Supporting Information
Table S1 Affiliation of the main OTUs (16S rRNA gene sequence-based OTUs) obtained from the eight clone libraries (2008-2009, 2001-2003, 1991-1993,
1987-1988, 1981-1983, 1972-1973, 1956-1960, 1951-1952) constructed either using primers that target all cyanobacteria (upper table) or using primers that are specific for Planktothrix (lower table). The closest BLAST hit (from NCBI BLASTN) is presented for each OTU.

Order
Oscillatoriales
Nostocales
Chroococcales other cyanobacteria
Putative genera
Planktothrix
Leptolyngbya
Leptolyngbya
Leptolyngbya
Leptolyngbya
Anabaena
Anabaena
Anabaena
Nostoc
Anabaena
Anabaena
Aphanizomenon
Microcystis
Microcystis
Microcystis
Microcystis
Microcystis
Microcystis
Chroococcus
Chroococcus
Chroococcus
Chroococcus
Chroococcidiopsis
Chroococcidiopsis uncultured chroococcales uncultured chroococcales
Aphanocapsa
Cyanothece
Cyanothece
Cyanothece
Pleurocapsa
Pleurocapsa pleurocapsa
Pleurocapsa
Staniera
Stanieria
Stanieria
Xenococcus
Synechocococcus uncultured cyanobacteria uncultured cyanobacteria uncultured cyanobacteria uncultured cyanobacteria
Order
Oscillatoriales species
P. rubescens
Phylogenetic affiliation closest blast from NCBI
P. rubescens CCAP 1460/15 (HE975026.1) (99%)
Leptolyngbya badia CRS-1(EF429298.1) (92%)
Leptolyngbya sp. GSE-PSE30-01B (HQ132934.1) (97%)
Leptolyngbya laminosa ETS-08 (FM210757) (95%)
Leptolyngbya sp. 2LT21S03 (FM177494) (100%)
Anabaena sp. CCAP 1403/29 (KC297498.1) (99%)
Anabaena planctonica 1tu30s13 (EF638709.1) (97%)
A. mendotae 04-11 (FM242083) (92%)
Nostoc ellipsosporum (AJ630450.1) (99%)
A. mendotae 04-45 (FM242084) (99%)
A. bergii ANA360H (FJ234898) (99%)
Aphanizomenon gracile (AJ293124.1) (99%)
Microcystis aeruginosa NIES-843 (NR_074314.1) (99%)
Microcystis wesenbergii VN212 (AB666077.1) (90%)
Microcystis wesenbergii VN484 (EF208676.1) (95%)
Microcystis aeruginosa VN224 (AB666058.1) (95%)
Microcystis sp. CCAP 1450/13 (HE975021.1) (93%)
Microcystis wesenbergii VN484(AB666080.1) (93%)
Chroococcus sp. JJCM (AM710384.1) (94%)
Chroococcus turgidus HUW 799 (DQ460703.1) (94%)
Chroococcus minutus CCALA 055 (GQ375047.1) (94%)
Chroococcus minutus CCALA 055 (GQ375047.1) (92%)
Uncultured Chroococcidiopsis sp. clone AS4_2 (FJ805930.1) (95%)
Chroococcidiopsis sp. BB82.3 (AJ344553) (97%)
Uncultured Chroococcidiopsis sp. clone AS4_2 (FJ805930 ) (95%)
Uncultured cyanobacterium clone AL58_2CY_3 (HQ419017) (99%)
Aphanocapsa sp. HBC6 ( EU249123.1 ) (95%)
Cyanothece sp. WH 8904 (AY620239.1) (95%)
Cyanothece sp. PCC 8801 strain PCC 8801 (NR_074265.1) (94%)
Cyanothece sp. PCC 8801 (NR_074265) (94%)
Pleurocapsa sp. PCC 7327 (CP003590) (94%)
Pleurocapsa sp. PCC 7327 (CP003590) (93%)
Pleurocapsa concharum VP4-07 (FR798930) (96%)
Pleurocapsa sp. CALU 1126 (DQ293994.1) (97%)
Stanieria cyanosphaera PCC 7437 (CP003653.1) (96%)
Stanieria cyanosphaera PCC 7437 (CP003653.1) (94%)
Stanieria cyanosphaera PCC 7437 (CP003653.1) (95%)
Xenococcus sp. CR_15M (EF545606) (96%)
23 OTUs see details in Domaizon et al 2013
Uncultured cyanobacterium clone Z27M62B (FJ485596.1) (99%)
Uncultured cyanobacterium clone Alchichica_AQ1_1_1C_12 ( JN825321) (94%)
Uncultured cyanobacterium (JX504288.1) (94%)
Uncultured cyanobacterium clone GG48 (JQ404419) (92%)
OTU name
OTU42
OTU82
OTU32
OTU93 *
OTU81
OTU19
OTU59
OTU62 *
OTU63 *
OTU66 *
OTU37
OTU64
OTU78 *
OTU10
OTU28
OTU33
OTU47 *
OTU58
OTU74
OTU7*
OTU14
OTU26
OTU31
OTU22
OTU29 *
OTU56 *
OTU83
OTU84 *
OTU34
OTU49 *
OTU57 *
OTU67 *
OTU70
OTU92 *
OTU71
OTU6
OTU20 *
OTU23
OTU30
OTU38
OTU21 *
OTU68 *
Phylogenetic affiliation closest blast from NCBI
P rubescens DP10.4.25 (FJ612410) (100%) representative sequence
(NCBI deposit) sequences
10-BF1-C
BF3-28C
36-BF12-C
23-BF24-C
32-BF24-C
8
1
1
2
1
BF3-60C
4-BF10-C
69-BF12-C
36
1
1
1 17-BF15-C
31-BF15-C
BF27-C-17
56-BF20-C
2
1
10
7 7-BF1-C
65-BF1-C
1-BF1-C
20-BF10-C
1
2
5
2 5-BF15-C
57-BF20-C
23-BF1-C
BF3-11C
1
8
3
4 BF3-52C
82-BF10-C
32-BF12-C
32-BF20-C
5-BF1-C
BF3-10-C
BF3-85C
8
1
7
2
9
1
9
BF3-90C
13-BF10-C
87-BF24-C
BF3-49C
3
4
3
1 BF27-C-42
42-BF24-C
41-BF1-C
29-BF12-C
2
2
2
1 49-BF15-C
54-BF15-C
53-BF15-C
1
1
BF3-35C
18-BF10-C
31-BF1-C
27-BF15-C
1
1
2
5
164
2008-2009
BF1
7
1
5
1
1
1
1
2
1
1 representative Number of
BF1-34PR-CYA9 105
2008-2009
BF1
66
2003-2005
BF3
2
2
1
1
4
2
1
2
2
1991-1993
BF10
6
1
3
5
1
1
4
1
3
2003-2005
BF3
0
1991-1993
BF10
0
1987-1989
BF12
1981-1983
BF15
1972-1973
BF20
1956-1961
BF24
1951-1952
BF27
1
Ability for N2 fixation (putative function)
1
3
1
2
1
1
1
1
1
1
1
1
14
1
2
2
1
1
1
1
1
1
6
3
1
1
2
2
1
2
1
4
1
6
5
4
2
1987-1989
BF12
3
1981-1983
BF15
0
1972-1973
BF20
4
1956-1961
BF24
0
5
1
3
1
1
1
1
1
1
2
1
1
1951-1952
BF27
32 y y y y y y y y y y c

Table S2 UNIFRAC results. Distance matrix obtained from the comparison of each pair of samples.
Bold * text denotes values in the upper quartile (i.e. most distant samples). bold* = 75-100% upper quartile ; bold = 50-75% ; non bold = <50% lower quartile. bf1 bf1 bf3 bf10 bf12 bf15 bf20 bf24 bf27
Environnment distance matrix bold * 75-100% bold non bold
50-75%
<50% bf3
0.171
Upper quartile
Lower quartiles bf10
0.181
0.202
bf12
0.271
0.272
0.218
bf15
0.351*
0.382*
0.354*
0.295
bf20
0.350*
0.357*
0.345
0.273
0.309
bf24
0.319
0.322
0.319
0.319
0.372*
0.218
bf27
0.229
0.217
0.298
0.272
0.345*
0.295
0.275

Leptolyngbia
Planktothrix
BF1 2009-2008
Nostoc
Anabaena
BF1 long reads
BF1 short reads other chroococales
Chroococcus
Microcystis
Synechococcus
Leptolyngbia
Planktothrix
0 20 40 60 80 number of reads assigned to cyanobacterial genera
BF10 1991-1993
Nostoc
Anabaena
BF10 long reads
BF10 short reads other chroococales
Chroococcus
Microcystis
Synechococcus
Leptolyngbia
Planktothrix
0 20 40 60 80 100 number of reads assigned to cyanobacterial genera
BF24 1956-1960
Nostoc
Anabaena
BF24 long reads
BF24 short reads other chroococales
Chroococcus
Microcystis
Synechococcus
0 20 40 60 80 100 number of reads assigned to cyanobacterial genera
Figure S1 Comparison of cyanobacterial assemblages obtained from sequencing of different amplicons that varied in length (~350bp or ~1400bp). Results are expressed as proportion of sequences affiliated to the main genera (identified from sedimentary DNA originating from three sediment layers (BF1, BF10, BF24), as assigned by BLASTN with identity >95%)

1957
1955
1953
1951
1949
1946
1975
1973
1972
1969
1967
1965
1962
1960
1991
1989
1987
1985
1983
1981
1979
1977
2009
2007
2005
2003
2001
1999
1997
1995
1993
1924
1922
1919
1917
1914
1912
1909
1944
1942
1940
1937
1935
1932
1930
1927
0
Average P. rubescens cells per g dry sediment
5000 10000 15000
Average P. rubescens cells per g dry sediment
16s rRNA
PC-IGS
Figure S2 Quantification of Planktothrix rubescens DNA preserved in the sediment samples estimated by qPCR from two types of assays targeting 16S rRNA (~80bp) and PC-IGS (~70 bp) genes.

Appendix S1 Age models for reference cores and sample dating.
Cores were sampled using a UWITEC gravity corer. They were split lengthwise, photographed and described. On each core (reference and working cores), descriptions included well-documented lithological markers: (i) dark layers related to historical floods events; (ii) gravity-reworked deposits and (iii) distinct white diatoms layers. Where they were present (from 1933), varves were counted directly. This was possible as each doublet was relatively thick (>2 mm). The working core LDB 09P2 was correlated to the reference core (LDB04P1) using both lithological markers and lamina counting , thus allowing an annual temporal resolution at least on the laminated part of cores. For age models,
210 Pb, 226 Ra, 137 Cs and 241 Am activities were measured by gamma spectrometry on the reference core
LDB04P1 at the Modane underground laboratory (LSM) environmental radioactivity facility (Reyss et al.
, 1995). Measurements were made on 1–3 g of dried sediments, using high-efficiency, very-lowbackground, well-type germanium detectors (Reyss et al.
, 1995). Six standards were used to calibrate the gamma detectors (Cazala et al.
, 2003). Generally 24–48 h of counting time were required to reach a statistical error below 10% for excess 210 Pb in the deepest samples and for the 1963 peaks of 137 Cs and 241 Am. The excess 210 Pb (i.e., formed in the atmosphere by the decay of 222 Rn) was calculated as the difference between the total 210 Pb and 226 Ra (supported 210 Pb) activities. This calculation is based on the assumption that the intermediate daughter product, 222 Rn, is in equilibrium with 214 Pb (i.e.,
226 Ra, Arnaud et al 2006).
The CFCS model (Constant Flux, Constant Sedimentation, Golberg 1963,
Krishnaswami et al.
, 1971) was chosen for these cores because sediment rate is dominated by constant sedimentation for recent sediments and because atmospheric flux of radionuclide 210 Pb is constant, as shown by the linear radionuclides decrease (Lake Bourget, R 2 =0.937). Whatever the model or procedure used, 210 Pb-based chronologies must always be confirmed by independent methods (Smith, 2001). Generally, 210 Pb dates are confirmed using 137 Cs profiles, when the 137 Cs profiles are sufficiently intact (Appleby & Oldfieldz 1983).
The 210 Pb exc
profiles in the reference core showed a regular and exponential decay in relation to depth.
CFCS age modeling indicated an average sedimentation rate of 4.0 mm y -1 for Lake Bourget.

The 137 Cs activity versus depth profile showed a small peak at 8.3-cm depth, without any increase in
241 Am activity, corresponding to the fallout from Chernobyl accident in 1986. This chronological marker implied an average sedimentation rate of 4.4 mm yr -1 for Lake Bourget between 1986 and
2005. The well-resolved peak of 137 Cs visible at 17.5-cm coinciding with a peak of 241 Am, confirmed that sediment at these depths were deposited at the period of maximum artificial radionuclide fallout due to nuclear weapon tests, in 1963. Age-depth relationships corresponded to an average sediment rate of 4.2 mm yr -1 between 1963 and 2005. The good agreement between average sedimentation rates deduced from 210 Pb exc
, 137 Cs and 241 Am profiles allowed the extrapolation of the regular sedimentation rate of 4.2 mm yr -1 giving 183 yr of sedimentation.
Appendix S2 Diatom-inferred TP annual concentrations (DI-TP; Berthon et al., 2013).
Annual mean total phosphorous concentrations ([TP]) were reconstructed from a diatom-based inference model based on 86 surface sediment samples collected in lakes around the Alps along a trophic gradient (Wunsam & Schmidt, 1995). The calibration dataset covered a large trophic gradient, with mean annual TP ranging from 2-266 µgP l -1 .
Several models were tested : Weighted Averaging with classical deshrinking WA cla
regression, Weighted Averaging with inverse deshrinking WA inv regression (Ter Braak & Van Dame, 1989), Weighted Averaging Partial Least Squares regression
(WAPLS – Ter Braak & Juggins 1993), Modern Analogue Technique (MAT) with squared chord distance and five analogues (Overpeck et al.
, 1985); and their relative performances were estimated using the r 2 and the Root Mean Square Error of Prediction (RMSEP) calculated after a cross validation method (bootstrapping, 500 permutations) on the calibration dataset.
The selected model was then applied to a diatom biostratigraphy of Lake Bourget from Millet et al.,
(2010). DI-TP reconstructions were performed using program C2 (version 1.7.2, Juggins, 2007). [TP] in the water column of Lake Bourget had been measured sporadically in the 70s and then monitored

from 2004. These monitoring data could hence be used to confirm the model’s relevancy (Database
SOERE-INRA of Thonon-les-Bains – France).
Compared to WA and WAPLS models, MAT models had the lowest performance. There were no relevant differences between WA and WAPLS models performances. Statistical results for DI-TP showed a relatively good predictive ability of the WA model with classical deshrinking (r 2 jackknife
=
0.604, RMSEP = 0.34 log unit). 91% of the fossil assemblage in all samples of Lake Bourget was common with the training set of the DI-TP. The bootstrapped average bias was low (0.92%)
Generally, DI-TP changes were consistent with expected dynamics. Values and timing of DI-TP maxima matched measured TP (Fig. Appendix 2) and DI-TP for current time periods were in close agreement with the monitored chemical data.
Fig. Appendix 2. Temporal changes of [TP] as reconstructed from the diatom-transfer functions
(Inferred TP (continuous line, left axis), and measured water [TP] (red open circles, right axis)) in
Lake Bourget.