Dry season climate

“dry-season_climate_1984-2017.csv” has the climate variables for drought-defined dry seasons from 1984 (start) to 2017 (closure) at the PAMOL Bulu Station. The dry season for a given year (row of table) may have started some weeks, typically in December, of the year prior. Estimates are mean daily values across the season, units given in the column headers except for ‘radi’: ‘n-days’, length of the season; ‘maxT’ and ‘minT’, maximum and minimum temperatures (dry bulb); ‘rain’, rainfall; ‘rft’, running 30-day rainfall total; ‘radio’, volume water evaporated in Gunn-Bellani [GB] radiometer; ‘radi(GB1)’ and ‘radi(GB2)’, radiation (W m–2), as converted from ‘radio’ for the two radiometers, ‘radi(comb)’, combined estimates, including the interpolated values for missing years (in bold). See S1 Appendix of the main paper for radiation calibrations and interpolation method.

> Table 1, Fig. 2; {S4 Appendix: Tables A & B, Figs A and B}

Mast fruiting and climate

“masting-climate_1989-2017.csv” has the masting and drought-defined dry season climate variables for 1989 to 2017: ‘mast’, intensity of mast fruiting on an ordinal scale (1, full mast [‘M’]; 0.5, moderate mast [‘m’]; 0, none); ‘start’ of dry season, being number of days since 1 December (in year prior), ‘dur’, duration of season, i.e. its length in days; minimum temperature (deg. C); ‘rain’, mean daily rainfall (mm); ‘radi’, mean daily radiation (W m–2). Climate values are included for 1987 and 1988 to allow derivation of 1- and 2-year lags. The binary-scale masting (1, 0) applies to 1989 to 2012.

> Fig. 1; {S3 Appendix: Figs B[a] and C}

Seasonal radiation

“radi_seasonal_1987-2017.csv” has mean daily radiation (W m–2) for 1987 to 2017 in each of the four seasonal quarters of the year (months): ‘radi1’, DJF; ‘radi2’, MAM; ‘radi3’, JJA; ‘radi4’, SON; ‘radit’, whole year (total).

> {S3 Appendix: Figs A, B[b] and D}

Radiation differences

“mast_radi-diffs_1989-2012.csv” has mean daily radiation (W m–2) in the current year of masting, as either drought-defined ‘radi’ or first seasonal-quarter defined [DJF] ‘radi1’ dry season, for 1989 to 2017. The ‘radi_1’ and ‘radi1_1’ are corresponding values at 1-year lag, with: ‘radi_diff’, the difference between successive years (current – prior); ‘radi_res(lin)’ and ‘radi1_res(lin)’, residuals from corresponding linear fits of ‘radi’ and ‘radi1’ to date; and ‘radi1_res(sth)’ and ‘radi1_1_res(sth)’, residuals from smoothed LOWESS fits of ‘radi1’ and ‘radi1_1’ to date.

> {S3 Appendix: Figs C and D}; {S4 Appendix: Table C, Figs C and D}

Dry season rainfall

“dryseas_rain&rtf_2003-2005.csv” and “dryseas_rain&rft_2010-2012.csv” have daily rainfall (mm), ‘rain03’ to ‘rain05’ and ‘rain10’ to ‘rain12’, recorded at Bulu, Ndian, over the annual periods 1 December to 31 March, which encompass the dry seasons of 2003 to 2005 and 2010 to 2012, as ‘date03’ to ‘date05’ and ‘date10’ to ‘date12’. The corresponding running 30-day rainfall totals are in columns headed ‘rft03’ to ‘rft05’ and ‘rft10’ to ‘rft12’. (December values are from the calendar year prior.)

> {S5 Appendix: Figs A and B}

Soil moisture

“soil_moisture_2003-2005.csv” has means and standard errors (n = 9) of soil moisture contents (% dry mass) between 2 December 2003 and 26 April 2005 (‘date’), and the matching 30-day running rainfall totals (‘rft’, mm). The 12 columns to the left have the individual sensor SMC values, headers for which combine sensor number, depth (cm) in parenthesis, and afterwards a ‘1’ (yes) or a ‘0’ (no) indicating whether the sensor values were used for finding the mean and se.

> Fig. 3

Raw data on phenology

“phenology_Mb_2010-2014.csv” has the raw field records on phenology over the years (end 2009) 2010 to 2014. Each column for these observations has a header which is a concatenation of ‘stadium’ and ‘date’. Scoring was on the scale 0 to 3 (see main text of paper for explanation). Stadia: ‘fall’, leaf fall; ‘flush’, leaf flushing; ‘flower’, flowering; ‘gr-pods’, (immature) green pods, and ‘br-pods’, (mature) brown pods. Other columns: ‘tree’, tagged tree number in P-plot; ‘diam’, stem diameter at breast height from last census (in 2005); ‘EE’ and ‘NN’, distances eastwards respectively northwards from origin of plot (see Newbery et al. 2013: reference in main paper).

> Fig. 4 [basis]

Summarized phenology

“phenol_Mb_2010.csv” and “phenol_Mb_2012.csv” have mean tree scores for recording dates between 19 December 2009 and 24 August 2010, and between 30 November 2011 and 30 July 2012, respectively, for five phenological stadia (‘stadium10’, ‘stadium12’): same as for the ‘Raw data on phenology’ file (n = 61 trees).

> Figs 4a and 4b

Leaf nutrients-1

“Mb_leaf-nutrients_1998.csv” has N and P concentrations (mg g–1) in litter trap samples of M. bisulcata leaves, over four fortnightly periods between 21.12.1997 and 15.02.1998 (‘dates’), in six HEM half-plots, of two subplots each, and nesting two types of trap: 1,’HL’, hessian lining; 2, ‘GM’, galvanized metal. Leaves separate naturally on shedding into in two ‘parts’: 1, leaflets; and 2, rachises. The layout and design of the half-plots is described in Newbery et al. (1997): reference in the main paper.

> Table 2

Leaf nutrients-2

“Mb_leaf-nutrients_2011.csv” has N, P, Ca, K and Mg concentrations (mg g–1) in hessian-bag litter trap samples of M. bisulcata leaves, of 29 January 2011 (sample collection for previous month), in 30 subplots across the eastern 25 ha of the main P-plot, with two replicate locations per subplot, and leaves separated into two ‘parts’: 1, leaflets; and 2, rachises. The ‘mass’ of each bulked sample is listed also. For plot details see Newbery et al. (2013), and for trap layout, Norghauer and Newbery (2015): references in the main paper.

> Table 2

Fruit nutrients

“seeds&pods_1998.csv” have the Ca, K and Mg concentrations (mg g–1) of M. bisulcata samples of seeds and pods (‘parts’) collected after the 1998 masting. For methodology see Green and Newbery (2002), where the complementary N and P values are also given.

> Table 3

Oil-palm yields

“oil-palm-yields_2003-2016.csv” has the monthly total fresh-fruit bunch masses (metric tons) combined across the five plantations of the Ndian Estate, Mundemba, SW Cameroon, for the years 2003 to 2016, as made available by PAMOL Plc. Dividing these values by ‘Area’ gives production in t/ha/month.

> {S7 Appendix: Fig. A}

Increment cores

“stem_incr_1989-2003.csv” has the annual stem wood increments from cores of 20 M. bisulcata trees: ‘mast’ is a binomial variable (1 = ‘M’; 0 = none). Columns are for the individual trees, whose header labels are tree number prefixed by ‘P’, the last standing for the main P-plot (see Newbery et al. 2013: reference in main paper). The units are mm ? 102 (e.g. 250 units = 2.5 mm).

> {S8 Appendix: Fig. A}

Note: Appendices S2 and S6 have no related data.