Data from: Seasonal shifts in trophic interaction strength drive stability of natural food webs
Data files
Jan 07, 2025 version files 354.19 KB
-
25_loops_IS_loop_weight.xlsx
101.92 KB
-
Biomass_stability.xlsx
23.01 KB
-
BodenseeCdef.txt
47.90 KB
-
Flux_IS.xlsx
83.76 KB
-
Jacobian_matrices.xlsx
54.60 KB
-
Openness.xlsx
34.72 KB
-
README.md
8.28 KB
Abstract
It remains challenging to understand why natural food webs are remarkably stable despite pronounced variability in environmental factors and population densities. We analysed the dynamics in the structure and stability of the pelagic food web of Lake Constance using seven years of high-frequency observations of biomasses and production, leading to 59 seasonally resolved quantitative food web descriptions. We analysed the dynamics in asymptotic food web stability using maximum loop weight, which revealed mechanisms governing stability. Maximum loop weight showed a recurrent seasonal pattern while indicating consistently high stability despite pronounced dynamics in biomasses and fluxes. This arose from rewiring of the food web structure along seasons, which counteracted destabilization by enhanced productivity. The rewiring originated from energetic constraints within loops and how loops were embedded into food web structure. The stabilizing dynamics originated from the counter-acting effect between high metabolic activity and competitiveness/susceptibility to predation within a diverse grazer community.
README: Seasonal shifts in trophic interaction strength drive stability of natural food webs
https://doi.org/10.5061/dryad.pnvx0k6xx
Description of the data and file structure
Files, variables and code
For a general description of Lake Constance and its food web, the overall data set, the raw data to obtain the biomasses, the production rates of phytoplankton and bacteria, and the corresponding measurement techniques see https://fred.igb-berlin.de/Lakebase and literature provided therein. More data and documentation will be provided in a data paper (Gaedke, in prep, intended as publication in Data Science).
The present study is based on 59 mass-balanced food web models established by Gaedke et al. (2002) (described by the fluxes among the compartments which also provide the gross growth efficiencies) and further measurements like plankton biomasses and production which have mostly been published in principle elsewhere but e.g. with different resolutions in time. Using this data base we calculated the interaction strengths (Jacobian matrices), measures of stability (Re(λmax), loop weights) and different types of openness in this study. The data describing the 59 mass-balanced food web models and those underlying the figures in the main text and the appendices are provided in the following data sets:
The data set flux_IS.xlsx provides for the 59 phases (first two columns, providing the year and the phase number) and each prey and predator guild (column 3 and 4):
1. the fluxes between all living compartments in units of mg C m-2 d-1 describing the mass-balanced food webs. The compartments are numbered as: 1 - phytoplankton, 2 – bacteria, 3 - heterotrophic flagellates, 4 - ciliates, 5 - rotifers, 6 - predominantly herbivorous crustaceans (daphnids, Bosmina sp., Eudiaptomus sp.), 7 - predominantly carnivorous crustaceans (cyclopoid copepods including their herbivorous nauplii, Leptodora kindtii, Bythotrephes longimanus) and 8 – fish.
2. the negative interaction strengths between all living compartments (d-1), and
3. the positive interaction strengths between all living compartments (d-1). The negative and positive interaction strengths are the off-diagonal values of the Jacobian matrices (the diagonal values were set to 0).
The file Jacobian_matrices.xlsx provides for each year and phase the same negative and positive interaction strengths as given in the data set flux_IS.xlsx but in a matrix format, which is more convenient for some but not for all further studies which may use these data. For the name of the compartments see above.
The data set biomass_stability.xlsx provides for the 59 phases (first columns, providing the year and the phase number):
1. the biomasses (mg C m-2) of all living compartments (trophic guilds, phyt - phytoplankton, bac – bacteria, HF - heterotrophic flagellates, cil - ciliates, rot - rotifers, herb.cru. -predominantly herbivorous crustaceans, carn.cru - predominantly carnivorous crustaceans) and the total plankton biomass averaged across all measurements conducted during the respective phase (the fish biomass is not provided per phase and year as it was kept constant at 752 mg C m-2),
2. the real part of the leading eigenvalues, Re(λmax) (d-1) , of the respective Jacobian matrices, called Real(lambda),
3. the weight of the heaviest loop in the food web of any length, LWmax (d-1),
4. the weight of the heaviest loop in the food web with length 3, called LWmax-3(d-1). This weight was used in the present analysis. It is three times slightly lower than that of a longer loop. Which loop of length 3 was the heaviest during a distinct phase is provided in Appendix Table 1,
5. the weight of the ciliate-flagellate-phytoplankton loop (d-1), called LWC-F-P,
6. the net primary production (i.e., the production of the resource guild) called Phyt Prod in mg C m-2 d-1, and
7. the gross growth efficiencies of the consumers (i.e., the ratio between production and ingestion), labelled HF Efficiency, …, Fish Efficiency (dimensionless). They are used to calculate loop weight (d-1).
The data set 25_loops_IS_weight.xlsx provides for each year and phase (first two columns) for the 25 positive loops of length 3
1. the three interaction strengths within the respective loop. For each loop the interaction strength IS-12 provides the negative interaction between the intermediate consumer and the resource guild, IS-23 stands for the negative interaction between the top predator and the intermediate consumer, and IS-13 for the positive interaction between the resource guild and the top predator, and
2. the loop weight.
The loops are coded according to their 3 trophic guilds going from the resource guild via the intermediate consumer to the top predator: 1- phytoplankton, 2 – bacteria, 3 - heterotrophic flagellates, 4 - ciliates, 5 - rotifers, 6 - predominantly herbivorous crustaceans, 7 predominantly carnivorous crustaceans. For example, the loop called “1-3-4” is the ciliates-flagellates-phytoplankton loop.
The data set openness.xlsx provides for each year and phase (first two columns) for the 6 loops which became the heaviest during individual phases
1. the overall openness OO of the entire loop (cf. methods in the main text) and
2. the openness in respect to the 4 in- or outgoing fluxes (O1 to O4) (both dimensionless).
The loops are coded according to their 3 trophic guilds going from the resource guild via the intermediate consumer to the top predator: 1- phytoplankton, 2 – bacteria, 3 - heterotrophic flagellates, 4 - ciliates, 5 - rotifers, 6 - predominantly herbivorous crustaceans, 7 predominantly carnivorous crustaceans.
Data for individual figures
The data for the figures in the main text are provided in the following way (the hash tagged numbers “#x” refer to the descriptions under point x of the respective data set, e.g., #4 of biomass_stability.xlsx refers to point 4 “the weight of the heaviest loop…”):
1. Fig. 2 by biomass_stability.xlsx #4,
2. Fig. 3 by 25_loops_IS_weight.xlsx #2,
3. Fig. 4 by biomass_stability.xlsx #5 and openness.xlsx, and
4. Fig. 5 by biomass_stability.xlsx #1, flux_IS.xlsx #1, 2 and 3 and openness.xlsx #1.
The data for the table and figures in the appendices are provided in the following way:
1. Table 1 by 25_loops_IS_weight.xlsx #2,
2. Fig. A1 by biomass_stability.xlsx #1 and flux IS.xlsx #1,
3. Fig. A2 by biomass_stability.xlsx #1 and #6, and flux IS.xlsx # 2 and 3,
4. Fig. A4 by biomass_stability.xlsx #2 and 4,
5. Fig. A6 by Table A1, biomass_stability.xlsx #2, 4 and 6,
6. Fig. A7 by openness.xlsx #1 and 25_loops_IS_weight.xlsx #2,
7. Fig. A8 by flux_IS.xlsx #1 and 25_loops_IS_weight.xlsx #2, and
8. Fig. A9 by flux_IS.xlsx #2, #3 and 25_loops_IS_weight.xlsx #2.
The following code was used:
1. To identify all possible loops in the food webs and to calculate the loop weight the programme "Program_identify_loops_and_calculate_ LW.R" was used. It uses the data file "BodenseeCdef.txt" as input file, containing the interaction matrices (fwnr - number of the food web (n=59), P - phytoplankton, B - bacteria, HF - heterotrophic flagellates, Cil - ciliates, R - rotifers, HC - herbivorous crustaceans, CC - carnivorous crustaceans, F - fish)
2. To produce Figure 2, 3 and 4 the code Figure2.R, Figure3.R and Figure4.R was used. To calculate the correlation coefficients the code spearmancorrelation.R was used.
The other calculations were done in Excel and the other figures with Powerpoint.
Reference
Gaedke, U., S. Hochstädter & D. Straile (2002) Interplay between energy limitation and nutritional deficiency: Empirical data and food web models. Ecological Monographs 72: 251-270.
Methods
For a general description of Lake Constance, the overall data set, the raw data to obtain the biomasses, the production rates of phytoplankton and bacteria, and the corresponding measurement techniques see https://fred.igb-berlin.de/Lakebase and literature provided therein. Plankton was sampled weekly during the growing season and approximately every second week in winter up to 20 years. Biomasses were calculated from taxonomically resolved microscopic counting of all plankton groups. Production rates were measured in in situ using advanced techniques. This data set was used to establish 59 mass-balanced food web models on which the present study is based.