Intermediate habitat fragmentation buffers droughts: How individual energy dynamics mediate mammal community response to stressors
Data files
Apr 29, 2025 version files 904.80 MB
-
mibcom_data.zip
900.70 MB
-
README.md
5.72 KB
-
UFZ_Drought_Monitor_daily_SM_L02_subset.nc
4.09 MB
Abstract
Biodiversity is threatened by land-use and climate change. Although these processes are known to influence species survival and diversity, predicting their combined effects on communities remains challenging. We here aim to disentangle the combined effects of drought-induced resource shortage and habitat fragmentation on species coexistence. To understand how both fragmentation and droughts affect individual movement and physiology, and ultimately influence population and community dynamics, we use an individual-based metabolic modelling approach to simulate a community of small mammals. Individuals forage in the landscape to ingest energy, which they then allocate to basal maintenance, digestion, locomotion, growth, reproduction, and storage. If individuals of several species are able to balance their energy intake and needs, and additionally store energy as fat reserve, they may overcome stress periods and coexist. We find that species recover best after a drought when they live in moderately fragmented landscapes compared to those with low or high fragmentation. In low fragmented landscapes, high local competition during resource shortages is problematic, while in highly fragmented landscapes, low energy balance and storage often lead to high mortality during drought. Intermediately fragmented landscapes balance these effects and show the least impact of droughts on species richness, a pattern that holds also when integrating observed drought time series from monitoring data in the model simulations. Due to the interacting negative impacts, we suggest that with ongoing global change, it is increasingly important to understand stressors simultaneously to identify measures that support species coexistence and biodiversity. Including individual energy dynamics allowed us to conflate the different global change effects through energy storage and energy allocation to different processes. Our presented community model, which integrates metabolic and behavioural reactions of individuals to different stressors and scales them to the community level, offers valuable insights with great potential to support nature conservation.
We here provide the model code for our individual-based metabolic community model along with the resulting data for the simulated experiments and the code for producing the manuscript figures for "Intermediate habitat fragmentation buffers droughts: How individual energy dynamics mediate mammal community response to stressors"
Files
The file "UFZ_Drought_Monitor_daily_SM_L02_subset.nc" contains data of the UFZ Drought Monitor / Helmholtz Centre for Environmental Research. It is a subset of the daily Soil Moisture Index (SMI) data from https://www.ufz.de/index.php?en=37937 for two time periods and 20 random locations, which were used to simulate observed droughts with the model. The data was provided by Dr. Andreas Marx and is based on Zink et al. 2015 (DOI 10.1088/1748-9326/11/7/074002).
The achive "mibcom_data.zip" contains the simulated data: the files "mibcom_controlled_community.csv", "mibcom_controlled_singlespecies.csv", "mibcom_controlled_singlespecies_strategy1.csv", "mibcom_controlled_singlespecies_strategy2.csv", "mibcom_data_community.csv" contain data simulated with the individual-based metabolic community model "mibcom.nlogo" (see Zenodo repository). The term "controlled" in the file name indicates that one drought per simulation was simulated with defined drought length and magnitude. The term "data" instead hints to the use of drought monitor data for the simulation of drought occurrences. Further, "community" or "singlespecies" defines whether all species are simulated together or each species isolated in a simulation. By default, individual behaviour during drought is similar to without drought, which means that individuals increase activity to find as much food as before. The file title "strategy1" defines simulations where individuals instead decrease their activity during drought and "strategy2" defines simulations where individuals maintain similar activity (see manuscript and model description for more details).
The model "mibcom.nlogo" was programmed using the programming language NetLogo (version 6.1.0, see Zenodo repository). See the publication for a detailed model description (ODD) and full documentation of the modelling (TRACE).
We visualised the simulated data using a Phyton notebook (Python version 3.8, "code_figures.ipynb", see Zenodo repository).
Columns
In the simulated data files, the columns "clump", "drought_scenario_combi", "specs-included", and "drought_type" define the scenario.
The variable "clump" defines the landscape fragmentation and has three levels: 0.9999 (low), 0.999 (medium), 0.99 (high).
The variable "drought_scenario_combi" defines the drought scenario with length and magnitude and has seven levels: 1: 3 days and 98% resource reduction; 2: 9 days and 98% resource reduction; 3: 9 days and 95% resource reduction; 4: 22 days and 95% resource reduction; 5: 22 days and 90% resource reduction; 6: 57 days and 90% resource reduction; 7: 57 days and 80% resource reduction.
The column "specs-included" can be either "all" for community simulations or, in the single-species simulations, the column defines which species is simulated from 0 (small) to 9 (large).
The column "drought_type" defines whether a single defined drought was simulated ("controlled") or monitor data was used for drought simulation ("data").
When monitor data was used, the additional columns "Drought_file" and "Location" indicate the used observed drought data (years 1952-1962 or 2009-2019, and drought threshold 2%, 5%, 10%, or 20%, see Python notebook file for details) and the index for the used location within this data (20 different random locations).
The column "[run number]" consecutively numbers all simulations in a file.
The last simulated time step is defined by "[step]". This indicates the number of simulated days.
The result column "spec_num" returnes how many species remained in a simulation in the last time step.
All other results are given for each species seperately defined by the number of the species usually at the end of the column name.
The columns "mean_rep_success" return the mean of the number of offspring a female successfully weans. The columns "mean_rep_success_hr" return the mean of the number of offspring a female successfully weans and that find their own homerange (excluding those offspring that dy during home range search).
The remaining variables are timeseries saved as lists over the entire simulation or over the drought period: "number" is the population size, "fmr" is the field metabolic rate, i.e. total energy use per day (g food), "in" is the total energy intake per day (g food), "balance" is the energy balance, i.e. energy costs vs. energy intake, "repro" is the energy investment in reproduction (relative to field metabolic rate), "drought_hr" is the home range size development during drought (radius in 10m), "drought_stor" is the energy storage fill during drought (relative to maximum storage capacity), "sublist locoX drought_start (drought_start + drought_length + drought_recover + drought_recover + 2)" is the energy investment in locomotion during drought (relative to field metabolic rate) and "sublist competX drought_start (drought_start + drought_length + drought_recover + drought_recover + 2)" is the number of competitors per foraging patch during drought.
Missing data is listed as "n/a" and occurs when a simulation does not include information about a specific species as it is either not initialized in the scenario (i.e., single species scenarios) or went extinct.
In this study, we used a relatively novel dynamic individual-based metabolic simulation model for a mammal community (see also https://doi.org/10.5061/dryad.4qrfj6qjf). The model is based on allometric relationships and movement in home ranges, allowing for a variety of species. The model was thoroughly tested and validated using real-world patterns from the literature. An extensive model development description is available in the format of a TRACE document with the publication. We used the model to simulate scenarios of different habitat fragmentation in combination with drought events and analyzed the resulting data.