iLand Release 2.1: Forest Management, Deadwood, and More

We are excited to release iLand 2.1! Our last big release was 2.0, roughly one year ago. We will tryp to keep this cadence of releases, and release each summer a new version that includes all additions and changes that went into the model over a year. So expect 2.2 in summer 2026!

This post summarizes the major changes and new features introduced in iLand version 2.1 since version 2.0 (July 2024). This release focuses on three key areas: a powerful new forest management library, enhanced deadwood dynamics, and various other improvements.

New Forest Management Library (abe-lib)

(c30858ae). See abe-library for more details!

A major addition in this release is the abe-lib JavaScript library, which provides a flexible and powerful way to define and execute complex forest management activities. This library is designed to be used within the ABE (Agent-Based Extension) module and allows for the creation of sophisticated silvicultural treatment programs (STPs).

Key Features

• Modular Design: The library is organized into modules for different types of activities, including harvest, thinning, and planting.
• Customizable Activities: Users can create custom activities by combining and parameterizing the provided functions.
• Signal-Based Scheduling: Activities can be scheduled based on time, specific events, or signals emitted by other activities, allowing for dynamic and adaptive management plans.
• Patch-Based Operations: ABE supports now powerful tools for selecting and managing patches of forest based on various criteria (think selecting sub-areas of a stand based on criteria, and then apply management specifically on these areas).

Getting Started

To use the new library, you can include it in your ABE scripts. Here are some examples of how to use abe-lib to create STPs:

Example: Clearcut and Plant

Example: Structured Thinning and Target DBH Harvest

Available Modules

• lib.harvest: Provides functions for various harvesting methods, including clearcut, femel, and target DBH harvest.
• lib.thinning: Includes functions for selective thinning, such as negative selection and dynamic thinning.
• lib.planting: Offers tools for planting specific species in selected areas.
• lib.selectOptimalPatches: A powerful tool for selecting patches based on user-defined criteria, such as max_light.

For more detailed examples, please refer to the abe-lib/example-use.js file.

Deadwood Dynamics

(4a7d19a5) See dead trees for more details!

This release introduces significant enhancements to the modeling of deadwood, allowing for individual tracking of dead trees in addition to the existing cohort/pool-based approach. This provides more detailed insights into the quantity, quality (decay class), dimension, and species of snags and downed stems, which is particularly useful for habitat quality assessments.

Trees enter the individual tracking system when their DBH at the time of death exceeds model.settings.soil.swdDBHSingle (cm). When such trees die, their biomass is separated into SWD pools (branches, foliage, fine- and coarse roots), while the stem biomass is tracked as a standing snag. If the cause of death is wind or management (e.g., cutting), the stem is immediately considered downed.

DeadTreeList Object

The DeadTreeList class in the ABE JavaScript API provides a powerful way to access, query, and analyze individual deadwood elements.

Example: Working with Deadwood

Snag Variables

GeoTIFF Support

(9afc3b57)

iLand now seamlessly integrates with GeoTIFF files for reading and writing most GIS data, which is more memory-efficient and faster than using ASCII based formats. This also enhances interoperability with other geospatial tools and workflows.
iLand reads projection information from the *first* GeoTIFF it encounters and uses this information for files written. It is thus very important to have consistent projection in all GIS files loaded in a project!

Note also that data type support is limited, currently iLand can work with: INT16, INT32, FLOAT, DOUBLE.

macOS Support

(909c1658)

iLand now supports macOS, including ARM-based M1/M2 processors, alongside Windows and Linux. This expands iLand's accessibility to a wider range of users and computing environments. Support is a bit experimental - if you encounter problems please get in touch on Discord!

Other Changes

• Bark Beetle Module: The bark beetle module has been updated to consider the proximity of attractive deadwood when calculating infestation risk.
• Fire Module: The fire module now includes a more detailed calculation of combustible fuel, including contributions from the forest floor and deadwood. The new calcCombustibleFuel function calculates the total combustible fuel (kg/ha), and sets the reference variables for forest floor and deadwood. This allows for a more realistic fire behavior simulation. (e2caeb6a)
• User Interface: The user interface now includes an option to visualize standing and lying deadwood on the map.

Minor Changes and Bug Fixes

Many smaller changes and bug fixes went into iLand over the last year. Please check out the GitHub repository for more details!

We are happy to release version 2.0 of iLand! While a lot of time passed since the last major release in 2016, model development has never cased, and continued with a good pace. The new version comes with new submodules (more below), updates of the user interface and numerous small improvements and bug fixes.
This release includes more than just a new model version: 12 years after the first iLand paper 2012 in Ecological modeling, a new iLand paper is released today in the same journal (aptly named “The individual-based forest landscape and disturbance model iLand: Overview, progress, and outlook”). The new paper (https://doi.org/10.1016/j.ecolmodel.2024.110785) takes stock of the state of the model and provides a comprehensive starting point for all things iLand.
The community of iLand users is continuing to grow. iLand has been applied on three continents and several labs use the model and contribute to iLand development and its applications. We therefore improve communication within the community with a new Discord-based channel, and by switching model development to GitHub.

iLand 2.0

iLand 1.0 was released in 2016 and was our first “feature-complete” version, that is a version that included all the components that we planned to have originally. Since then, iLand saw further advancements in both model development and model application. Here we’d like to highlight just a few highlights:
Model development
• A general module to simulate biotic disturbances (BITE, https://doi.org/10.1016/j.envsoft.2021.104977), from fungi to mammals
• A module to incorporate the interactions of forests with permafrost and soil-surface organic layers in boreal forests (POLE-FM, https://gmd.copernicus.org/articles/16/2011/2023/)
Model application
• iLand has been successfully applied in temperate and boreal forests on three continents, namely Europe, North America, and Asia
• Over the years the parameters of 150 species and provenances have been collected and recently published (Thom et al 2024, Data in Brief, https://doi.org/10.1016/j.dib.2024.110662)
• We are currently aware of 51(!) scientific publications using iLand

Further updates

With the release of iLand 2.0 we put a particular emphasis on new users of the model:
• Our new paper (https://doi.org/10.1016/j.ecolmodel.2024.110785) provides a comprehensive overview of the current state of iLand
• The iLand book (https://iland-model.org/iland-book/) is an online book with 130+ pages and was specifically created to provide a compact and accessible guide for working with iLand. It accompanies other model documentation streams (such as the website (https://iland-model.org), and scientific papers (https://iland-model.org/publications)) and is a resource for both iLand beginners and advanced users
• To facilitate communication within the growing number of model users, we started a Discord community channel for iLand: https://tinyurl.com/iland-discord
Join the channel using this link: <new Discord invite link>!
• We moved the development of iLand to a public GitHub repository (https://github.com/edfm-tum/iland-model). The repository includes the full development history (imported from the previous Subversion system) and makes it easier for everyone to have a look at the code (of specific versions), update the code and contribute to iLand development!

Within the frame of the RESIN project, iLand applications in Central Europe are expanding to Slovakia. A key aim of RESIN is to understand the disturbance regime of Central Europe by contrasting dynamics in different landscapes of the area. One of the focal sites for this analysis is in the Tatra mountains of Slovakia. After considerable preparation (data gathering, a field trip to the area from the BOKU team, etc.) we’ve now started to implement the Tatra study landscape in the iLand model.

To facilitate the simulation in the Tatra mountains the RESIN collaborators Tomas Hlasny, Laura Dobor, and Ivan Barka visited BOKU for a two-day workshop in December of 2016. With the help of their local expertise we made great strides towards bringing the forests of the western Carpathians into the iLand framework. 

At the workshop, discussions concerned local data availability and issues of model use, but also addressed the important question of testing the model against independent data when applying it to a new study environment. In addition, we’ve already ventilated a number of exciting ideas on how to put the model to use once everything is set up and evaluated. 

 After 7.5 years of working on iLand, we’re very happy to announce the release of iLand 1.0.

Why 1.0 now?

Well, when we set out to thinking about an entirely new model, and all the functionalities it should have, and all the processes it should contain, it was basically the current version of iLand that we had in our minds: A model to simulate fully dynamic and spatially explicit forest landscape dynamics at high spatial resolution, including process-based and interacting disturbance modules for wind, bark beetles, and wildfire, and capturing the complex and dynamic interactions between managers and emerging forest development. Voilà – here we are!
So what is new compared to the previous version?

Agent-based model of forest management

Over the last months (and years) we’ve developed a highly detailed and dynamic model of forest management and incorporate it into iLand. The Agent-Based forest management Engine (ABE) simulates managers as agents that dynamically adapt their actions to changes in the environment. The management agents are able to consider management unit-level goals and constraints (e.g., maximum sustainable harvest levels) and combine them with detailed silvicultural knowledge applied at the stand level in the form of stand treatment programs. Given stimuli like changes in the environment agents can adapt their behavior reactively (i.e., after the manifestation of change) or proactively (i.e., in anticipation of change). Furthermore, they are dynamically able to revise plans and spatial structure (such as stand boundaries) after a major reorganizing event such a disturbance has taken place. And all this is not restricted to one agent on the landscape – iLand is now well able to simulate multi-owner landscapes managed by a large number of different agents, with different objectives, preferences, and management systems. The basic approach and a few implications in dynamic simulations are described in more detail in this paper.

The concept of the agent based forest management engine. Agents interact via a (technical) interface with the ecosystem.

Bark beetle disturbance module

In addition to management we’ve been working on completing our initial triplet of disturbance modules by developing a module of bark beetle disturbance. The three quantitatively most important disturbance agents in Europe’s forests can thus now be simulated in iLand. The newly introduced bark beetle module dynamically simulates the interactions between climate, bark beetle disturbance, and forest development. It explicitly considers bark beetle phenology and development, spatially explicit dispersal of the beetles, colonization and tree defense reaction, as well as temperature-related overwintering success and predation by antagonists. Furthermore, the design of the models follows recent findings on multi-scale drivers of bark beetle outbreaks, considering drivers at the tree (defense, susceptibility), stand (thermal requirements and beetle phenology), landscape (host distribution, beetle dispersal) and regional (climate variation and extremes as triggers of outbreaks) scale. iLand is thus now able to explicitly simulate the amplifying feedbacks between wind and bark beetle disturbances. In fact we’ve used the model recently to quantify the contribution of interaction effects to recently observed disturbance events, and could show that wind – beetle interactions strongly amplify the climate sensitivity of the disturbance regime in Central Europe.

Simulated cumulative disturbances for wind and bark beetle under observed climate conditions (a) and assuming an increase in temperature of +4° (b). (Kalkalpen National Park, Austria)

As always, an extensive documentation of the new model features can be found in the iLand Wiki. Also, the new modules are included in the code and executable that can be obtained through the iLand download page.

We hope that you find these additions useful, and look forward to many applications of iLand 1.0 in the coming months and years. And now: onwards to 2.0!

This is a guest post by Dominik Thom.

After focusing on disturbance effects on tree diversity in our last blog, we aim to pour some more oil into the debate of this hot topic. This time we employed iLand to simulate tree vegetation on 13,865 ha of the Kalkalpen National Park in Austria, studying 36 unique combinations of disturbance and climate scenarios. Furthermore, we extended our analysis from purely tree related indicators of biodiversity (including tree diversity and canopy complexity) to forest-dwelling species (see here for more info). We developed empirical response functions for nine species groups including Araneae, Carabidae, ground vegetation, Hemiptera, Hymenoptera, Mollusca, saproxylic beetles, Symphyta, and Syrphidae, and linked them to our dynamic landscape simulation. To account for long lead-times of tree species adaptation to changing environmental conditions we simulated time a period of 1,000 years in each run. Check out the video of forest development trajectories under different climates below.

Our results on climate change impacts on biodiversity did not confirm the expectation of only negative climate effects. Indeed, the response of the 11 biodiversity indicators assessed here varied strongly: Aranea, canopy complexity, Carabidae, saproxylic beetles and tree species showed a decreasing trend with climate, while ground vegetation, Hemiptera, Hymenoptera, Mollusca, Symphyta and Syrphidae benefitted from climate change. Effects from disturbances on biodiversity were less pronounced but positive throughout, which is supported by a large body of the peer-reviewed literature on that topic. Moreover, we found an upwards shift in elevation of the spatial hotspots of biodiversity at our mountainous study landscape.

Our findings indicate that climate change-induced intensifications in disturbance regimes will likely compensate some of the negative effects of climate change on biodiversity. However, shifts in biodiversity hotspots may challenge conservation management, and thus need to be considered in future management activities. We will continue to study biodiversity-related topics in future simulation activities with iLand, in order to better understand the potential trajectories of forest biodiversity under changing environmental conditions.

The relationships and interactions between biodiversity and disturbance have long been a focus of ecological research (Connell 1978, Grime 1979). Furthermore, recent research has shown that biodiversity is the backbone for the provisioning of many important ecosystem services. One of the major hypothesis here is known as the insurance hypothesis (Yachi and Loreau 1999), which postulates that diversity buffers against fluctuations and disruptions. This has been shown conclusively for grassland ecosystems (see e.g., Cardinale et al. 2013), and is also implied for forests (where managers are frequently advised to propagate mixed forests to hedge against risks). Yet, how big this effect is, and if it changes over stand development (which is considerably more complex in forests compared to grassland ecosystems) remain big question marks.

Which is where iLand comes in. There are tremendous efforts to study diversity effects empirically and experimentally (Baeten et al. 2013), yet a comprehensive analysis over a large gradient of diversity levels covering centennial time scales is to date only possible in silicio. Within the framework of the FunDivEUROPE project we thus set up an analysis to study if and how tree species diversity mediates the impact of disturbances on the forest carbon cycle with iLand (ht to Mariana Silva Pedro, who’s actually done all the hard work).

We found that – as predicted by the insurance hypothesis – more diverse forests experienced a lower negative impact of disturbances on the C cycle. One particularly interesting finding was that this effect increased in strength with increasing disturbance impact, suggesting that diversity is at least partly able to buffer forests from the impacts of intensifying disturbance regimes (see Seidl et al. 2014). Also the predictability of the system increased with diversity, or, in other words, its stochasticity decreased, which is an important factor for a continuous provisioning of ecosystem services.

Interestingly, however, these effects were more pronounced in early seral stages of forest development, and weakened or even reversed in late seral forests. Competitive exclusion and diverging successional trajectories were responsible for this result, which underlines that diversity – disturbance relationships in forests are considerably more complex than in grassland systems. Here’s the link to the paper in case you want to know more.

We’ll be presenting the newest iLand developments at the Austrian “Klimatag” (i.e. climate day) in Vienna next week. Both Werner and I will be giving insights into a recent iLand-related project in which we focused on dynamically integrating human responses to a changing environment into the model. The objective of the MOCCA project (funded by the Austrian Climate Research Program) was to improve the simulation of managed forests under changing environmental conditions by not only considering ecological sensitivities (which are of course central to the iLand architecture) but also incorporating adaptive feedbacks between humans and forests. We approached this by integrating an agent-based model of forest management into the iLand framework, which we’ll be presenting at the Klimatag next week.

Here’s the talks:
Werner Rammer, Rupert Seidl, Filip Aggestam, Kristina Blennow, Bernhard Wolfslehner: Wie sensitiv reagieren WaldbewirtschafterInner auf klimabedingte Änderungen in ökologischen Prozessen? (April 30 2015, 10:35, Session D2)

Rupert Seidl, Werner Rammer: Simulation von dynamischen Rückkoppelungen zwischen Wald und seinen Bewirtschaftern unter sich wandelnden Klimabedingungen (April 30 2015, 14:45, Session F1)

And here’s a link to the overall program (which is quite diverse and features a lot of exciting talks and posters).

The endeavor of scientific publication is changing drastically. The article published in printed periodicals has - besides books, of course - been THE single most important means of making scientific advances public since more than 100 years (see Svante Arrhenius insights on the role of CO2 in the earth’s atmosphere, published in 1896 in the Philosophical Magazine and Journal of Science). In the last decade or so, these established journals have increasingly made use of the internet, and a growing share of the ever-expanding number of new journals is solely published online today. This has not only revolutionized the way we peruse the literature, it also opens up new possibilities with regard to the depth of information that is made available through an article.

It is, for instance, common to publish a detailed account of the (increasingly complex) methods used in a paper in a (digital only) online supplement. The latter is usually not restricted in length (since the cost of server space is &lt;&lt; the cost of page space for a journal), and particularly in highly ranked journals online supplements are frequently exceeding the actual printed article by a factor of 3 to 10 in length.

While having all the additional methodology available is nice, what is oftentimes as interesting to peers is access to the raw data that has been used in study. This not only ensures transparency and enables post-publication review, but also facilitates quantitative metaanalyses (which get increasingly important to synthesize the exponentially growing body of scientific literature). Groups such as the LTER Network have pioneered such data repositories in ecology (including the development of consistent metadata standards), and repositories such as Dryad, which not only archive datasets but make them citable by assigning a unique doi to them, are increasingly popular. In fact, two of the four journals I’ve published in in the last six months requested for us to archive the data in a public archive (a request we of course gladly complied with). I expect that this (softly forced) new openness with data and raw results will considerably advance the scientific endeavor and will both help detect fraud as well as open new views and novel perspectives on existing data.

So all is well with science in the digital age then? Well, yes and no. Although I see the direction in which publishing in the digital age is moving as largely positive, I’d argue that we ought to go much further in making use of the potential of digital information storage and management. While there are clearly many facets to this, I feel that we should increasingly also make methods (in the form of scripts, code, and/ or executables) available with our work, in order to grant peers true repeatability. For simulation studies this would mean that the model code, executable, and scripts to run the model would be available to anyone wanting to scrutinize the results. We in fact tried to archive iLand (code and software) and the respective auxiliary data alongside one of our recent application papers, and were greatly encouraged by the editor to do so. However, in the end (and after quite some back and forth between the journal, the editor, and us) we had to let it go, as the journal and its online archiving system was just not set up for handling such kind of information. I do think that in the future such options should be provided by the journals though, and should be encouraged by the editors.

For the time being we’ll continue to archive all the respective model information here on our website for those of you who’re interested. This is of course likely not going to be as permanent as a server backed by a big journal would be, but it is a start I guess. And talking about permanence opens another can of worms… will we be even able to run our software or compile our code in, say, ten years from now? Well, looking at my old 5 ¼ inch floppy disks I’m not so sure. But then again, the best way to keep software from falling into oblivion is to constantly use and improve it… so back to some more model development!

Quite a while ago I’ve been blogging about a new simulation experiment that we’ve been working on using iLand. Well, the good news is that the corresponding paper was now published in Ecological Applications. In this study we’ve used iLand to quantify the resilience (of structure, composition, and functioning) to disturbance, and tested how live tree legacies (i.e., remnant live trees that are carried over into the post disturbance landscape) modulate resilience.

In my post today I, however, want to focus on the model development that was necessary to implement this study, in particular the development of a wildfire module for iLand (as this is something that is relegated to the Online Supplement of the paper, I figured I’d highlight it here). So besides the wind module, the second fully published and tested disturbance module that is now available for iLand is on fire (pun intended). Our main development goals were (i) for the module to be fully integrated into the iLand environment in order to simulate dynamic vegetation – disturbance – vegetation feedbacks, (ii) to use a process-based framework addressing the principal processes ignition, spread, vegetation impact, and extinction explicitly.

In contrast to the wind module, where we’ve been putting quite some time and energy into developing a new and novel approach to modeling this disturbance agent, we’ve based the iLand wildfire module on previous experiences in wildfire modeling, in particular on the approaches of LADS, FireBGC v2.0, and LandClim.

If this sounds not that exciting for you I say: not so fast! The added value here comes from a well thought out combination of the strengths of these different approaches (LADS: grounded in empirical data and developed for large-scale applications, FireBGC: high process detail and integration with biogeochemical cycling, LandClim: parsimony and integration with vegetation dynamics at the landscape), and their incorporation in iLand. It’s particularly the high resolution of iLand (individual tree level) that makes addressing questions like the current one feasible. And besides: Yes, we do have fun in developing models, but not in reinventing the wheel. And since wildfire is arguably the best researched (and most frequently modeled) disturbance agent in temperate and boreal forests we chose this route for this particular agent.

After looking at a number of fire simulators in detail and developing the iLand fire module we were of course curious to see how it all works out. And since a thorough model evaluation is a prerequisite for any simulation study we engaged into an in depth parameterization and evaluation exercise with iLand, using independent empirical data from the HJ Andrews Experimental Forest (HJA). We used the extinction probability of the cellular automaton-based fire spread algorithm to parameterize fire complexity. Furthermore, we re-evaluated the crown kill function developed by Schumacher et al. (2006) for Rocky Mountains ecosystems and adapted it for our purposes in the Western Cascades of Oregon. With the model thus set up we tested, whether the fire regime of the last 500 years (reconstructed by means of dendroecology) could be realistically reproduced by the model. In these tests, the simulated mean fire (916 ha) corresponded well to the reconstructed mean fire size for the landscape (965 ha). The dynamically simulated mean fire return interval (MFRI) of 218 years for the entire HJA landscape was slightly lower than the reconstructed value (262 years). However, the expected pattern of decreasing MFRI with increasing elevation was reproduced by the dynamic fire simulations with iLand (see Figure, and Appendix B here for details).

Figure: (a) Simulated versus expected fire sizes. (b) Frequency of fires per hectare for the 6364 ha HJA landscape from fire history reconstructions and iLand simulations. Simulation results are the derived from ten replicates of a 500 year simulation series.

Overall we’re quite happy with how things turned out, and with having another disturbance module available in iLand. As always we’ll shortly roll out a new version of the model (executable + code + online documentation wiki) including all these new developments. But the show must go on of course… right now we’re back to where I personally started with my PhD, as we’re currently working on integrating also a bark beetle module into iLand. Stay tuned!

After the theoretically motivated post about scaling not too long ago I thought that it’d be insightful to illustrate the role of scale with an example. We recently did some work aiming to get a better understanding what drives the spatio-temporal differences in the disturbance regimes auf Austria (more info in the paper here). For this we had a nice dataset available courtesy of our colleages over at BFW, covering the full 3.99 Mill. ha of Austria’s forests for the period of 2002-2010. As we started wading through the exceedingly long list of potential explanatory variables for which we hypothesized an influence on disturbance regimes we quickly realized that we’d need to structure our analysis along some overarching theory/ hypothesis in order to make (ecological) sense of all the possible statistical relationships in such a dataset.

After some thinking we turned to a quite seasoned theory of tree death first put forward by Paul Manion in the 1980s. In his theory Manion suggests that slow “background” factors predispose a tree to be affected by disturbance and death, while fast, inciting factors are what trigger the actual occurrence of such an event. This distinction along the lines of different temporal scales has more recently also been proposed as an important approach to understand complex systems. Applying this approach sequentially in our study of the Austrian disturbance regime (i.e., first controlling for slowly changing spatial differences in predisposition (e.g., tree species composition), and subsequently investigating the effect of inciting factors such as the inter-annual changes in weather patterns) showed that both variable groups explained a significant portion of the spatio-temporal variance in disturbance. In other words, when we include variables at different (temporal) scales we increase our inferential power, and learn more about the system than if we were to focus solely on a single temporal scale.

What’s more is that the analysis at different scales also revealed quite interesting (and at first glance counter-intuitive) results with regard to ecological processes: As expected, we found the occurrence of gale-force gusts (fast variable) to be positively related to damage by wind disturbance. However, when considering windiness over longer periods of time (mean windiness, i.e. a slow variable of average wind exposure), we found that the sign changed, and – all else being equal – trees in windy regions accrued less damage from wind disturbance. This indicates a significant acclimation effect of forests through structural adaptation of the tree’s anchorage to wind loading. It also shows that evaluating risk (and changes therein) by simply extrapolating observations made at short time scales might lead to errors by neglecting the feedbacks and adaptive capacity of ecosystems. Thus, for addressing such questions I say: consider going multiscale!