Microbes colonizing a surface often experience colony growth dynamics characterized by an initial phase of spatial clonal expansion followed by collision between neighboring colonies to form potentially genetically heterogeneous boundaries. For species with life cycles consisting of repeated surface colonization and dispersal, these spatially-explicit “expansion-collision dynamics” generate periodic transitions between two distinct selective regimes, “expansion competition” and “boundary competition”, each one favoring a different growth strategy. We hypothesized that this dynamic could promote stable coexistence of expansion- and boundary-competition specialists by generating time-varying, negative frequency-dependent selection that insulates both types from extinction. We tested this experimentally in budding yeast by competing an exo-enzyme secreting “cooperator” strain (expansion-competition specialists) against non-secreting “defectors” (boundary-competition specialists). As predicted, we observed cooperator-defector coexistence or cooperator dominance with expansion-collision dynamics, but only defector dominance otherwise. Also as predicted, the steady-state frequency of cooperators was determined by colonization density (the average initial cell-cell distance) and cost of cooperation. Lattice-based spatial simulations give good qualitative agreement with experiments, supporting our hypothesis that expansion-collision dynamics with costly public goods production is sufficient to generate stable cooperator-defector coexistence. This mechanism may be important for maintaining public-goods cooperation-and-conflict in microbial pioneer species living on surfaces.

#### 2v9Nospatial

Simulation without spatial expansion-collision dynamics were ran with cells randomly shuffling among sites on the lattice every generation. Simulation were initiated with 2v9 dilution with initial cell-cell distance equals 3.

#### 2v8Nospatial

Simulation were run with cells randomly shuffling among sites on the lattice every generation. Simulation were initiated with 2v8 dilution with initial cell-cell distance equals 2. Note: Sheet 1 is a part of data for these simulations, Sheet 2 has entire date for this treatment and sheet 3 was used to generate the heatmap.

#### 2v7Nospatial

Simulation without expansion-collision dynamics were run with cells randomly shuffling among sites on the lattice every generation. Simulation were initiated with 2v7 dilution with initial cell-cell distance equals 1.

#### 2v4.9068Nospatial

Simulations without expansion-collision dynamics were ran with cells randomly shuffling among sites on the lattice every generation. Simulation were started with 0 initial cell-cell distance with a dilution of 2V4.9068.

#### 2v9_rareinvasion

Rare invasion simulations were ran with expansion-collision dynamics with initial cell-cell distance equals 3. Sheet 2 is the result for a single cooperator that invades a defector population. Sheet 3 is the result for a single defector that invades a defector population. Lines in Fig 6C (red for cooperator and green for defector) were drawn where the invasion probability first equals or exceeds 0.6, representing the boundary between invasion and non-invasion (a probability of 0.5 corresponds with selective neutrality) based on the yellow colored cells in the spreadsheet.

#### 2v8_rareinvasion

Rare invasion simulations were ran with expansion-collision dynamics with initial cell-cell distance equals 2. Sheet 2 is the result for a single cooperator that invades a defector population. Sheet 4 and sheet 5 are the result for a single defector that invades a defector population. Lines in Fig 6B (red for cooperator and green for defector) were drawn where the invasion probability first equals or exceeds 0.6, representing the boundary between invasion and non-invasion (a probability of 0.5 corresponds with selective neutrality) based on the yellow colored cells in the spreadsheet.

#### 2v7_rareinvasion

Rare invasion simulations were ran with expansion-collision dynamics with initial cell-cell distance equals 1. Sheet 2 is the result for a single cooperator that invades a defector population. Sheet 3 is the result for a single defector that invades a defector population. Lines in Fig. 6A (red for cooperator and green for defector) were drawn where the invasion probability first equals or exceeds 0.6, representing the boundary between invasion and non-invasion (a probability of 0.5 corresponds with selective neutrality) based on the yellow colored cells in the spreadsheet.

#### 2v9_nobenefit

Simulations were ran with expansion-collision dynamics where initial cell-cell distance equals 3 and benefit equals 0.

#### 2v8_nobenefit

Simulations were ran with expansion-collision dynamics where initial cell-cell distance equals 2 and benefit equals 0.

#### 2v7_nobenefit

Simulations were ran with expansion-collision dynamics where initial cell-cell distance equals 1 and benefit equals 0.

#### 2v4.9068_nobenefit

Simulations were ran with expansion-collision dynamics where initial cell-cell distance equals 0 and benefit equals 0.

#### 2v9_8neighbors/24neighbors

Simulations (sheet 'parameter2', 'outgrid2', 's' and 'heatmap2') were ran with expansion-collision dynamics with initial cell-cell distance equals 3 and difusivity equals 1/9. Spreadsheet 'parameter3', 'outgrid3', 's' and 'heatmap' are results for the conditions where initial cell-cell distance is 3 and difusivity equals 1/25.

2v9.xlsx

#### 2v8_8neighbors

Simulations (spreadsheet 'parameter2', 'outgrid2', 's' and 'heatmap2') were ran with expansion-collision dynamics with initial cell-cell distance equals 2 and difusivity equals 1/9.

#### 2v9_1neighbor/12neighbors/16neighbors

Simulations were ran with expansion-collision dynamics with initial cell-cell distance equals 3 and difusivity equals 1 (sheet '1_heatmap', 1/13 (spreadsheet '12_heatmap'), and 1/17 (sheet '16_heatmap').

2v9.xlsx

#### 2v7_8neighbors

Simulations were ran with expansion-collision dynamics with initial cell-cell distance equals 1 and difusivity equals 1/9.

#### 2v7_NoWeighingDispersal

Simulations were ran with expansion-collision dynamics and the initial cell-cell distance equals 1 and the assumption of weighted dispersal was relaxed.

#### 2v8_NoWeighingDispersal

Simulations were ran with expansion-collision dynamics and the initial cell-cell distance equals 2 and the assumption of weighted dispersal was relaxed.

#### 2v9_NoWeighingDispersal

Simulations were ran with expansion-collision dynamics and the initial cell-cell distance equals 3 and the assumption of weighted dispersal was relaxed.

#### cooperation1cdDVD_deterministic_evenlydistributed4

Simulation code for conditions with expansion-collision dynamics

#### cooperation1cdDVD_Noweighingdispersalv1

Simulation code for expansion-collision dynamics with the assumption of weighted dispersal relaxed.

#### cooperation1cdDVD_deterministic_evenlydistributed_rareinvasion

Simulation code for rare invasion of a single cooperator or a single defector with expansion-collision dynamics.