shmup/ants
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add richer ant documentation

Browse source
This commit is contained in:
Jared Miller 2026-03-09 12:12:28 -04:00
parent a3476d8ea7
commit b2fb120b6b
Signed by: shmup
GPG key ID: 22B5C6D66A38B06C
4 changed files with 1415 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

273
docs/ALARM-AND-DEFENSE.md Normal file
View file

@ -0,0 +1,273 @@
# alarm pheromones, threat response, and defense
how ants detect threats, communicate danger, and defend the colony. relevant to
features #2 (repellent pheromone) and #6 (alarm pheromone) in REALISM-IDEAS.md.
## what triggers alarm pheromone release
main triggers — all physical/biological, not environmental chemicals:
- physical disturbance of the nest or individual ant
- predator detection (olfactory, visual, or tactile)
- crushing or injury of nestmates (damaged ants release alarm compounds
from ruptured glands)
- intrusion by non-nestmates into the colony
ants exhibit "enemy specification" — dangerous species are more effective at
evoking alarm than less threatening ones. this is not a generic startle reflex.
### specific predator triggers
- army ants (Neivamyrmex spp.): minor workers of Pheidole desertorum and
P. hyatti initiate panic alarm leading to nest evacuation specifically when
they detect approaching army ants. distinct from response to other threats.
- the spider Habronestes bradleyi exploits Iridomyrmex purpureus alarm
pheromone (6-methyl-5-hepten-2-one) as a kairomone — the predator is
literally attracted by the alarm signal, turning the defense against the ants.
### what about plants, chemicals, environmental hazards?
no strong evidence for specific plants triggering alarm pheromone release.
some alarm compounds (citronellal) exist in plant essential oils, so
cross-reactivity is plausible but undocumented.
some alarm compounds (citral, 2-heptanone, 4-methyl-3-heptanone) have
antifungal properties. the alarm system may have originally evolved for
pathogen defense, with the danger-signaling function coming later.
### environmental threat responses
- flooding: Solenopsis invicta detects rising water and responds with
colony-wide raft formation — workers link legs to create buoyant living
rafts, brood placed on top. instinctive, coordinated.
- chemical contamination: cadmium exposure degrades olfactory sensitivity in
fire ants, reducing bait search efficiency. at higher doses, reverses
attraction to food odorants entirely. ants detect contamination through
impaired function more than active avoidance.
- fire: no specific "fire alarm" behavior. fire substantially changes ant
communities, recovery takes years. treated as a disturbance ecology question,
not real-time detection.
## alarm is graduated, not binary
graduated in at least three ways:
### concentration-dependent intensity
in Pogonomyrmex badius (harvester ant):
low intensity:
- increased locomotion
- antennae/head waving
- looping movements
- periodic gaster-lowering to substrate
high intensity:
- faster locomotion
- tighter circling
- mandible opening (gaping)
- reduced antennae waving (attention shifts from sensing to combat readiness)
### multi-component chemical modulation
in Oecophylla longinoda (weaver ant):
- major components (1-hexanol, hexanal) trigger alert and attraction
- minor, less volatile components act as markers for attack
- creates a staged escalation: volatiles spread first (alert), heavier
compounds arrive later (attack cue)
- hexanal is the most volatile — spreads fastest, causes head-raising and
jaw-opening
- hexanol is less volatile, recruits nestmates to the source
most alarm compounds fall in the C6-C10 molecular weight range, selected for
high volatility and rapid fade-out.
### context-dependent response (distance from nest)
- near the nest: alarm pheromone triggers defensive/aggressive behavior
(attack the threat)
- far from the nest (foraging area): same pheromone triggers flight/dispersal
(flee from the threat)
same chemical, different interpretation based on spatial location.
## alarm cascading and propagation
### positive feedback mechanics
alarmed ants produce alarm pheromone, which recruits and alarms additional
ants, who produce more pheromone. structurally similar to trail pheromone
reinforcement.
in territory-conflict models, peaceful ants encountering alarm pheromone
transition to aggressive state, producing their own alarm pheromone — classic
autocatalytic cascade.
### spread speed
determined by component volatility. volatile components (hexanal) expand the
active space rapidly; heavier components diffuse more slowly.
Wilson & Bossert (1963) established the theoretical framework: the "active
space" (zone where concentration >= detection threshold) expands rapidly then
contracts as the compound evaporates. for typical alarm compounds, the signal
dies out within a few minutes unless reinforced.
number of alarmed nodes decays linearly with network distance from the source.
### built-in dampening (alarm does NOT spiral out of control)
- volatility is the primary brake: alarm compounds are specifically selected
for rapid evaporation (low molecular weight, C6-C10). the signal
self-extinguishes.
- no reinforcement = fade-out: if the threat is removed, no new pheromone is
deposited, and the existing signal evaporates within minutes.
- linear decay with distance: the cascade weakens with each hop through the
network rather than amplifying.
the system is designed for fast on, fast off — the opposite of trail pheromone
which is selected for persistence.
## defensive strategies by species
### flee
- Pheidole desertorum, P. hyatti: detect army ant (Neivamyrmex) approach,
evacuate nest with brood. panic alarm.
### multi-phase defense
- Pheidole obtusospinosa: super majors block nest entrance with enlarged heads
(phragmosis), then switch to aggressive combat outside.
### autothysis (self-explosion)
- Colobopsis explodens and ~15 Colobopsis spp.: minor workers rupture their
bodies, releasing bright yellow, sticky, toxic secretion. used as an INITIAL
resort, not a last resort.
### chemical spray
- Formica rufa and other Formicinae: spray formic acid from acidopore (tip of
gaster). bite wound first, then spray acid into the wound. effective against
arthropods and even wood-boring beetles.
### trap-jaw strike
- Odontomachus bauri: mandibles close at 35-64 m/s — one of the fastest
movements in the animal kingdom. strike against substrate launches the ant
into the air for escape. strike against intruder for ejection.
### entrance blockade (phragmosis)
- Colobopsis majors, P. obtusospinosa super majors: use enlarged heads as
physical plugs at nest entrances.
### caste-based defensive labor
- P. obtusospinosa: super majors specialize in entrance-blocking (passive)
and combat (active). minors handle brood evacuation. regular majors do both.
- Colobopsis: minor workers are the suicide bombers (autothysis). major
workers are the entrance blockers.
## alarm interaction with other pheromones
### alarm + trail pheromone
alarm does NOT simply suppress trail-following. it can redirect it:
- ants may follow trails toward a threat for defense
- or away from a threat for evacuation
- context (distance from nest, threat intensity) determines which program wins
alarm and trail pheromones operate on different chemical channels (different
compounds, different glands — mandibular for alarm vs Dufour's/poison for
trail in many species). an ant can potentially process both simultaneously.
### foraging disruption
alarmed ants leave their nest pile and stop normal foraging. but given alarm
pheromone volatility (fades in minutes), foraging disruption is inherently
short-lived for localized threats.
recovery time: not well-quantified, but the volatility constraint means the
chemical signal clears within minutes. behavioral recovery follows shortly
after. the system is tuned for rapid return to baseline.
## simulation relevance
for the alarm pheromone channel (feature #6):
key parameters:
- separate channel from trail pheromone (world.A is available, or a second
world texture per INFRASTRUCTURE.md)
- high diffusion rate, fast decay (minutes, not hours)
- response is context-dependent: fight near nest, flee far from nest
- graduated intensity via concentration thresholds, not just on/off
- positive feedback with built-in decay (volatile = self-extinguishing)
- caste-specific responses if caste system is implemented
the multi-component timing (fast alert component + slow attack component) could
be modeled as:
- option A: two sub-channels with different diffusion rates
- option B: single channel with behavioral thresholds (low = alert, high =
attack)
- option B is simpler and captures the essential dynamics
for the repellent pheromone (feature #2, already has infrastructure):
- deposited at trail junctions to depleted food, not along entire failed paths
- ants encountering it U-turn or zigzag
- longer half-life than trail pheromone (~2x)
- junction detection is the hard part — requires knowing when an ant is at
a bifurcation point vs mid-trail
## sources
Alarm Communication
AntWiki antwiki.org/wiki/Alarm_Communication
Alarm pheromone processing in the ant brain
PMC2912167
Alarm pheromone composition in fungus-growing ants
PMC5371636
Alarm pheromone and alarm response of clonal raider ant
PMC9941220
Insect alarm pheromones in response to predators
Basu 2021 (WSU)
Alarm Pheromone — ScienceDirect Topics
Multi-phase defense by Pheidole obtusospinosa
PMC3014660
Colobopsis explodens
Wikipedia
Formica rufa
Wikipedia
Trap-jaw ant mandible mechanism
J Exp Biol 226(10) jeb245396
Ant territory formation model with alarm pheromones
ScienceDirect S0025556425001245
Fire ant flood raft behavior
AMDRO
Cadmium olfactory neurotoxicity in fire ants
ScienceDirect S0269749124016592
Wilson & Bossert 1963
Theoretical framework for pheromone active space dynamics
The Ants Chapter 7 — AntWiki

508
docs/FOOD-QUALITY.md Normal file
View file

@ -0,0 +1,508 @@
# food quality perception and evaluation in ants
reference doc for food quality mechanics in the simulation. all claims sourced
from published research. confidence levels noted where information is
extrapolated or less well-established.
## 1. sensory detection: what ants can taste
### sensory organs
ants detect food using gustatory (taste) sensilla distributed across multiple
body parts. the primary taste organs are:
- **foreleg tarsi** — contact chemoreceptors. in fire ants (Solenopsis invicta),
the foreleg tarsi play a MORE important role in sucrose detection than the
antennal flagellum. sensilla chaetica, trichoid II, and basiconica I/II all
have a clear pore at their tip for chemoreception.
- **antennal flagellum** — both olfactory and gustatory sensilla. used for
close-range assessment and during trophallaxis.
- **maxillary and labial palps** — mouthpart sensilla for evaluation during
ingestion.
- **pharynx** — internal gustatory sensilla that evaluate food as it enters the
crop.
**confidence: high** — well-established insect morphology, confirmed across
multiple ant species.
### gustatory receptor (GR) genes
the number of GR genes varies by species and correlates with dietary breadth:
| species | common name | GR genes |
|--------------------------|---------------------|----------|
| Linepithema humile | Argentine ant | 96 |
| Apis mellifera | honeybee | 10 |
| Drosophila melanogaster | fruit fly | 68 |
ant GR genes fall into four clades: CO2 receptors, GR43a-like (internal
fructose/nutrient sensors), sugar receptors, and bitter receptors. generalist
species tend to have expanded bitter receptor families, presumably broadening
the range of plant secondary metabolites they can detect.
**confidence: high** — genomic data from sequenced ant genomes.
### sugars
**preference hierarchy**: sucrose > glucose >> fructose (in fire ants)
- fire ant workers strongly prefer sucrose and glucose but show only weak
attraction to fructose.
- SinvGr43a is a fructose-responsive gustatory receptor in S. invicta, but it
acts primarily as an INTERNAL nutrient sensor (linked to neuropeptide
regulation and lipid metabolism) rather than a peripheral taste receptor.
- concentration matters: ants discriminate between sucrose concentrations.
1.0 M sucrose elicits strong behavioral responses (trail-laying, feeding),
while 0.01 M sucrose does not. in Lasius niger experiments, 0.2 M sucrose
led to lower food acceptance than 1.0 M.
- threshold detection in related insects: ~10 mM for antennal sensilla,
~100 mM for tarsal sensilla (moth data — ant-specific thresholds likely
similar order of magnitude but not precisely established).
**confidence: high** for preference hierarchy and concentration discrimination.
moderate for exact threshold values in ants specifically.
### amino acids and proteins
- ants discriminate essential amino acids (EAAs) from non-essential ones.
- when deficient in both carbs and EAAs and offered sucrose+EAA vs
sucrose+non-EAA solutions, ants focused foraging on the EAA solution
regardless of amino acid:carbohydrate ratio.
- S. invicta workers showed strong preference for leucine (an EAA) over other
tested amino acids, with preference intensifying at higher concentrations.
- when choosing between high-protein foods, ants preferred free amino acids
over whole proteins. no preference emerged with high-carb foods.
**confidence: high** — multiple controlled experiments across species.
### salts and minerals
- Solenopsis richteri workers prefer zinc, magnesium, and ammonium.
- sodium preference varies and shows a geographical gradient: ants farther from
the ocean consume more sodium. non-predatory species consume more sodium than
predatory species (predators get sodium from prey).
- salts and acids are attractive at low concentrations but aversive at high
concentrations (inverted U response curve).
**confidence: high** — field and lab studies.
### toxins and deterrents
- quinine is aversive to Lasius niger.
- high concentrations of caffeine in sucrose reduced feeding in Oecophylla
smaragdina (weaver ants).
- alkaloids reduced feeding in Ectatomma ruidum.
- leaf-cutter ants (Atta, Acromyrmex) avoid leaves containing anti-fungal
terpenoids that would harm their cultivar fungus.
**confidence: high** — behavioral assays with known compounds.
## 2. nutritional assessment and post-ingestive feedback
### speed of assessment
ants can compensate for nutritional deficiencies in their colony in under 10
minutes. this is fast enough that it likely involves rapid nutrient sensing
rather than slow learning/feedback loops.
### post-ingestive feedback mechanism
the term "post-ingestive feedback" refers to the process by which nutrients
interact with receptors on enteroendocrine cells in the gut after ingestion.
these cells secrete hormones that signal the brain and other tissues about
nutrient composition, food texture, and meal size.
mechanistic details (primarily from Drosophila, likely conserved in ants):
- Dh44 neurons are necessary and sufficient for post-ingestive nutrient sensing
- gut-to-brain signaling uses neuropeptide pathways
- the internal fructose sensor GR43a (mentioned above) is part of this system,
linking circulating nutrient levels to feeding behavior and lipid metabolism
**confidence: moderate** — the gut-sensor mechanism is well-established in
Drosophila. the specific molecular pathways in ants are inferred by homology
rather than directly demonstrated. the behavioral outcomes (rapid compensation)
are directly measured in ants.
### learning and memory about food quality
- ants can learn which foods are nutritious vs empty calories. foraging
motivation and food quality affect both route memory formation speed and the
likelihood of returning to a food source.
- two parameters dominate quality assessment at a food site:
1. **amount of food available** — initially dominates the decision to return
2. **reliability of food encounter** — takes precedence after a few visits
- ants may learn the location of higher-quality food faster, with most ants
eventually congregating at the best source.
**confidence: high** — direct behavioral experiments.
## 3. the geometric framework for nutrition
### core concept
the geometric framework (developed by Simpson & Raubenheimer) models nutrition
as a multi-dimensional space where each axis represents a nutrient. animals have
an "intake target" — an optimal point in this space — and regulate their feeding
to approach it.
### how it applies to ant colonies
- colonies have separate appetites for protein and carbohydrate, enabling them
to compensate for changes in nutrient density and to select among
nutritionally complementary foods.
- **workers need carbohydrates** (energy for foraging, maintenance).
- **larvae need protein** (growth, development).
- this creates a fundamental tension: the colony must collect both, but the
ratio shifts with brood load.
### colony-level regulation
- in Monomorium pharaonis (pharaoh's ant), colonies defended a slightly
carbohydrate-biased intake target.
- when confined to imbalanced protein:carbohydrate (P:C) diets, colonies used a
"generalist equal-distance strategy": overharvesting BOTH protein and
carbohydrate to reach the target ratio, rather than prioritizing one.
- ants regulate macronutrient intake at both individual and colony levels,
maintaining their specific elemental body composition.
### what happens when the balance is off
- when carbohydrate-supplemented, fire ant colonies consumed less cricket and
specifically avoided high-lipid ovaries.
- when amino-acid-supplemented, they consumed less male cricket (lower lipid,
higher protein).
- this demonstrates independent regulation of at least protein, carbohydrate,
and lipid.
**confidence: high** — the geometric framework is well-validated across multiple
ant species and other social insects.
## 4. food quality and recruitment behavior
### the pharaoh's ant baseline (Monomorium pharaonis)
- trail-marking ants deposited significantly more pheromone when returning from
high-quality food (1.0 M sucrose) vs low-quality food (0.01 M sucrose).
- at low food quality, there was no significant difference in marking intensity
between fed and unfed trail-marking ants — the quality signal disappeared.
### Lasius niger (black garden ant)
- deposits up to 22x more pheromone within 10 cm of a food source compared to
near the nest.
- uses an all-or-nothing individual response to food quality (binary: mark or
don't mark), which contrasts with Pharaoh's ant graded response.
- L. niger is proficient at visual-based orientation, so it's less reliant on
pheromone trails than pharaoh's ants.
- the presence of existing pheromone trails does NOT influence an individual
ant's subjective reward evaluation — they assess food quality independently.
### general principles across species
- the more rewarding a food source, the higher the pheromone concentration on
the trail.
- some species use multiple pheromones: a long-lasting exploration pheromone
(weak recruitment) and a shorter-lasting exploitation pheromone (strong
recruitment). the exploitation pheromone is deposited preferentially for
high-quality food.
### tandem running (Temnothorax spp.)
- tandem running is a one-to-one recruitment method where a leader guides a
follower from nest to food.
- tandem running is favored when food sources are hard to find, differ in
energetic value, and are long-lasting.
- colonies can adaptively allocate foragers across sources of different quality
using tandem running.
- followers learn specific routes from leaders — 90% of tandem leaders guided
followers along routes they had originally learned as followers themselves.
- gene expression: learning and memory genes are specifically upregulated in
scouts and tandem-followers.
### response to environmental change
- ants strongly upregulate pheromone deposition immediately after experiencing
an environmental change (e.g., food source moves or changes quality).
- VULNERABILITY: pheromone-based positive feedback can trap colonies at local
optima. if a poor feeder is established first, the pheromone trail can
outcompete incipient trails to a better source added later. this is a known
failure mode of stigmergic systems.
**confidence: high** — extensive experimental literature across species.
## 5. food source evaluation and decision-making
### comparing multiple food sources
- ants integrate food quality with foraging cost (distance, danger).
- the marginal value theorem (MVT) predicts: leave a food patch when the
current rate of energy gain drops to the average expected rate for the habitat.
- in practice: ants stay longer at patches that are farther apart or when
current patches are poor (both increase travel-cost-to-benefit ratio).
### distance vs quality tradeoff
- closer low-quality food vs farther high-quality food: ants can get trapped at
the closer source due to pheromone positive feedback (see vulnerability above).
- learning speed differences help: ants learn routes to higher-quality food
faster, which can partially overcome the distance disadvantage.
- small differences in learning speed for different food qualities can drive
efficient collective foraging at the colony level.
### memory and reassessment
- ants DO remember food source locations and quality.
- assessment updates over multiple visits — initial visits weight "amount of
food" heavily, later visits weight "reliability" more.
- private information (individual memory) can sometimes trap colonies at local
optima, independent of pheromone effects.
- ants do NOT appear to actively correct erroneous pheromone trails — trails
decay naturally rather than being "erased."
**confidence: high** for behavioral patterns. the MVT application to ants is
well-supported theoretically but ants don't perfectly optimize — they use
heuristics that approximate MVT predictions.
## 6. trophallaxis and food quality communication
### what gets transferred
trophallactic fluid in Camponotus floridanus contains far more than just food:
- **nutrients** — sugars, amino acids, lipids
- **proteins** — both digestion-related and non-digestion-related. many are
regulators of growth, development, and behavioral maturation.
- **juvenile hormone III (JH)** — a key developmental regulator. when workers'
food was supplemented with JH, larvae they reared via trophallaxis were TWICE
as likely to complete metamorphosis and became larger workers.
- **JH esterase paralogs** — enzymes that break down JH, providing a
regulatory counterbalance.
- **cuticular hydrocarbons (CHCs)** — nestmate recognition cues.
- **small RNAs (microRNAs)** — potential gene expression regulators.
### quality information transfer
- food receivers perceive the odor of food delivered by the donor and associate
it with the food reward.
- through individual experience, workers evaluate the characteristic information
of food and assess its quality.
- social information can OVERRIDE individual assessment: carpenter ants
receiving social instructions will consume food they would otherwise reject,
even toxic food, despite noxious effects. social instruction overrides
individual evaluation.
### the crop as a social stomach
- the crop (foregut) stores liquids separately from the midgut.
- food intended for sharing is kept available for trophallaxis without being
fully digested.
- this allows ants to act as mobile food storage and distribution units.
**confidence: high** — proteomic and molecular analysis of trophallactic fluid
is well-established, particularly in Camponotus floridanus.
## 7. species-specific food quality concerns
### fire ants (Solenopsis invicta)
- omnivorous, regulate protein/carb/lipid independently.
- prefer sucrose and leucine, with preference intensifying at higher
concentrations.
- prefer single-component solutions over multi-component mixtures.
- larvae display independent appetites for solid protein, amino acid solution,
and sucrose solution.
- when infected with SINV-1 virus: reduced foraging, declined lipid intake,
shifted preference toward carbohydrate-rich foods.
### leaf-cutter ants (Atta, Acromyrmex)
food quality is evaluated at TWO levels: for the ant AND for the fungal
cultivar.
- **leaf selection criteria**: plant chemistry, nutrient content, tenderness,
vein thickness, trichome density, endophyte load.
- prefer young leaves with soft cuticles, fewer defenses, higher nutritional
value.
- **fungal feedback**: ants detect chemical signals from the fungus. if a leaf
type is toxic to the fungus, the colony stops collecting it. this is a
learned colony-level response.
- fungus gardens preferentially break down simpler, more digestible substrates
first.
- the fungus produces specific enzyme profiles in response to different plant
substrates (different protein expression for different leaves).
- foraged material (fruits, flowers, leaves) is combined to maximize cultivar
performance — a multidimensional nutritional optimization.
- trace mineral management: concentrations of toxic trace minerals (Cu, Mn, Zn)
in foraged leaves peak near the macronutrient intake target, suggesting
active regulation of micronutrient toxicity.
### harvester ants (Pogonomyrmex spp.)
- seed specialists. selection based on multiple factors:
- caloric reward (energy density)
- seed size (prefer 3-30 mg in P. rugosus)
- protein and energy content (P. salinus preferentially selects
Lepidium papilliferum seeds for their higher protein/energy content)
- handling time and travel cost
- individual foraging choices are labile — converge on the most energetically
profitable species over time.
- colony dietary history influences individual seed preferences.
### honeypot ants (Myrmecocystus spp.)
- specialized repletes (living storage vessels) store liquid carbohydrates.
- foragers collect sweet exudates from cynipid galls and regurgitate to repletes
via trophallaxis.
- replete honey composition: primarily glucose and fructose, ~67g sugar per
100g, pH 3.85.
- the crop keeps food separated from the midgut so it remains available for
redistribution without being digested.
- when food is scarce, the process reverses: repletes regurgitate stored sugar
to feed the colony.
### Temnothorax spp. (acorn/rock ants)
- use tandem running rather than mass recruitment.
- stored excess food is sufficient for reducing protein foraging — colonies with
food reserves are less responsive to protein opportunities.
- small colonies respond more strongly to larval demand signals than large ones.
**confidence: high for fire ants and leaf-cutters** (extensively studied).
moderate for harvester ants (good behavioral data, less molecular detail).
moderate for honeypot ants (descriptive natural history is solid, mechanistic
data is thinner).
## 8. larval nutritional signaling
### the demand chain
nutritional information flows through a hierarchical demand chain:
larvae → nurses → foragers → environment
1. **larvae signal hunger** via non-volatile contact pheromones and begging
behaviors (physical solicitation).
2. **nurses respond** by soliciting food from foragers or from stored reserves.
3. **foragers adjust** foraging effort and target macronutrient composition based
on upstream demand.
### larval pheromone effects (dose-dependent)
- workers in colonies WITH larvae increase foraging activity compared to
broodless colonies.
- workers suppress ovarian activation in the presence of larvae (progressive
effect — stronger in smaller colonies).
- the response is dose-dependent: more larvae = stronger foraging drive +
stronger ovarian suppression.
### specificity of demand
- the chain is nutrient-specific. ants can match foraging to deficiencies in
single amino acids, suggesting the demand signal carries information about
WHAT is needed, not just "feed me."
- workers shift foraging toward protein-rich sources when larval demand is high
(more brood = more protein foraging).
- in Temnothorax longispinosus, stored excess food alone is sufficient to reduce
protein foraging — the colony tracks its reserves.
### information propagation in large colonies
- in small colonies, direct larva-worker contact may suffice.
- in larger colonies, larval pheromones may propagate through the trophallaxis
network: transferred from nurse to forager via oral fluid exchange, carrying
the chemical signal deeper into the nest.
- this is less well-characterized mechanistically than the direct-contact
pathway.
**confidence: high** for the existence and behavioral effects of the demand
chain. moderate for the specific molecular identity of larval hunger pheromones.
the propagation mechanism in large colonies is plausible but not definitively
demonstrated.
## simulation implications
key parameters to model for food quality in the simulation:
1. **food quality value (0-255, already allocated in world.R bits 6-13)**
- maps to sugar concentration / caloric density
- affects pheromone deposition intensity (graded or binary per species model)
- affects recruitment strength
2. **colony nutritional state**
- protein vs carbohydrate balance (two-axis model from geometric framework)
- brood load shifts target toward protein
- deficit in either axis biases forager preferences
3. **individual forager memory**
- food source location + quality rating
- updates over visits (amount → reliability weighting shift)
- learning speed proportional to food quality
4. **trophallaxis network effects**
- quality information propagates through social feeding
- social information can override individual assessment
- larval demand propagates upstream through nurses to foragers
5. **pheromone trail modulation**
- trail intensity proportional to food quality (above some threshold)
- dual-pheromone option: exploration (weak, long-lived) vs exploitation
(strong, short-lived)
- trap risk: established trails to poor sources resist switching
6. **distance-quality tradeoff**
- not a simple linear comparison — pheromone feedback creates path dependence
- closer poor food can dominate farther good food due to positive feedback
- learning speed differences partially compensate
## sources
- [Preference and effect of gustatory sense on sugar-feeding of fire ants](https://peerj.com/articles/11943/)
- [A fructose-sensitive gustatory receptor in fire ants](https://www.sciencedirect.com/science/article/abs/pii/S0965174825001845)
- [Detection of sweet tastants by insect gustatory receptors](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3910600/)
- [Ant foragers compensate for nutritional deficiencies in the colony](https://www.cell.com/current-biology/fulltext/S0960-9822(19)31458-7)
- [Flexible, but not enough: how an omnivorous ant copes with macronutrient imbalances](https://nsojournals.onlinelibrary.wiley.com/doi/abs/10.1002/oik.11557)
- [Nutritional geometry of Monomorium pharaonis](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0218764)
- [Carbohydrate regulation in relation to colony growth](https://journals.biologists.com/jeb/article/211/14/2224/17598/Carbohydrate-regulation-in-relation-to-colony)
- [Ant nutritional ecology: linking nutritional niche plasticity](https://www.sciencedirect.com/science/article/abs/pii/S221457451400090X)
- [Modulation of pheromone trail strength with food quality in pharaoh's ant](https://www.sciencedirect.com/science/article/abs/pii/S0003347207002278)
- [Lasius niger pheromone deposition near food sources](https://link.springer.com/article/10.1007/s00040-024-00995-y)
- [Trail pheromone does not modulate subjective reward evaluation in L. niger](https://pmc.ncbi.nlm.nih.gov/articles/PMC7540218/)
- [The role of multiple pheromones in food recruitment](https://journals.biologists.com/jeb/article/212/15/2337/18424/The-role-of-multiple-pheromones-in-food)
- [Trail pheromones of ants (review)](https://resjournals.onlinelibrary.wiley.com/doi/10.1111/j.1365-3032.2008.00658.x)
- [Food information acquired socially overrides individual assessment](https://link.springer.com/article/10.1007/s00265-016-2216-x)
- [Social transmission of information via trophallaxis](https://royalsocietypublishing.org/doi/10.1098/rspb.2017.1367)
- [Oral transfer of chemical cues, growth proteins and hormones in social insects](https://elifesciences.org/articles/20375)
- [Molecular evolution of JH esterase-like proteins in trophallactic fluid](https://www.nature.com/articles/s41598-018-36048-1)
- [Preferences for sugars and amino acids in nectar-feeding ants](https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/j.1365-2656.2004.00789.x)
- [Effects of macro- and micro-nutrients on feeding responses by ants](https://www.nature.com/articles/s41598-024-56133-y)
- [Dietary diversity, sociality, and the evolution of ant gustation](https://www.frontiersin.org/journals/ecology-and-evolution/articles/10.3389/fevo.2023.1175719/full)
- [Feeding preferences for sugars and amino acids in fire ants](https://www.mdpi.com/2075-4450/17/3/258)
- [Regulation of diet in the fire ant](https://link.springer.com/article/10.1023/A:1020835304713)
- [Route learning during tandem running in Temnothorax](https://journals.biologists.com/jeb/article/223/9/jeb221408/223803/Route-learning-during-tandem-running-in-the-rock)
- [Teaching in tandem-running ants](https://pubmed.ncbi.nlm.nih.gov/16407943/)
- [Tandem-running and scouting characterized by learning/memory gene upregulation](https://pubmed.ncbi.nlm.nih.gov/30903719/)
- [Temnothorax adjusts tandem running when distance exposes them to greater risks](https://link.springer.com/article/10.1007/s00265-018-2453-2)
- [Ant larvae regulate worker foraging behavior and ovarian activity dose-dependently](https://pmc.ncbi.nlm.nih.gov/articles/PMC5015688/)
- [Stored excess food reduces protein foraging in Temnothorax](https://link.springer.com/article/10.1007/s00265-025-03683-4)
- [Foraging and feeding independently regulated in clonal raider ant](https://link.springer.com/article/10.1007/s00265-021-02985-7)
- [Ants adjust pheromone deposition to changing environment](https://pmc.ncbi.nlm.nih.gov/articles/PMC4590477/)
- [Private information can trap colonies in local feeding optima](https://journals.biologists.com/jeb/article/219/5/744/16615/Private-information-alone-can-trigger-trapping-of)
- [Small differences in learning speed drive efficient collective foraging](https://link.springer.com/article/10.1007/s00265-018-2583-6)
- [Re-visiting plentiful food sources in desert ants](https://pmc.ncbi.nlm.nih.gov/articles/PMC3389614/)
- [Fungal cultivar of leaf-cutters produces specific enzymes per plant substrate](https://pmc.ncbi.nlm.nih.gov/articles/PMC5118115/)
- [Multidimensional nutritional niche of leaf-cutter fungus provisioning](https://pmc.ncbi.nlm.nih.gov/articles/PMC9292433/)
- [Evolutionary innovation of nutritional symbioses in leaf-cutters](https://pmc.ncbi.nlm.nih.gov/articles/PMC4553616/)
- [Seed selection by Pogonomyrmex rugosus](https://pubmed.ncbi.nlm.nih.gov/27257121/)
- [Flexible seed selection by Pogonomyrmex occidentalis](https://link.springer.com/article/10.1007/BF00164118)
- [Honeypot ant (Myrmecocystus) — Wikipedia](https://en.wikipedia.org/wiki/Honeypot_ant)
- [Evolutionary origins of repletism in ants](https://blog.myrmecologicalnews.org/2023/05/31/shedding-light-on-the-evolutionary-origins-of-repletism-in-ants/)

407
docs/PHEROMONES.md Normal file
View file

@ -0,0 +1,407 @@
# Ant Pheromone Biology
Reference doc on ant pheromone types, chemistry, triggers, and environmental
factors. Organized for simulation design — what matters for modeling, what
the real biology says, and where the data is thin.
## Pheromone types
### Trail pheromones
Guide nestmates to food, new nest sites, or other resources. Encode distance
and quality information through concentration gradients.
Gland sources vary by subfamily:
- Poison gland (most Myrmicinae — stinging ants)
- Dufour's gland (most Formicinae — non-venomous ants)
- Pygidial gland, sternal glands, hindgut (various)
Known compounds by species:
species compound gland
Atta texana, A. cephalotes methyl 4-methylpyrrole-2-carboxylate poison
Atta sexdens rubropilosa 3-ethyl-2,5-dimethylpyrazine poison
Tetramorium caespitum 70:30 2,5-dimethylpyrazine + EDMP poison
Tetramorium meridionale indole (major) + 4 pyrazines poison
Monomorium pharaonis (E,E)-faranal poison
Myrmica spp. EDMP + homofarnesenes poison + Dufour's
Solenopsis invicta various poison
Volatility is a feature — trails evaporate in minutes to hours, so paths to
depleted food naturally fade. The evaporation rate is effectively the colony's
spatial memory duration.
### Alarm pheromones
Fast-acting signals that trigger either panic (flee) or aggression (attack)
depending on species, colony size, and context.
Triggers:
- Physical disturbance of nest or individual
- Predator detection (olfactory, visual, or tactile)
- Crushing/injury of nestmates (damaged ants release alarm compounds)
- Intrusion by non-nestmates
Known compounds:
compound species gland
4-methyl-3-heptanone Atta, Pogonomyrmex, Ooceraea mandibular
formic acid Formica, Camponotus poison
n-undecane Camponotus, many Formicinae Dufour's
2-heptanone Iridomyrmex pruinosus mandibular
3-octanol Acromyrmex echinatior mandibular
3-octanone Acromyrmex octospinosus mandibular
(S)-(-)-citronellal Platythyrea punctata mandibular
(S)-(-)-actinidine Platythyrea punctata mandibular
Behavioral response sequence: stop movement -> swing antennae -> alerted
posture -> species-specific response (flee or attack). Larger species and
larger colonies tend toward aggression; smaller species tend toward evacuation.
#### Carpenter ant alarm system (formic acid + n-undecane)
Two-component blend from two different glands:
- Formic acid (poison gland) -> avoidance behavior, move away from source
- n-Undecane (Dufour's gland) -> attraction toward source, approach to attack
Each component alone produces a distinct response. Together they create the
full alarm sequence: alert, then approach and aggress. Formic acid also
modulates nestmate recognition — puts ants into a heightened discriminatory
state where they're more likely to attack non-nestmates.
Processed in ~5 specific "alarm-sensitive" glomeruli clustered in the
dorsalmost part of the antennal lobe, relayed to the lateral horn.
#### What triggers alarm? Plants? Predators?
No strong evidence for specific plants that trigger false alarm responses,
though some alarm compounds (citronellal) are also found in plant oils, so
cross-reactivity is plausible. Some alarm compounds (citral, 2-heptanone,
4-methyl-3-heptanone) have antifungal properties, suggesting a dual role in
pathogen defense — the alarm system may have originally evolved as an
antimicrobial response.
Main natural triggers are physical: nest disturbance, predator contact, and
injured nestmates releasing their contents. The "danger zone" concept maps
best to areas where nestmates have been injured or where persistent threats
exist, not to specific environmental chemicals.
### Repellent / negative pheromone
Deposited at trail junctions leading to unrewarding branches. Prevents the
colony from getting stuck on depleted food. Key details from Pharaoh's ant
studies:
- Lasts ~2x longer than attractive trail pheromone (~78 min vs ~33 min)
- Ants encountering it U-turn or zigzag
- Deposited specifically at bifurcation points, not along entire failed paths
Source: Nature 438, 442 (2005)
### Queen pheromones
Complex, multifunctional. Known effects:
- Attract workers for queen attendance
- Promote brood care
- Induce nestmate discrimination
- Inhibit larval sexual development (through worker behaviors)
- Suppress reproduction in other queens and workers
- Induce worker policing of reproductive workers
- Mediate queen acceptance/rejection in polygyne colonies
Chemistry is poorly characterized compared to trail/alarm pheromones. Involves
cuticular hydrocarbons and at least one dedicated odorant receptor (HsOr263 in
Harpegnathos saltator). Difficult to isolate because the signal may be a
complex blend rather than a single compound.
### Necrophoresis / death pheromones
Trigger corpse removal for nest hygiene. Dual-signal system:
1. Loss of "life signals": living ants produce dolichodial and iridomyrmecin
on their cuticle. These disappear within ~60 minutes of death. Their absence
is the initial trigger.
2. Gain of "death signals": oleic acid and linoleic acid accumulate as the
corpse decomposes. These are the classical necrophoresis triggers.
Timing: only 15% of freshly killed corpses get removed. Rises to 80% for
corpses 1-6 days post-mortem as fatty acids accumulate. The colony doesn't
panic-clean — it waits for chemical confirmation.
### Propaganda pheromones
Used by slave-making (dulotic) ants during raids. Polyergus queens enter
Formica host nests, kill the resident queen, acquire her chemical profile,
and release substances from an enlarged Dufour's gland that reduce worker
aggression. Raiding workers release:
- Manipulative alarm signals (cause panic, disorganized defense)
- Chemical weapons (directly repellent)
- Appeasement substances (reduce aggression toward the raider)
Stolen brood are chemically imprinted after eclosion, causing them to identify
the slave-maker colony as home.
### Cuticular hydrocarbons (CHCs) — colony recognition
Not classical volatile pheromones but contact chemicals on the cuticle. Each
colony has a distinctive CHC profile — a blend of dozens of hydrocarbons that
workers learn and use to distinguish nestmates from non-nestmates. Non-nestmates
are attacked. The postpharyngeal gland stores and distributes CHCs, homogenizing
the colony odor through trophallaxis (food sharing).
### Recruitment pheromones
Functionally distinct from trail pheromones in some species. In Leptogenys
diminuta, two gland sources serve different roles:
- Poison gland secretions: orientation cues (where to go)
- Pygidial gland secretions: recruitment stimulus (leave the nest and follow)
This separation means "follow this path" and "come help" are independent
signals that can be modulated separately.
### Brood pheromones
Signals from eggs, larvae, and pupae that attract worker care. Help workers
assess developmental stage and nutritional needs. Part of the demand chain
that links larval nutritional needs -> nurse behavior -> forager food
preferences. Less well-characterized than adult pheromones.
### Sex / mating pheromones
Released by virgin queens (gynes) to attract males during nuptial flights.
Less studied than other types. Identified in Polyergus breviceps.
## Food quality
### What "quality" means to ants
Primarily macronutrient content — the protein-to-carbohydrate (P:C) ratio.
Different colony members need different things:
- Workers need carbohydrates (energy for foraging, maintenance)
- Larvae and queens need protein (growth, egg production)
- Foragers collect for the colony's current needs, relayed through a demand
chain: larvae -> nurses -> foragers
Sugar concentration is the primary quality axis for foraging workers. In the
key Pharaoh's ant study, 1.0 M sucrose = "high quality" and 0.01 M sucrose =
"low quality."
Species-specific sugar preferences exist — carpenter ants (Camponotus modoc)
and European fire ants (Myrmica rubra) show selective preferences for specific
mono-, di-, and trisaccharides. Not all sugars are equal.
### How ants assess quality
Two mechanisms:
1. Contact chemoreception — tasting with mouthparts and antennae
2. Post-ingestive feedback — evaluating nutritional content after consumption
Colony-level nutritional regulation follows a "geometric framework" model:
colonies actively balance P:C intake, shifting forager preferences based on
current deficits. A protein-starved colony will recruit more aggressively
to protein sources.
### How quality modulates pheromone deposition
The Pharaoh's ant study (Jackson, Holcombe & Ratnieks 2004/2007):
- Trail marking occurs at ~40% frequency among both fed and unfed foragers
- High quality food (1.0 M sucrose) -> significantly more high-intensity
continuous marking
- Low quality food (0.01 M sucrose) -> no significant difference in marking
intensity between fed and unfed ants
- This is a graded individual response — ants modulate marking intensity,
not all-or-nothing
Contrast with Lasius niger, where trail strength modulation IS all-or-nothing
at the individual level — an ant either marks or doesn't, based on quality.
The difference: Pharaoh's ants live in dark enclosed spaces and rely heavily
on pheromone trails, so they must always produce some trail. L. niger can use
visual orientation and afford to skip trail-laying entirely for poor food.
Ants also adjust deposition based on environmental uncertainty — they modulate
trail marking in response to changing conditions and their error probability
(Czaczkes et al. 2015).
### Simulation relevance
For the sim, food quality maps to a 0-255 value stored in cell metadata bits.
When an ant picks up food, it reads the quality and stores it as cargoQuality.
The deposition multiplier would scale pheromone output — high quality food
gets stronger trails, attracting more ants. The colony naturally converges on
the best source first, then shifts when it depletes.
"Different types of food" is less important than "different concentrations."
Real ants care about sugar molarity and protein content, not food identity.
For the sim, a single quality scalar captures the essential dynamics.
## Pheromone detection
### Olfactory receptor genes
Ants have 300-500 odorant receptor (Or) genes — 4-5x more than most insects
(Drosophila has ~60). This expansion predates complex sociality but facilitated
it. A specialized 9-exon Or gene subfamily detects cuticular hydrocarbons and
candidate pheromones.
### Antennal lobe glomeruli
Each Or gene roughly corresponds to one glomerulus in the antennal lobe.
species glomeruli count
Apterostigma cf. mayri ~630
Camponotus (carpenter) ~430-460
Apis mellifera (honeybee) ~163
Drosophila melanogaster ~43
Workers have more glomeruli than males, reflecting greater need for chemical
discrimination. Worker and queen antennal lobes differ in composition and
Or expression.
## Environmental effects on pheromones
### Temperature
High temperatures accelerate degradation. Above ~40C, workers cannot
discriminate previously-marked substrate. Above ~30C, foraging activity drops
partly because trails decay too quickly to be useful.
Different compounds have different thermal stability: in Tapinoma nigerrimum,
most gaster secretions vanished at 25C, but iridodials persisted up to 55C.
Aphaenogaster senilis secretions resisted elevated temperatures better.
Pheromone persistence = f(temperature, time since deposition). Both interact —
higher temperature accelerates decay at all time points.
### Humidity
Higher humidity slows evaporation of polar pheromone components. Specific
experimental data is sparser than for temperature. The effect is real but
less dramatic than temperature in most contexts.
### Substrate surface
Porous surfaces absorb pheromone and release it slowly (longer persistence).
Smooth impermeable surfaces allow faster evaporation (shorter persistence).
Direct experimental data on ants is thin — the Pharaoh's ant study
(~9 min on plastic, ~3 min on paper) is one of the few quantitative
comparisons. The physics is well understood but species-specific measurements
are rare.
### Volatility as design feature
Trail pheromone volatility provides automatic negative feedback — trails to
depleted food fade without any active erasure. The evaporation rate sets the
colony's spatial memory duration. Too fast = no useful trails. Too slow =
colony gets stuck on depleted paths. Evolution tuned this balance per species
and habitat.
## Colony-level dynamics
### Colony size effects
Larger colonies have higher collective response thresholds. The relationship
is driven by social feedback — short-range excitatory and long-range
inhibitory interactions. In army ants, increasing colony size causes a
qualitative behavioral shift: organized search patterns in small colonies
give way to spontaneous mass raids in large ones.
Per-capita interaction rate is roughly scale-invariant — connectivity scales
hypometrically with colony size.
### Colony state modulates pheromone dynamics
- Protein-starved colonies shift forager preferences toward protein, changing
trail deposition patterns (more intense marking to protein sources)
- Carbohydrate-rich diets increase social immunity
- Weakly volatile "aggregation" pheromones mark the nest site as a constant
baseline signal anchoring the colony spatially
### "Colony mood"
Not a scientific term, but maps onto real dynamics:
- Alarm state propagates as more individuals detect alarm compounds and
release their own (positive feedback cascade)
- Foraging motivation modulated by colony nutritional state — hungry colonies
produce stronger recruitment signals
- The exploration/exploitation balance shifts with colony experience and
environmental conditions
## Sources
Pharaoh's ant pheromone modulation
Jackson, Holcombe & Ratnieks 2004/2007
ScienceDirect S0003347207002278
Negative pheromone
Nature 438, 442 (2005)
Alarm pheromone processing (carpenter ants)
PMC2912167
Alarm pheromone in clonal raider ant
PMC9941220
Formic acid + nestmate recognition
J Exp Biol 224(20) jeb242784
Necrophoresis dual signals
PNAS 0901270106
Corpse chemical changes
J Chem Ecol 10.1007/s10886-013-0365-1
Slave-maker chemical warfare
PLOS ONE 10.1371/journal.pone.0147498
Queen pheromone properties
Behav Ecol Sociobiol 10.1007/s00265-023-03378-8
Trail pheromone review
Morgan 2009, Physiological Entomology
10.1111/j.1365-3032.2008.00658.x
Trail pheromone compound list
Natural Product Communications 10.1177/1934578X1400900813
Odorant receptors in social insects
Nature Communications 10.1038/s41467-017-00099-1
Pheromone-sensitive glomeruli
Proc R Soc B 10.1098/rspb.2006.3565
Olfactory system review
PMC8002415
Ant olfactory receptor expansion
Vanderbilt (2012)
Temperature effects on trail following
J Chem Ecol 10.1007/s10886-012-0130-x
Colony interaction scaling
Frontiers in Ecology and Evolution 10.3389/fevo.2022.993627
Collective sensory threshold
PNAS 10.1073/pnas.2123076119
Ant nutritional ecology
ScienceDirect S221457451400090X
Nutrient regulation
Myrmecological News (Csata & Dussutour 2019)
Exocrine glands of ants
Chemoecology 10.1007/BF01256548
Sugar preferences
PMC8371376
Pheromone deposition under uncertainty
Czaczkes et al. 2015, PMC4590477

227
docs/TERRAIN-AND-DECAY.md Normal file
View file

@ -0,0 +1,227 @@
# terrain, substrates, and pheromone decay
how environmental factors affect pheromone persistence. relevant to feature #7
(substrate-dependent decay) and the worldBlur shader's per-cell decay rate.
## substrate effects on persistence
the key study is Jeanson et al. (2003) on Monomorium pharaonis:
substrate chemical half-life behavioral preference half-life
plastic ~9 min ~25 min
paper ~3 min ~8 min
a 3x difference in chemical half-life between two smooth artificial surfaces.
mechanism: paper is porous and wicks the compound away from the surface,
reducing the airborne concentration ants detect. plastic is non-porous, so the
compound sits on top and remains available.
no comparable controlled study exists for natural substrates (soil, rock, sand,
leaf litter, wood). inferred from physical chemistry:
- porous substrates (soil, sand, wood, leaf litter) behave more like paper —
absorb compounds, accelerate apparent decay
- non-porous substrates (rock, packed clay) behave more like plastic — keep
compounds on the surface, slower decay
- soil moisture complicates things further (see humidity section)
## species variation in baseline trail longevity
trail pheromone persistence varies enormously across species:
species trail longevity notes
Solenopsis invicta ~100 sec / <2 min extremely volatile compounds
Monomorium pharaonis ~9 min (plastic) multiple pheromone types
Aphaenogaster albisetosus minutes short-lived
Pachycondyla sennaarensis ~30 min to half gone in 1 hr
Monomorium spp. (general) ~1 day optimal varies
Camponotus (carpenter ants) days hindgut-produced
Daceton armigerum 7+ days poison gland secretion
Eciton spp. (army ants) weeks long-chain, low-volatility
the range is >100x. species in stable environments with permanent food sources
use long-lasting compounds. species exploiting ephemeral food use volatile ones.
Pharaoh's ants also use multiple pheromone types with different decay profiles:
a long-lasting attractive pheromone, a short-lived attractive pheromone, and a
short-lived repellent pheromone (~78 min vs ~33 min half-life).
## temperature
the definitive paper is van Oudenhove et al. (2011/2012), studying
Tapinoma nigerrimum and Aphaenogaster senilis.
key findings:
- above ~40C: workers cannot discriminate marked substrate — pheromone is
effectively destroyed
- above ~30C: foraging activity drops independently, partly because trails
decay too quickly to be useful
- between 25-40C: decay accelerates but trails remain functional
- the 40C threshold is a behavioral cliff, not a smooth curve
species differ in thermal resilience:
- T. nigerrimum (mass-recruiting): secretions highly volatile. most compounds
vanished even at 25C. only iridodials persisted up to 55C.
- A. senilis (group-recruiting): secretions less volatile, resisted elevated
temperatures better. at 55C, only nonadecene and nonadecane (long-chain
hydrocarbons) persisted.
pheromone persistence = f(temperature, time since deposition). both interact —
higher temperature accelerates decay at all time points.
diurnal implications: hot midday temperatures degrade trails laid in morning.
desert species tend to use less volatile compounds (evolutionary compensation
via chemistry rather than behavioral compensation via deposition rate).
## humidity and moisture
less quantitative data than temperature.
- higher humidity slows evaporation of polar pheromone components
- a cuticle covered with water may hinder both reception and emission of
pheromones — wet conditions impair laying AND sensing, not just persistence
- army ants (Eciton burchellii) increased speed by 30% in response to increased
humidity and rain sounds near the trail, but watering the trail directly did
not cause load-dropping
- extreme humidity (either direction) suppresses foraging entirely
wet porous substrates (damp soil, wet leaf litter) would absorb pheromone
faster than dry porous substrates, but no controlled study confirms this
quantitatively.
no direct evidence of ants selecting substrates specifically for pheromone
persistence, though trail-clearing behavior (see below) effectively creates
favorable substrate.
## physical trail infrastructure
### leaf-cutter ant highways (Atta spp.)
the standout example of ants modifying their environment for trail quality:
- colonies clear an average of 2,730 meters of trail per year
- individual trails can exceed 200 meters
- networks extend for kilometers cumulatively
- construction/maintenance costs: ~11,000 ant-hours per year
what they do: remove leaf litter, cut passes through overhanging vegetation,
shift soil to level surfaces. selective clearing — flat objects are ignored,
upright/folded obstructions are removed.
coordination: trail clearing happens WITHOUT information exchange between
workers. independent effort that adds up to emergent infrastructure. clearing
is triggered by freshly laid pheromone on an obstructed path.
### minim workers as trail maintainers
the smallest workers (minims) are always present on trails but never carry
leaves. they deposit pheromone at 83.3% frequency vs 20% for non-minims.
they're dedicated trail maintainers — keeping the chemical signal strong while
larger workers (2.2-2.9mm head width) handle physical clearing.
### physical + chemical reinforcement loop
physical clearing creates smooth packed soil (relatively non-porous) which
retains pheromone better than leaf litter. minim workers then maintain high
pheromone concentration. cleared trails retain pheromone better -> strong
pheromone attracts more traffic -> more traffic means more clearing and
reinforcement. positive feedback on two axes simultaneously.
energetics: not always profitable. depends on workforce composition and patrol
vs carry ratio. can amortize within days or take weeks/months.
## emergent highway formation
the trail network that emerges from substrate-dependent persistence:
1. substrate quality: non-porous > porous
2. temperature: shade > sun
3. physical modification: cleared > uncleared
4. traffic: popular > unpopular (reinforcement)
5. food quality: rich source > depleted source
a trail across cool, shaded, packed earth near a rich food source dominates
over a trail across hot, sun-exposed leaf litter near a marginal source. no
ant "decides" this — the pheromone math works out.
trail bifurcation: at branch points, trail asymmetry (angle, width) influences
decisions alongside pheromone presence. neither geometry nor pheromone alone
dominates — non-hierarchical interaction.
rapid decay as feature: in fire ants, trail pheromone drops below detection in
~2 minutes. this forces continuous reinforcement, meaning only actively
profitable routes persist. fast decay = responsive colony.
## simulation relevance
for the world texture's terrain type bits (3-5 in world.R):
terrain type decay multiplier real-world analog
0 (default) 1.0x generic surface
1 0.5x (slower) packed earth / rock
2 1.5x (faster) leaf litter / porous
3 2.0x (faster) sand / loose soil
4-7 reserved future use
the blur shader would read terrain type per cell and multiply the base decay
rate. ants wouldn't "know" about terrain — they'd just find that their trails
last longer on some surfaces, and positive feedback would do the rest.
temperature could be a global uniform rather than per-cell (simpler), or
per-cell if the simulation adds sun/shade regions.
## sources
Jeanson et al. 2003
Pheromone trail decay rates on different substrates in Pharaoh's ant
Physiological Entomology 10.1046/j.1365-3032.2003.00332.x
van Oudenhove et al. 2011
Temperature limits trail following through pheromone decay
Naturwissenschaften 10.1007/s00114-011-0852-6
van Oudenhove et al. 2012
Substrate temperature constrains recruitment and trail following
J Chem Ecol 10.1007/s10886-012-0130-x
Bruce et al. 2019
Infrastructure construction without information exchange in Atta
Proc R Soc B 10.1098/rspb.2018.2539
Bruce et al. 2017
Energetics of trail clearing in Atta
Behav Ecol Sociobiol 10.1007/s00265-016-2237-5
Robinson et al. 2008
Decay rates of attractive and repellent pheromones in foraging trail network
Insectes Sociaux 10.1007/s00040-008-0994-5
Morgan 2009
Trail pheromones of ants (review)
Physiological Entomology 10.1111/j.1365-3032.2008.00658.x
Effect of trail pheromones and weather on Eciton burchellii
ResearchGate 225346583
Minor workers maintain leafcutter ant pheromone trails
ResearchGate 248591651
Trail pheromone of Pachycondyla sennaarensis
PMC3281317
Monomorium trail pheromone longevity
ScienceDirect S1226861509001034
Uncovering the complexity of ant foraging trails
PMC3291321
Effect of trail bifurcation asymmetry and pheromone on trail choice
PMC4204274