Contrary to most soils, permafrost soils have the
atypical feature of being almost entirely deprived of soil fauna. Abiotic
constraints on the fate of permafrost carbon after thawing are increasingly
understood, but biotic constraints remain scarcely investigated. Incubation
studies, essential to estimate effects of permafrost thaw on carbon cycling,
typically measure the consequences of permafrost thaw in isolation from the
topsoil and thus do not account for the effects of altered biotic
interactions because of e.g. colonization by soil fauna. Microarthropods
facilitate the dispersal of microorganisms in soil, both on their cuticle
(ectozoochory) and through their digestive tract (endozoochory), which may
be particularly important in permafrost soils, considering that microbial
community composition can strongly constrain permafrost biogeochemical
processes.
Here we tested how a model species of microarthropod (the Collembola
Folsomia candida) affected aerobic CO2 production of permafrost soil over a 25 d
incubation. By using Collembola stock cultures grown on permafrost soil or
on an arctic topsoil, we aimed to assess the potential for endo- and
ectozoochory of soil bacteria, while cultures grown on gypsum and sprayed
with soil suspensions would allow the observation of only ectozoochory.
The presence of Collembola introduced bacterial amplicon sequence variants (ASVs) absent in the
no-Collembola control, regardless of their microbiome manipulation, when
considering presence–absence metrics (unweighted UniFrac metrics), which
resulted in increased species richness. However, these introduced ASVs did
not induce changes in bacterial community composition as a whole (accounting
for relative abundances, weighted UniFrac), which might only become
detectable in the longer term.
CO2 production was increased by 25.85 % in the presence of
Collembola, about half of which could be attributed to Collembola
respiration based on respiration rates measured in the absence of soil. We
argue that the rest of the CO2 being respired can be considered a
priming effect of the presence of Collembola, i.e. a stimulation of
permafrost CO2 production in the presence of active microarthropod
decomposers. Overall, our findings underline the importance of biotic
interactions in permafrost biogeochemical processes and the need to explore
the additive or interactive effects of other soil food web groups of which
permafrost soils are deprived.
Introduction
Carbon fluxes from soils are largely governed by the rate of decomposition
of organic matter. Soil fauna is a crucial component in organic matter
decomposition (García-Palacios et al., 2013; Griffiths et al., 2021),
both directly through mechanical degradation of litter and corpses into
smaller pieces and indirectly through feeding habits controlling the
abundances of other decomposer groups such as fungi, microbial eukaryotes or
bacteria (Hanlon and Anderson, 1979; Kaneda and Kaneko, 2008; Frouz et al.,
2020; Potapov et al., 2020). In arctic soils, the scarcity of macrofaunal
decomposers (e.g. earthworms – Blume-Werry et al., 2020) coupled with the high
abundance of microbivorous microarthropods such as Collembola (Potapov et
al., 2022) results in a particularly strong impact of Collembola on
decomposition through microbial population control (Koltz et al., 2018;
Crowther et al., 2012; Seastedt and Crossley, 1984).
Contrary to other important groups of soil fauna, such as earthworms,
millipedes (Golovatch and Kime, 2009; Berman et al., 2015) or woodlice
(Sfenthourakis and Hornung, 2018), Collembola are ubiquitous in arctic soils,
where they can reach high densities – up to 130 000 individuals per square metre
in high-Arctic Greenland (Sørensen et al., 2006). However, Collembola are
mostly abundant in the topsoil and to our best knowledge have never been
observed in the perennially frozen subsoil, the permafrost. Permafrost soils
are a prominent feature of Arctic landscapes, and the huge carbon stock they
represent and positive feedback to climate change that their thawing will
likely induce cause concern (Masson-Delmotte et al., 2021). The frozen
conditions over long periods of time have eliminated most fauna and
non-microbial life from these environments, and despite the possibility of
resuscitating organisms such as plants (Yashina et al., 2012), nematodes
(Shatilovich et al., 2018) or rotifers (Shmakova et al., 2021), newly thawed
permafrost soils generally harbour an extremely simplified soil food web
entirely deprived of metazoans. Despite this particularity, the absence of
Collembola and the consequence of their possible introduction into newly thawed permafrost on its biogeochemical cycling have been mostly overlooked.
In contrast to Collembola, some microorganisms survive and/or thrive in
permafrost conditions, and microbial adaptations to frozen conditions have
been studied (Mackelprang et al., 2011, 2017; Hultman et al., 2015; Bottos
et al., 2018). However, not all microbes survive these conditions, and the
combination of environmental constraints exerted over long periods of time
(Mackelprang et al., 2017) and strong dispersal limitations (Bottos et al.,
2018) results in microbial communities that can be deprived of some functions
(Knoblauch et al., 2018; Monteux et al., 2020; Barbato et al., 2022). The
re-introduction of such functions can result in drastic changes in
permafrost processes, and sizable impacts on greenhouse gas production have
been observed in vitro for CH4 and CO2 (Knoblauch et al., 2018;
Monteux et al., 2020) and confirmed in situ for N2O (Marushchak et al.,
2021). Upon thawing, this re-introduction of missing functions or ecological
rescue (Calderón et al., 2017) requires microorganisms to migrate into
this newly available habitat, which could happen for instance laterally
through airborne dispersal (Harding et al., 2011) – e.g. for permafrost
exposed to the air in abrupt thaw processes (Inglese et al., 2017) – or
vertically through percolation in the soil column when the active layer
becomes deeper (Monteux et al., 2018; Johnston et al., 2019).
Collembola are another possible vector for the dispersal of microorganisms
into newly thawed permafrost (Buse et al., 2014). With the deepening of the
active layer, Collembola migration into newly thawed permafrost is unlikely
since they mostly reside in topsoil layers, but it will likely occur in soil-mixing events (Väisänen et al., 2020) such as thermokarst, active-layer detachment or thaw slump processes where newly thawed permafrost is
exposed to surface conditions. Like all large organisms, Collembola host a
variety of microorganisms, their microbiome (Agamennone et al., 2015; Leo et
al., 2021). Because microarthropods, such as Collembola, can move across
large distances compared to fungi or bacteria, the microbial species in
collembolan microbiomes might be among the first to colonize and establish
in newly thawed permafrost when Collembola access it. This could occur
through a combination of two main processes: ectozoochory, where the
microbiome of the cuticle disperses into the new habitat, and endozoochory,
where microorganisms disperse after transiting through the gut of the
animals. To our best knowledge, whether Collembola affect the biogeochemical
functioning of newly thawed permafrost and whether and how they can serve
as a vector for microbial colonization have not been explored yet.
We incubated permafrost from the Yedoma domain, which represents a large
carbon stock in parts of Siberia and North America (Strauss et al., 2017)
and has previously been shown to lack certain microbial functions (Monteux
et al., 2020), in the presence or absence of a model species of
microarthropod (Collembola Folsomia candida), and assessed CO2 production and bacterial
community composition over a 25 d aerobic incubation. We used Folsomia candida Collembola
from a stock culture, as well as Collembola subjected to manipulation of
their cuticle microbiome or cuticle and gut microbiomes to test the
following hypotheses:
Collembola presence in permafrost will alter bacterial community
composition, through their grazing.
Collembola more closely exposed to topsoil bacteria will change permafrost
bacterial community composition further than those exposed less intricately.
In other words, we expect a gradual change between clean Collembola,
Collembola with cuticle microbiome manipulated (ectozoochory), and Collembola
with both gut and cuticle microbiomes manipulated (endozoochory).
Collembola presence in permafrost will increase CO2 production, through
both Collembola respiration and a stimulation of microbial activity
(“priming effect”).
The gradual introduction of distinct bacterial communities hypothesized
above (2) will result in increased CO2 production, owing to the
functional limitations of the Yedoma permafrost microbial community in terms of
CO2 production.
Materials and methodsExperimental designSoils
The Yedoma sediment used to assess the impacts of Collembola on permafrost
CO2 production and bacterial community composition originated from the
Cold Regions Research and Engineering Laboratory (CRREL) permafrost tunnel (Fox,
Alaska, USA). The sediment was sampled from the upper silt unit and is an
Upper Pleistocene silty deposit, previously described in detail (Shur et
al., 2004; Mackelprang et al., 2011, 2017; Monteux et al., 2020). This
sediment was chosen due to its microbial communities being vulnerable to
invasions and exhibiting functional limitations, allowing the discernment of impacts
of introduced microorganisms on broad proxies such as CO2 production
(Monteux et al., 2020). Approximately 35 g (fresh weight) of homogenized
sediment was set in 200 mL glass jars, sealed with parafilm to allow for gas
but not moisture or microorganism exchange, and pre-incubated at
10 ∘C for 11 d before inoculation.
To manipulate the Collembola microbiome and make it more similar to that
found in natural settings, we collected a topsoil (0–15 cm depth) from a
subarctic meadow (Kärkevagge, 30 km west of Abisko, northern Sweden;
68∘24′23.8′′ N, 18∘18′51.6′′ E) in September 2019 and kept
it frozen until the cultivation of Collembola. We could not obtain topsoil
from the same location as the Yedoma permafrost due to practical constraints
during the Covid-19 pandemic.
Collembola
A strain of the Collembola species Folsomia candida was obtained from Vrije Universiteit
Amsterdam and cultured 6 months prior to the onset of the experiment.
Folsomia candida is a parthenogenetic ground-dwelling Collembola, which has been routinely
used as a model organism in soil ecology. Stock cultures were maintained on
a gypsum and coal medium and fed baker's yeast; traces of mould were removed; fresh yeast and water were added once to twice a week, and fresh stock
cultures were started monthly.
Collembola inoculation treatments
Two months prior to the onset of the experiment, separate stock cultures
were established on gypsum and coal medium supplemented with a 2–3 cm layer
of topsoil to obtain Collembola whose skin and gut microbiome were both
colonized with topsoil microorganisms (“topsoil stock culture”). In
parallel, similar stock cultures using permafrost sediment were established
as an additional control (“permafrost stock culture”). These stock cultures
with soil or sediment were supplemented with yeast and water like the
gypsum stock culture to maintain high adult population densities prior to
the experiment. One day before inoculating the incubation jars, soil
suspensions were made from the topsoil and permafrost sediment (5 and 10 g,
respectively, in 100 mL ddH2O), shaken at 150 rpm for 1 h and
filtered (Ahlstrom-Munksjö grade 006, 1.5 µm pore size) to
manipulate the skin microbiome of Collembola in the ectozoochory treatments.
All incubation jars containing permafrost were randomly assigned a treatment
on the day of inoculation from among the following, replicating each treatment
across six jars (Fig. 1). The treatments consisted in adding to the jars
with permafrost
no Collembola (hereafter, “no-Collembola control”),
Collembola grown on gypsum stock culture (“Collembola”),
Collembola grown on gypsum stock culture and sprayed with topsoil suspension (“ectozoochory”),
Collembola grown on gypsum stock culture and sprayed with permafrost
suspension (“ectozoochory control”),
Collembola grown on topsoil stock culture (“endozoochory”),
Collembola grown on permafrost stock culture (“endozoochory control”).
To obtain a baseline allowing the quantification of the effect of Collembola on
permafrost CO2 production and bacterial community composition, Yedoma
permafrost was incubated in the absence of Collembola (no-Collembola control).
For all other treatments, the permafrost was supplemented with adult
Collembola isolated from the stock cultures using a handheld vacuum
cleaner mounted with a 10 mL pipet tip. The Collembola were transferred into
a black plastic tray, allowing them to be spread and the adults to be picked out,
excluding juveniles as much as possible. From this tray, a similar number of
Collembola (30–80 individuals, in the range of values used in the
literature, e.g. by Kaneda and Kaneko, 2008) was sampled with the vacuum
cleaner and then inoculated into each jar by pouring them into a plastic
funnel, with the Collembola provenance or manipulation depending on the
treatment. Prior to transferring them into incubation jars, Collembola used
in the ectozoochory treatment and its control were sprayed with the
corresponding soil suspensions. To limit cross-contamination, separate
funnels, as well as separate 10 mL pipet tips on the vacuum cleaner, were
used for the different treatments. A picture of the inside of each jar was
taken to count the exact number of Collembola, then the jars were closed
with rubber septa and flushed with moisturized CO2-free air and incubated.
To estimate Collembola respiration per individual, Collembola were also
incubated in six jars under identical conditions, except that the soil was
replaced with a few drops of autoclaved ddH2O to prevent dehydration
(hereafter, no-soil calibration).
Summary of the experimental design. Folsomia candida Collembola were
grown in a stock culture on gypsum and fed with yeast and then subjected to
different treatments before being inoculated into jars filled with Yedoma
permafrost sediment (30–80 individuals per jar). Ectozoochory was assessed
by spraying the Collembola with soil suspensions, while endozoochory was
assessed by growing Collembola in stock cultures supplemented with soil. The
jars were incubated in the dark for 25 d at 10 ∘C; CO2
production was measured throughout the incubation period, and bacterial
community composition was determined at the end of the incubation.
Incubation
We dark-incubated all flasks for 25 d under aerobic conditions at
10 ∘C. This incubation temperature is similar to summer
active-layer temperatures in permafrost-affected areas and within the
thermal tolerance range of psychrophilic microorganisms (D'Amico et al.,
2006). We used a short (25 d) incubation period to ensure a relatively
stable Collembola population level, by limiting uncertain numbers of
newly hatched Collembola individuals, since the eggs of Folsomia candida take 18–20 d to
hatch at 16 ∘C (Marshall and Kevan, 1962) and presumably longer
than 25 d at 10 ∘C.
MeasurementsCO<inline-formula><mml:math id="M31" display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula> production
Headspace air was sampled with a syringe to measure CO2 concentrations
(EGM-5 IRGA, PP Systems, Amesbury, Massachusetts, USA) at intervals ensuring
CO2 concentrations remained below 20 000 ppm to prevent a toxic
CO2 build-up (i.e. after 3, 7, 14 and 25 d). After each measurement,
the jars were flushed with 0.45 µm filtered CO2-free air
moisturized by bubbling it through two 5 L bottles of ddH2O, for 3 min
at 1 to 2 L min-1, i.e. with at least 15 times the volume of the jar.
CO2 concentrations were adjusted for changes in temperature and
atmospheric pressure to calculate CO2 production rates (τ) as
follows:
τi=CO2i×PiV/RTiΔti,
where (Δt)i is the time interval between measurement
(subscript i) and previous flushing, Pi atmospheric pressure at
measurement time, V the headspace volume, R the ideal gas constant, and
Ti the temperature. To calculate cumulative CO2 production over
the entire incubation, we summed up the quantity of CO2 present in the
headspace at each sampling, within each jar.
Bacterial community
Microcentrifuge tubes (1.5 mL) were filled with soil and snap-frozen in dry
ice to analyse microbial communities from the jars harvested at the end of the incubation. The frozen
tubes were kept at -20∘C for up to 4 months before
freeze-drying and then homogenized by bead beating (Precellys CK68 15 mL
tubes, 2×30 s at 4500 rpm). DNA was extracted from 183 to 285 mg
of homogenized freeze-dried soil using the DNeasy PowerSoil Pro Kit (Qiagen)
according to the manufacturer's instructions, and DNA concentrations in the
extracts were measured on a Qubit 1.0 fluorometer.
The V4–V5 region of the 16S ribosomal RNA gene was targeted in PCR
amplification using primers 515F (5′-GTGYCAGCMGCCGCGGTAA-3′) and 926R (5′-CCGYCAATTYMTTTRAGTTT-3′) with Illumina sequencing adapters, using
12.5 µL Phusion Taq Green PCR Master Mix (Thermo Scientific),
0.25 µM of each primer, 2 µL of DNA extract diluted to
5 ng µL-1 and nuclease-free water in 25 µL reaction
volume. PCR conditions were as follows: initial denaturation (98 ∘C, 3 min), 25 cycles of denaturation (98 ∘C, 15 s), annealing
(50 ∘C, 30 s) and elongation (72 ∘C, 40 s), and a
final elongation (72 ∘C, 10 min), after which PCR products were
checked by electrophoresis on 1 % agarose SB gel. A total of 20 µL of each PCR
product was cleaned, and the products' DNA concentrations were normalized using a
SequalPrep Normalization Plate Kit (Invitrogen), according to manufacturer's
instructions. Three DNA extraction blanks and two PCR blanks were included
as negative controls, as well as a mock community as the positive control
(ZymoBIOMICS, diluted to 5 and 0.5 ng µL-1 in two replicates
each).
A second PCR step was performed to add Nextera dual-indexing barcodes, using
30 µL reaction volume, 1 µM of each primer and 5 µL of cleaned PCR product. PCR conditions were as follows: initial denaturation
(98 ∘C, 3 min), eight cycles of denaturation (98 ∘C,
30 s), annealing (55 ∘C, 30 s) and elongation (72 ∘C,
40 s), and a final elongation (72 ∘C, 10 min), after which PCR
products were checked by electrophoresis on 1 % agarose SB gel. A total of 25 µL of each PCR product was cleaned, and the products' DNA concentrations were normalized using
a SequalPrep Normalization Plate Kit (Invitrogen). Serial elution across
columns was used to increase concentration of the pooled products, i.e.
using only 8×20µL elution buffer instead of 96×20µL. The eluted DNA was pooled; its concentration was measured on a
Qubit fluorometer, and the size distribution of the amplicons was measured by
automated electrophoresis (Agilent 2100 Bioanalyzer). The library was then
sent for sequencing on an Illumina MiSeq with V3 chemistry (2×300 bp, 15 % PhiX spike-in) at the SNP&SEQ Technology Platform in
Uppsala. Demultiplexing was performed by the sequencing facility, and data were
deposited at ENA with accession number PRJEB51992.
Data analysisBioinformatics
All bioinformatics and statistics were performed in R v4.1.3 (R Core Team,
2022), unless specified otherwise. The whole analysis pipeline is found at the Bolin Centre Code Repository (Monteux, 2022) and the processed data and figure-generating script at Zenodo (Monteux et al., 2022). In short, amplicon sequence variants (ASVs) were created with
DADA2 1.18.0 (pseudo-pooling; Callahan et al., 2016) after removing primers
and adapters with Cutadapt (v3.10; Martin, 2011). Taxonomy was assigned to
ASVs with the RDP naïve Bayesian classifier (v1.8; Wang et al., 2007),
and ASVs resolved to the genus rank were further assigned a species rank by
the exact string-matching algorithm implemented in DADA2 (assignSpecies), using SILVA
v138.1 reference data (Quast et al., 2013). Putative contaminant ASVs were
manually selected from those identified in silico using the decontam algorithm (Davis et al.,
2018) with combined prevalence- and frequency-based methods using the
default threshold of 0.1 and separate prevalence- and frequency-based
methods with a threshold of 0.05. Eight contaminant ASVs amounting to up to
0.024 % of the total reads were removed. Appropriateness of the
bioinformatics analysis parameters was judged by visually assessing the
composition of mock communities (microbial community DNA standard,
ZymoBIOMICS) at the genus level, leading to ASVs amounting to up to fewer than
10 reads and/or present in fewer than 3 samples being removed from the
dataset. Sequencing depth was not associated with experimental treatments
(ANOVA F5,30= 0.701, P= 0.627); therefore read numbers were converted
to proportional abundances within samples to normalize sample sizes.
Diversity analyses
The effect of the different treatments on bacterial communities was
visualized using principal coordinates analysis (PCoA) and tested with
permutational multivariate analyses of variance (PERMANOVAs, adonis function in the
vegan package; Oksanen et al., 2022) after verifying homoscedasticity (betadisper). Two distance matrixes
were computed for that purpose, using weighted and non-weighted UniFrac
distances (Lozupone et al., 2011) to distinguish between compositional
effects accounting for bacterial relative abundances and for only
presence–absence, respectively. Pairwise contrasts were subsequently
computed using the wrapper provided in the pairwiseAdonis R package (Arbizu,
2022).
In addition, the same effect was tested with the more robust and sensitive
manyglm approach (Wang et al., 2012), which avoids certain pitfalls of
distance-based methods (Warton et al., 2012). The manyglm models were fitted with
negative binomial distribution, after visually checking that assumptions
were met, on the non-normalized ASV count data using the default PIT-trap
resampling with 1999 bootstrap permutations and a likelihood-ratio testing
method. An analysis of deviance was carried out (anova.manyglm), subsequently using the
provided wrapper for pairwise comparisons with free step-down P-value
adjustment.
CO<inline-formula><mml:math id="M89" display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula> production analysis
A repeated-measures ANOVA was carried out, using Greenhouse–Geisser ϵ
correction to degrees of freedom to account for the violation of the
sphericity assumption and thus to assess the interactive effect of the Collembola
manipulation treatments over time. Since this interactive effect was not
statistically significant (P= 0.56, Table A1), we removed the time dimension
and used cumulative CO2 production at the end of the incubation period
in further analyses.
To assess our hypothesis that Collembola presence would increase CO2
production we used two-sample t tests with unequal variances to compare the
cumulative CO2 production at the end of the incubation between the
no-Collembola control and all other jars.
To explore differences between the Collembola microbiome manipulations, we
used a one-way ANOVA followed by treatment contrasts to assess the
difference from the control (i.e. all treatment and controls compared to the
no-Collembola control) and selected orthogonal contrasts to assess the
effects of each treatment (i.e. each treatment compared to its own control) using the “emmeans” package (Lenth, 2016).
Estimated Collembola respiration
Based on photographs taken upon inoculating the incubation jars with
Collembola and 3 d after inoculating, we counted the exact number of
live Collembola (i.e. ignoring Collembola which had apparently not moved
between when the two pictures were taken). Only a few animals did not survive the
transfer, and no dead animals were observed at the end of the incubation.
The number of Collembola per jar varied across treatments (F4,25= 3.19, P= 0.030), although no pairs of treated jars significantly differed
from each other (when including the no-soil calibration jars, F5,30= 4.048, P= 0.006; Fig. B1); we therefore needed to account for
differing Collembola numbers. We averaged the respiration per individual in
the no-soil calibration set for each of the CO2 concentration
measurement times and multiplied this amount by the number of Collembola
present in each jar to estimate the amount of CO2 produced by
Collembola basal respiration in treatment jars.
Response-ratio calculations
A response ratio of CO2 production in Collembola treatments was
calculated by dividing cumulative CO2 production by the average of the
no-Collembola control. This was performed both for the net CO2 production
after subtracting the estimated Collembola basal respiration (RRsoil)
and for the gross CO2 production (RRgross). This allows the
partition of the difference in altered CO2 production between what was
respired by the Collembola and a putative priming effect on soil CO2
production.
ResultsBacterial communities
Bacterial community composition was overall largely unaffected by the
presence of Collembola. Taking relative abundances into account, no effect
of the different treatments was identified with PERMANOVA (P> 0.05, Fig. 2a), but a significant treatment effect was observed using the
presence–absence distance metrics (PERMANOVA F5,30= 2.91, P= 0.001,
Fig. 2b). More specifically, treatments containing Collembola differed
from the no-Collembola control (Fig. 2b) and to some extent from each
other (Table C1). Using manyglm, a significant treatment effect was observed
(manyglm analysis of deviance P< 0.001; Table 1). The ectozoochory
treatment and its control (Collembola sprayed with topsoil or permafrost
soil suspension, respectively) differed significantly from the no-Collembola
control in the post hoc test (0.01 <P< 0.05), while the
other treatments including Collembola tended to differ from the
no-Collembola control (0.05 <P< 0.1, Table 1). Similarly to
the presence–absence analysis, treatments that included Collembola did not
significantly differ from each other in the manyglm analysis. Overall, soil
with Collembola exhibited higher alpha-diversity metrics (richness
estimators Chao1, abundance-based coverage estimator – ACE, number of observed ASVs, as well as Shannon and
Fisher diversity indexes; Fig. 3, Table C2) than the no-Collembola control, but those did not differ among the Collembola treatments.
Response of permafrost bacterial communities to the
addition of Collembola undergoing different microbiome manipulation treatments.
(a) Principal coordinates analysis (PCoA) of normalized abundance of
bacterial ASVs (amplicon sequence variants) based on weighted UniFrac
phylogenetic distance (i.e. accounting for relative abundances); (b) PCoA of normalized abundance of bacterial ASVs based on non-weighted UniFrac
phylogenetic distance (i.e. presence–absence only). Ellipses indicate the 80 %
confidence interval around treatment centroids, and axes are scaled to the
percentage of explained variance. The asterisk indicates that the
no-Collembola control (dark brown) in (b) significantly differs from all other
treatments (pairwise PERMANOVA P< 0.05) and endo- and ectozoochory treatments did not differ from the Collembola treatment. Pairwise comparisons are
summarized in Table C1.
Response of permafrost bacterial alpha diversity to the
addition of Collembola undergoing different microbiome manipulation treatments.
(a) Number of observed ASVs (amplicon sequence variants) per sample;
(b) abundance-based coverage estimator (ACE). Coloured symbols and
error bars are means and standard errors within a treatment (n=6); small
grey symbols are individual samples. Error bars are only shown when
exceeding the symbols' size. Asterisks denote that the no-Collembola control
differs from all other treatments (estimated marginal means pairwise
comparison, 0.001 < Holm-adjusted P< 0.01).
CO<inline-formula><mml:math id="M128" display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula> production
Overall, the Collembola addition resulted in higher CO2 production than
the no-Collembola control, but the ectozoochory and endozoochory treatments
did not result in higher CO2 production than their respective controls
(using gross CO2 production, as in RRgross; Table A2, Fig. A1).
When accounting for Collembola basal respiration (as in RRsoil), only
the ectozoochory treatment (P= 0.005) and to a lesser extent its control
(P= 0.069) differed from the no-Collembola control (Table A2, Fig. A1).
We hypothesized increased CO2 production in the zoochory treatments
through an effect on bacterial communities; however that effect was absent.
We therefore tested for the overall effect of Collembola presence across all
treatments compared to the no-Collembola control (Fig. 4). Using a t test
with unequal variances, we observed a 25.85 % increase in CO2
production in the presence of Collembola (RRgross; Fig. 4a). When
subtracting estimated Collembola basal respiration, this increase was on
average 13.22 % (RRsoil; Fig. 4b); thus roughly half of the
observed increase in CO2 production could be attributed to Collembola
basal respiration.
Response ratios of permafrost cumulative CO2 production
rates to addition of Collembola. (a) RRgross: measured
CO2 production rates. (b) RRsoil: measured CO2
production rates with estimated Collembola basal respiration subtracted. All
values are divided by the mean of the no-Collembola control; small grey
circles are individual values; large symbols are means; error bars are
standard errors of the mean (n= 6 and 30 for brown and black symbols,
respectively); violin plots denote the shape of the distributions with
the maximum width proportional to the number of observations. Statistics are
derived from two-sample t tests with unequal variances.
Discussion
We showed that Collembola stimulated CO2 production from permafrost
sediment through “priming” while subtly altering the presence of certain
bacterial ASVs, irrespectively of the microbiome manipulation imposed on the
Collembola. Colonization of permafrost by Collembola therefore has the
potential to result in bacterial community changes and to increase C
turnover from tundra soils.
Introduction of bacteria by Collembola
Collembola introduced new bacteria into thawed permafrost soil, resulting in
an increase in alpha diversity and significant differences in bacterial
community composition, on a presence–absence basis. This pattern was
consistent in all treatments where Collembola were added but did not depend
on the microbiome manipulation treatment assigned to the Collembola (Figs. 2b, 3). Despite their effects on the presence of certain bacteria,
Collembola did not impose broader community changes as bacterial
communities were not affected when considering relative abundances as we had
hypothesized (Fig. 2a, Table 1). However, given the subtle effect of
Collembola on bacteria during this short incubation, this suggests that on
longer timescales Collembola may modify bacterial communities more
dramatically.
Differences between ecto- and endozoochory treatments
Although Collembola introduced new bacteria into permafrost soil, our
results did not support that they did so differently through ecto- and/or
endozoochory. This suggests that Collembola presence mostly introduced
bacteria that were part of the core Collembola microbiome or that the way
in which their microbiome changed was not well represented by our
treatments. For example, spraying may not have altered the cuticle
microbiome as Collembola cuticle is particularly hydrophobic (Hensel et al.,
2013a, b; Nickerl et al., 2013). This should however not have been an issue
for the endozoochory treatment, in which Collembola were in contact with the
topsoil for a prolonged (> 1 month) period of time. Despite this
prolonged exposure, no additional effect of the endozoochory treatment on bacterial
community composition was observed, compared to ectozoochory or
Collembola-only treatments. Collembola typically harbour a gut microbiome
reflecting their diet and direct environment (Xiang et al., 2019; Leo et
al., 2021), and observing soil-coloured faeces in the topsoil and permafrost
stock cultures suggests that soil bacteria were present in the gut of
Collembola used in the endozoochory treatments. In contrast with the
permafrost stock cultures, where the yeast was necessary as an additional
food source to maintain stable populations, Collembola populations remained
stable in topsoil stock cultures similar to those used in the endozoochory
treatment even in the absence of yeast, further indicating their consumption of
soil particles and associated microorganisms. Therefore, the absence of
the effect of the ectozoochory treatment on permafrost bacterial communities may
indicate that Collembola cuticle does not serve as a vector for bacterial
dispersal, possibly due to its omniphobic structure. Further, the lack of
difference between the effects of Collembola and endozoochory treatments
suggests that few to none of the bacteria present in this topsoil are able
to establish in – or survive transit through – collembolan guts. This suggests
against generalist “hitch-hiking” bacteria using microfauna guts as a means
of dispersal, emphasizing instead that such hitch-hiking bacteria are
similar regardless of the feeding context of the Collembola.
It is not unlikely that the observed limited response of the bacterial community
to zoochory treatments was transient and that Collembola presence would
have eventually affected community composition also in terms of relative
abundances. A longer incubation period may have resulted in stronger effects
on bacterial community composition, as observed by Coulibaly et al. (2019),
but this would have been at the expense of controlling the number of
Collembola in the jars and thus of being able to account for their basal
respiration. Using an RNA-based approach to target the “active” bacterial
community may be another way to detect such transient effects. It should
also be noted that our semi-quantitative approach to bacterial community
composition did not allow us to assess putative changes in absolute
abundances, although such changes are not necessarily observed in similar
studies (Kaneda and Kaneko, 2008).
Collembola effects on fungal communities may have indirectly affected
bacterial communities and thus represent a potential additional mechanism
by which Collembola have affected the presence of certain bacteria. It is
not unlikely that fungi would have responded more strongly to the presence
of mostly fungivorous Folsomia candida. Using fungi as the focus organism in this study,
however, may have had its difficulties. For example, the use of Candida albicans as a food
source for the Collembola stock cultures may have strongly distorted fungal
community composition, thus requiring questionable bioinformatic
workarounds. Further, assessing the effects on fungal communities may not
allow for distinguishing between the consequences of Collembola as a vector for
microbial migration and as a grazing consumer. Although assessing responses
of the whole soil food web may have given better insights into potential
mechanisms by which Collembola affect thawed permafrost soil, we here chose
to focus on their effects on bacteria given their preponderant role on
realizing soil functions in this system (Monteux et al., 2020).
Collembola effects on CO<inline-formula><mml:math id="M151" display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula> production
Our hypothesis stating that soil CO2 production would be increased in the
presence of Collembola was supported by our results. However, since no
differences in bacterial community composition were detected between the
different Collembola treatments, the rationale for our follow-up hypothesis
(attributing an increased influence on CO2 production to the treatments
exhibiting stronger changes in bacterial community) was not met. We
therefore did not deem it relevant to formally evaluate this hypothesis. Our
results suggest that it would be rejected (Fig. A1), but we could not say
whether that is because the rationale is wrong or because the manipulation
of bacterial communities did not yield the expected effect.
Overall, Collembola presence increased respiration by 25 %, half of which
was attributed to respiration of Collembola themselves. We interpret the
remaining half (13.22 %) as a priming effect, i.e. a stimulation of
soil-organic-matter-derived CO2 production induced by the presence of Collembola. The
mechanisms behind this priming effect remain unclear and could for instance
relate to stimulation of microbial activity by the input of more labile
substrates by the Collembola (digestion by-products). Overall, our results
add to a list of contrasting findings on the direction and magnitude of
priming of soil-organic-matter decomposition by Collembola presence in soils in general.
Earlier studies that compared CO2 production in the presence or absence of
Collembola or other faunal groups have mostly focused on litter (e.g. Hanlon
and Anderson, 1979). The scarce studies on soil CO2 production
including Collembola reported contrasting findings, with some studies
observing increases in soil CO2 emissions in the presence of Collembola by
up to +400 % (Addison and Parkinson, 1978; Bakonyi, 1989; Kaneda and
Kaneko, 2008; Wang et al., 2017), while others found no significant changes
(Theenhaus et al., 1999; Lubbers et al., 2020; Lucas et al., 2020). The
Collembola effect on CO2 production in litter can be species-specific
(Hanlon and Anderson, 1979), and our results could support this as we found
a similar increase to those in three studies where Folsomia candida was used (Bakonyi, 1989;
Kaneda and Kaneko, 2008; Wang et al., 2017). Kaneda and Kaneko (2008)
reported a density-dependent increase in CO2 production with the
addition of F. candida, which showed no effect at the density used in our study but an
increase with 5–10 times higher density (i.e. 400 individuals in 30 g of soil
compared to our 30–80). Accordingly, we did not find any trend supporting a
density-dependent response in our data. Nevertheless, we suggest that
permafrost soils colonized by Collembola in particular, as well as by soil fauna in
general, will likely see an increase in their rates of decomposition, in
CO2 production in particular, due to both priming and the basal
respiration of the newly established food web.
Our experimental setup did not allow us to distinguish further between soil
and Collembola respiration. We estimated basal respiration, though this may
not perfectly reflect reality, considering that Collembola in jars without
soil may not have behaved similarly to those in jars with soil. Faced with a
scarcity of food sources, they may have slowed their metabolism to endure
that stress, which could result in decreased basal respiration and thus lead
us to overestimate the magnitude of priming. We could not find reference
values in the literature for Folsomia candida respiration rates, but Addison and Parkinson (1978) estimated the CO2 production of two high-Arctic collembolan
species to 182.6 and 250.6 µL CO2 per gram live weight per hour at
10 ∘C, for Hypogastrura tullbergi and Folsomia regularis, respectively. Assuming a CO2 density of
1.977 kg m-3 and 170 µg per adult individual of Folsomia candida, our values range
between 47 and 220 µL CO2 per gram live weight per hour. Addison and
Parkinson (1978) refer to Mitchell (1973) for a description of the method
used for estimating respiration rates, where the measurements appear to be
carried out over a period of 20 h. When considering only the first
measurement date in our data (3 d after inoculation), the rates are
closer to the values reported by Addison and Parkinson (1978) with 157.5±18.5µL CO2 per gram live weight per hour (mean ± SE,
n= 6). In future studies of this mechanism, it would be interesting to use
stable isotope methods to partition the Collembola-derived vs. soil-organic-matter-derived CO2 production more accurately.
Beyond the CO2 production increase due to priming by soil fauna, our
Collembola addition treatments may have induced “microbial community
priming”, in the form of the release of microbial functional limitations.
The Yedoma sediment used in this study indeed lacks certain microbial taxa
and functions, and the onset of an exotic microbial community can result in
large increases in CO2 production (Monteux et al., 2020). Considering
the limited to non-significant (depending on the distance metrics) effect of
Collembola addition on bacterial communities, we do not think this mechanism
explains our observed results, although we do not rule out this explanation.
Conclusions
Collembola presence modified bacterial communities in newly thawed
permafrost and primed its CO2 production, highlighting the importance
of non-microbial decomposers for the fate of soil organic matter in thawing
permafrost. An emerging theme in permafrost research is the missing
functions in permafrost microbial communities due to a lack of certain
microbial groups, hampering the production of CO2, CH4 and
N2O (Knoblauch et al., 2018; Monteux et al., 2020; Marushchak et al.,
2021). It is also becoming clearer that the functionality of permafrost
microbes can vary across space (Barbato et al., 2022); therefore the
modalities of microbial dispersal into newly thawed permafrost likely affect
the fate of the permafrost's organic matter and the rate of release of greenhouse gases.
Our findings suggest that Collembola may accelerate the release of
greenhouse gases, possibly in part through the introduction of
microorganisms, although those did not seem to depend strongly on the
preceding Collembola environment. In nature, thawing permafrost mostly
occurs at the bottom of the seasonally thawing active layer, often below the
water table and at depths far from the topsoil where most Collembola reside.
Different soil fauna, such as earthworms, may access such newly thawed
permafrost, but their current geographical range does not overlap much with
the permafrost region. However, rapid thaw events such as hillslope
thermokarst, active-layer detachments or retrogressive thaw slumps expose
newly thawed permafrost to surface conditions, thus making it susceptible to
colonization by Collembola. Fauna-induced dispersal of microorganisms is
most likely to take place in these environments and should be further
investigated in realistic field conditions.
It remains unclear whether the increase in CO2 production we attribute
to priming stems from increased functionality of the microbial community due
to bridging of functional limitations or from other mechanisms. Mechanical
breakdown of organic material and digestion processes might provide
microorganisms with more easily available substrates, while disruption of
the soil pore structure may also result in increased microbial decomposition
activity. Several studies indicate increases in CO2 production with the
presence of soil fauna; however to our best knowledge no studies have
specifically partitioned to which extent such increases were due to faunal
respiration or to priming effects. We argue that elucidating this question
would be an important next step towards opening the “black box” that soil
systems still often represent, thus helping to mechanistically address the
effects of global changes.
Response ratios of permafrost cumulative CO2 production
rates to addition of Collembola. (a) RRgross: measured CO2 production rates. (b) RRsoil: measured CO2 production rates with estimated Collembola
basal respiration subtracted. All values are divided by the mean of the
no-Collembola control; small grey circles are individual jar values; large
symbols are means; error bars are standard errors of the mean (n= 6). Black
symbols denote significant difference from the control; different symbols
denote different statistical significance (estimated marginal means treatment contrasts with
Holm adjustment for multiple comparisons, Appendix Table A2; dot (⋅) denotes 0.05 <P< 0.1; * 0.01 <P< 0.05; ** 0.001 <P< 0.01; ***P< 0.001).
Effect of Collembola addition on daily CO2 production over
time, repeated-measures ANOVA.
P[GG] indicates repeated-measures ANOVA P using Greenhouse–Geisser correction
to degrees of freedom to account for violation of the assumption of
sphericity (Mauchly's test). P values below 0.05 are denoted in bold; ***P< 0.001; * 0.01 <P< 0.05. Num. DOFs: numerator degrees of freedom; Den. DOFs: denominator degrees of freedom; GES: generalized Eta-squared measure of effect size; GG ϵ: Greenhouse–Geisser ϵ.
Effects of Collembola additions on response ratios of
cumulative CO2 production at the end of the incubation, excluding
(RRsoil) or including (RRgross) estimated Collembola respiration.
Holm adjustment for P values of non-orthogonal contrasts. P values below 0.05
are denoted in bold; ***P< 0.001; ** 0.001 <P< 0.01; * 0.01 <P< 0.05; dot (⋅) denotes 0.05 <P< 0.1. DOFs: degrees of freedom; Res. DOFs: residual degrees of freedom.
RRsoilANOVADOFsRes. DOFsSum sqMean sqF valuePTreatment5300.0030.0013.7280.010*Contrast vs. no-Collembola control EstimateSEDOFst ratioAdjusted PCollembola 0.0120.008301.5730.379Ectozoochory control 0.0200.008302.5250.068.Ectozoochory 0.0280.008303.6260.005**Endozoochory control 0.0030.008300.4430.764Endozoochory 0.0070.008300.8870.764Orthogonal contrasts EstimateSEt valuePCollembola vs. control 0.0060.0041.5730.126Ectozoochory vs. ectozoochory control 0.0020.0040.4440.660Endozoochory vs. endozoochory control 0.0040.0041.1010.280RRgrossANOVADOFsRes. DOFsSum sqMean sqF valuePTreatment5300.0050.0014.1840.005**Contrast vs. no-Collembola control EstimateSEDOFst ratioAdjusted PCollembola 0.0240.009302.7540.030*Ectozoochory control 0.0300.009303.3620.009**Ectozoochory 0.0380.009304.3310.001***Endozoochory control 0.0200.009302.2700.031*Endozoochory 0.0230.009302.6290.030*Orthogonal contrasts EstimateSEt valuePCollembola vs. control 0.0120.0042.7540.010**Ectozoochory vs. ectozoochory control 0.0020.0040.3590.722Endozoochory vs. endozoochory control 0.0040.0040.9690.340
Number of adult Collembola individuals per incubation
jar. Small grey circles are individual jar values; large symbols are means;
error bars are standard errors of the mean (n= 6); violin plots denote
the shape of the distributions with maximum width proportional to the number
of observations. Different black letters denote statistically significant
pairwise differences (estimated marginal means pairwise contrasts with Holm adjustment for
multiple comparisons, 0.05 <P< 0.1).
Effects of Collembola additions on bacterial community
composition (permutational multivariate ANOVAs and pairwise contrasts),
using distance matrixes based on relative abundances (weighted UniFrac) or
presence–absence (non-weighted UniFrac). Holm adjustment for P values of
pairwise contrasts. P values below 0.05 are denoted in bold; ***P< 0.001; ** 0.001 <P< 0.01; * 0.01 <P< 0.05; dot (⋅)
denotes 0.05 <P< 0.1.
PERMANOVADOFsRes. DOFsF modelR2PWeighted UniFrac (relative abundance)5301.2610.1740.101 Non-weighted UniFrac (presence–absence)5303.0180.335<0.001Pairwise contrastDOFF modelR2Adjusted PControl vs. Collembola13.2360.2440.025*Control vs. endozoochory control14.3560.3030.025*Control vs. ectozoochory16.7770.4040.022*Control vs. ectozoochory control15.6580.3610.025*Control vs. endozoochory14.7770.3230.012*Collembola vs. endozoochory control11.7670.1500.102Collembola vs. ectozoochory11.7510.1490.115Collembola vs. ectozoochory control11.3480.1190.148Collembola vs. endozoochory11.4340.1250.148Endozoochory control vs. ectozoochory13.0680.2350.025*Endozoochory control vs. ectozoochory control12.7110.2130.025*Endozoochory control vs. endozoochory12.1560.1770.025*Ectozoochory vs. ectozoochory control11.4970.1300.113Ectozoochory vs. endozoochory11.9580.1640.025*Ectozoochory control vs. endozoochory11.8200.1540.025*
Effects of Collembola additions on bacterial alpha
diversity (ANOVAs and pairwise contrasts), with total number of observed
ASVs, abundance-based coverage estimator (ACE), and Shannon's and Fisher's
diversity indexes. Holm adjustment for P values of pairwise contrasts.
P values below 0.05 are denoted in bold; ***P< 0.001; ** 0.001 <P< 0.01; * 0.01 <P< 0.05; dot (⋅) denotes 0.05 <P< 0.1.
All code used to process the raw DNA data and produce the figures and tables
presented in the text is found at 10.57669/monteux-2022-collembola-1.1.1 (Monteux, 2022).
The code used to generate figures from the processed DNA data is also found
at 10.5281/zenodo.7015077 (Monteux et al., 2022).
Data availability
All 16S sequencing data are found at ENA under project accession number
PRJEB51992.
All processed data used to generate the findings presented in the text are found
at 10.5281/zenodo.7015077 (Monteux et al., 2022) and 10.57669/monteux-2022-collembola-1.1.1 (Monteux, 2022).
Sample availability
Due to practical constraints, the exact permafrost material and Collembola
cultures used throughout the paper are no longer available. Permafrost
from the same location may be obtained through contacting the US Army
CRREL, while Folsomia candida strains may be obtained by contacting Janine Mariën. Frozen permafrost
samples from the end of the incubation period, as well as aliquots of DNA
extracts used for this study, may be obtained from the corresponding author
upon reasonable request.
Author contributions
SM, JM and EK designed the experiment. JM provided Folsomia candida strains and guidance for
culturing. EK provided topsoil from Kärkevagge. SM performed the
experiment and data analysis and wrote the manuscript with input from all
co-authors.
Competing interests
The contact author has declared that none of the authors has any competing interests.
Disclaimer
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Special issue statement
This article is part of the special issue “Global change effects on terrestrial biogeochemistry at the plant–soil interface”. It is not associated with a conference.
Acknowledgements
This study was funded by a grant from Formas (Dnr 2017-01182) awarded to
Eveline J. Krab and by Kempestiftelserna awarded to Sylvain Monteux.
We thank the Department of Forest Mycology and Plant Pathology, SLU, for
hosting the molecular work and Thomas H. Douglas from the US Army Cold Regions
Research and Engineering Laboratory's permafrost tunnel (Alaska) for
assistance and permission to sample. We also thank Frida Keuper from the French
National Research Institute for Agriculture, Food and Environment and Ellen Dorrepaal from Umeå University for providing the permafrost sample.
Sequencing was performed by the SNP&SEQ Technology Platform in Uppsala.
The facility is part of the National Genomics Infrastructure (NGI) Sweden
and Science for Life Laboratory. The SNP&SEQ Technology Platform is also supported
by the Swedish Research Council and the Knut and Alice Wallenberg
Foundation.
Financial support
This research has been supported by the Svenska Forskningsrådet Formas (Dnr 2017-01182 awarded to Eveline J. Krab) and Kempestiftelserna (awarded to Sylvain Monteux).
The article processing charges for this open-access publication were covered by Stockholm University.
Review statement
This paper was edited by Emily Solly and reviewed by two anonymous referees.
ReferencesAddison, J. A. and Parkinson, D.: Influence of Collembolan Feeding
Activities on Soil Metabolism at a High Arctic Site, Oikos, 30, 529–538,
10.2307/3543348, 1978.Agamennone, V., Jakupovic, D., Weedon, J. T., Suring, W. J., van Straalen,
N. M., Roelofs, D., and Roling, W. F. M.: The microbiome of Folsomia
candida: an assessment of bacterial diversity in a Wolbachia-containing
animal, FEMS Microbiol. Ecol., 91, fiv128,
10.1093/femsec/fiv128, 2015.Arbizu, P. M.: pairwiseAdonis: Pairwise multilevel comparison using adonis, R package version 0.4, Github [code], https://github.com/pmartinezarbizu/pairwiseAdonis, last access 22 August 2022.Bakonyi, G.: Effects of Folsomia candida (Collembola) on the microbial
biomass in a grassland soil, Biol. Fert. Soils, 7, 138–141,
10.1007/BF00292572, 1989.Barbato, R. A., Jones, R. M., Douglas, T. A., Doherty, S. J., Messan, K.,
Foley, K. L., Perkins, E. J., Thurston, A. K., and Garcia-Reyero, N.: Not
all permafrost microbiomes are created equal: Influence of permafrost thaw
on the soil microbiome in a laboratory incubation study, Soil Biol.
Biochem., 108605, 10.1016/j.soilbio.2022.108605, 2022.
Berman, D. I., Meshcheryakova, E. N., and Mikhaljova, E. V.: Cold Hardiness
and Range of the Myriapod Angarozonium Amurense (Polyzoniidae, Diplopoda,
Arthropoda) in Permafrost Environments, Cryoletters, 36, 237–242, 2015.Blume-Werry, G., Krab, E. J., Olofsson, J., Sundqvist, M. K., Väisänen, M., and Klaminder, J.: Invasive earthworms unlock arctic plant nitrogen limitation, Nat. Commun., 11, 1766, 10.1038/s41467-020-15568-3, 2020.Bottos, E. M., Kennedy, D. W., Romero, E. B., Fansler, S. J., Brown, J. M.,
Bramer, L. M., Chu, R. K., Tfaily, M. M., Jansson, J. K., and Stegen, J. C.:
Dispersal limitation and thermodynamic constraints govern spatial structure
of permafrost microbial communities, FEMS Microbiol. Ecol., 94, fiy110,
10.1093/femsec/fiy110, 2018.Buse, T., Ruess, L., and Filser, J.: Collembola gut passage shapes microbial
communities in faecal pellets but not viability of dietary algal cells,
Chemoecology, 24, 79–84, 10.1007/s00049-013-0145-y, 2014.Calderón, K., Spor, A., Breuil, M.-C., Bru, D., Bizouard, F., Violle,
C., Barnard, R. L., and Philippot, L.: Effectiveness of ecological rescue
for altered soil microbial communities and functions, ISME J., 11,
272–283, 10.1038/ismej.2016.86, 2017.Callahan, B. J., McMurdie, P. J., Rosen, M. J., Han, A. W., Johnson, A. J.
A., and Holmes, S. P.: DADA2: High-resolution sample inference from Illumina
amplicon data, Nat. Methods, 13, 581–583,
10.1038/nmeth.3869, 2016.Coulibaly, S. F. M., Winck, B. R., Akpa-Vinceslas, M., Mignot, L., Legras,
M., Forey, E., and Chauvat, M.: Functional Assemblages of Collembola
Determine Soil Microbial Communities and Associated Functions, Front.
Environ. Sci., 7, 52, 10.3389/fenvs.2019.00052, 2019.Crowther, T. W., Boddy, L., and Hefin Jones, T.: Functional and ecological consequences of saprotrophic fungus–grazer interactions, ISME J., 6, 1992–2001, 10.1038/ismej.2012.53, 2012.D'Amico, S., Collins, T., Marx, J.-C., Feller, G., Gerday, C., and Gerday,
C.: Psychrophilic microorganisms: challenges for life, EMBO Rep., 7,
385–389, 10.1038/sj.embor.7400662, 2006.Davis, N. M., Proctor, D. M., Holmes, S. P., Relman, D. A., and Callahan, B.
J.: Simple statistical identification and removal of contaminant sequences
in marker-gene and metagenomics data, Microbiome, 6, 226,
10.1186/s40168-018-0605-2, 2018.Frouz, J., Novotná, K., Čermáková, L., and Pivokonský,
M.: Soil fauna reduce soil respiration by supporting N leaching from litter,
Appl. Soil Ecol., 153, 103585,
10.1016/j.apsoil.2020.103585, 2020.García-Palacios, P., Maestre, F. T., Kattge, J., and Wall, D. H.:
Climate and litter quality differently modulate the effects of soil fauna on
litter decomposition across biomes, Ecol. Lett., 16, 1045–1053,
10.1111/ele.12137, 2013.
Golovatch, S. and Kime, R.: Millipede (Diplopoda) distributions: a review,
Soil Organisms, 81, 565–597, 2009.Griffiths, H. M., Ashton, L. A., Parr, C. L., and Eggleton, P.: The impact
of invertebrate decomposers on plants and soil, New Phytol., 231,
2142–2149, 10.1111/nph.17553, 2021.Hanlon, R. D. G. and Anderson, J. M.: The effects of collembola grazing on
microbial activity in decomposing leaf litter, Oecologia, 38, 93–99,
10.1007/BF00347827, 1979.Harding, T., Jungblut, A. D., Lovejoy, C., and Vincent, W. F.: Microbes in
High Arctic Snow and Implications for the Cold Biosphere, Appl. Environ.
Microb., 77, 3234–3243, 10.1128/AEM.02611-10, 2011.Hensel, R., Helbig, R., Aland, S., Voigt, A., Neinhuis, C., and Werner, C.:
Tunable nano-replication to explore the omniphobic characteristics of
springtail skin, NPG Asia Mater., 5, e37,
10.1038/am.2012.66, 2013a.Hensel, R., Helbig, R., Aland, S., Braun, H.-G., Voigt, A., Neinhuis, C.,
and Werner, C.: Wetting Resistance at Its Topographical Limit: The Benefit
of Mushroom and Serif T Structures, Langmuir, 29, 1100–1112,
10.1021/la304179b, 2013b.Hultman, J., Waldrop, M. P., Mackelprang, R., David, M. M., McFarland, J.,
Blazewicz, S. J., Harden, J., Turetsky, M. R., McGuire, A. D., Shah, M. B.,
VerBerkmoes, N. C., Lee, L. H., Mavrommatis, K., and Jansson, J. K.:
Multi-omics of permafrost, active layer and thermokarst bog soil
microbiomes, Nature, 521, 208–212, 10.1038/nature14238,
2015.Inglese, C. N., Christiansen, C. T., Lamhonwah, D., Moniz, K., Montross, S.
N., Lamoureux, S., Lafrenière, M., Grogan, P., and Walker, V. K.:
Examination of Soil Microbial Communities After Permafrost Thaw Subsequent
to an Active Layer Detachment in the High Arctic, Arct. Antarct. Alp. Res., 49, 455–472, 10.1657/AAAR0016-066, 2017.Johnston, E. R., Hatt, J. K., He, Z., Wu, L., Guo, X., Luo, Y., Schuur, E.
A. G., Tiedje, J. M., Zhou, J., and Konstantinidis, K. T.: Responses of
tundra soil microbial communities to half a decade of experimental warming
at two critical depths, P. Natl. Acad. Sci. USA, 116, 15096–15105,
10.1073/pnas.1901307116, 2019.Kaneda, S. and Kaneko, N.: Collembolans feeding on soil affect carbon and
nitrogen mineralization by their influence on microbial and nematode
activities, Biol. Fert. Soils, 44, 435–442,
10.1007/s00374-007-0222-x, 2008.Knoblauch, C., Beer, C., Liebner, S., Grigoriev, M. N., and Pfeiffer, E.-M.:
Methane production as key to the greenhouse gas budget of thawing
permafrost, Nat. Clim. Change, 8, 309–312,
10.1038/s41558-018-0095-z, 2018.Koltz, A. M., Classen, A. T., and Wright, J. P.: Warming reverses top-down effects of predators on belowground ecosystem function in Arctic tundra, P. Natl. Acad. Sci. USA, 115, E7541–E7549, 10.1073/pnas.1808754115, 2018.Lenth, R.: Least-Squares Means: The R Package lsmeans, J. Stat. Softw., 69, 1–33, 10.18637/jss.v069.i01, 2016.Leo, C., Nardi, F., Cucini, C., Frati, F., Convey, P., Weedon, J. T.,
Roelofs, D., and Carapelli, A.: Evidence for strong environmental control on
bacterial microbiomes of Antarctic springtails, Sci. Rep.-UK, 11, 2973,
10.1038/s41598-021-82379-x, 2021.Lozupone, C., Lladser, M. E., Knights, D., Stombaugh, J., and Knight, R.:
UniFrac: an effective distance metric for microbial community comparison,
ISME J., 5, 169–172, 10.1038/ismej.2010.133, 2011.Lubbers, I. M., Berg, M. P., De Deyn, G. B., van der Putten, W. H., and van
Groenigen, J. W.: Soil fauna diversity increases CO2 but suppresses N2O
emissions from soil, Glob. Change Biol., 26, 1886–1898,
10.1111/gcb.14860, 2020.Lucas, J. M., McBride, S. G., and Strickland, M. S.: Trophic level mediates
soil microbial community composition and function, Soil Biol.
Biochem., 143, 107756, 10.1016/j.soilbio.2020.107756,
2020.Mackelprang, R., Waldrop, M. P., DeAngelis, K. M., David, M. M., Chavarria,
K. L., Blazewicz, S. J., Rubin, E. M., and Jansson, J. K.: Metagenomic
analysis of a permafrost microbial community reveals a rapid response to
thaw, Nature, 480, 368–371, 10.1038/nature10576, 2011.Mackelprang, R., Burkert, A., Haw, M., Mahendrarajah, T., Conaway, C. H.,
Douglas, T. A., and Waldrop, M. P.: Microbial survival strategies in ancient
permafrost: insights from metagenomics, ISME J., 11, 2305–2318,
10.1038/ismej.2017.93, 2017.Marshall, V. G. and Kevan, D. K. McE.: Preliminary Observations on the Biology of Folsomia candida Willem, 1902 (Collembola: Isotomidae), Can. Entomol., 94, 575–586, 10.4039/Ent94575-6, 1962.Martin, M.: Cutadapt removes adapter sequences from high-throughput
sequencing reads, EMBnet.journal, 17, 10–12,
10.14806/ej.17.1.200, 2011.Marushchak, M. E., Kerttula, J., Diáková, K., Faguet, A., Gil, J.,
Grosse, G., Knoblauch, C., Lashchinskiy, N., Martikainen, P. J.,
Morgenstern, A., Nykamb, M., Ronkainen, J. G., Siljanen, H. M. P., van
Delden, L., Voigt, C., Zimov, N., Zimov, S., and Biasi, C.: Thawing Yedoma
permafrost is a neglected nitrous oxide source, Nat. Commun., 12, 7107,
10.1038/s41467-021-27386-2, 2021.
Masson-Delmotte, V., Zhai, P., Pirani, A., Connors, S. L., Péan, C.,
Berger, S., Caud, N., Chen, Y., Goldfarb, L., Gomis, M. I., Huang, M.,
Leitzell, K., Lonnoy, E., Matthews, J. B. R., Maycock, T. K., Waterfield,
T., Yelekçi, Ö., Yu, R., and Zhou, B. (Eds.): Climate Change 2021:
The Physical Science Basis. Contribution of Working Group I to the Sixth
Assessment Report of the Intergovernmental Panel on Climate Change,
Cambridge University Press, 2021.Mitchell, M. J.: An improved method for microrespirometry using gas
chromatography, Soil Biol. Biochem., 5, 271–274,
10.1016/0038-0717(73)90011-4, 1973.Monteux, S., Weedon, J. T., Blume-Werry, G., Gavazov, K., Jassey, V. E. J.,
Johansson, M., Keuper, F., Olid, C., and Dorrepaal, E.: Long-term in situ
permafrost thaw effects on bacterial communities and potential aerobic
respiration, ISME J., 12, 2129–2141,
10.1038/s41396-018-0176-z, 2018.Monteux, S., Keuper, F., Fontaine, S., Gavazov, K., Hallin, S., Juhanson,
J., Krab, E. J., Revaillot, S., Verbruggen, E., Walz, J., Weedon, J. T., and
Dorrepaal, E.: Carbon and nitrogen cycling in Yedoma permafrost controlled
by microbial functional limitations, Nat. Geosci., 13, 794–798,
10.1038/s41561-020-00662-4, 2020.Monteux, S.: Data analysis pipeline for bacterial community composition and CO2 fluxes in a permafrost-Collembola incubation, Software version 1.1.1, Bolin Centre Code Repository [code], 10.57669/monteux-2022-collembola-1.1.1, 2022.Monteux, S., Mariën, J., and Krab, E.: Data supporting “Dispersal of bacteria and stimulation of permafrost decomposition by Collembola”, Zenodo [data set], 10.5281/zenodo.7015077, 2022.Nickerl, J., Helbig, R., Schulz, H.-J., Werner, C., and Neinhuis, C.:
Diversity and potential correlations to the function of Collembola cuticle
structures, Zoomorphology, 132, 183–195,
10.1007/s00435-012-0181-0, 2013.Oksanen, J., Simpson, G. L., Blanchet, F. G., Kindt, R., Legendre, P., Minchin, P. R., O’Hara, R. B., Solymos, P., Stevens, M. H. H., Szoecs, E., Wagner, H., Barbour, M., Bedward, M., Bolker, B., Borcard, D., Carvalho, G., Chirico, M., Caceres, M. D., Durand, S., Evangelista, H. B. A., FitzJohn, R., Friendly, M., Furneaux, B., Hannigan, G., Hill, M. O., Lahti, L., McGlinn, D., Ouellette, M.-H., Cunha, E. R., Smith, T., Stier, A., Braak, C. J. F. T., and Weedon, J.: vegan: Community Ecology Package, R package [code], https://CRAN.R-project.org/package=vegan, last access: 22 August 2022.Potapov, A., Bellini, B. C., Chown, S. L., Deharveng, L., Janssens, F.,
Kováč, Ĺ., Kuznetsova, N., Ponge, J.-F., Potapov, M., Querner, P.,
Russell, D., Sun, X., Zhang, F., and Berg, M. P.: Towards a global synthesis
of Collembola knowledge: challenges and potential solutions, soil organisms,
92, 161–188, 10.25674/so92iss3pp161, 2020.Potapov, A. M., Beaulieu, F., Birkhofer, K., Bluhm, S. L., Degtyarev, M. I., Devetter, M., Goncharov, A. A., Gongalsky, K. B., Klarner, B., Korobushkin, D. I., Liebke, D. F., Maraun, M., Mc Donnell, R. J., Pollierer, M. M., Schaefer, I., Shrubovych, J., Semenyuk, I. I., Sendra, A., Tuma, J., Tůmová, M., Vassilieva, A. B., Chen, T.-W., Geisen, S., Schmidt, O., Tiunov, A. V., and Scheu, S.: Feeding habits and multifunctional classification of soil-associated consumers from protists to vertebrates, Biol. Rev., 97, 1057–1117, 10.1111/brv.12832, 2022.Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P.,
Peplies, J., and Glöckner, F. O.: The SILVA ribosomal RNA gene database
project: improved data processing and web-based tools, Nucl. Acids Res.,
41, D590–D596, 10.1093/nar/gks1219, 2013.R Core Team: R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, https://www.R-project.org/, last access: 22 August 2022.Seastedt, T. R. and Crossley Jr., D. A.: The Influence of Arthropods on Ecosystems, BioScience, 34, 157–161, 10.2307/1309750, 1984.Sfenthourakis, S. and Hornung, E.: Isopod distribution and climate change,
Zookeys, 801, 25–61, 10.3897/zookeys.801.23533, 2018.Shatilovich, A. V., Tchesunov, A. V., Neretina, T. V., Grabarnik, I. P.,
Gubin, S. V., Vishnivetskaya, T. A., Onstott, T. C., and Rivkina, E. M.:
Viable Nematodes from Late Pleistocene Permafrost of the Kolyma River
Lowland, Dokl. Biol. Sci., 480, 100–102,
10.1134/S0012496618030079, 2018.Shmakova, L., Malavin, S., Iakovenko, N., Vishnivetskaya, T., Shain, D.,
Plewka, M., and Rivkina, E.: A living bdelloid rotifer from 24,000-year-old
Arctic permafrost, Curr. Biol., 31, R712–R713,
10.1016/j.cub.2021.04.077, 2021.Shur, Y., French, H. M., Bray, M. T., and Anderson, D. A.: Syngenetic
permafrost growth: cryostratigraphic observations from the CRREL tunnel near
Fairbanks, Alaska, Permafrost Periglac., 15, 339–347,
10.1002/ppp.486, 2004.Sørensen, L. I., Holmstrup, M., Maraldo, K., Christensen, S., and
Christensen, B.: Soil fauna communities and microbial respiration in high
Arctic tundra soils at Zackenberg, Northeast Greenland, Polar Biol., 29,
189–195, 10.1007/s00300-005-0038-9, 2006.Strauss, J., Schirrmeister, L., Grosse, G., Fortier, D., Hugelius, G.,
Knoblauch, C., Romanovsky, V., Schädel, C., Schneider von Deimling, T.,
Schuur, E. A. G., Shmelev, D., Ulrich, M., and Veremeeva, A.: Deep Yedoma
permafrost: A synthesis of depositional characteristics and carbon
vulnerability, Earth-Sci. Rev., 172, 75–86,
10.1016/j.earscirev.2017.07.007, 2017.Theenhaus, A., Scheu, S., and Schaefer, M.: Contramensal interactions
between two collembolan species: effects on population development and on
soil processes, Funct. Ecol., 13, 238–246,
10.1046/j.1365-2435.1999.00303.x, 1999.Väisänen, M., Krab, E. J., Monteux, S., Teuber, L. M., Gavazov, K.,
Weedon, J. T., Keuper, F., and Dorrepaal, E.: Meshes in mesocosms control
solute and biota exchange in soils: A step towards disentangling (a)biotic
impacts on the fate of thawing permafrost, Appl. Soil Ecol., 151,
103537, 10.1016/j.apsoil.2020.103537, 2020.Wang, M., Zhang, W., Xia, H., Huang, J., Wu, Z., and Xu, G.: Effect of
Collembola on mineralization of litter and soil organic matter, Biol. Fert.
Soils, 53, 563–571, 10.1007/s00374-017-1200-6, 2017.Wang, Q., Garrity, G. M., Tiedje, J. M., and Cole, J. R.: Naive Bayesian
Classifier for Rapid Assignment of rRNA Sequences into the New Bacterial
Taxonomy, Appl. Environ. Microb., 73, 5261–5267,
10.1128/AEM.00062-07, 2007.Wang, Y., Naumann, U., Wright, S. T., and Warton, D. I.: mvabund – an R
package for model-based analysis of multivariate abundance data, Methods
Ecol. Evol., 3, 471–474,
10.1111/j.2041-210X.2012.00190.x, 2012.
Warton, D. I., Wright, S. T., and Wang, Y.: Distance-based multivariate
analyses confound location and dispersion effects, Methods Ecol. Evol., 3, 89–101, 10.1111/j.2041-210X.2011.00127.x,
2012.Xiang, Q., Zhu, D., Chen, Q.-L., Delgado-Baquerizo, M., Su, J.-Q., Qiao, M.,
Yang, X.-R., and Zhu, Y.-G.: Effects of diet on gut microbiota of soil
collembolans, Sci. Total Environ., 676, 197–205,
10.1016/j.scitotenv.2019.04.104, 2019.Yashina, S., Gubin, S., Maksimovich, S., Yashina, A., Gakhova, E., and
Gilichinsky, D.: Regeneration of whole fertile plants from 30,000-y-old
fruit tissue buried in Siberian permafrost, P. Natl. Acad. Sci. USA, 109, 4008–4013,
10.1073/pnas.1118386109, 2012.