Eight sampling sites scored against six soil ecosystem functions using semi-quantitative criteria — presence and abundance of named indicator organisms in the metabarcoding data, not computed numerical indices. The scoring is deliberate-qualitative at this stage; a path to calculated indices (e.g. Bongers Maturity Index for Function 6) is discussed in the function definitions and limitations. Click any column header for the function definition; click any row label for the detailed site scorecard.
| Site | N cycling | C / OM processing | Aggregation (4 layers) | Plant–microbe symbiosis | Disease & pest suppression | Food web maturity |
|---|---|---|---|---|---|---|
| ForsthausUnten | ✓ | ~ | ~ | ✗ | ~ | ✗ |
| ForsthausOben | ✓ | ✓ | ~ | ? | ~ | ~ |
| Alter Obstgarten | ✓ | ✓ | ~ | ? | ~ | ~ |
| Market Garden Mix | ✓ | ✓ | ~ | ? | ~ | ✓ |
| Market Garden 70 | ✓ | ✓ | ~ | ? | ~ | ~ |
| Haus Hügel | ✓ | ✓ | ~ | ? | ✓ | ✓ |
| Spannweid Mix | ✓ | ✓ | ✓ | ~ | ~ | ~ |
| Spannweid 14 | ✓ | ~ | ~ | ? | ~ | ~ |
Function 1 (N cycling) is complete at every site. The Nitrososphaera + Nitrospira chain is present even at the most disturbed site (ForsthausUnten). This is the most robust function detected and the easiest to score with confidence.
Function 4 (Plant–microbe symbiosis) is the systematic weakness — and almost certainly methodology-driven. Six of eight sites score ? (underdetected), one ✗, one ~. Whether real AMF / rhizobial / PGPR communities are actually present beneath the detection threshold is the single most important question this scorecard raises. A targeted SSU AMF survey at a subset of sites would resolve it.
Function 6 (Food web maturity) maps cleanly onto management duration. Market Garden Mix and Haus Hügel — the two longest-managed E-sites — both score complete. ForsthausUnten (newest, with disturbance legacy) scores ✗. This is the function most aligned with the Bongers / Ferris nematode-maturity tradition, and the ordering matches that tradition's expectations.
Function 3 (Aggregation) shows the four-layer model working as predicted. Layers 1–3 are present at almost every site; Layer 4 is the holdout. The Market Garden 70 spade-test correlation is the clearest direct evidence that the indicator organisms for Layers 1–3 correspond to a physically measurable structural outcome.
Three sites stand out for distinct reasons. Spannweid Mix is the only site scoring complete on aggregation due to its AMF signal. Haus Hügel is the only site scoring complete on disease / pest suppression due to its multi-predator stack. Market Garden Mix is the only site scoring complete on food web maturity due to its breadth across phyla and trophic levels.
No single site scores complete on all six functions. The closest approximations are Market Garden Mix and Spannweid Mix (each complete on four functions). This is consistent with an in-development system and gives a defensible "where we are in the journey" framing.
2024 state: bare soil with Gülle (slurry) legacy from previous land manager. Pumpkin / clover culture only just beginning to establish at sampling.
N cycling — ✓ Complete. Nitrososphaera viennensis at 2.14% of bacterial reads (highest of any A-site) plus Nitrospira japonica present. Nitrification chain intact despite the otherwise disturbed state of the site.
C / OM processing — ~ Partial. Decomposer community present but skewed: dominated by Podospora appendiculata (dung specialist, 3.69%) and Keratinophyton wagneri (keratin decomposer) — both signatures of animal-derived organic matter rather than diverse plant residues. Trichoderma not detected. Plant-litter decomposer guilds underrepresented.
Aggregation — ~ Partial. Layer 1 (bacterial EPS) present but Actinobacteriota count lower than E-sites; Layer 2 (fungal hyphae) present; Layer 3 (wormcast) weak — Enchytraeidae signal minimal; Layer 4 (AMF / GRSP) not detected.
Plant–microbe symbiosis — ✗ Absent. No AMF, no named rhizobia, no PGPR named species. AMF and Bacillus are systematically underdetected by standard primers, so part of this absence is methodology. However, at sampling time the site had effectively no established plant rooting (bare soil, Gülle legacy, pumpkin/clover only just emerging), so the biological basis for low plant–microbe symbiosis is independently defensible. Scored ✗ rather than ? on the bare-soil reasoning.
Disease & pest suppression — ~ Partial. Clonostachys intermedia detected. No predatory invertebrates named at this site.
Food web maturity — ✗ Absent / Insufficient. Lowest invertebrate diversity in the entire dataset (17 OTUs, 1 named species). Truncated food web — springtails present (5 OTUs) but nematode community minimal, no detected predatory invertebrates. Consistent with the site's history.
2024 state: four-year-old Swiss Kunstwiese (sown grass-and-legume meadow), established spring 2020. (Since converted to pumpkin in 2025 and winter grain in 2026.)
N cycling — ✓ Complete. Full Nitrososphaera + Nitrospira chain present.
C / OM processing — ✓ Complete. Full decomposer suite, Humicola udagawae present (cellulose decomposition).
Aggregation — ~ Partial. Layers 1–3 present and well-developed (Actinobacteriota strong, Marionina argentea detected). Layer 4 (AMF / GRSP) not detected.
Plant–microbe symbiosis — ? Underdetected. No named AMF, no named rhizobia. A four-year legume-rich Kunstwiese should plausibly host an active rhizobial community; absence in the data is likely a methodology artefact (16S species-level resolution + DNA extraction bias for Bacillus, and ITS underdetection for AMF). Scored as underdetected.
Disease & pest suppression — ~ Partial. No Clonostachys or Trichoderma named here; no predatory invertebrates named. Some Dorylaimida OTUs (potentially predatory) but unnamed.
Food web maturity — ~ Partial. 30 invertebrate OTUs, the most diverse nematode community of any site (18 OTUs), including bacterial feeders and plant parasites. K-strategist representation moderate. Lacks the higher trophic levels seen at Market Garden and Haus Hügel.
Note on plant-parasitic nematodes: Pratylenchus neglectus and Helicotylenchus minzi detected at low levels. Typical for established Kunstwiese; worth watching after the 2025 conversion to annual crops.
2024 state: old orchard newly under management, cover crop established autumn 2023 (~10 months at sampling). Since diversified into vegetable / flower succession.
N cycling — ✓ Complete. Full nitrification chain.
C / OM processing — ✓ Complete. Trichoderma reesei detected, Humicola present, Cellvibrio spp. detected (strongest at this and Market Garden sites).
Aggregation — ~ Partial. Layer 1 strong (highest Verrucomicrobiota count of any site at 32 OTUs, dense Actinobacteriota). Layer 2 present. Layer 3 (Marionina argentea) present. Layer 4 not detected.
Plant–microbe symbiosis — ? Underdetected. No named AMF, no named rhizobia. With ~10 months of cover crop establishment at sampling and subsequent diversified vegetable / flower planting, some AMF colonisation would be plausibly expected; ITS underdetection of Glomeromycota likely dominates over real absence. Scored as underdetected.
Disease & pest suppression — ~ Partial. Trichoderma reesei present. Clarkus papillatus (predatory nematode, Mononchidae) detected — one of only two sites with a confirmed nematode predator. No named predatory mites or beetles.
Food web maturity — ~ Partial. 24 invertebrate OTUs. Predatory nematode present. Striking bacterial functional diversity (3.27, equal to Market Garden Mix) despite short management duration.
Hypothesis from this scorecard: The autumn 2023 cover crop appears to have rapidly built bacterial diversity to a level matching the longest-managed E-site, while fungal and invertebrate communities remain in early stages. If reproducible, this identifies cover crops as a high-leverage early intervention.
2024 state: longest-managed site, loamy soil, intensive market garden operation. Mix probe across the whole field.
N cycling — ✓ Complete. Nitrososphaera plus two Nitrospira species (N. japonica and N. moscoviensis) — only site with two Nitrospira spp., suggesting functional redundancy in nitrite oxidation.
C / OM processing — ✓ Complete. Full decomposer suite, Trichoderma reesei, Humicola, Cellvibrio spp. Highest fungal OTU richness in the dataset (174).
Aggregation — ~ Partial. Layers 1–3 exceptionally strong (Actinobacteriota dense, full fungal hyphal network, Marionina argentea at 11.92% of invertebrate reads — strongest enchytraeid signal in the dataset, plus detected Lumbricidae). Layer 4 absent.
Plant–microbe symbiosis — ? Underdetected. No AMF detected. The absence at the longest-managed site is methodologically ambiguous. Two possible interpretations: (a) AMF establishment lags behind other community development under high-disturbance market garden conditions (frequent crop turnover, repeated tillage events), or (b) the ITS primer limitation dominates and AMF are present but invisible. Scored as underdetected given the ambiguity.
Disease & pest suppression — ~ Partial. Trichoderma reesei, Veigaia nemorensis (predatory mite) detected. No predatory nematode named, no ground beetle named.
Food web maturity — ✓ Complete. 45 invertebrate OTUs across 8 phyla (highest of dataset). All four trophic levels represented. Microhabitat-complexity indicators present (rotifers, tardigrades, flatworms, gastrotrichs).
2024 state: single stripe within Market Garden, 5/5 on spade test for aggregate structure at sampling. (Subsequently dropped to 3 in autumn 2024, recovered to 4+ by 2026; now under winter grain.)
N cycling — ✓ Complete. Two Nitrospira species plus Nitrososphaera.
C / OM processing — ✓ Complete. Standard decomposer suite well-represented.
Aggregation — ~ Partial. Layers 1–3 strong (highest bacterial evolutionary diversity in the dataset at 20.44; aggregate-building bacteria all present). Layer 4 absent. The 5/5 spade test at sampling time correlates directly with the strength of Layers 1–3 here.
Plant–microbe symbiosis — ? Underdetected. Same pattern as Market Garden Mix — no AMF detected, methodology ambiguity dominates.
Disease & pest suppression — ~ Partial. Veigaia nemorensis detected.
Food web maturity — ~ Partial. 29 invertebrate OTUs (lower than Mix due to spatial narrowness of single-stripe sampling). Tardigrades present. Score reflects the spatial-coverage effect, not necessarily ecological inferiority.
2024 state: several years under management, peat-and-loamy soil. Strong arthropod community at sampling.
N cycling — ✓ Complete. Full nitrification chain.
C / OM processing — ✓ Complete. Standard decomposer suite. Microlunatus panaciterrae at its peak abundance here (0.71%), suggesting active P cycling alongside C cycling.
Aggregation — ~ Partial. Layers 1–3 present. Layer 4 absent.
Plant–microbe symbiosis — ? Underdetected. No AMF, no named rhizobia. With several years of management and established cover, AMF would be ecologically expected; absence is more plausibly a methodology artefact than real ecology.
Disease & pest suppression — ✓ Complete. The most complete predator suite in the dataset: Veigaia nemorensis (predatory mite, Mesostigmata), Histiostoma feroniarum (mite, exclusively here, 5.32% of invertebrate reads), Clarkus papillatus (predatory nematode), and Poecilus cupreus (ground beetle). Four predators across multiple trophic levels — this combination did not appear at any other site.
Food web maturity — ✓ Complete. 34 invertebrate OTUs across multiple phyla. All four trophic levels well-represented. K-strategists (predatory mites, Mononchidae) detected.
2024 state: several years under management, contains both peat and loamy soil zones within one field. Mix probe across the whole field.
N cycling — ✓ Complete. Full chain present.
C / OM processing — ✓ Complete. The most functionally diverse fungal community of any site (functional diversity 2.08, evolutionary diversity 34.67 — both highest in dataset). Mortierella alpina detected only here (high-quality OM indicator). Humicola udagawae at its peak abundance (1.64%).
Aggregation — ✓ Complete. The only site scoring complete on all four layers. Layer 4 detection via Claroideoglomus claroideum (AMF, Glomeromycota). All other layers present.
Plant–microbe symbiosis — ~ Partial. AMF detected (one species), making this the only site with any AMF signal. Likely partial detection (ITS primer limitations) rather than the full community.
Disease & pest suppression — ~ Partial. Clonostachys intermedia at its strongest signal (0.40%, the highest in the dataset). Tyrannicordyceps fratricida (mycoparasitic fungus) detected. Oppiella nova (oribatid mite, bioindicator). No predatory nematode named.
Food web maturity — ~ Partial. 22 invertebrate OTUs (lower than expected). Oribatid mites (sensitive K-strategists) detected. Tardigrades present. Lacks the trophic completeness of Market Garden Mix or Haus Hügel.
Caveat for Spannweid Mix: The mix probe captures two distinct soil types (peat + loam) within a single sample, which partly inflates fungal diversity metrics. The AMF signal may originate specifically from the loamy zone.
2024 state: single stripe within Spannweid field, deliberate Urtica dioica (nettle) culture established autumn 2022.
N cycling — ✓ Complete. Nitrososphaera particularly strong here (2.16%, the highest of any site), consistent with nitrogen-rich nettle litter inputs.
C / OM processing — ~ Partial. Decomposer community present but distinctively specialised — Acidobacteriota dominance (41 OTUs, the highest of E-sites). Rhizophlyctis rosea (chytrid, moist-soil cellulose decomposer) detected here and at ForsthausUnten only. Narrower functional breadth than mix sites.
Aggregation — ~ Partial. Layers 1–3 present. Layer 4 absent (despite Spannweid Mix detection — the AMF signal there likely came from outside this stripe).
Plant–microbe symbiosis — ? Underdetected. No AMF detected, despite AMF detection in the Spannweid Mix from the same field. May reflect real ecological reality of the nettle stripe (different rhizosphere chemistry, possibly less AMF-favourable), or ITS methodology artefact. Scored as underdetected for consistency.
Disease & pest suppression — ~ Partial. Clonostachys intermedia detected. No predatory invertebrates named.
Food web maturity — ~ Partial. 22 invertebrate OTUs. Tardigrades and a centipede (Chilopoda, the only such detection in the dataset) present.
Methodological observation: Despite minimal intervention, Spannweid 14 does not score higher than the more intensively managed mix sites. Its 20 unique OTUs constitute a specialised community shaped by nettle chemistry, not a generally richer one. Reinforces that "less management" and "more biodiversity" are not equivalent.
Specific feedback would be most valuable on the following:
Capacity vs. activity. eDNA detects functional capacity, not active function. The presence of Nitrososphaera indicates capacity for ammonia oxidation; it does not prove that ammonia is currently being oxidised. RNA-based methods, enzyme assays, or stable-isotope tracing would close this gap but at substantially higher cost.
Single sample per site (N=1). The 2024 design has no within-site replication, so differences between sites are descriptive rather than statistically testable. A 2–3-replicate design at a future sampling round would allow inferential statements.
AMF underdetection. The biggest single methodological gap. Standard ITS primers do not capture Glomeromycota well; SSU-targeted sequencing would be required for reliable Function 4 scoring. This single change would substantially improve the scorecard.
Bacillus underdetection. Endospore-forming Bacillus species are difficult to lyse during DNA extraction, leading to systematic under-representation. Relevant for Function 4 (PGPR) and Function 5 (some biocontrol Bacilli).
Earthworm species resolution. 18S marker resolves Lumbricidae at family level only. For Function 3 Layer 3 scoring, this prevents distinguishing between ecologically very different species (epigeic, endogeic, anecic). Aporrectodea / Lumbricus / Octolasion-level resolution would require COI-based eDNA or physical sampling.
Nematode resolution. The 18S marker resolves nematode taxa at phylum level robustly and at family / genus level reasonably well; species-level resolution varies, with cryptic species often lumped and some lineages systematically over- or under-amplified. Comparative studies (Griffiths et al. 2018 and others) report around 13–15% species-level overlap between eDNA and morphological identification on the same samples. For functional-group assignment (bacterial feeder / fungal feeder / predator / plant parasite / omnivore) the family-level resolution is sufficient and the data is fit for purpose. For absolute density estimation, strict species census, or direct comparison with traditional agricultural-soil nematode surveys, microscopy remains the standard and eDNA complements rather than replaces it.
Functional gene metagenomics is not done. The current data is taxonomic metabarcoding, not metagenomic. nifH, amoA, nirK/S, nosZ gene sequencing would provide direct evidence of function rather than indicator-organism inference. More expensive, but for an external-facing rigorous framework it might be worth a one-time validation pass.
Scoring thresholds are presently judgement-based, not quantitative. "Partial" vs. "Complete" is currently a qualitative call. For the scorecard to be more rigorous, each function should ideally have a quantitative threshold (e.g. "complete = ≥3 of 4 indicator organisms detected at ≥X% relative abundance").
The framework has not been externally validated. Applied to one dataset of eight samples. To be defensible as a tool, it would need to be applied to additional datasets (other regenerative farms, other soil types, ideally including conventional contrasts).
This document is a first attempt to score the eight 2024 HofLabor sampling sites against a transparent, organism-by-organism framework of soil ecosystem functions, rather than against opaque composite indices. The objective is to test whether a mechanism-based scorecard can produce useful, defensible per-site readouts from standard metabarcoding data (16S + ITS + 18S), and to identify where additional or different methods are needed to close gaps.
HofLabor's working hypothesis is that biodiversity is a production factor: a functioning soil ecosystem substitutes biologically for what conventional agriculture supplies through chemical inputs. Testing this hypothesis requires not just diversity counts but evidence that the specific biological functions implicated in input substitution — nitrogen cycling, aggregate building, disease suppression, plant–microbe symbiosis — are actually present at each site.
The scorecard is explicitly a draft. The function definitions, indicator organisms, and scoring criteria are open to revision. The questions section above lists the specific points where feedback would be most valuable.
Six functions were selected as the framework. For each function:
The scoring reflects functional capacity, not active function. eDNA detects DNA, not activity. The presence of an indicator organism indicates capacity for the corresponding function; it does not prove the function is currently being performed at that location and time.
What it is. The biological conversion of organic and atmospheric nitrogen into plant-available forms (ammonia → nitrite → nitrate) and back (denitrification under anaerobic conditions). In an input-free system, complete nitrogen cycling is essential, because all nitrogen must come from biological fixation, atmospheric deposition, and the mineralisation of organic matter.
Indicators (16S): Ammonia oxidisers — Nitrososphaera viennensis (Thermoproteota archaea) and other AOA; Nitrosomonas / Nitrosospira (AOB, rare in our data). Nitrite oxidisers — Nitrospira spp. (incl. potential comammox). Free-living and symbiotic N-fixers — Rhizobium, Bradyrhizobium, Azospirillum, Frankia.
Limitation: The N-fixation step is poorly resolved by 16S taxonomy alone. Reliable N-fixation assessment requires nifH gene sequencing (metagenomic).
What it is. The biological breakdown of plant residues, root exudates and other organic matter inputs, and the stabilisation of a fraction of that material as soil organic matter. Without external inputs, the rate and completeness of OM processing determines nutrient release and SOM build-up.
Indicators (ITS, 16S): Cellulose decomposers — Trichoderma, Humicola, Cellvibrio. Polymer decomposers / dominant decomposer guilds — Sordariomycetes (Hypocreales, Sordariales), Dothideomycetes. OM-quality indicators — Mortierellomycetes (high-quality OM), Mortierella alpina specifically. Polysaccharide decomposers — Verrucomicrobium, Flavisolibacter.
What it is. The physical assembly of mineral particles into stable aggregates, on which water infiltration, drainage, aeration, root penetration and SOM stabilisation all depend. Aggregation operates at four mechanistic scales, each driven by a distinct organism guild.
Layer 1 — Microaggregates (bacterial EPS): Filamentous Actinobacteriota (count), Nakamurella flavida, Microlunatus panaciterrae, Sphingomonas spp., Gemmatimonas aurantiaca.
Layer 2 — Macroaggregates (fungal hyphae): Sordariomycetes diversity, filamentous saprotroph richness, Humicola udagawae.
Layer 3 — Wormcast aggregates: Lumbricidae (18S, family-level only), Enchytraeidae (Marionina argentea, Enchytraeus dichaetus).
Layer 4 — Glomalin-related soil protein (GRSP) and AMF: Glomeromycota presence (Claroideoglomus, Rhizophagus, Funneliformis, etc.). Per current understanding, GRSP is an operationally defined mixture of soil organic matter, not a pure AMF glycoprotein, but AMF remain the most-studied contributors to this fraction.
Limitation: Layer 4 is systematically underdetected by standard ITS primers. SSU-targeted Glomeromycota sequencing would be needed for reliable assessment.
What it is. Cooperative interactions between plant roots and soil microbes that deliver nutrients, water, hormones, and immune priming to the plant in exchange for carbon. The dominant categories are arbuscular mycorrhizal symbiosis (most plants), rhizobial nodulation (legumes), and free-living plant-growth-promoting rhizobacteria (PGPR).
Indicators: AMF — Glomeromycota (see Function 3, Layer 4). Rhizobia — Rhizobium, Bradyrhizobium, Sinorhizobium, Mesorhizobium. PGPR — Pseudomonas spp., Bacillus spp., Azospirillum, certain Burkholderia.
Limitation: AMF and Bacillus are both underdetected by standard primers. Rhizobial detection requires named-species resolution that 16S often does not provide reliably.
What it is. Biological capacity to limit populations of plant-pathogenic fungi, bacteria, nematodes, and arthropod pests. In a fungicide-free, insecticide-free system, this is the primary protection mechanism.
Mycoparasitic / antagonistic fungi: Clonostachys intermedia (broad-spectrum, Fusarium suppressor), Trichoderma spp.
Predatory soil fauna: Predatory nematodes — Mononchidae (e.g. Clarkus papillatus), Nygolaimidae. Predatory mites — Mesostigmata (e.g. Veigaia nemorensis), Sarcoptiformes predators. Ground beetles — Carabidae (e.g. Poecilus cupreus). Predatory pseudoscorpions, centipedes.
Entomopathogenic fungi: Beauveria bassiana, Metarhizium spp. (mostly absent in our data; Tyrannicordyceps fratricida at one site as nearest analogue).
What it is. The structural completeness of the soil food web, measured by trophic level coverage, predator presence, and representation of disturbance-sensitive K-strategists. A mature food web is correlated with lower disturbance, longer management continuity, and (generally) higher functional redundancy.
Indicators: Trophic completeness — organisms detected at primary decomposer, microbial grazer, secondary consumer (predator), and plant-feeding levels. Predator diversity — count of distinct predator OTUs across mites, nematodes, beetles, centipedes. K-strategist presence — high c-p nematodes (Mononchidae, Dorylaimida), oribatid mites (e.g. Oppiella nova), tardigrades, slow-reproducing arthropods. Total invertebrate OTU richness as a coarse summary.
Reference framework, and the gap to it. Conceptually aligned with the Bongers Maturity Index (1990) and the Ferris Enrichment, Structure, and Channel Indices (2001). However, the scores in this draft are qualitative — based on trophic completeness, predator presence, and K-strategist indicators — not computed from c-p values multiplied by proportional abundances. A calculated Bongers MI (and the Ferris EI/SI/CI) using the 2024 eDNA data would be feasible as a next iteration. The main methodological caveat: eDNA read counts are not equivalent to nematode population counts (PCR amplification bias, DNA copy-number variation, body-size differences), so an MI computed from eDNA reads is an approximation — directionally informative but not numerically identical to one computed from morphological counts. This is an active methods question in the literature, with papers both proposing and critiquing eDNA-based MI calculations.