Title of the dataset:
Data underlying the research on salt marsh seed retention, the seed bank and seed viability

Authors:
Marin van Regteren, Orcid:0000-0003-0329-422X

Related publication:
"Limited seed retention during winter inhibits vegetation establishment in spring, affecting lateral marsh expansion capacity"
Article ID: ECE35781 published in Ecology and evolution


Description:
Seed distribution (Field): 1800 layers of sediment are checked for supplied and naturally abundant seeds
Seed viability (lab): all seeds are checked for viability

This dataset contains the following files:
Seed distribution
Seed viability (lab)

Explanation of variables:
Treatment	Control, December or March for, respectively, no seed addition (C), seed addition in December (D), of seed addition in March (M)
Plot	Plot number, 10 replicate block were used along the coast/marsh edge
Edge/intertidal flat	Plots were placed at the marsh edge or 10 meter before the marsh edge on the bare intertidal flat
Replicate	Three sediment cores were taken in each plot for analysis of seed distribution (later on pooled for analyses)
Depth from-to (cm)	Depth of the sediment layer assessed. The upper 5 cm were analysed per 1 cm, the lower 10 cm of the sediment samples were analysed per 2 cm
ID	An ID that combines treatment, plot, edge/flat and depth, in order to pool the replicates
Depth	Mean of sampling depth
Elevation (m NAP)	Bed level elevation of each plot (an average of 4 RTK-DGPS measurements) in meter above NAP
Salicornia_seedling	Number of established Salicornia procumbens seedlings (upper layer only)
Salicornia_seed	Number of Salicornia procumbens seeds 
Salicornia_seed_germinated	Number of Salicornia procumbens seeds germinated before viability check
Salicornia_seed_germinated_lab	Number of Salicornia procumbens seeds germinated during viability check
Salicornia_seed_germinated_total	Total number of Salicornia procumbens seeds germinated 
Aster_seed	Number of Aster tripolium seeds germinated during the viability check
Aster_seed_germinated_lab	Number of Aster tripolium seeds 
Spartina_seed	Number of Spartina anglica seeds 
Spartina_seed_seedlings	Number of Spartina anglica seedlings
Spartina_seed_total	Total number of Spartina anglica seeds 
Spartina_seed_germinated_lab	Number of Spartina anglica seeds  germinated during viability check
Spartina_seed_germinated_total	Total number of Spartina anglica seeds germinated 


Methods, materials and software:
To assess seed retention and the seed bank, in May 2018, three randomly placed sediment cores with 5.8 cm diameter were taken, up to 15 cm deep in each plot. The three sediment cores together covered a total surface area of 80 cm2. This resulted in 180 sediment cores in total, 60 cores for each treatment (control, December and March). Of these 60 cores, 30 cores were collected at the vegetation edge and 30 cores were collected at the intertidal flat to estimate location effects. 
The top 5 cm of the soil was sliced into layers of 1 cm, the next 10 cm (5-15 cm deep) of soil was sliced into layers of 2 cm thick. This resulted in the following layers: 0-1, 1-2, 2-3, 3-4, 4-5, 5-7, 7-9, 9-11, 11-13 and 13-15 cm. In the statistical analysis, the centre of each layer is used to represent the depth. Each sediment layer was sieved over a 500 m mesh to filter out all seeds from the sediment. Everything larger than 500 m was transferred to a transparent gridded tray, placed on a light table and analysed for seeds. For the surface layer (0-1 cm), the number of seedlings for each of the three plant species was recorded. In each succeeding layer, the number of seeds for each of the three plant species was recorded. Thereafter, all seeds were transferred to the climate chamber for germination to determine seed viability, following the germination protocol by Ter Heerdt et al. (1996). 

Three replicate sediment cores from each plot were pooled to obtain seed abundance data. For the top 5 cm, abundances were counted per 80 cm3, for the lower 10 cm abundances were counted per 160 cm3. For samples between 5 and 15 cm deep, abundances were estimated per 80 cm3, by dividing total abundance with the thickness of the layer (2 cm). Viability was calculated based on the fraction of the recovered seeds that germinated successfully in laboratory conditions. This was repeated for all treatment and depth combinations. Samples without seeds present were excluded from the viability analysis. 
The elevation of the experimental plots at Zwarte Haan was used to estimate total bed-level change during and after winter. The high-resolution bed level change at Westhoek was calculated as the difference in distance to the bed from the instrument probe at the beginning and end of each measurement burst (10 min frequency). The measurements of bed level change and water level were then averaged over, respectively, 1 h and 10 minute intervals.
The abundance of A. tripolium and S. anglica seeds in the control and December treatments as well as below 3 cm in the March treatment was insufficient for statistical analyses. The results of A. tripolium and S. anglica seed abundance will, however, be presented and qualitatively discussed. Analyses on seed viability and depth distribution of the seed bank were performed on S. procumbens. 
Firstly, seed abundance was analyzed with a Generalized Additive Model (GAM) (Wood 2006). GAM was selected as the modelling instrument as the effect of depth was not linear. Furthermore, a GAM was applied to smoothen expected autocorrelation between depth layers. The fixed factors were: smoothed depth (interaction with treatment), location and treatment.  Block was considered to be a random effect (Wood 2008). The smooth function used penalized regression splines with 5 knots, where the number of knots reflects the degrees of freedom (i.e., the flexibility of the curve) required by the spline (Wood 2008). An offset was used to correct for varying sample thickness at different depths. Because a Poisson distribution resulted in an overdispersed fit, the data was analyzed with a log-linked negative binomial distribution. To correct for the sampled depth layer thickness, both model outcome and observed counts were eventually expressed in numbers per 80 cm3 soil. 
Next, viability, expressed as the fraction of successfully germinated seeds/seedlings of the total seeds, was analyzed using a GAM with a binomial distribution with a logit link function. The fixed factors were: smoothed depth (interaction with treatment), location and treatment. The block was again considered to be a random effect. The smooth function used penalized regression splines with 8 knots. Only sampled depth layers with S. procumbens seeds present were used for the viability analysis. 
Finally, seedling establishment was analyzed with a Generalized Least Squares (GLS) to allow for unequal variances (heteroscedasticity) (Pinheiro and Bates 2000). The fixed factors were: species, location and weighted treatment. Block was considered to be a random effect. 
All confidence intervals were plotted as 1.96*standard error (i.e. 95% confidence). Likelihood Ratio tests (LR) were used to assess the significance of individual factors. Model assumptions were assessed graphically (Zuur & Ieno 2016). All statistical analyses were performed with the statistical program R (R Core Team 2015) using additional packages: ggplot2 for plotting (Wickham 2009), nlme and lme4 for linear mixed effect models (Bates et al. 2015; Pinheiro et al. 2018), and mgcv for GAM models (Wood 2011).


This dataset is published under the CC BY (Attribution) license.
