.
VSA earthworm count shows a positive, statistically significant, Spearman’s correlation coefficient with soil’s Ntot, number of macrofauna groups and stone content, respectively rs=0.37, 0.47 and -0.42. Although the correlation with susceptibility to compaction was not statistically significant, a negative and statistically significant correlation was found with bulk density (rs=-0.32) (not shown in Figure 11). Correlations with all other soil properties tested were weak.
Differences in the correlations resulting from different ranking procedures are slight.
VSA degree of clod development shows positive Spearman’s correlations with SOC, Ntot and number of macrofauna groups, respectively rs=0.41, 0.42 and 0.54, statistically significant for α=0.05. All other correlations were weak and not statistically significant.
Differences in the correlations resulting from different ranking procedure are slight.
VSA soil colour shows a positive moderate Spearman’s correlations with number of macrofauna groups, microorganism C and exc. K, respectively rs=0.44, 0.39 and -0.33, statistically significant. All other correlations are weak and not statistically significant.
Differences in the correlations resulting from different ranking are slight and/or not relevant.
VSA surface ponding shows a positive Spearman’s correlation with Ntot, rs=0.42, statistically significant for α=0.05 (Figure 14). It also depicts weak/moderate positive correlations with SOC and susceptibility to compaction, not statistically significant, and a moderate negative correlation with bulk density rs=-0.32 (data not shown in Figure 14), statistically significant for α=0.05.
Differences in the correlations resulting from different ranking are slight and/or not relevant.
VSA wind and water erosion only shows a positive moderate Spearman’s correlation with number of macrofauna groups, rs=0.43, statistically significant for α=0.05. Correlations with stone content, exc. K and sand are weak/moderate and not statistically significant.
Differences in the correlations resulting from different ranking are slight and/or not relevant.
The sample size of the field campaign of 2016 is bigger than the one in 2018, n=264 vs. n=40. Also, a larger set of VSA indicators was assessed in 2018 albeit covering all VSA indicators assessed in 2016.
In relation to measured soil properties, raw data for soil properties systematically measured at the 2016 campaign (pH, infiltration rate, and LOC) is not available at the moment, but about to being compiled. Only organic matter (OM) from 2016 (n=106, from 5 CSSs) is available. Measured properties in the campaign of 2018 compose a different set of data and, with exception of OM, cannot be directly compared. Correlations with number of macrofauna groups are discussed separately, at the end of the end of this section.
Statistically significant correlations of VSA indicators with organic matter, in 2016, are weak with VSA structure (rs=0.24), and moderate with soil colour (rs=0.38), stability (rs=0.40) and porosity (rs=0.34). These correlations are not very dissimilar from those obtained in 2018, with a lower N (40) and greater pedoclimatic zones coverage (11 CSSs), respectively rs=0.45, 0.27 (not statistically significant), 0.34 and 0.37.
From the campaign of 2016 (n=264), VSA soil structure have a positive moderate Spearman’s correlation with porosity (rs=0.42), and no correlation with stability (rs=0.05), while porosity has a weak correlation with stability (rs=0.21). Roughly the same pattern exists in the data from the campaign of 2018 (n=40), although the strength of the association is much higher (respectively, 0.72, 0.28 (not statistically significant) and 0.51). Further analysis of the correlations between these and other VSA indicators, and with measured soil properties (including texture), may throw some light on the properties and relations governing these correlations (or the lack of them).
VSA tillage pan, subsoil compaction in the campaign of 2018, shows a weak and negative correlation with OM in 2016 campaign (rs=-0.15) and a moderate positive correlation in 2018 (rs=0.44). The differences in the correlation coefficients reflects the differences in the data sets (size and pedoclimatic distribution) and the question of representability of the sample may arise. Subsoil compaction is the result of direct load applied to the soil beyond the shear stresses it can resist (its bearing capacity), thus management related. The soil bearing capacity varies and it is a function of soil texture, water content, aggregate stability, among others that may have more or less importance depending on local context. Soil compaction is commonly associated with loads applied through traffic of tractors and machinery, intensive grazing, and aggravated by soil disruption through tillage. A compacted soil presents loss of porosity with the rearrangement of the soil particles (with decrease of natural aggregates) and the formation of more or less continuous masses of hard soil of higher bulk density than the original soil. In 2016, VSA tillage pan correlated positively, and statistically significant, with VSA porosity, rs=0.28 (n=264), and this correlation is in fact the highest correlation of tillage pan with other VSA indicators. This correlation is much higher in the data from 2018, rs=0.69, and again it is the highest correlation of subsoil compaction with other VSA indicators. From the data of 2018, neither VSA subsoil compaction nor VSA porosity are explained by susceptibility to compaction (the correlations with these two VSA indicators are respectively rs=-0.17 and 0.01). The lack of correlation with susceptibility to compaction may be due to the methodological approach: sampling for bulk density was performed at 15 cm depth while porosity was assessed to a depth of 20 cm and subsoil compaction to a depth of 50 cm (by looking for evidence of hardpans); meaning that the model (susceptibility to compaction) may be well appropriate to describe compaction and loss of porosity but we failed to measure it at the appropriate depths. Correlation of subsoil compaction with bulk density is inexistent (rs=0.04). Other statistically significant correlations exist between VSA tillage pan (subsoil compaction) with other VSA indicators and measured soil properties (namely with microorganism C content in the dataset of 2018), in both years (2016 and 2018), but given the mechanisms of subsoil compaction, any interpretation would be pure speculation.
The moderate correlation of OM with VSA soil colour in 2016 (rs=0.38, n=106 from 5 CSSs) support the claim of VSA soil colour as a good indicator of OM soil status. However, a weak and negative, but statistically significant correlation of VSA soil colour with LOC (rs=-0.13, n=230 from 13 CSSs) exists, meaning a weak trend for better VSA soil colour score where LOC is depleted. Nevertheless, we suspect that LOC status classification may have played an important role and masked a hypothetical stronger correlation (that can be sorted out if CSSs give access to LOC raw data). For the 2018 campaign, VSA soil colour shows only a weak correlation with OM (rs=0.27), not statistically significant, an even lower correlation with Ntot (rs=0.21), moderate statistically significant correlation with microorganism C (rs=0.39), and moderate, negative, statistically significant correlation with exc. K (rs=-0.33). The correlation with exc. K may be a statistical artefact, due to a low sample size and the positive correlation usually observed of exc. K with soil bacteria counts (e.g. Higashida and Takao, 1986). On the other hand, the correlation with microorganism C remains open, waiting for further correlation analysis of soil colour with LOC. Another circumstantial observation linking microorganism C abundance and soil colour lies on the moderate positive correlation between VSA soil colour and measured pH in the campaign of 2016 (rs=0.33, n=256), the highest correlation between VSA indicators and pH; pH is known for the marked effect on soil microorganisms’ communities, especially bacteria, both in terms of diversity and abundance (Rousk et al. 2010). If we analyse the correlations of VSA soil colour with other VSA indicators from the campaign of 2016 (n=264), there are moderate correlations with VSA porosity and stability, respectively rs=0.38 and 0.37, while correlations with other VSA indicators are all weak, although statistically significant, and again, these VSA indicators, porosity and stability, show relatively high correlations with pH (rs=0.31 and 0.21, respectively), meaning that factors (soil properties) governing soil colour may, to some extent, govern aggregate stability in water and porosity (part of Deliverable 6.2).
VSA biodiversity (earthworm count) correlation with OM in the campaign of 2016 is weak (rs=0.17) and not statistically significant. A similar correlation of VSA earthworm count with OM was observed in 2018 (rs=0.22) but, for the related soil property Ntot, the correlation between the two is moderate and statistically significant rs=0.37. This is interesting because the equivalent correlation with SOC (considering SOC= SOM x 0.58), is very weak (rs=0.22), despite the strong linear relationship between Ntot and SOM, r=0.98 for n=45 (12 case study sites) and statistically significant for α=0.001 (Teixeira and Basch, 2019). If we consider the correlations between earthworm count and SOC, and between earthworm count and Ntot, performed after classification of SOC and Ntot according to SQAPP thresholds, we have respectively rs= 0.14 and 0.37. When we compare pairs of Ntot and SOC ranks we observe differences in 9 out of 40 pairs, and in all occasions a higher Ntot rank, meaning that the C/N ratio was lower for those pairs; the remaining 31 pairs had equal scores. We also found that the number of earthworms is positively associated to a measurable lower C/N ratio. This raises the question whether it is the earthworms that cause a lower C/N ratio or whether earthworms prefer soils with lower organic matter C/N ratios? The correlation between VSA earthworm count with LOC in 2016, although weak (rs=-0.24) is statistically significant, meaning that higher earthworm count VSA scores are associated with poorer LOC status (to be further analysed). Another interesting finding, in the 2018 campaign, is the inexistence of a correlation between VSA earthworm count and microorganism C (rs=-0.01) and the moderate correlation with VSA stability (rs=0.40), the only statistical significant correlation of VSA earthworms with other VSA indicators, backing the for long established fact that earthworm casts are stable aggregates (e.g. Shipitalo and Protz, 1989). This correlation (earthworm count with slaking) was also observed in 2016 and, although much weaker (rs=0.15), it was statistically significant; texture may play a substantial role in the mechanisms (to be further assessed).
In the dataset of 2018, VSA earthworm count shows negative moderate correlations with stone content and bulk density (not susceptibility to compaction, with rs=0.21), respectively rs=-0.42 and -0.32, both statistically significant, meaning that earthworms thrive better where there’s less mechanical impediments.
VSA number and color of soil mottles was only assessed in the 2018 campaign. Correlations between VSA soil mottles scores and measured properties were not statistically significant for any measured soil property, and all correlations are weak with the exception of a moderate correlation with macrofauna (rs=0.34). Correlations between VSA soil mottles and other VSA indicators were only statistically significant with VSA subsoil compaction (rs=0.39) and degree of clod development (rs=0.43). The correlation with subsoil compaction is expected due to the reduction of soil aeration and waterlogging associated to compaction but another cause may be a high water table (poor drainage), and thus only a moderate correlation. The correlations with degree of clod development and subsoil compaction should be further investigated.
Similarly, VSA degree of clod development was only assessed in 2018. The only correlations of VSA degree of clod development with measured properties that are statistically significant are with OM (SOC) rs=0.41 and with related Ntot, rs=0.42. With exception of the correlation with VSA earthworm count, that was weak and not statistically significant, correlations with other VSA indicators, are all moderate/strong, ranging from 0.36 with VSA surface ponding to 0.78 with VSA porosity. The correlation of VSA degree of clod development with subsoil compaction in the data from 2018 campaign is rs=0.66. The important correlations with most VSA indicators make this VSA indicator very important for a quick soil assessment, and especially by the fact that it may constitute a good visual indicator (surface indicator) of subsoil compaction, especially because of the generalized believe that compaction leaves no telltale signs on the soil surface. Tilled surfaces of compacted soils will show broken pieces of the masses of hard soil, of higher bulk density than the original soil that may persist for longer on the soil surface after rainfall (higher rainfall accumulation until smoothing the surface).
Baseline indicator VSA surface ponding shows in the data from 2016 (n=240), a positive, weak but statistically significant correlation with all other VSA indicators. In the data from 2018 (n=40), correlations are weak or inexistent with most VSA indicators with the exception of porosity and degree of clod development, where a moderate and statistically significant correlation exists, both rs=0.36. Correlations with measured properties in the data from 2016, are moderate with soil pH (rs=0.30) and weak with infiltration rate (rs=0.19); and very weak (non-existing) with OM and LOC. In the data of 2018, with the exception of Ntot (rs=0.42), correlations with all other measured properties were either weak or non-existing. An interesting finding is the relatively high correlation with VSA susceptibility to compaction, rs=0.27, not statistically significant, and even higher correlation with bulk density rs=-0.32, statistically significant, meaning better VSA scores are higher where bulk density is lower. These results reflect the known effect of the measured properties but are puzzling in relation to OM, very weak correlation in 2016 and moderate in 2018, but lower than Ntot; otherwise a better porosity leads to higher soil infiltration capacity; soils with a lower bulk density (less compacted) preserve higher porosity and pore continuity and thus higher soil infiltration capacity.
Baseline VSA susceptibility to water and wind erosion in the data of 2016 (n=256) shows a positive, weak but statistically significant correlation with all other VSA indicators with exception of VSA structure and tillage pan, with which there are no correlations. Differently from 2016, data in 2018 (n=40) shows moderate positive correlations, statistically significant, with structure and subsoil compaction, no correlations with VSA mottles (only measured in 2018) and surface ponding, and only a weak and not statistically significant correlation with VSA earthworm count (rs=0.22). Correlation with measured properties in the data of 2016 is only statistically significant with infiltration rate (rs=0.19). In the data from 2018, only the correlation with stone content is moderate (rs=-0.35), and not statistically significant.
pH in the campaign of 2016 (n=256), showed weak to moderate positive correlations, statistically significant with most VSA indicators, varying from rs=0.12, with VSA susceptibility to erosion, to rs=0.33 with VSA soil colour, the exception being with VSA biodiversity (earthworm count) where no correlation exists. Although the results from 2016 cannot be directly compared with the results from 2018, due to different protocol (pH measured in CaCl2) and thresholds used for pH classification, weak correlations, not statistically significant, are observed for some VSA indicators namely VSA structure and stability (slaking), both with a rs=0.19.
Infiltration rate in the campaign of 2016 (n=264) showed only weak correlations with some VSA indicators, although statistically significant, varying from rs=0.14, with VSA porosity, to rs=0.21 with VSA structure. The lack of correlation with VSA stability (slaking) and VSA biodiversity (earthworm count), is noteworthy because: 1) it allows to question the claim of a universal effect of earthworm count on water infiltration; 2) it allows to question the use of VSA stability (slaking), irrespectively of soil properties, namely its texture, and the matric water potential at the time of the VSA assessment, as an indicator of status regarding water infiltration.
LOC in the campaign of 2016 (n=230) showed only weak and negative correlations with VSA tillage pan (rs=-0.15), soil colour (rs=-0.13) and biodiversity (earthworm count) (rs=-0.24), already discussed above. No correlations were found with the rest of VSA indicators.
Measured soil chemical properties
Exchangeable K and available P, with 2 exceptions, show only weak (or inexistence of) correlations with VSA indicators. Exc. K shows only a moderate, negative and statistically significant correlation with VSA soil colour (rs=-0.33). Available P shows a negative, moderate but statistically significant correlation with VSA subsoil compaction (rs=-0.33). This correlation of VSA subsoil compaction with available P may be explained by waterlogging that occurs where soil is compacted (lower VSA score), with the transformation of ferric phosphates to more soluble forms of ferrous phosphates, and thus a higher content of available P (e.g. Patrick and Mahapatra, 1968). Correlation between exc. K and VSA soil colour requires further analysis. Electric conductivity (EC) was measured only in 2018 (n=40). Correlations of EC with VSA indicators are weak or inexistent. EC values at the 40 locations were all below 2 dS m-1 (value considered the upper threshold for good condition: below this value salinity is not a threat). The highest, not significant correlation coefficient (rs=0.28) was found between EC and VSA stability (slaking test), but disappears when ranking is performed after EC classification. Nevertheless, it shows a potential positive correlation with stability in the EC interval of the 2018 data.
Texture and particle size proportion (of sand, silt and clay) correlations with VSA indicators are not discussed because of the low sample size (n=40), range intervals, and the opposite/ contradictory results from an equally low sample (n=56, from Estonia, Romania and The Netherlands (3)) from 2016 campaign. An extended and meaningful study can be performed if the CSSs forward the raw data from the 264 sites.
Number of macrofauna groups was measured only in 2018 (n=28) and the correlation results with VSA indicators show that it may become a good, non-specific, overall indicator of soil quality status, especially if local indexes, based on local macrofauna, easily recognizable, can be developed, to lower the level of expertise needed that the present protocol requires. Moderate positive correlations, statistically significant, exist for several VSA indicators, including VSA earthworm count – in comparison, VSA earthworm count has a statistical significant correlation only with VSA stability. The small sample size restrains us from further analysis.
Note: For full references to papers quoted in this article see