Together, the world’s oceans and land surfaces store away more than half of the carbon dioxide (CO₂) emitted by fossil fuel burning and cement production. The current strength of the land sink, which accounts for roughly half of this storage, is partly due to a fertilizing effect of increased atmospheric CO₂ on plant growth. Over the past decades, much research has been dedicated to the resilience and optimization of this “ecosystem service”, seeking to answer three questions: first, what is its long-term fate, given among others that CO₂ fertilization follows a saturating curve, while the respiratory CO₂ emission from dead biomass decomposition increases progressively with temperature? Second, how can we keep our ever-increasing land use from unintentionally harming, or even tailor it to support climate change mitigation? And third, which side effects beyond the impact on the carbon cycle do we need to consider when doing so?

There is an ongoing debate about the first question, which is however outside the scope of our paper. One popular answer to the second question is to plant more trees, although warnings have been issued  about oversimplified claims regarding the potential of this measure, besides neglecting the services of non-forest ecosystems. Such claims could even further support a fatal reluctance in terminating fossil fuel emissions. One answer to the third question is that when assessing the climate effects of a land-use change, among other parameters we also need to carefully consider the albedo – that is, the fraction of incoming sunlight energy reflected back by a surface. Most forests absorb more light than most other land ecosystems, thus providing a biogeophysical “heating” that counteracts their biogeochemical “cooling” effect due to CO₂ storage.

In addition to their own optical properties, trees interrupt (“mask”) what would otherwise be a particularly bright snow surface on smooth ground. Partly because of this, the climate change mitigation potential of afforestation is most disputed in northern latitudes and less so in the tropics. Apart from this spatial dimension, the competition between CO₂ and albedo has a temporal one: Suddenly changing the surface albedo from an old to a new value exerts, in first approximation, a likewise step change in the warming or cooling. A similar sudden change of the CO₂ uptake of that surface, in contrast, exerts an initially weak warming or cooling effect that depends on the cumulative uptake, and thus becomes stronger each year - until the surface’s capacity to store or release any more carbon has been exhausted. Such “saturation” might only start after decades; in the few extreme cases which produced today’s fossil fuel reserves, accumulation continued for millennia. As a result, a compensation time (where CO₂ effects override albedo effects), or a full pathway with a time axis, is a more appropriate way to report a land-use change’s combined albedo-CO₂ effects, than a single value of radiative forcing or CO₂-equivalents suggesting one final answer. While influential early contributions to the debate were aware of this, we miss a clear commitment to the importance of the timeline in much of the present literature.

The current state of knowledge inspired us to dive into two underexamined questions. First, is it a general rule that increasing CO₂ uptake comes at the cost of a lower albedo, even when putting aside the well-known albedo contrast between forests and tree-free ecosystems? And second, if CO₂ effects dominate over long timescales and albedo effects during the first decades – is there a land use strategy to yield a net cooling effect all the time?

The workhorse among the data sources of the study was FLUXNET, particularly its branches AMERIFLUX and ICOS –networks of stations monitoring the greenhouse gas, water and anergy budget of representative ecosystems around the globe – from heavily agriculturally managed to largely natural ones. Comparing their measured albedo and CO₂ flux values showed indeed a roughly anti-correlation-like pattern. This pattern still persisted when looking at forests only, tree-free surfaces only, and snow-free climates only. In other words, the albedo “cost” of increasing CO₂ uptake applies in a seemingly universal manner not only when re- or afforesting, but also when replacing one forest by another, or one tree-free ecosystem by another, in the absence of any snow.

While this is not encouraging news for these climate mitigation efforts, the good news is that the pattern did not apply to the data cloud as such, but to its upper limit. In other words: Ecosystems with a higher CO₂ uptake have a lower maximum limit to their possible albedo, and vice versa; but most examined sites failed to be anywhere near that maximum. At these sites, it may be possible to either increase albedo while maintaining CO₂ uptake, increase CO₂ uptake while maintaining albedo, or even increase both. One example, which has meanwhile been examined in case studies, is the comparatively dark soil surface in many regions of intensive agricultural crop production, which is left exposed during the fallow period. Where climate allows, a vegetation cover persisting throughout the year could both increase the albedo and sequester more CO₂ (into living biomass in the short term and soil organic carbon in the long term). One way to establish such a longer-lasting vegetation cover are cover crops, which are increasingly applied not deliberately because of their albedo or CO₂ uptake, but because of biodiversity and fertilizer-saving effects.

The reasons for a site to not have an optimal combination of albedo and CO₂ uptake may be manifold. But the above example illustrates that one of them can be an intense but inefficient human management of sites that would have provided better climate services under their natural vegetation – and might do so even under usage for crop production, if only that usage is “smart” and inspired by nature. Notwithstanding the underlying mechanisms and possible ecological and economical limitations of each single case, we conceptually simulated a land use change motivated by climate mitigation, assuming that the vegetation cover and thus CO₂ uptake and albedo of each site can be converted to the one of any other site within the same climate. This approach allowed us to estimate the global net effect of strategies that strive to optimize CO₂ uptake, albedo, or both at the same time. According to these simulations, optimizing albedo would provide immediate cooling, which however wears off over time. From about 25 years after conversion onwards, there is an increasing risk that the warming effect of the cumulative lack in CO₂ storage prevails. If we wanted to correct our decision at this point in the future, we would at first create even further warming due to the lower albedo of CO₂-efficient land use systems – a veritable deadlock. In contrast, if we favour land-use sequestering much CO₂ right from the start of the simulation, about 20 years of initial unwanted warming effect would be followed by a cooling effect still increasing at the end of the century – and after less than 40 years surpassing even the strongest cooling effect of the “albedo optimization” strategy.

But can we afford the initial warming? We have lingered so long in stopping fossil fuel emissions that even optimistic IPCC scenarios include a temporary overshoot of the Paris target around the mid-21st century . To avoid triggering the tipping points which motived this same target, any extra warming from land-use change, even with the best intention, should obviously not occur during this overshoot period. Therefore, the window of opportunity for pure "CO₂ optimization" is rapidly closing during the next years. In the near future, it might be better to moderately increase both CO₂ uptake and albedo at sites where it is possible, and elsewhere go for the solution minimizing the unwanted side effect. This “balanced” scenario yielded a global cooling effect at all timescales, albeit initially not quite as strong as “albedo optimization” and towards the end of the century not as strong as “CO₂ optimization”.