Physical Inputs, Internal Responses, and
Repair in Medusozoa
A selective evidence synthesis with a descriptive
Aurelia regeneration time course
Working manuscript · 18 September 2026
Literature synthesis and secondary analysis of public data. Unnumbered draft; not submitted,
published, or independently peer reviewed.
Abstract
Background. Physical conditions, cellular signaling, and tissue organization influence cnidarian
repair, but these findings do not necessarily establish a retained internal state after a separate
environmental cue ends. We ask whether a brief, naturally realistic mechanical or electrical
input can produce a measurable post-input change that affects closure, remodeling, or organ
regeneration.
Methods. We integrate a supplied synthesis of 30 primary-research and associated-data
records with a descriptive reanalysis of the public single-cell atlas of Aurelia coerulea
rhopalium regeneration. Nine pooled post-amputation samples contain 205,747 captured cells.
We distinguish cell-type proportions, within-cell-type normalized RNA, and transcript detection,
without treating cells as independent animals.
Results. Published experiments connect mechanical stretch to Hydra organizer activity,
electrical conditions to reversible morphogenesis, and wound-associated molecular functions to
Aurelia blastema formation. Repair endpoints are separable: wound closure and remodeling
need not entail replacement of a functional organ. In the deposited Aurelia atlas, the
wound-induced-cell (WIC) cluster constitutes 5.246% of captured cells immediately after
amputation, 38.341% at 0.5 days, 36.136% at day 1, and 0.663% at day 2. Its preselected
20-gene signature also has its highest within-cluster mean at 0.5 days, although individual
candidate markers differ. Later estimates become sparse, reaching four wound-induced cells at
day 7.
Conclusion. The combined evidence supports a temporally structured post-injury repair
response and experimentally consequential interactions between physical conditions and
cellular programs. It does not yet establish the complete
separate-cue-to-retained-state-to-repair sequence, intracellular hydration as its mediator, or a
naturally realistic heartbeat exposure. The synthesis identifies what is measured, what the new
calculations add, and which causal connections remain open.
Keywords: Medusozoa; regeneration; mechanobiology; bioelectricity; wound-induced cells;
single-cell RNA sequencing; post-input persistence.
1. Introduction
Regeneration requires more than survival after damage. Tissue boundaries can close, surviving
structures can rearrange, and missing organs can be replaced; these outcomes need not share
the same requirements. In cnidarians, mechanical stretch can influence organizer signaling,
imposed electrical conditions can redirect morphogenesis, and wound-associated molecular
programs contribute to organ rebuilding. These findings make the interaction between physical
inputs and cellular responses a substantive research question. [1] [2] [3]
The central question is more specific than whether jellyfish respond to their surroundings: can a
brief, naturally realistic mechanical or electrical input produce a measurable internal
change after the input stops, and does that change affect a defined repair outcome?
Injury itself must be distinguished from a separate cue. Amputation creates both a wound and
the conditions for its response; it is not a clean test of a noninjurious pulse followed by a
retained state.
The relevant internal variables also require definition. Cytoplasmic mechanics, calcium-probe
activity, gene expression, cell-population composition, and tissue architecture are not
interchangeable measurements. Mammalian-cell studies establish intracellular mechanical and
membrane-repair components, whereas jellyfish extracellular-matrix measurements address a
different compartment. Extracellular ATP effects on jellyfish epithelial closure do not, by
1


themselves, measure intracellular hydration. [4] [5] [6] [8]
This paper makes two linked contributions. First, it organizes the existing evidence by exposure,
measured response, repair endpoint, and inferential limit. Second, it adds completed
calculations from the Aurelia regeneration atlas to a question that the earlier synthesis left for
future analysis. The resulting account is a selective synthesis with a descriptive secondary
analysis, not a new biological experiment or a claim that one mechanism has been
demonstrated across all included species.
2. Materials and methods
2.1 Literature scope and evidence classification
The literature component uses the supplied 18 September 2026 synthesis and its saved source
register. Its 30 reference records include primary studies and associated data, not 30
independent replications. The original synthesis used focused searches across six questions; this
compilation does not claim a new exhaustive search, systematic-review selection procedure, or
meta-analysis. The uploaded 13-page PDF is byte-identical to the previously saved synthesis.
Medusozoan evidence spans freshwater Hydra polyps and marine medusae from different
classes and life stages. Mammalian cellular studies and a colonial chordate study are retained as
explicitly external comparisons, not as jellyfish experiments. Findings are classified as published
experimental observations or perturbations, computational inferences, descriptive
secondary-analysis results, or proposed connections. Absence of a required comparison is
reported as unresolved in the located evidence, rather than as experimental refutation.
2.2 Aurelia data and experimental units
The article-linked processed-data/code deposit was verified at Zenodo record 19466460, version
3. Its Seurat object contains nine raw RNA count layers and 29,766 genes in their union. The
BioProject accession PRJNA1224934 corroborates organism and study identity; its public run
metadata were inspected without downloading sequencing reads. The official supplementary
PDF and Dataset S1 workbook were obtained through Europe PMC. [3] [31] [32] [33]
The processed samples are H0, H12, H24, H48, H72, H96, H120, H144, and H168, corresponding
to 0, 0.5, 1, 2, 3, 4, 5, 6, and 7 days after amputation (dpa). The methods pool tissue from three
medusae before single-cell capture at each stage. This yields one measured pool per stage, not
three independently measured animal replicates. Day 0 is already post-amputation. Archive
records share one BioSample and do not provide animal-level identities. A day-14
single-cell-classified run exists in the archive but is absent from this processed nine-stage atlas
and was not added to the analysis. [3] [32]
The separate bulk experiment describes 51 libraries: three biological replicates for intact
controls and each of 16 post-amputation stages. No usable bulk expression matrix was located
in the inspected materials. Bulk replication was therefore not transferred to the single-cell
results. [3] [33]
2.3 Label audit and descriptive calculations
The stored 16-group clustering, RNA_snn_res.0.1, was used instead of the object's default
62-group clustering. Cluster 2 in this solution was used as the authors' wound-induced-cell (WIC)
group. Across 77 preselected atlas markers, all 154 recomputed within/outside-cluster detection
fractions matched Dataset S1 to its three-decimal rounding. Every stage's captured-cell total
matched Table S2, summing to 205,747 cells. RNA unique molecular identifier (UMI) totals and
detected-gene totals also matched deposited metadata for every cell. These are technical
consistency checks on the same dataset, not independent biological validation. [31] [33]
Captured-cell composition was calculated as WIC-assigned cells divided by all retained captured
cells at a stage. For each gene and cell, RNA expression was normalized as log1p(10000 ×
gene UMIs / total cell RNA UMIs). Means retain zero-count cells. Detection is the fraction of
cells with at least one gene UMI and is reported separately.
The 20-gene signature is the equal-gene mean of the top 20 positive cluster-2 markers ranked
by Dataset S1 fold change before inspecting their time courses. It is a descriptive signature, not
a validated pathway-activity or stemness score. Three other audit-panel genes absent from
some feature lists were excluded rather than assigned artificial zeros; all signature genes and
named annotation candidates remained available across stages.
Only maxima at actual sampled times are reported. No animal-level confidence intervals,
cell-level significance tests, interpolated onset times, or decay constants were estimated. Stored
2


gene annotations are treated as candidates unless an authoritative gene-ID/symbol
correspondence is available. Pseudotime is not substituted for elapsed sampling time.
3. Literature evidence: inputs, internal responses,
and distinct repair endpoints
3.1 Mechanical conditions can influence organizer activity
In Hydra tissue spheroids, suppressing osmotic inflation prevented normal Wnt3 organizer
activity and regeneration, while sustained Wnt3 expression or pre-existing organizer tissue
could bypass the usual mechanical requirement. This provides evidence for a functional
connection between mechanical conditions and organizer state. Restoring normal medium,
however, restores continuing water influx and stretch. Measurements during that renewed
regime are not observations after a completed finite stretch pulse. [1]
Other experiments extend the mechanical picture without completing that post-input
comparison. Microaspiration characterizes deformation, flow, and rupture of regenerating Hydra
tissue, but does not establish subsequent organizer activation by a brief alternative stimulus.
Repeated local strain events occur near future head sites during regeneration, with
strain-to-morphogen coupling partly represented by a model. Altered contraction rates in Aurelia
can change stable body shape, supporting an active mechanical contribution to form rather than
a demonstrated retained state after an external cue. [9] [10] [11]
Intracellular water remains relevant as a possible physical component. In mammalian cultured
cells, osmotic volume changes alter cytoplasmic mechanical behavior, and much of the
measured indentation relaxation occurs on a sub-second scale under the reported conditions.
This is evidence for intracellular poroelastic behavior, not a measured lifetime for regenerative
memory in a jellyfish. Mesogleal rheology addresses extracellular material and cannot replace
an intracellular measurement. [4] [8]
3.2 Electrical conditions alter tissue state; withdrawal matters
Prolonged alternating electrical stimulation can reversibly drive regenerated Hydra tissues
toward a spheroid-like state. Calcium-probe activity increases while Wnt3 organizer signal
declines during reversal; setting voltage to zero permits renewed Wnt3 expression and
regeneration. The response depends on stimulus conditions: changing frequency at the same
reported voltage can alter the outcome. These observations establish electrical control and a
genuine post-removal regrowth outcome, but do not isolate a lasting intracellular mediator
produced by a brief naturalistic pulse. [2]
Calcium activity alone is not an adequate marker of regenerative benefit. The same study's
extracellular-potassium comparison increased calcium-probe activity and caused tissue folding
without the corresponding recoverable regenerative outcome. Initial wound sealing also
preceded electrical exposure in the spheroid experiments. A change in activity must therefore
be interpreted alongside tissue state and the endpoint actually tested. [2]
Implanted-electrode pacing in Aurelia establishes control of contractions and swimming during
stimulation. Later embodied-computing work extends analysis of stimulated body dynamics, but
neither supplies a post-off repair assay. Electrode voltage cannot be equated with ambient field
strength at a naturally exposed animal, especially across different electrode geometries and
freshwater versus seawater conditions. [12] [13]
An informative external comparison is Botryllus schlosseri, a colonial chordate. A finite
electrical-treatment regimen was associated with later differences in normal budding, colony
output, and other traits. The 35-day budding comparison used seven colonies per group. RNA
samples labeled two and 24 hours post-treatment provide reported molecular differences, but
the main text does not unambiguously assign them to a particular session schedule or define
time zero relative to its first or final session. Separate treatment/control tanks and the absence
of a reported electrode-connected zero-current sham leave alternatives unresolved. This is
neither a jellyfish wound-regeneration experiment nor a demonstration that the measured RNA
changes mediate the later outcome. [23]
3.3 Moving water permits regrowth under some conditions, but the
mechanism is unresolved
In amputated Aurelia ephyrae, continuing water motion in a permissive setup was associated
with arm regeneration over the recovery period. Outgrowth could contain canals, muscle,
neurons, and sometimes a rhopalium, with regenerated arms participating in coordinated
pulsing. Frequencies and completeness varied. These findings concern more than sealing, but
3


not complete restoration in every animal. [14] [15]
Water motion, swimming, posture, loading, feeding access, and transport were not
independently isolated. Reduced oxygen could also promote regeneration in the permissive
setup, so increased oxygen supply alone is not an adequate explanation of the current effect.
Nutritional interventions provide further candidates without resolving their interaction with flow.
The earlier mechanically driven symmetrization experiments concern rearrangement of
surviving tissue, a different endpoint from arm replacement. The current-screen data can
describe tested associations, but cannot supply unmeasured intracellular variables or absent
factorial controls. [14] [15] [16]
3.4 Closure, remodeling, and functional replacement are separable
The distinction between repair endpoints is experimentally consequential. In Clytia, hydroxyurea
treatment impaired manubrium outgrowth without preventing initial closure or umbrella
remodeling. This pharmacological contrast is useful without assuming every drug effect is
exclusively proliferation-specific. Extracellular ATP also affects Clytia epithelial closure and actin
organization, but was not established as a pulse-and-washout demonstration of a retained state.
Mammalian calcium- and ATP-dependent membrane resealing concerns the barrier of individual
cells, not the same compartment or experiment. [17] [6] [5]
Table 1. Selected repair endpoints and their interpretation.
Organism and context
Observed endpoint
Important distinction
Clytia medusae
Closure
 by
 about
 12
 h;
feeding-functional manubrium
by day 4 in 43/44 untreated
animals. [17]
Closure/remodeling can occur
without
 normal
 organ
outgrowth.
Cladonema
 bulb-retaining
tentacles
Closure by 24 h; prey capture
in 55% at 48 h and 100% at 72
h,
 n
 =
 36
 per
 reported
time-point assay. [18]
Functional
 replacement
exceeds sealing; contributions
of repair-specific and resident
cells differ.
Haliclystus tentacles, preprint
Both tested types sealed by 48
h; secondary tips regrew by
168 h while primary tips did
not within that window. [7]
Developmental/tentacle
contexts differ; this is not
permanent
 or
 class-wide
incapacity.
Tripedalia
 rhopalia,
abstract-only evidence
Rhopalial/neural
 regrowth,
renewed
 pacemaking,
 and
lens-eye light responses were
reported. [19]
Structural
 and
 functional
recovery
 were
 assessed;
learned-state control of repair
was not.
The Haliclystus result remains preliminary: primary and secondary assays used small animal
groups, with within-animal comparisons, and proliferation was not isolated by a selective
blockade/rescue. Life-cycle reversal is another distinct endpoint. The historical Turritopsis
nutricula report describes medusa-to-hydroid reversal, not a demonstrated water mechanism
or indefinite survival of every individual. [7] [30]
4. Aurelia rhopalium regeneration: published
perturbations and new descriptive results
4.1 What the original experiments establish
Li and colleagues identified an early wound-induced-cell population associated with rhopalium
regeneration and tested molecular requirements for blastema formation. After knockdown,
blastemas formed in 0/16 Soxf2-treated individuals and 0/16 Otx5b-treated individuals,
compared with 15/16 controls; Wntless knockdown yielded 4/16. These are the authors'
animal-level perturbation results, not new experiments or tests from the present analysis. They
support gene functions under the reported conditions without establishing selective necessity of
the entire WIC population, which was not specifically ablated. [3]
Regenerating rhopalia recovered anatomical structures and supported progressive swimming
recovery. The article describes wound healing at 0–1 dpa, blastema development at 1–3 dpa,
later statocyst and ocellus formation, and continued growth to day 14. Swimming recovery is a
functional outcome, but not an assay of complete recovery of every sensory capacity. Inferred
4


cell trajectories and expression-based communication analyses remain distinct from direct
lineage tracing. [3]
4.2 Observed captured-cell time course
The reanalysis resolves the early dynamics in actual sampled time rather than only broad phase
labels. The maximum WIC fraction occurs at 0.5 dpa, while the maximum captured WIC count
occurs at day 1. These are different summaries because total captured cells also vary. Neither is
an absolute tissue abundance estimate. Results derive from the authors' processed atlas and
supplementary marker reference. [31] [33]
Table 2. Captured-cell composition and within-WIC RNA signature.
Days
Captured total
WIC count
WIC share (%)
Within-WIC
20-gene mean
0
10,904
572
5.246
0.589
0.5
21,079
8,082
38.341
0.853
1
29,796
10,767
36.136
0.567
2
17,943
119
0.663
0.341
3
20,291
135
0.665
0.379
4
15,207
21
0.138
0.517
5
25,300
41
0.162
0.437
6
37,454
23
0.061
0.116
7
27,773
4
0.014
0.217
The article's narrative describes cluster-2 proportions as peaking during blastema formation.
The deposited stage summaries put the numerical maximum at 0.5 dpa, with a similarly high
fraction at day 1. The new draft therefore reports the exact sampled stages and flags the broad
phase wording for clarification; it does not silently move the samples or reinterpret captured
fractions as tissue cell counts. [3] [31]
5


Figure 1. Captured WIC fraction and the preselected 20-gene signature. Expression means
within WICs are separated from means across all captured cells. Lines connect observations, not
inferred trajectories. Shaded late stages contain only 4–41 WICs each. There are no animal-level
error bars because separately measured replicate pools are unavailable.
The signature's highest within-WIC mean is at 0.5 dpa. Later apparent rebounds have small
denominators: the day-4 mean describes 21 cells and the day-7 mean four. The article's neural
cluster 5 has a captured share of 3.26% at day 0, a sampled maximum of 11.32% at day 4, and
9.09% at day 7. Sensory hair-cell cluster 8 changes from 5.36% to 1.01% between days 0 and 7.
These shifts do not demonstrate WIC-to-neuron conversion. [3] [31]
4.3 Marker expression is not interchangeable with population
representation
Dataset S1 contains protein-description annotations, not a certified Aurelia gene-symbol
crosswalk. The gene IDs below identify the measured rows; their biological names remain
candidates. The published gene-symbol knockdowns and these RNA rows should not be treated
as an independently verified gene-specific causal chain without that crosswalk. [3] [33]
Table 3. Candidate-marker mean log-normalized RNA within WICs.
Gene
Mean 0 d
Mean 0.5 d
Mean 2 d
Within-WIC
sampled
maximum
Aco03402
0.347
1.869
0.120
0.5 d
Aco10552
0.293
0.778
0.175
0.5 d
Aco03861
0.168
0.331
0.093
0.5 d
Aco07511
0.778
2.248
1.226
0.5 d
Aco20833
1.698
1.152
0.256
0 d
Candidate labels are Col6a5 (Aco03402), Otx5b (Aco10552), Fzd10 (Aco03861), Wntless
(Aco07511), and Plg (Aco20833). Table 3 abbreviates the row identifiers by omitting the gene-
prefix.
The Plg-annotated candidate gene-Aco20833 demonstrates why the distinction matters: its
highest within-WIC mean occurs at day 0, whereas its mean across all captured cells is highest
at day 0.5. A whole-sample maximum need not represent the same expression timing inside a
cell type. Conversely, the Col6a5, Otx5b, Fzd10, and Wntless candidates listed here have their
highest within-WIC mean at day 0.5.
Figure 2. Candidate-marker normalized means and detection fractions are separate
measurements. At day 0.5, Col6a5-, Otx5b-, and Wntless-annotated candidates are detected in
51.4%, 26.8%, and 67.4% of WICs. The day-7 Wntless-candidate detection of 75% represents
only 3/4 captured cells, not 75% of animals or a reliable late activation estimate.
Both Plg candidates and both Piwi-like candidates are retained in the complete result tables
because their timing differs. SoxF2/SOX18 annotations do not establish equivalence to the
supplement's Sox2 label. Wnt1 and Wnt3 lacked exact description matches in the inspected
workbook. An opsin candidate has rare day-0 UMIs, but an uncertified gene/subcluster mapping
6


cannot independently confirm or refute the paper's reported photoreceptor onset. [33]
5. Integrated interpretation
5.1 A changing repair program is not yet a same-cell persistence
measurement
The combined Aurelia evidence is stronger than a morphological description alone: published
gene perturbations identify functional requirements, while the reanalysis quantifies a rapidly
changing early captured population and its RNA profile. Their combination supports a temporally
structured post-injury response. It does not show that one identified cell retains a state across
days, because sequencing destructively samples different cells at each stage.
This distinction is compatible with active maintenance. A post-input state need not be passive or
require the organism to become inactive. It would need an identified variable that remains
different from an appropriate control after a verified external cue ends. A lesion, continuing
exposure, recurrent muscle activity, a transcriptional program, and altered neural
responsiveness can all last for different reasons; duration alone does not identify a common
mechanism.
Recent Hydra cell-state work supports tracking changing positional identities and organizer
organization over time rather than treating recovery of shape as the only endpoint. A separate
calcium/polarity study supplies mechanochemical context but was available here only through
abstract and figure captions. Neuromechanical and neural-network models generate additional
hypotheses; computational aftereffects must remain identified as predictions rather than
measured post-input cellular memory. [21] [22] [28] [29]
5.2 Learning, natural exposure, and repair require an explicit
connection
Tripedalia learning experiments establish associative changes in behavior and
isolated-rhopalium responsiveness. The reported post-training response is relevant, but the
study does not define a stimulus-free delay demonstrating retention for hours or days, nor does
it test repair. Combining evidence that a species learns with separate evidence that it
regenerates a rhopalium does not establish that learned state controls regeneration. [20] [19]
Naturalistic exposure is a separate measurement problem. Prolonged low-frequency playback
produced later statocyst lesions in two scyphozoan species, demonstrating a post-exposure
tissue effect under those conditions. It was high-level exposure with an injury endpoint, not
evidence of a brief beneficial cue; particle motion at the animal was not measured. Aglantha
circuit physiology addresses internal signaling, not a calibrated remote heartbeat-detection
threshold. [24] [25]
Marine bioelectric source measurements and immersed human ECG recordings also do not
establish the electric gradient received by a jellyfish. Source geometry, distance, conductive
medium, duration, and receiver sensitivity are required. Heartbeat-associated mechanical
motion and cardiac electrical activity must be considered separately. The located evidence
supplies no defensible heartbeat-to-jellyfish detection distance or repair effect. [26] [27]
5.3 The remaining causal comparison
The strongest open question is not whether physical conditions matter; they already do in the
scoped experiments. It is whether a separate, completed, naturally realistic cue produces
an identified post-off state that changes a specified repair outcome in the same biological
comparison.
That connection would require verified receiver exposure and cessation, a post-input internal
measurement, an outcome that distinguishes closure from remodeling or functional
replacement, and evidence that the internal change contributes to the outcome rather than
merely accompanies it. These are evidential requirements, not a laboratory protocol or a claim
that the missing experiment has been performed. Intracellular hydration remains one possible
variable to measure, not an inferred cause supplied by a lumen, calcium probe, gene signature,
or mesogleal measurement.
6. Limitations
The literature is selective, biologically heterogeneous, and includes single-study findings,
historical primary evidence, an unreviewed Haliclystus preprint, and access-limited records.
7


Tripedalia regeneration, Aglantha circuitry, and historical Turritopsis reversal are abstract-only
in this source set; the 2026 Hydra polarity record adds figure captions, not a full-text
assessment. Thirty-three cited research/data records are not thirty-three independent studies or
replications. [7] [19] [22] [25] [30]
For the secondary analysis, one pooled sample per stage prevents estimation of between-animal
or between-pool variation. Dissociation, capture efficiency, stage-associated batch differences,
and sequencing depth may affect comparisons. Median UMIs vary across stages; normalization
does not replace a coverage-matched sensitivity analysis, which was not performed. Candidate
annotations and very small late WIC groups further limit marker interpretation. RNA
measurements do not directly establish protein activity.
Neither the single-cell atlas nor the located literature completes a pulse-withdrawal,
intracellular-hydration, or heartbeat-to-repair comparison. A missing variable cannot be
recovered by reinterpretation of a different measurement. These limits constrain the proposed
connection without erasing the directly observed mechanical, cellular, and regenerative
findings.
7. Conclusion
Physical conditions and cellular programs have experimentally consequential roles in
medusozoan tissue behavior, and closure, remodeling, and organ replacement are separable
outcomes. The Aurelia reanalysis adds a reproducible early repair-cell time course: captured
WIC share and the selected within-WIC signature are highest at 12 hours, with high
representation still present at 24 hours and a sharp fractional decline by 48 hours. Individual
marker timing is not uniform.
The resulting research model is specific: physical conditions, internal activity, cell composition,
and tissue reconstruction can interact, but each must be measured on its own terms. The
complete separate-cue-to-retained-state-to-repair sequence remains a testable open connection
rather than a demonstrated common mechanism.
Data availability and manuscript status
External data and supplements are linked below. Reproducible extraction, audit, calculation, and
test instructions are retained in research/aurelia-timecourse/README.md; the compilation
folder contains the evidence ledger, reference crosswalk, input hashes, and rendering
instructions. The supplied synthesis and original analysis remain unchanged. No raw-read
alignment, new laboratory experiment, corpus revision, app change, or publication was
performed. This manuscript is an AI-assisted working compilation, not an independently
peer-reviewed paper.
References
1. Mechanical oscillations orchestrate axial patterning through Wnt activation in Hydra
(2021-12-10; open full text; publication status: peer reviewed)
2. Electric-Induced Reversal of Morphogenesis in Hydra (2019-09-16; open full text; publication
status: peer-reviewed; single-laboratory result, no independent replication located)
3. Deciphering the ancestral mechanisms of neural regeneration through single-cell analyses of
jellyfish rhopalium repair (2026-05-20; open full text; publication status: peer-reviewed)
4. The cytoplasm of living cells behaves as a poroelastic material (2013-01-06; author
manuscript open full text; publication status: peer reviewed)
5. Repair of injured plasma membrane by rapid Ca2+-dependent endocytosis (2008-03-10; open
full text; publication status: peer reviewed)
6. Epithelial wound healing in Clytia hemisphaerica provides insights into extracellular ATP
signaling mechanisms and P2XR evolution (2023-11-01; open full text; publication status:
peer-reviewed)
7. Regression of juvenile tentacles is driven by loss of cell proliferation in Haliclystus
sanjuanensis, a cnidarian with limited metamorphosis (2026-04-01; open full text; publication
status: bioRxiv Version 1; not peer reviewed as of 2026-09-18)
8. Micro- and Macrorheology of Jellyfish Extracellular Matrix (2012-01-04; open full text;
publication status: peer-reviewed)
8


9. Mechanical characterization of regenerating Hydra tissue spheres (2024-05-23; open full text;
publication status: peer reviewed)
10. Mechanical strain focusing at topological defect sites in regenerating Hydra (2025-03-03;
open full text; publication status: peer reviewed)
11. Jellyfish shape as a mechanical balance (2025-03-17; open full text; publication status:
peer-reviewed; associated correction concerns acknowledgments)
12. Low-power microelectronics embedded in live jellyfish enhance propulsion (2020-01-29;
open full text; publication status: peer-reviewed; later locomotion follow-ups exist, repair
consequence not tested)
13. Harnessing natural embodied intelligence for spontaneous jellyfish cyborgs (2025-05-23;
open full text; publication status: peer-reviewed 2025 follow-up; no repair endpoints)
14. A conserved strategy for inducing appendage regeneration in moon jellyfish, Drosophila, and
mice (2021-12-07; open full text; publication status: peer reviewed)
15. Figure 2 source data: Aurelia factor screen (2021-12-07; open spreadsheet; publication
status: published with peer-reviewed article)
16. Self-repairing symmetry in jellyfish through mechanically driven reorganization (2015-06-15;
open full text; publication status: peer-reviewed)
17. Pattern regulation in a regenerating jellyfish (2020-09-07; open full text and source data;
publication status: peer reviewed)
18. Distinct stem-like cell populations facilitate functional regeneration of the Cladonema
medusa tentacle (2023-12-21; open full text and data; publication status: peer reviewed)
19. Regeneration of the Rhopalium and the Rhopalial Nervous System in the Box Jellyfish
Tripedalia cystophora (2018-02-01; abstract via EPA HERO bibliographic record; publisher full
text inaccessible; publication status: peer-reviewed)
20. Associative learning in the box jellyfish Tripedalia cystophora (2023-09-22; full text;
publication status: peer-reviewed)
21. Dynamic cell fate transitions during morphallactic regeneration (2026-06-02; open full text;
publication status: peer-reviewed)
22. The dual nature of body-axis formation in Hydra regeneration: Polarity-morphology
concurrency (2026; primary abstract and figure captions; full article not reviewed; publication
status: published journal research; experiment/model distinction retained)
23. Electrical stimulation promotes longevity and regeneration in a colonial chordate
(2026-05-26; open full text; publication status: peer-reviewed; single-study comparison)
24. Evidence of Cnidarians sensitivity to sound after exposure to low frequency noise
underwater sources (2016-12-21; open full text; publication status: single study in two
scyphozoan species; no independent replication located in searches through 2026-09-18)
25. Central Circuitry in the Jellyfish Aglantha digitale: I. The Relay System (1995-11-01; abstract
accessible; full text paywalled; publication status: older foundational circuit study; accessible
page is abstract-only)
26. Bioelectric Fields of Marine Organisms: Voltage and Frequency Contributions to Detectability
by Electroreceptive Predators (2013; open author-hosted PDF; publication status:
component/source evidence only; receiver tests concern elasmobranch literature, not jellyfish)
27. A novel method for noninvasive bioelectric measurement utilizing conductivity of seawater
(2021-03-29; open full text; publication status: component/source measurement only; not
jellyfish evidence)
28. Neuromechanical wave resonance in jellyfish swimming (2021-03-08; open full text;
publication status: peer-reviewed; experiment plus computational model)
29. From single neurons to behavior in the jellyfish Aurelia aurita (2019-12-23; open full text;
publication status: model predictions must not be treated as direct stimulus-response
measurements)
30. Reversing the Life Cycle: Medusae Transforming into Polyps and Cell Transdifferentiation in
Turritopsis nutricula (Cnidaria, Hydrozoa) (1996; primary abstract only; publisher full text
9


inaccessible in this pass; publication status: peer-reviewed; historical species name retained)
31. Authors' processed-data/code Zenodo record 19466460 v3 (2026-04-08; open; publication
status: published Zenodo record; version 3)
32. PRJNA1224934 raw-archive metadata (2025-02-17; public metadata; raw archive record;
publication status: public archive metadata record)
33. Official Europe PMC supplementary bundle for PMC13189125 (2026-05-20; official Europe
PMC supplementary bundle; publication status: official supplementary bundle)
10

