(A) Three-dimensional (3D) reconstruction (dorsal view) of the pallium of a her4.1:dRFP/gfap:nlsGFP double transgenic adult immunostained for GFP, dRFP, Sox2, and the proliferation marker proliferating cell nuclear antigen (Pcna) [labeling the same cells as the proliferation marker minichromosome maintenance 5 (21)]. The lateral and the medial pallial domains (Dl and Dm) are highlighted with dotted lines. Close-ups of the boxed area with different combinations of channels illustrating the four Sox2⁺ progenitor subtypes that make up the pallial germinal zone are shown: quiescent NSCs (qNSCs; blue arrow), activated NSCs (aNSCs; yellow arrow), qNPs (white arrow), and aNPs (pink arrow). (B) Distribution of gfap⁺ and her4.1⁺ NSCs, alone and in combination, among Sox2⁺ cells. (C) Respective distributions of her4.1⁺ and gfap⁺ cells among the gfap⁺ and the her4.1⁺ populations. Paired t test: P = 0.34. (D) Relative proportions of quiescent (Pcna⁻) and proliferating (Pcna⁺) gfap⁺ NSCs, her4.1⁺ NSCs, and NPs (Sox2⁺). The data were analyzed using a repeated measures mixed model. Overall tests: P < 0.001; pairwise comparisons: ***P < 0.001 after Holm’s adjustment. (E) Distribution of gfap⁺ and her4.1⁺ NSCs together with Sox2⁺ NPs among proliferating (Pcna⁺) progenitors. (F) Summary of markers characterizing Dm pallial progenitors. (B to E) n = 7 brains were analyzed. Error bars, SEM.

(A) Cre recombination was sparsely induced in 3-mpf her4.1:ERT2CreERT2;ubi:Switch double transgenic adults by 4-OHT, resulting in mCherry expression in recombined NSCs and their progeny. Analyzed time points (arrows) span between 6 and 507 dpi. (B) Dorsal view of a representative pallium showing sparsely induced cells at 6 dpi (dotted area to the pallial Dm territory of interest; fig. S1E). Boxed areas are magnified to illustrate the different cell types traced (yellow arrows). Proliferating progenitors were labeled by a 24-hour 5-bromo-2′-deoxyuridine (BrdU) pulse. NSCs are Gs⁺ and Sox2⁺, NPs are Sox2⁺ only. In contrast to aNSCs and aNPs, qNSCs and qNPs did not incorporate BrdU during the pulse. (C) Cell type composition of the mCherry⁺ population at 6 dpi. n = 6 brains. (D) Example of clustering at 183 dpi. Left: Dorsal view of a 3D reconstruction of a hemisphere. The spots registering the coordinates of traced cells (red) are shown (white dots). Right: 2D map of the highest probability clonal partition identified by the statistical inference. Axes correspond to the x and y coordinates (in μm) of the 2D projection of the cell centers. Cells belonging to the same clone are embedded in the same shaded area. (E) Average number of traced (mCherry⁺) Gs⁺/Sox2⁺ NSCs per hemisphere across all time points analyzed. One-way analysis of variance (ANOVA): F_(8,37) = 1, P = 0.45; all pairwise comparisons: least significant difference (LSD) test followed by Holm’s adjustment. Error bars, 95% confidence interval (CI). (F) Average number of traced (mCherry⁺) Sox2⁺ cells per hemisphere across all time points analyzed. One-way ANOVA: F_(8,37) = 1.65, P = 0.14; all pairwise comparisons: LSD test followed by Holm’s adjustment. Error bars, 95% CI. (E and F) n = 6, 3, 3, 3, 4, 6, 7, 8, and 6 brains at 6, 18, 30, 64, 91, 125, 183, 307, and 507 dpi, respectively.

(A) Dorsal view of a pallial hemisphere exemplifying some clones recovered at 64 dpi (magnification of the boxed area shows one clone). The analyzed region is outlined with dotted lines. (B) 3D reconstruction of clones at different time points illustrating the progressive increase in their NSC content until 183 dpi. (C) Time evolution of the number of NSCs (i.e., Sox2⁺ cells) per NSC-containing clone. (D) Time evolution of the proportion of NSC-containing clones among all clones. (E) Left: 3D reconstruction of a multicelled fully neuronal clone at 507 dpi. Right: Optical section along the plane defined by the dotted line on the 3D reconstruction (left). Note that the clone does not contain NSCs anymore and is completely detached from the ventricular zone (dotted line). (F) Time evolution of the ratio of neurons to NSCs within the entire population of traced cells. (G) Schematic illustrating the simplified model used for maximum likelihood estimation. The inferred activation frequency and relative proportions of both rNSCs and oNSCs are given in red and blue, respectively (rNSC ➔ rNSC + oNSC, ν = 0.007 day⁻¹; oNSC ➔ oNSC + oNSC, λ = 0.006 day⁻¹; oNSC ➔ loss, μ = 0.017 day⁻¹). (C and D) Box-and-whisker plots, experimental data. The central bold bar and the upper and lower edges of the boxes represent the means and SEM of the most likely clonal composition, respectively; the whiskers of the box respectively correspond to the 95% CIs of the most extreme clonal assignments still in agreement with clonality (see Supplementary Materials). They reflect the combined uncertainty stemming from the clonal reconstruction and the finite sample size. (C, D, and F) Clones as defined by the algorithm. n = 6, 3, 3, 3, 4, 6, 7, 9, and 7 brains at 6, 18, 30, 64, 91, 125, 183, 307, and 507 dpi, respectively. Solid lines depict modeling predictions.

(A) Dorsal view (3D reconstruction) of a pallial hemisphere imaged from a live casper;gfap:dTomato fish. The region analyzed is surrounded by the dotted line. The green and blue squared areas are enlarged in (B) and (D), respectively. (B) Example of an NSC that underwent two symmetric amplifying divisions. Green spots are superimposed to the NSCs of interest; tracks (green) with red dots indicating imaging sessions where the NSCs were still present (blue arrows point to the moments of image acquisition). (C) Relative proportions of the different types of NSC divisions inferred from their tracking in live fish (color coded). n, loss of gfap expression (NP/future neuron). n = 2 brains. (D) Example of a direct differentiation of an NSC (loss of gfap expression) (blue arrows). Asterisks (*) mark surrounding NSCs as visual landmarks. (E) Relative proportions of NSC fates leading to expansion (symmetric amplifying divisions; green), maintenance (asymmetric divisions; red), and consumption (symmetric neurogenic divisions and direct neuronal differentiations; blue). Light colors are the results of our simulations. Note the overall balance in NSC fates. n = 2 brains. Error bars, SD and credibility interval (see Supplementary Theory) for the in vivo data and the modeling, respectively. (F) Modeling of NSC dynamics. The inferred proportions and frequencies (in red, values per cell) of both rNSC and oNSC fates upon activation are indicated (oNSC ➔ oNSC + n, κ = 0.018 day⁻¹; oNSC ➔ n + n, μ₁ = 0.004 day⁻¹; oNSC ➔ n, μ₂ = 0.013 day⁻¹). The corresponding curves are plotted in green above experimental data in Fig. 3 (C, D, and F). Note that NSC/NSC fates monitored by live imaging include both rNSC/oNSC and oNSC/oNSC divisions.

(A) Quantification of the total number of NSCs (as assessed by Sox2 expression) per hemisphere in the Dm territory of interest. One-way ANOVA: F_(8,37) = 8.06, P < 0.001; all pairwise comparisons: LSD test followed by Holm’s adjustment. *P < 0.05, **P < 0.01, and ***P < 0.001. n = 6, 3, 3, 3, 4, 6, 7, 8, and 6 brains at 3.1, 3.6, 4, 5.1, 6, 7.1, 9.2, 13.2, and 19.6 mpf, respectively. Error bars, SEM. (B) Scatter plot showing the absence of correlation between the total numbers of NSCs (Sox2⁺ cells) in the pallial region analyzed and the number of NSCs (Sox2⁺ cells) in the traced her4.1 lineage. n = 6, 3, 3, 3, 4, 6, 7, 8, and 6 brains at 6, 18, 30, 64, 91, 125, 183, 307, and 507 dpi, respectively. Cell numbers are per hemisphere. (C) Timeline of the experiment. Maximal recombination of the ubi:Switch reporter was induced by treating her4.1:ERT2CreERT2;ubi:Switch transgenic fish with 4-OHT repetitively over 5 days. (D) Left: Dorsal view of pallia immunostained for mCherry and Sox2 at 6, 91, and 190 dpi. Right: Segmented Sox2⁺ NSCs (spots). Red spots, NSCs within the her4.1 lineage (Sox2⁺ and mCherry⁺); white spots, mostly newly formed NSCs (Sox2⁺ and mCherry⁻). (E) Quantification of Sox2⁺/mCherry⁺, Sox2⁺/mCherry⁻, and Sox2⁺ cells in the Dm territory of interest (numbers of cells are per hemisphere). One-way ANOVA: Sox2⁺/mCherry⁺ cells, F_(2,13) = 0.59 and P = 0.5697; Sox2⁺/mCherry⁻ cells, F_(2,13) = 6.02 and P = 0.0141; Sox2⁺ cells, F_(2,13) = 9.06 and P = 0.0034. All pairwise comparisons: LSD test followed by Holm’s adjustment. *P < 0.05 and **P < 0.01. Error bars, SEM. n = 5, 6, and 5 brains at 6, 91, and 190 dpi, respectively. (F) Proposed hierarchical organization of pallial NPs.

(A) Dorsal views of the analyzed pallial Dm region at different time points of the clonal analysis. Only the mCherry channel is displayed to highlight the clones. NSCs and neurons are marked by white and blue spots, respectively. (B) Evolution over time of the average number of traced neurons. One-way ANOVA: F_(8,37) = 10.12, P < 0.001; all pairwise comparisons: LSD test followed by Holm’s adjustment (the results of pairwise comparisons are given in data file S1). Error bars, SEM. n = 6, 3, 3, 3, 4, 6, 7, 8, and 6 brains at 6, 18, 30, 64, 91, 125, 183, 307, and 507 dpi, respectively. (C) Time evolution of the average number of neurons per clones. (D) Evolution over time of the ratio of neurons to NSCs in the NSC-containing clones. (E) Evolution of the number of neurons per NSC-containing clone across all time points analyzed. (F) Evolution over time of the average clone size. (C to F) Box-and-whisker plot: The horizontal bar indicates the median, the lower and upper edges of the box represent the first and third quartiles, respectively. The whiskers extend to the most extreme data point. Kruskal-Wallis test: P < 0.0001; all pairwise comparisons: Behrens-Fisher tests (for the sake of the graphs readability, the results of pairwise comparisons are given in data file S1). (C to G) n = 6, 3, 3, 3, 4, 6, 7, 9, and 7 brains at 6, 18, 30, 64, 91, 125, 183, 307, and 507 dpi, respectively.